article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
w uma contact binaries are very common eclipsing variables in which the eclipsing light curves have nearly equal minima .
@xcite classified w uma contact binaries into a- or w - type on the basis of their light curves .
the secondaries of w uma contact binaries have an unusual mass - luminosity relationship which was first recognized by @xcite .
@xcite proposed that it is caused by the energy transfer from the primary ( the more massive component ) to the secondary ( the less massive component ) within a convective envelope .
but the mechanism causing energy transfer between the two components of w uma contact binaries and the effect of the energy transfer on the components are not clear .
w uma contact binaries are an important class of eclipsing variables in several respects . in studies of galactic structure ,
w uma contact binaries play an important role because they have high spatial frequency of occurrence , ease of detection , and provide a standard candle for distance determinations @xcite .
more importantly , w uma contact binaries are interesting objects due to the mass and energy transfer between two components .
understanding the energy and mass transfer in the common envelope is necessary to develop a correct theory of the structure and evolution of w uma systems .
therefore , the investigation of the mechanism causing energy transfer is a core problem for understanding the structure and evolution of w uma contact binaries @xcite .
the energy transfer in w uma systems has been investigated by many authors @xcite .
@xcite calculated the relative energy transfer rate of w uma contact binaries using the normal mass - luminosity relation for independent stars and found that the relative energy transfer rate depends only on the mass ratio of w uma contact binaries .
@xcite found that the relative energy transfer rate increases with increasing mass ratio continuously based on 22 contact binaries .
@xcite found that the energy transfer rate depends not only on the mass ratio but also on the evolutionary degree of the primary . by studying a catalogue data of 159 systems
, @xcite found that the energy transfer rate is a function of the mass and luminosity ratio .
the treatment of the energy transfer is very important to construct theoretical models of w uma contact binaries .
although it seems probable that the energy transfer occurs in the common envelope of w uma systems , it is not clear at present where and how the transfer is taking placed in the common envelope . the structure and evolution of w uma contact binaries have been investigated by several authors in recent years @xcite .
it is found that loss of contact is avoided if the energy transfer is assumed to be sufficiently effective @xcite .
@xcite discussed the region of energy transfer in the common envelope of w uma contact binaries , and showed that the energy transfer may take place in the radiative region of the common envelope .
@xcite suggested that the mechanism of the energy transfer may be differential rotation which has been observed by helioseismology in the solar convection zone @xcite .
the energy transfer from the primary to the secondary in w uma contact binaries will restructure the secondary and make it oversized and overluminous for its mass @xcite .
the radii of the secondaries of w uma contact binaries ( including a- and w - types ) are obviously altered from those of zero - age main sequence ( zams ) stars @xcite .
this might be the result of the energy transfer from the primary to the secondary @xcite .
the temperatures and densities of the secondaries of w uma contact binaries are also affected by the energy transfer .
@xcite calculated the response functions which describe the effect of energy transfer on the radii and effective temperatures of the stars .
but the relation between the energy transfer and the reaction of the secondaries of w uma systems is not completely understood . in this paper ,
the physical parameters of 133 w uma contact binaries have been collected from the literature . using these data ,
the energy transfer and its effect on the secondaries in w uma systems are investigated .
table 1 . the new or recently obtained physical parameters of contact binaries .
+ [ cols="<,^,^,^,^,^,^,^,^,^,^,^,^ " , ] + columns : stars - gcvs name of star ; @xmath4-orbital period ; @xmath5-mass of the primary ; @xmath6-mass of the secondary ; @xmath7-luminosity of the + primary ; @xmath8-luminosity of the secondary.@xmath9-radius of the primary;@xmath10-radius of the secondary ; @xmath11-effective temperature of the + primary ; @xmath12- effective temperature of the secondary + references in table 1 : ( 1 ) awadalla & hanna 2005 ; ( 2 ) gazeas et al . 2006 ; ( 3 ) zdemir et al . 2002 ; ( 4 ) yakut et al . 2005 ; ( 5 ) zoa et al .
2001 ; ( 6 ) yang et al .
2005 ; ( 7 ) szalai et al .
2007 @xcite defined a relative energy transfer rate which is the ratio of the transferred luminosity to the surface luminosity of the primary . * based on the assumption that the primaries are zams , the relative energy transfer rate @xmath1 can be written as @xmath13 where @xmath14 are the surface luminosity of the primary and the secondary ; @xmath15 and @xmath16 are their nuclear luminosities , respectively @xcite .
we can give another relative energy transfer rate as @xmath17 using equation ( 1 ) , we have @xmath18 @xmath19 where @xmath20,@xmath21 , @xmath0 is mass ratio and @xmath22 is the exponent of mass - luminosity relation and it is equal to 3.42 .
if it is assumed that the components in each w uma system are identical in the effective temperature , and that the components of each w uma contact binary just fill the inner roche lobes ( i.e. @xmath23 ) , equation ( 4 ) and equation ( 5 ) can be written as @xmath24 @xmath25 the theoretical curves and the observational data are shown in figure 2 and figure 3 with a solid line and the open stars ( w - subtypes ) or the solid squares ( a - subtypes ) , respectively .
it is seen in figure 2 and figure 3 that although the theoretical curves can reflect the distribution of the @xmath1 @xmath3 @xmath0 and @xmath2 @xmath3 @xmath0 of the observed data , the observational points are largely scattered and some observed systems are significantly deviated from the solid line .
the deviation might be caused by the applicability of the basic assumptions .
* in order to find the applicability of the basic assumptions , we take @xmath26 , @xmath27 , and @xmath22 to be different values . at first , we must inspect the applicability of a basic assumption that the components of each w uma system just fill their inner roche lobes ( i.e. @xmath23 ) .
in fact , most observed systems are over - contact binaries , and they should not satisfy this relation . the relation between the logarithms of the radius ratio ( @xmath28 ) and the logarithms of the mass ratio @xmath0 of the observed systems is shown in figure 4 . as seen from figure 4 ,
the logarithm of the radius ratio is almost linearly changed with the logarithms of the mass ratio .
a least - squares solution leads to the following relation , @xmath29 according to equation ( 8) , @xmath30 , which is indeed different from roche approximation relation @xmath31 .
these relations are also plotted in figure 4 with a dashed line and a solid line , respectively . using the relation , @xmath30 , of the observed systems , equation ( 4 ) and equation ( 5 )
are shown in figure 2 and figure 3 with a dot - dashed line , respectively .
as seen from figure 2 and figure 3 , the dot - dashed lines are very similar to the solid ones .
therefore , the assumption that the components of each w uma systems just fill their inner roche lobes is acceptable .
secondly , the different values of @xmath22 are adopted by the different investigators ( 4.0 , @xcite , 4.6 , @xcite ) . if @xmath32 ( the largest value ) , the relative distributions between the observed systems and the theoretical curves are similar to those shown by figures 2 and 3 , although the observed points and theoretical curves have been shifted since they are shifted with the same direction .
this suggests that the exponent of the mass - luminosity relation has little effects on the relative distribution between the observed systems and the theoretical curves .
so , the assumption that the components in w uma systems are zams is also acceptable .
finally , we inspect the assumption of the two components with equal temperature in each w uma system .
the distribution of @xmath27 for the observed w uma systems is located in a region from 0.88 to 1.097 .
we take @xmath27 to be 0.9 and 1.1 .
we plot the resulting curves in figure 2 and figure 3 with two dashed lines . as seen from figure 2 and figure 3
, these two dashed lines can cover the large scattering of the observed systems .
therefore , the deviation caused by unequal effective temperatures of the components of w uma systems are larger than those caused by other two assumptions .
this suggests that the equal - temperature assumption for the components of w uma systems is the most unreasonable one in three basic simplifying assumptions .
the temperature difference is also an important parameter for investigating the classification and the energy transfer of w uma contact binaries .
it is defined by @xcite as @xmath33 the temperatures of w uma binary components can differ substantially , and this difference was explained by different proportions of the energy exchange in superadiabatic and adiabatic part of envelopes @xcite . @xcite investigated the relations between the temperature difference and other observational parameters , and found that the temperature difference for w - type systems is not correlated with the mass ratio , the fill - out parameter or the color . the relation between the relative energy transfer rate ( @xmath1 ) and the temperature difference is shown in figure 5 .
w uma contact binaries seem to populate a strip limited by two solid lines , and there is a tendency for decreasing temperature difference with increasing relative energy transfer rate .
this suggests that the temperature difference is correlated with the relative energy transfer rate ( @xmath1 ) .
this also indicates that the temperature of the secondary increases with increasing relative energy transfer rate and even exceeds the temperature of the primary if the relative energy transfer rate ( @xmath1 ) is large enough , and that a thermal coupling exists in the two components of w uma contact binaries .
main - sequence stars with @xmath34 have little or no convective envelope ; however , main - sequence stars with @xmath35 have a convective envelope and a radiative core @xcite .
@xcite argued that convection is by no means essential to heat transport in the common envelope of w uma systems by employing eggleton s stellar evolution code @xcite .
this means that the temperature difference should not depend on the mass of the primary .
the distribution of the temperature difference @xmath36 @xmath3 the primary s mass @xmath5 is shown in figure 6 . as seen from figure 6 ,
the distribution of temperature difference of the observed systems with @xmath37 is similar to that of the systems with @xmath38 .
this suggests that there is no correlation between the temperature difference and the mass of the primary , and that the efficiency of energy transfer in the common envelope of w uma systems is indeed not significantly enhanced by convective motion .
the mass - radius relations of the secondaries for a- and w - type of w uma contact binaries are different from that of zams stars @xcite .
this is a result of the energy transfer from the primary to the secondary @xcite .
but the relation between the energy transfer and the radius of the secondary is not clear .
the relation between the relative radius change of the secondary * ( @xmath39 ) * and the relative energy transfer rate ( @xmath2 ) of w uma contact binaries is plotted in figure 7 .
@xmath40 is the radius of the main sequence secondary in w uma systems without the effect of the energy transfer , i.e. it is the radius of * zams with * mass @xmath6 . according to @xcite ,
it can be expressed as @xmath41 where @xmath6 is the mass of the secondary in solar units . as seen in figure 7 ,
the relative radius change of the secondary is correlated with the relative energy transfer rate @xmath2 , i.e. * the radius of the secondary in a w uma system increases with increasing @xmath2*. this indicates that the energy transfer exercises a decisive influence on the secondaries of a - type and w - types . by using a linear fitting ,
the relation between the relative radius change of the secondary ( @xmath42 ) and the relative energy transfer rate ( @xmath2 ) can be written as @xmath43 this correlation is a result of the response of the secondary to the energy transfer from the primary .
it is the energy transfer that makes the radii of the secondaries of w uma contact binaries deviate from those of zams stars @xcite .
however , a w uma system vw cep has a radius smaller than that given by equation ( 10 ) .
vw cep is one of triple systems @xcite .
so the smaller secondary of vw cep might be attributed to an inaccurate spectroscopic solution due to the presence of the additional companion .
@xcite argued that the over - luminosity of the secondary is related to its density .
this is also due to the effect of the energy transfer from the primary to the secondary on the density of the secondary .
the relation between the relative density change of the secondary * ( @xmath44 ) * and the relative energy transfer rate ( @xmath2 ) is presented in figure 8 . as shown from figure 8 , the relative density change increases with increasing relative energy transfer rate .
this suggests that the more the secondary gets energy from the primary , the lower the density of the secondary becomes .
a secondary obtaining more energy from the primary would swell more greatly , and then its density would become smaller .
in this paper , we investigate * the energy transfer * of w uma contact binaries based on a sample of 133 w uma contact binaries , and then we study the effects of the energy transfer on the secondaries of w uma systems . * based on the assumptions(the components are contact configurations with nearly uniform effective temperature and the primaries are zams ) , the relations are given between the relative energy transfer rates and the mass ratio of w uma systems .
the theoretical curves can reflect the distribution of @xmath1 @xmath3 @xmath0 and @xmath2 @xmath3 @xmath0 .
but some observation systems are significantly deviated from these curves . by comparing the observational data and these resulting curves
, it is found that the deviations are mainly resulted from the difference in the effective temperatures of the components in w uma systems .
this means that the assumption that the components are uniform in effective temperature should be restrainedly applied to investigate the energy transfer in w uma systems . * the distribution of the temperature difference @xmath3 the mass of the primary suggests that the convection does not affect the efficient of energy transfer between two components in the common envelope of w uma contact binaries .
this suggests that the energy transfer in w uma systems does not depend on the property of the common envelope of w uma contact binaries and the convection is by no means essential to heat transport in the common envelope of w uma systems .
this also suggests that the energy transfer might occur in radiative region of common envelope of w uma contact binaries @xcite or the mechanism of energy transfer might be the differential rotation @xcite or circulation currents ( * ? ? ?
* ; * ? ? ?
* and references therein )
. the energy transfer from the primary to the secondary would lead w uma systems to be not in thermal equilibrium , then lead w uma contact binaries to suffer thermal relaxation oscillations ( tro ) .
however , the energy transfer is also related to the evolutionary degree of the primary , i.e. the higher is the evolutionary degree of the primary , the lower is the energy transfer rate @xcite .
this suggests that with the evolution of w uma systems , the thermal relaxation oscillation might be disappeared if the evolutionary degree of the primary is high enough .
if the energy transfer rate decreases in the evolved w uma systems , the rate of mass transferred from the secondary to the primary should become smaller and smaller with the evolution of w uma systems , and the decrease in mass ratio of the systems would become slower and slower , so that the lifetime of w uma systems might become longer than the prediction of the theory models @xcite .
figure 7 shows that the relative radius change of the secondary increases with increasing relative energy transfer rate @xmath2 . @xcite and @xcite suggested that the the deviation of the radius of the secondary from that of zams stars is probably the result of the energy transfer from the primary to the secondary in w uma contact binaries .
the energy transfer from the primary to the secondary is more than the energy generated in the core of the secondary , so the radius and density of the secondaries is significantly influenced by the energy transfer and their radius and density deviates from those of zams stars .
@xcite calculated the effects of energy transfer on the radius and temperature of stars and gave the response functions for the radius and temperature of the stars . in present paper , we have given a relation between the relative radius change of the secondary and the relative energy transfer rate @xmath2 .
this relation probably provides a useful information in the structure of the secondaries in w uma contact binaries , and can help us to understand the structure and evolution of w uma systems .
the authors are grateful to an anonymous referee for his / her valuable suggestions and insightful remarks , which have improved this paper greatly .
this work was partly supported by the chinese natural science foundation ( 10673029 , 10773026 , 10433030 and 10521001 ) , and by the yunnan natural science foundation ( 2007a113 m and 2005a0035q ) . | based on the physical parameters of 133 w ursae majoris ( w uma)-type contact binaries , the energy transfer and its effects on the secondary in w uma contact binaries are investigated .
relations are given between the mass ratio ( @xmath0 ) for w uma contact binaries and the relative energy transfer rates , i.e. @xmath1 , the ratio of the transferred luminosity to the surface luminosity of the primary and @xmath2 , the ratio of the transferred luminosity to the nuclear luminosity of the secondary . * the theoretical curves(@xmath1 @xmath3 @xmath0 and @xmath2 @xmath3 @xmath0 ) are derived based on the various assumptions that the two components in each w uma system are nearly identical in effective temperature , they just fill their inner roche lobes , and the primaries are zams stars .
although these curves can reflect the distribution of @xmath1 @xmath3 @xmath0 and @xmath2 @xmath3 @xmath0 , some observational systems are significantly deviated from these curves .
it is mainly resulted from the difference in the effective temperatures of the components in w uma systems.*the radius and the density of the secondary are related to the relative energy transfer rate @xmath2 : the higher is @xmath2 , the greater is the expansion and the lower is the density of the secondaries in w uma systems .
in addition , it is found that the temperature difference of w uma binary components is correlated with the relative energy transfer rate @xmath1 and decreases with increasing @xmath1 .
this might suggest that there is a thermal coupling between two components in w uma contact binaries , and that the classification of w uma contact binaries into a- or w - types depends on the energy transfer from the primary to the secondary .
the temperature difference of w uma binary components is poorly correlated with the mass of the primary .
this suggests that the properties of the common envelope of w uma contact binaries might not have a significant effect on the energy transfer between two components .
ctextemp_psfig.sty [ firstpage ] binaries : eclipsing stars : statistics stars : evolution |
in this section we will follow @xcite .
let @xmath12 , @xmath13 , be the generators of @xmath9 in the symmetric irreducible representation @xmath14 of dimension @xmath15 . they satisfy @xmath16=if_{abc}t_c~\label{comm}\end{aligned}\ ] ] and @xmath17 ( where @xmath18 are the usual gell - mann matrices ) be the generators of @xmath9 in the fundamental representation @xmath19 of dimension @xmath20 .
they also satisfy @xmath21 the @xmath22dimensonal generator @xmath12 can be obtained by taking the symmetric product of @xmath23 copies of the fundamental @xmath24dimensional generator @xmath25 , viz @xmath26 in the continuum @xmath0 is the space of all unit vectors @xmath27 in @xmath28 modulo the phase .
thus @xmath29 , for all @xmath30 , define the same point on @xmath0 .
it is obvious that all these vectors @xmath29 correspond to the same projector @xmath31 .
hence @xmath0 is the space of all projection operators of rank one on @xmath28 .
let @xmath32 and @xmath33 be the hilbert spaces of the @xmath9 representations @xmath14 and @xmath19 respectively .
we will define fuzzy @xmath0 through the canonical @xmath9 coherent states as follows .
let @xmath34 be a vector in @xmath35 , then we define the projector @xmath36 the requirement @xmath37 leads to the condition that @xmath34 is a point on @xmath0 satisfying the equations @xmath38=0~,~n_a^2=\frac{4}{3}~,~d_{abc}n_an_b=\frac{2}{3}n_c.\end{aligned}\ ] ]
we can write @xmath39 we think of @xmath40 as the coherent state in @xmath33 ( level @xmath41 matrices ) which is localized at the point @xmath34 of @xmath0 . therefore the coherent state @xmath42 in @xmath32 ( level @xmath43 matrices ) which is localized around the point @xmath34 of @xmath0 is defined by the projector @xmath44 we compute that @xmath45 hence it is natural to identify fuzzy @xmath0 at level @xmath15 ( or @xmath46 ) by the coordinates operators @xmath47 they satisfy @xmath48=\frac{2i}{n}f_{abc}x_c~,~x_a^2=\frac{4}{3}(1+\frac{3}{n})~,~d_{abc}x_ax_b=\frac{2}{3}(1+\frac{3}{2n})x_c.\end{aligned}\ ] ] therefore in the large @xmath7 limit we can see that the algebra of @xmath49 reduces to the continuum algebra of @xmath50 .
hence @xmath51 in the continuum limit @xmath52 .
the algebra of functions on fuzzy @xmath46 is identified with the algebra of @xmath53 matrices @xmath54 generated by all polynomials in the coordinates operators @xmath49 .
recall that @xmath15 .
the left action of @xmath9 on this algebra is generated by @xmath14 whereas the right action is generated by @xmath55 .
thus the algebra @xmath54 decomposes under the action of @xmath9 as @xmath56 a general function on fuzzy @xmath46 is therefore written as @xmath57 @xmath58 are @xmath9 polarization tensors in the irreducible representation @xmath59 . @xmath60 and @xmath61 are the square of the isospin , the third component of the isospin and the hypercharge quantum numbers which characterize @xmath9 representations .
the derivations on fuzzy @xmath46 are defined by the commutators @xmath62 $ ] .
the laplacian is then obviously given by @xmath63 $ ] .
fuzzy @xmath46 is completely determined by the spectral triple @xmath64 .
now we can compute @xmath65 @xmath66 are @xmath9 polarization tensors defined by @xmath67 furthermore we can compute @xmath68p_n=<\vec{n},n|[t_a , f]|\vec{n},n>=({\cal l}_af_n)(\vec{n})~,~{\cal l}_a =- if_{abc}n_b{\partial}_c.\end{aligned}\ ] ] and @xmath69 the star product on fuzzy @xmath46 is found to be given by @xcite @xmath70
we will introduce fuzzy gauge fields @xmath71 , @xmath13 , through the covariant derivatives @xmath72 , @xmath13 , as follows @xmath73 @xmath72 are @xmath53 matrices which transform under the action of @xmath74 as follows @xmath75 where @xmath76 . hence @xmath71 are @xmath53 matrices which transform as @xmath77 $ ] . in order that the field @xmath78 be a @xmath11 gauge field on fuzzy @xmath46 it must satisfies some additional constraints so that only four of its components are non - zero .
these are the tangent components to @xmath46 .
the other four components of @xmath78 are normal to @xmath46 and in general they will be projected out from the model .
let us go back to the continuum @xmath0 and let us consider a gauge field @xmath71 is a function on continuum @xmath0 as opposed to the @xmath71 in the fuzzy setting which is an @xmath53 matrix .
] , @xmath13 , which is strictly tangent to @xmath0 . by construction
this gauge field must satisfy @xmath79 @xmath80 is the projector which defines the tangent bundle over @xmath0 .
the normal bundle over @xmath0 will be defined by the projector @xmath81 .
explicitly these are given by @xmath82 in above we have used the fact that the generators in the adjoint representation @xmath83 satisfy @xmath84 .
remark that we have the identities @xmath85 .
hence the condition ( [ cond ] ) takes the natural form @xmath86 this is one condition which allows us to reduce the number of independent components of @xmath71 by one .
we know that there must be three more independent constraints which the tangent field @xmath71 must satisfy since it has only @xmath87 independent components . to find them we start from the identity @xcite @xmath88.\label{24}\end{aligned}\ ] ] thus @xmath89.\end{aligned}\ ] ] by using the fact that @xmath90 we obtain @xmath91 hence it is a straightforward calculation to find that the gauge field @xmath71 must also satisfy the conditions @xmath92 in the case of @xmath2 the projector @xmath80 takes the simpler form @xmath93 and hence @xmath94 . from equation ( [ plk55 ] )
we have on @xmath0 @xmath95 if we choose to sit on the `` north pole '' of @xmath0 , i.e @xmath96 then we can find that @xmath97 and as a consequence @xmath98 .
so @xmath99 , @xmath100 correspond to the normal directions while @xmath99 , @xmath101 correspond to the tangent directions . indeed by substituting @xmath96 in equation ( [ cond2 ] ) and using @xmath102 where @xmath103 and @xmath104 where @xmath105 and @xmath106 we get @xmath107 which is what we want .
in fact ( [ cond2 ] ) already contains ( [ cond1 ] ) .
in other words it contains exactly the correct number of equations needed to project out the gauge field @xmath71 onto the tangent bundle of @xmath0 .
let us finally say that given any continuum gauge field @xmath71 which does not satisfy the constraints ( [ cond1 ] ) and ( [ cond2 ] ) we can always make it tangent by applying the projector @xmath80 .
thus we will have the tangent gauge field @xmath108 similarly the fuzzy gauge field must satisfy some conditions which should reduce to ( [ cond1 ] ) and ( [ cond2 ] ) . as it turns out constructing a tangent fuzzy gauge field using an expression like ( [ cond ] ) is a highly non - trivial task due to @xmath5 gauge covariance problems and @xmath6 operator ordering problems . however implementing ( [ cond1 ] ) and ( [ cond2 ] ) in the fuzzy setting is quite easy since we will only need to return to the covariant derivatives @xmath72 and require them to satisfy the @xmath9 identities ( [ idd ] ) , viz @xmath109 so @xmath72 are almost the @xmath9 generators except that they fail to satisfy the fundamental commutation relations of @xmath9 given by equation ( [ comm ] ) .
this failure is precisely measured by the curvature of the gauge field @xmath71 , namely @xmath110+f_{abc}d_c\nonumber\\ & = & i[t_a , a_b]-i[t_b , a_a]+f_{abc}a_c+i[a_a , a_b].\end{aligned}\ ] ] the continuum limit of this object is clearly given by the usual curvature on @xmath0 , viz @xmath111 $ ] . to check that this fuzzy gauge field @xmath71 has the correct degrees of freedom we need to check that the identities ( [ cond3 ] ) reduce to ( [ cond1 ] ) and ( [ cond2 ] ) in the continuum limit @xmath112 .
this fact is quite straightforward to verify and we leave it as an exercise . next we need to write down actions on fuzzy @xmath46 .
the first piece is the usual yang - mills action @xmath113 by construction it has the correct continuum limit .
@xmath114 is the normalized trace @xmath115 .
the second piece in the action is a potential term which has to implement the constraints ( [ cond3 ] ) in some limit .
indeed we will not impose these constraints rigidly on the path integral but we will include their effects by adding to the action a very special potential term .
in other words we will not assume that @xmath72 satisfy ( [ cond3 ] ) .
to the end of writing this potential term we will introduce the four normal scalar fields on fuzzy @xmath46 by the equations ( see equations ( [ cond3 ] ) ) @xmath116 and @xmath117 we add to the yang - mills action the potential term @xmath118 in the limit where the parameters @xmath119 and @xmath120 are taken to be very large positive numbers we can see that only configurations @xmath71 ( or equivalently @xmath72 ) such that @xmath121 and @xmath122 dominate the path integral which is precisely what we want .
this is the region of the phase space of most interest .
this is the classical prediction .
however in the quantum theory we will find that the parameter @xmath119 must be related to @xmath120 in some specific way in order to kill exactly the normal components of @xmath71 .
this result ( which we will show shortly in the one - loop quantum fuzzy theory ) is the quantum analogue of the classical continuum statement that equation ( [ cond2 ] ) contains already ( [ cond1 ] ) .
the total action is then given by @xmath123 ^ 2+if_{abc}d_ad_bd_c\bigg]+\frac{3n}{4g^2}tr_n\phi + \beta tr_n{\phi}^2+m^2 tr_n{\phi}_c^2.\end{aligned}\ ] ] this is essentially the same action considered in @xcite . however this action is different from the action considered in @xcite which is of the form @xmath124
^ 2+\frac{2i}{3}f_{abc}d_ad_bd_c\bigg].\end{aligned}\ ] ] the first difference is between the cubic terms which come with different coefficients .
the second more crucial difference is the presence of the potential term in our case .
the linear term in @xmath125 is actually a part of the yang - mills action .
the equations of motion derived from the action @xmath126 are @xmath127=0.\end{aligned}\ ] ] these are solved by the fuzzy @xmath46 configurations @xmath128 and also by the diagonal matrices @xmath129 we think of these diagonal matrices ( including the zero matrix ) as describing a single point in accordance with the ikkt model @xcite .
more interestingly is the fact that these equations of motion are also solved by the fuzzy @xmath130 configurations @xmath131 indeed in this case @xmath132=[d_i , f_{ib}]$ ] . for @xmath133
this is equal to @xmath134 because @xmath135 whereas for @xmath136 this is equal to zero because @xmath137 . in above @xmath138 are the generators of @xmath10 in the irreducible representation @xmath139 .
the equations of motion derived from the action @xmath140 are on the other hand given by @xmath141+\frac{1}{2g^2}f_{abc}f_{bc}+2\beta\{\phi , d_a\}+m^2\bigg(2d_{abc}\{{\phi}_c , d_b\}-\frac{2n+3}{3}{\phi}_a\bigg)=0.\end{aligned}\ ] ] now the only solutions of these equations of motion are the @xmath46 configurations ( [ cp2nf ] ) .
thus the potential term has eliminated the diagonal matrices ( [ cp2di ] ) as possible solutions .
in fact this classical observation will not hold in the quantum theory for all values of the parameter @xmath120 since there will always be a region in the phase space of the theory where the vacuum solution is not @xmath142 but @xmath143 .
however when we take @xmath120 to be very large positive number then we can see that @xmath142 becomes quantum mechanically more stable .
hence by neglecting the potential term we can not at all speak of the space @xmath46 since it will collapse rather quickly under quantum fluctuations to a single point .
the potential term has also eliminated the fuzzy @xmath130 configurations ( [ cp2s2 ] ) as possible solutions .
in fact even if we set @xmath144 in the above equations of motion the fuzzy @xmath130 configurations ( [ cp2s2 ] ) are not solutions . the other major difference between @xmath140 and @xmath126
is that if we expand around the fuzzy @xmath46 solution @xmath142 by writing @xmath145 and then substitute back in @xmath140 and @xmath126 we find that @xmath126 does not yield in the continuum limit the usual pure gauge theory on @xmath0 .
it contains an extra piece which resembles the chern - simons action ( although it is strictly real ) .
we skip here the corresponding elementary proof ( see the appendix ) .
@xmath140 will yield on the other hand the desired pure gauge theory on @xmath46 in the limit @xmath146 and hence it has the correct continuum limit .
if we do not take the limit @xmath147 then @xmath140 will describe a gauge theory coupled to @xmath87 adjoint scalar fields which are the normal components of @xmath71 .
the only motivation for @xmath126-as far as we can see- is its similarity to the fuzzy @xmath2 action which looks precisely like @xmath126 with the replacement @xmath148 .
this fuzzy sphere action was obtained in string theory in the limit @xmath149 when we have open strings moving in a curved background with an @xmath150 metric in the presence of a neveu - schwarz b - field .
however it is found that perturbation theory with @xmath126 is simpler than perturbation theory with @xmath140 .
more importantly it is found that @xmath126 allows for some new topology change which does not occur with @xmath140 .
in particular the transitions @xmath151 and @xmath152 are possible in the quantum theory of @xmath126 . using the background field method
we find that the one - loop effective action in the gauge @xmath153 is given by ( see the appendix ) @xmath154&=&{s}_i[d]+\frac{1}{2}tr_8tr\log{\omega}^i_{ab}-tr\log{\cal d}_a^2.\label{for}\end{aligned}\ ] ] where @xmath155&=&\frac{1}{g^2}tr_n\bigg(-\frac{1}{4}[d_a , d_b]^2+i{\alpha}_i f_{abc}d_ad_bd_c\bigg)+{\rho}_i tr_n\phi + { \beta}_itr_n{\phi}^2+m_i^2 tr_n{\phi}_c^2.\nonumber\\\end{aligned}\ ] ] and @xmath156 the trace @xmath157 corresponds to the left and right actions of operators on matrices whereas @xmath158 is the trace associated with @xmath159dimensional rotations . given a matrix @xmath160 the operator @xmath161 is given by @xmath162 $ ] , for example @xmath163 $ ] .
for @xmath126 we have @xmath164 , @xmath165 while for @xmath140 we have @xmath166 , @xmath167 , @xmath168 and @xmath169 .
[ [ fuzzy - bf - cp2-phase ] ] fuzzy @xmath0 phase + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let us first neglect the potential term in @xmath170 , i.e we will set @xmath171 or equivalently @xmath172 in @xmath140 .
the effective potential is given by the formula ( [ for ] ) , viz @xmath154&=&s_i[d]+\frac{1}{2}tr_8tr\log{\omega}^i_{ab}-tr\log{\cal d}_a^2\end{aligned}\ ] ] where the background field is chosen such that @xmath173 the reason is simply because we want to study the stability of the fuzzy @xmath46 vacuum @xmath142 against quantum fluctuations . hence @xmath174 is an order parameter which measures in a well defined obvious sense the radius of @xmath46 .
let us compute the classical potential in this configuration .
we have @xmath175&=&\frac{3|n|^2}{g^2}\bigg[\frac{1}{4}{\phi}^4-\frac{{\alpha}_i}{2}{\phi}^3+\frac{g^2{\rho}_i}{3n}({\phi}^2 - 1)\bigg].\label{clcp2}\end{aligned}\ ] ] the main quantum correction is equal to the trace of the logarithm of the laplacian @xmath176 which is given by the simple formula @xmath177 where @xmath178 $ ] .
there are two cases to consider . for both @xmath126 and @xmath140
we obtain the quantum correction ( see the appendix ) @xmath179&=&\frac{1}{2}tr_8tr\log({\phi}^2{\cal l}_a^2{\bf 1}_8)-tr\log({\phi}^2{\cal l}_a^2)\nonumber\\ & = & + 6n^2\log{\phi}+{\rm constant}.\end{aligned}\ ] ] [ [ case-1 ] ] case @xmath180 + + + + + + + + + + + + + + + + + + + + for @xmath126 we have @xmath181 and @xmath182 .
thus the quantum effective potential is @xmath183}{6n^2 } & = & \frac{2}{3n^2g^2}\bigg[\frac{1}{4}{\phi}^4-\frac{1}{3}{\phi}^3\bigg]+\log{\phi}+{\rm constant}.\label{rmconstant}\end{aligned}\ ] ] the quantum minimum of the model is given by the value of @xmath174 which solves the equation @xmath184 .
it is not difficult to convince ourselves that this equation of motion will admit a solution only up to an upper critical value @xmath185 of the gauge coupling constant @xmath186 beyond which the configuration @xmath187 collapses . at this value
@xmath185 the potential @xmath188 becomes unbounded from below .
the conditions which will yield the critical value @xmath185 are therefore @xmath189+\frac{1}{\phi}=0~,~ v_{\rm eff}^{''}=\frac{2}{3n^2g^2}\bigg[3{\phi}^2 - 2{\phi}\bigg]-\frac{1}{{\phi}^2}=0.\label{firs}\end{aligned}\ ] ] we find immediately @xmath190 above the value @xmath185 we do not have a fuzzy @xmath46 , in other words the space @xmath46 evaporates at this point .
this critical point separates two distinct phases of the model , in the region above @xmath191 we have a `` matrix phase '' while in the region below @xmath185 we have a `` fuzzy @xmath46 '' phase in which the model admits the interpretation of being a @xmath11 gauge theory on @xmath0 . by going from small values of @xmath186 ( @xmath192 corresponding to the `` fuzzy @xmath193 phase '' ) towards large values of @xmath186
we get through the value @xmath185 where the space @xmath46 decays .
looking at this process the other way around we can see that starting from large values of @xmath186 ( @xmath194 corresponding to the `` matrix phase '' ) and going through @xmath191 we generate the space @xmath46 dynamically .
it seems therefore that we have generated quantum mechanically the spectral triple which defines the space @xmath46 .
furthermore we note that in the `` matrix phase '' we have a @xmath74 gauge theory reduced to a point where @xmath7 is the size of the matrices since the minimum there is given by diagonal matrices .
the important point is that in this phase the gauge group is certainly not @xmath11 .
let us recall that in the `` fuzzy @xmath46 phase '' we had a @xmath11 gauge theory .
hence across the transition line between the `` fuzzy @xmath46 phase '' and the `` matrix phase '' the structure of the gauge group also changes .
thus we obtain in this model in correlation with the topology change across the critical line a novel spontaneous symmetry breaking mechanism . for the purpose of comparing with the numerical results of @xcite we define the coupling constant @xmath195 such that @xmath196
. then the critical value ( [ crit ] ) is seen to occur at @xmath197 this is precisely the result of the monte carlo simulation reported in equation @xmath198 of @xcite .
[ [ case-2 ] ] case @xmath199 + + + + + + + + + + + + + + + + + + + + for @xmath140 we have @xmath200 and @xmath167 and hence the effective potential is given by @xmath201}{6n^2 } & = & \frac{2}{3n^2g^2}\bigg[\frac{1}{4}{\phi}^4-\frac{1}{2}{\phi}^3+\frac{1}{4}{\phi}^2\bigg]+\log{\phi}+{\rm constant}.\label{rmconstant1}\end{aligned}\ ] ] a direct calculation yields the critical values @xmath202 this @xmath185 is smaller than the @xmath191 obtained in ( [ crit ] ) and hence the fuzzy @xmath46 is more stable in the model @xmath126 than it is in the model @xmath140 which is largely due to the linear term proportional to @xmath125 in @xmath140 .
in other words attempting to put true gauge theory on fuzzy @xmath46 causes the space to decay more rapidly .
however for @xmath126 the true vacuum is the fuzzy sphere @xmath130 and not the fuzzy @xmath46 as we will now discuss [ [ a - stable - fuzzy - sphere - phase ] ] a stable fuzzy sphere phase + + + + + + + + + + + + + + + + + + + + + + + + + + + we know that there is also a fuzzy sphere solution ( [ cp2s2 ] ) for the model @xmath126 .
we consider then the background field @xmath203 we want now to study the stability of this vacuum against quantum fluctuations .
the @xmath174 is now an order parameter which measures the radius of the fuzzy sphere @xmath130 .
the classical potential in this configuration is @xmath204 & = & \frac{2c_2}{g^2}\bigg[\frac{1}{4}{\phi}^4-\frac{1}{3}{\phi}^3\bigg].\label{clcp2s2}\end{aligned}\ ] ] in above @xmath205 is the casimir of @xmath10 in the irreducible representation @xmath139 ( @xmath15 ) .
it is clear that @xmath206 and @xmath207 in the large @xmath23 limit .
hence the action ( [ clcp2s2 ] ) around the classical minimum @xmath208 is much smaller than the classical action ( [ clcp2 ] ) . in other words
the fuzzy sphere is more stable than the fuzzy @xmath46 in this case .
the quantum corrections are given in this case by @xmath209 in above @xmath210 following the same arguments of the previous section ( only now it is @xmath10 representation theory which is involved ) we have in the large @xmath23 limit @xmath211 we can also argue that we have @xmath212 in other words the configurations ( [ lmnb ] ) although they are fuzzy sphere configurations they know ( through their quantum interactions ) about the other @xmath9 structure present in the model .
classically this @xmath9 structure is not detected at all by these configurations in the classical potential ( [ clcp2s2 ] ) .
the effective potential becomes in this case @xmath183}{6n^2 } & = & \frac{1}{12g^2}\bigg[\frac{1}{4}{\phi}^4-\frac{1}{3}{\phi}^3\bigg]+\log{\phi}+{\rm constant}.\label{rmconstant2}\end{aligned}\ ] ] a direct calculation yields the critical value @xmath213 in terms of the coupling @xmath214 define by @xmath215 the critical value @xmath191 reads @xmath216 this is again what is measured in the monte carlo simulation of the model @xmath126 as it is reported in equation @xmath217 of @xcite .
therefore we have a fuzzy sphere phase above @xmath218 and a matrix phase below @xmath218 .
the model @xmath126 can also be in a fuzzy @xmath46 phase for @xmath219 below the second value of ( [ crit ] ) which for large enough @xmath23 is much smaller than the value @xmath219 with @xmath220 given by the second equation of ( [ 763 ] ) .
however we have seen in the previous paragraph that this @xmath193 will decay rather quickly to a single point which ( by the discussion of the present section ) can only happen by going first across a fuzzy sphere phase .
we have then the transition pattern @xmath221 . in the limit where @xmath222 is kept fixed
we can see that the above critical value ( [ 763 ] ) is infinitely large which means that the model @xmath126 is mostly in the fuzzy sphere phase .
the matrix phase shrinks to zero and the fuzzy sphere is completely stable in this limit since the fuzzy @xmath46 phase can occur only at very small values of the coupling constant @xmath223 .
[ [ the - large - mass - limit ] ] the large mass limit + + + + + + + + + + + + + + + + + + + + now we include the effect of the potential term @xmath225 .
the relevant model is given by the action @xmath140 .
naturally the calculation becomes more complicated in this case .
the classical potential in the configuration @xmath187 is @xmath226}{6n^2 } & = & \frac{2}{3n^2g^2}\bigg[\frac{1}{4}{\phi}^4-\frac{1}{2}{\phi}^3+\frac{1}{4}{\phi}^2+\frac{g^2\beta}{9}({\phi}^2 - 1)^2+\frac{g^2m^2}{27}({\phi}^2-\phi)^2\bigg].\label{cl}\end{aligned}\ ] ] the most important quantum correction is given by the determinant of @xmath227 by using the identities ( [ idd ] ) and ( [ 24 ] ) we find that the extra contributions are given explicitly by the expressions @xmath228.\end{aligned}\ ] ] @xmath229.\label{plkmass}\end{aligned}\ ] ] in the continuum large @xmath7 limit the first extra correction behaves as @xmath230 let us introduce the projector @xmath231 .
this is a rank one normal projector which projects vector fields along the normal direction @xmath232 .
recall the rank four tangent projector @xmath233 and the rank four normal projector @xmath234
. then we must necessarily have @xmath235 where @xmath236 is a rank three normal projector which projects vector fields along the normal directions @xmath237 , @xmath238 .
it is given by @xmath239 .
we have the decomposition @xmath240 .
hence @xmath241 similarly in the continuum large @xmath7 limit the first three terms of @xmath242 takes the form ( by using also the identity @xmath243 ) @xmath244 let us remark that the coefficients in front of the projectors @xmath245 and @xmath246 are the masses of the normal components of the gauge field and hence they must be positive .
for example the mass of the normal components @xmath247 is given by @xmath248 where @xmath249 .
this is positive definite for all values @xmath250 of the radius of @xmath46 if @xmath251 is such that @xmath252 and @xmath253
. thus @xmath251 must be in the range @xmath254 .
since @xmath119 must be positive we obtain the condition @xmath255 the mass of the normal component @xmath256 is given by @xmath257 . the requirement that this mass must be positive definite gives now the condition that the radius @xmath174 can only be in the range @xmath258 we remark that for all allowed values of @xmath251 we have @xmath259 and @xmath260 so we can still access the limits @xmath261 and @xmath262 although there is now a forbidden gap between these two important regions .
the mass of the tangent components @xmath263 is given by @xmath264 .
this is not always positive in the range ( [ range ] ) .
however this mass formally vanishes in the limit @xmath265 where the most probable value of the radius of @xmath46 is @xmath266 .
finally the last correction of the inverse propagator coming from the addition of the potential @xmath225 ( which is given explicitly by the last term in ( [ plkmass ] ) ) is also negative .
remark that this correction is proportional to @xmath267 and as a consequence we will not need to compute it explicitly ( see below ) .
we are now ready to compute the determinant .
we have @xmath268 where @xmath269 thus @xmath270.\end{aligned}\ ] ] from the last two terms we get in the large @xmath271 limit the two delta functions @xmath272 and @xmath273 and as a consequence the determinant reduces to @xmath274 in above it is consistent to neglect the mass term @xmath275 since in the large mass limit @xmath265 this term is subleading as we have discussed .
the eigenvalues of the operator @xmath276 were computed in the appendix .
we found that the second term in @xmath277 ( which is proportional to @xmath278 ) and the third term ( which is proportional to @xmath279 ) can be neglected in the large @xmath23 limit compared to @xmath280 .
for example the eigenvalues of @xmath281 are given by @xmath282 with @xmath283 whereas the eigenvalues of @xmath276 are found to be at most linear in @xmath284 and hence in the large @xmath23 limit ( where large values of @xmath284 which are of the odrer of @xmath23 are expected to contribute the most ) we can make the approximation @xmath285 thus the quantum effective potential is @xmath286}{6n^2}&=&\frac{s_1[d]}{6n^2}+\frac{1}{6n^2}\bigg(\frac{1}{2}tr_8tr\log({\phi}^2p^t)-tr\log({\phi}^2)\bigg)+ .... \end{aligned}\ ] ] the last term comes from the ghost contribution .
the final result is @xmath287+\frac{1}{3}\log\phi.\end{aligned}\ ] ] the calculation of the critical values in terms of the mass parameters @xmath288 and @xmath251 is done in the same way as before and it yields the following equations .
the critical radius occurs at the solutions of the equation @xmath289{\phi}_*^2-\frac{9}{8}\big[1+\frac{4\hat{m}^2}{27}\big]{\phi}_*+\frac{1}{4}-\frac{\hat{m}^2}{9}(2\gamma -\frac{1}{3})=0.\label{lkj}\end{aligned}\ ] ] in the limit @xmath265 we get the solution @xmath290 the choice of the plus sign instead of the minus sign is so that when @xmath251 goes to zero ( in other words @xmath291 ) this solution will reduce to the first equation of ( [ crit1 ] ) .
this agreement is due to the fact that the limit @xmath292 is formally equivalent to the limit @xmath293 ( since @xmath294 ) .
indeed for very small values of @xmath271 we get the potential @xmath295+\log\phi.\end{aligned}\ ] ] this will also lead to the equation ( [ lkj ] ) which for @xmath293 admits the solution given by the first equation of ( [ crit1 ] ) .
the critical value of the coupling constant @xmath191 ( or equivalently @xmath296 ) is given on the other hand by the equation @xmath297.\end{aligned}\ ] ] hence in the limit @xmath265 we get the behavior @xmath298 the equation of motion @xmath299 could admit in general four real solutions where the one with the least energy can be identified with the radius of fuzzy @xmath46 .
this solution is found to be very close to @xmath180 .
however this is only true up to an upper value of the gauge coupling constant @xmath186 ( or equivalently a lower bound of @xmath195 ) for every fixed value of @xmath271 beyond which the equation of motion ceases to have any real solutions . at this value
the fuzzy @xmath46 collapses under the effect of quantum fluctuations and we cross to a pure matrix phase .
as the mass @xmath271 is sent to infinity it is more difficult to reach the matrix phase and hence the presence of the mass makes the fuzzy @xmath46 solution @xmath187 more stable .
in fact when @xmath300 the critical value @xmath301 approaches zero .
[ [ the - transition - bf - cp2longrightarrowbf - s2 ] ] the transition @xmath224 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we repeat the large mass analysis for the model @xmath126 . in other words
we add the potential @xmath225 to the action @xmath126 and study the effective potential when @xmath302 .
the interest in this action lies in the fact that it admits ( at least for @xmath303 ) a fuzzy sphere solution and hence we can contemplate a transition ( at the level of the phase diagram ) between fuzzy @xmath46 and fuzzy @xmath130 when we take the limit @xmath304 . as
before we consider fuzzy @xmath46 configurations @xmath187 , @xmath13 . for @xmath305 ( in other words non - zero values of @xmath271 and @xmath119 ) these configurations
are in fact the true vacuum as we have discussed previously .
when @xmath172 the fuzzy @xmath130 configurations become the true minimum .
the calculation of the quantum corrections with non - zero @xmath225 is exactly identical to what we have done in the previous paragraphs and thus we end up with the effective potential @xmath306+\frac{1}{3}\log\phi.\end{aligned}\ ] ] in the large @xmath271 limit we get the same critical value ( [ phi3 ] ) .
the critical value of @xmath186 ( or equivalently @xmath195 ) is found on the other hand to be given by @xmath307 so again in the large mass limit the fuzzy @xmath46 phase is stable even for the model @xmath126 . however we know from our previous discussion that in the limit @xmath293 the minimum of the model @xmath126 should tend to the fuzzy sphere solutions .
thus it is important to consider also the fuzzy sphere configurations @xmath308 .
the classical potential in these configurations becomes @xmath309}{6n^2 } & = & \frac{1}{12g^2}\bigg[\frac{1}{4}{\phi}^4-\frac{1}{3}{\phi}^3+\frac{\hat{m}^2\gamma c_2}{2n^2}({\phi}^2-\frac{|n|^2}{c_2})^2+\frac{\hat{m}^2}{2}\bigg(\bigg(\frac{2n+3}{6n^2}\bigg)^2{\phi}^2+\frac{c_2}{3n^4}{\phi}^4\bigg)\bigg ] .
\ ] ] the quantum corrections in the limit @xmath293 should be given by ( [ enarray1 ] ) and ( [ enarray2 ] ) .
we get then the effective potential @xmath310}{6n^2}&=&\frac{1}{12g^2}\bigg[\frac{1}{4}{\phi}^4-\frac{1}{3}{\phi}^3+\frac{\hat{m}^2\gamma c_2}{2n^2}({\phi}^2-\frac{|n|^2}{c_2})^2+\frac{\hat{m}^2}{2}\bigg(\bigg(\frac{2n+3}{6n^2}\bigg)^2{\phi}^2+\frac{c_2}{3n^4}{\phi}^4\bigg)\bigg]\nonumber\\ & + & \log\phi.\end{aligned}\ ] ] the critical values are @xmath311+o(\hat{m}^4).\end{aligned}\ ] ] @xmath312}{\longrightarrow}\frac{256\gamma}{3}\hat{m}^2+o(\hat{m}^4).\label{smallmass}\end{aligned}\ ] ] so when @xmath313 this @xmath296 goes to zero which is consistent with the result ( [ 763 ] ) .
this equation tell us how we actually approach this critical value @xmath314 .
the intersection of this equation with ( [ pre16 ] ) gives a one - loop estimation of the value @xmath315 at which the vacuum of the model @xmath126 goes from a fuzzy sphere @xmath130 to a fuzzy @xmath46 as we increase the mass parameter @xmath316 .
equivalently the intersection point occurs at the value @xmath315 at which the vacuum of the model @xmath126 goes from a fuzzy @xmath46 to a fuzzy sphere @xmath130 as we decrease @xmath316 .
in this article we have studied the one - loop effective potential for two models of @xmath11 gauge theory on fuzzy @xmath46 . the first model is given by the action @xmath305 and the second model is given by the action @xmath140 .
each model is characterized by @xmath1 parameters .
@xmath317 the gauge coupling constant @xmath318 or equivalently @xmath319 , @xmath320 the mass @xmath271 of the normal components of the @xmath159dimensional gauge field and @xmath321 the parameter @xmath322 which gives an extra mass for the normal component in the direction @xmath232 . the term in the action proportional to @xmath119 is not needed in the classical theory while in the quantum theory the parameter @xmath119 must be in the range ( [ range0 ] ) .
the order parameter ( the variable ) of the effective potential is the radius of the fuzzy @xmath46 and thus by studying the stability of this potential we can test the stability of the space as a whole against the effect of quantum fluctuations of the gauge field theory .
the second term in the effective potential ( the log term ) is not convex which implies that there is a competition between the classical potential and the logarithmic term which depends on the values of @xmath271 and @xmath186 .
the parameter @xmath119 plays no further role at this stage .
we found that there exists values of the gauge coupling constant @xmath186 and the mass @xmath271 for which the fuzzy @xmath46 solutions are not stable .
this instability is believed to be related to ( or is a reflection of ) the perturbative uv - ir mixing phenomena of the quantum gauge field theory .
the connection between the two effects was established explicitly for the case of the lower dimensional coadjoint orbit @xmath323 which is the case of the fuzzy sphere @xcite .
see also @xcite .
the phase structure of the @xmath11 models on fuzzy @xmath46 which are studied in this article reads as follows .
[ [ the - model - s_1 ] ] the model @xmath140 + + + + + + + + + + + + + + + + + + + + + + + + + + + .
this is the correct model which describes @xmath11 gauge theory in the continuum limit at least classically .
the minimum of the model ( for non - zero potential ) can only be fuzzy @xmath46 .
there are two phases . in the fuzzy @xmath46 phase we have a @xmath11 gauge theory on fuzzy @xmath46 whereas in the matrix phase the fuzzy @xmath46 configurations @xmath187 collapse and we end up with a @xmath74 gauge theory on a single point .
we have described in this article the qualitative behavior of a first order phase transition which occurs between these two regions of the phase space .
however it is obvious from the critical line ( [ pre16 ] ) that when the mass @xmath271 of the four normal scalar components of the @xmath159dimensional gauge field on fuzzy @xmath324 goes to infinity it is more difficult to reach the transition line . in this limit
the fuzzy @xmath46 phase dominates while the matrix phase shrinks to zero .
therefore we can say that we have a nonperturbative regularization of @xmath11 gauge theory on fuzzy @xmath325 .
[ [ the - model - s_0v_0 ] ] the model @xmath305 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + . this is a string - theory - inspired gauge model which does not go in the continuum limit to the usual @xmath11 gauge theory on @xmath0 even classically .
indeed it can be shown that it contains in the continuum limit ( in addition to the usual yang - mills term ) a chern - simons - like term ( see the appendix ) .
however this model has a more interesting phase structure since it allows for the ( quantum ) transitions between fuzzy @xmath130 and fuzzy @xmath46 .
the main reason behind this remarkable feature lies in the fact that when @xmath265 the absolute minimum of the model is the fuzzy @xmath46 configurations @xmath187 whereas in the limit @xmath293 the absolute minimum of the model is the fuzzy sphere configurations @xmath326 and @xmath327 .
the phase diagram of this model with the particular value @xmath328 is plotted in figure @xmath180 for illustration .
the phase diagram consists of @xmath1 phases . :
this is the region with @xmath329 and above the line ( [ pre16 ] ) where the absolute minimum of the model is the fuzzy @xmath46 configurations @xmath187 and where the field theory is some @xmath11 gauge theory on fuzzy @xmath46 .
recall that @xmath315 is the value of the mass parameter @xmath316 at which the two curves ( [ pre16 ] ) and ( [ smallmass ] ) intersect . the fuzzy @xmath46 phase dominates the phase diagram when @xmath330 .
: this phase shrinks to zero when @xmath330 .
this occurs at the points of the phase diagram which are below the two lines ( [ pre16 ] ) and ( [ smallmass ] ) . : these are the points which have @xmath331 and which are above the line ( [ smallmass ] ) where the absolute minimum of the model is the fuzzy @xmath130 configurations @xmath326 , @xmath327 and where the field theoy is a @xmath11 gauge theory on fuzzy @xmath130 with complicated coupling to @xmath332 adjoint scalars .
generalization to higher gauge groups @xmath333 with and without fermions should be straightforward if we are only interested in the effective potential and topology change .
similarly generalization to higher coadjoint orbits @xmath334 with @xmath335 should also be straightforward since the corresponding actions will be exactly of the same form as @xmath126 and @xmath140 and only we need to work with the group theory of @xmath336 s instead of @xmath9 .
in particular we expect that there will more possibilities for topology change in higher coadjoint orbits which relate to the fact that the group @xmath336 contains besides @xmath10 the groups @xmath9 , @xmath337 and many others ( for high enough @xmath7 ) as subgroups .
thus we may see transitions like @xmath338 , @xmath339 , @xmath340 , ... as well as transitions between the subspaces @xmath2 , @xmath0 , @xmath341 , .. and transitions from and to the matrix ( single point ) phase .
[ phase ] [ [ acknowledgements ] ] acknowledgements + + + + + + + + + + + + + + + + the work of d.d is supported by the college of science - research center project no : phys/2006/04.the work of d.dou is also supported in part by the associate scheme of abdus salam ictp . the work of badis ydri is supported by a marie curie fellowship from the commission of the european communities ( the research directorate - general ) under contract number mif1-ct-2006 - 021797 .
b.y would like also to thank the staff at humboldt - universitat zu berlin for their help and support .
in particular he would like to thank michael muller - preussker and wolfgang bietenholz .
first we can compute explicitly the following classical actions @xmath342 ^ 2 + 4if_{abc}d_ad_bd_c+3d_a^2\bigg).\end{aligned}\ ] ] @xmath343.\end{aligned}\ ] ] @xmath344.\end{aligned}\ ] ] in above we have used the identity ( [ 24 ] ) and the identity @xmath345 we can compute ( with @xmath346 $ ] ) @xmath347 next we will study the quantization of the action @xmath348&=&s_i[d]+tr_nj_ad_a~,~s_i[d]=\bar{s}_i[d]+\bar{v}_i[d]\end{aligned}\ ] ] where @xmath349&=&\frac{1}{g^2}tr_n\bigg(-\frac{1}{4}[d_a , d_b]^2+i{\alpha}_i f_{abc}d_ad_bd_c\bigg)\nonumber\\ \bar{v}_i[d]&=&{\rho}_i tr_n\phi + v_i={\rho}_i tr_n\phi + { \beta}_i tr_n{\phi}^2 + m_i^2 tr_n{\phi}_c^2.\label{okok}\end{aligned}\ ] ] for @xmath350 we have @xmath164 , @xmath165 while for @xmath351 we have @xmath166 , @xmath167 , @xmath168 and @xmath169 .
@xmath352 is a source .
we adopt the background field method to the problem of quantization of this model .
we write @xmath353 where @xmath354 is the background field and @xmath355 is the fluctuation field .
we will fix the gauge by adding to the action the gauge - fixing and fadeev - popov terms , viz we compute @xmath357 ^ 2&=&tr_n\bigg([b_a , b_b]^2 + 2[b_a , b_b][q_a , q_b]+2[b_a , q_b]^2 + 2[b_a , q_b][q_a , b_b]\nonumber\\ & + & 4[b_a , b_b][b_a , q_b]+o(q^3)\bigg).\end{aligned}\ ] ] and @xmath358 we find ( by using the identity @xmath359[b_b , q_a]=tr_n[b_a , q_a]^2-tr_n[b_a , b_b][q_a , q_b]$ ] ) the following expression @xmath349&=&\bar{s}_i[b]+\frac{i}{g^2}tr_n\bigg[b_{ab}-(1-\frac{3{\alpha}_i}{2})f_{abc}b_c , b_a\bigg]q_b-\frac{1}{2g^2}tr_n\bigg([b_a , q_b]^2\nonumber\\ & -&[b_a , q_a]^2 + 2iq_a\bigg[b_{ab}-(1-\frac{3{\alpha}_i}{2})f_{abc}b_c , q_b\bigg]\bigg)+o(q^3)\nonumber\\ & = & \bar{s}_i[b]+\frac{i}{g^2}tr_n\bigg[b_{ab}-(1-\frac{3{\alpha}_i}{2})f_{abc}b_c , b_a\bigg]q_b+\frac{1}{2g^2}tr_nq_a\bigg({\cal b}_c^2{\delta}_{ab}\nonumber\\ & -&{\cal b}_a{\cal b}_b-2i{\cal b}_{ab}+2i(1-\frac{3{\alpha}_i}{2})f_{abc}{\cal b}_c\bigg)q_b+o(q^3).\label{47}\end{aligned}\ ] ] @xmath360 is the curvature of the background curvature @xmath354 , in other words @xmath361+f_{abc}b_c$ ] .
remark also how this action simplifies for @xmath126 , i.e for @xmath181 .
this `` technical '' simplification is a major advantage in considering @xmath126 instead of @xmath140 . given a matrix @xmath160 the operator @xmath362 is given by @xmath162 $ ] , for example @xmath363 $ ] .
we can also compute @xmath364 @xmath365 @xmath366 @xmath367 and @xmath368 are the normal scalar fields corresponding to the background covariant derivative @xmath354 , viz @xmath369 and @xmath370 where @xmath371 .
let us now introduce the actions @xmath372=\bar{s}_i[d]+\bar{v}_i[d]+tr_nj_ad_a$ ] and @xmath373=\bar{s}_i[b]+\bar{v}_i[b]+tr_nj_ab_a$ ] . by using the above ingredients we have immediately the result @xmath348+s_{\rm g.fixing}+s_{\rm gh}&=&{s}_i[b , j]+tr_n{\cal j}_b^iq_b+\frac{1}{2g^2}tr_nq_a{\omega}_{ab}^iq_b+o(q^3)\nonumber\\ & + & tr_nb^+{\cal b}^2_ab.\end{aligned}\ ] ] in above @xmath374 and @xmath375
are given respectively by in the following we will assume that the background fields @xmath354 satisfy the equations of motion @xmath378 and we will choose the gauge @xmath379 .
we then obtain @xmath348+s_{\rm g.fixing}+s_{\rm gh}&=&{s}_i[b , j]+\frac{1}{2g^2}tr_nq_a{\omega}_{ab}^iq_b+o(q^3)+tr_nb^+{\cal b}^2_ab.\end{aligned}\ ] ] the fluctuation fields @xmath355 and the ghosts can be integrated out since they are gaussian and one obtains therefore the effective action @xmath380&=&{s}_i[b , j]+\frac{1}{2}tr_8tr\log{\omega}^i_{ab}-tr\log{\cal b}_a^2.\end{aligned}\ ] ] where @xmath381 for @xmath126 we have @xmath181 and @xmath182 and hence @xmath382=\frac{3|n|^2}{g^2}\bigg[\frac{1}{4}{\phi}^4-\frac{1}{3}{\phi}^3\bigg]+\frac{1}{2}tr_8tr\log{\omega}^0_{ab}-tr\log{\phi}^2{\cal l}_a^2\nonumber\\ & & { \omega}^0_{ab}={\phi}^2{\cal l}_c^2{\delta}_{ab}-{\phi}(\phi -1)\bigg[{\cal j}_c^2-{\cal l}_c^2-(adt_c)^2\bigg]_{ab}~,~\vec{\cal j}=\vec{\cal l}+{ad}\vec{t}.\end{aligned}\ ] ] the total @xmath9 angular momentum @xmath383 corresponds to the tensor product of the irreducible representations @xmath59 where @xmath283 ( corresponding to @xmath384 ) with the adjoint representation @xmath83 ( corresponding to @xmath385 ) . by using young tableaux
we obtain the decomposition @xmath386 the dimension of an irreducible representation @xmath387 of @xmath9 is @xmath388 and the quadratic casimir is @xmath389 .
thus we can immediately compute @xmath390 in this diagonal matrix the dimensions of the first block is @xmath391 , the second block is @xmath392 , the third block is @xmath393 , the @xmath87th block is @xmath394 and the @xmath395th block is @xmath396 .
it is obvious from the above equation ( [ c2p ] ) that the second term in @xmath397 is at most linear in @xmath284 while the first term is quadratic and hence in the large @xmath23 limit ( where large values of @xmath284 which are of the odrer of @xmath23 are expected to contribute the most ) we can make the approximation @xmath398 thus the quantum effective potential is @xmath399&=&s_0[d]+\frac{1}{2}tr_8tr\log({\phi}^2{\cal l}_a^2{\bf 1}_8)-tr\log({\phi}^2{\cal l}_a^2)\nonumber\\ & = & \frac{3|n|^2}{g^2}\bigg[\frac{1}{4}{\phi}^4-\frac{1}{3}{\phi}^3\bigg]+\frac{1}{2}(8)(n^2)\log{\phi}^2-(n^2)\log{\phi}^2+{\rm constant}\nonumber\\ & = & \frac{3|n|^2}{g^2}\bigg[\frac{1}{4}{\phi}^4-\frac{1}{3}{\phi}^3\bigg]+6n^2\log{\phi}+{\rm constant}.\end{aligned}\ ] ] recall that @xmath15 and @xmath371 and hence @xmath400}{6n^2 } & = & \frac{2}{3n^2g^2}\bigg[\frac{1}{4}{\phi}^4-\frac{1}{3}{\phi}^3\bigg]+\log{\phi}+{\rm constant}.\label{rmconstant}\end{aligned}\ ] ] the only difference ( as far as this quantum correction is concerned ) with case @xmath180 is that we have now a shifted laplacian @xmath402 so the result for the determinant already obtained will not be altered .
we end up thus with the effective potential @xmath400}{6n^2 } & = & \frac{2}{3n^2g^2}\bigg[\frac{1}{4}{\phi}^4-\frac{1}{2}{\phi}^3+\frac{1}{4}{\phi}^2\bigg]+\log{\phi}+{\rm constant}.\label{rmconstant1}\end{aligned}\ ] ] badis ydri , _ fuzzy physics _
, phd thesis ( 2001 ) , _ hep - th/0110006_. h.grosse,c.klimik,p.pre najder,_commun.math.phys .
_ 180 ( 1996 ) 429 , _
int.j.theor.phys . _ 35 ( 1996 ) 231 .
d.oconnor , _ mod.phys.lett .
_ a18 ( 2003 ) 2423 .
c.klimik , _ commun.math.phys .
_ 199 ( 1998 ) 257 .
a. p. balachandran , s. kurkcuoglu and s. vaidya,lectures on fuzzy and fuzzy susy physics ,
_ hep - th/0511114 _ .
a. connes , _ noncommutative geometry _ , academic press , london , 1994 .
g. landi , _ an introduction to noncommutative spaces and their geometry _ , springer ( 1997 ) .
j. m. gracia - bondia , j. c. varilly , h. figueroa , _ elements of noncommutative geometry _ , birkhauser ( 2000 ) .
h. aoki , s. iso , h. kawai , y. kitazawa , t. tada , _
_ 99 ( 1998 ) 713 - 746
. h. aoki , s. iso , h. kawai , y. kitazawa , t. tada , a. tsuchiya , _
_ 134 ( 1999 ) 47 - 83
. h. aoki , n. ishibashi , s. iso , h. kawai , y. kitazawa , t. tada , _ nucl.phys .
_ b565 ( 2000 ) 176 - 192 .
badis ydri , _ quantum equivalence of nc and ym gauge theories in 2d and matrix theory _ , _ hep - th/0701057_. badis ydri , _ the one - plaquette model limit of nc gauge theory in 2d _ , _ hep - th/0606206 _ , to be published in npb . | many gauge theory models on fuzzy complex projective spaces will contain a strong instability in the quantum field theory leading to topology change .
this can be thought of as due to the interaction between spacetime via its noncommutativity and the fields ( matrices ) and it is related to the perturbative uv - ir mixing .
we work out in detail the example of fuzzy @xmath0 and discuss at the level of the phase diagram the quantum transitions between the @xmath1 spaces ( spacetimes ) @xmath0 , @xmath2 and the @xmath3dimensional space consisting of a single point @xmath4 .
the approach of fuzzy physics @xcite @xmath5 to quantum geometry and @xmath6 to non - perturbative field theory insists on the use of finite dimensional matrix algebras with suitable laplacians ( metrics ) to describe the geometry @xcite .
fields will be described by the same matrix algebras or more precisely by the corresponding projective modules @xcite and the action functionals will be given by finite dimensional matrix models similar to the ikkt models @xcite .
as it turns out a topology change can occur naturally if we try to unify spactime and fields using this language of finite dimensional matrices .
this is precisely the picture which emerges from the perturbative and non - perturbative studies of noncommutative gauge theory on the fuzzy sphere @xcite .
indeed we found in the one - loop calculation @xcite as well as in numerical simulations @xcite and the large @xmath7 analysis @xcite that the noncommutative gauge model on the fuzzy sphere written in @xcite ( which is obtained in the zero - slope limit of string theory ) and its generalizations undergo first order phase transitions from the `` fuzzy sphere '' phase to a `` matrix '' phase where the fuzzy sphere vacuum collapses under quantum fluctuation .
the matrix phase is the space consisting of a single point .
this topology change from the two dimensional sphere to a single point and vice versa is intrinsically a quantum mechanical process and it is related to the perturbative uv - ir mixing phenomena .
this result was extended to the case of fuzzy @xmath8 in @xcite . in this article
we will go one step further and generalize this result to higher fuzzy complex projective spaces .
in particular we will work out the case of fuzzy @xmath0 in detail .
we find the possibility of first order phase transitions between @xmath0 and @xmath2 , between @xmath0 and a matrix phase and between @xmath2 and a matrix phase .
this richer structure of topology changes is due to the fact that @xmath9 contains also @xmath10 as a subgroup besides the trivial abelian subgroups @xmath11 . as we will explain generalization of our calculation to higher fuzzy complex projective spaces is obvious and straightforward .
for other approaches to topology change using finite dimensional matrix algebra and fuzzy physics see @xcite .
this article is organized as follows . |
the shortest timescales in the variability of the compact objects always was a very interesting topic of x - ray astronomy .
the efforts to measure the highest frequency in the variability of cyg x-1 can be traced back to rocket experiments in 70-ties ( e.g. @xcite , @xcite ) .
it was found that cyg x-1 demonstrates statistically significant variability of the x - ray flux at least up to frequencies of @xmath310 hz .
since then cyg x-1 was extensively observed by various satellite missions searching for millisecond and submillisecond time scales variability ( heao-1 , see e.g. @xcite ; exosat , e.g. @xcite ; ginga , e.g. @xcite ) . at the present time the rossi x - ray timing explorer observatory ( rxte , @xcite ) is the most powerful experiment for the investigation of the short time scale x - ray flux variations .
it combines high telemetry rate ( up to 512 kbps ) , large effective area of the detectors ( @xmath36400 @xmath12 in the soft x - ray band ) , high time resolution ( down to 1@xmath13sec ) and comparatively low deadtime distortions .
the comparison of the characteristics of several x - ray experiments can be found in e.g. @xcite .
the shape of the power density spectrum ( pds ) of cyg x-1 in its low / hard spectral state is well established ( in the range from @xmath14 to @xmath350100 hz ) since exosat observations it can be roughly represented by a constant from @xmath15 hz up to some break frequency ( @xmath16 ) , then it steepens to the power law with index @xmath31 ( @xmath17 ) and then it breaks again to the power law with the index @xmath31.61.8 , @xmath18 ( see e.g. rxte results in @xcite )
. short observations ( @xmath310 ksec ) of cyg x-1 by rxte / pca with high time resolution ( @xmath19s ) showed the absence of significant variability of cyg x-1 at time scales smaller than @xmath33 msec ( @xcite ) .
in this paper we report the systematic study of very high frequency variability of cyg x-1 in x - rays using large number of rxte observations ( total exposure time close to 190 ksec for the hard state and @xmath310 ksec for the soft state ) and carefully treating the deadtime effects .
for our analysis we used the data from the proportional counter array ( pca ) aboard rxte . to study the source variability in the hard spectral state , we chose a compact group of observations of cyg x-1 performed on 1517 dec .
1996 ( proposal p10236 ) with total exposure time of @xmath2080 ksec and 56 available observations of cyg x-1 from the proposal p30157 performed between dec .
1997 and dec .
1998 with total exposure time of @xmath21110 ksec .
the soft spectral state data were accumulated on 418 june 1996 data ( proposal p10512 , see also @xcite ) with total exposure time of @xmath2110 ksec .
all observations were performed with 5 pcus switched on .
the power density spectra were calculated from the light curves with the bin duration equal to the intrinsic time resolution of the data ( @xmath22s@xmath23s in proposals p10236 and p10512 and @xmath24s@xmath25s in proposal p30157 ) .
some deviations from the noise level predicted by the pca dead time models are observed in the power spectra constructed in the pca energy sub - bands with the amplitude of @xmath26 ( rms / mean)@xmath27/hz .
we therefore followed recommendations of the rxte gof and analyzed the light curves in the total pca energy band , where these deviations seem to be considerably less significant ( william zhang , private communication ) .
the power spectra calculated for individual time series were normalized to units of squared fractional rms ( @xcite ) @xmath28 and averaged . in the above formula
@xmath29 is the fourier amplitude at the frequency @xmath30 ; @xmath31 is total observed number of counts in the time series ; @xmath32 averaged observed count rate for the time series .
note that the ideal poissonian noise component is already subtracted , therefore @xmath33 for a poissonian distribution of counts in the time series . in a real detector
various effects could lead to modification of the poissonian noise level by an amount @xmath34 .
two major effects contribute to @xmath34 in the case of pca detector : @xmath35 where @xmath36 is a modification of the noise level due to counts dead time , @xmath37 additional noise component resulting from vetoing of the pca detectors by the very large events ( vle ) . to the first approximation the very large events lead to appearance of an additional ( positive ) component ( @xcite ) @xmath46 ^ 2\end{aligned}\ ] ]
where @xmath47 the duration of the vle window , @xmath48 - the vle count rate in one pcu .
note that the above expression does not include the binning and sampling effects and is valid in the limit @xmath49 .
finally , the total noise component model in the first approximation would be : @xmath50\ ] ] @xmath51 where @xmath52 - the _ total observed _ count rate in one pcu , @xmath53 number of pcus .
the @xmath54 and @xmath55 are given by the eq .
( 1 ) and ( 2 ) respectively .
the factor @xmath56 accounts for the fact that the count streams from @xmath53 independent units were merged together .
the factor @xmath57 accounts for reduction of the observed count rate due to very large events .
this formula is valid under the following conditions : @xmath58 , @xmath59 and @xmath60 .
it is important to stress out that in the adopted dead time model @xmath52 is the the _ total observed _ count rate of the events causing the dead time @xmath41 ( as opposite to the observed count rate in the analyzed time series ) . in the case of pca detector
it should include total good xenon rate , propane and remaining counts .
we also note that for sufficiently small count rates , @xmath6135 kcnts / s / pcu dependence of @xmath54 and , therefore , of @xmath34 upon @xmath40 vanishes . the model has been tested on a series of monte - carlo simulations and has proven to be sufficiently accurate in the parameters range of interest .
note that we did not include in our model an additional background term ( @xcite or @xcite ) . for bright sources ( @xmath621 kcnts / s / pca ) its contribution can be neglected .
= 0.18 cm cccccccccccc obs.id.&@xmath42 , @xmath13s&&&&&@xmath63/dof + & & & & & + & & exp.&fit&tot@xmath64.&gx@xmath65&fit&exp.@xmath66&fit&exp.@xmath67&fit & + + # 1&15.25879&810@xmath68&@xmath69&29.2&20.8&@xmath70&150&@xmath71&265&@xmath72&152/151 + # 2&15.25879&810@xmath68&@xmath73&19.6&15.1&@xmath74&60&@xmath75&239&@xmath76&180/157 + # 3&122.0703&810@xmath68&8.7(fixed ) & 23.6&18.3&@xmath77(fixed)&&76.0(fixed)&157&@xmath78&43/46 + + @xmath68
e.g. @xcite @xmath64 measured total count rate ( good xenon , propane counts and remaining counts ) per pcu @xmath65 measured good xenon count rate per pcu @xmath66 preflight values for all pca , see @xcite @xmath67 measured count rate ( housekeeping data ) = 8.5 cm = 8.5 cm = 8.5 cm to verify our model for the noise component we compared it with the data of pca observations of an extremely bright source sco x-1 ( @xmath3100 000 cnts / s / pca ) for which all flavors of the dead time distortions are much more prominent than for cygnus x1 .
we used 3 different observations of sco x-1 10059 - 01 - 01 - 00 ( feb .
14 , 1996 ; hereafter # 1 ) , 10059 - 01 - 03 - 00 ( feb .
19 , 1996 ; # 2 ) and 10056 - 01 - 02 - 02 ( may 26 , 1996 ; # 3 ) , having different value of the vle window @xmath47 and different time resolution . ideally ,
once the values of the dead time @xmath79 and vle window @xmath47 are calibrated , the model should reproduce the noise level with count rates @xmath52 and @xmath80 set to the measured values . however , as we discuss below , this is not strictly the case .
we therefore followed the approach adopted by @xcite and fitted the power spectra at high frequencies , @xmath81 hz , with the model for the noise component plus the model for the source component leaving the noise model parameters free .
the source contribution in this particular case was modeled as a superposition of two lorentzians representing khz qpos .
this approach has a disadvantage that it requires a priory assumptions about the shape of the power spectrum of the source .
in particular , if the source has very weak and very flat power spectrum component it might be not detected by our procedure .
= 8.5 cm = 8.5 cm = 8.5 cm in fig .
[ scox1_ds2 ] , [ scox1_ds1 ] and [ scox1_qpos ] we present the power spectra of sco x-1 in observations # 1 # 2 and # 3 along with the best fit noise models and residuals .
the best fit parameters and their expected values are given in table [ dt_params ] .
for observation # 3 we fixed parameters @xmath82 and @xmath47 because the limited frequency range of the power spectrum ( @xmath834096 hz ) does not allow us to constrain their values from the fit . from fig .
[ scox1_ds2 ] , [ scox1_ds1 ] and [ scox1_qpos ] one can see that with appropriate tuning of the parameters the adopted noise model is capable to reproduce the observed shape of the noise component
. in general the best fit parameters of the noise level model ( table [ dt_params ] ) look reasonable and are close to the expected values .
the best fit value of @xmath47 , in observation # 2 is by @xmath84 higher than the preflight value .
it is possible , however , that the duration of the vle window differs from the preflight values ( alan smale , private communication ) .
the most apparent difference is a large , by a factor of @xmath9 , discrepancy between the best fit and observed total count rate .
but we should note that the observed goodxenon count rate is much closer to the best fit @xmath52 value ( see table [ dt_params ] ) .
the reason of this discrepancy is not clear , it might be due to unaccounted processes in the pca detectors at high count rates .
we note however , that dependence of the noise component ( in units of squared fractional rms ) on the total count rate is a specific feature of a non - paralyzable dead time process at sufficiently high count rates . at lower count rate ( @xmath8530005000 cnts
/ s ) and in particular in the case of cyg x-1 ( @xmath861000 - 1500 cnts / s ) this dependence vanishes .
we adopted the noise level model described in the previous section for the subsequent analysis of the cyg x-1 data .
since the cyg x-1 observations were performed with time resolution coarser than @xmath87sec , the photon dead time and duration of the vle window are not constrained by the data .
we fixed them at the best fit values determined from the sco x-1 data : @xmath88s and @xmath89s .
as mentioned above , at the count rates typical for cyg x-1 , @xmath8610001500 cnts / s , the noise level does not depend on the count rate .
we therefore fixed the @xmath52 parameter at the averaged observed value .
the only free parameter of the noise level model was @xmath80 .
similarly to the procedure applied to the sco x-1 data , we included the source component into the model .
the source component was represented by a power law @xmath90 .
we used the high frequency part of the power spectra , from 100 hz up to the nyquist frequency , for the fits .
the high frequency part of the observed power spectra of cyg x-1 before subtraction of the noise level @xmath91 , the best fit model and the residuals of the data from the model are shown in fig .
[ p10236_noise ] , [ p30157_noise ] and [ soft_noise ] .
the overall power spectra in a broad frequency range from 1 mhz to @xmath92 khz after subtraction of the best fit noise level model are shown in fig.[broadband ] .
one can see that the model describes the data points reasonably well : @xmath9338/55 dof for observations p10236 ( fig .
[ p10236_noise ] ) , @xmath9356/49 dof for observations
p30157 ( fig . [ p30157_noise ] ) and @xmath9346/49 for observations p10512 ( soft state , fig.[soft_noise ] ) .
the best fit values of the slope of the source power spectrum are : @xmath94 for p10236 , @xmath95 for p30157 and @xmath96 for p10512 .
the intrinsic variability of the source has been statistically significantly detected up to frequencies of @xmath3150300 hz ( fig .
[ p10236_noise][soft_noise ] ) with the fractional rms amplitude in frequency band @xmath3100 - 400 hz of @xmath97% ( hard spectral state ) and @xmath212% ( soft state ) . in order to increase statistics we averaged all the hard state data increasing the total exposure time up to @xmath98 ksec .
the power law approximation of the averaged power spectrum above 100 hz gives the power law index of @xmath99 .
the power law approximation of the same data in the 2060 hz frequency range results in @xmath100 , although the pds clearly has a more complicated shape than a power law ( fig.[outcome ] ) .
we therefore conclude that we detected statistically significant steepening of the power spectrum of cyg x-1 in the hard state at the frequency of @xmath34080 hz .
note , that some indication of such steepening can be found in @xcite , although short duration of the used observations ( @xmath101 ksec ) was not sufficient to study the high frequencies in detail . in order to illustrate the high frequency behaviour of the power spectra , we plot in fig.[outcome ] the power spectra multiplied by @xmath102 . as it can be seen from fig.[outcome ] the particular shape of the rollover above 100 hz
is not significantly constrained by the data .
assuming that power spectrum continues with the same slope to the higher frequencies an observatory of the extra / laste class ( effective area @xmath103 m@xmath27 , @xcite ) would be able to detect statistically significant signal up to several khz in @xmath3tens of ksec exposure time and help us to say something more about the exact shape of the rollover of the pds .
the 2@xmath104 upper limit on the amplitude of a possible qpo component ( lorentz profile with quality @xmath105 ) in the 500 - 2000 hz frequency range in the hard state power spectrum is @xmath32% for @xmath106 and @xmath30.9% for @xmath107 .
these upper limits were obtained using the observations from the propsal p10236 performed on 15 - 17 dec .
1996 in which the shape of the power density at the lower frequency did not vary significantly .
the upper limit on the continuum noise component at the high frequency is somewhat more difficult to estimate .
a statistical ( @xmath108 ) error on the fractional rms in the 500 - 2000 hz for the sum of all hard state observations is @xmath109 . assuming that the noise level model is exact the upper limit on the fractional rms would coincide with the above number .
however , as we mentioned in the previous section a weak and very flat source power spectrum component might be not detected given the procedure used to determine the instrumental noise level .
we analized @xmath3190 ksec of observations of cyg x-1 in the low / hard state and reanalyzed @xmath310 ksec of high / soft state observations .
we detected statistically significant variability up to frequencies of @xmath3300400 hz with the fractional rms amplitude in the 100400 hz frequency range @xmath213% in the hard state and @xmath212% in the soft state .
the @xmath108 upper limits on the fractional rms amplitude of a qpo feature with lorentzian profile in the 5002000 hz frequency range are @xmath32% for @xmath110 and @xmath30.9% for @xmath111 in the hard state and 2.5% ( @xmath110 ) and 1.5% ( @xmath111 ) in the soft state . assuming that the model for the instrumental noise level is exact , the @xmath108 upper limits on the continuum variability in the 4002000 hz frequency range are @xmath31% and @xmath6% for the hard and soft state respectively .
we detected statistically significant steepening of the power spectrum in the hard state at frequencies @xmath112 hz . the power law approximation to the spectrum at frequencies higher than @xmath380100 hz gives the slope of @xmath1132.32.4 . in the soft state the slope of power spectrum changes from @xmath8 to @xmath9 at the frequency of @xmath10 hz without any evidence of further steepening up to the frequency of @xmath114 hz .
thus the sensitivity and time resolution of the rxte allow one to probe interesting range of the variability time scales for accreting black holes .
several characteristic frequencies fall approximately in this range .
first of all the frequency of rotation at the last marginally stable orbit for the non - rotating ( schwarzschild ) black hole is @xmath115 hz .
for the maximally rotating kerr black hole this frequency ranges from @xmath31200 hz to @xmath116 hz for the co - rotating and counter rotating orbits respectively .
if keplerian rotation itself makes significant contribution to the variability of the x ray flux ( e.g. due to the doppler shift and boosting of the emission from the `` hot spots '' in the disk ) this difference should be manifested in this frequency range ( sunyaev 1972 ) .
secondly typical sound crossing time of the disk height ( in the case of optically thick geometrically thin standard accretion disk ) corresponds to the frequency of hundreds of hz .
if strong turbulence is present in the accretion flow as anticipated by the most popular theoretical models ( see e.g. @xcite for review ) then it may affect the variability at higher frequencies .
the dependence of the amplitude of variations on the frequency is then a valuable tool to study the turbulence in the accretion flow .
thirdly light crossing time of the central region @xmath117 corresponds to the frequencies of the order of khz .
the light crossing time for the region of the main energy release ( @xmath118 1020 @xmath119 ) corresponds to the frequencies of 300500 hz . the character of the the variability may change drastically around this frequency or ( e.g. in the comptonization models involving multiple scatterings of the photons ) at a several times lower frequencies ( e.g. @xcite , @xcite , @xcite , @xcite ) . the rxte data ( i ) demonstrate that power specifically associated with any of these characteristic frequencies is not large ( see given above upper limits on the narrow features on the pds in the frequency range from 5002000 hz )
i.e. there are no `` resonances '' at these frequencies , ( ii ) limits the total power associated with high frequency variability ( from 400 hz to 2000 hz ) to a percent level , ( iii ) show that in the hard state of the source prominent , previously unknown , break in the pds occurs at the frequency of @xmath3 70 hz , ( iv ) extend the range of significantly detected variations up to the frequencies of @xmath3200300 hz .
the theoretical models of the accretion onto the black holes are now to be tested against these results .
the study of the high frequency noise might become an additional tool to distinguish black holes from weakly magnetized neutron stars ( sunyaev & revnivtsev 2000 ) .
in the subsequent publication ( revnivtsev et al .
2000 , in preparation ) we perform similar analysis for the accreting neutron stars . the authors thank rashid sunyaev for numerous discussions , william zhang for helpful discussions of pca deadtime , alan smale and keith jahoda for the useful comments on the pca operations .
this research has made use of data obtained through the high energy astrophysics science archive research center online service , provided by the nasa / goddard space flight center .
m.revnivtsev acknowedges partial support by rfbr grant 00 - 15 - 96649 . | we analyzed a large number of rxte / pca observations of cyg x-1 in the hard and soft spectral states with total exposure time of @xmath0 and @xmath1 ksec respectively and time resolution better than @xmath2s in order to investigate its variability down to a few milliseconds time scales .
the model of modifications of the power density spectra due to the dead time effect was tested using rxte observations of an extremely bright source
sco x1 .
the results of these tests demonstrated , that although some problems still remain , for sources like cyg x-1 ( which is an order of magnitude less bright than sco x-1 ) present knowledge of the instrument is sufficient to reliably analyze the power spectra in the khz frequency range . in both spectral states of cyg
x-1 we detected statistically significant variability up to @xmath3150300 hz with a fractional rms amplitude above 100 hz at the level of @xmath4 .
the power spectrum in the hard state shows a steepening at frequency of @xmath5 hz with the power law slope changing from @xmath6 to @xmath7 . in the soft state the slope of power spectrum changes from @xmath8 to @xmath9 at the frequency of @xmath10 hz without any evidence of further steepening up to @xmath11 hz .
these break frequencies represent the highest characteristic frequencies detected in the power density spectrum of cyg x-1 so far . |
the optic nom @xcite is a deployable , long - arm , non - contact profilometer consisting of a scanning pentaprism and a digital autocollimator ; this type of profilometer is also known as a deflectometer . a deflectometer can be used as a non - contact probe to provide accurate measurement of the height profile across an optical surface .
the pentaprism traverses along a stiff linear guide - bar above an optical surface , relaying the autocollimator beam to provide a set of surface slope ( angle ) measurements along that path .
pentaprism position information is provided via a linear encoder and so a height profile of the optical surface can be calculated .
a diagram of the set - up is given in fig .
[ fig : nom ] .
the instrument is being used to confirm the base radius of curvature ( @xmath2 ) and conic constant ( @xmath3 ) of prototype european extremely large telescope ( e - elt ) segments that are being manufactured at optic glyndr .
this profilometry technique originated in the synchrotron community @xcite to provide accurate measurements of x - ray focussing optics .
these mirrors for synchrotron x - rays generally fall under two categories : ( 1 ) long ( up to 1.5 m ) and narrow ( a few 10 s of mm ) with very long focal lengths ( i.e. very large radii of curvature with @xmath2 from 100 s of metres up to a few kilometres ) and ( 2 ) nano or micro - focussing x - ray mirrors that can have higher surface curvature ( i.e. @xmath4 m ) but are only @xmath5 mm long . in both these cases the change in slope over the surface of the mirrors is usually < 10 milliradians so the limited angular measurement range of the digital autocollimator is not an issue and angular `` stitching '' techniques are not required in the data analysis . for our purpose
we do not require the extreme accuracy and precision that is needed for x - ray optics , but we need to measure large optical surfaces ( e.g. > 1 m diameter areas ) with slope variations that exceed the range of most commercially available , sub - arcsecond accuracy , digital autocollimators .
also , in contrast to the x - ray optic community , the optic nom measurements are subjected to a significantly higher level of environmental noise due to the fact that the instrument is portable and used in - situ over a robotic polishing machine .
this paper describes a numerical method that can be used to process sets of deflectometer line data that have been taken over an optical surface , where the individual line scans in that data set have been taken in difference coordinate frames , and fit an aspheric ( conic - section ) surface to them .
simulated datasets have been used to test the fitting method in order to determine the accuracy of the method and any limiting behaviour .
the software has been implemented using matlab @xcite .
to @xmath6 and @xmath7 to @xmath8 ) to enable measurements over the entire surface ( e.g. dotted path ) . ]
for a primary mirror ( m1 ) as large as that proposed for the e - elt , it is not feasible to manufacture the mirror in one continuous piece ; it must be constructed from many smaller mirrors that are of a size more practical to manufacture , measure and transport .
the primary aperture of the e - elt at the time of this study was a 42 m diameter ellipsoid .
the surface form of the primary mirror is given by the formula : @xmath9 ( ref : @xcite , p. 16
, the ellipsoid formula ) . where @xmath2 is the base radius of curvature , @xmath3 is the conic constant and @xmath10 , @xmath11 , @xmath12 are the coordinates in the m1 ( primary mirror ) reference frame .
the following production and measurement accuracies were specified by eso @xcite : * the first polished segment should have a base radius of curvature and conic constant which satisfy @xmath13 mm @xmath14 mm and @xmath15 . *
the accuracy of the knowledge of the nominal radius of curvature , @xmath16 shall be better or equal to 14 mm rms . * the maximum allowable surface error ( before removal of low and mid - spatial frequency terms ) over the useful area of any prototype segment is 50 nm rms ( 40 nm goal ) .
the 42 m primary consists of 1148 hexagonal mirror segments of approximately 1.4 m diameter .
the set of prototype segments to be manufactured and measured were a group of 7 segments towards the outer edge of the primary mirror assembly , i.e. each mirror segment is an off - axis ellipsoid .
one of the challenges in manufacturing the primary mirror is that each individual segment must appear to have been `` cut - out '' from this larger ellipsoid - each segment must be made to the same @xmath2 and @xmath3 ( within tolerance ) and this surface form must be correctly aligned to the geometry of the segment ( which are slightly irregular hexagons ) .
interferometric testing provides accurate surface - form maps in relation to the geometry of the segment , however it can not provide a direct measurement of @xmath2 .
a more direct measure ( i.e. one not requiring a reference optic ) can be provided using a deflectometer .
therefore , the primary task of the optic nom is to provide a confirmation of the base radius of curvature ( @xmath2 ) and conic constant ( @xmath3 ) of prototype e - elt mirror segments during the end stages of optical polishing .
in the following discussion , reference is made to the m1 ( primary mirror ) coordinate reference frame and the segment ( denoted by the segment number : e.g. s1 , s4 etc . )
coordinate reference frame .
the position and surface form of the segments is defined in m1 coordinates as described by equation [ eq : ellipform ] .
when measuring a segment by interferometry , the segment is supported in its gravitationally symmetric position : the mirror segment is oriented so that the normal to the surface at its centre coordinate is pointing vertically upwards ( along the z - axis ) . in comparison ,
in the m1 coordinate reference frame , the normal vector of a general segment is not pointing along z ; only a segment centred at the centre of m1 would possess this property .
when measuring with the deflectometer the mirror segment is , on average , oriented with its centre - normal pointed vertically upwards . however , as previously described , the range of surface slope on a segment exceeds the measurement range of the autocollimator and so the segment must be pitched and rolled ( see fig .
[ fig : nom ] ) to allow scans to be taken over the entire surface .
the range of these angular re - positionings is of the order of @xmath17 degrees . with
the segment tilted in various orientations for different line - scans the measurement coordinate frame is no longer co - aligned with the segment coordinate frame .
this makes co - locating all the separate line scans to a single surface - which needs to be a best fit to this data - non - trivial .
figure [ fig : scanlines](a ) gives an example of a set of scan - lines taken over a segment surface , fig.[fig : scanlines](b ) shows the effect that the tilting method used to overcome the limited measurement range has on the data and fig.[fig : scanlines](c ) shows how the line scans appear when each scan is properly corrected for z - height offsets and rotations after the optimisation / fitting procedure .
a suite of matlab programs have been written to read - in , process and find the best - fit conic surface for the nom line - scan datasets .
the data flowchart for these processes is shown in fig .
[ fig : dataflow ] .
as previously described , the challenge with fitting this data to a surface is that due to the limited angular range of the autocollimator , each scan - line taken on the mirror surface is in an unknown different coordinate reference frame and so the fitting routine must translate and rotate the scans to correct for this .
also , given that the @xmath12 ( height ) data of the surface is obtained via integration of slope data ( of the form d@xmath12/d@xmath10 = tan@xmath18 ) , the absolute @xmath12 position of each scan is unknown ( the unknown integration constant ) , so the relative @xmath12 height of all the scans with reference to each other must also be part of the optimisation routine . .
the final step ( 10 ) is beyond the scope of this work . ] for a given set of scan - lines the parameters that require fitting are : @xmath2 and @xmath3 ( globally ) and also 6 parameters per scan ( 3 translations and 3 rotations ) .
this means there are @xmath19 parameters to fit per set of line - scans ( where @xmath1 is the number of line scans in a set ) . for the line - scan pattern shown in fig .
[ fig : scanlines ] , this means there are @xmath20 parameters to optimise in the conic fitting . to perform this optimisation
a trust - region - reflective algorithm is used to determine the minimum of a cost - function . as expected ,
a gradient minimisation over so many dimensions is computationally costly and prone to stopping in local minima . to help the optimiser , `` good '' initial values of @xmath2 and @xmath3
must be provided and boundary limits applied to the individual scan - line transformations . in addition a global search strategy is used to circumvent the local - minima problem starting the optimiser with many different sets of initial conditions and selecting the most optimal
. this fitting routine can be run on a standard desktop pc ( for this benchmark the processor is an intel^^ xeon^^ cpu e3 - 1240 v3 3.40 ghz ) . as an example of the run - time , a single run of the optimisation / fitting routine ( i.e. using only a single @xmath21 pair as an initial starting value ) on a typical dataset simulated here ( 24 scan - lines containing 2562 data points in total ) which requires 146 parameters to optimise takes @xmath22 seconds to converge .
an important step in setting up this optimisation routine is the formulation of an appropriate cost function , @xmath23 .
the quantity that we minimise is the sum of the squared differences between the height of each point in each line scan ( @xmath24 ) and the calculated @xmath12 of an ellipsoid surface at the same ( @xmath25 , @xmath26 ) coordinate positions for a given @xmath2 and @xmath3 ( called @xmath27 here ) . to enable
the comparison between @xmath24 and @xmath27 , either the line scans must be transformed into the m1 coordinate frame or the ellipsoid formula ( equation [ eq : ellipform ] ) transformed into the relevant segment coordinate system .
the more reliable of these for the cost function is the latter - a comparison in the segment coordinate frame .
one of the reasons for this is that in the segment reference frame the cost function ( @xmath24 @xmath27 ) is less weighted by the outer scan lines ( furthest from the segment centre ) for the same angular displacement , another is that the cost function is smoother since most of the different parameter values are closer to zero than in the m1 frame .
figure [ fig : refframes ] presents an illustration of this issue .
the transformation of the ellipsoid formula from m1 to segment reference frame uses the pre - defined centre coordinate of the segment under manufacture and the measured centre coordinate determined experimentally from the segment geometry .
the @xmath12 value of the segment centre , in segment coordinates is set to be @xmath28 .
the form of the conic equation used to calculate @xmath27 is : @xmath29 where @xmath2 and @xmath3 are the ellipsoid parameters , @xmath10 and @xmath11 are the coordinates in the segment frame and @xmath30 is the rotation angle around the x - axis that brings the normal vector of the segment at its centre in the m1 frame to a vertical orientation ( i.e. along the positive z - axis ) ; this is also illustrated in fig .
[ fig : refframes ] . the equation can be formulated in this way because the ellipsoid formula ( equation [ eq : ellipform ] ) is symmetric about the z - axis , and for the purposes of comparing the theoretical surface with the real data the data can be transformed so that the segment centre lies on the y - axis .
the optimization routine being used implicitly computes the sum of squares as part of the algorithm , so our cost function , @xmath23 , is multi - valued and creates a length @xmath31 vector of z differences , one for each measured ( @xmath32,@xmath33,@xmath34 ) value : @xmath35 \label{eq : f } \end{split}\ ] ] where @xmath1 is the number of line scans and @xmath36 and where @xmath25 , @xmath26 , @xmath24 are the rotated and translated line scan measurements given by @xmath37 here @xmath38 and @xmath39 where @xmath40 , @xmath41 , and @xmath42 are the measured coordinate position and height data for point @xmath43 and @xmath44 is the line scan number which contains the point @xmath43 .
@xmath45 , @xmath46 and @xmath47 are the standard rotation matrices around the @xmath10 , @xmath11 and @xmath12 axes respectively for the angle parameters @xmath48 , @xmath49 , @xmath50 for line scan number @xmath44 .
a trust - region - reflective algorithm is used in the optimiser and it is set up as a bounded problem - initial values are allocated to the parameters that require fitting and limits are also defined for those parameters .
the optimisation is based around a gradient minimisation ; the cost function gradient is calculated using the central finite differencing method . for this problem
the upper and lower boundary limits were chosen to be symmetric around the initial value , @xmath51 , of each parameter @xmath52 : @xmath53 for the 6 degrees of freedom in positioning each scan line into a common ( the segment ) coordinate frame the @xmath54 , @xmath55 , @xmath56 limits are set at @xmath57 mm and the 3 rotations are limited to @xmath58 radians ; the initial values are set to zero .
these limits have been chosen to be suitable for the measurement set - up as previously described and the expected displacements and rotations of the line scans are well within these boundary limits . for @xmath2 and @xmath3 , the global parameters across all scan lines , the initial values , @xmath59 and @xmath60 , are set to a good estimate based on other measurements ( e.g. interferometric measurements ) and the boundary limits in the program are set to @xmath61 mm and @xmath62 . section [ sec : performance ] demonstrates the method used to reach the correct solution even if good initial estimates for @xmath2 and @xmath3 are not available .
( i.e. d@xmath12/d@xmath2 ) over @xmath10 and @xmath11 .
both plots show the same jacobian but calculated using different forms of the quadratic equation solution - the left plot is calculated before the function calculations were optimised for numerical precision and the right plot is after the reformulations to improve numerical noise .
these have been calculated for s1 ( near the outer edge of m1 ) , which is the worst case ( in terms of the numerical noise ) due to the magnitude of the y - coordinates . towards the centre of m1
, the centre coordinates describing the segment positions are more moderate in magnitude resulting in a smoother jacobian . ] in order to allow the optimisation algorithm to operate efficiently , it is important to present it with a cost function that is as smooth as possible .
any noise in the cost function causes fluctuating gradient measurements which at best will cause the optimisation to take a longer path to the solution , and at worst may cause it to terminate prematurely in a false local minimum .
an example of problems with numerical noise is shown in fig .
[ fig : numnoise](left plot ) : the initial formulation of the problem results in such significant loss of numerical precision that there is no discernible gradient for the optimiser to follow and it terminates in an apparent local minimum very rapidly ( it hardly moves from the initial starting values ) .
figure [ fig : numnoise](right plot ) shows the same gradient function after applying several improvements as described below .
numerical noise in the cost function is still present , but it is now significantly below the level of the actual function evaluation values , and low enough that the derived gradients can be followed by the optimiser .
a standard formulation of equations in the cost function can result in a significant loss of numerical precision due to cancellation between terms of the ellipsoid equation . by reformulating the calculation of the surface using a different form of the quadratic equation ( @xcite , ch.5 , pp.227 )
the numerical accuracy of the calculation is vastly improved .
the form of the cost function ( @xmath63 ) used is calculated from equation [ eq : zref ] . in addition , the order of arithmetic operations throughout the cost function was carefully analysed in the context of the expected magnitudes of the inputs and improved where possible to maximise the precision of the cost function . the choice of how the gradient is calculated has a significant effect on the solution convergence - the central ( or symmetric ) finite differencing method has been chosen which is computationally more costly than the forward finite differencing method but the central method has higher numerical precision ( @xcite , ch.5.7 , pp.229 ) .
use of the forward method results in a solution of much lower accuracy and is insufficient for the solution accuracy required here .
the default step size for jacobian calculations in matlab is quite small ( 1e-8 ) , although it can be set to any value from 0 to inf .
increasing the step size will sample over a size greater than the granularity of the numerical noise features and so a smoother function is produced which is then less likely to present the optimiser with false ( noise based ) local minima .
however , the step size also needs to be minimised to provide an accurate estimation of the gradient function and ensure that the optimiser will converge on a solution of the required accuracy . with the initial cost function formulation , a step size of 1e-3 ( @xmath641e9*eps(@xmath65 ) ) , where @xmath66 , was needed to produce the required smoothness , but this results in very poor gradient estimates .
machine epsilon ( eps ) is the approximate relative error due to rounding in floating point arithmetic of a 64 bit double representing that value . with the improved cost function , a step size of 1e-7 ( @xmath641e4*eps(@xmath65 ) )
could be chosen which provides a good balance between noise reduction and accuracy of estimation .
the optimisation algorithm allows only a single value for the gradient estimation step size , @xmath67 , to be chosen to apply to all the parameters .
the parameters in our problem differ by several orders of magnitude , so for the optimisation process the inputs are normalised such that their bounds map to [ -1 1 ] .
a step size of 1e-7 is then suitable for the entire n - dimensional gradient space minimisation . to guard against solutions that may have converged in a local minimum rather than the global minimum a global search strategy
is used , starting the optimisation routine using many sets of different initial conditions and comparing the results .
an efficient way to implement this strategy is to generate a randomised sobol sequence @xcite over the required n - dimensional search space . by using @xmath68 ( where @xmath69 is the number of sets of initial parameters ) points to generate the sobol set
the points will be uniformly distributed over the required search space . for this work
we have generally been using this to generate a set of @xmath21 initial values to verify the solution convergence . , @xmath3 parameter space that occupies the same rms surface error contours map solutions having the same rms surface error .
the inner 10 contour levels are ( in nm ) : 40 , 50 , 100 , 150 , 200 , 300 , 400 , 500 , 700 and 900 . given @xmath66 mm and @xmath70 for a `` perfect '' first segment , the eso allowable surface error of 50 nm rms for subsequent segments corresponds to the plotted elliptical boundary limits in ( @xmath2 ( mm ) , @xmath3 ) of ( 83990.7 , -0.995295 ) and ( 84009.3 , -0.991291 ) . ]
for the real data , sources of error such as environment noise ( vibration , dust , air turbulence ) , alignment errors , encoder errors etc . lead to measurement errors in the autocollimator slope angle . in principle
these slope errors could be estimated using classical error propagation , but this alone does not give us the error on the parameters we are trying to determine : @xmath2 ( base radius of curvature ) and @xmath3 ( conic constant ) .
the slope angles are converted into gradients and integrated to @xmath12 values . the cumulative integration error could be estimated and would vary depending on the numerical integration method .
as previously described , the ellipsoid formula ( equation [ eq : ellipform ] ) is fitted to the set of line scans . due to the limited range of the autocollimator , line scans covering the full extent of the segment
can only be obtained by tip , tilt and rotation of the segment on its support structure . due to this ,
the parameters in the fitting routine include @xmath2 and @xmath3 , but also separate spatial translation and rotation parameters for each line scan , to allow for the fact that we only approximately know their relative locations . an analytical propagation of errors to the parameters @xmath2 and @xmath3 is not possible in this situation ( a non - linear iterative fitting process ) ; the error in @xmath2 and @xmath3 from this fitting process is better estimated using monte carlo methods and will depend on the optimisation algorithm and the exact cost function formulation as well as the errors in @xmath10 and @xmath12
. in theory , errors in our estimates for the translations / rotations of the line scans can be corrected by the optimiser but in practice they will also affect the ability of the algorithm to fully converge . to test the behaviour and accuracy of the fitting routine simulated nom scans are generated .
as shown in fig .
[ fig : scanlines](a ) , a set of scan coordinates defined in @xmath10 and @xmath11 in the segment reference frame are passed to the nom scan simulator along with the surface specification in @xmath2 and @xmath3 , the centre coordinate in the m1 reference frame and the 1-sigma level of z - noise to add to the generated @xmath12 values . based on a measurement procedure of ensuring each scan is centred within the surface slope range of the autocollimator
, the data for each scan line is then rotated in the @xmath10 and @xmath11 axes around the centre point of the scan line so that the @xmath10 , @xmath11 slopes at this centre point are zero .
note that the specification of the @xmath10 , @xmath11 coordinates of the scan lines are chosen to ensure that the slope measurement along its length does not exceed the total measurement range of the autocollimator .
realistic levels of random noise on @xmath12 are chosen based on real data acquired with the optic nom .
the simulated dataset then looks like fig .
[ fig : scanlines](b ) and these data are passed to the fitting routine . in terms of the whole process shown in fig .
[ fig : dataflow ] , the simulated dataset comes in at step 5 and represents the data as if it has gone through the previous steps and the data taking process ( not indicated ) .
for all the performance tests presented here the simulated datasets were generated in the format as shown in fig .
[ fig : scanlines](a ) ( i.e. a scan pattern consisting of a grid of 7 lines across each of the x and y axis to produce 24 scan lines that are within the measurement range of the nom autocollimator ) .
the data points within a scan line are separated by 5 mm .
since the data used in these tests are simulated , the @xmath2 and @xmath3 is known and the level of random noise injected into the data is known .
sanity checks on the reconstructed / fitted data are made by checking that the final value of the cost function ( which is basically a sum of the residuals ) corresponds to the level of noise that was injected ( i.e. an rms of the residuals should be close to the 1-sigma level of noise that was used ) . if the residuals are higher than expected it is usually because the optimiser has terminated too early .
this can happen if the initial parameters were beyond the bounds of the optimiser ( see section [ sec : outbounds ] ) , or the initial values were located unfavourably in the gradient - space so that the route taken by the optimiser needed more steps than the allocated maximum number of iterations ; it is for these reasons that a multi - start strategy is best used for unknown data . the reconstructed data ( as per fig . [
fig : scanlines](c ) ) as well as the residual plot ( fig .
[ fig : scanlines](d ) ) can also be plotted to double - check that the scan lines have been sensibly located on the best - fit surface and that there is nothing unexpected in the residuals ( i.e. that they are consistent with the expected noise ) .
finally , in order to determine whether a solution has converged within the limits of the allowed rms surface error a contour plot of rms surface error was derived over the @xmath2 , @xmath3 parameter space .
the eso allowable surface error is defined to be 50 nm rms @xcite . for segment s1
, whose centre coordinate is positioned near the outer edge of the primary mirror this contour plot is given in fig .
[ fig : contours ] . the rms surface error is calculated based on the 2562 data points in the line scan pattern ( fig .
[ fig : scanlines](a ) ) of the typical simulated scan set used , with the minimum surface error ( injected @xmath12 noise only ) set at @xmath66 mm and @xmath70 . the rms surface error versus @xmath21 varies with position on m1 , so a different contour map must be calculated for each segment set to enable an accurate evaluation of the @xmath2 , @xmath3 solution convergence .
an example of this for a segment closer to the inner edge ( named s12 here ) is given in section [ sec : innerseg ] .
comparing the @xmath2 , @xmath3 limits on the 50 nm rms contour for s12 gives : rlimits = 83991.1 84008.9 mm ( diff = 17.8 mm ) and klimits = -1.009890 - -0.976677 ( diff = 0.033213 ) .
the kdiff for s1 is 0.00392392 ( an order of magnitude smaller ) and the rdiff is 18.6 mm .
when using this analysis method on a real data - set there are certain unknowns that the analysis must take into account and be robust to . it is assumed that approximate values are known for the key parameters that define the shape of the surface ( e.g. base radius of curvature and conic constant ) , so that these can be used as reasonable initial values in the optimisation process .
however , even if there is no prior knowledge of these parameters a wider search strategy can be employed using more steps in the process ; the time taken to reach a satisfactory solution is increased accordingly . a number of different data sets were created to test the fitting method and these are described in the following sub - sections
. the boundary box in the optimiser is set at @xmath61 mm and @xmath62 ( where @xmath59 and @xmath60 are the initial values provided to start the optimiser ) ; so if the initial values fall beyond the real solution @xmath71 the @xmath2 and @xmath3 boundary limits then the routine will fail to find the correct solution because it is beyond its search - space .
it is unlikely that the user would ever have this level of uncertainty when providing a good initial value , however an example is given ( in section [ sec : outbounds ] ) of what happens in this case . and @xmath3 ( indicated by the green star on the inset top plot ) and at the same outer segment position ( s1 ) .
each dataset is analysed using the same set of 32 @xmath72 points as initial values to demonstrate the multi - start strategy .
the inset plot at the top is a zoom - in on the central region of the main plot ; the median result of each group is indicated with a `` + '' marker .
the main plot is shown to demonstrate that all the results converged within the inner contour ( 40 nm ) , and within the eso allowable error . ] to investigate how the fitting routine behaves in the presence of different levels of noise , datasets were generated from a surface of a fixed base radius of curvature ( @xmath2 ) and conic constant ( @xmath3 ) ( as per equation [ eq : ellipform ] ) with only the 1-sigma level of random noise on the @xmath12 height data varying between the datasets . for these experiments : @xmath73 mm and @xmath74 and the 1-sigma level of random noise on the @xmath12 data = 1.2 nm , 3.0 nm , 6.0 nm and 12 nm .
these noise levels are based around the maximum measurement error of the optic nom autocollimator which is @xmath17 arcsec and random position error of @xmath75 m for a single point measurement . for a 5 mm step size between measurements
, this angle measurement error converts to a z error of 6.0 nm ( i.e. @xmath76 ) . in a real - life measurement
25 autocollimator measurements are acquired in rapid succession for each data point reducing this error to 1.2 nm .
the position uncertainty converts to a @xmath12 error of approximately 4 nm ( varies depending on segment and position on segment ) .
setting the 1-sigma level of z - noise in the simulations to 6.0 nm is close to an expected maximum error level and using 3.0 nm represents a more modest level of noise .
the lower and upper limits of noise ( 1.2 and 12 nm ) have been chosen to represent an environment / set - up of excellent stability and accuracy and a more noisy / less stable environment than we assume are typical of the optic nom .
the data generated is for s1 which is positioned near the outer edge of m1 and so is the most off - axis ellipsoid . using the scan pattern as previously described
, 5 datasets were generated using each of the 4 specified levels of random noise .
a randomised @xmath21 sobol set containing 32 points was generated over an r - range of 83000 to 85000 mm and a k - range of -1.1 to -0.8 .
these 32 points provide the initial @xmath2 , @xmath3 values for the multi - start strategy ; the same set of initial values was used to start the fitting / optimisation process for all the datasets .
every dataset is run through the fitting process 32 times , each time using a different initial value for @xmath2 and @xmath3 as provided by the sobol set .
the results of the final fitted @xmath2 , @xmath3 for each of these datasets is shown in fig .
[ fig : diffnoise ] , over - plotted on the contour map of @xmath2-@xmath3-rms surface error . the median result for each of the different noise groups
is indicated .
the best - fit @xmath2 , @xmath3 to the simulated nom data is taken as this median value .
the median , rather than the mean is used as it is a better representation of the majority of the fitting results since we know that some @xmath2 , @xmath3 initial values will have been poorly placed in the gradient space and that there will be a few spurious / outlier results .
using the same @xmath2 and @xmath3 ( 84000 mm and -0.993295 ) , a z - noise level of 3.0 nm and the same set of 32 @xmath21 initial values , the fitting process was repeated on 5 datasets generated at a different m1 centre coordinate in order to see if there are any differences in fitting data from the inner edge of m1 i.e. closer to the edge of the central cut - out and the least off - axis ellipsoid compared to the outer edge of m1 ( s1 ) .
the centre coordinate of s1 ( outer edge ) in the m1 frame is [ 0 , 18469.942 ] mm and of this ( inner edge ) segment is [ 0 , 6393.5 ] mm . as described in section 4 , a new contour map of @xmath2-@xmath3-rms surface error was generated to compare results at this new m1 mirror position .
the results are shown in fig .
[ fig : innerseg ] . and @xmath3 values ( lower cost function value indicates a better fit ) given an over - wide search space for the multi - start optimisation strategy .
the square outlined points show the initial values that were passed to the optimiser for this dataset ; the green shaded rectangular area indicates the boundary limit around the correct solution initial values outside of this boundary will not be able to converge to the correct solution .
the small grey points show the end - point @xmath2,@xmath3 values of the optimiser with a light - grey line linking it to its initial value .
for the 8 start values within the boundary ( filled green squares ) the optimiser reached the correct solution ( on the contour plot there are 8 almost overlapping points at ( 84000 , -0.993295 ) ) . ] an extended range 32 point sobol set of initial @xmath21 values was generated to demonstrate the results when the routine is provided with initial values that are outside the boundary limits around the real solution .
these boundary limits are set within the optimisation function to be : @xmath77 mm around @xmath59 and @xmath78 around @xmath60 .
the extended sobol set was generated using a rrange of 80000 to 90000 and a krange of -1.5 to 0 .
the nom dataset for testing was generated with the same fixed @xmath2 and @xmath3 as previous ( 84000 mm , -0.993295 ) , z - noise = 3.0 nm and at the s1 position .
the results of fitting this dataset for each of the initial values in this extended range set is shown in fig .
[ fig : extrange ] .
it demonstrates the use of the multi - start strategy not only to avoid local minima issues , but also as a tool to `` zoom - in '' on the correct answer even if there is very poor knowledge of the shape of the optic from which the data was taken .
firstly the results with the lowest value of the cost function will give a good indication of @xmath2 , @xmath3 values that are at or approaching the correct results .
also , as in the case shown in fig .
[ fig : extrange ] , there are 8 initial values that lie within the boundary box centred on the correct @xmath2 , @xmath3 value and so there is a `` point '' on fig .
[ fig : extrange ] that consists of 8 almost overlapping results that started from those points and have converged on the correct solution .
once an indication of the correct result is gleamed from an `` over - wide search '' , the optimisation can be repeated with a smaller range of @xmath2 and @xmath3 around the region where the correct solution is believed to lie .
a 128-point set of @xmath21 pairs was generated using a randomised sobol set and used to generate 128 different nom datasets .
the @xmath2 range for this sample was 60000 to 90000 mm and the @xmath3 range -1.5 to 0 .
as a side - note , this range includes the current specification for the e - elt of @xmath79 mm and @xmath80 @xcite .
an arbitrary off - axis position was chosen for the simulated segment data , this was fixed for all the 128 datasets at cm = [ 0 , 12000 ] mm , which is around the mid - section of a primary mirror of diameter 40000 mm .
the 1-sigma level for random noise on the data - points was also fixed at 3.0 nm . given the large amount of processing time that would be required to perform a many point multi - start strategy as demonstrated in section [ sec : fixrk ] ,
it was assumed for these analyses that a reasonable knowledge of the @xmath21 value of the mirror is known and so the initial values given to the optimiser was set to : @xmath81 mm and @xmath82 ; where @xmath83 and @xmath84 are the values used to generate the simulated datasets .
the results of these optimisations are shown in fig . [
fig : varrk ] .
the average of the residual rms surface error is 2.95 nm which corresponds well to the expected noise in the data ( 3.0 nm ) .
the median value of the rdiff ( fitted r value ( @xmath85 ) - r value used to generate the data ( @xmath83 ) ) is 0.202 mm ; the median value of the kdiff is 6.65e-5 .
these are well within acceptable tolerances . on a `` real '' optic ,
the surface form is unlikely to be perfectly described by equation [ eq : ellipform ] with just random noise on the data points and line - scan positions .
it is likely that the shape will be slightly modified by higher order terms such as astigmatism . for the manufacturing of the eso e - elt segments
there is a certain amount of allowable misfigure in terms of zernikes @xcite , since the active mirror support structure can remove a certain amount of form error . for these tests we chose two of the higher order allowable errors to add to the simulated datasets to investigate how this affected the result on fitted @xmath2 and @xmath3 ;
the misfigure in zernike terms we added to this data are astigmatism ( z = 4 ) and higher order `` trefoil '' ( z = 9 ) . the misfigure terms were considered individually and were added in to the ellipsoid data either at the maximum allowed amount of error for that term or at half the allowed amount .
a 32-point multi - start strategy was used to analyse each dataset , the results are plotted in fig . [
fig : formerror ] . , @xmath3 position of the generated data .
only 2 ( out of 160 ) results from the maximum allowed z9 multi - start set fall outside the boundary . ]
as expected the presence of form error in the data modifies the best fit @xmath2 and @xmath3 .
the presence of either astigmatism or trefoil results in a fitted @xmath2 and @xmath3 that is shifted towards a higher @xmath2 and @xmath3 than the values that were used to generate the data .
again , as might be expected , astigmatism had a more significant effect on the fitted results .
this is partly to do with the fact that more error in astigmatism is allowed in the data but also because it can be seen how the two axes of different curvature on an astigmatic surface can add to the ellipsoid surface ( also with two axes of different curvature ) to produce a surface which closely resembles an ellipsoid surface with slightly modified @xmath2 and @xmath3 parameters .
figure [ fig : formerror ] shows that the rms surface error due to the difference between the fitted @xmath2 , @xmath3 and the actual @xmath2 , @xmath3 is within allowable limits . if the maximum allowed form errors are known to be present within the data ( i.e. it is known from the interferometric data ) it is advisable to subtract these form errors from the profilometer data and fit for @xmath2 and @xmath3 again to get a result with an improved accuracy .
this work has demonstrated that it is possible to converge on an @xmath2 , @xmath3 solution of sufficient accuracy for the described purpose by using a gradient minimisation optimisation over many ( 100 + ) dimensions to place scan - lines in a common coordinate frame of reference and simultaneously fit the best aspheric surface .
the simulations performed here have assumed a fixed scan pattern of 7 line sections across x and y with data points taken every 5 mm along each of these lines ( divided into 24 line scans due to the limited autocollimator measurement range ) .
this pattern was chosen since it represented one of the best compromises between data acquisition time and adequate sampling of the optic being measured .
the number of degrees of freedom in the optimisation is altered by the number of scan lines ( 6n + 2 where n is the number of scan lines ) .
the behaviour of the optimisation process with respect to number of scan lines has not been completely investigated .
the optimisation process is robust to a wide range of random noise on the z ( height ) data , a 1-sigma level of 1.2 to 12 nm has been investigated here and shown to satisfy requirements .
the process can cope with small ( @xmath86 m ) random offsets of entire line scan positions since x , y , z position is part of the optimisation for each line ( in order to account for line shifts when rotating the line scans into their common reference frame and the unknown integration constant ) . exact knowledge ( i.e. within the noise ) of the segment centre position in the data is assumed in order to have accurate mapping to the required segment centre coordinate in the m1 reference frame .
any large or systematic errors in the coordinate system or elsewhere in the data should be corrected or removed before fitting to obtain the most accurate result ; if this is not possible their influence on the fitting process will need to be investigated .
although the focus of these fitting tests were for the old e - elt specification ( @xmath73 mm , @xmath74 , 42000 mm diameter primary ) this fitting / optimisation process has also been shown to be effective for surfaces described by different @xmath2 ( 60000 to 90000 mm ) and @xmath3 ( 0 to -1.5 ) and different off - axis positions ( i.e. different segment centre coordinates in the m1 frame ) .
it also performs within specification in the presence of the allowed higher - order surface error terms ( only z4 and z9 tested here ) as defined for the e - elt segment manufacturing , although for the best accuracy it is recommended to subtract those terms from the data before ( re)fitting , especially if high levels of those errors are known to be present .
assuming the optic to be measured can be supported on a suitably stiff structure that allows the optic to be rotated around its x and y axes without introducing significant ( i.e. below the measurement error ) distortion on the surface form then the fitting method described provides a useful way to effectively extend the angular measurement range of an autocollimator in a non - contact profilometer and allows the line scans taken across the optic to be reconstructed to the best - fit aspheric ( conic ) surface .
s. alcock , k. sawhney , s. scott , u. pedersen , r. walton , f. siewert , t. zeschke , f. senf , t. noll , and h. lammert , `` the diamond nom : a non - contact profiler capable of characterizing optical figure error with sub - nanometre repeatability , '' nuclear instruments and methods in physics research a * 616 * , 224 ( 2010 ) . | we present a description of our method to process a set of autocollimator - based deflectometer 1-dimensional line - scans taken over a large optical surface and reconstruct them to a best - fit conic - section surface .
the challenge with our task is that each line - scan is in a different ( unknown ) coordinate reference frame with respect to the other line - scans in the set .
this problem arises due to the limited angular measurement range of the autocollimator used in the deflectometer and the need to measure over a greater range ; this results in the optic under measurement being rotated ( in pitch and roll ) between each scan to bring the autocollimator back into measurement range and therefore each scan is taken in a different coordinate frame .
we describe an approach using a @xmath0 dimension optimisation ( where @xmath1 is the number of scan lines taken across the mirror ) that uses a gradient - based non - linear least squares fitting combined with a multi - start global search strategy to find the best - fit surface .
careful formulation of the problem is required to reduce numerical noise and allow the routine to converge on a solution of the required accuracy .
= 1 120.3940 metrology ; 120.6650 surface measurements , figure ; 220.1250 aspherics .
2016 optical society of america .
one print or electronic copy may be made for personal use only .
systematic reproduction and distribution , duplication of any material in this paper for a fee or for commercial purposes , or modifications of the content of this paper are prohibited .
the version of this paper as published in applied optics can be viewed here : http://dx.doi.org/10.1364/ao.55.002827 |
we begin with an example .
let @xmath14 be an integer , and consider the cube @xmath15^n \subset \rr^n$ ] .
suppose that @xmath16 is a random vector that is distributed uniformly in the cube @xmath17
. then @xmath18 are independent , identically - distributed random variables of mean zero and variance one .
consequently , the classical central limit theorem states that the distribution of the random variable @xmath19 is close to the standard normal distribution , when @xmath20 is large .
moreover , suppose we are given @xmath21 with @xmath22 .
then under mild conditions on the @xmath23 s ( such as lindeberg s condition , see , e.g. , ( * ? ? ?
* section viii.4 ) ) , the distribution of the random variable @xmath24 is approximately gaussian , provided that the dimension @xmath20 is large . for background on the classical central limit theorem
we refer the reader to , e.g. , @xcite and @xcite .
let us consider a second example , no less fundamental than the first .
we denote by @xmath25 the standard euclidean norm in @xmath5 , and let @xmath26 around the origin in @xmath5 .
we also write @xmath27 for the unit sphere in @xmath5 .
suppose that @xmath28 is a random vector that is distributed uniformly in the ball @xmath29
. then @xmath30 are identically - distributed random variables of mean zero and variance one , yet they are not independent .
nevertheless , it was already observed by maxwell that for any @xmath31 , the distribution of the random variable @xmath32 is close to the standard normal distribution , when @xmath20 is large .
see , e.g. , @xcite for the history of the latter fact and for more information .
there is a wealth of central limit theorems in probability theory that ensure normal approximation for a sum of many independent , or weakly dependent , random variables .
our first example , that of the cube , fits perfectly into this framework .
the approach we follow in this paper relates more to the second example , that of the euclidean ball , where the `` true source '' of the gaussian approximation may be attributed to geometry .
the geometric condition we impose on the distribution of the random variables is that of convexity .
we shall see that convexity may substitute for independence in certain aspects of the phenomenon represented by the classical central limit theorem .
a function @xmath33 is log - concave if @xmath34 for all @xmath35 and @xmath36 .
that is , @xmath37 is log - concave when @xmath38 is concave on the support of @xmath37 .
examples of interest for log - concave functions include characteristic functions of convex sets , the gaussian density , and several densities from statistical mechanics . in this manuscript
, we consider random vectors in @xmath5 that are distributed according to a log - concave density .
thus , our treatment includes as a special case the uniform distribution on an arbitrary compact , convex set with a non - empty interior .
we say that a function @xmath33 is isotropic if it is the density of a random vector with zero mean and identity covariance matrix .
that is , @xmath37 is isotropic when @xmath39 for all @xmath40 .
any log - concave function with @xmath41 may be brought to an isotropic position via an affine map , that is , @xmath42 is isotropic for some affine map @xmath43 ( see , e.g. , @xcite ) .
suppose that @xmath2 and @xmath44 are two random variables attaining values in some measure space @xmath45 ( here @xmath45 will always be @xmath46 or @xmath5 or a subspace @xmath47 ) .
we define their total - variation distance as @xmath48 where the supremum runs over all measurable sets @xmath49 .
note that @xmath50 equals the @xmath51-distance between the densities of @xmath2 and @xmath44 , when these densities exist .
let @xmath52 stand for the unique rotationally - invariant probability measure on @xmath53 , also referred to as the uniform probability measure on the sphere @xmath53 .
there exist sequences @xmath54 for which the following holds : let @xmath14 , and let @xmath2 be a random vector in @xmath5 with an isotropic , log - concave density
. then there exists a subset @xmath55 with @xmath56 , such that for all @xmath57 , @xmath58 where @xmath59 is a standard normal random variable .
[ thm_basic ] we have the bounds @xmath60 and @xmath61 for @xmath62 and @xmath63 from theorem [ thm_basic ] , where @xmath64 are universal constants .
the quantitative estimate we provide for @xmath62 is rather poor .
while theorem [ thm_basic ] seems to be a reasonable analog of the classical central limit theorem for the category of log - concave densities , we are still lacking the precise berry - esseen type bound .
a plausible guess might be that the logarithmic dependence should be replaced by a power - type decay , in the bound for @xmath62 .
theorem [ thm_basic ] implies the result stated in the abstract of this paper , which does not require isotropicity ; indeed , recall that any log - concave density can be made isotropic by applying an appropriate affine map .
thus , any log - concave density in high dimension has at least one almost - gaussian marginal . when the log - concave density is also isotropic , we can assert that , in fact , the vast majority of its marginals are approximately gaussian
. an inherent feature of theorem [ thm_basic ] is that it does not provide a specific unit vector @xmath65 for which @xmath66 is approximately normal .
this is inevitable : we clearly can not take @xmath67 in the example of the cube above , and hence there is no fixed unit vector that suits all isotropic , log - concave densities .
nevertheless , under additional symmetry assumptions , we can identify a unit vector that always works .
borrowing terminology from banach space theory , we say that a function @xmath68 is unconditional if @xmath69 that is , @xmath37 is unconditional when it is invariant under coordinate reflections .
there exists a sequence @xmath0 for which the following holds : let @xmath14 , and let @xmath70 be an unconditional , isotropic , log - concave function .
let @xmath16 be a random vector in @xmath5 that is distributed according to the density @xmath37 .
then , @xmath71 where @xmath59 is a standard normal random variable .
[ thm_uncond ] we provide the estimate @xmath72 for @xmath62 from theorem [ thm_uncond ] .
multi - dimensional versions of theorem [ thm_basic ] are our next topic . for integers
@xmath73 with @xmath74 , let @xmath75 stand for the grassmannian of all @xmath76-dimensional subspaces in @xmath5 .
let @xmath77 be the unique rotationally - invariant probability measure on @xmath75 .
whenever we refer to the uniform measure on @xmath75 , and whenever we select a random @xmath76-dimensional subspace in @xmath5 , we always relate to the probability measure @xmath77 defined above . for a subspace @xmath78 and a point @xmath79 , let @xmath80 stand for the orthogonal projection of @xmath81 onto @xmath82 .
a standard gaussian random vector in a @xmath76-dimensional subspace @xmath83 is a random vector @xmath2 that satisfies @xmath84 for any measurable set @xmath85 .
there exists a universal constant @xmath86 for which the following holds : let @xmath87 be an integer , and let @xmath2 be a random vector in @xmath5 with an isotropic , log - concave density .
let @xmath88 and suppose that @xmath89 is an integer .
then there exists a subset @xmath90 with @xmath91 such that for any @xmath92 , @xmath93 where @xmath94 is a standard gaussian random vector in the subspace @xmath82 .
[ thm_multi ] that is , most @xmath76-dimensional marginals of an isotropic , log - concave function , are approximately gaussian with respect to the total - variation metric , provided that @xmath95 .
note the clear analogy between theorem [ thm_multi ] and milman s precise quantitative theory of dvoretzky s theorem , an analogy that dates back to gromov ( * ? ? ?
* section 1.2 ) .
readers that are not familiar with dvoretzky s theorem are referred to , e.g. , ( * ? ? ?
* section 4.2 ) , to @xcite or to @xcite .
dvoretzky s theorem shows that @xmath76-dimensional geometric projections of an @xmath20-dimensional convex body are @xmath96-close to a euclidean ball , provided that @xmath97 .
theorem [ thm_multi ] states that @xmath76-dimensional marginals , or measure - projections , of an @xmath20-dimensional convex body are @xmath96-close to gaussian when @xmath98 .
thus , according to dvoretzky s theorem , the geometric shape of the support of the marginal distribution may be approximated by a very regular body a euclidean ball , or an ellipsoid whereas theorem [ thm_multi ] demonstrates that the marginal distribution itself is very regular ; it is approximately normal .
more parallels between theorem [ thm_multi ] and dvoretzky s theorem are apparent from the proof of theorem [ thm_multi ] below .
we currently do not know whether there exists a single subspace that satisfies both the conclusion of theorem [ thm_multi ] and the conclusion of dvoretzky s theorem simultaneously ; both theorems show that a `` random subspace '' works with large probability , but with respect to different euclidean structures . the logarithmic dependence on the dimension is known to be tight in milman s form of dvoretzky s theorem .
however , we have no reason to believe that the quantitative estimates in theorem [ thm_multi ] are the best possible . there are several mathematical articles where theorem [ thm_basic ] is explicitly conjectured .
brehm and voigt suggest theorem [ thm_basic ] as a conjecture in @xcite , where they write that this conjecture appears to be `` known among specialists '' .
anttila , ball and perissinaki formulated the same conjecture in @xcite , independently and almost simultaneously with brehm and voigt .
anttila , ball and perissinaki also proved the conjecture for the case of uniform distributions on convex sets whose modulus of convexity and diameter satisfy certain quantitative assumptions .
gromov wrote a remark in ( * ? ? ?
* section 1.2 ) that seems related to theorem [ thm_basic ] and theorem [ thm_multi ] , especially in view of the techniques we use here . following @xcite and @xcite , significant contributions regarding the central limit problem for convex sets
were made by bastero and bernus @xcite , bobkov @xcite , bobkov and koldobsky @xcite , brehm and voigt @xcite , brehm , hinow , vogt and voigt @xcite , koldobsky and lifshits @xcite , e. and m. meckes @xcite , e. milman @xcite , naor and romik @xcite , paouris @xcite , romik @xcite , s. sodin @xcite , wojtaszczyk @xcite and others .
let us explain a few ideas from our proof .
we begin with a general principle that goes back to sudakov @xcite and to diaconis and freedman @xcite ( see also the expositions of bobkov @xcite and von weizscker @xcite .
a sharpening for the case of convex bodies was obtained by anttila , ball and perissinaki @xcite ) .
this principle reads as follows : suppose @xmath2 is any random vector in @xmath5 with zero mean and identity covariance matrix
. then most of the marginals of @xmath2 are approximately gaussian , if and only if the random variable @xmath99 is concentrated around the value one . in other words ,
typical marginals are approximately gaussian if and only if most of the mass is concentrated on a `` thin spherical shell '' of radius @xmath100 and width much smaller than @xmath100 .
therefore , to a certain extent , our task is essentially reduced to proving the following : let @xmath14 be an integer and let @xmath2 be a random vector with an isotropic , log - concave density in @xmath5 .
then for all @xmath101 , @xmath102 where @xmath103 are universal constants .
[ cor_202 ] a significantly superior estimate to that of theorem [ cor_202 ] , for the case where @xmath96 is a certain universal constant greater than one , is given by paouris @xcite , @xcite .
it would be interesting to try and improve the bound in theorem [ cor_202 ] also for smaller values of @xmath96 .
returning to the sketch of the proof , suppose that we are given a random vector @xmath2 in @xmath5 with an isotropic , log - concave density .
we need to show that most of its marginals are almost - gaussian .
select a random @xmath76-dimensional subspace @xmath78 , for a certain integer @xmath76 .
we use a concentration of measure inequality in a way similar to milman s proof of dvoretzky s theorem to show that with large probability of choosing the subspace @xmath82 , the distribution of the random vector @xmath104 is approximately spherically - symmetric .
this step is carried out in section [ section3 ] , and it is also outlined by gromov ( * ? ? ?
* section 1.2 ) . fix a subspace @xmath82 such that @xmath104 is approximately spherically - symmetric . in section [ section4 ]
we use the fourier transform to conclude that the approximation by a spherically - symmetric distribution actually holds in the stronger @xmath105-sense , after convolving with a gaussian . in section [ section5 ] we show that the gaussian convolution has only a minor effect , and we obtain a spherically - symmetric approximation to @xmath104 in the total - variation , @xmath51-sense .
thus , we obtain a density in the subspace @xmath82 that has two properties : it is log - concave , by prkopa - leindler , and it is also approximately radial .
a key observation is that such densities are necessarily very close to the uniform distribution on the sphere ; this observation boils down to estimating the asymptotics of some one - dimensional integral . at this point
, we further project our density , that is already known to be close to the uniform distribution on a sphere , to any lower - dimensional subspace . by maxwell s principle
we obtain an approximately gaussian distribution in this lower - dimensional subspace .
this completes the rough sketch of our proof . throughout this paper , unless stated otherwise , the letters @xmath106 etc
. denote positive universal constants , that are not necessarily the same in different appearances .
the symbols @xmath107 etc . denote universal constants that are assumed to be sufficiently large , while @xmath108 etc .
denote sufficiently small universal constants .
we abbreviate @xmath109 for the natural logarithm , @xmath110 for expectation , @xmath111 for probability and @xmath112 for volume .
_ acknowledgements .
_ i would like to thank charles fefferman , emanuel milman and vitali milman for interesting discussions on related subjects , and to boris tsirelson for mentioning the central limit problem for convex sets in his graduate course at tel - aviv university .
here we gather some useful facts pertaining mostly to log - concave densities . for more information about log - concave functions ,
the reader is referred to , e.g. , @xcite , @xcite and @xcite .
the raison dtre of log - concave densities on @xmath5 stems from the classical brunn - minkowski inequality and its generalizations .
let @xmath78 be a subspace , and let @xmath33 be an integrable function .
we denote the marginal of @xmath37 with respect to the subspace @xmath82 by @xmath113 where @xmath114 is the affine subspace in @xmath5 that is orthogonal to @xmath82 and passes through @xmath81 .
the prkopa - leindler inequality ( see @xcite , @xcite , @xcite or the first pages of @xcite ) , which is a functional version of brunn - minkowski , implies that @xmath115 is log - concave whenever @xmath37 is log - concave and integrable .
therefore , when @xmath37 is isotropic and log - concave , @xmath115 is also isotropic and log - concave
. a further consequence of the prkopa - leindler inequality , is that when @xmath37 and @xmath116 are integrable log - concave functions on @xmath5 , so is their convolution @xmath117 .
( the latter result actually goes back to @xcite , @xcite and @xcite . )
let @xmath14 be an integer , and let @xmath2 be a random vector in @xmath5 with a log - concave density .
assume that @xmath118 is an even , convex function , such that @xmath119 for all @xmath120 . denote @xmath121 .
then , 1 .
@xmath122 for all @xmath123 .
additionally , let @xmath124 , and let @xmath125 satisfy @xmath126 then , 1 .
@xmath127 for all @xmath128 .
[ lem_1010 ] lemma [ lem_1010 ] is the well - known borell s lemma ( see its elegant proof in @xcite or ( * ? ? ?
* theorem iii.3 ) ) .
let @xmath33 be an integrable function .
for @xmath65 and @xmath129 we define @xmath130 and @xmath131 the function @xmath132 is continuous in @xmath4 and @xmath133 , non - decreasing in @xmath133 , and its derivative @xmath134 is the radon transform of @xmath37 .
thus , in principle , one may recover the function @xmath37 from a complete knowledge of @xmath132 . clearly , for any subspace @xmath78 , @xmath135 moreover , let @xmath65 , let @xmath136 be the one - dimensional subspace spanned by @xmath4 , and denote @xmath137 .
then @xmath138 for all points @xmath129 where , say , @xmath139 is continuous .
let @xmath14 be an integer , and let @xmath33 be an isotropic , log - concave function .
fix @xmath65 . then , 1 . for @xmath123
we have @xmath140 .
2 . for @xmath141
we have @xmath142 [ lem_1028 ] _ proof : _ let @xmath2 be a random vector with density @xmath37 . then @xmath143 .
we use lemma [ lem_1010](i ) , with the function @xmath144 , to deduce the desired inequalities .
@xmath145 the space of all isotropic , log - concave functions in a fixed dimension is a compact space , with respect to , e.g. , the @xmath51-metric .
in particular , one - dimensional log - concave functions are quite rigid .
for instance , suppose that @xmath146 is an isotropic , log - concave function .
then ( see hensley @xcite and also , e.g. , ( * ? ? ?
* lemma 5.5 ) or @xcite ) , @xmath147 we conclude that for any log - concave , isotropic function @xmath148 , @xmath149 to prove ( [ eq_547 _ ] ) , we set @xmath136 and @xmath150 . then @xmath116 is isotropic and log - concave , hence @xmath151 by ( [ eq_547 ] ) . note that @xmath116 is continuous in the interior of its support , since it is a log - concave function . according to ( [ eq_621 _ ] )
, the function @xmath152 is the derivative of the function @xmath153 , and ( [ eq_547 _ ] ) follows .
our next proposition is essentially taken from anttila , ball and perissinaki @xcite , yet we use the extension to the non - even case which is a particular case of a result of bobkov ( * ? ? ?
* proposition 3.1 ) .
a function @xmath154 is @xmath155-lipshitz , for @xmath156 , if @xmath157 for all @xmath158 .
let @xmath14 be an integer .
let @xmath129 and let @xmath33 be an isotropic , log - concave function .
then , the function @xmath159 is @xmath160-lipshitz on @xmath53 . here
, @xmath161 is a universal constant .
[ lem_bobkov ] the proof of proposition [ lem_bobkov ] in @xcite involves analysis of two - dimensional log - concave functions . a beautiful argument yielding proposition [ lem_bobkov ] , for the case where @xmath37 is an even function , appears in @xcite .
the approach in @xcite is based on an application of busemann s theorem in dimension @xmath162 , which leads to the conclusion that @xmath163 is a norm on @xmath5 for any fixed @xmath123 .
it is well - known that for large @xmath20 , the uniform probability measure @xmath52 on the unit sphere @xmath53 satisfies strong concentration inequalities .
this concentration of measure phenomenon is one of the main driving forces in high - dimensional convex geometry , as was first demonstrated by milman in his proof of dvoretzky s theorem ( see @xcite or ( * ? ? ?
* section 4.2 ) ) .
our next proposition is essentially taken from milman s work , though the precise formulation we use is due to gordon @xcite , @xcite ( see also @xcite , @xcite or ( * ? ? ?
* theorem 6 ) ) .
let @xmath14 be an integer , let @xmath156 , @xmath164 , and let @xmath165 be an @xmath155-lipshitz function . denote @xmath166 .
assume that @xmath167 is an integer .
suppose that @xmath168 is a random subspace , i.e. , @xmath82 is distributed according to the probability measure @xmath77 on @xmath75 .
then , with probability greater than @xmath169 , @xmath170 here , @xmath171 are universal constants .
[ dvoretzky ] our use of `` dvoretzky s theorem type '' arguments in the next lemma is inspired by the powerful methods of paouris in @xcite , @xcite , @xcite .
let @xmath14 be an integer , let @xmath172 and let @xmath33 be an isotropic , log - concave function .
assume that @xmath173 is an integer , and let @xmath82 be a random @xmath174-dimensional subspace in @xmath5 .
then with probability greater than @xmath175 , @xmath176 here , @xmath177 is a universal constant .
[ lem_dvo ] _ proof : _ we may assume that @xmath20 exceeds a given universal constant , since otherwise , for a suitable choice of a small universal constant @xmath178 , there is no @xmath174 with @xmath179 .
fix a real number @xmath133 . according to proposition [ lem_bobkov ] ,
the function @xmath180 is @xmath160-lipshitz on @xmath53 .
let @xmath181 be a random subspace , uniformly distributed in @xmath182 .
we would like to apply proposition [ dvoretzky ] with @xmath183 and @xmath184 .
note that for this choice of parameters , @xmath185 provided that @xmath178 is a sufficiently small , positive universal constant , and that @xmath20 is greater than some universal constant .
hence the appeal to proposition [ dvoretzky ] is legitimate . from the conclusion of that proposition , with probability larger than @xmath186 of selecting @xmath82 , @xmath187 for any fixed @xmath129 , the estimate ( [ eq_434 ] ) holds with probability greater than @xmath186 .
denote @xmath188 .
then , with probability greater than @xmath189 , we obtain @xmath190 indeed , the estimate for the probability follows from the inequality @xmath191 . fix an @xmath174-dimensional subspace @xmath83 that satisfies ( [ eq_443 ] ) .
select @xmath192 .
we will demonstrate that for any @xmath129 , @xmath193 to that end , note that when @xmath194 , by lemma [ lem_1028 ] , @xmath195 hence ( [ eq_618 ] ) holds for @xmath194 . we still need to consider the case where @xmath196 . in this case ,
@xmath197 and hence there exists @xmath198 with @xmath199 . according to ( [ eq_547 _ ] ) from section [ section2 ] , the function @xmath200 is @xmath201-lipshitz for @xmath202 .
therefore , by using ( [ eq_443 ] ) , we conclude ( [ eq_618 ] ) also for the case where @xmath196 .
thus ( [ eq_618 ] ) holds for all @xmath129 , under the assumption that @xmath82 satisfies ( [ eq_443 ] ) .
recall that @xmath192 are arbitrary , hence we may take the supremum over @xmath203 and the infimum over @xmath204 in ( [ eq_618 ] ) .
we discover that whenever the subspace @xmath82 satisfies ( [ eq_443 ] ) , it necessarily also satisfies ( [ eq_616 ] ) . the probability for a random @xmath174-dimensional subspace @xmath78 to satisfy ( [ eq_443 ] )
was shown to be greater than @xmath205 .
the lemma thus follows .
@xmath145 _ remark .
_ for the case where @xmath37 is even , lemma [ lem_dvo ] follows from a direct application of dvoretzky s theorem in milman s form . indeed , in this case
, @xmath206 is a norm , and lemma [ lem_dvo ] asserts that this norm is almost hilbertian when restricted to certain random subspaces .
a large portion of this section is devoted to proving the following proposition
. there exist universal constants @xmath207 and @xmath177 for which the following holds : let @xmath14 be an integer and let @xmath33 be an isotropic , log - concave function .
assume that @xmath208 suppose that @xmath44 is a random vector in @xmath5 with density @xmath37 .
then for all @xmath209 , @xmath210 [ cor_735 ] for @xmath14 and @xmath211 we define @xmath212 to be the function @xmath213 then @xmath214 is the density of a gaussian random vector in @xmath5 with expectation zero and covariance matrix that equals @xmath215 , where @xmath216 is the identity matrix .
we write @xmath217 for the group of orthogonal transformations of @xmath5 .
let @xmath14 be an integer , let @xmath218 , and let @xmath33 be an isotropic , log - concave function .
assume that @xmath219 denote @xmath220 , where @xmath221 stands for convolution .
then , @xmath222 [ lem_fourier ] _ proof : _ we will show that the fourier transform of @xmath37 is almost spherically - symmetric . as usual
, we define @xmath223 where @xmath224 .
let @xmath225 , and fix @xmath226 with @xmath227
. denote by @xmath228 the one - dimensional subspaces spanned by @xmath229 , respectively . from ( [ eq_621 _ ] ) of section [ section2 ]
we see that @xmath230 for @xmath231 and for all @xmath133 in the interior of the support of the log - concave function @xmath232 . by integrating by parts we obtain @xmath233 e^{-2 \pi i r t } dt } \\ & = & 2 \pi i r \int_{-\infty}^{\infty } \left[m_{f } \left ( \frac{\xi_1}{|\xi_1| } , t \right ) - m_{f } \left ( \frac{\xi_2}{|\xi_2| } , t \right ) \right ] e^{-2 \pi i r t } dt , \phantom{aaaaaaaa}\end{aligned}\ ] ] as the boundary terms clearly vanish . from lemma [ lem_1028 ]
we have @xmath234 according to ( [ f_xi ] ) , ( [ eq_933 ] ) and to our assumption ( [ eq_848 ] ) , we conclude that for any @xmath225 and @xmath235 with @xmath227 , @xmath236 \leq r e^{-2 \alpha n } , \nonumber\end{aligned}\ ] ] where we made use of the fact that @xmath237 . a standard computation ( e.g. @xcite )
shows that @xmath238 . recall that we define @xmath239 , and hence @xmath240 .
we thus deduce from ( [ eq_1042 ] ) that for any @xmath226 , @xmath241 let @xmath79 , and let @xmath242 be an orthogonal transformation . by using the inverse fourier transform ( see , e.g. ( * ? ? ?
* chapter i ) ) and applying ( [ eq_1103 ] ) , we get @xmath243 e^{2 \pi i \langle x , \xi \rangle } d\xi \right| } \\ & \leq & \int_{\rr^n } e^{-2 \pi^2 |\xi|^2 } |\xi| e^{-2 \alpha n } d \xi \leq e^{-2 \alpha n } \int_{\rr^n } e^{-\pi @xmath79 and @xmath242 are arbitrary , from ( [ eqn_1110 ] ) we conclude ( [ eq_750 ] ) .
@xmath145 let @xmath244 be a log - concave function with @xmath245 , that is continuous on @xmath246 and @xmath247-smooth on @xmath248
. for @xmath249 , denote by @xmath250 the unique @xmath251 for which @xmath252 and also @xmath253 @xmath250 is well - defined , under the above assumptions on @xmath37 and @xmath254 .
_ proof : _ we need to explain why a solution @xmath133 to equation ( [ eq_106 ] ) exists and is unique , for all @xmath255 . to that end , note that @xmath37 is a log - concave function with finite , positive mass , hence it decays exponentially fast at infinity ( this is a very simple fact ; see , e.g. , ( * ? ? ?
* lemma 2.1 ) ) . therefore , the function @xmath256 satisfies @xmath257 the function @xmath258 is continuous , non - negative , not identically zero , and tends to zero at @xmath259 and at @xmath260 .
consequently , @xmath258 attains its positive maximum at some finite point @xmath261
. then @xmath262 and @xmath263 , since @xmath258 is @xmath247-smooth .
on the other hand , @xmath37 is log - concave , and @xmath264 is strictly log - concave , hence @xmath258 is strictly log - concave on its support .
therefore , there is at most one point where @xmath258 is non - zero and @xmath265 vanishes .
we conclude that there exists exactly one point @xmath266 such that @xmath267 and @xmath268 = 0.\ ] ] thus a finite , positive @xmath133 that solves ( [ eq_106 ] ) exists and is unique .
@xmath145 let us mention a few immediate properties of the quantity @xmath250 .
first , @xmath269 for all @xmath255 .
second , suppose that @xmath37 is a continuous , log - concave function on @xmath246 , @xmath247-smooth on @xmath248 , with @xmath41 .
then , @xmath270 indeed , if @xmath271 then ( [ eq_1127 ] ) is trivial .
otherwise , @xmath272 and @xmath273 , hence @xmath37 is necessarily positive on @xmath274 $ ] by log - concavity .
therefore @xmath38 is finite and continuous on @xmath274 $ ] , and @xmath247-smooth in @xmath275 .
additionally , @xmath38 is concave , hence @xmath276 is non - increasing in @xmath277 . from the definition ( [ eq_106 ] ) we deduce that @xmath278 for all @xmath279 , and ( [ eq_1127 ] ) follows .
furthermore , since @xmath276 is non - increasing on the interval in which it is defined , then @xmath280 for @xmath281 for which @xmath252 . we conclude that for any @xmath282 , @xmath283 note that @xmath250 behaves well under scaling of @xmath37 .
indeed , let @xmath37 be a continuous , log - concave function on @xmath246 , @xmath247-smooth on @xmath248 , with @xmath41
. for @xmath284 , denote @xmath285 . from the definition ( [ eq_106 ] )
we see that for any @xmath255 , @xmath286 let @xmath287 , and let @xmath288 be continuous , log - concave functions , @xmath247-smooth on @xmath248 , such that @xmath289 and @xmath290 .
assume that for any @xmath123 , @xmath291 then , @xmath292 [ lem_257 ] _ proof : _ set @xmath293 . according to ( [ scale ] ) ,
both the conclusions and the requirements of the lemma are invariant when we replace @xmath37 and @xmath116 with @xmath294 and @xmath295 , respectively .
we apply this replacement , and assume from now on that @xmath296 .
inequality ( [ eq_1127 ] ) and our assumption that @xmath272 show that @xmath297 for @xmath298 .
we combine this inequality with ( [ eq_1132 _ ] ) to obtain the bound @xmath299 for all @xmath298 .
in particular , @xmath116 is positive on @xmath300 $ ] .
denote @xmath301 .
then for all @xmath298 , @xmath302 next , we claim that @xmath303 indeed , assume by contradiction that ( [ eq_1129 ] ) does not hold
. then there exists @xmath304 for which @xmath305 . from our assumptions , @xmath37 and @xmath116
are log - concave , hence @xmath306 and @xmath307 are concave , and hence @xmath308 and @xmath309 are non - increasing on @xmath310 .
therefore , for @xmath311 , @xmath312 denote @xmath313 .
then @xmath314 \subset [ 0 , 1]$ ] and by ( [ eq_1149 ] ) , @xmath315 \ , - \ , \left [ f_0(t_0 ) - g_0(t_0 ) \right ] \
, > \ , 4 e^{-2n } \cdot ( t_1 - t_0 ) \ , = \ , 4 e^{-4n},\ ] ] in contradiction to ( [ eq_1135 ] ) .
thus , our momentary assumption that ( [ eq_1129 ] ) does not hold was false , and hence ( [ eq_1129 ] ) is proved . from the definition ( [ eq_106 ] )
we see that @xmath316 .
recall once again that @xmath309 is non - increasing . by applying the case @xmath317 in ( [ eq_1129 ] )
, we conclude that for @xmath318 , @xmath319 from ( [ eq_1135 _ ] ) we conclude that @xmath320 for all @xmath321 .
the definition ( [ eq_106 ] ) shows that @xmath322 recalling the scaling argument above , we see that we have actually proved that @xmath323 whenever the assumptions of the lemma hold .
however , these assumptions are symmetric in @xmath37 and @xmath116 .
hence , @xmath324 for any functions @xmath325 that satisfy the assumptions of the lemma . since @xmath326 for @xmath287 , the lemma is proved .
our next lemma is a standard application of the laplace asymptotic method , and is similar to , e.g. , ( * ? ? ?
* lemma 2.1 ) and ( * ? ? ?
* lemma 2.5 ) .
we will make use of the following well - known bound : for @xmath327 , @xmath328 the inequality in ( [ eq_1218 ] ) may be proved , for example , by computing the laplace transform of the gaussian density and applying markov s inequality ( e.g. , ( * ? ? ?
* section 1.3 ) ) .
let @xmath287 be an integer , and let @xmath329 be a continuous , log - concave function , @xmath247-smooth on @xmath330 , with @xmath331 .
then for @xmath332 , @xmath333 where @xmath334 and @xmath177 are universal constants .
[ lem_1153 ] _ proof : _ we begin with a scaling argument .
a glance at ( [ scale ] ) and ( [ eq_248 ] ) assures us that both the validity of the assumptions and the validity of the conclusions of the present lemma , are not altered when we replace @xmath37 with @xmath335 , for any @xmath336 .
hence , we may switch from @xmath37 to @xmath337 , and reduce matters to the case @xmath296 .
thus @xmath338 . multiplying @xmath37 by an appropriate positive constant
, we may assume that @xmath339 .
we denote @xmath340 , where we set @xmath341 whenever @xmath342 .
since @xmath339 , then @xmath343 .
additionally , @xmath344 because @xmath296 .
the function @xmath345 is concave , and therefore it attains its maximum at @xmath201 .
let @xmath346 be the minimal positive numbers for which @xmath347 and @xmath348 .
such @xmath349 and @xmath350 exist since @xmath345 is continuous , @xmath343 and @xmath351 when @xmath352 ( because of @xmath353 ) and when @xmath354 ( because of @xmath38 , since @xmath37 is log - concave with @xmath41 ) .
we may suppose that @xmath355 ; for an appropriate choice of a large universal constant @xmath160 , the right hand side of ( [ eq_248 ] ) is negative for @xmath356 , and hence the lemma is obvious for @xmath356 .
denote @xmath357 and @xmath358 . since @xmath296 , necessarily @xmath359 and @xmath360 .
then , for @xmath361 , @xmath362 since @xmath38 is concave and hence @xmath363 . from ( [ eq_113 ] ) we obtain , in particular , the inequality @xmath364 for @xmath365 .
recalling that @xmath366 , we see that @xmath367 for all @xmath368 .
therefore @xmath369 and @xmath370 , and consequently @xmath371 since @xmath355 , then ( [ s0_above ] ) implies that @xmath372 . recall that the function @xmath345 is concave , hence @xmath373 is non - increasing .
the relations @xmath374 thus imply that @xmath375 examination of ( [ eq_113 ] ) shows us that @xmath376 for @xmath377 . by definition ,
we thus conclude from ( [ eq_302 ] ) that @xmath378 for @xmath379 .
fix @xmath101 . then , @xmath380 where we used ( [ eq_1218 ] ) to estimate the last integral .
next , observe again that @xmath381 for all @xmath382 , by ( [ eq_113 ] )
. we use ( [ eq_302 ] ) , as well as the fact that @xmath348 , to obtain @xmath383 consequently , @xmath384 by ( [ eq_1218 ] ) . since @xmath385 , we deduce from ( [ eq_1046 ] ) that @xmath386 .
recall that @xmath373 is non - increasing , that @xmath344 and that @xmath387 for @xmath388 .
therefore , @xmath389 whenever @xmath390 .
thus we realize that @xmath391 for @xmath123 .
hence , @xmath392 let @xmath393 .
then , by the definition of @xmath349 and @xmath350 , @xmath394 the inequalities we gathered above will allow us to prove ( [ eq_248 ] ) .
note that ( [ eq_248 ] ) is trivial when @xmath395 ; for an appropriate choice of a large constant @xmath160 , the right - hand side of ( [ eq_248 ] ) is negative in this case .
we may thus restrict our attention to the case where @xmath396 .
hence , @xmath397 and @xmath398 , by ( [ s0_above ] ) .
we add ( [ eq_121 ] ) , ( [ eq_909 ] ) and ( [ eq_935 ] ) to get @xmath399 division of ( [ eqn_603 ] ) by ( [ total ] ) yields , @xmath400 in order to establish ( [ eq_248 ] ) and complete the proof , it is sufficient to show that @xmath401 according to ( [ s0_above ] ) , we know that @xmath402 . in the case where @xmath403 we have @xmath404 and hence the estimate ( [ eq_624 ] ) follows from ( [ eq_613 ] ) by choosing the `` @xmath405 '' term in the minimum in ( [ eq_613 ] ) . in the complementary case , we have @xmath406 since @xmath407 for @xmath128 . in this case ,
@xmath408 , and ( [ eq_624 ] ) follows by selecting the `` @xmath409 '' term in ( [ eq_613 ] ) .
hence ( [ eq_624 ] ) is proved for all cases .
the proof is complete .
@xmath145 the following lemma is standard , and is almost identical , for example , to ( * ? ? ?
* appendix v.4 ) . for a random vector @xmath2 in @xmath5
, we denote its covariance matrix by @xmath410 .
let @xmath14 be an integer , let @xmath411 and let @xmath2 be a random vector in @xmath5 with @xmath412 and @xmath413 .
assume that the density of @xmath2 is log - concave , and that @xmath414 then , 1 .
for all @xmath415 2 .
@xmath416 provided that @xmath417 . here , @xmath418 are constants that depend solely on @xmath419 and @xmath420 .
[ crude_lem ] _ proof : _ by a simple scaling argument , we may assume that @xmath421 ; otherwise , replace the function @xmath422 with the function @xmath423 .
in this proof , @xmath424 etc .
stand for constants depending only on @xmath419 and @xmath420 .
we begin by proving ( ii ) . since @xmath425 , lemma [ lem_1010](i ) implies that @xmath426 therefore , @xmath427 provided that @xmath428 . from ( [ eq_342 ] )
we deduce ( ii ) . to prove ( i ) , it is enough to consider the case where @xmath429 . in this case , by ( ii ) ,
@xmath430 and ( i ) follows from ( [ eq_855 ] ) for the range @xmath431 . by adjusting the constants , we establish ( i ) for the entire range @xmath101 .
@xmath145 let @xmath14 be an integer , let @xmath432 , and let @xmath33 be a log - concave function that is the density of a random vector with zero mean and with covariance matrix that equals @xmath433 .
then @xmath434 where @xmath177 is a universal constant .
[ zero_large_lem ] _ proof : _ the inequality @xmath435 is proved in ( * ? ? ? * theorem 4 ) . by our assumptions , @xmath436 .
markov s inequality entails @xmath437 therefore , @xmath438 since @xmath439 ( see , e.g. , @xcite ) .
@xmath145 _ proof of proposition [ cor_735 ] : _ recall our assumption ( [ eq_848 _ ] ) and our desired conclusion ( [ eq_309 ] ) from the formulation of the proposition .
we assume that @xmath20 is greater than some large universal constant , since otherwise ( [ eq_309 ] ) is obvious for an appropriate choice of constants @xmath440 .
denote @xmath239 , the convolution of @xmath37 and @xmath441 .
then @xmath116 is log - concave , and is the density of a random vector with mean zero and covariance matrix @xmath442 . by lemma [ zero_large_lem ] ,
@xmath443 we set @xmath444 where @xmath445 is the constant from ( [ zero_large ] ) .
our assumption ( [ eq_848 _ ] ) is precisely the basic requirement of lemma [ lem_fourier ] , for @xmath446 . by the conclusion of that lemma , @xmath447 since @xmath448 , according to the definition of @xmath449 and ( [ zero_large ] ) .
the function @xmath116 is @xmath450-smooth , since @xmath239 with @xmath441 being @xmath450-smooth .
additionally , since @xmath451 then for some @xmath452 , @xmath453 ( see , e.g. , ( * ? ? ?
* lemma 2.1 ) ) . for @xmath65 and @xmath123
, we write @xmath454 .
then @xmath455 is log - concave , continuous on @xmath246 , @xmath450-smooth on @xmath248 and integrable on @xmath456 by ( [ eq_408 ] ) .
in addition , @xmath457 by ( [ zero_large ] ) .
fix @xmath458 , and denote @xmath459 . according to ( [ eq_742 _ ] ) , for any @xmath65 and @xmath123 , @xmath460 thus the functions @xmath455 and @xmath461 satisfy the assumptions of lemma [ lem_257 ] , for any @xmath65 . by the conclusion of that lemma , for any @xmath65 ,
@xmath462 because @xmath463 .
we deduce that for any @xmath464 and @xmath65 , @xmath465 for @xmath101 let latexmath:[$a_{\eps } = \ { x \in \rr^n ; \big | @xmath101 , @xmath467 note that ( [ eq_313 ] ) is obvious for @xmath468 , since in this case @xmath469 for an appropriate choice of universal constants @xmath470 .
we still need to deal with the case @xmath471 . to that end , note that @xmath455 satisfies the requirements of lemma [ lem_1153 ] for any @xmath472 by the discussion above .
we will integrate in polar coordinates and use ( [ eq_303 ] ) as well as lemma [ lem_1153 ] .
this yields @xmath473 since @xmath474 .
this completes the proof of ( [ eq_313 ] ) .
let @xmath475 be a sequence of independent , real - valued , standard gaussian random variables . by the classical central limit theorem , @xmath476 consequently , @xmath477 for some universal constant @xmath478 .
denote @xmath479 .
then @xmath2 is distributed according to the density @xmath441 in @xmath5 .
we record the bound just mentioned : latexmath:[\[\frac{1}{c^{\prime } } \leq prob \ { @xmath44 be another random vector in @xmath5 , independent of @xmath2 , that is distributed according to the density @xmath37 .
since the density of @xmath2 is an even function , then for any measurable sets @xmath481 with @xmath482 and @xmath483 , @xmath484 additionally , the random vector @xmath485 has @xmath116 as its density , because @xmath239 .
therefore ( [ eq_313 ] ) translates to @xmath486 since @xmath2 and @xmath44 are independent , we conclude from ( [ bound_x ] ) , ( [ eq_130 ] ) and ( [ eq_132 ] ) that for all @xmath209 , @xmath487 and similarly , @xmath488 next , we estimate @xmath489 . recall that the density of @xmath490 is log - concave , @xmath491 and @xmath492 .
we invoke lemma [ crude_lem](ii ) , based on ( [ eq_132 ] ) , and conclude that @xmath493 , under the legitimate assumption that @xmath494 . denote @xmath495 . then @xmath496 and @xmath497 for @xmath498 .
therefore , ( [ eq_136 ] ) and ( [ eq_141 ] ) imply that for any @xmath499 , @xmath500 after adjusting the constants , we see that @xmath501 recall that @xmath44 is distributed according to the density @xmath37 , which is an isotropic , log - concave function .
we may thus apply lemma [ crude_lem](i ) , based on ( [ eq_346 ] ) , and conclude ( [ eq_309 ] ) .
the proposition is proved .
we proceed to discuss applications of proposition [ cor_735 ] .
the following lemma is usually referred to as the johnson - lindenstrauss dimension reduction lemma @xcite .
we refer , e.g. , to ( * ? ? ?
* lemma 2.2 ) for an elementary proof . recall that we denote by @xmath80 the orthogonal projection of @xmath81 onto @xmath82 , whenever @xmath81 is a point in @xmath5 and @xmath78 is a subspace .
let @xmath74 be integers , and let @xmath502 be a random @xmath76-dimensional subspace .
let @xmath79 be a fixed vector .
then for all @xmath101 , @xmath503 where @xmath103 are universal constants .
[ lem_find_reference ] _ proof of theorem [ cor_202 ] : _ we use the constant @xmath504 from proposition [ cor_735 ] , and the constant @xmath178 from lemma [ lem_dvo ] .
let @xmath505 and fix @xmath506 .
we may assume that @xmath507 ; otherwise , @xmath20 is smaller than some universal constant and the conclusion of the theorem is obvious .
we assume that @xmath2 is a random vector in @xmath5 whose density is an isotropic , log - concave function to be denoted by @xmath37 .
let @xmath508 be a fixed subspace that satisfies @xmath509 denote @xmath137 .
then ( [ eq_609 _ ] ) translates , with the help of ( [ proj_ok ] ) from section [ section2 ] , to @xmath510 the function @xmath116 is an isotropic , log - concave function , and it is the density of @xmath104 .
we invoke proposition [ cor_735 ] , for @xmath174 and @xmath116 , based on ( [ eq_609 ] ) . by the conclusion of that proposition , @xmath511 under the assumption that the subspace @xmath82 satisfies ( [ eq_609 _ ] ) .
suppose that @xmath512 is a random @xmath174-dimensional subspace in @xmath5 , independent of @xmath2 .
recall our choice of the integer @xmath174 . according to lemma [ lem_dvo ] , with probability greater than @xmath513 , the subspace @xmath514 satisfies ( [ eq_609 _ ] ) .
we conclude from ( [ eq_625 ] ) that @xmath515 where the last inequality holds as @xmath516 and @xmath517 . since @xmath2 and @xmath518 are independent , then by lemma [ lem_find_reference ] , @xmath519 to summarize , with probability greater than @xmath520 we have 1 .
@xmath521 , and also 2 .
@xmath522 .
hence , @xmath523 note that @xmath524 and @xmath525 , and recall that @xmath526 was arbitrary , and that @xmath527 . by adjusting the constants
, we deduce from ( [ eq_841 ] ) that the inequality in the conclusion of the theorem is valid for all @xmath101 .
the theorem is thus proved .
@xmath145 the following lemma may be proved via a straightforward computation . nevertheless , we will present a shorter , indirect proof that is based on properties of the heat kernel , an idea we borrow from ( * ? ? ?
* theorem 3.1 ) .
let @xmath14 be an integer and let @xmath528 .
then , @xmath529 where @xmath161 is a universal constant .
[ lem_929 ] _ proof : _ the integral on the left - hand side of ( [ eq_1021 ] ) is never larger than @xmath530 .
consequently , the lemma is obvious when @xmath531 or when @xmath532 , and hence we may assume that @xmath533 .
moreover , in this case both the left - hand side and the right - hand side of ( [ eq_1021 ] ) are actually symmetric in @xmath420 and @xmath534 up to a factor of at most @xmath530 .
therefore , we may assume that @xmath535 ( the case @xmath536 is obvious ) .
for @xmath537 and for a measurable function @xmath68 , we define @xmath538 whenever the integral converges .
then @xmath539 is the heat semigroup on @xmath5 .
we will make use of the following estimate : for any smooth , integrable function @xmath68 and any @xmath540 , @xmath541 an elegant proof of the inequality ( [ eq_1032 _ ] ) , in a much more general setting , is given by ledoux ( * ? ? ? * section 5 ) .
it is straightforward to verify that @xmath542 consequently , ( [ eq_1032 _ ] ) implies that @xmath543 it is well - known and easy to prove that @xmath544 .
since @xmath545 , then ( [ eq_534 ] ) implies ( [ eq_1021 ] ) .
the lemma is proved .
we are now able to prove theorem [ thm_uncond ] by combining the classical berry - esseen bound with theorem [ cor_202 ] .
_ proof of theorem [ thm_uncond ] : _ we may assume that @xmath20 exceeds a given universal constant .
let @xmath37 and @xmath2 be as in the assumptions of theorem [ thm_uncond ] .
according to theorem [ cor_202 ] , @xmath546 the case @xmath547 in ( [ eq_115 _ ] ) shows that @xmath548 , under the legitimate assumption that @xmath20 exceeds a certain universal constant . by ( [ eq_115 _ ] ) and by lemma [ lem_1010](ii ) , @xmath549 let @xmath550 be independent bernoulli random variables , that are also independent of @xmath2 , such that @xmath551 for @xmath552
. for @xmath129 and @xmath553 denote @xmath554 we write @xmath555 for @xmath7 and @xmath129 . by the berry - esseen bound ( see , e.g. , ( * ? ? ?
* section xvi.5 ) or ( * ? ? ?
* section 2.1.30 ) ) , for any @xmath79 , @xmath556 where @xmath161 is a universal constant . since @xmath37 is unconditional , the random variable @xmath557 has the same law of distribution as the random variable @xmath558 .
for @xmath129 we set @xmath559 we denote the expectation over the random variable @xmath2 by @xmath560
. then @xmath561 by the complete probability formula . for @xmath552 ,
the random variable @xmath562 has mean zero , variance one , and its density is a log - concave function .
consequently , @xmath563 , and by lemma [ lem_1010](i ) , for any @xmath564 , @xmath565 therefore , with probability greater than @xmath566 of selecting @xmath2 , @xmath567 fix @xmath129 .
we substitute into ( [ eq_1230 ] ) the information from ( [ eq_109 ] ) , and from the case @xmath568 in ( [ eq_115 _ ] ) .
we see that with probability greater than @xmath569 of selecting @xmath2 , @xmath570 since always @xmath571 and @xmath572 , we conclude that @xmath573 according to lemma [ lem_929 ] , for any @xmath79 , @xmath574 and therefore by ( [ eq_115 _ _ ] ) @xmath575 recall that @xmath576 and that @xmath133 is an arbitrary real number .
we apply jensen s inequality , and then combine ( [ eq_117 ] ) and ( [ eq_609 _ _ ] ) to obtain @xmath577 the random variable @xmath578 has mean zero , variance one and a log - concave density .
its cumulative distribution function @xmath579 satisfies ( [ eq_200 ] ) .
therefore , we may invoke ( * ? ? ?
* theorem 3.3 ) , and conclude from ( [ eq_200 ] ) that @xmath580 where @xmath59 is a standard gaussian random variable .
the theorem follows , with @xmath581 .
@xmath145 _ remarks . _ 1 .
suppose that @xmath37 is a log - concave density in high dimension that is isotropic and unconditional . in theorem [ thm_uncond
] , we were able to describe an explicit one - dimensional marginal of @xmath37 that is approximately normal .
it seems possible to identify some multi - dimensional subspaces @xmath78 , spanned by specific sign - vectors , such that @xmath115 is guaranteed to be almost - gaussian .
we did not pursue this direction .
2 . under the assumptions of theorem [ thm_uncond ]
, we proved that @xmath66 is approximately gaussian when @xmath582 .
a straightforward adaptation of the proof of theorem [ thm_uncond ] shows that @xmath66 is approximately gaussian under the weaker assumption that @xmath583 are rather small ( as in lindeberg s condition ) .
3 . theorem [ thm_basic ] , with a worse bound for @xmath62 , follows by combining theorem [ cor_202 ] with the methods in @xcite , and then applying ( * ? ? ?
* theorem 3.3 )
. we will deduce theorem [ thm_basic ] from the stronger theorem [ thm_multi ] in the next section .
the next few pages are devoted to the proof of the following lemma .
we begin with an addendum to lemma [ lem_1153 ] . rather than appealing to the laplace asymptotic method once again , we will base our proof on an elegant observation by bobkov regarding one - dimensional log - concave functions .
_ proof : _ if @xmath591 then @xmath592 almost everywhere and ( [ eq_329 ] ) is trivial .
thus , we may suppose that @xmath593 .
moreover , we may assume that @xmath37 is continuous on @xmath456 and @xmath247-smooth on @xmath248 , by approximation ( for example , convolve @xmath37 with @xmath594 on @xmath46 , restrict the result to @xmath246 , and let @xmath96 tend to zero ) .
since @xmath41 then @xmath37 decays exponentially fast at infinity , and @xmath595 .
multiplying @xmath37 by a positive constant , we may assume that @xmath596
. for @xmath251 , denote , @xmath597 then @xmath598 is a log - concave function with @xmath599 .
recall the definition of @xmath600 , that is , ( [ eq_106 ] ) from section [ section4 ] . according to that definition , @xmath601 .
. then @xmath603 by ( [ eq_1127 ] ) from section [ section4 ] , and hence @xmath604 where @xmath605 is defined in the formulation of the lemma . since @xmath606 and since @xmath37 is continuous and vanishes at infinity , the number @xmath607 is finite , greater than @xmath600 , and satisfies @xmath608 . from ( [ eq_1127_2 ] ) of section [ section4 ] we see that @xmath609 .
therefore , @xmath610 where @xmath611 because @xmath598 is log - concave , @xmath612 and @xmath601 .
let @xmath613 stand for the inverse function to @xmath614 .
a useful fact we learned from bobkov s work ( * ? ? ? *
lemma 3.2 ) is that the function @xmath615 is concave on @xmath310 .
( to see this , differentiate @xmath345 twice , and use the inequality @xmath616 . )
since @xmath598 attains its maximum at @xmath600 , then @xmath345 attains its maximum at @xmath617 .
the function @xmath345 is non - negative and concave on @xmath310 , hence for @xmath618 and @xmath619 , @xmath620 equivalently , for @xmath621 and @xmath209 , the inequality @xmath622 implies the bound @xmath623 .
we have shown that @xmath624 satisfies @xmath625 , and hence we conclude that @xmath626 .
recalling that @xmath627 , the lemma follows .
@xmath145 _
proof : _ for @xmath65 set @xmath632 by log - concavity , @xmath633 is a ( possibly infinite ) interval in @xmath246 containing zero . for @xmath123 and @xmath65
we denote @xmath634 .
then @xmath635 is log - concave . since @xmath636 ,
then , e.g. , by ( * ? ? ?
* lemma 2.1 ) we know that @xmath37 decays exponentially fast at infinity and @xmath637 .
next , we integrate in polar coordinates and use lemma [ lem_1153 _ ] .
this yields @xmath638 @xmath145 _ proof : _ assume the contrary . since @xmath3 is convex , then there exists @xmath65 such that @xmath641 .
hence , @xmath642 denote @xmath136 , the one - dimensional line spanned by @xmath4 , and let @xmath150 .
then @xmath116 is log - concave and isotropic , hence @xmath151 by ( [ eq_547 ] ) of section [ section2 ] .
since @xmath116 is the density of the random variable @xmath66 and @xmath643 , then @xmath644 an appeal to ( * ? ? ?
* lemma 3.3 ) a result that essentially goes back to grnbaum and hammer @xcite shows that @xmath645 after adding ( [ eq_957 ] ) to ( [ eq_956 ] ) , we arrive at a contradiction to ( [ eq_1018 ] ) .
this completes the proof .
let @xmath287 be an integer , let @xmath584 , and let @xmath33 be an isotropic , log - concave function . consider the sets @xmath649 and @xmath650 .
then , @xmath651 where @xmath161 is a universal constant .
[ lem_1208 ] _ proof : _ let @xmath652 be the probability measure on @xmath5 whose density is @xmath37 . by corollary [ cor_302 ] , @xmath653 the set @xmath654 is convex , since @xmath37 is log - concave . according to ( [ eq_304 ] ) and lemma [ lem_lovasz ] , @xmath655 by the definition , @xmath656 .
since @xmath657 , then @xmath658 because @xmath654 is convex and @xmath659 .
we use ( [ eq_159 ] ) and lemma [ zero_large_lem ] for @xmath421 .
this implies the estimate @xmath660 where we also used the standard estimate @xmath661 .
the inclusion ( [ eq_159 ] ) and the convexity of @xmath654 entail that @xmath662 therefore , according to the prkopa - leindler inequality , @xmath663 we combine ( [ eq_944 ] ) , ( [ eq_1015 ] ) , ( [ eq_218 ] ) and ( [ eq_304 ] ) to obtain @xmath664 for some universal constant @xmath478 ( the verification of the last inequality is elementary and routine ) .
the lemma is thus proved .
@xmath145 _ proof of lemma [ lem_915 ] : _ by approximation
, we may assume that @xmath37 is continuously differentiable .
denote @xmath665 ( with @xmath666 when @xmath667 ) .
then @xmath345 is a concave function .
consider the sets @xmath668 and @xmath669 .
the first step of the proof is to show that @xmath670 note that @xmath272 by ( * ? ? ?
* theorem 4 ) , and hence @xmath671 for all @xmath672 .
consequently , @xmath345 is finite on @xmath654 , and @xmath673 is well - defined on @xmath654 . in order to prove ( [ eq_317 ] ) , let us pick @xmath672 such that @xmath674 .
set @xmath675 . to prove ( [ eq_317 ] ) , it suffices to show that @xmath676 by the definition of @xmath3 . according to the definition of @xmath654
, it is enough to prove that @xmath677 we thus focus on proving ( [ eq_551 ] ) .
we may assume that @xmath678 since otherwise ( [ eq_551 ] ) holds trivially .
by concavity , @xmath679 is finite for @xmath680 , and @xmath681 since @xmath258 is concave , then @xmath265 is non - increasing . consequently , @xmath682 for @xmath683 .
hence , @xmath684 as @xmath584 and @xmath287
. recall that @xmath685 by ( * ? ? ?
* theorem 4 ) and that @xmath686 .
we conclude from ( [ eq_256 ] ) that @xmath687 , and ( [ eq_551 ] ) is proved .
this completes the proof of ( [ eq_317 ] ) . for @xmath79 and
@xmath336 denote @xmath688 .
fix @xmath672 such that @xmath689 .
then for any @xmath690 we have @xmath691 and hence @xmath692 , by ( [ eq_317 ] ) .
consequently , @xmath693 recalling that @xmath694 , we obtain @xmath695 we will also make use of the crude estimate @xmath696 that follows , for example , from lemma [ lem_1010](i ) as @xmath697 . according to ( * ? ? ?
* theorem 4 ) , @xmath698 since @xmath672 .
recall that @xmath699 .
we use ( [ eq_659 ] ) , ( [ eq_659 _ ] ) and ( [ eqn_521 ] ) to conclude that @xmath700 denote @xmath701 .
we have shown that ( [ eq_738 ] ) holds for any @xmath702 .
thus , @xmath703 note that @xmath704 .
corollary [ cor_302 ] and lemma [ lem_1208 ] show that @xmath705 by ( [ eq_335 ] ) and ( [ eq_607 ] ) , @xmath706 since @xmath707 , then according to ( [ eq_607 ] ) and ( [ eq_608 ] ) , @xmath708 dx \leq \hat{c } n^{-\alpha /10}. \label{eq_336}\ ] ] the lemma follows by adding inequalities ( [ eq_335 ] ) and ( [ eq_336 ] ) . @xmath145
lemma [ lem_915 ] allows us to convolve our log - concave function with a small gaussian .
the proof of the next lemma is the most straightforward adaptation of the proof of lemma [ lem_fourier ] .
we sketch the main points of difference between the proofs .
let @xmath287 be an integer , let @xmath584 , and let @xmath33 be an isotropic , log - concave function .
assume that @xmath709 denote @xmath710 , where @xmath221 stands for convolution .
then , @xmath711 [ lem_fourier2 ] _ sketch of proof : _ for @xmath226 with @xmath712 , @xmath713 and consequently @xmath714 , by ( [ eq_934 ] ) and lemma [ lem_1028 ] .
note that @xmath715 ( see , e.g. , @xcite ) .
therefore latexmath:[\[\left| \hat{g}(\xi_1 ) - \hat{g } ( \xi_2 ) \right| \leq r e^{-2 \pi^2 n^{-\alpha } r^2 } e^{-2 \alpha
n \log n } \ \ \ \text{when } \ \ later , we will combine the following proposition with lemma [ lem_dvo ] in order to show that a typical marginal is very close , in the total - variation metric , to a spherically - symmetric concentrated distribution . a random vector @xmath2 in @xmath5
has a spherically - symmetric distribution if @xmath718 for any measurable set @xmath719 and an orthogonal transformation @xmath242 .
there exist universal constants @xmath720 for which the following holds : let @xmath287 be an integer , and let @xmath148 be an isotropic , log - concave function .
let @xmath2 be a random vector in @xmath5 with density @xmath37 .
assume that @xmath721 then there exists a random vector @xmath44 in @xmath5 such that _ proof : _ recall that @xmath724 for some universal constant @xmath725
. we will define two universal constants : @xmath726 \ \ \ \text{and } \ \ \ c_1 = \max \ { 5 \alpha_0 , 2 c_0 \}\ ] ] where @xmath449 is the constant from proposition [ cor_735 ] and @xmath727 is the constant from ( [ eq_1040 ] ) . throughout this proof , @xmath728 and
@xmath727 will stand for the universal constants just mentioned .
we assume that inequality ( [ eq_848 _ _ ] ) the main assumption of this proposition holds , with the constant @xmath729 as was just defined
. we may apply proposition [ cor_735 ] , based on ( [ eq_848 _ _ ] ) , since @xmath730 . by the conclusion of that proposition ,
@xmath731 let @xmath732 be a gaussian random vector in @xmath5 , independent of @xmath2 , with @xmath733 and @xmath734 .
then @xmath735 , and , for example , by lemma [ lem_1010](i ) , we know that @xmath736 consequently , the event @xmath737 holds with probability greater than @xmath738 . by applying ( [ eq_954 ] ) we obtain that for @xmath101 , @xmath739 ( in obtaining the last inequality in ( [ eq_1002 ] ) , one needs to consider separately the cases @xmath740 and @xmath741 ) .
the density of @xmath732 is @xmath742 .
denote by @xmath743 the density of the random vector @xmath744 .
since @xmath745 and @xmath746 , then ( [ eq_848 _ _ ] ) implies the main assumption of lemma [ lem_fourier2 ] for @xmath747 . by the conclusion of that lemma , for all @xmath748 and @xmath749 , @xmath750
denote , for @xmath79 , @xmath751 the spherical average of @xmath116 .
the function @xmath752 is a spherically - symmetric function with @xmath753 , and from ( [ eq_1122 ] ) , @xmath754 according to ( [ eq_1014 ] ) and the case @xmath755 in ( [ eq_1002 ] ) , @xmath756 by the definition of @xmath757 , where @xmath758 for any measurable function @xmath759 .
let @xmath44 be a random variable that is distributed according to the density @xmath752 .
then @xmath44 satisfies the conclusion ( ii ) of the present proposition , since @xmath752 is a radial function .
additionally , ( [ eq_1002 ] ) shows that @xmath44 satisfies ( iii ) , since the random variables @xmath760 and @xmath761 have the same distribution .
it remains to prove ( i ) . to that end , we employ lemma [ lem_915 ] .
the assumptions of lemma [ lem_915 ] are satisfied for @xmath762 , since @xmath763 .
we use ( [ eq_1043 ] ) and the conclusion of lemma [ lem_915 ] to obtain @xmath764 as @xmath765 .
this completes the proof of ( i ) .
@xmath145 let @xmath74 be integers , let @xmath766 , let @xmath528 and let @xmath2 be a random vector in @xmath5 with a spherically - symmetric distribution .
suppose @xmath78 is a @xmath76-dimensional subspace .
assume that for @xmath101 , @xmath767 then , @xmath768 where @xmath94 is a standard gaussian random vector in @xmath82 , and @xmath470 are constants depending only on @xmath420 and @xmath534 .
[ computation ] _ proof : _ in this proof we write @xmath769 etc . to denote various positive constants depending only on @xmath420 and @xmath534 .
we may clearly assume that @xmath770 and @xmath771 , as otherwise the result of the lemma is trivial with @xmath772 .
let @xmath44 be a random vector , independent of @xmath2 , that is distributed uniformly in @xmath53 .
let @xmath94 be a standard gaussian vector in @xmath82 , independent of @xmath2 and @xmath44 .
we will use a quantitative estimate for maxwell s principle by diaconis and freedman @xcite .
according to their bound , @xmath773 for any @xmath123 .
since @xmath2 is independent of @xmath44 and @xmath94 , then also @xmath774 for @xmath123 , the density of @xmath775 is the function @xmath776 .
lemma [ lem_929 ] implies that @xmath777 , for some universal constant @xmath778 .
hence , @xmath779 where we used ( [ eq_317 _ ] ) . note that the random vectors @xmath2 and @xmath780 have the same distribution , since the distribution of @xmath2 is spherically - symmetric . by combining ( [ eq_647 ] ) and ( [ eq_723 ] ) , @xmath781 because @xmath782 .
this completes the proof .
let @xmath14 and @xmath783 be integers , and let @xmath2 be a random vector in @xmath5 with an isotropic , log - concave density .
then there exists a subset @xmath90 with @xmath91 such that for any @xmath92 , @xmath784 where @xmath94 is a standard gaussian random vector in @xmath82 , and @xmath470 are universal constants .
[ thm_405 ] _ proof : _ we use the constant @xmath729 from proposition [ prop_842 ] , and the constant @xmath178 from lemma [ lem_dvo ] .
we begin as in the proof of theorem [ cor_202 ] .
denote the density of @xmath2 by @xmath37 .
set @xmath785 we may assume that @xmath20 exceeds a certain universal constant , hence @xmath507 .
fix a subspace @xmath786 that satisfies @xmath787 denote @xmath137 .
then @xmath116 is log - concave and isotropic , and by combining ( [ eq_1032 ] ) with ( [ proj_ok ] ) from section [ section2 ] , @xmath788 we invoke proposition [ prop_842 ] , for @xmath174 and @xmath116 , based on ( [ eq_1102 ] ) .
recall that @xmath116 is the density of @xmath104 . by the conclusion of proposition [ prop_842 ]
, there exists a random vector @xmath44 in @xmath82 , with a spherically - symmetric distribution , such that @xmath789 and @xmath790 fix @xmath791 , and let @xmath792 be a @xmath76-dimensional subspace . since the distribution of @xmath44 is spherically - symmetric , we may apply lemma [ computation ] for @xmath793 and @xmath794 , based on ( [ eq_1105 ] ) . by the conclusion of that lemma , @xmath795 where @xmath796 is a standard gaussian random vector in @xmath518 .
we combine the above with ( [ eq_1104 ] ) , and obtain @xmath797 ( note that @xmath798 . ) in summary , we have proved that whenever @xmath82 is an @xmath174-dimensional subspace that satisfies ( [ eq_1032 ] ) , then all the @xmath76-dimensional subspaces @xmath792 satisfy ( [ eq_1128 ] ) .
suppose that @xmath799 is a random @xmath174-dimensional subspace .
we will use lemma [ lem_dvo ] , for @xmath800 and @xmath801 . note that @xmath516 , hence @xmath802 , by the definition of @xmath174 above .
therefore we may safely apply lemma [ lem_dvo ] , and conclude that with probability greater than @xmath803 , the subspace @xmath82 satisfies ( [ eq_1032 ] ) .
therefore , with probability greater than @xmath803 of selecting @xmath82 , all @xmath76-dimensional subspaces @xmath792 satisfy ( [ eq_1128 ] ) .
next , we select a random subspace @xmath518 inside the random subspace @xmath82 .
that is , fix @xmath804 , and suppose that @xmath805 is a random subspace , that is distributed uniformly over the grassmannian of @xmath76-dimensional subspaces of @xmath82 .
since @xmath82 is distributed uniformly over @xmath806 , it follows that @xmath518 is distributed uniformly over @xmath75 .
we thus conclude that @xmath518 which is a random , uniformly distributed , @xmath76-dimensional subspace in @xmath5 satisfies ( [ eq_1128 ] ) with probability greater than @xmath803 .
recall that @xmath807 for a universal constant @xmath808 , and that our only assumption about @xmath76 was that @xmath809 .
the theorem is therefore proved .
_ proof of theorem [ thm_multi ] : _ observe that @xmath810 under the assumptions of theorem [ thm_multi ] .
the theorem thus follows from theorem [ thm_405 ] , for an appropriate choice of a universal constant @xmath86 .
@xmath145 an additional notion of distance between multi - dimensional measures is known in the literature under the name of `` @xmath813-distance '' ( see , e.g. , @xcite , @xcite ) . for two random vectors @xmath2 and @xmath44 in a subspace @xmath78 , their @xmath813-distance is defined as @xmath814 the @xmath813-distance between @xmath2 and @xmath44 compares only one - dimensional marginals of @xmath2 and @xmath44 , hence it is weaker than the total - variation distance . the following proposition is proved by directly adapting the arguments of naor and romik @xcite .
let @xmath88 , and assume that @xmath815 is an integer .
suppose that @xmath2 is a random vector in @xmath5 with an isotropic , log - concave density .
let @xmath816 be an integer , and let @xmath168 be a random @xmath76-dimensional subspace . then , with probability greater than @xmath817 of choosing @xmath82 , @xmath818 where @xmath94 is a standard gaussian random vector in the subspace @xmath82 . here
, @xmath103 are universal constants .
[ prop_t ] _ sketch of proof : _ let @xmath819 be the spherical average of @xmath37
. for @xmath820 , set @xmath821 . according to theorem [ cor_202 ] ,
@xmath822 denote @xmath823 and fix @xmath458 .
we apply lemma [ computation ] ( for @xmath824 and @xmath811 ) based on ( [ eqn_953 ] ) , to obtain the inequality @xmath825 valid for any @xmath129 .
let us fix @xmath129 . by proposition [ lem_bobkov ] , the function @xmath826 is @xmath727-lipshitz .
we apply proposition [ dvoretzky ] for @xmath827 and then we use ( [ e_1152 ] ) to conclude that with probability greater than @xmath828 of selecting @xmath82 , @xmath829 here we used the fact that @xmath830 . recall that @xmath831 is arbitrary .
let @xmath832 for @xmath833 where @xmath834 is the inverse function to @xmath614 .
then , with probability greater than @xmath835 of selecting @xmath82 , the estimate ( [ e_101 ] ) holds for all @xmath836 . by using , e.g. , ( * ? ? ?
* lemma 6 ) we see that with probability greater than @xmath837 of selecting @xmath82 , @xmath838 the proposition follows from ( [ e_128 ] ) and the definition of the @xmath813-distance .
@xmath145 at first glance , the estimates in proposition [ prop_t ] seem surprisingly good : marginals of almost - proportional dimension are allegedly close to gaussian . the problem with proposition [ prop_t ] hides , first , in the requirement that @xmath839 , and second , in the use of the rather weak @xmath813-distance .
bastero , j. , bernus , j. : _ asymptotic behavior of averages of @xmath76-dimensional marginals of measures on @xmath5 .
_ preprint .
available at + http://www.unizar.es/galdeano/preprints/2005/preprint34.pdf bobkov , s. g. , koldobsky , a. : _ on the central limit property of convex bodies_. geometric aspects of functional analysis , israel seminar ( 200102 ) , lecture notes in math .
, 1807 , springer , berlin , ( 2003 ) , 4452 .
davidovi , ju .
s. , korenbljum , b. i. , hacet , b. i. : _ a certain property of logarithmically concave functions .
_ ( russian ) dokl .
nauk sssr , 185 , ( 1969 ) , 12151218 .
english translation in soviet math .
, 10 , ( 1969 ) , 477480 .
gordon , y. : _ on milman s inequality and random subspaces which escape through a mesh in @xmath5 .
_ geometric aspects of functional analysis , israel seminar ( 198687 ) , lecture notes in math . , 1317 ,
springer , berlin , ( 1988 ) , 84106 .
johnson , w. b. , lindenstrauss , j. : _ extensions of lipschitz mappings into a hilbert space .
_ conference in modern analysis and probability ( new haven , conn . , 1982 ) , contemp .
, 26 , amer .
soc . , providence , ri , ( 1984 ) , 189206 .
koldobsky , a. , lifshits , m. : _ average volume of sections of star bodies . _ geometric aspects of functional analysis , israel seminar ( 199600 ) , lecture notes in math .
, 1745 , springer , berlin , ( 2000 ) , 119146 .
ledoux , m. : _ spectral gap , logarithmic sobolev constant , and geometric bounds .
_ appears in eigenvalues of laplacians and other geometric operators , surveys in differential geometry 9 , somerville , ma , ( 2004 ) , 219240 .
lovsz , l. , vempala , s. : _ the geometry of logconcave functions and sampling algorithms .
_ to appear in random structures & algorithms .
available at + http://www-math.mit.edu/`~`vempala/papers/logcon.pdf milman , v. d. : _ a new proof of a. dvoretzky s theorem on cross - sections of convex bodies . _
( russian ) funkcional .
i priloen . , 5 , no .
4 , ( 1971 ) , 2837 .
english translation in funct .
appl . , 5 , ( 1971 ) , 288295 .
milman , v. d. , pajor , a. : _ isotropic position and inertia ellipsoids and zonoids of the unit ball of a normed @xmath20-dimensional space .
_ geometric aspects of functional analysis , israel seminar ( 198788 ) , lecture notes in math . ,
1376 , springer , berlin , ( 1989 ) , 64104 .
schechtman , g. : _ a remark concerning the dependence on @xmath842 in dvoretzky s theorem .
_ geometric aspects of functional analysis , israel seminar ( 198788 ) , lecture notes in math . ,
1376 , springer , berlin , ( 1989 ) , 274277 .
sodin , s. : _ tail - sensitive gaussian asymptotics for marginals of concentrated measures in high dimension .
_ to appear in geometric aspects of functional analysis , israel seminar , lecture notes in math .
available at http://arxiv.org/abs/math.mg/0501382 sudakov , v. n. : _ typical distributions of linear functionals in finite - dimensional spaces of high - dimension .
_ ( russian ) dokl .
sssr , 243 , no . 6 , ( 1978 ) , 14021405 .
english translation in soviet math .
dokl . , 19 , ( 1978 ) , 15781582 . | we show that there exists a sequence @xmath0 for which the following holds : let @xmath1 be a compact , convex set with a non - empty interior .
let @xmath2 be a random vector that is distributed uniformly in @xmath3 .
then there exist a unit vector @xmath4 in @xmath5 , @xmath6 and @xmath7 such that @xmath8 where the supremum runs over all measurable sets @xmath9 , and where @xmath10 denotes the usual scalar product in @xmath5 . furthermore
, under the additional assumptions that the expectation of @xmath2 is zero and that the covariance matrix of @xmath2 is the identity matrix , we may assert that most unit vectors @xmath4 satisfy ( @xmath11 ) , with @xmath12 and @xmath13 .
corresponding principles also hold for multi - dimensional marginal distributions of convex sets . |
particle acceleration at collisionless shocks is believed to be a common phenomenon in space plasmas in a diversity of environments , ranging from the inner heliosphere to the central regions of distant galaxies . in the heliosphere ,
evidence to support the belief that such a mechanism can efficiently produce non - thermal particles includes direct measurements of accelerated populations in various energy ranges at the earth s bow shock ( e.g. scholer et al .
1980 ; mbius et al .
1987 ; gosling et al . 1989 ) and interplanetary shocks ( for the pre - ulysses era see , for example , sarris and van allen 1974 ; gosling et al . 1981 ; decker , pesses and krimigis , 1981 ; tan et al .
the motivations for developing theories of shock acceleration are therefore obvious , and a variety of approaches have emerged .
one possible means for the generation of non - thermal particles is the fermi mechanism , often called diffusive shock acceleration ; it is this process that is the focus of this paper .
there are various approaches to modelling diffusive shock acceleration . among these
are convection - diffusion differential equation approaches ( e.g. kang and jones 1995 ) , and the kinematic monte carlo technique of ellison and jones ( e.g. , see jones and ellison , 1991 , and references therein ) , which describes the injection and acceleration of particles from thermal energies ; both of these can address spectral and hydrodynamic properties .
this latter simulational approach is the central tool for the analysis here , and is ideally suited to the interpretation of the modest time - resolution shock data acquired by ulysses . hybrid and full plasma codes ( e.g. quest 1988 ; burgess 1989 , winske et al .
1990 ; trattner and scholer 1991 ; giacalone , burgess and schwartz 1992 ; liewer , goldstein , and omidi 1993 ; kucharek and scholer 1995 ) provide contrasting probes of shock environs , with an emphasis primarily on plasma structure and wave properties in the environs of shocks
. comparisons of predictions from theoretical models with observed phase - space distribution data are informative probes .
the first detailed theory / data comparison along these lines was performed by ellison , mbius and paschmann ( 1990 ) in the case of the quasi - parallel portion of the earth s bow shock , comparing predictions of the monte carlo method with particle distributions of protons , and a @xmath0 , @xmath1 and @xmath2 ion mix obtained by the ampte experiment .
the agreement between model predictions and data was impressive .
ellison , mbius and paschmann ( 1990 ) concluded that a successful fit was possible only in the non - linear acceleration regime , when the dynamic effects of the accelerated particles are crucial to the determination of the shock structure . since this pioneering work , successful comparisons of other theoretical techniques with data from the earth s bow shock have been performed .
these include the hybrid plasma simulations of trattner and scholer ( 1991 ) and giacalone et al .
( 1993 ) , and solutions to the convection - diffusion equation ( kang and jones 1995 ) , both of which have yielded good agreement with both the bow shock data and the monte carlo technique ( ellison et al . 1993 ) .
such a comparison between theory and experiment was extended to the domain of interplanetary shocks in the work of baring , ogilvie , ellison and forsyth ( 1997 , hereafter boef97 ; see also kang & jones 1997 , for application of their convection - diffusion equation technique ) . in this development
, impressive agreement was found between the monte carlo predictions and spectral data obtained by the solar wind ion composition spectrometer ( swics ) aboard ulysses , in the case of two shocks observed early in the ulysses mission .
such agreement was possible only with the assumption of strong particle scattering ( i.e. near the bohm diffusion limit ) in the highly oblique candidate shocks . for a third shock , detected a month later , the comparison failed , with significant differences arising in the 500 - 800 km / sec range of the phase space distribution .
boef97 attributed this discrepancy to the omission of pick - up ions from the model : such an extra component would be expected to provide a substantial contribution to the accelerated population in this particular event .
this paper addresses the role of pick - up ions in such shocks via modeling the accelerated population for the specific interplanetary shock observed on day 292 of 1991 , detected by the swics and hi - scale instruments aboard the ulysses spacecraft at around 4.5 au , as reported in gloeckler et al .
phase space distributions from the simulations are compared with swics and hi - scale data , yielding acceptable fits for the and populations , using standard prescriptions for the injected pick - up ion distribution , by adjusting a single turbulence parameter . using this same ,
the simulation results also successfully account for the observation of energetic protons farther upstream of the forward shock than lower energy pick - up protons , since a rigidity - dependent diffusion is used in the models .
the monte carlo simulation technique used here has been invoked in previous applications to shocks in the heliosphere , as discussed above , and is extensively described in a number of papers ( ellison , jones , & eichler 1981 ; ellison , jones & reynolds 1990 ; jones & ellison 1991 ; baring , ellison & jones 1994 ; ellison , baring , & jones 1996 ) .
it follows closely bell s ( 1978 ) test particle approach to diffusive acceleration .
particles are injected upstream and allowed to convect into the shock , meanwhile diffusing in space so as to effect multiple shock crossings , and thereby gain energy through the shock drift and fermi processes . for the interplanetary shock that is the focus of this paper , throughout and shall denote the upstream and downstream flow speed in the shock rest frame , respectively .
the particles gyrate in a laminar electromagnetic field .
their trajectories are obtained by solving a fully relativistic lorentz force equation in the shock rest frame , in which there is , in general , a * u @xmath3 b * electric field in addition to the magnetic field .
the effects of magnetic turbulence are modeled by scattering these ions in the rest frame of the local fluid flow .
these collisions , effectively mimicking diffusive transport instigated by alfvn waves , are assumed to be elastic in the local fluid frame , an assumption that is valid so long as the flow speed far exceeds the alfvn velocity .
otherwise , the scattering centers are not anchored in the fluid frame , and the collisions are inelastic , thereby introducing significant second - order fermi ( stochastic ) acceleration . for interplanetary shocks in general , and in particular for the day 292 , 1991 shock considered here , the elastic scattering assumption is somewhat violated due to a low shock alfvenic mach number . yet the stochastic acceleration contribution is expected to be relatively small , perhaps at most of the order of 25% , so its inclusion is deferred to later work .
the simulation can routinely model either large - angle or small - angle scattering . in this paper ,
large - angle scattering is employed , motivated by previous observations ( e.g. , hoppe et al .
1981 ; balogh et al . 1993 ) indicating very turbulent fields in ip shocks . between scatterings ,
particles are allowed to travel a distance in their local fluid frame that is exponentially distributed about the scattering mean free path , with : @xmath4 here is the gyroradius of an ion of momentum , mass , and charge .
also is the gyroradius of an ion with a speed equal to the far upstream flow speed normal to the shock plane , , where denotes the direction normal to the shock plane .
the mean free path scale is set proportional to with constant of proportionality such that .
following previous monte carlo work , for simplicity we set , a specialization that is appropriate for interplanetary plasma shocks ( e.g. , see ellison et al .
1990 ; mason , gloeckler & hovestadt 1983 ; giacalone , burgess & schwartz 1992 for discussions on the micro - physical expectations for ) , yet is easily generalizable in the simulation . since for physically meaningful diffusion resulting from gyro - resonant wave - particle interactions , the case is also motivated on fundamental grounds .
cross - field diffusion emerges naturally from the simulation , since at every scattering , the direction of the particle s momentum vector is randomized in the local fluid frame , with the resulting effect that the gyrocenter of a particle is shifted randomly by a distance of the order of one gyroradius in the plane orthogonal to the local field .
transport perpendicular to the field is then governed by a kinetic theory description , so that the ratio of the spatial diffusion coefficients parallel ( ) and perpendicular ( ) to the mean magnetic field is given by ( see forman , jokipii & owens 1974 ; ellison , baring & jones 1995 , for detailed expositions ) .
hence , couples directly to the amount of cross - field diffusion , and is a measure not only of the frequency of collisions between particles and waves , but also of the level of turbulence present in the system , i.e. is an indicator of .
note that is the bohm diffusion limit of quasi - isotropic diffusion , presumably corresponding to . as will become evident in the section [ sec : model ] , by virtue of its connection to cross - field diffusion , plays an important role in determining the injection efficiency of low energy particles .
we note that other cross - field transport effects such as that incurred by so - called field - line wandering ( e.g. see giacalone & jokipii 1999 ) can be incorporated parametrically in the monte carlo technique , and will be addressed in future work .
while and serve as parameters here , in principal they can be calculated for a given wave field .
however , in practice , precision is limited by _ in situ _
magnetometer data , and exact determination of 3-dimensional particle diffusion properties from a 1-dimensional field data stream is impossible .
yet insights can be gained by using established theoretical formalism , such as random fluctuation theory ( e.g. jokipii & coleman 1968 ) , or equivalently quasi - linear diffusion theory ( see , e.g. , kulsrud & pierce 1969 ; jones , birmingham , & kaiser 1978 ) , to estimate components of the diffusion tensor along and orthogonal to the mean field .
these components can be expressed in terms of integrals of components of the power spectral tensor for field fluctuations .
such slab models for turbulence can be extended to consider higher - dimensional field fluctuations ( e.g. see bieber et al .
guidance for the behavior and values of and can be gleaned from such investigations , and an analysis along these lines to facilitate the interpretation of ulysses ip shock spectral data will be the subject of future work .
the simulation technique makes no distinction between accelerated particles and thermal ones . therefore the injection efficiency is determined solely by the a particle s ability to diffuse back upstream after it has encountered the shock for the first time . in the case of the earth s bow shock ( ellison et al .
1990 ) , this permits the development of a nonlinear model that includes the effects of the accelerated particles on the dynamics of the shock .
however , in the relatively weak ip shock considered here , a steep power - law is present , and the poor injection efficiency reduces the number density of energetic particles , eliminating nonlinear hydrodynamic effects .
upstream plasma quantities are input from observational data , as discussed below , and downstream quantities are determined using the full mhd rankine - hugoniot relations .
the simulation is capable of producing energy spectra / fluxes , and hence phase space distributions , at any location upstream or downstream of the shock and in any reference frame including that of a spacecraft .
this makes the simulation ideal for comparison with observational data .
our case study focuses on the forward shock of a cir encountered by ulysses on day 292 of 1991 , for which downstream particle distributions were obtained for comparison with data published in gloeckler et al .
( 1994 ) . to determine parameters for this shock that are appropriate input for the monte carlo simulation , the observational results detailed in gloeckler et al .
( 1994 ) are used along with the data compilations of balogh et al .
( 1995 ) , gonzlez - esparza et al .
( 1996 ) , and hoang et al .
these sources indicate that the sonic mach number of the shock was 2.53 , with an unstated uncertainty .
this controls the temperature of the thermal solar wind protons , but does receive contributions from electrons and alpha particles .
other parameters include the angle the upstream magnetic field makes with the shock normal , nt , and magnetic compression ratio . here and are the _ total _ magnetic field magnitudes upstream and downstream of the shock , respectively .
in addition , the normalization of solar wind distributions was established using @xmath5 as the solar wind proton density , and @xmath5 as that for solar wind ( within the uncertainties of values listed table 1 in gloeckler et al .
1994 ) .
although fluctuations in the plasma data preclude precise determination of the velocity compression ratio of the shock , gloeckler et al . inferred a value of .
this inference is somewhat inconsistent with the observed power - law behavior of the phase space density at high energies , which suggests a compression ratio nearer 2 .
the choice of compression ratio is controlled by the shock mach number , receiving contributions from sonic and alfvnic mach numbers , and respectively , which can be written in the form ( e.g. baring et al .
1997 ) : @xmath6 where is the shock speed in units of 100 km / sec , and are the proton and electron temperatures in units of k , is the upstream field strength in units of gauss , and is the proton density in . using upstream values of and , as tabulated in hoang et al .
( 1995 ) , we find that for km / sec , . adding in the magnetic field gives . to yield a compression ratio for adiabatic index
, one must have a slightly lower value of ( i.e. perhaps higher field ) . in this paper
, we adopted a value of , which lies within the margin of error of gloeckler et al.s inferred value , to generate power - law slopes commensurate with the data .
due to significant observational uncertainties , ranges of parameters are permissible . in the simulation runs , the temperature of the particles
was chosen to be four times that of the solar wind protons .
no guidance for this assumed upstream temperature was provided in gloeckler et al .
( 1994 ) , nor in the aforementioned ulysses data compilation papers .
however , in the ip shock theory / data comparison of baring et al .
( 1997 ) , for the one shock with good measurements of the distribution , namely the 91097 event , the upstream particle thermal velocity was comparable to ( actually slightly greater ) than that of the solar wind protons , implying . adopting the same temperature ratio in this work was motivated by this precedent , and the expectation that the adiabatic expansion of the solar wind from 2.7au to 4.5au should roughly preserve the ratios of the temperatures per nucleon for these two ions .
note that the correction to the sonic mach number in eq .
( [ eq : machnos ] ) incurred by solar wind helium is then approximately 4% . to facilitate transformation of simulation results measured in the shock rest frame to the approximate spacecraft frame , so as to compare directly with gloeckler et al.s published distributions , a value for the solar wind plasma speed of km / s was assumed _ after _ the shock encounter .
this identifies the downstream fluid frame .
no values for the shock speed were listed in gloeckler et al .
( 1994 ) , as is common practice in other expositions ( e.g. see burton et al . 1992 ; baring et al .
hence selection of a value for consistent with other data became necessary . from the long - term solar wind radial velocity traces in figure 2 of gonzlez - esparza
( 1996 ) we can set a lower bound on the solar wind speed of km / s prior to the shock passage , i.e. upstream . assuming collinear velocities , i.e. that the solar wind is not deflected by the shock on large scales , we have the speed relations and in addition to , where is the unknown shock speed in the spacecraft frame .
these solve to yield km / s and km / s for for km / s ; higher values of and would be realized for smaller choices of .
the other input for the monte carlo simulation is that for the incoming pick - up ion distributions .
gloeckler et al . (
1994 ) used a simplified form for these distributions ; here we adopt the slightly more developed expression used in ellison , jones & baring ( 1999 ) that is modeled on the seminal work of vasyliunas & siscoe ( 1976 ) .
this form incorporates the gravitational focusing of interstellar neutrals , the physics of their ionization as a function of distance from the sun ( approximately at solar minimum ) , and adiabatic losses incurred during propagation away from the sun , and is similar in conception to pick - up ion distributions used in le roux , potgieter & ptuskin ( 1996 ) .
our pick - up ion model therefore provides both the detailed shape and normalization of these superthermal distributions , which are distinctly different for and , largely due to the patently different ionization and charge exchange rates for their corresponding interstellar neutrals .
[ fig : distrib ] downstream distributions for thermal and accelerated ions from the monte carlo simulation are compared with swics and hi - scale measurements taken in the frame of the spacecraft on the downstream side of the day 292 , 1991 shock in fig .
[ fig : distrib ]
. the ulysses data are those exhibited in fig . 1 of gloeckler et al .
( 1994 ) , and the simulation data are transformed to the spacecraft frame as described above . the solar wind and pick - up ion parameters and abundances
are fairly tightly specified , so that the theoretical model has one largely free parameter , the ratio of the particle mean free path to its gyroradius , . as noted in ellison , baring & jones ( 1995 ) and baring et al .
( 1997 ) , the efficiency of acceleration of thermal ions in oblique shocks , i.e. the normalization of the non - thermal power - law , is sensitive to the choice of , so this parameter was adjusted to obtain a reasonable `` fit '' to the ulysses data .
ellison , jones & baring ( 1999 ) note that for their application to the solar wind termination shock , the efficiency is not as sensitive to when superthermal pick - up ions enter the problem , a property that we also replicate here . here , for
, the accelerated pick - up ion phase space density is orders of magnitude above that of the thermal ions , for both and . in fig .
[ fig : distrib ] , the theory / data comparison is overall not quite as good as the ones in baring et al .
( 1997 ) for ulysses ip shock data at around 23 au , where pick - up ions are less of a factor .
yet the fits here do model the accelerated pick - up ions very well , for , a value that is slightly higher than those inferred in the fits of baring et al .
( 1997 ) , but is still consistent with a moderate level of field turbulence ( i.e. here ) .
note that since the inferred value of for and data is most sensitive to the assumed shock obliquity ( among other input parameters ) , the observational uncertainty in maps over to an uncertainty of around in the value of .
there are clear differences between the simulation results and the observations .
the distribution of thermal solar wind ions appears slightly narrower than the published observations .
more notable though is the fact that the accelerated thermal ions are injected somewhat less efficiently in the simulation than in the observations , an inefficiency characteristic of highly oblique shocks that is present also for the _ solar wind _ protons , though not explicitly displayed in fig .
[ fig : distrib ] . the efficiency of acceleration of thermal ions could be increased via several means : ( i ) by lowering the shock obliquity angle , for which there is a large observational uncertainty ; ( ii ) by decreasing , corresponding to increased turbulence , without altering the pick - up ion acceleration efficiencies substantially , and ( iii ) increasing the temperature of the thermal ions somewhat , though this would reduce the compression ratio and accordingly steepen the non - thermal continuum .
however , such parameter adjustments only incur small changes to the widths of the thermal distributions .
trial runs indicate that an agreeable fit can be obtained to the entire particle distribution simply by reducing to , without changing other parameters .
it is not clear what plasma characteristic would establish a species - dependent , a circumstance that contrasts the inference of in the 91097 event analyzed by baring et al .
( 1997 ) .
an instructive diagnostic on the acceleration model is to probe the spatial scale of diffusion .
this is performed by examining upstream distributions of high energy particles . in their 1994 paper , gloeckler et al .
discussed an energy - dependent rise in fluxes of non - thermal particles _ prior _ to the shock crossing .
this was cited as indicating the existence of a pre - acceleration mechanism .
the monte carlo simulation was run with flux measurement planes placed upstream of the shock at different distances , as well as downstream .
this enabled determination of the number fluxes of a particular energy ion at different positions upstream of the shock .
this was converted to observer s times using the approximate speed of the ulysses spacecraft in the solar wind .
results for two different ion energies are displayed in fig . 2 , together with corresponding data from fig . 3 of gloeckler et al .
( 1994 ) for identical energy windows . was chosen by gloeckler et al . since good spatial resolution at 5 kev was possible .
observe that the y - axis in fig .
2 represents ratios of flux to time - averaged flux at the respective energies , so that the normalization does not explicitly exhibit the property that the absolute flux at 200 kev is always much less than that at 5 kev when downstream of the shock .
note also that the ulysses data normalization was established by averaging over 3 days of accumulated data , whereas the model normalization was adjusted to match observed fluxes around 1/2 day downstream of the shock .
the essential result of this comparison is that the spatial scale of the exponential decline of ions upstream of the shock is more or less identical to that of the model , for our choice of .
theoretically , high energy particles with a mean free path according to eq .
( [ eq : mfp ] ) establish an exponential dilution in space / time due to random scattering of the particles seeding upstream leakage against a convective flow . from standard diffusion - convection theory ,
the spatial scale for this dilution is for a shock of obliquity , where is the component of the spatial diffusion coefficient along the mean field , , and is prescribed by eq .
( [ eq : mfp ] ) .
hence , this upstream dilution scale is proportional to the proton s energy .
for the 200 kev ions , the simulation results are clearly well correlated with the data prior to the shock , modulo plasma fluctuations . on the other hand , for the lower energy 5kev ions ,
the exponential decay has a very short time scale , a factor of 40 smaller than for the 200 kev ions , and drops to background levels very quickly .
so , although the simulation results are consistent with the observed results , it is impossible to draw more definitive conclusions without an improvement in data time resolution .
the 200 kev ions , however , have much longer mean free paths and travel much farther upstream , and are well correlated .
the comparison of phase space distributions from the monte carlo simulation of diffusive shock acceleration with those observed by swics and hi - scale in the day 292 , 1991 event reveals a good deal of consistency between theory and experiment for both the proton and spectra above speeds around 600 km / sec . at these speeds
, the injection of pick - up protons dominates that of solar wind protons , and the acceleration of solar wind is inefficient relative to that of pick - up .
the characteristic of the shock most relevant for these high energy components is its compression ratio : to match the observed steep spectra , is required , on the low end of the range quoted by gloeckler et al .
( 1994 ) .
the normalization of the energetic ion power - laws is best fit with a model `` turbulence '' parameter , where , corresponding to a ratio of components of the spatial diffusion tensor . for such values , cross - field diffusion is insufficient for the monte carlo model to account for observed accelerated , which requires so that the bohm limit is approximately realized .
while the high energy protons and are modeled fairly well , there are significant discrepancies in the high velocity wings of the thermal protons .
these deviations are slightly more marked than those found by baring et al .
( 1997 ) in ip shocks without significant pick - up ion components .
clearly some non - diffusive element of heating may be present in the plasma shock that is not incorporated in the simulation .
to explore such possibilities , it is planned to incorporate cross - shock and shock layer charge separation potentials in the monte carlo code in future extensions , to effect a more accurate modeling of the shock structure .
it must be noted , though , that uncertainties in the shock parameters significantly impact inferences of discrepancies between theory and experiment . the flux increases of energetic protons seen upstream of the shock are quite consistent with the expected upstream leakage associated with diffusive shock acceleration .
the value of inferred from the spectral fit scales the upstream diffusive lengthscale , and the accompanying exponential decline in predicted flux is commensurate with the ulysses data presented in gloeckler et al .
hence , the observed upstream flux precursor is not clear evidence of a pre - acceleration mechanism , as claimed by gloeckler et al .
, though it is quite possible that some pre - acceleration mechanism may be acting .
a most enticing feature of this work is that a single model parameter can couple injection efficiency to the spatial scales of ions upstream of the shock .
whether or not this identifies diffusion as the dominant operating mechanism remains to be seen , but it is clear that processes acting on the scales of a few gyroradii control both injection and upstream transport in this interplanetary shock .
-5pt balogh , a. , forsyth , r. j. , ahuja , a. , et al .
the interplanetary magnetic field from 1 to 5 au - ulysses observations .
@xmath713 , ( 6 ) 1524 .
balogh , a. , gonzlez - esparza , j. a. , forsyth , r. j. , et al .
interplanetary shock waves : ulysses observations in and out of the ecliptic plane .
@xmath772 , 171180 .
1995 . baring , m. g. , ellison , d. c. & jones , f. c. monte carlo simulations of particle acceleration at oblique shocks .
@xmath790 , 547552 .
1994 . baring , m. g. , ogilvie , k. w. , ellison , d. c. , & forsyth , r. j. acceleration of solar wind ions by nearby interplanetary shocks : comparison of monte carlo simulations with ulysses observations .
@xmath7476 , 889902 .
bell , a. r. the acceleration of cosmic rays in shock fronts .
i. @xmath7182 , 147156 .
bieber , j. w. , matthaeus , w. h. , smith , c. w. , et al .
proton and electron mean free paths : the palmer consensus revisited .
@xmath7420 , 294306 .
burgess , d. alpha particles in field - aligned beams upstream of the bow shock - simulations .
@xmath716 , 163166 .
burton , m. e. , smith , e. j. , goldstein , b. e. , et al .
ulysses - interplanetary shocks between 1 and 4 au .
@xmath719 , 1,2871,289 . 1992 .
decker , r. b. , pesses , m. e. , & krimigis , s. m. shock - associated low - energy ion enhancements observed by voyagers 1 and 2 .
@xmath786 , 8,8198,831 .
ellison , d. c. , baring , m. g. , & jones , f. c. acceleration rates and injection efficiencies in oblique shocks .
@xmath7453 , 873882 .
ellison , d. c. , baring , m. g. & jones , f. c. nonlinear particle acceleration in oblique shocks .
@xmath7473 , 1,0291,050 .
ellison , d. c. , giacalone , j. , burgess , d. , & schwartz , s. j. simulations of particle acceleration in parallel shocks : direct comparison between monte carlo and one - dimensional hybrid codes .
@xmath798 , 21,08521,093 .
ellison , d. c. , jones , f. c. & baring , m. g. direct acceleration of pickup ions at the solar wind termination shock : the production of anomalous cosmic rays .
@xmath7512 , 403416 .
ellison , d. c. , jones , f. c. & eichler , d. monte carlo simulation of collisionless shocks showing preferential acceleration of high a / z particles .
j. geophys . -
zeitschrift fuer geophysik 50 , 110113 .
ellison , d. c. , jones , f. c. & reynolds , s. p. first - order fermi particle acceleration by relativistic shocks .
@xmath7360 , 702714 . 1990 .
ellison , d. c. , mbius , e. , & paschmann , g. particle injection and acceleration at earth s bow shock - comparison of upstream and downstream events .
@xmath7352 , 376394 .
forman , m. a. , jokipii , j. r. & owens , a. j. cosmic - ray streaming perpendicular to the mean magnetic field .
@xmath7192 , 535540 .
giacalone , j. , burgess , d. , & schwartz , s. j. ion acceleration at parallel shocks : self - consistent plasma simulations , in esa , study of the solar - terrestrial system , 6570 .
giacalone , j. , burgess , d. , schwartz , s. j. & ellison , d. c. ion injection and acceleration at parallel shocks - comparisons of self - consistent plasma simulations with existing theories .
@xmath7402 , 550559 . 1993 .
giacalone , j. & jokipii , j. r.the transport of cosmic rays across a turbulent magnetic field .
@xmath7520 , 204214 .
gloeckler , g. , geiss , j. , roelof , e. c. , et al .
acceleration of interstellar pickup ions in the disturbed solar wind observed on ulysses .
@xmath799 , 17,63717,643 . 1994 .
gonzlez - esparza , j. a. , balogh , a. , forsyth , r. j. , et al .
interplanetary shock waves and large - scale structures : ulysses observations in and out of the ecliptic plane .
@xmath7101 , 17,05717,072 . 1996 .
gosling , j. t. , asbridge , j. r. , bame , s. j. , et al .
interplanetary ions during an energetic storm particle event - the distribution function from solar wind thermal energies to 1.6 mev .
@xmath786 , 547554 .
gosling , j. t. , thomsen , m. f. , bame , s. j. , & russell , c. t. on the source of diffuse , suprathermal ions observed in the vicinity of the earth s bow shock . @xmath794 , 3,5553,563 . 1989 .
hoang , s. , lacombe , c. , mangeney , a. , et al .
interplanetary shocks observed by ulysses in the ecliptic plane as a function of the heliocentric distance .
@xmath715 ( 8/9 ) , 371374 .
hoppe , m. m. , russell , c. t. , frank , l. a. , et al .
upstream hydromagnetic waves and their association with backstreaming ion populations - isee 1 and 2 observations .
@xmath786 , 4,4714,492 .
jokipii , j. r. & coleman , p. j. cosmic - ray diffusion tensor and its variation observed with mariner 4 .
@xmath773 , 5,4955,501 .
jones , f. c. , birmingham , t. j. , & kaiser , t. b. partially averaged field approach to cosmic ray diffusion .
fluids @xmath721 , 347360 .
jones , f. c. and ellison , d. c. the plasma physics of shock acceleration .
@xmath758 , 259346 .
kang , h. , & jones , t. w. diffusive shock acceleration simulations : comparison with particle methods and bow shock measurements . @xmath7447 , 944961 . 1995 .
kang , h. , & jones , t. w. diffusive shock acceleration in oblique magnetohydrodynamic shock : comparison with monte carlo methods and observations .
@xmath7476 , 875888 .
kucharek , h. & scholer , m. injection and acceleration of interstellar pickup ions at the heliospheric termination shock . @xmath7100 , 1,7451,754 .
kulsrud , r. & pierce , w. p. the effect of wave - particle interactions on the propagation of cosmic rays .
@xmath7156 , 445469 .
le roux , j. a. , potgieter , m. s. , & ptuskin , v. s. a transport model for the diffusive shock acceleration and modulation of anomalous cosmic rays in the heliosphere .
@xmath7101 , 4,7914,804 .
liewer , p. c. , goldstein , b. e. , & omidi , n. hybrid simulations of the effects of interstellar pickup hydrogen on the solar wind termination shock . @xmath798 , 15,21115,220 .
mason , g. m. , gloeckler , g. , & hovestadt , d. temporal variations of nucleonic abundances in solar flare energetic particle events .
i - well - connected events . @xmath7267 , 844862
mbius , e. , scholer , m. , sckopke , n. , et al .
the distribution function of diffuse ions and the magnetic field power spectrum upstream of earth s bow shock . @xmath714 , 681684 .
quest , k. b. theory and simulation of collisionles parallel shocks .
@xmath793 , 9,6499,680 .
sarris , e. t. , & van allen , j. a. effects of interplanetary shock waves on energetic charged particles .
@xmath779 , 4,1574,173 .
scholer , m. , hovestadt , d. , ipavich , f. m. , & gloeckler , g. conditions for acceleration of energetic ions greater than 30 kev associated with the earth s bow shock . @xmath785 , 4,6024,606 .
tan , l. c. , mason , g. m. , gloeckler , g. , & ipavich , f. m. downstream energetic proton and alpha particles during quasi - parallel interplanetary shock events . @xmath793 , 7,2257,243 .
trattner , k. j. , & scholer , m. diffuse alpha particles upstream of simulated quasi - parallel supercritical collisionless shocks .
@xmath718 , 1,8171,820 .
vasyliunas , v. m. , & siscoe , g. l. on the flux and the energy spectrum of interstellar ions in the solar system .
@xmath781 , 1,2471,252 .
winske , d. , thomas , v. a. , omidi , n. , & quest , k. b. re - forming supercritical quasi - parallel shocks .
ii - mechanism for wave generation and front re - formation .
@xmath795 , 18,82118,832 . 1990 . | the acceleration of interstellar pick - up ions as well as solar wind species has been observed at a multitude of interplanetary ( ip ) shocks by different spacecraft .
the efficiency of injection of the pick - up ion component differs from that of the solar wind , and is expected to be strongly enhanced at highly oblique and quasi - perpendicular shock events , in accord with inferences from _ in situ _ observations .
this paper explores theoretical modeling of the phase space distributions of accelerated ions obtained by the ulysses mission for the day 292 , 1991 shock associated with a corotating interaction region , encountered before ulysses fly - by of jupiter .
a monte carlo simulation is used to model the acceleration process , adapting a technique that has been successfully tested on earlier ip shocks possessing minimal pick - up ion presence .
phase space distributions from the simulation technique for various low mass ions are compared with swics and hi - scale data to deduce values of a `` turbulence parameter '' that controls the efficiency of injection , and the degree of cross - field diffusion .
acceptable fits are obtained for the and populations using standard prescriptions for the pick - up ion distribution ; spectral data was only fit well for scenarios very close to the bohm diffusion limit .
it is also found that the simulation successfully accounts for the observation of energetic protons farther upstream of the forward shock than lower energy pick - up protons , using the same turbulence parameter that is required to achieve reasonable spectral fits .
interplanetary shocks , co - rotating interaction regions , ulysses mission , diffusive shock acceleration , shock turbulence |
omega centauri , the most massive globular cluster in the milky way , is unique in terms of its metallicity content , internal kinematics , and structure . unlike other galactic globular clusters , @xmath0 cen shows a wide spread in metallicity ( e.g. norris , freeman , & mighell 1996 ) , with a main metal - poor component at [ fe / h]@xmath5 , a second smaller peak at [ fe / h]@xmath6 , and a long tail extending up to [ fe / h]@xmath7 .
the metal - rich population holds a low velocity dispersion and no sign of rotation , in contrast to the rotating metal - poor population .
furthermore , the metal - rich stars in @xmath0 cen are largely enhanced in @xmath8-process elements relative to those in globular clusters and field stars with similar metallicities ( e.g. norris & da costa 1995 ) , thereby suggesting that the ejecta from low - mass , asymptotic giant branch ( agb ) stars had to be retained and incorporated into the next - generation stars . in spite of its large mass ( @xmath9 m@xmath10 ) , it has been demonstrated by gnedin et al .
( 2002 ) that @xmath0 cen is not unique in its ability to retain the agb ejecta as found for other clusters .
an isolated formation of @xmath0 cen is thus unlikely , because the enriched gas would easily be lost by encountering the galactic disk .
the most viable explanation for the uniqueness of @xmath0 cen is that it was once the dense nucleus of a dwarf galaxy ( freeman 1993 ) . a gravitational potential provided by progenitor s stellar system and dark matter (
as suggested from dwarf spheroidal galaxies in the local group , mateo 1998 ) would help retaining the enriched gas and let the cluster being self - enriched at least over a few gigayears .
if this hypothesis is the case for the origin of @xmath0 cen , the question arises : _ where and in what form does the stellar system of its progenitor galaxy remain ? _ dinescu ( 2002 ) first investigated this issue , by examining the possible signature of the progenitor s tidal debris among nearby metal - poor stars in the catalog of beers et al .
( 2000 , b00 ) .
she identified a group of stars with @xmath11[fe / h]@xmath12 , which departs from the characteristics of the inner galactic halo but has retrograde orbits similar to @xmath0 cen .
her simplified disruption model of the progenitor galaxy demonstrated that trailing tidal debris , having orbital characteristics similar to the cluster , can be found in the solar neighborhood , although the concrete spatial distribution and kinematics of the debris stars remain yet unclear .
this work motivates us to conduct an n - body simulation for the tidal disruption of @xmath0 cen s progenitor galaxy , to obtain the characteristic structure and kinematics of its debris stars and compare with various observations showing signatures of recent merging events in the milky way ( gilmore , wyse , & norris 2002 , gwn ; kinman et al .
2002 , k02 ; chiba & beers 2000 , cb ) .
in particular , we show that a recently identified stream of stars at heliocentric radial velocity of @xmath3 km s@xmath2 ( gwn ) is a natural outcome of the current disruption model , without affecting local halo kinematics near the sun and microlensing optical depth towards the large magellanic cloud ( lmc ) .
we calculate the dynamical evolution of an orbiting dwarf galaxy in a fixed external gravitational potential representing the milky way .
the potential consists of three parts : a spherical hernquist bulge @xmath13 , a miyamoto - nagai disk @xmath14 , and a logarithmic dark halo @xmath15 , where @xmath16 is the galactocentric distance and @xmath17 denote cylindrical coordinates . each is given as , @xmath18 , @xmath19 , and @xmath20 , where @xmath21 m@xmath10 , @xmath22 kpc , @xmath23 m@xmath10 , @xmath24 kpc , @xmath25 kpc , @xmath26 km s@xmath2 , and @xmath27 kpc .
this choice yields a circular velocity of 228 km s@xmath2 at the solar circle of @xmath28 kpc and a flat rotation curve outside @xmath29 .
we set self - gravitating particles in the dwarf galaxy following a king model , where the central density , central velocity dispersion , and core radius are given as 0.3 m@xmath10 pc@xmath30 , 18.1 km s@xmath2 , and 0.56 kpc , respectively .
in addition , a particle with the mass of @xmath31 m@xmath10 representing @xmath0 cen is placed at the center of the galaxy to trace its orbit .
this setting yields the total mass of the galaxy as @xmath32 m@xmath10 .
a part of the mass is provided by stars , which is roughly estimated from the mean metallicity of stars in @xmath0 cen ( @xmath33[fe / h]@xmath34 ) , combined with the metallicity - luminosity relation for the local group dwarfs ( ct et al . 2000 ) and the mass - to - light ratio ( assuming @xmath35 obtained for @xmath0 cen , meylan et al .
1995 ) , yielding @xmath36 m@xmath10 . thus
, our model galaxy is largely dominated by a dark component , in agreement with the observed large @xmath37 in dwarfs ( mateo 1998 ) .
the galaxy is represented by a collection of @xmath38 particles and the self - gravity is calculated in terms of a multiple expansion of the internal potential to fourth order ( zaritsky & white 1988 ) .
in the course of its orbital motion , a dwarf galaxy is disrupted by galactic tides , whereas its dense core is expected to survive and follow @xmath0 cen s orbit .
while our calculation in a fixed galactic potential neglects dynamical friction against progenitor s orbit , the effect is only modest for the system of @xmath39 m@xmath10 ( zhao 2002 ) , especially during a few orbital periods required for tidal disruption .
we thus examine two representative orbits for the progenitor , model 1 and 2 : model 1 follows the current orbit of @xmath0 cen , whereas for model 2 , we calculate an orbit back to the past over @xmath40 gyr from its current position and velocity by fully taking into account dynamical friction and then set a progenitor galaxy on its non - decaying orbit .
these two models provide us with satisfactory information on the generic properties of a tidally disrupted progenitor and we postulate that the realistic nature of their debris is midway between these model predictions .
we calculate @xmath0 cen s orbit , based on the distance from the sun @xmath41 kpc , position @xmath42 , proper motion @xmath43 mas yr@xmath2 , and heliocentric radial velocity @xmath44 km s@xmath2 ( dinescu , girard , & van altena 1999 ) .
this orbit for model 1 is characterized by frequent disk crossings with a period of @xmath45 yr , retrograde motion , and apo and pericentric distances @xmath46 kpc . for model 2 , we obtain @xmath47 yr and @xmath48 kpc . in both experiments ,
we adopt the same progenitor mass @xmath49 and place it at apocenter to maximize its survival chances .
figure 1 shows the spatial distribution of the tidally disrupted debris .
upper ( middle ) panel shows model 1 ( model 2 ) after the 1.37 ( 1.86 ) gyr orbital excursion of the progenitor galaxy .
lower panel shows the orbit of the galaxy center .
in the course of the orbital motion of the galaxy , its structure is made elongated along the orbit induced by galactic tides , in particular at its pericenter passages , and then the particles are spread out to form the tidal streams along the orbit . a rosette - like feature of the debris becomes steady after about eight orbital periods .
model 1 results in more compact distribution than model 2 , which reflects the difference in orbital radii .
figure 2 shows the velocity distributions of the debris particles in cylindrical coordinates @xmath50 . as is evident , model 1 and 2
provide essentially the same debris kinematics : most remarkable is a sharply peaked @xmath51 distribution at @xmath1 km s@xmath2 , arising from a retrograde orbit of a progenitor .
these kinematics suggest that the difference in model 1 and 2 resides only in the spatial extent of the debris .
in order to assess the reality of the debris stars in light of observed stellar kinematics in the milky way , we analyze the kinematics of both the simulated debris and other galactic stars generated randomly by a monte carlo method .
the metal - poor halo is modeled as a flattened spheroid @xmath52 , where @xmath53 is an axis ratio ranging 0.55 - 0.7 , anisotropic velocity ellipsoid @xmath54 km s@xmath2 , and small mean rotation @xmath55 km s@xmath2 , as found for halo stars with [ fe / h]@xmath56 near the sun ( cb ) .
thin and thick disks are modeled as @xmath57 , where @xmath58 kpc and @xmath59 ( 1 ) kpc for thin ( thick ) disk .
both disks rotate at 200 km s@xmath2 , having velocity ellipsoids of @xmath60 km s@xmath2 and @xmath61 km s@xmath2 for thin and thick disks , respectively ( cb ) .
the relative fraction of each component is fixed using observed local densities near the sun , in such a manner that the halo and thick - disk densities at @xmath62 kpc are 0.2 % and 2 % of the thin - disk density , respectively ( e.g. yamagata & yoshii 1992 ) . in our model of @xmath0 cen s progenitor galaxy ,
the self - gravitating particles represent both stars and dark matter .
we note that a correct estimate for the fraction of stars is uncertain , because their @xmath37 ratio as well as the amount of dark matter in the progenitor is unavailable . as a useful method to incorporate this ambiguity for the current kinematic analysis
, we set a parameter @xmath63 as the fraction of the debris particles relative to halo stars near the sun , _ when all of the particles are regarded as stellar ones_. by this , the normalization of the halo density is obtained for the given debris particles .
a typical value of @xmath63 for the conversion of the simulated particles to the stars is estimated in the following manner .
model 1 ( model 2 ) yields 21 ( 74 ) particles at @xmath64 kpc , giving the mass density of @xmath65 m@xmath10 pc@xmath30 near the sun , whereas the total mass density and metal - poor halo density have been derived as @xmath66 m@xmath10 pc@xmath30 ( gates , gyuk , & turner 1995 ) and @xmath67 m@xmath10 pc@xmath30 ( gould , flynn , & bahcall 1998 ) , respectively .
then , if the debris stars ( with @xmath68 m@xmath10 ) are distributed in the same manner as the simulated particles ( with @xmath69 m@xmath10 ) , which would be a reasonable approximation in view of the dissipationless nature of stars , the mass density of the debris stars in the solar neighborhood can be estimated as @xmath70 m@xmath10 pc@xmath30 , which is about 1 % of the halo density .
thus , @xmath63 , defined here as debris fraction at @xmath71 kpc , is expected to be of order of a few percents .
recently , gwn reported a spectroscopic survey of @xmath72 f / g stars down to @xmath73 mag , in the direction against galactic rotation @xmath74 and @xmath75 , where radial velocities , @xmath76 , in combination with distances largely reflect orbital angular momentum .
the @xmath76 distribution of the stars a few kpc from the sun shows two stellar streams at @xmath77 km s@xmath2 and @xmath3 km s@xmath2 , which are not explained by known galactic components .
while the stream at @xmath78 km s@xmath2 was reproduced by their model of a merging satellite in prograde rotation , the stream at @xmath79 km s@xmath2 remains yet unexplained .
figure 3a shows the @xmath76 distribution for the debris stars of model 2 and halo stars of @xmath80 ( i.e. without disks ) at @xmath81 kpc , @xmath82 , and two fields for @xmath83 .
figure 3b shows when disk stars are incorporated . as is evident , the debris stars from @xmath0 cen s progenitor form a local peak at @xmath79 km s@xmath2 , which is provided by many stars having @xmath84 km s@xmath2 .
this is in good agreement with the @xmath79 km s@xmath2 stream discovered by gwn .
a more flattened halo than the case @xmath80 yields a higher peak , since the density contrast of the debris relative to halo is made higher in this survey region .
it is worth noting that model 1 yields essentially the same @xmath76 distribution as model 2 , reflecting the same velocity distribution , although to attain the same peak height at @xmath3 km s@xmath2 , @xmath63 be a few factor larger and the selected range of @xmath85 be a few degree higher than the respective values in model 2 , because of less number of debris stars near the sun .
this rule applies to other considerations below as well .
majewski ( 1992 ) suggested that the outer halo at the north galactic pole ( ngp ) shows a retrograde rotation @xmath86 km s@xmath2 at @xmath87 kpc .
also , k02 reported that their sample of horizontal branch stars at @xmath88 kpc shows a retrograde rotation at @xmath89 km s@xmath2 . on the other hand ,
halo stars near the sun show no retrograde rotation ( cb ) . to investigate the role of the debris stars in this issue , we select those of model 2 and randomly generated stars at @xmath90 and @xmath91 kpc ( resembling k02 s selection ) .
since the observational determination of full space velocities involves rather inaccurate information of proper motions compared to radial velocities , we convolve the velocity distribution of stars with a gaussian distribution for velocity errors , having 1 @xmath92 of a typical 30 km s@xmath2 error .
the resulting velocity distribution shows a non - gaussian feature owing to the presence of the debris stars : the @xmath51 distribution holds an extra peak at @xmath1 km s@xmath2 in addition to the @xmath93 km s@xmath2 peak , where the former amplitude becomes comparable to the latter one at @xmath63 of a few percents , whereas for @xmath94 and @xmath95 , the velocity distributions are made slightly asymmetric .
however , the change of @xmath96 by the inclusion of the debris stars with @xmath97 % amounts to only @xmath98 ( @xmath99 ) km s@xmath2 for @xmath100 ( @xmath101 ) , which are still insufficient for explaining the reported @xmath102 km s@xmath2 . also , if we extend the selection of the stars at higher @xmath103 or instead consider model 1 , the changes of @xmath96 become smaller than the above mentioned values , because there are no debris stars in our current model
thus , it is safe to conclude that the debris stars contribute only in part to a reported retrograde motion at the ngp .
we select the nearby debris and halo stars at @xmath104 kpc , @xmath105 kpc , and @xmath106 kpc ( as was drawn by cb ) , convolve the velocities with a gaussian error distribution of @xmath107 km s@xmath2 , and compare with the corresponding stars with [ fe / h]@xmath108 in b00 .
it follows that the non - gaussian feature in velocities is much weaker than that at the ngp : the change of @xmath96 for @xmath97 % is only @xmath109 ( @xmath110 ) km s@xmath2 in model 1 ( model 2 ) .
this is due to the smaller debris fraction near @xmath111 than at high @xmath112 .
the effects of the debris stars on the microlensing optical depth towards lmc , @xmath113 , are modest as well . following the gould ( 1999 ) prescription for @xmath113 and investigating the debris within @xmath114 centered at lmc , we arrive at @xmath115 , thereby indicating that @xmath113 provided by the debris stars is much smaller than the observed @xmath113 of @xmath116 .
we have demonstrated that our fiducial models of an orbiting dwarf galaxy that once contained @xmath0 cen predict a sequence of tidal streams in retrograde rotation and their existence is imprinted in kinematics of nearby stars , especially in the direction against galactic rotation ( gwn ) and at the ngp ( k02 ) , while local halo kinematics and microlensing towards lmc remain unchanged .
the streams are mostly distributed inside the solar circle , as suggested from the current orbit of @xmath0 cen ( dinescu 2002 ) .
in contrast to the sgr dwarf galaxy having polar orbit , the orbit of @xmath0 cen s progenitor galaxy is largely affected by a non - spherical disk potential , where the orbital plane exhibits precession with respect to the galactic pole , causing self - crossing of tidal streams in the disk .
the projection of the orbit perpendicular to the disk shows an x-like feature , thereby leaving denser streams at high @xmath112 than at low @xmath112 for a given radius .
this explains the significance of the debris at the ngp compared to the solar neighborhood .
existing kinematic studies of galactic stars to search for a signature of @xmath0 cen s progenitor galaxy are yet confined to nearby stars , where the significance of the debris streams is modest , as shown here .
searches of stars inside the solar circle are more encouraging ( fig .
1 ) , in particular in the directions of @xmath117 and @xmath118 , where we expect the presence of high - velocity streams at @xmath119 km s@xmath2 and @xmath120 km s@xmath2 , respectively . future radial velocity surveys of these fields including the sample of the sloan digital sky survey or planned radial velocity experiment are worth exploring in this context .
also , detailed abundance studies of candidate stream stars will be intriguing , because such stars may exhibit different abundance patterns from field halo stars , as found in dwarf galaxies ( shetrone , ct , & sargent 2001 ) .
a yet unsettled issue is the origin of a progenitor satellite orbiting inside the solar circle , because dynamical friction alone from the present - day _ smooth _ galactic components is insufficient for shrinking the orbit if it was born at a large distance ( say , @xmath121 kpc ) from galactic center ( zhao 2002 ) .
one of the possibilities to preclude it may be that the merging of a satellite occurred while the milky way was still in the process of halo formation via hierarchical merging of several subgalactic clumps ; successive gravitational interaction among clumps may help reducing the orbital angular momentum of a progenitor efficiently . also , a progenitor may have formed in the vicinity of the proto - galaxy , where the environment of a strong tidal field promotes the formation of a compact m32-like galaxy ( burkert 1994 ) and its high density affords the survival chances until the epoch of the galactic disk formation .
then , if a progenitor retained gas , growing galactic tides induce the infall of gas into the progenitor center and trigger the formation of a globular cluster there ( bekki & chiba 2002 ) similar to @xmath0 cen .
the story is yet speculative but worth pursuing based on sophisticated numerical codes for galaxy formation . | we present the kinematic properties of a tidally disrupted dwarf galaxy in the milky way , based on the hypothesis that its central part once contained the most massive galactic globular cluster , @xmath0 cen .
dynamical evolution of a self - gravitating progenitor galaxy that follows the present - day and likely past orbits of @xmath0 cen is calculated numerically and the kinematic nature of their tidal debris is analyzed , combined with randomly generated stars comprising spheroidal halo and flat disk components .
we show that the retrograde rotation of the debris stars at @xmath1 km s@xmath2 accords with a recently discovered , large radial velocity stream at @xmath3 km s@xmath2 towards the galactic longitude of @xmath4 .
these stars also contribute , only in part , to a reported retrograde motion of the outer halo at the north galactic pole .
the prospects for future debris searches and the implications for the early evolution of the galaxy are briefly presented . |
the dirac phase factor @xmath1 with a phase containing integrals over potentials ( of the general form @xmath2 ) @xmath1 is the standard and widely used solution of the usual gauge transformation equations of electrodynamics ( with @xmath3 and @xmath4 vector and scalar potentials respectively ) . in a quantum mechanical context
, it connects wavefunctions of two systems ( with different potentials ) that experience the same classical fields , i.e. either systems that are gauge - equivalent ( a trivial case with no physical consequences ) , or systems that exhibit phenomena of the aharonov - bohm type ( magnetic or electric ) @xmath1 and then this dirac phase has nontrivial observable consequences .
however , it has not been realized that the gauge transformation equations , viewed in a more general context , can have _ more general solutions _ than simple dirac phases , and these lead to wavefunction-_phase - nonlocalities _ that have been widely overlooked and that seem to have important physical consequences . in this paper
we will briefly demonstrate these generalized solutions and will present cases ( and closed analytical results for the wavefunction - phases ) that actually connect ( or map ) two quantal systems that are * neither physically equivalent nor of the usual aharonov - bohm type*. we will also explore the consequences of the new ( _ nonlocal _ ) contributions ( that appear in the wavefunction - phases ) and will see that they are numerous and important ; they are also of a different type in static and in time - dependent configurations ( and in the latter cases they seem to lead to relativistically _ causal _ behaviors , that apparently resolve earlier paradoxes arising in the literature from the use of standard dirac phase factors ) .
let us first remind the reader of a property that is more general than usually realized@xmath0 the solutions @xmath5 of the @xmath6-dependent schrdinger ( or dirac ) equation ( se ) for a quantum particle of charge @xmath7 that moves ( as a test particle ) in two distinct sets of ( predetermined and classical ) vector and scalar potentials ( @xmath8 ) and ( @xmath9 ) , that are generally spatially- and temporally - dependent [ and such that , at the spacetime point of observation @xmath10 , the magnetic and electric fields are the same in the two systems ] , are formally connected through @xmath11 with the function @xmath12 required to satisfy @xmath13 the above property can be immediately proven by substituting each @xmath14 into its corresponding ( @xmath15 ) time - dependent se ( namely with the set of potentials ( @xmath16))@xmath0 one can then easily see that ( [ basic1 ] ) and ( [ basic11 ] ) guarantee that both ses are indeed satisfied together ( after cancellation of a global phase factor in system 2 ) . [ in addition , the equality of all classical fields at the observation point , namely @xmath17 @xmath18 for the magnetic fields ( mfs ) and @xmath19 for the electric fields ( efs ) , is obviously consistent with all equations ( [ basic11 ] ) @xmath1 provided , at least , that @xmath20 is such that interchanges of partial derivatives with respect to all spatial and temporal variables ( at the point @xmath21 ) are allowed ] .
the above fact is of course well - known within the framework of the theory of quantum mechanical gauge transformations ( the usual case being with @xmath22 , hence a mapping from a system with no potentials)@xmath23 but in that framework , these transformations are supposed to connect ( or map ) two _ physically equivalent systems _ ( more rigorously , this being true for ordinary gauge transformations , in which case the function @xmath12 , the so - called gauge function , is unique ( single - valued ) in spacetime coordinates ) . in a formally similar manner ,
the above argument is also often used in the context of the so - called singular gauge transformations , where @xmath24 is multiple - valued , but the above equality of classical fields is still imposed ( at the observation point , that always lies in a physically accessible region)@xmath23 then the above simple phase mapping ( at all points of the physically accessible spacetime region , that experience equal fields ) leads to the standard phenomena of the aharonov - bohm ( ab ) type , reviewed below , where _ unequal fields in physically - inaccessible regions _ have observable consequences . however , we should keep in mind that that above property ( ( [ basic1 ] ) and ( [ basic11 ] ) taken together ) can be _ more generally valid _ @xmath1 and , as already stated , we will present cases ( and closed analytical results for the appropriate phase function @xmath12 ) that actually connect ( or map ) two systems ( in the sense of ( [ basic1 ] ) ) that are * neither physically equivalent nor exhibiting the usual ab behaviors*. and naturally , because of the above provision of field equalities at the observation point , it will turn out that any nonequivalence of the two systems will involve _ remote _ regions of spacetime , namely regions that do _ not _ contain the observation point @xmath10 ( and in which regions , as we shall see , the classical fields experienced by the particle may be _ different _ in the two systems ) . returning to the standard cases , usual @xmath24 s are given in terms of dirac phases , namely integrals over potentials .
i.e. in static cases , and if , for simplicity , we start from system 1 being completely free of potentials ( @xmath22 ) , the wavefunctions of the particle in system 2 ( moving only in a static vector potential @xmath25 ) will acquire an extra phase with an appropriate gauge function @xmath26 that must satisfy@xmath27 the standard ( and widely - used ) solution of this is the line integral @xmath28 ( which , by considering two paths encircling an enclosed inaccessible magnetic flux , formally leads to the well - known magnetic ab effect@xcite ) .
it should however be stressed that the above is only true if @xmath29 is valid for * all * points @xmath30 of the region where the particle moves , i.e. if the particle in system 2 moves ( as a narrow wavepacket ) always outside mfs ( @xmath31 everywhere ) .
similarly , if the particle in system 2 moves only in a spatially uniform scalar potential @xmath32 , the appropriate @xmath24 must satisfy @xmath33 the standard solution being @xmath34 that gives the extra phase acquired by system 2 ( this result formally leading to the electric ab effect@xcite by applying it to two equipotential regions , such as two metallic cages held in distinct time - dependent scalar potentials ) .
once again however it should be stressed that the above is only true if @xmath35 is valid at * all * times @xmath6 of interest , i.e. if the particle in system 2 moves ( as a narrow wavepacket ) always outside efs ( @xmath36 at all times ) .
( in the electric ab setup , the above is ensured by the fact that @xmath6 lies in an interval of a finite duration @xmath37 for which the potentials are turned on , in combination with the narrowness of the wavepacket@xmath23 and the assumption is that , during @xmath37 , the particle has vanishing probability of being at the edges of the cage where the potential starts having a spatial dependence .
the reader is referred to appendix b of peshkin@xcite that demonstrates the intricasies of the electric ab effect , to which we return with an important comment at the end of this paper ) . for potentials more general than in the above cases , ( and if , for notational simplicity , we restrict our attention to only one spatial variable @xmath38)@xmath39it is usually stated that the general gauge function that connects ( through a phase factor @xmath40 ) the wavefunctions of a quantum system with no potentials to those in a general set @xmath41 is the obvious combination ( and a natural extension ) of the above two forms , namely @xmath42 which , however , is generally * incorrect * for @xmath38 and @xmath6 uncorrelated variables@xmath0 it does * not * generally * * satisfy the standard system * * * * ( [ basic11 ] ) ( viewed as a system of partial differential equations ( pdes ) ) , namely @xmath43 indeed@xmath0 ( i ) when the @xmath44 operator acts on eq.([wrong ] ) , it gives the correct @xmath45 from the 1st term , but it also gives some annoying additional nonzero quantity from the 2nd term ( that survives because of the @xmath38-dependence of @xmath4 ) ; hence it invalidates the first of the basic system ( [ basicpde ] ) .
( ii ) similarly , when the @xmath46 operator acts on eq.([wrong ] ) , it gives the correct @xmath47 from the 2nd term , but it also gives some annoying additional nonzero quantity from the 1st term ( that survives because of the @xmath6-dependence of @xmath48 ) ; hence it invalidates the second of the basic system ( [ basicpde ] ) .
it is only when @xmath48 is @xmath6-independent , and @xmath4 is spatially - independent , that eq.([wrong ] ) is correct .
it is also interesting to note that the line integrals appearing in ( [ wrong ] ) do _ not _ form a path ( in spacetime ) that connects the initial to the final point ( see below ) .
[ an alternative form that is also given in the literature is again eq.([wrong ] ) , but with the variables that are not integrated over implicitly assumed to belong to the initial point ( hence a @xmath49 replaces @xmath6 in @xmath3 , and an @xmath50 replaces @xmath38 in @xmath4 ) .
however , one can see again that the system ( [ basicpde ] ) is _ not _ satisfied ( the above differential operators , when acted on @xmath24 , give @xmath51 and @xmath52 , hence _ not _ the values of the potentials at the point of observation @xmath53 as they should ) , _ this not being an acceptable solution either_. and in this case also there is no spacetime - path connecting the initial @xmath54 to the final point @xmath53 either , as the reader can easily verify ] .
what is the problem here , or , better put , what is the deeper reason for the above inconsistency ?
the short answer is the uncritical use of dirac phase factors that come from path - integral treatments ( where @xmath38 and @xmath6 are not * uncorrelated * variables , but actually correlated to produce a path @xmath55 ) .
the general inadequacy of ( [ wrong ] ) was actually one of the main points that has motivated this work . by looking for the most general form of @xmath24 that solves the basic system of pdes we have recently found _ generalized results _ that actually _ correct _
eq.([wrong ] ) in 2 ways@xmath0 through the proper appearance of @xmath50 and @xmath49 ( as in eq.([lambdastatic1 ] ) and eq.([lambdastatic2 ] ) of next section ) @xmath1 which happens to give a path - sense ( that connects the initial to the final point ) in either of the two solutions ( see fig.1 ) , being therefore consistent with feynman s path integral result in the special case of narrow - wavepacket states @xmath1 but most importantly , through the additional presence of _ novel nonlocal terms _ that had so far been overlooked@xcite .
these generalized results are the * exact * solutions of the system ( [ basicpde ] ) but , even most importantly , the formulation ( and methodology of solution ) that produces them , if applied to @xmath56 ( in the 2-d static case ) and also to @xmath57 ( in the full dynamical 2-d case ) , leads to the exact ( nontrivial ) forms of the phase function @xmath24 that , apart from satisfying ( in all cases ) the system ( [ basicpde ] ) , seems to also have far reaching consequences for the wavefunction - phases in the schrdinger picture ( the most important being their causal behavior ) . summarizing
, we will see in this paper that the full form of a general @xmath24 goes beyond the usual dirac phases@xmath0 apart from integrals over potentials , it also generally contains terms of classical fields that act _ nonlocally _ ( in spacetime ) on the solutions of the @xmath6-dependent se . as a result , the phases of wavefunctions in the schrdinger picture are affected nonlocally by mfs and efs @xmath1 nonlocal contributions that have apparently escaped from path - integral approaches .
we will then focus on two types of application of the new formulation@xmath0 ( i ) application to particles passing through static mfs or efs will lead to cancellations of ab phases at the observation point ; these cancellations will be linked to behaviors at the semiclassical level ( to early experimental observations by werner & brill or to recent reports of batelaan & tonomura ) but will be shown to be far more general ( valid not only for narrow wavepackets but also for completely delocalized quantum states ) . by using them we will provide a new interpretation of semiclassical results and we will point out a number of sign errors in popular reports in the literature@xmath0 we will clearly show that semiclassical phase - differences picked up by classical trajectories ( deflected by fields ) are _ opposite _ ( and _ not _ equal , as usually stated or implied ) to the corresponding ab phase ( due to the flux enclosed by the same trajectories ) .
( ii ) application to @xmath6-dependent situations will provide a remedy for a number of misconceptions ( on improper use of simple dirac phase factors ) propagating in the literature ( feynman , erlichson and others ) , and will lead to nontrivially extended phases that contain an ab part and a nonlocal field - part@xmath0 their competition will be shown to recover relativistic causality in earlier paradoxes ( such as the van kampen thought - experiment ) and will provide a fully quantitative formulation of peshkin s qualitative discussion ( on expected causal behavior ) in the electric ab effect ( discussion that was also based on a simple dirac phase factor ) .
the temporal nonlocalities found in this work demonstrate in part _ a causal propagation of phases of quantum wavefunctions in the schrdinger picture _
( through the well - known causal propagation of fields ) , something that may open a new and direct way for addressing @xmath6-dependent double - slit experiments and the associated causal issues .
let us first consider 1-d cases and find the proper @xmath58 that takes us from ( maps ) a system in a set ( @xmath59,@xmath60 ) to a set ( @xmath61,@xmath62 ) .
as already emphasized , we must assume that at the point @xmath53 of observation we have equal efs , i.e. @xmath63 , but we will _ not _ exclude the possibility of the two systems passing through _ different _ efs in other regions of spacetime ( that do _ not _ contain the observation point ) .
in fact , this possibility will _ come out naturally _ from a careful solution of the basic pdes , namely@xmath64 @xmath65 this system is underdetermined in the sense that we only have knowledge of @xmath24 at an initial point @xmath54 and with no further boundary conditions ( hence multiplicities of solutions being generally expected , see below ) . by following a careful procedure of integrations@xcite
we finally obtain 2 distinct solutions ( depending on which eq .
we integrate first)@xmath0 the first solution is @xmath66 with@xmath67 required to be chosen so that the quantity@xmath68 is indep . of @xmath38 , and
the second solution is @xmath69 with@xmath70 to be chosen in such a way that@xmath71is indep . of @xmath6 .
we can directly verify that ( [ lambdastatic1 ] ) or ( [ lambdastatic2 ] ) are indeed solutions of the basic pdes .
[ for ( [ lambdastatic1 ] ) we have ( even for @xmath72)@xmath0 @xmath73 satisfied trivially ( because @xmath74 is indep . of @xmath38 ) , and @xmath75 , and then with the substitution @xmath76 we obtain @xmath77 . since the 2nd and 4th terms _
cancel each other _ , and the 1st term is @xmath78 we obtain @xmath79 @xmath80 we have directly shown therefore that the basic system of pdes is indeed satisfied by our _ generalized _ solution ( [ lambdastatic1 ] ) * even for any nonzero * @xmath81 ( in regions @xmath82 ) .
( note however that at the point of observation @xmath83 , signifying the essential fact that the fields in the two systems are identical ( recall that @xmath84 ) at the point of observation @xmath53 ) .
it can similarly be shown that ( [ lambdastatic2 ] ) is also a solution ] . in ( [ lambdastatic1 ] ) and ( [ lambdastatic2 ] )
the placement of @xmath50 and @xmath49 gives a path - sense to the line integrals in each solution ( each path consisting of 2 perpendicular line segments connecting @xmath54 to @xmath85 with solution ( [ lambdastatic1 ] ) having a clockwise and solution ( [ lambdastatic2 ] ) a counter - clockwise sense , see red and green arrow paths in fig.1)@xmath86this way a natural _ rectangle _ is formed , within which the enclosed electric fluxes in spacetime appear to be crucial ( showing up as nonlocal contributions of the efs - difference from regions @xmath87 of space and time _ that are remote to the observation point _
@xmath53 ) .
these nonlocal terms in @xmath24 have a direct effect on the wfs phases at @xmath53 .
the actual manner in which this happens is determined by the functions@xmath67 or@xmath70these must be chosen in such a way that they satisfy their respective conditions . in fig.1a
we show an extended _ vertical _ striped-@xmath88-distribution ( the case of a 1-d capacitor that is arbitrarily charged for all time ) , where , for @xmath38 located outside ( and on the right of ) the capacitor , the simplest proper choices are @xmath89 and @xmath90 ( since the quantity @xmath91 is already indep . of @xmath38 ( a displacement of the @xmath53-corner of the rectangle to the right does not change the enclosed electric flux @xmath1 hence the choice of @xmath89 ) but _ _ is not a constant__@xmath0 this enclosed flux depends on @xmath6 ( since it _ does change _ with a displacement of the @xmath53-corner upwards ) - hence the choice of @xmath92 above ) .
these choices then of @xmath93 and @xmath92 lead ( through ( [ lambdastatic1 ] ) and ( [ lambdastatic2 ] ) ) to new ( generalized ) solutions for this particular field - configuration .
we then note that the _ difference _ of the two solutions ( [ lambdastatic1 ] ) and ( [ lambdastatic2 ] ) is _ zero _ ( the flux determined by the potential - integrals is exactly cancelled by the nonlocal term of efs ) , a cancellation effect that is important and that will be generalized below . for other shapes of @xmath88 the choices of @xmath93 and @xmath92
will be different@xmath0 for an extended _ horizontal _ strip ( the case of a nonzero ef in all space that has a finite duration @xmath37 ) , proper choices ( for observation instant @xmath94 ) are @xmath95 and @xmath96 ( since the electric flux enclosed in the observation rectangle now depends on @xmath38 , but not on @xmath6 ) @xmath1 or a more involved example would correspond to a triangular shape ( see fig.1b for the corresponding magnetic case to be discussed later ) , where the enclosed flux depends on _ both _ @xmath38 and @xmath6 ( but can be shown to be separable , see next section ) .
as for the last constant terms @xmath97 and @xmath98 ( what we will call multiplicities ) , these are only present when @xmath24 is expected to be multivalued , i.e. in cases of motion in multiple - connected spacetimes , and are then related to the fluxes in the inaccessible regions : in the electric ab setup , the prototype of multiple - connectivity in spacetime , it turns out@xcite that @xmath99 enclosed electric flux , and if these values are substituted in ( [ lambdastatic1 ] ) and ( [ lambdastatic2 ] ) they cancel out the new nonlocal terms and lead to the usual electric ab result . in simple - connected spacetimes , it can be rigorously shown@xcite that solutions ( [ lambdastatic1 ] ) and ( [ lambdastatic2 ] ) are equal ( with @xmath93 being equal to the @xmath6-indep .
bracket of ( [ lambdastatic2 ] ) , and @xmath92 being equal to the @xmath38-indep .
bracket of ( [ lambdastatic1 ] ) ) , the nonlocal terms having therefore the tendency to exactly cancel the ab terms ( this being true _ for arbitrary shapes and analytical form of _
@xmath100 ) .
the same method applied to static 2-d cases ( now for the system of pdes @xmath101 @xmath102 ) finally gives 2 general solutions@xcite@xmath0 the first is @xmath103 with @xmath93 such that @xmath104 @xmath105 @xmath106 and the second is @xmath107 with @xmath108 such that @xmath109 @xmath105 @xmath110 these results apply to cases where the particle goes through _ different _ perpendicular mfs ( recall @xmath111 ) _ in spatial regions remote to the observation point _ @xmath112 .
one can again show that the 2 solutions are equal for simple - connected space , and for multiple - connectivity the values of multiplicities @xmath113 and @xmath114 cancel out the nonlocalities and reduce the above to the usual result of mere @xmath115-integrals along the 2 paths ( i.e. two simple dirac phases ) . for striped @xmath116-distributions , functions @xmath93 and @xmath108
must be chosen in ways compatible with their above conditions ( as in the earlier @xmath53-cases)@xmath23 by then taking the _ difference _ of ( [ lambda(x , y)1 ] ) and ( [ lambda(x , y)4 ] ) we obtain that the ab phase ( originating from the _ closed _ line integral of @xmath115 s ) is exactly cancelled by the nonlocal term of mfs .
this is reminiscent of the cancellation of phases observed in the early experiments of werner & brill@xcite for particles passing through a mf ( a cancellation between the ab phase and the semiclassical phase picked up by the trajectories ) , and our method seems to provide a natural explanation@xmath0 as our results are general ( and for delocalized states in simple - connected space they basically demonstrate the uniqueness of @xmath24 ) , they are also valid and applicable to states that describe wavepackets in classical motion , as _ was _ the case in werner & brill s work .
the above cancellations can then be understood as a compatibility between the ab fringe - displacement and the trajectory - deflection due to the lorentz force ( i.e. the semiclassical phase picked up due to the optical path difference of the two deflected trajectories _
exactly cancels _ ( is _ opposite in sign _ from ) the ab phase picked up by the same trajectories due to the flux that they enclose ) .
this opposite sign seems to have been rather unnoticed@xmath0 in feynman s fig.15 - 8 @xcite , or in felsager s
fig.2.16 @xcite , classical trajectories are deflected after passing through a strip of a mf placed on the right of a double - slit apparatus .
both authors determine the semiclassical phase picked up by the deflected trajectories and find it consistent with the ab phase .
one can see on closer inspection , however , that the two phases actually _ have opposite signs _
( see our own fig.2 and the discussion that follows below , where this is proved in detail ) .
similarly , in the very recent review of batelaan & tonomura@xcite , their fig.2 shows wavefronts associated to deflected classical trajectories where it is stated that the phase shift calculated in terms of the lorentz force is the same as that predicted by the ab effect in terms of the vector potential .
once more , however , it turns out that the sign of the classical phase - difference is really opposite to the sign of the ab phase ( see proof below ) .
the phases are not equal as stated by the authors . and
it turns out that even electric analogs of the above cases also demonstrate this opposite - sign relationship ( see proofs further below ) .
all the above examples can be viewed as a manifestation of the cancellations that have been found in the present work for general quantum states ( but in those examples they are just special cases for wavepacket - states in classical motion ) .
let us give a brief elementary proof of the above claimed opposite sign - relationships@xmath0 indeed , in our fig.2 , the ab phase due to the flux enclosed between the two classical trajectories ( of a particle of charge @xmath7 ) is @xmath117 with @xmath118 the flux quantum , and @xmath119 the enclosed flux between the two trajectories ( for small trajectory - deflections ) , with the deflection originating from the presence of the magnetic strip @xmath116 and the associated lorentz forces . on the other hand ,
the semiclassical phase difference between the same 2 classical trajectories is @xmath120 , with @xmath121 being the de broglie wavelength ( and @xmath122 being the speed of the particle , taken almost constant ( as usually done ) due to the small deflections ) , and with @xmath123 being @xmath124 ( @xmath125 being the ( displaced ) position of the central fringe on the screen ) .
we have therefore @xmath126 now , the lorentz force ( exerted only during the passage through the thin magnetic strip , hence only during a time interval @xmath127 ) has a component parallel to the screen ( let us call it @xmath38-component ) that is given by @xmath128 which shows that there is a change of kinematic momentum ( parallel to the screen ) equal to @xmath129 or , equivalently , a change of parallel speed @xmath130 which is the speed of the central fringe s motion ( i.e. its displacement over time along the screen ) .
although this has been caused by the presence of the thin deflecting magnetic strip , this displacement is occuring uniformly during a time interval @xmath131 and this time interval must satisfy @xmath132 ( as , for small displacements , the wps travel most of the time in uniform motion , i.e. @xmath133 ) .
we therefore have that the central fringe displacement must be @xmath134 and noting that @xmath135 , we finally have @xmath136 by susbstituting ( [ xc ] ) into ( [ semiphase ] ) , the lengths @xmath137 and @xmath138 cancel out , and we finally have @xmath139 which with @xmath140 the flux quantum , and @xmath141 the enclosed flux ( always for small trajectory - deflections ) gives ( through comparison with ( [ abphase ] ) ) our final proof that @xmath142 the electric analog of the above exercise is also outlined below , now with a homogeneous ef ( pointing downwards everywhere in space , but switched on for only a finite duration @xmath37 ) on the right of a double - slit apparatus ( see our fig.3)@xmath0 in this case the electric lorentz force @xmath143 is exerted on the trajectories only during the small time interval @xmath144 which we take to be much shorter ( @xmath145 ) than the time of travel @xmath146 ( we now have a thin electric strip in _ time _ rather than the thin magnetic strip in space that we had earlier ) .
the electric type of ab phase is now @xmath147 with @xmath148 being the electric potential difference between the two trajectories , hence @xmath149 ( again for small trajectory - deflections ) . on the other hand , the semiclassical phase difference between the two trajectories is again given by ( [ semiphase ] ) , but the position @xmath125 of the central fringe
must now be determined by the ef force @xmath150 the change of kinematic momentum ( always parallel to the screen ) is now @xmath151 , hence the analog of ( [ dvx ] ) is now @xmath152 which if combined with ( [ dvx2 ] ) ( that is obviously valid in this case as well , again for small deflections , due to the @xmath153 ) , and always with @xmath146 , gives that the central fringe displacement must be @xmath154 , and using again @xmath155 , we finally have the following analog of ( [ xc ] ) @xmath156 by substituting ( [ xcnew ] ) into ( [ semiphase ] ) , the lengths @xmath137 and @xmath138 again cancel out , and we finally have @xmath157 , which with @xmath140 the flux quantum , and through comparison with ( [ abele ] ) leads once again to our final proof that @xmath158 we note therefore that even in the electric case , the semiclassical phase difference ( between two trajectories ) picked up due to the lorentz force ( exerted on them ) is once again opposite to the electric ab phase picked up by the same trajectories ( due to the electric flux that they enclose )
. we should point out once again , however , that although the above elementary considerations apply to semiclassical motion of narrow wavepackets , in this paper we have given _ a more general understanding of the above opposite sign - relationships _ that applies to general ( even completely delocalized ) states , and that originates from our generalized werner & brill cancellations . in a slightly different vein , the cancellations that we found above give an explanation of why certain classical arguments ( invoking the past @xmath6-dependent history of an experimental setup ) seem to be successful in giving at the end an explanation of ab effects ( namely a phase consistent with that of a static ab configuration ) .
however , there is again an opposite sign that seems to have been largely unnoticed in such arguments as well ( i.e. in silverman@xcite , where in his eq.(1.34 ) there should be an extra minus sign ) .
finally , on other shapes of @xmath116 , see fig.1b for an example of a homogeneous@xmath159 distributed in a triangular shape ( now the part of the magnetic flux contained inside the observation rectangle depending on _ both _
@xmath38 _ and _ @xmath160 ) .
it turns out that this flux can be written as a sum of separate @xmath38- and @xmath160-contributions , and for an equilateral triangle of side @xmath161 we obtain that proper functions ( for the solutions ( [ lambda(x , y)1 ] ) and ( [ lambda(x , y)4 ] ) ) are @xmath162 and @xmath163 . these , if substituted in ( [ lambda(x , y)1 ] ) and ( [ lambda(x , y)4 ] ) , lead to new and nontrivial nonlocal solutions ( or , correspondingly , to _
nonlocal phases _ of wavefunctions ) . in cases of circularly
shaped distributions ( when the enclosed flux is not separable ) it is advantageous to solve the pdes directly in polar coordinates ( for corresponding results see @xcite ) @xmath1 while for general shapes , one may need to first transform to an appropriate coordinate system , and only then apply the above methodology ( i.e. strategy , for solving the resulting pdes ) .
finally , for the @xmath6-dependent 2-d case we have to solve@xmath164 @xmath165 @xmath166 in order to see how the solutions _ combine _ the spatial and temporal nonlocal effects found above .
we now have 3!=6 alternative routes to follow for integrating the system and , at the end , 12 different results are derived , where the @xmath6-propagation of @xmath116 and of @xmath167 and @xmath168 in all space is nontrivially important . by leaving out all the long details@xcite
we merely show one solution , where only @xmath169 appears ( the @xmath6-dependence of @xmath116 having already been incorporated in the behavior of @xmath167 and @xmath168 through faraday s law ) , namely @xmath170 @xmath171 with conditions : @xmath172 @xmath173 @xmath174 @xmath175 , and @xmath176 @xmath110 in the above , @xmath177 accounts for possible multiplicities at @xmath49 .
this solution , together with its spatial dual @xmath178now with @xmath179 replacing the above @xmath115-terms , and with @xmath180 replacing the above @xmath88-terms , and with @xmath181 being replaced by a @xmath182 that must satisfy@xmath0 @xmath183 @xmath184 $ ] , are both crucial for the discussion of the thought - experiment that follows@xmath0 in @xcite van kampen considered a magnetic ab setup , but with an inaccessible magnetic flux that is @xmath6-dependent@xmath0 he envisaged turning on the flux very late , or equivalently , observing the interference of the two wavepackets on a distant screen very early , earlier than the time it takes light to travel the distance to the screen ( i.e. @xmath185 ) , hence using the ( instantaneous nature of the ) ab phase to transmit information ( on the presence of a confined flux somewhere in space ) _
superluminally_. indeed , the ab phase at any @xmath6 is determined by differences of @xmath186 with @xmath187 ( basically a special case of ( [ wrong ] ) ) . however , if we use , instead , our results above ( that contain the additional nonlocal terms ) , it turns out@xcite that , for a spatially - confined flux @xmath188 and for @xmath189 @xmath190 , functions @xmath191 , @xmath192 and @xmath193 can all be taken zero@xmath39(their conditions are all satisfied ) , the point being that at instant @xmath6 , the @xmath194-field has not yet reached the spatial point @xmath112 of the screen @xmath1 a generalization of the striped cases that we saw earlier but now to the case of 3 spatio - temporal variables ( with now the spatial point @xmath112 being outside the light - cone defined by @xmath6 ( see fig.4))@xmath23 as the electric flux is independent of the upper limits @xmath38 and
@xmath6 , this construction rigorously gives @xmath195 .
moreover , the ab multiplicities ( at @xmath49 ) lead to cancellation of the @xmath116-terms ( always at @xmath49 ) , with the final result ( after subtraction of the 2 solutions ) being @xmath196 which , with @xmath197 the instantaneous enclosed magnetic flux and with the help of faraday s law @xmath198 gives @xmath199 although @xmath200 is generally @xmath6-dependent , we obtain the intuitive ( causal ) result that , for @xmath185 ( i.e. if the physical information has not yet reached the screen ) , the phase - difference turns out to be @xmath6-independent , and leads to the magnetic aharonov - bohm phase that we _ would _ observe at @xmath49 .
_ the new nonlocal terms have conspired in such a way as to exactly cancel the causality - violating ab phase _ ( that would be proportional to the instantaneous @xmath188 ) .
this gives a resolution of the van kampen paradox within a canonical formulation , without using any vague electric ab argument ( as there is no multiple - connectivity in @xmath53-plane ) .
an additional physical element is that , for the above cancellation , it is not only the @xmath88-fields but also the @xmath6-propagation of the @xmath116-fields ( the full radiation field ) that plays a role@xcite .
use of the other 10 solutions can also address bound - state analogs ( in @xmath6-driven 1-d nanorings ) or even electric analogs of the van kampen case@xmath0 in peshkin s review@xcite , on the electric ab effect,@xmath39the author correctly states one can not wait for the electron to pass and only later switch on the field to cause a physical effect .
although peshkin uses his eq.(b.5 ) and ( b.6 ) ( based on ( [ wrong ] ) ) , he carefully states that it is not the full solution ; actually , if we view it as an _ ansatz _ , then it is understandable why he needs to enforce a _ condition _ ( his eq.(b.8 ) , and later ( b.9 ) ) on the ef outside the cages ( in order for certain ( annoying ) terms ( resulting from a minimal substitution due to the incorrect ansatz ) to vanish and for ( b.5 ) to be a solution ) .
but then he notes that the extra condition can not always be satisfied ( hence ( b.5 ) is not really the solution for all times ) , drawing from this the above qualitatively correct conclusion on causality .
as it turns out , our treatment gives exactly what peshkin describes in words ( with the total radiation field outside the cages being once again crucial in recovering causality ) , but in a direct and fully quantitative manner , and with _ no ansatz _ based on an incorrect form .
we should also point out that improper uses of simple dirac phases appear often in the literature@xmath0 even in feynman@xcite it is stated that the simple phase factor @xmath201 is valid even for dynamic fields ; this is also explicitly stated in erlichson s
review@xcite @xmath1 silverman@xcite being the only report with a careful wording about ( [ wrong ] ) being only restrictedly valid ( for @xmath6-indep .
@xmath3 and @xmath30-indep .
@xmath4 ) , although even there the nonlocal terms have been missed . at the level of the basic lagrangian @xmath202
there are no fields present , and the view holds in the literature@xcite that efs or mfs can not contribute directly to the phase .
this view originates from the path - integral treatments widely used ( where the lagrangian determines directly the phases of propagators ) , but , nevertheless , our canonical treatment shows that fields _ do _ contribute nonlocally , and they are actually crucial in recovering relativistic causality .
moreover , path - integral discussions@xcite of the van kampen case use wave ( retarded)-solutions for @xmath48 ( hence in lorenz gauge ) and are incomplete ; our results take advantage of the retardation of _ fields _ @xmath203 and @xmath204 * * ( true in _ any _ gauge ) , and _ not _ of potentials .
in addition , troudet@xcite correctly states that his path - integral treatment is good for not highly - delocalized states in space , and that in case of delocalization the proper treatment would be much more complicated , and would require a much more complete analysis .
such an analysis has actually been provided in the present work .
it should be added that the van kampen paradox seems to be still thought of as remarkable@xcite .
the present work has provided a natural and general resolution , and most importantly , through nonlocal ( and relativistically causal ) propagation of wavefunction - phases . on a broader significance of the new solutions
we conclude that a causal behavior may exist at the level of quantum mechanical phases , enforced by the nonlocal terms ( through the well - known causal behavior of fields ) .
the nonlocal terms found in this work at the level of @xmath24 reflect a causal propagation of wavefunction - phases _ in the schrdinger picture _
( at least a part of them , the one containing the fields , that competes with the ab types of phases containing the potentials ) .
this nonlocality and causality of quantum phases is an entirely new concept ( given the local nature but also the nonrelativistic character of the se ) and deserves to be further explored .
possible immediate applications would be in @xmath6-dependent slit - experiments recently discussed using a completely different method ( with modular variables in the heisenberg picture)@xcite .
it has been recently noted@xcite that physics can not currently predict how we dynamically go from the single - slit diffraction to the double - slit diffraction pattern ( whether it is in a gradual and causal manner or not ) .
application of our nonlocal terms to such questions ( i.e. by introducing scalar potentials on the slits in a @xmath6-dependent way ) provides a completely new formulation for addressing causal issues of this type .
finally , one can always wonder what the consequences of these new nonlocalities would be , if these were included in other systems of high - energy or condensed matter physics _ _ with a gauge structure__@xmath23 alternatively , it is worth noting that , if @xmath88 s were substituted by gravitational fields and @xmath116 s by coriolis force fields arising in non - inertial frames of reference , the above nonlocalities ( and their apparent causal nature ) could possibly have an interesting story to tell about quantum mechanical phase behavior in a relativistic / gravitational framework .
students kyriakos kyriakou and georgios konstantinou of the univ . of cyprus and areg ghazaryan of yerevan state university
are acknowledged for having carefully reproduced all results .
georgios konstantinou is also acknowledged for having drawn figures 2 , 3 and 4 .
cleopatra christoforou of the department of mathematics and statistics of the univ . of cyprus
is acknowledged for a discussion concerning the mathematical method followed .
the main findings of this work first appeared in the proceedings of 2 conferences in tel aviv and in bristol ( _ 50 years of the aharonov - bohm effect _
_ and 25 years of the berry s phase _ ) recently published in k. moulopoulos , _
a _ * * 43 * * _ _ , _ _ 354019 _ _ ( 2010 ) .
the present article contains additional material , emphasizing less the mathematical structure and more the physics of sign - errors during passage through classical fields , as well as providing further clarifications on the structure of the entire theory , to make it simpler and accessible to a wider audience . *
fig . 1 . *
( color online)@xmath0 examples of field - configurations ( in simple - connected spacetime ) where the nonlocal terms are nonzero : ( a ) a strip in 1 + 1 spacetime , where the electric flux enclosed in the observation rectangle is dependent on @xmath6 but independent of @xmath38 ; ( b ) a triangular distribution in 2-d space , where the part of the magnetic flux inside the observation rectangle depends on both @xmath38 and @xmath160 .
the appropriate choices for the corresponding functions @xmath93 and @xmath92 for case ( a ) , or @xmath93 and @xmath108 for case ( b ) , are given in the text .
* figure 2 . *
( color online)@xmath0 the standard double - slit apparatus with an additional strip of a perpendicular magnetic field @xmath116 of width @xmath206 placed between the slit - region and the observation screen .
the deflection shown is for a negative charge @xmath7 ( and in the text it is assumed small , due to @xmath207 ) .
* figure 3 . *
( color online)@xmath0 the analog of fig.2 ( again for a negative charge @xmath7 ) but with an additional electric field parallel to the observation screen that is turned on for a time interval @xmath37 ( with @xmath145 , and @xmath6 the time of travel ) .
* figure 4 . *
( color online)@xmath0 the analog of paths of fig.1 but now in 2 + 1 spacetime for the van kampen thought - experiment , when the instant of observation @xmath6 is so short that the physical information has not yet reached the spatial point of observation @xmath112 .
the two solutions ( that , for wavepackets , have to be subtracted in order to give the phase difference at @xmath208 ) are described in the text , and are here characterized through their electric field @xmath88-line - integral behavior@xmath0 electric field path ( i ) ( the red - arrow route ) denotes the dual solution , and electric field path ( ii ) ( the green - arrow route ) denotes the primary solution given in the beginning of section iv . | we report on previously overlooked solutions of the usual gauge transformation equations that exhibit a new form of nonlocal quantal behavior with the well - known relativistic causality of classical fields affecting directly the phases of wavefunctions .
the new nonlocalities compete with aharonov - bohm behaviors and they provide@xmath0 a correction to a number of erroneous results in the literature , a new interpretation of semiclassical observations and further extensions to delocalized states , a natural remedy of earlier paradoxes , and a new formulation in the study of time - dependent slit - experiments . |
let @xmath14 \rightarrow v$ ] , where @xmath15is some linear space and@xmath16 \rightarrow \left [ 0,t\right]\]]be such that @xmath17 . by the
_ historic frame path _ @xmath18 of @xmath19 , we mean the path - valued path@xmath20 .
\label{algebraic frame process definition}\]]@xmath21 takes its values in the space of paths from @xmath22 $ ] to @xmath5 , i.e. @xmath23 } $ ] ( where @xmath24 } $ ] denotes the space of all paths from @xmath25 $ ] to @xmath5 ) . the map@xmath16 \rightarrow \left [ 0,t\right]\]]determines the frame length of @xmath21 at time @xmath26 so that the evaluation of @xmath21 at time @xmath26 is a @xmath5-valued path of length @xmath27 . if @xmath28 , then @xmath29 .
we give two examples to illustrate the relevance of the _ historic frame path : _ from a control theory point of view , a differential equation is interpreted as follows : the system we want to control has state @xmath30 at time @xmath31 ( where @xmath30 lives on some manifold @xmath32 ) .
the initial state is given by @xmath33 .
the driving signal @xmath34 that determines the state of the system at time @xmath31 lives on some banach space @xmath5 .
it filters through a map @xmath35to produce an effect in the state space @xmath32 ( here @xmath36 denotes the tangent space of @xmath32 ) .
we call @xmath37 the _ vector field map _ and write @xmath38subject to @xmath37 and @xmath19 satisfying certain conditions , @xcite gives meaning to ( [ differential equation ] ) ( the exact result is quoted in theorem [ universal limit theorem ] below ) . in the classical theory of dynamical systems with memory ( _
delay differential equations _ ) , the system response at time @xmath39 may depend on the entire solution trajectory up to time @xmath39 , that is the vector field map @xmath37 is a function of the path segment @xmath40 , so that @xmath41the delay here occurs as an argument of the vector field map .
one could imagine a dynamical system with a different type of memory where the delay occurs in the driving signal @xmath19 : _
_ _ _ since each state @xmath42 is determined by the initial state @xmath43 and the signal path segment @xmath44 , an approach where @xmath45 is driven by the _
entire _ historic trajectory @xmath46 could prove to be interesting .
in other words , we might like to give meaning to ( [ differential equation ] ) in the sense of lyons98 if the driving signal @xmath19 evaluated at time @xmath39 equals @xmath47 and @xmath48 is some @xmath5-valued path . in this case
, @xmath49 is the frame process defined in ( [ algebraic frame process definition ] ) associated to @xmath48 with variable frame length @xmath50 .
we consider a volterra gaussian process , say fractional brownian motion ( fbm ) with hurst parameter @xmath51 .
@xmath2 is a real - valued brownian motion .
the evaluation of fbm at time @xmath31 is obtained by applying the wiener integral to an appropriate deterministic @xmath52- kernel @xmath53 is given by@xmath54 \mathbf{1}_{\left ] 0,t\right [ } \left ( s\right)\]]where @xmath55 , @xmath56this is a functional of the path - segment @xmath57
i.e. a functional of the brownian frame process of frame length @xmath58 .
this thesis investigates some of the rough path properties ( c.f .
@xcite ) of the _ historic frame path _ on the wiener space @xmath59 \right ) , \sigma _ { \left\vert .\right\vert _ { \infty } } , \mathbb{p}\right ) $ ] \right ) $ ] denotes the space of continuous functions on @xmath60 $ ] that are @xmath61 at @xmath62 .
@xmath63 denotes the completion of the @xmath64-algebra generated by the @xmath8-norm and @xmath65 is the wiener measure .
] , defined for @xmath66 \right ) $ ] at @xmath67 $ ] as@xmath68here , we have a constant frame length @xmath69 .
when viewed as a ( borel ) random variable on wiener space , @xmath70 will be called the _ brownian frame _ random variable at @xmath26 ( associated to the brownian motion @xmath2 supported on @xmath71 \right ) , \sigma _ { \left\vert .\right\vert _ { \infty } } , \mathbb{p}\right ) $ ] ) .
the theory of rough paths as developed in @xcite shows how to construct solutions to differential equations driven by paths that are not of bounded variation but have controlled roughness .
the @xmath6-variation ( definition _ _ _ _ [ defn : p - varn ] below ) is taken as a measure of a banach space valued path s roughness .
the analysis is independent of the dimension of the banach space and so appears particularly well adapted to the case where the banach space in question is a path - space .
we give a brief overview : let @xmath5 be a banach space with norm @xmath72 .
we denote by @xmath73 the truncated tensor algebra of @xmath5 , that is @xmath74where @xmath75 ( @xmath76 copies ) . a _ continuous _ functional @xmath77is called _ multiplicative _ if it satisfies _ _
identity__@xmath78and @xmath79 [ ex : level 1 multiplicative lift]every continuous @xmath5-valued path @xmath19 determines a multiplicative functional @xmath80 in @xmath81 .
the converse is not true : with every multiplicative functional @xmath82 we may associate a collection of @xmath5-valued paths @xmath83 . a wide class of _ multiplicative functionals _ is given by [ theorem : chen s theorem]suppose @xmath19 is a bounded variation path in @xmath5 .
then @xmath19 has a canonical _ multiplicative _ lift @xmath84 to @xmath85 given by its sequence of iterated integrals:@xmath86we call @xmath87 the _ chen lift of _ @xmath19 . in general , if @xmath88 is a multiplicative functional with values in @xmath89 and for some @xmath90 taking values in @xmath73 with @xmath91 , we have that@xmath92then we say that @xmath93 is the projection of @xmath94 onto @xmath85 .
we also say that @xmath94 is _ a _ multiplicative functional lying above @xmath93 or that @xmath94 is _ a lift _ of @xmath93 .
it is important to note that in general for a given @xmath93 , neither existence nor uniqueness of @xmath94 is obvious ( theorem [ firsttheoremlyons ] below deals with the existence of a unique lift for a particular class of @xmath93 ) . for each @xmath95 , @xmath96 is assumed to be equipped with a _
compatible tensor norm _
@xmath97 : if @xmath98 and @xmath99 , then @xmath100 is said to be _ compatible _ if for any @xmath101 with @xmath102 , we have @xmath103and @xmath104 [ defn : p - varn ] let @xmath105 \right ) $ ] denote the set of all finite dissections of @xmath106 $ ] , that is @xmath107 \right ) : = \left\ { \left\ { t_{0},t_{1,}, ... ,t_{n}\right\ } : t_{0}=0<t_{1}<t_{2}< ... <t_{n}=t\text { and } n% \text { is finite}\right\ } .\]]for @xmath108 , the @xmath6__-variation functional of level _ _ @xmath76 of a function @xmath109with@xmath110is defined as @xmath111 \right ) \right\ } ^{\frac{i}{p}}.\]]@xmath87 is said to have finite @xmath6__-variation _ _ if@xmath112is finite . a multiplicative functional of finite @xmath6-variation with values in @xmath113 ( where @xmath114 ) is called a rough path of roughness @xmath6 .
the set of all @xmath6-rough paths is denoted as @xmath115 for any @xmath116 @xmath117 , the @xmath6-variation distance @xmath118 is defined as@xmath119 @xmath120 is a complete metric space as are all projections of @xmath121 onto @xmath85 for @xmath122 ( lemma 3.3.3 in @xcite ) .
however , @xmath121 is not a linear space in general , the sum of two multiplicative functionals fails to be multiplicative . if @xmath123 , then @xmath124 . for @xmath125 , @xmath126 ^{\frac{\acute{p}}{p% } }
& = & \left [ \sum_{j=1}^{n}\left\vert a_{j}\right\vert \right ] \left [ \sum_{j=1}^{n}\left\vert a_{j}\right\vert \right ] ^{\frac{\acute{p}}{p}-1 } \\ & = & \sum_{j=1}^{n}\left\vert a_{j}\right\vert \left [ \sum_{i=1}^{n}\left\vert a_{i}\right\vert \right ] ^{\frac{\acute{p}}{p}-1}\geq \sum_{j=1}^{n}\left\vert a_{j}\right\vert ^{\frac{\acute{p}}{p}}.\end{aligned}\]]we now fix @xmath127 and a dissection @xmath128 in @xmath105 \right ) $ ] ( definition defn : p - varn above ) .
if we take @xmath129 for @xmath130 and then take the @xmath8 over @xmath105 \right ) $ ] , the result follows ( we set @xmath131 for @xmath132 ) .
[ defn : smooth rough path]suppose @xmath19 is a bounded variation path in @xmath5 .
we will denote by @xmath87 the chen lift of @xmath19 to @xmath133 and call @xmath87 a _ smooth rough path_. the collection of smooth rough paths is contained in @xmath134 .
[ defn : geometric rough path]the closure of the smooth rough paths under @xmath118 is called the space of _ geometric rough paths of roughness _ @xmath6 and denoted @xmath135 . the _ _ first theorem _ _ from lyons98 gives sufficient conditions for the existence of a unique lift : [ firsttheoremlyons]let @xmath136 be a continuous multiplicative functional in @xmath137 of finite @xmath6-variation where @xmath138 .
there exists a multiplicative extension @xmath139 to @xmath140 , @xmath141 which is of finite @xmath6-variation .
the extension is unique in this class .
the _ universal limit theorem _ below establishes the connection between multiplicative functionals and solutions to differential equations : [ universal limit theorem]let @xmath37 be a linear map from @xmath5 to the space of lipschitz-@xmath142 vector fields on @xmath32 , that is @xmath143 denotes the space of @xmath95-times differentiable functions from @xmath144 to @xmath32 whose @xmath145 derivative is @xmath146-lipschitz]where @xmath147 . then consider the it map @xmath148 defined for bounded variation paths by @xmath149define the one form @xmath26 by@xmath150for any geometric multiplicative functional @xmath151 there is exactly one geometric multiplicative functional extension @xmath152 such that if @xmath153 @xmath154 then @xmath155 satisfies the rough . to emphasise the difference between a differential equation and a _ rough _ differential equation we write @xmath156instead of @xmath157(c.f definition 4.1.1 in @xcite ) * * ] differential equation@xmath158such a solution exists on a small interval @xmath159 $ ] whose length can be controlled entirely in terms of the control of the roughness of @xmath160 and of @xmath37 . the it map is uniformly continuous and the map @xmath161 is the unique continuous extension of the it map from @xmath162 to @xmath163 . in chapter @xmath164
we establish the existence of a continuous brownian frame process @xmath0 into @xmath165 \right ) , \left\vert .\right\vert _ { \infty } \right ) $ ] .
@xmath0 algebraically agrees with the object defined in ( [ brownian frame process ] ) , i.e. for any @xmath166 \right ) $ ] , @xmath167 .
we call @xmath0 the brownian @xmath8__-frame process__. _ _ _ _ for @xmath168 @xmath169 is shown to have finite @xmath9-variation .
furthermore , the dyadic polygonal approximations of @xmath0 converge to @xmath170 in @xmath9-variation .
if @xmath171 , @xmath165 \right ) _ { p},\left\vert .\right\vert _ { p}\right ) $ ] denotes the banach space of continuous functions of finite @xmath6-variation equipped with the @xmath6-variation norm @xmath172for @xmath7 , we establish the existence of a continuous brownian frame process @xmath173 into @xmath165 \right ) _ { p},\left\vert .\right\vert _ { p}\right ) $ ] that algebraically agrees with the object defined in ( [ brownian frame process ] ) off a nullset .
that is , @xmath174 is equal to @xmath175 for @xmath176-a.e .
@xmath166 \right ) .$ ] @xmath177 is called the @xmath6__-variation frame process . _ _ for @xmath178 , we show that the dyadic polygonal approximations of @xmath179 converge to @xmath173 in @xmath9-variation . in chapter @xmath180 , we examine further properties of @xmath181 ( @xmath7 ) : we find constants , @xmath182 ( @xmath183 is a constant strictly greater than @xmath184 ) and @xmath185 , so that the random variable @xmath186has gaussian tails .
more precisely , we find constants @xmath187 and @xmath185 such that @xmath188 in chapter @xmath189 we are concerned with the @xmath8-frame process @xmath170 .
the _ universal limit theorem _ ( theorem [ universal limit theorem ] above ) tells us how to solve differential equations driven by rough paths _ provided the driving signal is the level _
@xmath190 _ projection of some geometric _ @xmath9__-rough path .
_ _ since @xmath191 \rightarrow c\left ( \left [ 0,1\right ] , \left\vert .\right\vert _ { \infty } \right)\]]has finite @xmath9-variation for @xmath11 , by theorem firsttheoremlyons , any lift of @xmath0 to @xmath192 that preserves finite @xmath9-variation establishes an integration theory for @xmath0 ( here , @xmath193 \right ) , \left\vert .\right\vert _ { \infty } \right ) $ ] ) . from chapter 2
we know that the ( smooth ) dyadic polygonal approximations @xmath194 of @xmath0 converge to @xmath0 in @xmath195-variation norm .
so it is natural to ask whether the lifts of @xmath196 converge in @xmath192 , thereby providing a natural lift of @xmath0 to @xmath197 .
we show that while the canonical lifts of the dyadic polygonal approximations to @xmath0 converge , the convergence is to an object that lives _ outside _ the injective tensor product @xmath198 : after proving that @xmath199 is isomorphic to @xmath200 \times \left [ 0,1\right ] \right ) $ ] , we show that the lvy area @xmath201 of @xmath202 which is the limit of the antisymmetric component of @xmath203 as @xmath204 is continuous _ off _ the diagonal of the unit square .
however , it is shown to have a jump - discontinuity _ on _ the diagonal and so does not exist in @xmath199 .
we believe that a further examination of the obstruction which is intimately linked to the quadratic variation of brownian motion will prove interesting and important .
we work on the classical wiener space @xmath71 , \mathbb{r}\right ) , \sigma _ { \left\vert .\right\vert _ { \infty } } , \mathbb{p}% \right ) $ ] . here ,
@xmath205 \right ) : = \left\ { f\in c\left ( \left [ -1,1% \right ] \right ) : f\left ( -1\right ) = 0\right\}\]]and @xmath206 denotes the completion of the borel @xmath64-algebra generated by the sup - norm @xmath207 \right\ } , \]]with respect to the wiener measure @xmath65 .
@xmath2 denotes the coordinate process @xmath208 \right ) \rightarrow \mathbb{r}% : b_{t}\left ( f\right ) : = f\left ( t\right ) \text { \ \ } , t\in \left [ -1,1\right]\ ] ] ( on @xmath209 \right ) $ ] ) which ( under wiener measure @xmath176 ) is a brownian motion . @xmath210 \right ) _ { p}$ ] denotes the space of continuous real - valued paths on @xmath25 $ ] of finite @xmath6-variation , i.e.@xmath211 \right ) _ { p}=\left\ { f\in c\left ( \left [ a , b\right ] \right ) : \mathcal{v}_{p}\left ( f^{1}\right ) < \infty \right\ } , \]]where @xmath212 ( c.f .
theorem [ theorem : chen s theorem ] in chapter 1 ) and the @xmath6-variation functional of level @xmath190
@xmath213 is defined in definition [ defn : p - varn ] in chapter 1 .
similarly , @xmath214 \right ) _ { p}$ ] denotes the space of continuous functions of finite @xmath6-variation that are @xmath61 at @xmath215 , i.e. @xmath216 \right ) _ { p}=\left\ { f\in c_{0}\left ( \left [ a , b\right ] \right ) : \mathcal{v}_{p}\left ( f^{1}\right ) < \infty \right\ } .\ ] ] when @xmath217 \right )
_ { p}$ ] , we say that @xmath37 has finite @xmath6-variation .
we write @xmath218 in place of @xmath219 and in this way view @xmath220 as a functional on @xmath210 \right )
_ { p}$ ] . from chapter 1
we recall that @xmath221 is the @xmath6-variation norm on @xmath222 \right ) _ { p}$ ] for which we write@xmath223for the banach space @xmath224 \right ) _ { p}$ ] equipped with @xmath225 we write @xmath165 \right ) _ { p},\left\vert .\right\vert _ { p}\right ) $ ] .
when the norm is not explicitly mentioned , @xmath224 \right ) _
{ p}$ ] is viewed as a subspace of @xmath165 \right ) , \left\vert .\right\vert _ { \infty } \right ) $ ] . for @xmath226 and @xmath227
, we define the function @xmath228 as@xmath229 the evaluation at @xmath67 $ ] of the ( deterministic ) _ frame operator _
@xmath230on the set of functions from @xmath231 $ ] to @xmath232 denoted as @xmath233 } $ ] is defined as@xmath234 } \rightarrow \mathbb{r}^{% \left [ 0,1\right ] } \\
\mathcal{t}_{h}^{f}=\left ( f\left ( h-1+u\right ) \right ) _ { u\in \left [ 0,1% \right ] } % \end{array}% \right .
.\]]we will be interested in @xmath235 as a function on wiener space and so will be considering the restriction of @xmath235 to @xmath236 \right ) $ ] . if @xmath237 , we write @xmath238 to denote the number of elements in @xmath239 , i.e. @xmath240 . the @xmath241-function is defined as @xmath242
. @xmath243 ^{d}\right ) $ ] is the @xmath244-dimensional unit cube of dyadic rationals , i.e. @xmath245 ^{d}\right ) : = \left ( \bigcup_{n\in \mathbb{n}% } \left\ { \frac{k}{2^{n}}:0\leq k\leq 2^{n}\right\ } \right ) ^{d } , \label{eqn : dyadic n - cube}\]]and for @xmath246 ^{d}$ ] , @xmath247 means that @xmath248 finally , for any @xmath67 $ ] , @xmath249 denotes the unique integer such that @xmath250 . in the following section ,
we show that for every fixed @xmath251 $ ] , @xmath252 \right ) } $ ] is a borel random variable on wiener space , mapping into @xmath253 \right ) , \left\vert .\right\vert _ { \infty } \right ) $ ] . in order to distinguish between the algebraic frame operator @xmath254 ( which we recall is into @xmath255 } $ ] without a topology ) and the analytic object that is the borel random variable into @xmath165 \right ) , \left\vert .\right\vert _ { \infty } \right ) $ ] , we write @xmath256 \right ) } .\]]we show that @xmath0 has continuous sample paths .
as mentioned in chapter @xmath190 , @xmath0 is called the @xmath8-frame - process . for @xmath257
we show that for any fixed @xmath67 $ ] the function @xmath258 \right ) _ { p } \\ 0 & \text{otherwise}% \end{array}% \right .
\label{p - variation frame process}\]]is a borel random variable on wiener space .
we produce hlder - type moment bounds for @xmath259 and @xmath260 . in the case of @xmath261 , we use kolmogorov s lemma to show that there exists a modification with continuous sample paths . we denote this continuous modification as @xmath181 and call it the @xmath6-variation frame process .
we then show that if @xmath10 , the dyadic polygonal approximations of @xmath181 converge to @xmath262 in @xmath9-variation , @xmath65-a.s . in the case of @xmath202
we deduce that @xmath0 has finite @xmath9-variation for @xmath11 , @xmath65-a.s , and that the polygonal dyadic approximations of @xmath0 converge to @xmath0 in @xmath195-variation , @xmath65-a.s .
[ lemma : sup - norm process is borel]for any @xmath67 $ ] , @xmath263 is continuous and hence borel measurable .
fix @xmath264 .
suppose @xmath265 \right ) $ ] and that @xmath266 then @xmath267 } \left\vert f\left ( h-1+u\right ) -g\left ( h-1+u\right ) \right\vert \leq \left\vert f - g\right\vert _ { \infty } < \varepsilon \text{.}\ ] ] the @xmath8-norm frame process @xmath0 has continuous sample paths .
fix @xmath166 \right ) $ ] and @xmath264 . since @xmath268 $ ] is compact , @xmath37 is uniformly continuous , so that @xmath269 such that @xmath270 , i.e.@xmath271 : \left\vert x - y\right\vert < \delta _ { \varepsilon } } \left\vert f\left ( y\right ) -f\left ( x\right ) \right\vert \leq \varepsilon \text{,}\]]and so for @xmath272 , @xmath273 } \left\vert \mathcal{t}_{h_{2}}^{f}\left ( u\right ) -\mathcal{t}_{h_{1}}^{f}\left ( u\right ) \right\vert \\ & = & \sup_{u\in \left [ 0,1\right ] } \left\vert f\left ( h_{2}-1+u\right ) -f\left ( h_{1}-1+u\right ) \right\vert \\ & \leq & \sup_{x , y\in \left [ -1,1\right ] : \left\vert x - y\right\vert < \delta _ { \varepsilon } } \left\vert f\left ( y\right ) -f\left ( x\right ) \right\vert \leq \varepsilon .\end{aligned}\ ] ] we regard the @xmath6-variation functional @xmath274 as a functional on @xmath253 \right ) _ { p},\left\vert .\right\vert _ { \infty } \right ) $ ] and prove that it is lower semi - continuous as a map@xmath275 \right ) _
{ p},\left\vert .\right\vert _ { \infty } \right ) \rightarrow \left ( \mathbb{r}^{+},\left\vert .\right\vert \right ) .\ ] ] a real valued function @xmath37 defined on a topological space @xmath276 is _ lower semi - continuous _ if for any @xmath277 the set @xmath278 is closed .
so any lower - semicontinuous real - valued function is borel - measurable . _ _ [ lsc_lemma]a real - valued function @xmath37 is _ lower semi - continuous _
iff @xmath279 , then @xmath280 .
@xmath281 : fix any @xmath282 and consider @xmath283 .
suppose @xmath284 and @xmath279 .
since @xmath285 for all @xmath286 , it follows that @xmath287 .
but by assumption @xmath288 , so that @xmath289
. thus , @xmath290 is closed .
@xmath291 : let @xmath292 and fix @xmath293 .
an infinite subsequence of @xmath294 is contained in @xmath295 .
hence , @xmath296 for all @xmath293 .
hence , @xmath297 .
[ propn : semicontinuity of p - variation]we recall the definition of the @xmath298-variation functional ( c.f .
definition [ defn : p - varn ] ) : let @xmath299 \right ) $ ] denote the set of all finite dissections of @xmath106 $ ] . if @xmath171 , the @xmath6-variation functional @xmath300 on @xmath224 \right ) _ { p}$ ] , defined as@xmath301 \right ) _ { p}\rightarrow \mathbb{r}% ^{+}:\mathcal{v}_{p}\left ( f\right ) : = \sup \left\ { \sum_{d}\left\vert f\left ( t_{i+1}\right ) -f\left ( t_{i}\right ) \right\vert ^{p}:d\in \mathcal{d% } \left ( \left [ 0,1\right ] \right ) \right\ } ^{\frac{1}{p}}\]]is lower - semicontinuous . fix @xmath302 \right ) _
suppose that @xmath303 \right ) $ ] and that @xmath304 .
since @xmath305 $ ] is compact and @xmath37 and @xmath306 are continuous , we may assume w.l.o.g .
that the sequence @xmath307 is uniformly bounded by @xmath308 , i.e. that @xmath309 .
since @xmath37 has finite @xmath6-variation , @xmath310 , there exists a sequence of dissections @xmath311 with @xmath312 \right)\]]such that@xmath313 in other words : the sequence of step - functions defined as @xmath314converges to @xmath37 in @xmath6-variation .
we start by fixing @xmath264 .
w.l.o.g . we may assume that @xmath315 is non - decreasing ( otherwise we eventually end up with the trivial dissection @xmath316 ) so that we may choose @xmath317by definition of @xmath318 , @xmath319 such that @xmath320and since @xmath304 in @xmath224 \right ) $ ] and @xmath305 $ ] is compact , @xmath321 such that @xmath322fix @xmath323 and @xmath324 and consider any @xmath325 .
then @xmath326 by the mean value theorem , for any differentiable function @xmath327 on @xmath328 $ ] , @xmath329 } \left\vert g^{\prime } \left ( t\right ) \right\vert .
\label{mean value theorem}\]]in particular , we choose @xmath330 and @xmath331 and apply ( [ mean value theorem ] ) to the right hand side of ( some bounds ) , noting that @xmath332we then get@xmath333hence , @xmath334 } \right ) \right .
\\ & & \left .
+ 2^{p}\delta _ { m}pk^{p-1}\left\vert d_{m}\right\vert \right ) \\ & = & \inf_{m >
n_{\varepsilon } ^{1}}\inf_{n > n_{\varepsilon , m}^{2}}\mathcal{v}% _
{ p}^{p}\left ( \sum_{d_{m}}f_{n}\left ( t_{j}^{\left ( m\right ) } \right ) \mathbf{1}_{\left [ t_{j}^{\left ( m\right ) } , t_{j+1}^{\left ( m\right ) } \right ] } \right ) + \varepsilon \\ & \leq & \inf_{n > n_{\varepsilon , m}^{2}}\mathcal{v}_{p}^{p}\left ( f_{n}\right ) + \varepsilon .\end{aligned}\]]letting @xmath335 , we get@xmath336the result now follows from lemma [ lsc_lemma ] .
hence , fix @xmath171 .
let @xmath337 denote the imbedding map@xmath338 \right ) _
{ p},\left\vert .\right\vert _ { \infty } \right ) \rightarrow \left ( c\left ( \left [ 0,1\right ] \right ) _ { p},\left\vert .\right\vert _ { \infty } + \mathcal{v}_{p}\left ( .\right ) \right)\]]for every fixed @xmath67 $ ] , @xmath339 \right ) _ { p},\left\vert .\right\vert _ { \infty }
\right ) \rightarrow \left ( c\left ( \left [ 0,1\right ] \right ) _ { p},\left\vert .\right\vert _ { p}\right)\]]is @xmath206-measurable . for @xmath7 , @xmath340 \right ) , \left\vert .\right\vert _ { \infty } \right ) \rightarrow \left ( c\left ( \left [ 0,1\right ] \right ) _ { p},\left\vert .\right\vert _ { p}\right)\]]as defined in ( [ p - variation frame process ] ) is a borel random variable .
proposition [ propn : semicontinuity of p - variation ] proves that the imbedding map @xmath337 is _ lower - semicontinuous _ and hence @xmath341-measurable _ _ _ _ for any @xmath171 . _ _ _ _ in addition , by lemma [ lemma : sup - norm process is borel ] , for any fixed @xmath67 $ ] , @xmath342 \right ) , \left\vert .\right\vert _ { \infty } \right ) \rightarrow \left ( c\left ( \left [ 0,1\right ] \right ) , \left\vert .\right\vert _ { \infty } \right ) $ ] is continuous , so that the restriction of @xmath254 to @xmath71 \right ) _
{ p},\left\vert .\right\vert _ { \infty } \right ) $ ] is also continuous .
hence , @xmath343 is @xmath344-measurable .
a result by p. lvy ( _ theorem 9 _ in @xcite ) states that @xmath176-a.e .
brownian sample path has infinite @xmath6-variation if @xmath345 , but finite @xmath6-variation if @xmath7 , so that @xmath346 \right ) _ { p}\right ) = 1\text { if } p>2\text{.}\]]since @xmath206 is complete w.r.t .
@xmath176 , both @xmath209 \right ) _ { p}$ ] and @xmath347 \right ) \setminus c_{0}\left ( \left [ -1,1% \right ] \right ) _ { p}$ ] are in @xmath206 .
hence,@xmath348 \right ) _ { p } \\ 0 & \text{otherwise}% \end{array}% \right.\]]is borel - measurable . the following inequality ( _ lemma 3 _ in @xcite ) will be used frequently : [ lemma : l_p bound]suppose that @xmath349 for all @xmath350 .
then @xmath351@xmath352 is as defined in section 2.1 , equation ( c(alpha , p)-definition ) , i.e. @xmath353where @xmath226 and @xmath227 . by hlder s
inequality@xmath354 [ propn : pathwise p - varn bound]suppose that _ _ _ _ @xmath251 $ ] and that @xmath355 \right ) $ ] .
if @xmath227 and @xmath356 , then there is a finite constant @xmath357 such that@xmath358bounds both @xmath359and @xmath360and hence @xmath361here , @xmath362 is defined in section 2.1 , equation ( c(alpha , p)-definition ) . if @xmath363 , let @xmath364 ( @xmath365 ) denote the smallest ( largest ) integer that is greater ( less )
than @xmath19 .
any interval @xmath366 $ ] with @xmath367 is a countable union of disjoint dyadic intervals contained in @xmath368 of the form@xmath369and @xmath370where @xmath371 ( @xmath249 is defined in _ section 2.1 _ ) .
suppose that @xmath355 \right ) $ ] . in order to simplify notation
, we shift @xmath37 to the right by @xmath190 and work with @xmath372 \right ) .\]]with @xmath373 we associate the function @xmath374 @xmath375 \times \left [ 0,2\right ] \right ) $ ] .
since @xmath376 and @xmath377 and @xmath373 @xmath378is continuous , we have the following representation for @xmath379 : @xmath380 using ( [ representation ] ) we may now apply ( [ hoelder s inequality ] ) from lemma [ lemma : l_p bound ] to bound the @xmath381 power of @xmath382 by the @xmath381 powers of the increments over dyadics contained in @xmath383 @xmath384which by jensen s inequality is bounded by @xmath385 but @xmath386is contained in @xmath387 \cap \left\ { \frac{k}{2^{n}}:0\leq k\leq 2^{n+1};n\geq n\left ( t - s\right ) \right\ } .\]]hence , ( [ 1st p - bound ] ) and in turn @xmath388 is bounded by @xmath389we now use ( [ p - bound for continuous functions ] ) to bound 1 .
@xmath390 } \left\vert \hat{f}^{1}\left ( t , t+h\right ) \right\vert $ ] where @xmath67 .$ ] + for each fixed @xmath31 , ( [ p - bound for continuous functions ] ) applies .
note that ( [ p - bound for continuous functions ] ) is itself bounded by @xmath391which only depends on @xmath26 not on @xmath392 so that @xmath393 } \left\vert \hat{f}^{1}\left ( t , t+h\right ) \right\vert \notag \\
& \leq & 2^{\frac{p-1}{p}}c\left ( \alpha , p\right ) \left [ \sum_{i=0}^{\infty } \left ( i+1\right ) ^{\alpha p}\sum_{k=0}^{2^{i+n\left ( h\right ) + 1}-1}\left\vert \hat{f}^{1}\left ( \frac{k}{2^{n\left ( h\right ) + i}},\frac{% k+1}{2^{n\left ( h\right ) + i}}\right ) \right\vert ^{p}\right ] ^{\frac{1}{p}}. \label{sup - bound}\end{aligned}\ ] ] 2 .
@xmath394 where @xmath67 $ ] .
+ let @xmath128 be a dissection of @xmath106 $ ] . we may then rewrite @xmath128 as @xmath395where@xmath396and@xmath397since @xmath398and by jensen s inequality , we have @xmath399note that @xmath400contains pairwise disjoint intervals as do @xmath401and @xmath402hence , applying ( [ p - bound for continuous functions ] ) to every term in sum @xmath403 and then summing
, there is no double counting of increments over dyadics so that@xmath404the same argument applies to sum @xmath2 , sum @xmath405 and sum @xmath128 .
hence,@xmath406but this bound does not depend on the particular dissection @xmath128 chosen , so that @xmath407 if we now shift @xmath408 \right ) $ ] back into @xmath409 \right ) $ ] by replacing @xmath31 by @xmath410 we retrieve @xmath37 , and so taken together , @xmath411 and @xmath412 prove the proposition .
[ cor : corollary to propn : pathwise p - varn bound]suppose that @xmath413 and that @xmath355 \right ) $ ] . if @xmath227 and @xmath226 , then @xmath414bounds both @xmath415and @xmath416and hence @xmath417@xmath362 is defined in section 2.1 , equation ( c(alpha , p)-definition ) .
let @xmath418 and @xmath419 , so that @xmath420then apply proposition [ propn : pathwise p - varn bound ] .
we now prove the following hlder - type inequality : [ hoelder s condition]suppose that @xmath422 .
if @xmath423 , @xmath226 and @xmath424 , then there is a finite constant @xmath425 such that @xmath426 \leq d\left ( \alpha , \beta , p,\acute{p}\right ) \left ( h_{2}-h_{1}\right ) ^{\left ( \frac{1}{2}-\frac{1}{p% } \right ) \acute{p}}.\]]here , @xmath427 ^{\acute{p% } } c\left ( \beta , \frac{\acute{p}}{p}\right ) ^{\frac{\acute{p}}{p}}\sqrt{% \frac{2^{\acute{p}}}{\pi } } \gamma \left ( \frac{\acute{p}+1}{2}\right ) \sum_{n=0}^{\infty } \left ( n+1\right ) ^{\alpha \acute{p}+\beta \frac{\acute{p% } } { p}}2^{n\left ( \frac{1}{p}-\frac{1}{2}\right ) \acute{p}}.\]]if @xmath428 ,
@xmath429 \leq d\left ( \alpha , p\right ) \left ( h_{2}-h_{1}\right ) ^{\left ( \frac{1}{2}-\frac{1}{p}\right ) p}.\]]if @xmath430 @xmath426 \leq d\left ( \alpha , p\right ) ^{\frac{\acute{p}}{p}}\left ( h_{2}-h_{1}\right ) ^{\left ( \frac{1}{2% } -\frac{1}{p}\right ) \acute{p}}\]]here,@xmath431@xmath362 is defined in section 2.1 , equation ( c(alpha , p)-definition ) .
we are interested in the @xmath432moment of @xmath433since @xmath7 , the set @xmath209 \right ) _ { p}$ ] has full @xmath65-measure ( theorem 9 in @xcite ) , so that we may work on @xmath209 \right ) _ { p}$ ] instead of @xmath236 \right ) $ ] .
we fix @xmath166 \right ) _ { p}$ ] and consider the sample path @xmath175 of @xmath434 : for any @xmath67 $ ] , @xmath435 \right ) _ { p},\left\vert .\right\vert _ { p}\right ) .\]]let @xmath436 .
we first consider the case where @xmath437 . by corollary [ cor :
corollary to propn : pathwise p - varn bound ] , @xmath438 ^{\acute{p } } \\ & & \left ( \sum_{n=0}^{\infty } \underset{:=b_{n}}{\underbrace{\left ( n+1\right ) ^{\alpha p}\sum_{k=0}^{2^{n+n\left ( h\right ) + 1}-1}\left\vert f\left ( \frac{k+1}{2^{n+n\left ( h\right ) } } -1\right ) -f\left ( \frac{k}{% 2^{n+n\left ( h\right ) } } -1\right ) \right\vert ^{p}}}\right ) ^{\frac{\acute{p}% } { p}}.\end{aligned}\]]by lemma [ lemma : l_p bound],@xmath439furthermore , by jensen s inequality,@xmath440hence,@xmath438 ^{\acute{p}}c\left ( \beta , \frac{\acute{p}}{p}\right ) ^{\frac{\acute{% p}}{p } } \\ & & \sum_{n=0}^{\infty } \left ( n+1\right ) ^{\alpha \acute{p}+\beta \frac{% \acute{p}}{p}}\left ( 2^{n+n\left ( h\right ) + 1}\right ) ^{\frac{\acute{p}}{p}% -1 } \\ & & \sum_{k=0}^{2^{n+n\left ( h\right ) + 1}-1}\left\vert f\left ( \frac{k+1}{% 2^{n+n\left ( h\right ) } } -1\right ) -f\left ( \frac{k}{2^{n+n\left ( h\right ) } } % -1\right ) \right\vert ^{\acute{p}}.\end{aligned}\ ] ] in order to calculate the moment bound , we recall that @xmath441hence , we get@xmath442 = % \sqrt{\frac{2^{\acute{p}}}{\pi } } \gamma \left ( \frac{\acute{p}+1}{2}\right ) \left\vert t - s\right\vert ^{\frac{\acute{p}}{2}}. \label{gaussian moments}\]]so by ( [ gaussian moments ] ) , @xmath443 \\ &
\leq & 2^{\frac{\acute{p}}{2}}\left [ 2\left ( 2 + 2^{-\frac{1}{p}}\right ) c\left ( \alpha , p\right ) \right ] ^{\acute{p}}c\left ( \beta , \frac{\acute{p}}{% p}\right ) ^{\frac{\acute{p}}{p}}\sqrt{\frac{2^{\acute{p}}}{\pi } } \gamma \left ( \frac{\acute{p}+1}{2}\right ) \\ & & \sum_{n=0}^{\infty } \left ( n+1\right ) ^{\alpha \acute{p}+\beta \frac{% \acute{p}}{p}}\left ( 2^{n+n\left ( h\right ) + 1}\right ) ^{\frac{\acute{p}}{p}-% \frac{\acute{p}}{2 } } \\ & \leq & d\left ( \alpha , \beta , p,\acute{p}\right ) h^{\left ( \frac{1}{2}-\frac{% 1}{p}\right ) \acute{p}}\end{aligned}\]]where@xmath444 ^{\acute{p}}c\left ( \beta , \frac{\acute{p}}{p}% \right ) ^{\frac{\acute{p}}{p}}\sqrt{\frac{2^{\acute{p}}}{\pi } } \gamma \left ( \frac{\acute{p}+1}{2}\right ) \\ & & \times \sum_{n=0}^{\infty } \left ( n+1\right ) ^{\alpha \acute{p}+\beta \frac{\acute{p}}{p}}2^{n\left ( \frac{\acute{p}}{p}-\frac{\acute{p}}{2}% \right ) } \end{aligned}\]]which is finite if @xmath7 , @xmath226 and @xmath445 next , we consider the case where @xmath428 : by corollary [ cor : corollary to propn : pathwise p - varn bound],@xmath446using ( [ gaussian moments ] ) we find that @xmath447 \\ & \leq & 2^{\frac{p}{2}}\left ( 4 + 2^{\frac{p-1}{p}}\right ) ^{p}c\left ( \alpha , p\right ) ^{p}\sqrt{\frac{2^{p}}{\pi } } \gamma \left ( \frac{p+1}{2}\right ) \sum_{n=0}^{\infty } \left ( n+1\right ) ^{\alpha p}2^{\left ( n+n\left ( h\right ) + 1\right ) \left ( 1-\frac{p}{2}\right ) } \\ & \leq & \underset{=d\left ( \alpha , p\right ) } { \underbrace{\left ( 2^{\frac{p}{2% } } \left ( 4 + 2^{\frac{p-1}{p}}\right ) ^{p}c\left ( \alpha , p\right ) ^{p}\sqrt{% \frac{2^{p}}{\pi } } \gamma \left ( \frac{p+1}{2}\right ) \right ) \sum_{n=0}^{\infty } \left ( n+1\right ) ^{\alpha p}2^{n\left ( 1-\frac{p}{2}% \right ) } } } h^{\frac{p}{2}-1}.\end{aligned}\]]finally , if @xmath448 , by lyapunov s inequality we have that @xmath449 \\ & = & \mathbb{e}\left [ \left\vert \mathcal{t}_{h_{2}}^{b , p}-\mathcal{t}% _ { h_{1}}^{b , p}\right\vert _ { p}^{p\frac{\acute{p}}{p}}\right ] \\ & \leq & \mathbb{e}\left [ \left\vert \mathcal{t}_{h_{2}}^{b , p}-\mathcal{t}% _ { h_{1}}^{b , p}\right\vert _ { p}^{p}\right ] ^{\frac{\acute{p}}{p}}\leq d\left ( \alpha , p\right ) ^{\frac{\acute{p}}{p}}\left ( h_{2}-h_{2}\right ) ^{\left ( \frac{1}{2}-\frac{1}{p}\right ) \acute{p}}.\end{aligned}\ ] ]
we recall _ kolmogorov s lemma _ ( e.g. @xcite ) : [ thm : kolmogorov s lemma]let @xmath450 , @xmath451 ^{d}$ ] , be a banach space @xmath5-valued process for which there exist three strictly positive constants @xmath142,@xmath452,@xmath146 such that @xmath453 \leq c\left\vert t - s\right\vert ^{d+\varepsilon } ; \]]then the process@xmath454 ^{d}\right ) : u\geq t}x_{u}\]]is a continuous modification is called a modification of @xmath455 if for any @xmath451 ^{d}$ ] we have @xmath456 . ] of @xmath457 .
[ continuous version]for @xmath7 , the process @xmath458 \rightarrow \left ( c\left ( \left [ 0,1\right ] \right ) _ { p},\left\vert .\right\vert _ { p}\right ) $ ] has a continuous modification
the brownian frame process whose evaluation at time @xmath451 $ ] is @xmath459 \right ) : u\geq t}\mathcal{t}_{u}^{b , p}. \label{frame process definition}\ ] ] applying proposition [ hoelder s condition ] with @xmath460 and @xmath461we see that @xmath462 satisfies the conditions of kolmogorov s lemma so that ( [ frame process definition ] ) gives a continuous modification of @xmath462 .
[ defn : dyadic polygonal approximation]the dyadic polygonal approximations @xmath463 to a banach space @xmath5 valued path @xmath457 are defined as @xmath464here , @xmath465 [ lyons holder proposition]suppose @xmath466 is a continuous banach space @xmath5 valued stochastic process on a completed probability space @xmath467 for which there are constants @xmath468 , @xmath469 such that @xmath470 as well as a third constant @xmath405 , such that @xmath471 \leq c\left\vert t - s\right\vert ^{\kappa \acute{p}}\text { } \forall s , t\in % \left [ 0,1\right ] .
\label{lyons ' holder condition}\]]then the dyadic polygonal approximations @xmath472 have finite @xmath195-variation uniformly in @xmath473 , @xmath65-a.s . furthermore
, @xmath474 converges to @xmath160 in @xmath9-variation @xmath176-a.s . hence , for @xmath475 the @xmath6-variation norm frame process@xmath476 \rightarrow \left ( c\left ( \left [ 0,1% \right ] \right ) _ { p},\left\vert .\right\vert _ { p}\right)\]]has finite @xmath9-variation for @xmath477 @xmath176-a.s .
furthermore , the dyadic polygonal approximations @xmath478 as defined in ( [ dyadic polygonal approximations ] ) converge to @xmath181 in @xmath9-variation @xmath65-a.s . by proposition
[ hoelder s condition ] , for appropriate choices of @xmath479 and @xmath480 and @xmath422 , we have that @xmath481 \\ & \leq & \left\ { \begin{array}{ll } d\left ( \alpha , \beta , p,\acute{p}\right ) \left ( h_{2}-h_{1}\right ) ^{\left ( \frac{1}{2}-\frac{1}{p}\right ) \acute{p } } & \text{if } \acute{p}>p>2 \\ d\left ( \alpha , p\right ) \left ( h_{2}-h_{1}\right ) ^{\left ( \frac{1}{2}-% \frac{1}{p}\right ) p } & \text{if } \acute{p}=p>2 \\ d\left ( \alpha , p\right ) ^{\frac{\acute{p}}{p}}\left ( h_{2}-h_{1}\right ) ^{\left ( \frac{1}{2}-\frac{1}{p}\right ) p } & \text{if } p>\acute{p}>2% \end{array}% \right .
.\end{aligned}\]]since @xmath173 is by definition continuous and takes its values in the separable banach space @xmath165 \right ) _
{ p},\left\vert .\right\vert _ { p}\right ) $ ] , by choosing @xmath178 , we may apply proposition [ lyons holder proposition ] to @xmath181 with @xmath482 and @xmath483 [ propn : finite p - varn and convergence of dyad poly ] @xmath0 has finite @xmath9-variation for @xmath11 , @xmath65-a.s .
furthermore , the dyadic polygonal approxmiations @xmath484 as defined in ( [ dyadic polygonal approximations ] ) converge to @xmath0 in @xmath9-variation @xmath65-a.s .
fix @xmath286 and @xmath485 .
then fix @xmath486 @xmath487 so large that @xmath488since @xmath489 , using proposition [ hoelder s condition ] , we have that @xmath490 \\ & \leq & \mathbb{e}\left [ \left\vert \mathcal{t}_{h_{2}}^{b}-\mathcal{t}% _ { h_{1}}^{b}\right\vert _ { p}^{\acute{p}}\right ] \\ & = & d\left ( \alpha , p\right ) ^{\frac{\acute{p}}{p}}\left ( h_{2}-h_{1}\right ) ^{\left ( \frac{1}{2}-\frac{1}{p}\right ) \acute{p}}.\end{aligned}\]]by choice @xmath491so that the statement of proposition [ lyons holder proposition ] applies with @xmath482 on the set @xmath492 of full @xmath65-measure .
the statement of this proposition then holds on the set @xmath493 of full @xmath494-measure .
as before , we work on the wiener space @xmath71 \right ) , \sigma _ { \left\vert .\right\vert _ { \infty } } , \mathbb{p}\right ) $ ] .
@xmath496 \right ) $ ] denotes the sobolev space of differentiable functions with derivative in @xmath497 \right ) $ ] , i.e.@xmath498 \right ) : = \left\ { f\in c_{0}\left ( \left [ -1,1\right ] \right ) : \frac{df}{dx}\in l^{2}\left ( \left [ -1,1\right ] \right ) \right\ } \text{,}\]]equipped with its hilbert space norm@xmath499 \right ) } : = \left\vert \frac{df}{dx}\right\vert _ { l^{2}\left ( \left [ -1,1\right ] \right ) } \text{.}\]]in the wiener space setting , @xmath500 \right ) , \left\vert .\right\vert _ { w^{1,2}\left ( \left [ -1,1\right ] \right ) } \right ) $ ] is also known as the _ cameron - martin space_. @xmath501 \right ) } = \mathcal{o}$ ] denotes the closed unit ball of the _ cameron - martin space _ , i.e.@xmath502 \right ) : \left\vert \frac{df}{dx}\right\vert _ { l^{2}\left ( \left [ -1,1\right ] \right ) } \leq 1\right\ } .\]]@xmath503 denotes the continuous imbedding map @xmath504 \right ) , \left\vert .\right\vert _ { w^{1,2}\left ( \left [ -1,1\right ] \right ) } \right ) \hookrightarrow \left ( c_{0}\left ( \left [ -1,1\right ] \right ) , \left\vert .\right\vert _ { \infty } \right ) .\]]to see that @xmath505 is continous , fix @xmath264 and choose @xmath506 \right ) $ ] such that @xmath507 \right ) } < \frac{% \varepsilon } { \sqrt{2}}$ ] .
then@xmath508 } \left\vert j\left ( f\right ) \left ( t\right ) -j\left ( g\right ) \left ( t\right ) \right\vert \\ & = & \sup_{t\in \left [ -1,1\right ] } \left\vert \int_{-1}^{t}\left ( \frac{df}{du% } -\frac{dg}{du}\right ) du\right\vert \\ & \leq & \sup_{t\in \left [ -1,1\right ] } \left\vert t+1\right\vert ^{\frac{1}{2}% } \left [ \int_{-1}^{1}\left ( \frac{df}{du}-\frac{dg}{du}\right ) ^{2}du\right ] ^{\frac{1}{2 } } \\ & = & \sup_{t\in \left [ -1,1\right ] } \left\vert t+1\right\vert ^{\frac{1}{2}% } \left\vert f - g\right\vert _ { w^{1,2}\left ( \left [ -1,1\right ] \right ) } < \varepsilon .\end{aligned}\]]hence
, @xmath509 is @xmath510 \right ) } } $ ] -measurable where @xmath511 \right ) } } $ ] denotes the borel-@xmath64-algebra on @xmath496 \right ) .$ ] we say that a borel random variable @xmath512 on wiener space into some banach space @xmath513 is lipschitz in the direction of the cameron - martin space ( or @xmath514 \right ) $ ] _ _ -lipschitz _ _ ) _ _ _ _ if @xmath515 \right ) } \left ( \mathcal{f}\right ) : = \sup_{x\in c_{0}\left ( \left [ -1,1\right ] \right ) } \sup_{r>0}\sup_{y\in j\left ( \mathcal{o}\right ) \setminus \left\ { 0\right\ } } \frac{\left\vert \mathcal{f}\left ( x+ry\right ) -\mathcal{f}\left ( x\right ) \right\vert _ { v}}{r% } \]]is finite .
@xmath516 \right ) } \left ( \mathcal{f}% \right ) $ ] is called the lipschitz - norm of @xmath512 ( in the direction of @xmath496 \right ) $ ] ) .
@xmath517 denotes the cumulative distribution function of the gaussian measure on @xmath232 , i.e.@xmath518if @xmath403 and @xmath2 are subsets of a vector space @xmath5 , we define @xmath519and for any @xmath520,@xmath521a median @xmath522 of a real - valued random variable @xmath160 is defined as a value in @xmath232 with the property that@xmath523 in this chapter , we are concerned with the @xmath6-variation frame process @xmath179 ( @xmath7 ) as defined in ( [ frame process definition ] ) .
we find two constants @xmath182 ( where @xmath524 ) and @xmath185 so that the random variable @xmath186has gaussian tails , i.e.@xmath525a concentration of measure result for wiener space ( due to borell and independently roughly at the same time to sudakov _ _ _ _ and tsirelson ) is applied to the functional @xmath526 on wiener space in order to derive the above bound .
the following inequality due to _ borell _ ( theorem 3.1 in @xcite _ _ _ _ and _ _ _ _ @xcite ) _ _ _ _ is at the heart of our argument we refer to it as _ borell s inequality : _ if @xmath403 @xmath527 and @xmath528 , then we have the following lower bound for the enlargement of @xmath403 in the direction of the unit ball in the cameron - martin space @xmath529:@xmath530 it is a fact that @xmath531 is in @xmath341 ( @xcite ) .
we are interested in applying borell s inequality to real - valued non - negative random variables on wiener space .
[ lemma : deviation inequality ] suppose that @xmath512 is a non - negative random variable that has finite lipschitz norm in the direction of @xmath496 \right ) $ ] , i.e. @xmath515
\right ) } \left ( \mathcal{f}\right ) < \infty .\]]then@xmath532 + rlip_{w^{1,2}\left ( \left [ -1,1\right ] \right ) } \left ( \mathcal{f}\right ) \right ) \leq \frac{1}{\sqrt{2\pi } r}\exp \left ( -\frac{r^{2}}{2}\right ) .
\label{deviation inequality}\ ] ] let @xmath533 denote the median of @xmath512 . by _ borell s inequality , _
@xmath534but @xmath535 \right ) : \mathcal{f}% \left ( y\right ) \leq m_{\mathcal{f}}+rlip_{w^{1,2}\left ( \left [ -1,1\right ] \right ) } \left ( \mathcal{f}\right ) \right\}\end{aligned}\]](which is a borel set ) . since @xmath512 is _ non - negative _ , we have that @xmath536 \geq \int_{\mathcal{f}\geq m_{f}}% \mathcal{f}dp\geq \frac{1}{2}m_{\mathcal{f}}.\]]therefore , @xmath537 + rlip_{w^{1,2}\left ( \left [ -1,1\right ] \right ) } \left ( \mathcal{f}\right ) \right ) \geq \phi \left ( r\right ) .\]]since we have that @xmath538the result follows . we start with an observation on how to bound the lipschitz - norm of a semi - norm : [ lemma : lipschitz - norm of seminorm functional]if @xmath512 is a seminorm , then@xmath515 \right ) } \left ( \mathcal{f}\right ) \leq \sup_{y\in \mathcal{o}}\left\vert \mathcal{f}\left ( j\left ( y\right ) \right ) \right\vert .\ ] ] this follows from the triangle inequality .
next , we find the exact lipschitz - norm for the non - negative functional @xmath515 \right ) } \left ( \left\vert \mathcal{t}% _ { h_{2}}-\mathcal{t}_{h_{1}}\right\vert _ { \infty } \right ) .\ ] ] [ lemma : lipschitz bound frame process sup norm]for any fixed @xmath413,@xmath515 \right ) } \left ( \left\vert \mathcal{t}% _ { h_{2}}-\mathcal{t}_{h_{1}}\right\vert _ { \infty } \right ) = \left ( h_{2}-h_{1}\right ) ^{\frac{1}{2}}\text{.}\ ] ] we fix @xmath540 \right ) $ ] and note that @xmath541hence,@xmath393 } \left\vert \left ( \mathcal{t}% _ { h_{2}}^{\int_{-1}^{.}f\left ( s\right ) ds}-\mathcal{t}_{h_{1}}^{% \int_{-1}^{.}f\left ( s\right ) ds}\right ) \left ( t\right ) \right\vert \\ & = & \sup_{t\in \left [ 0,1\right ] } \left\vert \int_{h_{1}-1+t}^{h_{2}-1+t}f\left ( s\right ) ds\right\vert \\ & \leq & \left ( h_{2}-h_{1}\right ) ^{\frac{1}{2}}\left\vert f\right\vert
_ { l^{2}\left ( \left [ -1,1\right ] \right ) } .\end{aligned}\]]furthermore , @xmath542 is a seminorm on wiener space so that using lemma [ lemma : lipschitz - norm of seminorm functional ] , we have shown that @xmath515 \right ) } \left ( \left\vert \mathcal{t}% _ { h_{2}}-\mathcal{t}_{h_{1}}\right\vert _ { \infty }
\right ) \leq \left ( h_{2}-h_{1}\right ) ^{\frac{1}{2}}.\ ] ] in order to show that @xmath516 \right ) } \left ( \left\vert \mathcal{t}_{h_{2}}-\mathcal{t}_{h_{1}}\right\vert _ { \infty } \right ) $ ] is at least @xmath543 , we consider @xmath544 } , \]]and note that @xmath545 . furthermore
, the supremum@xmath393 } \left\vert \left ( \mathcal{t}% _ { h_{2}}^{\int_{-1}^{.}g\left ( s\right ) ds}-\mathcal{t}_{h_{1}}^{% \int_{-1}^{.}g\left ( s\right ) ds}\right ) \left ( t\right ) \right\vert \\ & = & \frac{1}{\sqrt{h_{2}-h_{1}}}\sup_{t\in \left [ 0,1\right ] } \left\vert \int_{h_{1}-1+t}^{h_{2}-1+t}\mathbf{1}_{\left [ h_{1}-1\leq s < h_{2}-1\right ] } ds\right\vert\end{aligned}\]]is attained at @xmath546 and is equal to @xmath547 which completes the proof .
we find an upper bound for the lipschitz - norm of @xmath548 : [ lemma : lipschitz bound frame process p - functional]for @xmath7 and any fixed pair @xmath549 and @xmath550 with @xmath422,@xmath515 \right ) } \left ( \mathcal{v}_{p}\left ( \mathcal{t}_{h_{2}}^{.}-\mathcal{t}_{h_{1}}^{.}\right ) \right ) \leq d_{p}\left ( h_{2}-h_{1}\right ) ^{\frac{1}{2}-\frac{1}{p}},\]]where@xmath551 we fix a dissection @xmath552 of @xmath231 $ ] and @xmath540 \right ) $ ] . using the representation ( [ reexpressing frame process eval on w1,2 ] ) , we find that@xmath553for any dissection piece @xmath554 $ ] , by the cauchy - schwarz inequality , we have that @xmath555 \right ) } \min \left ( \left ( t_{i+1}-t_{i}\right ) ^{\frac{1}{2}},\left ( h_{2}-h_{1}\right ) ^{\frac{1}{2}}\right ) .\end{aligned}\]]hence,@xmath556
\right ) } ^{p}\sum_{t_{i},t_{i+1}\in d}\min \left ( \left ( t_{i+1}-t_{i}\right ) ^{\frac{p}{2}},\left ( h_{2}-h_{1}\right ) ^{\frac{p}{2}% } \right ) .\end{aligned}\]]as in the proof of proposition [ propn : pathwise p - varn bound ] in chapter 2 , we write @xmath128 as a disjoint union of @xmath557 and @xmath558 where @xmath419 and @xmath559and @xmath560accordingly , we split the sum:@xmath561since there are at most @xmath562 terms in sum @xmath403 ( an interval of length @xmath164 can contain at most @xmath562 disjoint intervals of length at least @xmath26 ) , sum @xmath403 is bounded by @xmath563 . in order to bound sum @xmath2 , we start by listing the dissection pieces with endpoints in @xmath557 : @xmath564 , \left [ t_{i_{2}},t_{i_{2}+1}% \right ] , ... , \left [ t_{i_{m}},t_{i_{m}+1}\right ] \right\ } .\]]we now consider the following subsequence of @xmath565 : let @xmath566 be the first element in the sequence such that the summed length of all dissection pieces @xmath567 $ ] with index @xmath568 up to and including @xmath569 is at least @xmath26 and at most @xmath570 ( if no such index exists , then the sum of all dissection pieces in @xmath557 adds up to strictly less than @xmath26 , so that @xmath571 ) . in this way , recursively define @xmath572 such that the overall length of all dissection pieces whose endpoints are included in @xmath573 with index @xmath568 between @xmath574 and @xmath575 is at least @xmath26 and at most @xmath570 .
since an interval of length @xmath164 can contain at most @xmath562 disjoint intervals of length at least @xmath26 , the subsequence @xmath576 thus defined has at most @xmath577 elements . furthermore since @xmath7 , for every @xmath578@xmath579hence,@xmath580therefore,@xmath515 \right ) } \left ( \mathcal{v}_{p}\left ( \mathcal{t}_{h_{2}}-\mathcal{t}_{h_{1}}\right ) \right ) \leq \underset{:=d_{p}% } { \underbrace{2^{\frac{1}{2}+\frac{1}{p}}\left ( 1 + 2^{\frac{p}{2}}\right ) ^{% \frac{1}{p}}}}\left ( h_{2}-h_{1}\right ) ^{\frac{1}{2}-\frac{1}{p}}.\ ] ] if @xmath7 , @xmath581 and @xmath582 , then @xmath583where @xmath584and @xmath585@xmath586 is the constant given in proposition hoelder s condition .
@xmath118 is the constant given in lemma [ lemma : lipschitz bound frame process p - functional ] .
we recall ( [ deviation inequality ] ) from lemma [ lemma : deviation inequality],@xmath532 + rlip_{w^{1,2}\left ( \left [ -1,1\right ] \right ) } \left ( \mathcal{f}\right ) \right ) \leq \frac{1}{\sqrt{2\pi } r}\exp \left ( -\frac{r^{2}}{2}\right ) , \]]which we will apply to @xmath587 noting in what follows that @xmath181 is a continuous modification of @xmath261 . from chapter 2 ,
proposition hoelder s condition , we recall that @xmath588 \leq d\left ( \alpha , p\right ) \left ( h_{2}-h_{1}\right ) ^{\left ( \frac{1}{2}-\frac{1}{p}\right ) p},\]]so that by lyapunov s inequality@xmath589
\leq \mathbb{e}\left [ \left\vert \mathcal{s}_{h_{2}}^{b , p}-\mathcal{s}_{h_{1}}^{b , p}\right\vert _ { p}^{p}% \right ] ^{\frac{1}{p}}\leq d\left ( \alpha , p\right ) ^{\frac{1}{p}}\left ( h_{2}-h_{1}\right ) ^{\frac{1}{2}-\frac{1}{p}}.\]]from lemma [ lemma : lipschitz bound frame process sup norm ] and lemma _ _ _ _ [ lemma : lipschitz bound frame process p - functional ] , we have that on a set of full @xmath65-measure @xmath590 \right ) } \left ( \left\vert \mathcal{s}% _ { h_{2}}^{b , p}-\mathcal{s}_{h_{1}}^{b , p}\right\vert _ { p}\right ) & = & lip_{w^{1,2}\left ( \left [ -1,1\right ] \right ) } \left ( \left\vert \mathcal{% t}_{h_{2}}-\mathcal{t}_{h_{1}}\right\vert _ { p}\right ) \\ &
\leq & 2d_{p}\left ( h_{2}-h_{1}\right ) ^{\frac{1}{2}-\frac{1}{p}}.\end{aligned}\]]hence,@xmath591so that @xmath592
as in previous chapters , @xmath224 \right ) $ ] denotes the banach space of real - valued continuous functions equipped with the @xmath8-norm .
similarly , @xmath224 \times \left [ 0,1\right ] \right ) $ ] denotes the banach space of real - valued continuous functions on the unit square equipped with the @xmath8-norm .
if @xmath599 $ ] , then @xmath600 denotes the evaluation functional at @xmath19 , that is for any continuous function @xmath601 \rightarrow \mathbb{r}$ ] , @xmath602a _ partition _ of @xmath106 $ ] is defined as a countable collection @xmath603 such that @xmath604 if @xmath605 and @xmath606 .
the _ total variation _ of a signed borel measure @xmath607 on @xmath106 $ ] is defined as @xmath608 \right ) } : = \sup \left\ { \sum_{i=1}^{\infty } \left\vert \mu \left ( \left [ t_{i},t_{i+1}% \right ) \right ) \right\vert : \text{all partitions of } \left [ 0,1\right ] \right\ } .\ ] ] from proposition [ propn : finite p - varn and convergence of dyad poly ] in chapter 2 we know that the @xmath8-norm frame process @xmath0 has finite @xmath9-variation if @xmath11 .
theorem firsttheoremlyons in chapter 1 tells us that in order to establish a rough path integration theory for @xmath0 , we need to find a lift of @xmath202 to @xmath611 ( where @xmath612 \right ) $ ] ) .
proposition [ propn : finite p - varn and convergence of dyad poly ] furthermore asserts that the dyadic polygonal approximations @xmath196 converge to @xmath0 in @xmath9-variation norm .
chen s theorem ( theorem [ theorem : chen s theorem])gives us a multiplicative lift for every @xmath473 .
thus , one may be led to ask whether @xmath0 has a lift that is the limit of the dyadic polygonal smooth rough paths associated to the sequence @xmath613 . in this chapter , we show that this is _ not _ the case : if we consider the injective tensor product @xmath199 , while @xmath614takes its values in @xmath199 for every fixed @xmath473 , the limit as @xmath615 does not .
the reason is the following : while @xmath198 is shown to be isomorphic to @xmath224 \times % \left [ 0,1\right ] \right ) $ ] , we prove that in the limit as @xmath616 , the antisymmetric component of ( [ eqn : chen lift 2-tensor ] ) given by @xmath617exists and is continuous off the diagonal of the unit square . however , on the diagonal it fails to be continuous .
as we will see in the proof of proposition [ propn : representation of levy area random variable ] and proposition [ propn : discontinuity of la ] below , this breakdown in continuity is intimately linked to the fact that the lvy area of the frame process @xmath0 picks up the quadratic variation of the brownian sample path @xmath2 .
this result has the following geometric interpretation : from lvy s modulus of continuity ( e.g. @xcite ) , we know that for @xmath65-a.e . sample path @xmath166 \right ) $ ] , @xmath618hence , the planar path @xmath619converges uniformly to the path @xmath620 at a rate @xmath621 .
the lvy area ( example [ ex : r^2 ] below ) of @xmath622 is @xmath61 .
however , we will see that for @xmath623 , even though @xmath624 uniformly , the lvy area of the path @xmath625 converges to @xmath626 .
so the random path sequence @xmath627 exhibits a similar behaviour to the deterministic path sequence @xmath628 known from example 1.1.1 in @xcite : although @xmath629 converges to @xmath630 uniformly , the associated sequence of lvy areas converges to @xmath631 .
similarly , @xmath632 converges uniformly to a process taking values on the diagonal which does not generate area .
but the associated sequence of lvy areas tends to @xmath626 .
[ [ the - injective - tensor - algebra - cleft - left-01right - rightotimes-_vee - cleft - left-01right - right ] ] the injective tensor algebra @xmath224 \right ) \otimes _ { \vee } c\left ( \left [ 0,1\right ] \right ) $ ] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ suppose that @xmath5 is a banach space with topological dual @xmath596 .
the algebraic tensor product of @xmath5 with itself denoted as @xmath12 is the set of all elements @xmath633 where @xmath634 @xmath635 and @xmath95 is finite : for any @xmath636 , we have that @xmath637 the following proposition asserts that any element @xmath638 identifies a unique finite rank element in @xmath639 whose action on an element @xmath640 is given by @xmath641 since @xmath596 is also a banach space , by replacing @xmath5 by @xmath596 in the above proposition , every element of @xmath645 identifies a unique finite rank element of @xmath646 . in this way
, we define a natural duality pairing between @xmath645 and @xmath12 given by @xmath647 [ defn : symmetric + antisymmetric two - tensor]a tensor @xmath648 @xmath649 is called _ symmetric _ , if for any @xmath650 , we have @xmath651it is called _ antisymmetric _ if for any @xmath650 , we have @xmath652 [ ex : injective tensor product norm]suppose @xmath5 is a banach space . the closure of @xmath12 in @xmath644 is called the _ injective tensor product _ denoted as @xmath662 . as a consequence of the hahn - banach theorem ( theorem 4.3 b ) in rudin91 ) , for any @xmath663 , @xmath664hence , the injective tensor product norm is given by@xmath665 the injective tensor product norm is an example of a cross - norm and hence of a compatible norm .
it has the additional property that it is the smallest cross norm whose dual norm @xmath666 on @xmath667 given by@xmath668 is also cross ( @xcite ) .
[ [ a - representation - result - for - cleft - left-01right - rightotimes-_vee - cleft - left-01right - right ] ] a representation result for @xmath224 \right ) \otimes _ { \vee } c\left ( \left [ 0,1\right ] \right ) $ ] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ [ proposition : injective tensor algebra]the injective tensor product @xmath200 \right ) \otimes _ { \vee } c\left ( \left [ 0,1\right ] \right ) $ ] is isomorphic to the space of continuous functions on the unit square , i.e. @xmath670
\right ) \otimes _ { \vee } c\left ( \left [ 0,1\right ] \right ) \cong c\left ( \left [ 0,1\right ] \times \left [ 0,1\right ] \right ) .\ ] ] suppose @xmath671 \right ) \otimes _ { \vee } c\left ( % \left [ 0,1\right ] \right ) $ ] has a representation@xmath672by the _ radon - riesz theorem _ ( theorem 6.19 in @xcite ) the topological dual of @xmath165 \right ) , \left\vert .\right\vert _ { \infty } \right )
$ ] is isomorphic to the space of borel measures of finite total variation
denoted by @xmath673 \right ) $ ] with the norm given by the total variation @xmath674 \right ) } $ ] .
thus , the injective tensor norm has the following form @xmath675 \right ) } \right\ } .\]]but if @xmath676 \right ) } $ ] , then the product measure @xmath677 is in @xmath678 \times \left [ 0,1\right ] \right ) } $ ] .
hence,@xmath679 \right ) } \left ( 1\right ) \right\ } \\
& \leq & \sup \left\ { \sum_{i=1}^{n}\int_{0}^{1}\int_{0}^{1}f_{i}\left ( u\right ) g_{i}\left ( v\right ) \gamma \left ( du , dv\right ) : \gamma \in \mathcal{o}_{m^{1}\left ( \left [ 0,1\right ] \times \left [ 0,1\right ] \right ) } \left ( 1\right ) \right\}\end{aligned}\]]using ( [ eqn : hahn - banach ] ) , it follows that@xmath680 \times \left [ 0,1\right ] \right ) } } \left\vert \int_{0}^{1}\int_{0}^{1}\sum_{i=1}^{n}f_{i}\left ( u\right ) g_{i}\left ( v\right ) \gamma \left ( du , dv\right ) \right\vert\]]hence,@xmath681to prove the other direction , we note that since @xmath682 \right ) $ ] for @xmath683 , @xmath684 and hence @xmath685 is in @xmath686 \times \left [ 0,1\right ] \right ) $ ] . but
@xmath106 \times \left [ 0,1\right ] $ ] is compact , so that by the bolzano - weierstrass theorem there exists a pair @xmath687 \times \left [ 0,1% \right ] $ ] such that@xmath688 , y\in \left [ 0,1\right ] \right\ } .\]]but @xmath689 and @xmath690 are in @xmath691 \right ) } $ ] so that @xmath692 , y\in \left [ 0,1\right ] \right\ } \\ & = & \sup \left\ { \left\vert \int_{0}^{1}\int_{0}^{1}w\left ( u , v\right ) \delta _ { x}\left ( u\right ) \delta _ { y}\left ( v\right ) dudv\right\vert : x\in \left [ 0,1\right ] , y\in \left [ 0,1\right ] \right\ } \\ &
\leq & \left\vert w\right\vert _ { \vee } , \end{aligned}\ ] ] hence,@xmath693this concludes the proof that the imbedding map @xmath694 \right ) \otimes c\left ( \left [ 0,1\right ] \right ) , \left\vert .\right\vert _ { \vee } \right ) \hookrightarrow \left ( c\left ( \left [ 0,1\right ] \times \left [ 0,1\right ] \right ) , \left\vert .\right\vert _ { \infty } \right)\]]is continuous and norm - preserving . to prove that @xmath695 \right ) \otimes c\left ( \left [ 0,1\right ] \right ) \right ) } = c\left ( \left [ 0,1\right ] \times \left [ 0,1% \right ] \right ) , \]]we note that @xmath696 \right ) \otimes c\left ( % \left [ 0,1\right ] \right ) \right ) $ ] contains the algebra @xmath403 of bivariate polynomials , i.e.@xmath697this algebra separates points on @xmath106 \times \left [ 0,1\right ] $ ] .
hence , we may invoke the stone - weierstrass theorem to deduce that@xmath698 \times \left [ 0,1\right ] \right)\ ] ] and since @xmath403 is contained in @xmath696 \right ) \otimes c\left ( \left [ 0,1\right ] \right ) \right ) $ ] , that@xmath670 \right ) \otimes _ { \vee } c\left ( \left [ 0,1\right ] \right ) = \overline{j\left ( c\left ( \left [ 0,1\right ] \right ) \otimes c\left ( % \left [ 0,1\right ] \right ) \right ) } = c\left ( \left [ 0,1\right ] \times \left [ 0,1\right ] \right ) .\ ] ] [ [ multiplicative - level-2-path - lifts - the - relevance - of - an - area - process ] ] multiplicative level @xmath164 path lifts : the relevance of an area process
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ suppose @xmath19 is a banach space @xmath5-valued path .
from example [ ex : level 1 multiplicative lift ] in chapter 1 , we know that @xmath19 has a canonical multiplicative lift to @xmath699 given by @xmath700 . our aim here is to construct a multiplicative lift of @xmath701 to @xmath192 : from ( [ symmetrisation ] ) we know that every two - tensor can be written as the sum of a symmetric and an antisymmetric two - tensor .
our method of construction is to choose a symmetric two - tensor process , @xmath702 , and contingent on the choice of @xmath702 , an antisymmetric two - tensor process @xmath703 , in such a way that @xmath704satisfies the multiplicative two - tensor condition , i.e. that for any @xmath705@xmath706any two - tensor process @xmath707 that satisfies ( [ multiplicative two - tensor condition ] ) determines a multiplicative functional @xmath708 in @xmath192 . if we choose @xmath709@xmath710 has to satisfy @xmath711for @xmath712 to obey ( [ multiplicative two - tensor condition ] ) .
if we choose @xmath715 as the symmetric component , then the problem of finding a multiplicative lift of @xmath19 to @xmath192 reduces to finding an area process @xmath716 : such a process is not unique ; different choices of @xmath710define different lifts of @xmath19 .
[ ex : r^2]suppose @xmath720 is a path in @xmath721 and that @xmath37 and @xmath327 are both of bounded variation . in this case , we have @xmath722and so if @xmath723 and @xmath724 represent an orthonormal basis of @xmath725 such that @xmath726then @xmath727@xmath728 has a geometric meaning : it is the oriented area enclosed by the path @xmath19 run from @xmath719 to @xmath31 and the chord between the points @xmath729 and @xmath730 . given a path @xmath19 , for each fixed @xmath95 , the dyadic polygonal approximation @xmath731 of @xmath19 ( c.f .
definition [ defn : dyadic polygonal approximation ] in chapter 2 ) has bounded variation and so defines a sequence @xmath732 of smooth rough paths ( definition [ defn : smooth rough path ] , chapter 1 ) given by chen s theorem ( theorem theorem : chen s theorem , chapter 1 ) : the two - tensor coordintate @xmath733 of @xmath732 is given by @xmath734the antisymmetric component of which is @xmath735thus , if @xmath19 has a lvy area , we get a second representation of the lvy area as @xmath736 [ [ non - existence - of - the - lvy - area - of - mathcaltb - in - the - injective - tensor - algebra ] ] non - existence of the lvy area of @xmath0 in the injective tensor algebra ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ let us assume for the moment that the lvy area @xmath737 of the frame process @xmath0 exists in the injective tensor algebra . by proposition [
proposition : injective tensor algebra ] , this is equivalent to the existence of a continuous process @xmath738 on the unit square , i.e. the antisymmetric two - tensor @xmath737 has to be a @xmath224 \times \left [ 0,1\right ] \right ) $ ] -valued random - variable . by definition
[ defn : symmetric + antisymmetric two - tensor ] , for any @xmath739 \times \left [ 0,1\right ] $ ] and @xmath65-a.e .
sample path @xmath740 \right ) $ ] , we have that@xmath741hence , since we assumed that @xmath742 is continuous on @xmath106 \times \left [ 0,1\right ] $ ] , we require that for any sequence @xmath743 through @xmath106 , $ ] @xmath744however , we show that for any fixed @xmath745 $ ] and any sequence @xmath746 , for @xmath65-a.e . sample path @xmath66 \right ) $ ] , @xmath747thus contradicting ( [ continuity condition ] ) and hence disproving our initial assumption that @xmath201 exists in the injective tensor product .
[ [ existence - of - the - mathbbr - valued - random - variable - mathcala_01left - mathcaltbright - left - stright - for - dyadic - s - and - t ] ] existence of the @xmath232-valued random variable @xmath748 for dyadic @xmath719 and @xmath749 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ we show the existence of @xmath201 on the dyadic unit square @xmath243 ^{2}\right ) $ ] . by ( eqn : antisymmetry ) , if we have defined @xmath750 on @xmath751 , we have defined it on the entire unit square ( @xmath752 for @xmath745 $ ] ) . for @xmath65-a.e .
@xmath166 \right ) $ ] and @xmath753 ^{2}\right ) $ ] , the limit @xmath754exists .
moreover , on the open upper dyadic triangle @xmath755 ^{2}\right ) : 0\leq s
< t\leq 1\right\}\]](which _ excludes _ the diagonal ) , @xmath756has the following representation:@xmath757the integrals are understood in an it sense and the brownian motion @xmath758 is defined as @xmath759 .\]]by ( [ eqn : antisymmetry ] ) , this defines @xmath760 on the open lower dyadic triangle , @xmath755 ^{2}\right ) : 0\leq t < s\leq 1\right\ } , \]]via @xmath761on the dyadic diagonal @xmath762 \right ) \right\ } $ ] , @xmath763 over points @xmath767 in the upper dyadic triangle of the unit square as in ( [ eqn : first area representation ] ) , is equivalent to summing terms of the form @xmath768over points @xmath767 in the upper dyadic triangle of the unit square shifted by @xmath769 .
this set of shifted points can be written as the ( disjoint ) union of @xmath770@xmath771@xmath772and@xmath773 in this way , we can rewrite @xmath774 as @xmath775since @xmath776 , the process @xmath777 is adapted to the brownian filtration @xmath778 \right\}\]]and so @xmath779converges to the it integral @xmath780 as @xmath781 , @xmath65-a.s .
@xmath788 \\ & & \times \left [ \left ( b_{1}-b_{1-\frac{k}{2^{n}}}\right ) -\left ( b_{1}-b_{1-% \frac{k-1}{2^{n}}}\right ) \right ] \\ & = & \sum_{k=2^{n}\left ( 1-s\right ) + 1}^{2^{n}\left ( 2-s\right ) } \left ( \hat{b}% _ { 2-t\wedge \frac{k-1}{2^{n}}}-\hat{b}_{\frac{k-1}{2^{n}}-\left ( t - s\right ) } \right ) \left ( \hat{b}_{\frac{k}{2^{n}}}-\hat{b}_{\frac{k-1}{2^{n}}}\right)\end{aligned}\ ] ] [ [ non - existence - of - the - lvy - area - random - variable - mathcala_01left - mathcaltbright - in - the - injective - tensor - product ] ] non - existence of the lvy area random variable @xmath794 in the injective tensor product ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ in proposition [ propn : representation of levy area random variable ] , we showed that @xmath201 exists on the dyadic unit square @xmath243 ^{2}\right ) $ ] . in this section ,
we prove that @xmath201 has a unique continuous extension to the entire unit square _ excluding the diagonal_. we show that this extension has a jump discontinuity on the diagonal and hence , that the lvy area does not exist in the injectivetensor product .
[ propn : discontinuity of la]with @xmath65-probability @xmath190 , the lvy area @xmath201 as defined in proposition [ propn : representation of levy area random variable ] on the dyadic open upper triangle @xmath755 ^{2}\right ) : 0\leq s
< t\leq 1\right\}\]]has a unique uniformly continuous extension @xmath795 to the open upper triangle @xmath796
\times \left [ 0,1\right ] : 0\leq s < t\leq 1\right\}\]]and hence by ( [ eqn : antisymmetry ] ) to the open lower triangle of the unit square @xmath796 \times \left [ 0,1\right ] : 0\leq t
< s\leq 1\right\ } .\]]@xmath797 agrees with @xmath798 on the dyadic open upper triangle ( dyadic open lower triangle ) on a set @xmath799 of full @xmath176-measure .
@xmath797 has a jump discontinuity on the diagonal and so does not exist in the injective tensor algebra @xmath224 \times \left [ 0,1\right ] \right ) $ ] ( c.f .
proposition [ proposition : injective tensor algebra ] ) : there does not exist a continuous extension of @xmath800 to the unit square .
we start by proving the existence of a continuous extension of @xmath794 to the open upper triangle : in proposition [ propn : representation of levy area random variable ] , we showed that on the _ open _ upper dyadic triangle , @xmath755 ^{2}\right ) : 0\leq s < t\leq 1\right\ } , \]]@xmath801 has the following representation : @xmath802for a brownian motion @xmath32 , the it integral @xmath803 is a continuous martingale with respect to the filtration @xmath804 \right\ } $ ] , so that @xmath805 is continuous on a set of full @xmath65-measure . furthermore , the closed upper triangle @xmath796
\times \left [ 0,1\right ] : 0\leq s\leq t\leq 1\right\}\]]is compact . hence , @xmath806and @xmath807have uniformly continuous sample paths on the closed upper triangle .
@xmath808 , @xmath809 and @xmath810 are uniformly continuous in @xmath719 and @xmath31 on the closed upper triangle because of the @xmath176-a.s .
sample path continuity of brownian motion and compactness of the closed upper triangle .
this leaves us with having to prove that the terms @xmath811and@xmath812have uniformly continuous modifications on the closed dyadic upper triangle . since @xmath755 ^{2}\right ) : 0\leq s\leq t\leq 1\right\}\]]is dense in @xmath796 \times \left [ 0,1\right ] : 0\leq s\leq t\leq 1\right\ } , \]]this is equivalent to showing that @xmath813 and @xmath814 each have a unique uniformly continuous extension to the closed upper triangle .
we start by establishing the existence of a continuous modification of @xmath815 by using kolmogorov s lemma ( theorem [ thm : kolmogorov s lemma ] in chapter @xmath164 ) for a two - dimensional parameter set .
suppose that @xmath816 and @xmath817 are in @xmath818 ^{2}\right ) : 0\leq s < t\leq 1\right\ } $ ] and w.l.o.g .
suppose that @xmath819 .
for @xmath820 , we have that @xmath821by jensen s inequality .
we require the following result from @xcite , equations ( 3.24 ) and ( 3.25 ) : if @xmath160 is a measurable , adapted process satisfying @xmath822 < \infty , \]]for some real number @xmath823 and @xmath824 and @xmath825 is a standard , one - dimensional brownian motion on @xmath159 $ ] , then @xmath826 \leq \left ( m\left ( 2m-1\right ) \right ) ^{m}t^{m-1}\mathbb{e}\left [ \int_{0}^{t}\left\vert x_{u}\right\vert ^{2m}du\right ] .\]]we apply this to our setup where the brownian motion @xmath2 is defined on @xmath231 $ ] ( instead of on @xmath159 $ ] ) to find the following moment bounds : we find that @xmath827 \\ & \leq & \left ( m\left ( 2m-1\right ) \right ) ^{m}\frac{2^{m}}{\sqrt{\pi } } % \gamma \left ( \frac{2m+1}{2}\right ) \left\vert s_{2}-s_{1}\right\vert ^{m},\end{aligned}\]]where we used ( [ gaussian moments ] ) in chapter 2 . similarly , we find that @xmath828 \\ & \leq & \left ( m\left ( 2m-1\right ) \right ) ^{m}\frac{2^{m}}{\sqrt{\pi } } % \gamma \left ( \frac{2m+1}{2}\right ) \left\vert t_{2}-t_{1}\right\vert ^{m-1}\int_{t_{1}}^{t_{2}}\left\vert u - t_{2}+s_{1}\right\vert ^{m}du \\ & \leq & \left ( m\left ( 2m-1\right ) \right ) ^{m}\frac{2^{m}}{\sqrt{\pi } } % \gamma \left ( \frac{2m+1}{2}\right ) \left\vert t_{2}-t_{1}\right\vert ^{m},\end{aligned}\]]and@xmath829 \\ & \leq & \left ( m\left ( 2m-1\right ) \right ) ^{m}\frac{2^{m}}{\sqrt{\pi } } % \gamma \left ( \frac{2m+1}{2}\right ) \left\vert t_{2}-t_{1}\right\vert ^{m},\end{aligned}\]]and finally that @xmath830 \\ & \leq & \left ( m\left ( 2m-1\right ) \right ) ^{m}\frac{2^{m}}{\sqrt{\pi } } % \gamma \left ( \frac{2m+1}{2}\right ) \left\vert t_{2}-t_{1}\right\vert ^{m}.\end{aligned}\]]therefore , for @xmath831 , we have the bound hence , by kolmogorov s theorem ( c.f .
theorem [ thm : kolmogorov s lemma]in chapter 2 ) , @xmath833 has a continuous modification @xmath834 on the closed upper dyadic triangle @xmath755 ^{2}\right ) : 0\leq s\leq t\leq 1\right\ } .\ ] ] in summary , @xmath839is @xmath65-a.s .
uniformly continuous on the closed upper triangle
@xmath796 \times \left [ 0,1\right ] : 0\leq s\leq t\leq 1\right\ } .\]]the restriction of @xmath797 to the dyadic open upper triangle@xmath840 ^{2}\right ) : 0\leq s < t\leq 1\right\}\]]agrees with @xmath201 on a set @xmath841 of full @xmath65-measure . to see that @xmath797 has a jump discontinuity on the diagonal suppose that @xmath842 such that @xmath843 \times \left [ 0,1\right ] : 0\leq s < t\leq 1\right\}\]]and @xmath844 \times \left [ 0,1\right ] : 0\leq t < s\leq 1\right\}\]]for all @xmath286 .
since@xmath845and @xmath846@xmath847and @xmath848we find that @xmath849-a.s .
hence , by ( eqn : antisymmetry ) , @xmath850 , @xmath65-a.s . , which concludes the proof of the fact that @xmath851 has a jump discontinuity on the diagonal .
now , suppose that @xmath852 is a continuous extension of @xmath853 to the unit square @xmath106 \times \left [ 0,1% \right ] $ ] .
since the unit square is compact , @xmath852 is uniformly continuous . by the supposed continuity and ( [ eqn : antisymmetry ] )
, we have that @xmath854 .\]]since @xmath852 is an extension of @xmath800 there exists a set @xmath855 of full @xmath65-measure such @xmath856 ^{2}\right ) : 0\leq s < t\leq 1\right\ } } = \mathcal{a}% _ { 0,1}\text { on } \bar{\omega}_{0}\text{.}\]]but @xmath857 ^{2}\right ) : 0\leq s
< t\leq 1\right\ } } = \mathcal{a}% _ { 0,1}\text { on } \omega _
{ 0}\text{.}\]]the uniqueness of the uniformly continuous extension implies that @xmath858which can not be since @xmath859 does _ not _ vanish on the diagonal , whereas @xmath852 does . | the brownian frame process @xmath0 is defined as @xmath1 , \]]where @xmath2 is a real - valued brownian motion with parameter set @xmath3 $ ] .
this thesis investigates properties of the path - valued brownian frame process relevant to establishing an integration theory based on the theory of rough paths ( @xcite ) .
the interest in studying this object comes from its connection with gaussian volterra processes ( e.g. decreusefond05 ) and stochastic delay differential equations ( e.g. mohammed84 ) .
chapter 2 establishes the existence of @xmath0 .
we then examine the convergence of dyadic polygonal approximations to @xmath4 if the path - space @xmath5 where @xmath0 takes its values is equipped with first the @xmath6-variation norm ( @xmath7 ) and second the @xmath8-norm . in the case of the @xmath6-variation norm ,
the brownian frame process is shown to have finite @xmath9-variation for @xmath10 . in the case of the @xmath8-norm
, it is shown to have finite @xmath9-variation for @xmath11 .
chapter 3 provides a tail estimate for the probability that two evaluations of the brownian frame process are far apart in the @xmath6-variation norm . chapter 4 shows that @xmath0 does not have a lvy area if @xmath12 is equipped with the injective tensor product norm ( where @xmath13 \right ) $ ] ) . to my parents .
my thanks go to my supervisor , terry lyons , without whom this thesis would never have happened .
i am grateful to ben hambly and james norris for their careful reading and valuable comments . as a recipient of a doctoral training grant
i am endebted to _
epsrc_. i gratefully acknowledge the contribution of a _
jesus college _ _ old members graduate scholarship_. most importantly , i want to thank my parents for their generosity in all respects . |
while large cp violating asymmetry has recently been confirmed in b - meson decay @xcite , the standard model ( sm ) does not predict any appreciable cp violation ( cpv ) in the lepton sector .
however , physics beyond the sm could produce cpv in leptonic processes ; we would expect such effects to be enhanced for @xmath1 leptons due to their large mass .
parameterizing cpv in @xmath1-pair production by an electric dipole moment @xmath0 , several authors have found possible effects of order @xmath9 cm due to new physics models @xcite .
the best existing bounds on @xmath0 are indirect , requiring @xmath10 cm based on naturalness arguments @xcite and precision lep data @xcite.cm has also been derived from the experimental limits on @xmath11 decay @xcite .
we note , however , that this result assumes a particular ansatz for the lepton mixing matrix . in a recent preprint @xcite
, other authors have argued that the same constraint may be derived under weaker assumptions . ]
previous direct measurements have been performed at lep , where l3 @xcite found @xmath12 cm and opal @xcite found @xmath13 cm , using the process @xmath14 ; and by argus @xcite , which set limits @xmath15 cm and @xmath16 cm based on a study of @xmath17 production . in this paper
we present the first direct measurement of the @xmath1 lepton s electric dipole moment with a sensitivity in the @xmath18 cm range .
we improve on the sensitivity of argus by an order of magnitude , by reducing the systematic uncertainty in the extraction of @xmath0 , and by analyzing a much larger sample of events .
we adapt the so - called optimal observable method @xcite , which maximizes the sensitivity to @xmath0 . here
the optimal observables are defined as @xmath41 the mean value of the observable @xmath42 is given by @xmath43 where the integration is over the phase space ( @xmath44 ) spanned by the relevant kinematic variables .
the cross - term containing the integral of the product of @xmath45 and @xmath46 drops out because of their different symmetry properties .
the expression for the imaginary part is similar .
the means of the observables @xmath47 and @xmath48 are therefore linear functions of @xmath49 , @xmath50 eight different final states in the decay of @xmath1-pairs , @xmath51 , @xmath52 , @xmath53 , @xmath54 , @xmath55 , @xmath56 , @xmath57 , and @xmath58 , are analyzed , where all particles except @xmath59 and @xmath60 are positively or negatively charged .
because of the undetectable particles , we can not fully reconstruct the quantities @xmath32 and @xmath34 .
therefore , for each event we calculate possible kinematic configurations and obtain the mean value of @xmath61 , @xmath45 and @xmath46 by averaging over the calculated configurations . in the case when both @xmath1 leptons decay hadronically ( @xmath62 or @xmath63 ) , the @xmath1 flight direction is calculated with a two - fold ambiguity @xcite and we take the average of @xmath61 , @xmath45 and @xmath46 over the two solutions . in the case
when either one or both @xmath1 leptons decay leptonically ( @xmath64 or @xmath65 ) , a monte carlo ( mc ) treatment is adopted to take into account the additional ambiguity in the effective mass of the @xmath66 system ( @xmath67 ) . for each event we generate 100 mc configurations using a hit - and - miss approach by varying @xmath67 , and compute the averaged @xmath61 , @xmath45 and @xmath46 over successful tries in which the @xmath1 direction can be constructed kinematically . in the calculation
, we ignore the effect of undetected photons coming from initial state radiation , radiative @xmath1 decays and bremsstrahlung .
the resulting bias in @xmath0 is included among the systematic errors ; the effect on the final result is negligible .
in this analysis , we use 26.8 million @xmath1-pairs ( @xmath68 ) accumulated with the belle detector @xcite at the kekb accelerator @xcite .
kekb is an asymmetric energy @xmath69 collider with a beam crossing angle of 22 mrad .
its center - of - mass energy is 10.58 gev , corresponding to the @xmath70 resonance , with beam energies of 8 and 3.5 gev for electrons and positrons , respectively .
belle is a general - purpose detector with an asymmetric structure along the beam direction . among the detector elements , the central drift chamber ( cdc ) and the silicon vertex detector ( svd ) are essential to obtain the momentum vectors of charged particles . the combined information from the silica aerogel cherenkov counters ( acc ) , the time - of - flight counters ( tof ) , the csi electromagnetic calorimeter ( ecl ) , and the @xmath71 detector ( klm ) is used for particle identification .
the mc event generators koralb / tauola @xcite are used for @xmath1-pair production and decays .
the detector simulation is performed with a geant - based program , gsim .
actual data and mc generated events are reconstructed by the same program written by the belle collaboration .
.yield , purity and background rate obtained for the event selection described in the text , where the purity is evaluated by mc simulation and its error comes from mc statistics . [
cols="^,>,^,<",options="header " , ] finally , we obtain mean values for @xmath72 and @xmath73 over the eight different @xmath74 modes weighted by quadratically summed statistical and systematic errors , @xmath75 with corresponding 95% confidence limits @xmath76 we would like to thank professors k. hagiwara , o. nachtmann , and z. was for their constructive advice and many helpful discussions .
we also wish to thank the kekb accelerator group for the excellent operation of the kekb accelerator .
we gratefully acknowledge the support from the ministry of education , culture , sports , science , and technology of japan , grant - in - aid for jsps fellows 01655 2001 , and the japan society for the promotion of science ; the australian research council and the australian department of industry , science and resources ; the national science foundation of china under contract no .
10175071 ; the department of science and technology of india ; the bk21 program of the ministry of education of korea and the chep src program of the korea science and engineering foundation ; the polish state committee for scientific research under contract no .
2p03b 17017 ; the ministry of science and technology of the russian federation ; the ministry of education , science and sport of the republic of slovenia ; the national science council and the ministry of education of taiwan ; and the u.s .
department of energy .
99 k. abe et al .
( belle collaboration ) , phys .
87 ( 2001 ) 091802 ; phys .
d 66 ( 2002 ) 032007
. b. aubert et al .
( babar collaboration ) , phys .
87 ( 2001 ) 091801 ; phys .
d 66 ( 2002 ) 032003 .
t. huang , w. lu , and z. tao , phys .
d 55 ( 1997 ) 1643 .
w. bernreuther , a. brandenburg , and p. overmann , phys .
b 391 ( 1997 ) 413 ; ibid . ,
b 412 ( 1997 ) 425 .
k. akama , t. hattori , and k. katsuura , phys .
88 ( 2002 ) 201601 .
r. escribano and e. mass , phys .
b 301 ( 1993 ) 419 ; nucl . phys .
b 429 ( 1994 ) 19 ; phys .
b 395 ( 1997 ) 369 .
t. huang , z. h. lin , and x. zhang , phys .
b 450 ( 1999 ) 257 .
j. q. zhang , x. c. song , w. j. huo , and t. f. feng , hep - ph/0205309 . m. acciarri et al .
( l3 collaboration ) , phys .
b 434 ( 1998 ) 169 .
k. ackerstaff et al .
( opal collaboration ) , phys .
b 431 ( 1998 ) 188 . h. albrecht et al .
( argus collaboration ) , phys .
b 485 ( 2000 ) 37 .
w. bernreuther , o. nachtmann , and p. overmann , phys .
d 48 ( 1993 ) 78 .
d. atwood and a. soni , phys .
d 45 ( 1992 ) 2405 . k. ackerstaff et al .
( opal collaboration ) , z. phys .
c 74 ( 1997 ) 403 .
a. abashian et al .
( belle collaboration ) , nucl .
instr . and meth . a 479 ( 2002 ) 117 .
e. kikutani ed .
, kek preprint 2001 - 157 ( 2001 ) , to appear in nucl .
instr . and meth .
a. koralb(v.2.4)/tauola(v.2.6 ) : s. jadach and z. was , comp .
85 ( 1995 ) 453 ; ibid . , 64 ( 1991 ) 267 ;
36 ( 1985 ) 191 ; s. jadach , z. was , r. decker , and j.h .
khn , comp .
commun . 64
( 1991 ) 275 ; ibid .
, 70 ( 1992 ) 69 ; ibid . , 76 ( 1993 ) 361 .
k. hanagaki et al . , nucl .
instr . and meth . a 485 ( 2002 ) 490 . | we have searched for a cp violation signature arising from an electric dipole moment ( @xmath0 ) of the @xmath1 lepton in the @xmath2 reaction . using an optimal observable method and 29.5 fb@xmath3 of data collected with the belle detector at the kekb collider at @xmath4 gev
, we find @xmath5 and @xmath6 and set the 95% confidence level limits @xmath7 and @xmath8 . , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , and tau , electic dipole moment , cp violation , optimal observable 13.40.gp , 13.35.dx , 14.60.fg |
the binomial edge ideal of a graph was introduced in @xcite , and @xcite at about the same time .
let @xmath1 be a finite simple graph with vertex set @xmath2 $ ] and edge set @xmath3 .
also , let @xmath4 $ ] be the polynomial ring over a field @xmath5 .
then the * binomial edge ideal * of @xmath1 in @xmath6 , denoted by @xmath7 , is generated by binomials @xmath8 , where @xmath9 and @xmath10 . also , one could see this ideal as an ideal generated by a collection of 2-minors of a @xmath11-matrix whose entries are all indeterminates .
many of the algebraic properties of such ideals were studied in @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite and @xcite . in @xcite
, the authors introduced the binomial edge ideal of a pair of graphs , as a generalization of the binomial edge ideal of a graph .
let @xmath12 be a graph on the vertex set @xmath13 $ ] and @xmath14 a graph on the vertex set @xmath2 $ ] , and let @xmath15 be an @xmath16-matrix of indeterminates .
let @xmath17 $ ] be the polynomial ring in the variables @xmath18 , where @xmath19 and @xmath20 .
let @xmath21 for some @xmath22 and @xmath23 for some @xmath24 . to the pair @xmath25
, the following @xmath26-minor of @xmath27 is assigned : @xmath28=x_{it}x_{jl}-x_{il}x_{jt}.\ ] ] then , the ideal @xmath29 is called the * binomial edge ideal of the pair * @xmath30 .
some properties of this ideal were studied in @xcite . note that
if @xmath12 is just an edge , then @xmath31 is isomorphic to @xmath32 . in @xcite
, the authors posed a question about the castelnuovo - mumford regularity of the binomial edge ideal of graphs , which is if @xmath33 , where @xmath34 is the number of maximal cliques of @xmath1 .
they also answered this question for closed graphs , in @xcite . in @xcite
, the authors gained an upper bound for the regularity of the binomial edge ideal of a graph on @xmath35 vertices .
they showed that @xmath36 .
they also posed a conjecture which is @xmath37 , whenever @xmath1 is not @xmath38 , the path over @xmath35 vertices . in this paper , we investigate about the regularity of the binomial edge ideals and especially these two problems .
this paper is organized as follows . in section [ preliminaries ]
, we pose some definitions , facts and notation which will be used throughout the paper . in section [ the castelnuovo - mumford regularity of binomial edge ideals ] , we show that the conjecture of matsuda and murai is true for every graph which has a cut edge or a simplicial vertex .
hence , we prove this conjecture for all chordal graphs .
we also show that the regularity of the binomial edge ideal of the join of two graphs @xmath12 and @xmath14 ( not both complete ) is equal to @xmath39 . applying this fact
, we prove the conjecture of matsuda and murai , in the case of join of graphs , and hence for complete @xmath0-partite graphs . then , we generalize some results of schenzel and zafar about complete @xmath0-partite graphs . using a similar discussion as one appeared in @xcite
, we prove that the conjecture due to the authors is true for a class of chordal graphs including block graphs , which we call them `` generalized block graphs '' .
hence , we extend the recent results of ene and zarojanu about block graphs . throughout the paper ,
we mean by a graph @xmath1 , a simple graph .
moreover , if @xmath40 is the vertex set of @xmath1 ( which contains @xmath35 elements ) , then , for simplicity , we denote it by @xmath2 $ ] .
in this section , we review some notions and facts around graphs and binomial ideals associated to graphs , which we need throughout . in @xcite , the authors determined all graphs whose binomial edge ideal have a grbner basis with respect to the lexicographic order induced by @xmath41 , and called this class of graphs , * closed graphs*. there are also some combinatorial descriptions for these graphs , for example in @xcite , the authors proved that a graph @xmath1 is closed if and only if there exists a labeling of @xmath1 such that all facets of @xmath42 are intervals @xmath43\subseteq [ n]$ ] . here , @xmath42 is the clique complex of @xmath1 , the simplicial complex whose facets are the vertex sets of the maximal cliques of @xmath1 .
we say that a vertex of @xmath1 is a * free vertex * , if it is a free vertex of the simplicial complex @xmath42 , i.e. it is just contained in one facet of @xmath42 .
this is equivalent to say that such a vertex @xmath44 has this property that @xmath45 induces a complete subgraph of @xmath1 , and the vertex @xmath44 is also called a * simplicial vertex*. by @xmath45 , we mean the set of all neighbors ( i.e. adjacent vertices ) of the vertex @xmath44 in @xmath1 .
let @xmath1 and @xmath46 be two graphs on @xmath13 $ ] and @xmath2 $ ] , respectively .
we denote by @xmath47 , the _
join _ ( product ) of two graphs @xmath1 and @xmath46 , that is the graph with vertex set @xmath13\cup [ n]$ ] , and the edge set @xmath48,~w\in [ n]\}$ ] . in particular ,
the cone of a vertex @xmath44 on a graph @xmath1 is defined to be their join , that is @xmath49 , and is sometimes denoted by @xmath50 .
let @xmath51 be a set .
to simplify our notation throughout this paper , we introduce the join of two collection of subsets of @xmath51 , @xmath52 and @xmath53 , denoted by @xmath54 , as @xmath55 .
if @xmath56 are collections of subsets of @xmath51 , then we denote their join , by @xmath57 .
let @xmath1 be a graph and @xmath58 an edge of it .
then vertices @xmath44 and @xmath59 are called the endpoints of @xmath60 . if @xmath61 is a set of edges of @xmath1 , then by @xmath62 , we mean the graph on the same vertex set as @xmath1 in which the edges @xmath63 are omitted . here
, we simply write @xmath64 , instead of @xmath65 . an edge @xmath60 of @xmath1 whose deletion from the graph , implies a graph with more connected components than @xmath1 , is called a * cut edge * of @xmath1 .
now , we recall a notation from @xcite .
if @xmath66 are two vertices of a graph @xmath67 and @xmath58 is not an edge of @xmath1 , then @xmath68 is defined to be the graph on the vertex set @xmath51 , and the edge set @xmath69 . a vertex @xmath44 of @xmath1 whose deletion from the graph , implies a graph with more connected components than @xmath1 , is called a * cut point * of @xmath1 .
a * nonseparable * graph is a connected and nontrivial graph with no cut points
. a * block * of a graph is a maximal nonseparable subgraph of it .
a * block graph * is a connected graph whose blocks are complete graphs ( see @xcite for more information in this topic ) .
a disconnected graph is called a block graph , if all of its connected components are block graphs .
one can see that a graph @xmath1 is a block graph if and only if it is a chordal graph in which every two maximal cliques have at most one vertex in common .
this class was considered in ( * ? ? ?
* theorem 1.1 ) . here
, we introduce a class of chordal graphs including block graphs : let @xmath1 be a connected chordal graph such that for every three maximal cliques of @xmath1 which have a nonempty intersection , the intersection of each pair of them is the same . in other words , @xmath1 has the property that for every @xmath70 , if @xmath71 , then @xmath72 . we call @xmath1 , a * generalized block * graph . a disconnected graph is called a generalized block graph , if all of its connected components are generalized block graphs . by the above , it is clear that a block graph is also a generalized block graph .
hence , a tree is also a generalized block graph . the graph depicted in figure
[ generalized ] is a generalized block graph , which is not a block graph .
[ generalized ] suppose that @xmath1 is a graph on @xmath2 $ ] .
let @xmath73 be a subset of @xmath2 $ ] , and let @xmath74 be the connected components of @xmath75\setminus t}$ ] , the induced subgraph of @xmath1 on @xmath2\setminus t$ ] .
for each @xmath76 , we denote by @xmath77 the complete graph on the vertex set @xmath78 .
if there is no confusion , then we may simply write @xmath79 instead of @xmath80 .
set @xmath81 then , @xmath82 is a prime ideal , where @xmath83 , by ( * ? ? ?
* lemma 3.1 ) .
moreover , @xmath84}p_t(g)$ ] , by ( * ? ? ?
* theorem 3.2 ) .
so that , @xmath85\}$ ] , by ( * ? ? ?
* cororally 3.3 ) . if each @xmath86 is a cut point of the graph @xmath87\setminus t)\cup \{i\}}$
] , then we say that @xmath73 has * cut point property * for @xmath1 .
let @xmath88:t~\mathrm{has~cut~point~property~for}~g\}$ ] .
one has @xmath89 if and only if @xmath1 is a complete graph . on the other hand , denoted by @xmath90 ,
we mean the set of all minimal prime ideals of @xmath7 . then , one has @xmath91 if and only if @xmath92 , by ( * ? ? ?
* corollary 3.9 ) .
in this section , we deal with two recent conjectures on the castelnuovo - mumford regularity of the binomial edge ideal of a graph , one is due to matsuda and murai and the other is due to the authors . recall that the * castelnuovo - mumford regularity * ( or simply , * regularity * ) of a graded @xmath6-module @xmath93 is defined as @xmath94 denoted by @xmath34 , we mean the number of maximal cliques of the graph @xmath1 .
the following are those mentioned conjectures : + * conjecture a. * ( see @xcite ) let @xmath1 be a graph on @xmath35 vertices which is not a path
. then @xmath37 . +
* conjecture b. * ( see @xcite ) let @xmath1 be a graph . then @xmath33 .
+ the following bounds for the regularity were given in @xcite and @xcite : moreover , note that recently , ene and zarojanu computed the exact value of the regularity of the binomial edge ideal of a closed graph with respect to the graphical terms in @xcite , which also yields theorem [ kiani - saeedi ] . also , by theorem [ matsuda - murai ]
, it is clear that conjecture b is true for trees and unicyclic graphs , whose unique cycles have length greater that three .
_ let @xmath1 be a graph and @xmath60 a cut edge of @xmath1 such that its endpoints are the free vertices of @xmath64 .
then , we call @xmath60 , a * free cut edge * of @xmath1 .
suppose that @xmath61 is the set of all free cut edges of @xmath1 .
then , we call the graph @xmath95 , the * reduced graph * of @xmath1 , and denote it by @xmath96 . if @xmath1 does not have any free cut edges , then we set @xmath97 . _ [ main ] let @xmath98 be a graph on @xmath35 vertices which is a disconnected graph , or else + _ ( a ) _ has a simplicial vertex , or + _ ( b ) _ has a cut edge , or + _ ( c ) _ is the join of two graphs , or + _ ( d ) _ is an @xmath35-cycle .
+ then , we have @xmath37 . using the above proposition
, we partially answer a question that j. herzog had posed in a discussion with us .
he asked if @xmath107 , for all @xmath108 , where @xmath1 is a graph on @xmath109 vertices .
note that in ( * ? ? ?
* theorem 2.2 ) , the authors mentioned that @xmath107 , for all @xmath110 .
we use the induction on the number of edges of a forest .
if @xmath1 is a graph with no edges , then @xmath111 , and hence the result is obvious .
now , let @xmath1 be a forest on @xmath109 vertices and at least one edge .
let @xmath60 be an arbitrary edge of @xmath1 . so that @xmath60 is clearly a cut edge of @xmath1 .
thus , by proposition [ colon1 ] , for all @xmath108 , we have @xmath112 . but @xmath113 , since @xmath114 .
thus , we have @xmath115 , for all @xmath108 . by the induction hypothesis
, we have @xmath116 , for all @xmath108 .
thus , the desired result follows . in the statements of the above proposition ,
the equality does not occur necessarily , because the free resolution which is obtained for @xmath117 by mapping cone is not minimal necessarily .
but , in a more special case , as in the next proposition , we get the minimal free resolution and hence desired equalities .
the proof is as similar as proposition [ colon1 ] .
note that @xmath121 , since @xmath60 is a free cut edge of @xmath1 .
so , one may consider the short exact sequence @xmath122 instead of that was mentioned in the proof of proposition [ colon1 ] .
let @xmath123 be the minimal graded free resolution of @xmath124 .
now , consider the homomorphism of complexes @xmath125 , as the multiplication by @xmath100 .
in fact it is a lift of the map @xmath126 . obviously , the mapping cone over @xmath127 resolves @xmath117 .
in addition , it is minimal , because @xmath123 is minimal and all the maps in the complex homomorphism @xmath127 are of positive degrees .
since @xmath96 has @xmath129 connected components , @xmath1 has exactly @xmath130 free cut edges .
so , by using proposition [ colon2 ] repeatedly , we have that @xmath131 . on the other hand , @xmath132 , and hence @xmath133 .
_ proof of theorem [ construction ] .
_ by corollary [ reg - reduced ] , we have @xmath128 . by the assumption , conjecture b is true for @xmath134 , for all @xmath135 , so that @xmath136 .
thus , @xmath137 . on the other hand , @xmath138 , because @xmath96 has @xmath129 connected components and hence @xmath1 has @xmath130 free cut edges .
thus , @xmath33 , which implies that the conjecture is also true for @xmath1 .
@xmath139 + combining proposition [ colon2 ] and ( * ? ? ?
* theorem 2.2 ) , we get the following : [ reg - closed - ene ] let @xmath1 be a connected graph and @xmath99 the connected components of @xmath96 . if @xmath96 is closed , then @xmath140 where @xmath141 is the length of the longest induced path in @xmath134 .
[ linear strand1 ]
let @xmath1 be a graph and @xmath60 be a cut edge of @xmath1 .
then @xmath143 , for all @xmath144 .
in particular , if @xmath60 is a free cut edge of @xmath1 , then @xmath145 , for all @xmath144 . here , we focus on graphs containing a simplicial vertex .
this class of graphs includes a famous class of graphs , i.e. chordal graphs .
indeed , we show that conjecture a is true in this case . moreover , we prove conjecture b for a class of chordal graphs containing block graphs .
the following is one of the main theorems of this subsection : let @xmath154 be all the neighbors of the simplicial vertex @xmath44 , and @xmath63 be the edges joining @xmath44 to @xmath154 , respectively , where @xmath155
. without loss of generality , assume that @xmath156 .
note that for each @xmath157 , @xmath158 is a path between @xmath44 and @xmath159 in @xmath64 , so that for all @xmath157 , @xmath160 and @xmath161 are in the minimal monomial set of generators of @xmath162 .
also , all other paths between @xmath44 and @xmath159 in @xmath64 contain @xmath163 for some @xmath157 .
thus , all the monomials correspond to these paths , are divisible by either @xmath160 or @xmath161 for some @xmath157 .
hence , we have @xmath164 .
so that @xmath165 .
the binomial generators of @xmath166 correspond to the edges containing vertices @xmath167 , are contained in @xmath162 .
let @xmath168 .
then , we have @xmath169\setminus \{v , v_1,\ldots , v_{t-1}\}}}+(x_i , y_i:1\leq i\leq t-1)$ ] , since @xmath44 is a free vertex of @xmath42 .
thus , @xmath170\setminus \{v , v_1,\ldots , v_{t-1}\}}})$ ] .
but , @xmath171\setminus \{v , v_1,\ldots , v_{t-1}\}}})\leq n-2 $ ] , by theorem [ m - colon1 ] , since @xmath155 .
therefore , @xmath153 , as desired .
now , we go to the proof of theorem [ chordal ] : + _ proof of theorem [ chordal ] .
_ we use induction on the number of the vertices .
let @xmath1 be a graph on @xmath2 $ ] , with a simplicial vertex , which is not a path .
we consider two following cases : \(i ) suppose that @xmath1 has a simplicial vertex which is a leaf , say @xmath44 .
then , assume that @xmath59 is the only neighbor of @xmath44 , and @xmath58 is the edge joining @xmath44 and @xmath59 .
we have @xmath172\setminus v}})$ ] , since @xmath44 is an isolated vertex of @xmath64 .
thus , by theorem [ matsuda - murai ] , @xmath173 . on the other hand
, we have @xmath174 , by theorem [ m - colon1 ] .
note that @xmath44 is also an isolated vertex of @xmath175 , so that we can disregard it in computing the regularity .
thus , @xmath176 , by the induction hypothesis , since @xmath175 has @xmath59 as a simplicial vertex .
hence , @xmath177 .
thus , by proposition [ reg - chordal ] , we get @xmath178 .
\(ii ) suppose that all the simplicial vertices of @xmath1 have degree greater than one .
let @xmath44 be a simplicial vertex of @xmath1 and @xmath154 be all the neighbors of @xmath44 , and @xmath63 be the edges joining @xmath44 to @xmath154 , respectively , where @xmath155 .
using repeatedly proposition [ reg ] and lemma [ colon3 ] , we get @xmath179 .
note that @xmath180 is a graph on @xmath35 vertices in which @xmath44 is a leaf .
thus , by case ( i ) , we have @xmath181 , by case ( i ) .
thus , @xmath178 . therefore , by the above cases
, we get the desired result . @xmath139 + now , recall that a facet @xmath182 of a simplicial complex @xmath183 is called a * leaf * , if either @xmath182 is the only facet , or there exists a facet @xmath1 , called a * branch * of @xmath182 , such that for each facet @xmath46 of @xmath183 , with @xmath184 , one has @xmath185 .
one can see that each leaf @xmath182 has at least a free vertex .
a simplicial complex @xmath183 is called a * quasi - forest * if its facets can be ordered as @xmath186 such that for all @xmath187 , @xmath188 is a leaf of @xmath183 with facets @xmath189 .
such an order of the facets is called a * leaf order*. a connected quasi - forest is called a * quasi - tree*. our proof is similar to the proof of ( * ? ? ?
* theorem 2.9 ) , which is based on the technique applied in the proof of ( * ? ? ?
* theorem 1.1 ) .
so , we omit some details . by dirac s theorem ( see @xcite ) , @xmath42 is a quasi tree , since @xmath1 is connected and chordal .
let @xmath191 , and @xmath192 be a leaf order of the facets of @xmath42 .
we use induction on @xmath193 , the number of maximal cliques of @xmath1 .
if @xmath194 , then the result is obvious . let @xmath195 and @xmath196 be all the branches of the leaf @xmath197 . since @xmath1 is a generalized block graph , each pair of the facets @xmath198 intersect in exactly the same set of vertices , say @xmath199 , and also , @xmath200 , for all @xmath201 , as @xmath197 is a leaf .
hence , we have that @xmath202 , for all @xmath201 . on the other hand , for @xmath203 $ ] with @xmath91
, we have that @xmath204 if and only if @xmath205 ; because if @xmath206 and @xmath207 , then @xmath44 is not a cut point of the graph @xmath87\setminus t)\cup \{v\}}$ ] , as @xmath208 , so it is a contradiction .
thus , let @xmath209 , where @xmath210 similar to the proof of ( * ? ? ?
* theorem 2.9 ) and ( * ? ? ?
* theorem 1.1 ) , let @xmath211 be the graph obtained from @xmath1 , by replacing the cliques @xmath198 , by a clique on the vertex set @xmath212 .
one can see that @xmath213 and @xmath214\setminus a}}$ ] .
thus , we have @xmath215\setminus a}}$ ] . by the definition of generalized block graphs , it is not difficult to see that @xmath211 , @xmath75\setminus a}$ ] and @xmath216\setminus a}$ ] are also generalized block graphs
. then , considering the short exact sequence @xmath217 the result follows , as in ( * ? ? ?
* theorem 2.9 ) . here , we focus on the join of two graphs and deal with the conjectures for this class of graphs . consequently , we gain some results on complete @xmath0-partite graphs , which generalize some previous results .
note that the join of two complete graphs is also obviously complete , so that its binomial edge ideal has a linear resolution , by ( * ? ? ?
* theorem 2.1 ) , and hence its regularity is equal to @xmath26 .
the following theorem determines the regularity of the binomial edge ideal of the join of two graphs with respect to the original graphs , when they are not both complete graphs .
[ both disconnected 1 ]
suppose that @xmath223 and @xmath224 are two graphs on disjoint sets of vertices @xmath218=\bigcup_{i=1}^r [ n_{1i}]$ ] and @xmath219=\bigcup_{i=1}^s [ n_{2i}]$ ] , respectively , where @xmath225 .
then we have @xmath226\ } \big{)}\cup \big{(}(\bigcirc_{i=1}^{s}\mathcal{c}(g_{2i}))\circ \{[n_1]\ } \big{)}.\ ] ] let @xmath227 and @xmath228\}$ ] .
so , @xmath229\cup ( \bigcup_{i=1}^r t_{1i})$ ] , where @xmath230 , for @xmath231 .
we show that @xmath73 has cut point property .
let @xmath232 . if @xmath233 , for some @xmath231 , then @xmath87\setminus t)\cup \{j\}}={g_{1i}}_{([n_{1i}]\setminus t_{1i})\cup \{j\}}\sqcup ( \bigsqcup_{l=1,l\neq i}^r { g_{1l}}_{([n_{1l}]\setminus t_{1l})})$ ] . in this case
, @xmath234 is a cut point of @xmath235\setminus t_{1i})\cup \{j\}}$ ] , since @xmath230 . so that @xmath234 is also a cut point of @xmath87\setminus t)\cup \{j\}}$ ] .
if @xmath236 $ ] , then @xmath87\setminus t)\cup \{j\}}=j*\bigsqcup_{i=1}^r { g_{1i}}_{([n_{1i}]\setminus t_{1i})}$ ] .
so , @xmath234 is a cut point of @xmath87\setminus t)\cup \{j\}}$ ] , since @xmath87\setminus t)}$ ] is disconnected .
thus , in both cases , @xmath73 has cut point property . if @xmath237\}$ ] , then similarly , we have @xmath91 . for the other inclusion , let @xmath238 . if @xmath73 does not contain @xmath218 $ ] and @xmath219 $ ] , then @xmath75\setminus t}$ ] is connected , and hence no element @xmath239 of @xmath73 is a cut point of @xmath87\setminus t)\cup \{i\}}$ ] .
so , we have @xmath218\subseteq t$ ] or @xmath219\subseteq t$ ] .
suppose that @xmath218\subseteq t$ ]
. then , @xmath240\cup ( \bigcup_{i=1}^s t_{2i})$ ] , where @xmath241 $ ] , for @xmath242 .
let @xmath243 .
if @xmath244 , then , clearly , @xmath245 . if @xmath246 , then each @xmath247 , is a cut point of @xmath87\setminus t)\cup \{j\}}$ ] , since @xmath91 . so that @xmath234 is a cut point of @xmath248\setminus t_{2i})\cup \{j\}}$ ] , because @xmath247 and @xmath249 s are on disjoint sets of vertices .
thus , @xmath245 .
therefore , @xmath237\}$ ] . if @xmath219\subseteq t$ ]
, then similarly we get @xmath228\}$ ] .
_ proof of theorem [ reg - join ] .
_ let @xmath227 .
note that since @xmath1 is not a complete graph , @xmath7 does not have a linear resolution , by ( * ? ? ?
* theorem 2.1 ) . so that @xmath254 . on the other hand , by proposition [
induced ] , @xmath255 and @xmath256 , because @xmath12 and @xmath14 are induced subgraphs of @xmath1 .
so , @xmath257 . for the other inequality ,
first , suppose that @xmath12 and @xmath14 are both disconnected graphs .
let @xmath223 and @xmath224 be two graphs on disjoint sets of vertices @xmath218=\bigcup_{i=1}^r [ n_{1i}]$ ] and @xmath219=\bigcup_{i=1}^s [ n_{2i}]$ ] , respectively , where @xmath225 . by proposition
[ both disconnected 1 ] , @xmath258\ } \big{)}\cup \big{(}(\bigcirc_{i=1}^{s}\mathcal{c}(g_{2i}))\circ \{[n_1]\ } \big{)}$ ] .
so , @xmath259 , where @xmath260\subseteq t } } p_t(g)~~,~~q'=\bigcap_{\substack { t\in \mathcal{c}(g ) \\ [ n_1]\nsubseteq t } } p_t(g ) .
\nonumber\ ] ] thus , we have @xmath261)+\bigcap_{\substack { t\in \mathcal{c}(g ) \\ [ n_1]\subseteq t } } p_{t\setminus [ n_1]}(g_2 ) \nonumber\ ] ] and @xmath262\nsubseteq t } } p_t(g)\big { ) } = p_{\emptyset}(g)\cap \big{(}(x_i , y_i : i\in [ n_2])+\bigcap_{\substack { t\in \mathcal{c}(g ) \\ [ n_2]\subseteq t } } p_{t\setminus [ n_2]}(g_1)\big{)}. \nonumber\ ] ] so , one can see that @xmath263)+j_{g_2}$ ] , @xmath264)+j_{g_1}\big{)}$ ] and @xmath265)+j_{k_{n_2}}$ ] .
now , consider the short exact sequence @xmath266 by ( * ? ? ?
* corollary 18.7 ) , we have @xmath267 . on the other hand
, we have @xmath268 , @xmath269 ( by using a suitable short exact sequence as above ) , and @xmath270 .
hence , @xmath271 . now , suppose that @xmath12 or @xmath14 is connected .
we add an isolated vertex @xmath44 to @xmath12 and an isolated vertex @xmath59 to @xmath14 .
thus , we obtain two disconnected graphs @xmath272 and @xmath273 .
so , by the above discussion , we have @xmath274 .
but , clearly , we have @xmath275 and @xmath276 , so that @xmath277 .
thus , the result follows by proposition [ induced ] , since @xmath278 is an induced subgraph of @xmath279 .
@xmath139 [ sharpness ] _ by theorem [ reg - join ] , we have if @xmath1 is a ( multi)-fan graph ( i.e. @xmath280 , for some @xmath281 , which might be a non - closed graph ) , then @xmath282 .
this implies that if conjecture b is true , then the given bound is sharp .
_ we use induction on @xmath155 , the number of parts .
if @xmath288 , then @xmath1 is the join of two graphs each consisting of some isolated vertices .
so , the regularity of the binomial edge ideal of each of them is @xmath289 .
thus , by theorem [ reg - join ] , we have @xmath287 . now
, suppose that @xmath290 and the result is true for every complete @xmath291-partite graph which is not complete .
let @xmath292 be the partition of the vertices of @xmath1 to @xmath0 parts .
hence , we have @xmath293 , and @xmath294 is a complete @xmath291-partite graph . if @xmath294 is a complete graph , then @xmath295 , since , otherwise , @xmath1 is a complete graph , a contradiction .
so , by theorem [ reg - join ] , @xmath287 .
if @xmath294 is not complete , then by the induction hypothesis , we have @xmath296 . thus , again by theorem
[ reg - join ] , the result follows .
now , we go to the proof of our main theorems . + _ proof of theorem [ main ] . _
if @xmath1 is a disconnected graph with @xmath297 connected components @xmath221 , over @xmath298 vertices , respectively , then we have @xmath299 . by theorem [ matsuda - murai ] , we have @xmath300 .
so that @xmath301 , since @xmath297 .
so , the result follows in this case . now , suppose that @xmath1 has a cut edge @xmath60 .
then , @xmath64 is disconnected and hence @xmath173 . on the other hand , @xmath175 has two connected components , say @xmath12 and @xmath14 , which both have a simplicial vertex .
so that conjecture a is true for both of them , by theorem [ chordal ] .
note that @xmath12 and @xmath14 are not both paths , as @xmath1 is not .
thus , as mentioned above , we have @xmath302 .
so , by proposition [ colon2 ] , we get the result in this case too . now , combining theorem [ reg - chordal ] , theorem [ reg - join ] and ( * ? ? ?
* corollary 3.8 ) we get the result .
@xmath139 + _ proof of theorem [ main2 ] .
_ by theorem [ construction ] , theorem [ kiani - saeedi ] and theorem [ reg - generalized ] , the result follows .
@xmath139 + * acknowledgments : * the authors would like to thank professor j. herzog for some useful comments .
also , the authors would like to thank to the institute for research in fundamental sciences ( ipm ) for financial support .
the research of the first author was in part supported by a grant from ipm ( no .
92050220 ) . | we prove two recent conjectures on some upper bounds for the castelnuovo - mumford regularity of the binomial edge ideals of some different classes of graphs .
we prove the conjecture of matsuda and murai for graphs which has a cut edge or a simplicial vertex , and hence for chordal graphs .
we determine the regularity of the binomial edge ideal of the join of graphs in terms of the regularity of the original graphs , and consequently prove the conjecture of matsuda and murai for such a graph , and hence for complete @xmath0-partite graphs .
we also generalize some results of schenzel and zafar about complete @xmath0-partite graphs .
we also prove the conjecture due to the authors for a class of chordal graphs . |
gamma - ray bursts ( grb s ) have been a challenge to theorists and a source of fascination for all for over 30 years and many models have been suggested to explain them ( nemiroff 1993 ) . lately major progress has occurred in understanding grb s because of accurate localizations provided by the beppo - sax mission .
these locations allow rapid follow - up observations with optical , x - ray , and radio telescopes that have yielded exciting information about grb counterparts .
two bursts have been found to lie in galaxies having red - shifts of 0.83 and 3.42 and are inferred to have had enormous energies , @xmath410@xmath5 erg and @xmath6 erg for grb 970508 and grb 971214 respectively .
it is currently believed that most gamma - ray bursts occur at such great distances that their mean energy is at least 10@xmath7 erg _ in gamma - rays alone _ , times an uncertain beaming factor that might reduce the energy by up to 100 at the expense of requiring many more events .
this developing paradigm was challenged last month by the discovery ( galama et al .
1998ab ) of a supernova , sn 1998bw , type ib ( sadler et al .
1998 ; lidman et al .
1998 ) and later ic ( patat & piemonte 1998 ) , within the 8 arc minute error box of grb 980425 ( soffita et al . 1998 ) .
extrapolation of the supernova light curve implied an explosion time consistent with the grb , an extremely unlikely occurrence unless the two were associated .
further , the supernova was unusual , presenting a radio luminosity 100 times brighter than typical type ib s , brighter in fact than any supernova ever observed before ( wieringa et al .
moreover , relativistic expansion was inferred ( kulkarni et al 1998 ) , the spectrum was unusual ( lidman et al .
1998 ; patat & piemonte 1998 ) and the light was curve brighter ( galama et al .
1998b ) than typical for a ib or ic .
_ in toto _ the case for a grb - supernova association is compelling . however , the redshift to the barred spiral galaxy where the supernova occurred is only 0.0085 ( tinney et al .
1998 ) and the burst was not an extraordinarily bright one .
the duration and count rate for beppo sax were in fact comparable to grb 971214 at a red shift of 3.42 . from this
we infer that the gamma - ray burst , which lasted 30 seconds , had an energy that was about 10@xmath8 erg , or 10@xmath9 - 10@xmath10 times fainter than a typical cosmological grb .
the batse detector on compton gamma - ray observatory also saw the burst ( galama et al .
1998b ) for about 35 s and inferred a total energy of @xmath11 erg in gamma - rays .
batse saw no emission above 300 kev for this burst , making it another example of the so called no high energy " grb s , about 25% of the batse sample . at this luminosity ,
other grb s like grb 980425 would have been invisible had they occurred 20 times farther away , so unless this was an extremely serendipitous observation , there must be a very high spatial density of these events , perhaps thousands of times that of the classical " batse bursts ( _ modulo _ the beaming factors ) .
this requires a source that is very common in nature . indeed , since batse observations can be explained with an event rate of 10@xmath12/yr / l@xmath13 galaxy ( wijers et al . 1998 )
, this suggests an event at least 1% as frequent as supernovae . in order to explain the brilliance of sn 1998bw ,
if it is powered by the decay of radioactivity like other type i s , it is necessary to synthesize and eject @xmath14 m of @xmath0ni in the explosion .
since this was not a type ia supernova - and we know of no way to make grb s or strong radio sources based on type ia supernovae - we invoke massive stars .
but then the large @xmath0ni mass requires , in traditional models , both a very massive star and a high explosion energy .
the energy must also be large in order to accelerate the mass - several times that in a typical type ib supernova - to the observed high velocities and to make the light curve peak in only 17 days ( galama et al .
finally , we are prejudiced by the belief that grb s require stars so massive that the neutrino powered hot bubble " mechanism for supernova explosion fails ( woosley 1993 ) .
this also leads us to consider stars whose main sequence mass was over 30 m. as we were writing our paper , the preprint by galama et al .
( 1998b ) appeared which references similar conclusions , at least a massive stellar explosion with large energy , reached in a paper by iwamoto et al .
we have not seen that paper and our work has proceeded independently . in the following sections we describe the modeling of the supernova explosion ,
calculate the fraction and energy of relativistic mass ejected , and examine the model light curve and spectrum .
we also attempt to understand how the supernova might have made a grb .
the interaction of the supernova shock with circumstellar material has an appealing physical basis and might be expected to occur frequently , but the gamma - ray energy requirements even for this faint burst are large and are not obtained ( in spherical symmetry ) even for very violent explosions .
we do find models that agree well with the multi - band photometry of the supernova and from these are able to make predictions about the spectrum - unknown to us as of this writing .
the large explosion energy and lack of a straightforward way of making the grb in spherical symmetry suggest that something unusual happened in sn 1998bw . in the conclusions we discuss what it may have been .
the models we use , which might eventually be tuned to give better agreement , are based upon massive stars - 25 - 35 m on the main sequence , that have lost their hydrogen envelope and perhaps even their helium shell . for 25 m
, this may require membership in a close binary ; for 35 m , radiative mass loss will suffice .
once the helium core is uncovered rapid mass - dependent mass loss may commence ( langer 1989 ) that removes a portion of the helium shell .
we thus experiment with both the helium cores and the carbon - oxygen cores of these massive stars .
all calculations of the explosion and expansion were carried out using the kepler code ( weaver , zimmerman , & woosley 1978 ) . the light curve and approximate spectra
are calculated using a different approach ( @xmath153.4 , 3.5 ) .
our first model uses the 9.12 m helium core of a 25 m main sequence star similar to the one evolved to presupernova by woosley & weaver ( 1995 ) . because we are interested in the correct density distribution in the atmosphere of the star ( for shock acceleration ) , it was important that the surface of the helium star be fine zoned and in thermal and hydrostatic equilibrium .
it takes time for the star to relax into this equilibrium and this can not be accomplished by a star that is already exploding .
so rather than try to make a `` stripped down '' helium core , we used the 25 m model at carbon ignition to construct our model .
the hydrogen envelope was removed ( down to a hydrogen mass fraction of 0.01 ) and the rezoner allowed to prepare a very finely zoned surface as the outer helium layer expanded . a surface boundary pressure of 10@xmath16
dyne @xmath17 was necessary to keep the star numerically stable .
this did not appreciably affect the structure .
10@xmath18 m ( 8 zones ) into the atmosphere , the pressure exceeded this boundary value by 10 and the radius had decreased by only 9% .
this boundary pressure was of course removed when the star exploded .
the outer zone was @xmath19 m. this atmosphere was allowed to relax into thermal and hydrostatic equilibrium and the star was then evolved , without farther mass loss , through neon , oxygen , and silicon burning to the presupernova state . as a presupernova ,
the star had a luminosity of @xmath20 erg s@xmath21 and radius @xmath22 cm .
as before , the iron core mass was 1.78 m. this star was then exploded using a piston as described in woosley & weaver ( 1995 ) .
the final kinetic energy at infinity was varied ( table 1 ) .
this series of models is called he9 with a suffix to indicate explosion energy .
three other presupernova models were similarly constructed .
the next used the 6.55 m carbon - oxygen core of the 25 m star at carbon depletion .
fine surface zoning was again engineered with outer zones typically @xmath23 g. the radius of the star at explosion was 1.22 @xmath24 cm and the luminosity @xmath25 erg s@xmath21 .
models from this series are denoted co6 .
two additional models were extracted from a 35 m star at carbon ignition .
this gave a helium core of 14.13 m ( models he14 ) and a carbon - oxygen core of 11.03 m ( models co11 ) .
these models were all exploded using pistons parameterized so as to give a specified kinetic energy at infinity for the ejecta ( woosley & weaver 1995 ) .
typical values for the `` @xmath26 '' parameter were 10 - 20 .
the piston was located at the edge of the iron core in each case ( 1.78 m in the 25 m derived models ; 2.03 m , for the 35 m derived models ) .
nucleosynthesis was followed as in weaver , zimmerman , & woosley ( 1978 ) using the nuclear reaction set described in woosley & weaver ( 1995 ) .
the final kinetic energies and abundances of @xmath10he , @xmath27o , @xmath28si , and @xmath0ni are given in table 1 .
the first model ever proposed for gamma - ray bursts was supernova shock break - out ( colgate 1969 , 1974 ) .
the outer layers of the star are heated by the eruption of the strong shock wave , then release their energy as the layers expand .
we followed here the emergence of the shock using the kepler hydrodynamics code ( weaver , zimmerman , & woosley 1978 ) and a simple prescription for the opacity - electron scattering based upon a full solution of the saha equation ( ensman & woosley 1988 ) .
as previously noted , the zoning of the outer layers was very fine , logarithmically smooth down to 10@xmath29 m. the radiation transport for this early stage was calculated using a simple single temperature model of flux - limited radiative diffusion .
it is known that this approach underestimates the temperature of the burst by up to about two ( ensman & burrows 1992 ) , but the luminosity should not be far off and both should suffice for present purposes .
the results for a representative sample of our models are given in table 2 .
typical burst luminosities are 10@xmath30 - 10@xmath31 erg s@xmath21 for up to several seconds
. typical photon energies ( 3 kt ) are about a kev . while bright and potentially detectable , this burst of radiation is over two orders of magnitude softer and four orders of magnitude fainter than what was seen in grb 980425 .
as the shock progresses through the outer layers of the star , it accelerates .
if the density gradient is steep enough and the shock strong enough , a portion may even become become relativistic .
analytic solutions of ultra - relativistic shocks and semi - analytic solutions of mildly relativistic shocks exist ( johnson & mckee 1971 ; mckee & colgate 1973 ; gnatyk 1985 ) . for an exponentially declining density profile ,
the product of the lorentz factor ( @xmath32 ) and the velocity of the shock ( @xmath33 where @xmath34 is the speed of light ) is given by ( gnatyk 1985 ) : @xmath35 where @xmath36 is a geometric factor set to 2 for spherical symmetry , and @xmath26 is determined , via simulations , to be @xmath40.20 .
we can use this scaling relation to estimate the energy ejected as a function of @xmath32 for lower mass zones than we are able to carry in our present ( newtonian ) hydrodynamical calculation . in figs . 1 and 2 the quantities @xmath37 and q
= @xmath38 are plotted as functions of the mass outside of radius @xmath39 .
the density and radius are evaluated in the presupernova star ; @xmath32 and @xmath40 are evaluated after the matter has reached the coasting phase .
the scaling relation for @xmath32 is not precise because it neglects the internal energy deposited by the shock and the subsequent acceleration that energy causes ( fryer & woosley 1998a ) .
however the near constancy of q suggests that we can extrapolate the well determined sub - relativistic solution calculated here to higher @xmath32 s .
taking a representative range of q @xmath41 and a scaling relation between @xmath37 and external mass m = 10@xmath42 ( fig .
1 ) , we estimate the kinetic energy , @xmath32mc@xmath43 , contained in material having @xmath44 to be 10@xmath45 - 10@xmath46 erg . for @xmath32 of 3
the range is 10@xmath31 - 10@xmath47 erg .
this is several orders of magnitude less than required to produce the grb .
sub - relativistic matter is also unlikely to produce the burst . to carry 10@xmath8
erg requires a minimum of @xmath23 g. this matter will interact with its own mass before giving up its energy . for a mass loss rate of 10@xmath18 m y@xmath21 and speed 10@xmath16 cm s@xmath21 , the radius where this will happen is at least @xmath410@xmath48 cm .
the light crossing time for this region is @xmath4910@xmath10 s , so the burst would be too long and faint .
raising the mass loss can give a smaller interaction radius and shorter burst , but at the expense of becoming optically thick to the gamma - rays that are produced .
it seems likely that an enduring hard x - ray flash will be created - an analogue to what was seen in sn 1993 j ( leising et al .
1994 ; fransson , lundquist , & chevalier 1996 ) .
this lasted about a hundred days at 50 - 100 kev .
an additional concern is that the radio emission implies relativistic expansion even days after the grb occurred ( kulkarni & frail 1998 ) .
there is roughly @xmath50 erg in the outer 10@xmath51 m of ejecta of our models here , all moving at about 1/3 c. this could certainly provide a bright radio source , but the expansion would not be relativistic .
ubvri photometric observations of sn 1998bw have been reported by galama et al .
( 1998b ) and show the supernova falling in brightness when first observed ( 0.6 days after the grb 980425 ) ) , then rising to a maximum of @xmath52 ( a distance of 36 mpc based on the object s redshift , h@xmath53 km s@xmath21 mpc@xmath21 , and @xmath54 = 0.2 mag , is used throughout this discussion ) .
we have used these observations to estimate the bolometric luminosity " ( @xmath55 ) by integrating over the @xmath56 photometry . to do so , we extend the spectrum beyond the i - band using a blackbody tail and beyond the u - band with a spline .
the results are not sensitive to the treatment of the infrared , but there is some ambiguity in the treatment of the ultraviolet .
our procedure here is influenced by previous analyses of supernovae that had broad wavelength coverage ( e.g. , type ic sn 1994i ) .
type i supernovae of all subclasses are affected by line blanketing and it is important to cut off the ultraviolet spectrum quickly relative to the best fitting blackbody .
the photometric evolution of this object is consistent with other objects which have a rapidly falling ultraviolet spectrum .
the derived bolometric flux would only be in significantly error if there were a large amount of flux below 3000 .
this appears unlikely except at the earliest times ( less than three days after the grb ) .
the derived bolometric light curve is given in table 3 and in figs .
3 and 4 .
these observational data were used to discriminate among possbile models .
each model was evolved with the kepler hydro - code to 10@xmath57 seconds after explosion at which point a link was made to a multi - group radiation transport code , eddington ( eastman & pinto 1993 ) .
this code solves the time - dependent transport equation , in the co - moving frame , simultaneously determining the gas temperature by balancing heating and cooling .
the heating rate includes energy deposition by gamma - rays from radioactive decay .
gamma - ray transport was computed using a single energy group approximation to compute the transport each of gamma - ray line ( woosley et al .
1994 ) . for the eddington light curve calculations ,
the kepler grid , which consisted of 370 to 700 zones , was remapped onto a grid of 80 zones .
the composition was artificially moderately mixed " , which is to say a running boxcar average using a grid 1 m wide was calculated sliding the grid out through the star . for those models that had a helium shell ,
this was not sufficient mixing to bring @xmath0ni up into the helium . bringing @xmath0ni into the helium layer would probably produce a type ib , not ic supernova ( woosley & eastman 1997 ) .
the opacity included contributions from he i - ii , c i - vi , o i - viii , si i - x , s i - x , ca i - xii , fe i - xiv , co i - xiv and ni i - xiv .
processes included inner shell and valence shell photoionization , bremsstrahlung , electron scattering , and line opacity from 90,000 lines , which was represented using the expansion opacity described by eastman & pinto ( 1993 ) .
the light curve calculations assumed local thermodynamic equilibrium ( lte ) .
gas excitation and ionization was computed by solving the saha - boltzmann equation at the local temperature and density . because the density is so low here
, the assumption of lte is questionable .
this asuumption remains approximately valid because the gas is radiatively driven into thermal equilibrium .
but as the ejecta becomes more transparent , the assumption of lte gets progressively worse . in general
, we find that lte tends to overestimate the population of excited states , underestimate the ionization , and underestimate the gas temperature . for the present light curve calculations ( figs 3 - 5 ) , the frequency grid consisted of 500 groups covering the range @xmath58 angstroms .
because of this low resolution , spectral features computed by the light curve code are smeared , but the spectrum is still adequate for photometry .
the best fit to the light curve and photometry is for our lowest mass , highest energy explosions ( table 1 ) , those based on the 6 mcarbon - oxygen core .
even these models do not rise fast enough to agree with observations during the first few days .
more mixing of @xmath0ni to nearly the surface of the explosion would give a more rapid rise , but in one dimension this mixing would keep a larger volume hot and ionized at late time and increase the photospheric radius .
this would make the supernova too red .
another possibility is that the pre - explosive star had a helium layer and a larger radius .
the release of shock deposited energy by helium recombination would then give a brief `` plateau '' in the light curve as is often calculated for type ib models .
there are some indications in the data of the first few days that the supernova initially faded slightly
. this would be consistent with helium recombination .
alternatively the explosion was not spherically symmetric ( @xmath155 ) . in order to evaluate the effects of the lte approximation and low frequency resolution , we carried out a higher resolution , non - lte calculation of the spectrum of model co6c near maximum light ( fig .
6 ; 14.4 days ) .
this calculation assumed steady state between energy deposition and emission .
gamma - ray transport was computed with the monte carlo code fastgam ( pinto & woosley , 1988 ) using a frequency grid of 30,000 groups and a spatial grid of 41 radial zones .
ions included were he i - ii , c i - iv , o i - iv , si i - iv , s i - iv , ca i - iv , fe i - iv and co i - iv .
the broadband photometry predicted by this model was shown in fig . 5 as solid points .
the agreement with the observations is much improved over the predictions of the lte calculation . in particular , the predicted u band flux is a magnitude brighter in the non - lte calculation .
6 shows the spectrum predicted by the non - lte calculation of model co6c just prior to peak light , ( the calculation is at 14.4 days ) .
although we have not yet had access to any optical spectroscopy of sn 1998bw , the maximum light spectrum of co6c has many of the properties displayed in the maximum light spectrum described by patat & piemonte ( 1998 ) : it peaks near 5400 angstroms and shows strong absorptions by c ii , o i , o ii , si ii , s ii and ca ii . the model does have a he i @xmath59 absorption feature , which patat says was not present in sn 1998bw .
however , it is weak , highly blue shifted , and could easily be mistaken for something else .
also , the he i @xmath60 is very weak in model co6c , and blended with c ii and o ii , consistent with patat s report on sn 1998bw .
the velocities here are very high . in the unmixed model , most of the helium ( which came from photodisintegration in model co6c ) was moving between 0 and 12,000 km s@xmath21 ; carbon was appreciably abundant ( over 1% by mass ) only at speeds greater than 25,000 km s@xmath21 ; oxygen was abundant over 14,000 km s@xmath21 ; magnesium , 15,000 km s@xmath21 and up ; silicon 12,000 to 26,000 km s@xmath21 ; calcium , 12,000 - 15,000 km s@xmath21 ; and cobalt ( @xmath0ni ) was found between 0 and 14,000 km s@xmath21 .
this inverted speed distribution for helium and heavier elements might be a distinctive feature in the spectrum of a co - explosion as opposed to that of a helium star . in a helium star
there might be a bimodal distribution of helium in velocity . in a co star high velocity helium
is weak ( arbitrarily we defined the outer boundary of the co model as where helium went to 1% by mass in the 25 m star igniting carbon burning ) .
the velocities here are higher than reported by patat & piemonte ( 1998 ) . in a later paper , when spectroscopic data is available , we hope to treat the spectral evolution of sn 1998bw in greater detail . however , from the information at hand it seems that , photometrically at least , sn 1998bw is well modeled as the explosion of a carbon - oxygen core of 6 m with a kinetic energy of @xmath61 erg which naturally yields a @xmath0ni mass near 0.5 m. the fact that we used a co core without an appreciable layer of helium still in place is in part an expedient
. it may well be that a helium core of the same mass and explosion energy would have worked just as well .
if detailed spectroscopic analysis shows that the high velocities , e.g. , of model co6c , are not present , this may indicate an asymmetric explosion ( @xmath155 ) .
so why have we not observed events like this before ? or have we ?
wang and wheeler ( 1998 ) have compared the correlation of supernovae with grb s and find a positive correlation with type ic s , but no correlation between grb s and other supernovae .
there have only been 16 supernovae classified as type ic during the six year period 1992 - 1997 as listed in the asiago catalog ( barbon et al .
presumably many others have been missed , but they do not affect the argument . the batse sky coverage is about one third , so one might expect about 5 sn ic - grb correlations if all sn ic are grbs .
but there may also be considerable variation in the grb s from supernova to supernova .
perhaps only the stars with the highest mass and biggest explosion energies make a visible grb , or maybe they must be observed from a certain angle .
nevertheless , it would be interesting to search the known grb error boxes for subsequent supernovae - but when would the supernova be discovered ?
two weeks later , a month ?
we checked only three cases because we knew them to be unusually luminous type ic s .
these were sn 1992ar ( discovered as part of the calan / tololo survey , hamuy & maza 1992 ) ; sn 1997cy , ( discovered as part of the mount stromlo abell cluster supernova search , germany et al .
1997 ) ; and sn 1997ef ( nakano & sano 1997 ) .
sn 1992ar was discovered in late july , 1992 and grb 920616 occurred about two sigma from the sn s position .
sn 1997ef , discovered on november 25 , 1998 , has also been pointed out by wang and wheeler along with its coincidences ( within 3-sigma error boxes ) with grb 971115 and grb 971120 .
while it is interesting that both of these supernovae have a reasonable grb candidate , neither is a particularly compelling case because of the large separation between grb and the centrod of the error box .
however , the situation is different for sn 1997cy . this supernova ( not in wang and wheeler s list ) had a bizarre spectrum , with broad ic - like lines like observed in sn 1997ef and 1992ar , but also a h@xmath26 line with broad and narrow components .
sn 1997cy was also the most luminous supernova ever discovered , having @xmath62 at maximum .
grb 970514 , a burst with a smaller than typical error box ( 3-degrees ) , occurred less than a degree away at a time compatable with the discovery and pre - discovery images .
this object is the subject of a paper by germany et al .
so perhaps sn 1998bw is not an isolated case .
however , we want to state clearly that we do not believe that all or even a majority of type ib ( or ic ) supernovae make grb s .
most of these supernovae are very well modelled by a lower mass explosion ( 3 - 4 m helium core ) that makes about one - third as much @xmath0ni as sn 1998bw and expands with moderate energy @xmath410@xmath7 erg ( e.g. , woosley & eastman 1997 ) .
even the more massive stars and unusual explosions studied here might only make a grb when viewed at certain angles .
as we discuss in the next section , the grb is probably beamed while the supernova is certainly visible at all angles .
we expect a grb supernova association only in the unusual case and two - thirds of these will be missed by batse .
sn 1998bw was and continues to be an unusual supernova .
when modeled as a spherically symmetric explosion , it requires an energy over 20 @xmath63 erg , a @xmath0ni mass over 0.45 m , rapid expansion , high stellar mass , and high mass loss rate ( to explain the radio ) .
of course the most unusual property of sn 1998bw was its proximity to grb 980425 .
we have assumed here that the two are related and have looked for ways the supernova might make the burst . for our one - dimensional models we found none .
however , we do find good agreement with the multiband photometry of galama et al .
( 1998b ) , and the bolometric light curve integrated from that data , and the explosion of a 6 m core of carbon , oxygen , and heavy elements with final kinetic energy @xmath64 erg .
the explosion leaves behind a 1.78 m ( baryonic mass ) object , presumably a neutron star and makes about 0.5 m of @xmath0ni .
however the mass of the remnant and the explosion energy were not calculated in a consistent way , but were free parameters .
we do not think it is critical that our best fit was a carbon - oxygen core and not a helium core ; the key quantity is the energy to mass ratio .
type ic supernovae have weak helium lines chiefly as a consequence of weaker mixing between the helium and @xmath0ni shells than in type ib ( woosley & eastman 1997 ) .
even this very energetic explosion is too faint the first few days of the supernova .
there are two explanations for this - either there was a helium layer with a larger photospheric radius than the carbon - oxygen core used here that gave a brighter plateau " before the radioactive decay energy diffused out , or the explosion was asymmetric , ejecting some @xmath0ni almost to the surface at some angles - a very mixed model .
however , spherically symmetric mixing would have given a larger photosphere and perhaps a redder supernova than was observed ( woosley & eastman 1997 ) .
helium may be present in the spectrum even in our carbon - oxygen core models , but it is chiefly from photodisintegration and would be the slowest not the fastest moving ejecta .
high velocity helium would be a signature of a helium star .
all in all , though the parameters may be extreme , especially the explosion energy , one could model sn 1998bw in a qualitatively similar way to other type ib and ic supernovae , that is if it were not the origin of grb 980425 .
but we believe that it was .
so what happened ? can nature really provide @xmath65 erg to a supernova whose main sequence mass was over 25 m ?
current belief ( e.g. , burrows 1998 ; fryer 1998 ) is to the contrary .
if anything , the explosion actually becomes weaker as one goes to larger mass .
the iron core is larger and can potentially provide more neutrinos , but it is also close to criticality and the mass flux from the imploding mantle of the star is formidable .
it is very difficult to stop the implosion before the neutron star gives way and collapses to a black hole .
and so it may be that something else happened here , that the explosion was not spherical and powered by neutron star formation , but very asymmetric and powered by jets from black hole formation .
bodenheimer and woosley ( 1983 ) first considered such an outcome to black hole formation and found that a supernova still resulted .
woosley ( 1993 ) and hartmann & woosley ( 1995 ) emphasized jet production and proposed an association of this model with gamma - ray bursts .
initially this model was referred to as the `` failed supernova '' ( because the prompt supernova mechansim failed ) , and later as the `` collapsar model '' ( woosley 1996 ) , because it was the outcome of a collapsed star .
a model having very similar characteristics , called the `` hypernova '' , has been discussed by paczynski ( 1997 ) .
fryer & woosley ( 1998b ) have also discussed setting up very similar conditions in the merger , by common envelope , of a stellar mass black hole and the helium core of a massive supergiant star .
current two dimensional studies of the collapsar model by macfadyen and woosley ( 1998ab ) are encouraging .
specifically they find , in the collapse of a 14 m rotating helium star to a black hole , an accretion rate of over 0.1 m s maintained for about 10 s as the black hole grows from 2 m to 7 m. the kerr parameter , @xmath66 , grows to @xmath49 0.9 early on . for these conditions , popham ,
woosley , & fryer ( 1998 ) find that the annihilation of neutrinos radiated from the viscous disk deposits up to 10@xmath7 erg s@xmath21 along the rotational axis of the black hole .
thus energies as much as 10@xmath5 erg are potentially available .
some of this energy goes into accelerating relativistic jets along the rotational axes , but more may go into ejecting a lot of mass at lower speeds .
macfadyen & woosley have not calculated the evolution beyond 15 s. large amounts of energy can also potentially be extracted from the rotation of the black hole ( e.g. , meszaros & rees 1997 ) .
viewed this way , grb 980425 was a low energy analogue of the enormously more luminous
classic " grb s .
both are produced by black hole accretion , but in grb 980425 the jet energy was weaker and @xmath32 , along our line of sight , lower .
perhaps if we had viewed grb 980425 straight down the axis a more powerful , harder grb would have been seen . or maybe the helium core mass , rotation rate , and therefore black hole accretion rate were not so extreme in grb 980425 as in other grb s .
viewed from the side though , in any case , the emission from the high @xmath32 jet would have been supressed and spread over a much longer time , probably invisible .
but even at our angle there may have been , say , 10@xmath12 - 10@xmath29 m moving with @xmath67 10 .
colliding with the pre - explosive mass loss at about 10@xmath68 - 10@xmath48 cm , this would have made the observed burst ( meszaros & rees 1993 ) . if we had seen sn 1998bw at still lower latitudes , the grb would have been missed .
once spherical symmetry is abandoned an entirely different solution becomes possible for the supernova .
if matter can fall in to close to the black hole and come out again ( macfadyen & woosley 1998b ) , the production of @xmath0ni is not directly tied to the shock energy and pre - explosive density structure of the star .
it is as if @xmath0ni could be made `` convectively '' .
the one number we view with some confidence here is that sn 1998bw made about 0.5 m of @xmath0ni . but
suppose it could do so while only ejecting a few solar masses of heavy elements and helium .
then the correlation between @xmath0ni mass and explosion energy is lost .
sn 1998bw could have been a slower moving , lower energy explosion ( shared by a smaller ejected mass ) than we have calculated here and still have peaked as early as it did .
it is unfortunate that so many questions remain unresolved .
first , is it certain that sn 1998bw and grb 980425 are the same thing ?
future missions with smaller error boxes ( e.g. , hete- 2 ) should show if this is the case . finding other historic type ic supernovae in coincident with grb locations from batse would also lend credence to this identification .
we have given two possible examples .
there may be more .
can a combination of theory and obsevation still tell us what happened in this supernova / grb ? continued spectroscopic monitoring of the supernova will obviously be an important diagnostic as the supernova enters ( has in fact already entered ) its nebular phase . what widths and asymmetries are apparent in the lines of oxygen , iron , helium , silicon , calcium , and carbon ?
are the high velocities of model co6c really there ? what is the mass of the ejecta ?
multi - dimensional modeling of the explosion and radiation transport in the collapsar model should also show whether it can explain the observations .
if it does not , perhaps something even more interesting has occurred .
the authors gratefully acknowldge helpful conversations on the subjects of gamma - ray bursts and sn 1998bw with roger chevalier , dale frail , chris fryer , lisa germany , chryssa kouveliotou , andrew macfadyen , bob popham , and eileen sadler .
this research has been supported by the doe ( w-7405-eng-48 ) , nasa ( mit sc a292701 ) , and the nsf ( ast-97 - 31569 ) lcccccc co6a@xmath69 & 6.55 & 5.5 & 0.06 & 3.3 & 0.28 & 0.32 co6b & 6.55 & 15 & 0.14 & 3.1 & 0.36 & 0.42 co6c & 6.55 & 22 & 0.20 & 2.9 & 0.40 & 0.47 co6d & 6.55 & 28 & 0.26 & 2.8 & 0.42 & 0.49 co11a & 11.0 & 9.1 & 0.09 & 6.3 & 0.54 & 0.68 co11b & 11.0 & 25 & 0.21 & 5.9 & 0.70 & 0.84 he9a & 9.12 & 3.7 & 2.4 & 3.0 & 0.35 & 0.51 he9b & 9.12 & 7.7 & 2.4 & 2.9 & 0.39 & 0.58 he9c & 9.12 & 21 & 2.5 & 2.5 & 0.54 & 0.77 he14a & 14.1 & 4.2 & 2.8 & 6.2 & 0.46 & 0.73 he14b & 14.1 & 10 & 2.8 & 6.0 & 0.51 & 0.86 | recently a type ic supernova , sn 1998bw , was discovered coincident with a gamma - ray burst , grb 980425 .
the supernova had unusual radio , optical , and spectroscopic properties . among other things , it was especially bright for a type ic and rose quickly to maximum .
when modeled in the usual way as a spherically symmetric explosion , this requires a large mass of @xmath0ni , 0.45 - 0.60 m , a quite massive star , and a very large explosion energy .
we explore here models based upon helium stars in the range 9 - 14 m and carbon - oxygen stars 6 - 11 m which experience unusually energetic explosions ( kinetic energy @xmath1 erg ) .
bolometric light curves and multi - band photometry are calculated and compared favorably with observations .
no spectroscopic data are available at this time , but both lte and non - lte spectra are calculated for the model that agrees best with the light curve , a carbon - oxygen core of 6 m exploded with a kinetic energy of @xmath2 erg .
we also examine potential mechanisms for producing the observed gamma - ray burst ( grb ) - shock break - out and relativistic shock deceleration in circumstellar material . for
spherically symmetric models , both fail to produce a grb of even the low luminosity inferred for grb 980425 .
the high explosion energies required to understand the supernova are in contrast to what is expected for such massive stars and may indicate that a new sort of explosion has been identified , possibly the consequence of a collapsar ( woosley 1993 , 1996 ) whose main sequence mass was @xmath3 m ( helium core mass 14 m ) .
indeed a more likely explanation for what was seen is a highly asymmetric explosion in which the grb was produced by a relativistic jet , perhaps viewed obliquely , and only a fraction of the total stellar mass was ejected , the remainder accreting into a black hole . the ejected mass ( but not the @xmath0ni mass ) , explosion energy , and velocities
may then be smaller .
other associations between luminous type ic supernovae and grb s may exist and should be sought , but most type ib and type ic supernovae do not make grb s .
submitted to _ the astrophysical journal _ 0.2 in |
to show that our approach can accurately probe the mechanical properties of _ e. coli _ and _ b. subtilis _ cells , we applied short pulses of hydrodynamic force to the cells and monitored their response ( fig . 2 ) .
for this , the cells were grown in excess nutrients initially without flow . as the cells formed long , straight filaments , we applied a series of pulse - like flow .
each of these pulses lasted for less than 10 seconds so that the effect of growth is negligible .
2 and supporting video s1 show our typical experimental result .
the cells bent transiently at each pulse , and fully recovered their shape immediately after the applied force was removed .
we can explain the experimental results using the theory of linear elasticity .
in the theory , the long filamentous cell can be considered as a cylindrical beam bending by a hydrodynamic force ( si text ) .
shown in fig .
2b is the comparison between data and theory for the tip position ( thus the degree of cell bending ) of the cell during a representative experiment in fig .
2a . from these experiments
, we extracted a flexural rigidity of @xmath2 n@xmath3m@xmath4 , comparable to the result of @xmath5 n@xmath3m@xmath6 obtained by mechanical manipulation of _
e. coli _ cells using optical traps @xcite .
assuming a cell wall thickness of @xmath7 nm , this corresponds to a young s modulus of @xmath8 mpa .
it should be noted that the small mismatch between the bacterium s diameter and the width of the end of the channel has to be taken into account , see the si text for further details .
we also show in the si text that the turgor pressure does not contribute to the restoring torque .
similar experiments were performed for _ b. subtilis _ ( see si fig .
5 ) giving a young s modulus of @xmath9 mpa , consistent with past results @xcite .
from these results , we conclude that both _
e. coli _ and _ b. subtilis _ cells respond to fast mechanical perturbations as a linear elastic rod .
the elastic deformation and recovery of the cells to pulse - like forces is consistent with the view that _ e. coli _ and _ b. subtilis _ cells can be considered as elastic cylindrical beams @xcite . the linear elasticity theory
, however , can not explain the observations that bacteria can also deform plastically @xcite . for plastic deformations
, we noticed that the cells in refs .
@xcite were grown with geometric constraints for a duration comparable to or longer than their mass - doubling time .
since these geometric constraints may exert forces on the cells , we hypothesized that external forces were transduced to the cell - wall synthesis machinery during growth .
therefore , the force - transduction may cause differential growth , thus deformations , of growing cell walls . as a first step to testing our hypothesis
, we grew the cells with a constant hydrodynamic flow ( of the growth medium ) .
the filamentous cells grew into a curved shape because of the continuous hydrodynamic forces ( fig .
the curved shape of the cell , however , was not easily distinguishable from elastically deformed cells in fig . 2 . to confirm that the cells deformed plastically , we abruptly switched off the flow when the cells significantly bent
. the cells did not recover their straight shape immediately , i.e. , they deformed plastically ( fig .
3a and supporting video s2 ) .
we repeated the experiments with non - growing filamentous cells , and the cells did not show plastic deformations despite long - term application of forces ( see si fig .
we conclude that plastic deformations require both cell wall synthesis and application of forces for a duration over which substantial cell wall synthesis occurs . the plastic deformation experiments in fig .
3 unraveled an important feature of bacteria cell walls .
as the external forces were removed , the cells always partially snapped back .
this is clearly seen in the time trace of the cell s tip position ( fig .
the partial snap back suggests that the cell bending in fig . 3 has _ both _ elastic and plastic components . in other words , external forces not only cause elastic bending , but may also lead to differential growth which further bends the cell
the two components may be decoupled by removing the external forces , upon which the elastic component should vanish instantly . to test our hypothesis
, we developed a quantitative , phenomenological model that incorporates force transduction and growth .
we describe below our model and its comparison with data .
we found the data and theory in good quantitative agreement .
our starting point for understanding the data is the theory of dislocation - mediated growth that we developed recently @xcite .
the model was inspired by recent experimental observations that , in both gram - positive @xcite and gram - negative @xcite bacteria , mreb proteins , associated with cell wall growth , move processively along the cell s circumference ( fig .
processive circumferential insertions resemble the climb of edge dislocations in a two - dimensional crystal @xcite , and thus one can utilize the theory of dislocations well developed in materials science . in the previous version of the growth model
, we studied the motion of dislocations on a perfect cylinder in the absence of external forces @xcite .
therefore , the previous version can not account for deformations .
we extended our theory to allow for differential growth by transduction of external stress to dislocations ( growth machinery ) .
in our theory , tensile ( stretching ) stress can have two different effects : ( 1 ) increase the speed of processive motion of dislocations ( fig .
4a ) or ( 2 ) reduce the activation energy for nucleation ( fig .
compressive stress will have the opposite effect .
the forces on the elongation machinery are given by the peach - koehler force @xcite acting on dislocations : @xmath10 where @xmath11 is the levi - civita tensor , @xmath12 is the stress tensor and @xmath13 the burgers vector characterizing the dislocation . for the cylindrical geometry of the cells in our experiment
, the theory predicts that the @xmath14 component of the stress tensor is the one which couples to the circumferential growth , where @xmath15 is the coordinate along the cell s long axis , see fig .
1 . as the cell grows in the presence of a constant flow , a cell wall of thickness @xmath16 , length @xmath17 ( measured from the channel end to the cell tip ) and radius @xmath18 experiences a mechanical stress that depends on its position in the cell ( fig .
4a ) : _ ( ) = , [ eq : stress ] where @xmath19 is the azimuthal angle of the cylinder along the circumference ( @xmath20 for the cell front facing the flow , and @xmath21 for the back ) , @xmath22 the force per unit length induced by the flow , @xmath15 the position along the cylinder s axis and @xmath14 the mechanical stress induced by the flow .
the surface directly facing the incoming flow feels a tensile stress ( surface - stretching force ) , whereas the other side feels a compressive stress ( which acts to shorten the distance between adjacent glycan strands ) .
accordingly , the cell walls grow differentially , growing faster on the upstream side of the flow . in our experiments ,
the external hydrodynamic force would increase the peach - kohler force by @xmath23 pn for the cell front ( at the channel end , for @xmath24 @xmath25 m ) , and decrease the force by the same amount on the opposite side .
this force is comparable , for example , to the one generated by rna polymerase @xcite .
it is also comparable to the force due to the turgor pressure ( @xmath8 pn for _ e. coli _
@xcite ) , and thus we expect the mechanical stress to significantly change the activation energy for enzymatic reactions ( see @xcite , p. 165 ) .
this differential growth leads to a dynamically deformed cell shape . due to the large aspect ratio of the filamentous bacteria in our experiments , a relatively small asymmetry in growth leads to a large angular deflection . in the si text
we show that the angular deflection due to differential growth in a small segment of the cell is given by : = , [ amplify ] where @xmath26 is the net difference on opposite sides of the cell wall due to the asymmetric growth .
therefore , for a small segment of length @xmath17 the relative asymmetry @xmath27 is amplified " by the aspect ratio @xmath28 .
this amplification of differential growth often appears in nature , in a diverse range of systems such as the mechanics of plants and of the human gut @xcite .
a key result that emerged from our theory is a single dimensionless parameter @xmath29 that quantifies the relative contribution of the elastic and the plastic components of deformation .
the parameter @xmath29 depends only on the intrinsic physical parameters of the cell , independent of the experimental conditions .
we show in the supplementary information that = p r / y h , [ eq : e_v_p]where @xmath30 is the turgor pressure , @xmath18 the cell s radius , @xmath31 the young s modulus , and @xmath16 the thickness of the cell wall . for @xmath32 ,
the elastic effects dominate , and there should be full snapback . for @xmath33 ,
the nature of deformation is plastic , and there would be no snapback .
we find @xmath29 is of order 1 for the measured values for _ e. coli _
( see supplementary table 1 ) ,
i.e. , the snapback should be _
b. subtilis _ also has a similar @xmath29 value because _ b. subtilis _ has a proportionally thicker cell wall and higher turgor pressure ( supplementary table 1 ) .
illustration of the differential growth driven by coupling between mechanical stress and cell wall synthesis .
( a ) tensile ( stretching ) stress increases speed of processive motion of strand elongation machinery .
compressive stress has the opposite effect .
black arrows indicate the trajectories of activated dislocations .
( b ) similarly , tensile stress will lower the activation energy for insertion of a new glycan strand , and compressive stress will increase it .
( c ) asymmetric insertion of dislocations is enhanced by the large aspect ratio of the filamentous bacteria , leading to the large observed angular deflections.,width=8 ] our predictions can be tested quantitatively by measuring the magnitude of snapback and relating it to @xmath29 .
note that the local curvature is proportional to the local growth asymmetry on the front and back sides of the cell .
we find that the slope of the line connecting the tip and the hinge of the cell is a good measure for the integrated differential growth [ @xmath34 , see fig .
4 ] . with this ,
our theory ( si text ) predicts that for small deflections the angles before and after snapback should be proportional , with a slope depending on @xmath29 : large values of @xmath29 imply a small slope ( since the relative importance of plastic deformations is smaller for larger turgor pressure ) .
5 confirms this prediction .
it shows the snapback events of 33 _ e. coli _ and 33 _ b. subtilis _ cells performed under 8 different experimental conditions . as predicted , data for each species collapse onto a master line regardless of the growth medium , temperature , and stokes force .
the slope of the master line is approximately @xmath35 , consistent with a value of @xmath29 of order unity . that our prediction is verified by two species of bacteria as evolutionarily distinct as _ e. coli _ and _ b. subtilis
_ indicates that force transduction may have more general biological consequences for growth .
for example , recent studies indicate that mechanical force plays a major role in cell - shape deformation leading to tissue morphogenesis in higher organisms @xcite .
if mechanical stress regulates the rate of cell wall synthesis , it should play a central role in shape regulation . on a cylinder
the circumferential stress due to turgor pressure is twice as large as the stress along the cell s long axis .
this difference in stresses will provide a natural cue for orienting the growth machinery and coordinating glycan strand insertions . in gram - positive bacteria , the stress regulation will further lead to a constant thickness of the cell wall by a negative feedback mechanism : the mechanical stress is inversely proportional to the thickness of the cell wall , and thinner parts of the cell wall will have larger stresses due to turgor pressure , leading to enhanced local growth . of the tip before and after the snapback .
error bars are omitted from several series for clarity , but are on the order of those shown .
the snapback angle is approximately proportional to the initial angle of deflection when the angles are small , confirming the prediction of the dislocation - mediated growth theory ( si text ) .
the @xmath361/2 slope shows that the contributions by the two fundamental modes of deformation ( elastic vs. plastic ) are comparable with one another , regardless of the temperatures and growth conditions.[fig : snapback],width=8 ] note that , throughout the experiments , there were cell - to - cell variations in the bending rates , which was about @xmath37 % of the mean .
this suggests that the variability of elastic constants between different cells might be smaller than what was found in earlier studies @xcite .
we have developed a simple , novel experimental approach to directly probe the mechanical properties of bacterial cell walls .
our results indicate that the cell - wall growth rate depends on the mechanical stress , where faster growth is obtained for tensile ( stretching ) stress .
this mechanism is likely involved in shape regulation and the maintenance of a stable rod - shaped cell over time .
we found that bending forces cause the cell walls to grow differentially .
the differential growth leads to persisting plastic deformations , whose recovery requires cell wall synthesis .
this is in sharp contrast to the elastic deformations by pulse - like forces .
the elastic and plastic deformations represent two fundamental modes of cellular response to external mechanical forces , and can be decoupled experimentally as predicted by our theory .
these effects can be revealed only by applying bending forces , and can not be studied by means which can only change the uniform pressures that have to exist inside and outside the cell walls ( e.g. , osmotic shocks ) : since cell - wall growth is not the bottle - neck for the cell growth rate @xcite , the growth rate is not dictated by the cell - wall synthesis rate . in the future
, it would be illuminating to study deformation behavior in other types of bacteria such as tip - growers @xcite .
based on the mechanism described here , unlike _ e. coli _ or _ b. subtilis _
, we expect purely elastic deformations regardless of the duration of applied external forces . at the molecular level , direct observation of the asymmetric growth
would provide critical information for the mechanism of cell wall synthesis .
in particular , any changes in the speed of cell - wall synthesis rate or nucleation rate would be a direct test of our model .
this would be a challenging , but experimentally feasible based on the technology available to us @xcite .
finally , it would be fascinating to study similar mechanisms in higher organisms .
what is the role of forces in guiding the vast diversity of organism shapes in nature ?
previous works have shown the importance of mechanical forces in controlling the shapes of mammalian cells , where the properties of the extracellular matrix can have a large effect on the shape of the cells @xcite , their fate in differentiation @xcite , and tissue morphogenesis @xcite .
the bacterial model system studied here provides insights into the growth process itself , indicating that coupling of mechanical forces should be taken into account to understand robust morphological regulation of single - cell organisms .
+ to deform growing cells non - invasively by applying a well - defined force , we employed a microfluidic device , mother machine , " previously developed by some of us @xcite .
this device ensures balanced growth of individual cells for hundreds of generations .
furthermore , by controlling the flow speed in the device , the ambient buffer or the content of growth medium surrounding the cells can be switched within fractions of a second @xcite .
thus , we can tune the force applied to the growing cell since the force is directly proportional to the speed of the flow and the size of the cell @xcite .
the force which we achieve in this setup far exceeds that attainable with afm and optical tweezers .
also , our technique provides much more control of the applied forces in comparison , for example , with cells confined to microfluidic chambers of fixed or deformable shape . for the purpose of our experiments , it was beneficial to work with filamentous cells with cell division suppressed .
we grew the cells typically up to 50 @xmath1 m long .
this allowed us to precisely monitor the deformation of the cells in the presence of external mechanical forces .
for the _ e. coli _ experiments , we used strain mg1655 with sula under an inducible promoter ( gift from debu raychaudhuri , see a description of the plasmid pdb192 in ref .
@xcite ) and cytoplasmic yfp .
sula inhibits polymerization of ftsz , and upon induction with 1 @xmath38 concentration of iptg the cells became filamentous .
yfp is constitutively expressed in order to increase the signal - to - noise ratio of the images taken .
for the _ b. subtilis _
experiments , we used a ylao ( homolog of ftsw ) knockout with ylao reinserted under a d - xylose inducible promoter ( pxyl ) ( gift from michael elowitz ) .
this strain is typically grown in the presence of 0.5% xylose ; filamentation is induced by removing xylose from the growth media .
we used the hydrostatic pressure due to the controlled height difference between the inlet and outlet to control the flow through the device and achieve short ( subsecond ) response times of the flow velocity , as we verified using fluorescent beads . in the si text
we analyze the flow profile through the microfluidic device , both analytically and experimentally .
we calculate the resulting forces and torques on the cell due to the viscous flow , and provide details of the microscopy and image processing algorithms used .
we thank e. garner , s. t. lovett , s. wang , n. s. wingreen and a. wright for stimulating discussions .
we are also indebted to a. rotem and h. a. stone for advice regarding the microfluidics , and to e. efrati , j. w. hutchinson and j. paulose for useful discussions of the mechanical aspects and differential growth .
we thank s. taheri , c. nurnberg , e. hanna , a. kanagaki , y. caspi , d. raychaudhuri , j. bechhoefer and the nikon microscopy center for their help with the experimental part .
a.a . was supported by a junior fellowship of the harvard society of fellows and the milton fund .
work by d.r.n was supported by the national science foundation via grant dmr1005289 and through the harvard mrsec via grant dmr0820484 .
s.j . was supported by bauer fellowship ( harvard university ) , allen distinguished investigator award , pew award , and nsf career award .
to whom correspondence may be addressed : [email protected] ( a.a . ) or [email protected] ( s.j . ) .
10 campas , o , mahadevan , l , & joanny , j. f. ( 2012 ) actin network growth under load .
, 10491058 .
janmey , p. a & weitz , d. a. ( 2004 ) dealing with mechanics : mechanisms of force transduction in cells . , 364370 .
discher , d. e , janmey , p , & wang , y. l. ( 2005 ) tissue cells feel and respond to the stiffness of their substrate .
, 113943 .
minc , n , boudaoud , a , & chang , f. ( 2009 ) mechanical forces of fission yeast growth .
, 10961101 .
aharoni , h , abraham , y , elbaum , r , sharon , e , & kupferman , r. ( 2012 ) emergence of spontaneous twist and curvature in non - euclidean rods : application to _ erodium _ plant cells . , 238106 .
sun , s. x & jiang , h. ( 2011 ) physics of bacterial morphogenesis .
, 543565 .
lovering , a. l , safadi , s. s , & strynadka , n. c. j. ( 2012 ) structural perspective of peptidoglycan biosynthesis and assembly .
, 45178 .
scheffers , d .-
j & pinho , m. g. ( 2005 ) bacterial cell wall synthesis : new insights from localization studies bacterial cell wall synthesis : new insights from localization studies . .
young , k. d. ( 2010 ) bacterial shape : two - dimensional questions and possibilities .
, 223240 .
den blaauwen , t , de pedro , m. a , nguyen - distche , m , & ayala , j. a. ( 2008 ) morphogenesis of rod - shaped sacculi . , 32144 .
koch , a. l. ( 2001 ) _ bacterial growth and form_. ( springer , berlin ) .
takeuchi , s , diluzio , w. r , weibel , d. b , & whitesides , g. m. ( 2005 ) controlling the shape of filamentous cells of escherichia coli .
, 18191823 .
mnnik , j , driessen , r , galajda , p , keymer , j. e , & dekker , c. ( 2009 ) bacterial growth and motility in sub - micron constrictions .
, 1486114866 .
wang , s , arellano - santoyo , h , combs , p. a , & shaevitz , j. w. ( 2010 ) actin - like cytoskeleton filaments contribute to cell mechanics in bacteria .
, 91829185 .
yao , x , jericho , m , pink , d , & beveridge , t. ( 1999 ) thickness and elasticity of gram - negative murein sacculi measured by atomic force microscopy thickness and elasticity of gram - negative murein sacculi measured by atomic force microscopy . .
amir , a & nelson , d. r. ( 2012 ) dislocation - mediated growth of bacterial cell walls .
, 98339838 .
tuson , h. h , auer , g. k , renner , l. d , hasebe , m , tropini , c , salick , m , crone , w. c , gopinathan , a , huang , k. c , & weibel , d. b. ( 2012 ) measuring the stiffness of bacterial cells from growth rates in hydrogels of tunable elasticity . , 874891 . and references therein .
nelson , d. r. ( 2012 ) biophysical dynamics in disorderly environments . , 371402 .
garner , e. c , bernard , r , wang , w , zhuang , x , rudner , d. z , & mitchison , t. ( 2011 ) coupled , circumferential motions of the cell wall synthesis machinery and mreb filaments in b. subtilis .
, 222225 .
domnguez - escobar , j , chastanet , a , crevenna , a. h , fromion , v , wedlich - sldner , r , & carballido - lpez , r. ( 2011 ) processive movement of mreb - associated cell wall biosynthetic complexes in bacteria .
, 225228 .
van teeffelen , s , wang , s , furchtgott , l , huang , k. c , wingreen , n. s , shaevitz , j. w , & gitai , z. ( 2011 ) the bacterial actin mreb rotates , and rotation depends on cell - wall assembly . , 15822 .
hirth , j. p & lothe , j. ( 1982 ) _ theory of dislocations_. ( wiley , new york ) .
amir , a , paulose , j , & nelson , d. r. ( 2013 ) theory of interacting dislocations on cylinders .
, 042314 .
peach , m & koehler , j. s. ( 1950 ) the forces exerted on dislocations and the stress fields produced by them .
, 436439 .
wang , m. d , schnitzer , m. j , yin , h , landick , r , gelles , j , & block , s. m. ( 1998 ) force and velocity measured for single molecules of rna polymerase .
, 902907 .
liang , h & mahadevan , l. ( 2011 ) growth , geometry , and mechanics of a blooming lily .
, 55165521 .
savin , t , kurpios , n. a , shyer , a. e , florescu , p , liang , h , mahadevan , l , & tabin , c. j. ( 2011 ) on the growth and form of the gut .
, 5762 .
guillot , c & lecuit , t. ( 2013 ) mechanics of epithelial tissue homeostasis and morphogenesis . , 11851189 .
scott , m , gunderson , c. w , mateescu , e. m , zhang , z , & hwa , t. ( 2010 ) interdependence of cell growth and gene expression : origins and consequences .
, 10991102 .
brown , p. j , kysela , d. t , & brun , y. v. ( 2011 ) polarity and the diversity of growth mechanisms in bacteria .
, 790 798 .
jiang , x , takayama , s , qian , x , ostuni , e , wu , h , bowden , n , leduc , p , ingber , d. e , & whitesides , g. m. ( 2002 ) controlling mammalian cell spreading and cytoskeletal arrangement with conveniently fabricated continuous wavy features on poly(dimethylsiloxane ) . , 32733280 .
wang , p , robert , l , pelletier , j , dang , w. l , taddei , f , wright , a , & jun , s. ( 2010 ) robust growth of escherichia coli .
, 10991103 .
pelletier , j , halvorsen , k , ha , b .-
y , paparcone , r , sandler , s. j , woldringh , c. l , wong , w. p , & jun , s. ( 2012 ) physical manipulation of the escherichia coli chromosome reveals its soft nature .
, e26492656 .
berg , h. c. ( 1993 ) _ random walks in biology_. ( princeton university press , chichester ) .
de boer , p. a , crossley , r. e , & rothfield , l. i. ( 1990 ) central role for the escherichia coli minc gene product in two different cell division - inhibition systems .
, 112933 . * supporting information
the central element of the experimental setup consists of a microfluidic device ( the mother machine " ) , with dimensions that match well those of the bacteria ( width and height of approximately @xmath39 @xmath1 m , matching the bacteria s diameter ) , see fig . 1 of the main text . details of the fabrication of the devices are described elsewhere @xcite .
in contrast to the work of ref .
@xcite , where the robustness of the growth was studied and the bacteria were constantly dividing , here we suppress the cell division by regulating proteins involved in the division machinery ( sula for _ e. coli _ and ylao for _ b. subtilis _ ) , after inserting the cells into the microchannels . upon filamentation , a single bacterium will fill the @xmath40 @xmath1 m long channels , and upon further growth part of the bacterium will be exposed to the flow in the main chamber of the microfluidic device , see fig . 1 of the main text . in most experiments the flow was of lb broth , which was also used to supply the bacteria with nutrients , but we have also done the experiments using minimal and synthetic rich media , see fig . 5 of the main text .
next , we will characterize the flow and calculate the resulting forces on the bacteria .
throughout the experiment , the reynolds number is low enough ( @xmath41 ) such that the flow is laminar .
the bacteria reside very close to the surface , though , where the velocity vanishes , and for this reason it is important to accurately characterize the profile of the flow through the device . in the experimental setup , the pressure on the device is controlled . for a given pressure difference ,
the flow through the device can be analytically expressed as an infinite series @xcite .
1a shows the theoretical expectation for the flow through a cross - section of the main channel .
we mapped the flow with spinning - disk confocal microscopy , using fluorescent beads with a diamater of @xmath42 nm and an exposure time of 2 ms .
we extracted the velocity profile from the lengths of the tracks left by the beads .
a comparison of the measurement and the theoretical expectations , with no fitting parameters , is shown in fig .
1b . to estimate the drag force on a bacterium ( per unit length ) , we use the exact solution for the viscous drag force ( per unit length ) on a cylinder residing on a surface @xcite : f= 4 r , where @xmath18 is the bacterium s radius and @xmath43 is the gradient of the velocity profile close to the surface far away from the cell ( where the flow velocity is approximately linear in the height above the surface ) . in our setup
, the force is approximately @xmath44 pn/@xmath1 m . in this configuration
, the net lift on the cell due the flow can be shown to vanish , but there is a non - vanishing torque per unit length due to the flow @xcite : _ = 6r^2 .
this torque leads to a non - vanishing @xmath45 component of the stress - tensor .
this component , however , is not expected to couple to the cell wall growth and can therefore be neglected . moreover , the resulting shear stresses obey @xmath46 , and are significantly smaller than the @xmath14 stress which we shall later calculate .
similarly , @xmath47 due to the flow is also negligible .
we used a nikon ti microscope , with fluorescence light provided by the lumencore spectra x arc lamp and 100x objective .
we used the nikon perfect focus system to maintain the cells in focus during the lengths of each measurement , which typically lasted for 30 - 60 minutes for each field - of - view .
phase contrast images were taken every 1 second , and fluorescent images with a yfp filter were taken every 20 seconds , in order to minimize the effects of phototoxicity ( in figs .
2a and 3a of the main text we show that exponential growth with the physiological growth rate is achieved for these conditions : slightly less than 30 minutes in fig .
2a and @xmath48 minutes in fig .
3a ) . for this reason
exposure times were chosen as the minimal ones which provide good signal - to - noise ( snr ) ratios .
these were typically 100 ms for the phase contrast images and 20 - 40 ms for the florescent images .
we implemented an image processing algorithm to track the shape of the cell as a function of time , which was optimized for the purposes of our experiment .
the phase contrast images were used to initially find the location of the end of the microchannel ( not visible in the fluorescent images ) , manually . for the image processing ,
only the fluorescent channel images were used , due to their superior snr ( which was typically between 3 - 5 when comparing the center of the bacterium and the noise level ) .
the algorithm is described in the following .
starting at the end of the microfluidic channel , the position along the cell is followed , where at each running step we keep track of the direction of the local tangent to the cell .
initially , the tangent is pointing in the @xmath15 direction , as shown in point @xmath49 in fig . 2
. \2 . at every step of the algorithm ,
the _ first approximation _ to the next tracking point is chosen by taking the current pointer position and advancing it by a small increment ( 1/15 @xmath1 m , corresponding to one pixel ) in the direction of the current tangent .
the intensity profile is extracted in a line perpendicular to the tangent , see for example the line next to point @xmath50 in fig . 2 , and the inset showing the corresponding intensity profile .
the length of the perpendicular segment is chosen as @xmath39 @xmath1 m , to match the bacteria s diameter - we shall define it as the `` search range '' .
center - of - mass _ of the intensity in the search range is calculated , in order to minimize the fluctuations , _ which is chosen as the new pointer position .
the new tangent can now be chosen as the difference of the updated pointer and the pointer from previous runs , where in order to avoid large local fluctuations due to noise we choose to compare two pointers separated by 20 runs ( corresponding to approximately @xmath39 @xmath1 m ) .
\5 . to find the tip of the bacteria , we define the noise level as the median of the intensity over the entire field - of - view , and in the previous tracking algorithm we demand that the average intensity over the search range should be greater than a threshold of snr = 3 .
when the signal goes below the threshold , implying we have reached the end of the bacterium , we fit a 4th order polynomial to the extracted shape , and used the smooth polynomial in order to calculate the contour length of the bacterium .
the young modulus of peptidoglycan was estimated in refs .
@xcite,@xcite to be of the order of @xmath51 n / m@xmath4 .
the thickness of the cell wall , @xmath16 , is of the order of @xmath52 nm . in a similar fashion to the usual derivation of the deformation of an elastic rod , we can calculate the elastic deformation of a hollow rod of thickness @xmath53 , subject to a constant force per unit length perpendicular to its axis of symmetry .
this is equivalent to the problem of a rod being deformed under its own weight , with one fixed end and one free end , see ref .
@xcite p. 81 .
we outline the details of the calculation , since part of the reasoning will also be relevant to the discussion of the _ irreversible _ deformations in section [ plastic_sect ] .
a key ingredient is the ( geometric ) connection between the local radius - of - curvature and the local strains . as illustrated in fig . 3 , for a given radius - of - curvature @xmath54 at a given point @xmath15 along the cell
, the strain on a smaller region of the outer part of the cell wall will be given by : = 2r / r , [ curvature ] where @xmath55 is the arc traversed by the centerline .
here , we assume that the strains vanish for a straight cell , an assumption that will _ not _ hold when we will consider asymmetric insertions in section [ plastic_sect ] .
we also assumed that the strains are small , which does _ not _ imply that the deflections are small - for a cell with a large aspect ratio , one can have small strains throughout the cell leading to an overall large angular deflections . typical deflections @xmath56 calculated below are in the range @xmath57 , where @xmath17 is the protruding length of the bacterium . in the regime of linear
elasticity , the strains are proportional to the stresses .
the surrounding viscous flow creates primarily a @xmath14 component of the stress tensor within the cell wall ( see fig . 1 of the main text for the definition of the coordinate system ) , which can be found from the condition that the torques created by them must compensate for the torque created by the external forces .
we proceed to express the extra stress at a point with coordinate @xmath15 along the centerline of the cell s long axis and azimuthal angle @xmath19 , in terms of the flow parameters ( see fig . 6 for the definition of @xmath19 ) . using the linear relation of stress and strain in a hollow rod @xcite , we expect that for small deformations : _
( , y ) y r ( ) /r(y ) , [ stress_r]where @xmath54 is the local radius - of - curvature ( not to be confused with the radius of the bacteria , @xmath18 ) , and @xmath31 is the young modulus of the cell wall .
upon balancing the torque induced by this stress with the external torque at position @xmath15 , @xmath58 , we find that : \(y ) = , [ torque ] where @xmath59 is the ` moment - of - inertia ' of a cross - section , which for our case is found to be @xmath60 ( this result can also be obtained by differentiating the formula for a solid cylinder , @xmath61 ) .
note that the moment - of - inertia is a mathematical analogy with the theory of rigid bodies , a notation commonly used in the theory of rod elasticity .
@xmath62 is known as the flexural rigidity . for a constant force per unit length
@xmath22 we find that the stresses within the cell wall are : _ ( , y ) = , [ stress]this result should be multiplied by the thickness @xmath16 if one is interested in the two - dimensional stress .
indeed , one can check directly that the torques arising from these stresses at a cross - section at point @xmath15 are equal to the total torque due to the force @xmath22 integrated along the segment @xmath63 $ ] . upon combining eq .
( [ stress ] ) with eq .
( [ stress_r ] ) , the resulting curvature is given by : 1/r(y ) [ torques_elastic ] .
we note from differential geometry that the radius - of - curvature can be expressed in terms of the horizontal deflection of the bacterium @xmath56 , as : = .
[ full_elastic]upon neglecting the @xmath64 term in the denominator , valid for weak deflections , and focusing on the case of a constant force along the cell , we obtain a linear equation that can be readily integrated to give the resulting deflection of a point @xmath15 along the cylinder . under the assumption of a vanishing derivative at the end of the microchannel
, we find that : ( y ) = , [ z _ ] thus , the maximal deflection @xmath65 is given by : = .
[ elastic_bend ] thus , the horizontal deformation of the tip @xmath66 is approximately proportional to @xmath67 , where @xmath17 is the length of the cell exposed to the force .
based on the data in fig .
2b of the main text , we were able to calculate the stiffness of the _ e. coli _ cells , and extract a flexural rigidity of @xmath68 n @xmath3 m@xmath4 .
this result is similar to that obtained by very different methods such as afm @xcite and recent work in which the cells were transiently deformed by optical traps @xcite .
it should be pointed out that even though the perpendicular deflections are significant , the local strains and stresses are in fact small .
thus , it is justified to assume that we are in the regime of _ linear _ elasticity . due to the large aspect ratio of the filamentous bacteria ,
whose length @xmath17 is much larger than their radius @xmath18 , the deflections are enhanced by a factor of @xmath69 .
recently , bending of elastic fibers in a flow was studied , experimentally and theoretically , albeit in a different regime of parameters where the flow is highly confined and the fiber blocks most of it @xcite .
the previous analysis of the elastic deformations assumed a boundary condition with the cell fixed at the end of the microchannel , i.e. , that the tangent to the cell centerline would be parallel to the microchannel there .
however , looking at a typical cell ( see for example fig .
5 shows that there can be a finite derivative @xmath70 there , due to the small mismatch between the cell diameter and the channel diameter at the end of the microchannel , which we denote by @xmath71 .
we will find @xmath72 by analyzing the forces and torques , show that the assumption of a cell fixed at the microchannel end is an excellent one for the longer cells , and account for the deviations for the shorter cells .
the torque balance for the part of the cell outside the microchannel implies that @xmath73 is still given by eq .
( [ torques_elastic ] ) , and thus the shape of the cell is given by : \(y ) = + a y , [ z ] where @xmath72 is yet to be determined .
for the part of the cell inside the microchannel , external forces @xmath74 and @xmath75 are only applied at two points , shown in fig .
the derivative at the point where @xmath75 is applied vanishes ( since the cell is tangent to the microchannel there ) , so the position of the centerline of the cell is given by the usual solution of a beam bent by an external force @xcite : ( y ) = ( - ) , -y_0y 0[delta ] where @xmath76 corresponds to the end of the microchannel .
we therefore have : = y_0 ^ 3 . [ eq1_elastic ]
however , both @xmath75 and @xmath77 are unknown .
we fix these quantities by considering the torques around the end of the microchannel : f_2 y_0 = = _ 0^l f l dl=f l^2/2 .
[ eq2_elastic ] upon eliminating @xmath75 from eqs .
( [ eq1_elastic ] ) and ( [ eq2_elastic ] ) , we find that : y_0 ^ 2 = .
hence : y_0 = , f_2=y_0/=. since the derivative at the end of the microchannel is continuous , we have : a = = . for the case of a perfect match between the bacterium and the microchannel diameter ( @xmath78 ) ,
the maximal deflection was given by : _
= . the corrected maximal deflection will now be : _
[ correctdelta ] the correction is negligible for long enough cells , @xmath79 , but becomes significant for short cells .
it is also possible to take into account a _ non - uniform _ force per unit length outside the channel : in this case , the force per unit length @xmath22 and torque @xmath80 must be recalculated as a function of position on the cell .
the deflection can then be calculated by numerical integration .
2b in the main text uses this more general procedure when modeling the elastic deformations : \1 .
we use the full solution for the flow profile to calculate the force per unit length at every point along the cell , and the resulting torques at every point .
we solve the elastica problem for this torque distribution , and find the resulting deflection by integrating the equations numerically , including the non - linear terms due to the denominator of eq .
( [ full_elastic ] ) . for stiff , long cells using eq .
( [ correctdelta ] ) with a constant force per unit length @xmath81 pn/@xmath1 m would be an excellent approximation , since the flow would be approximately constant and the deformations of stiff cells would be still small such that the non - linear contributions are insignificant .
this is the case for the data shown in fig . 4 below .
transient bending by pulse - like forces provide simple , non - invasive means to directly measure the elastic properties of living bacterial cells .
4 shows the maximum deflection @xmath82 vs. the length ( @xmath17 ) of the cells exposed to the force . the solid red line is the fit using a full theoretical flow profile in fig . 1 , and we obtained the flexural rigidity @xmath83 .
this implies @xmath84 mpa for a cell radius @xmath85 @xmath1 m and cell - wall thickness @xmath86 nm , in good agreement with past measurements which found y = 10 50 mpa @xcite .
we also used eq .
[ correctdelta ] to fit both the average force per unit length , @xmath22 , and the flexural rigidity ( @xmath71 can be measured directly from the images ) .
we obtain a fit that is virtually identical to the full - flow result ( fig . 4 ) .
for @xmath87 @xmath1 m , we find @xmath88 pn/@xmath1 m and @xmath89 n@xmath3m@xmath4 . with the same assumptions from above we find @xmath90 mpa , consistent with the results using the full flow profile .
this suggests that our flow - based scheme of imparting force may provide a simple and robust measurement of the young s modulus of various bacteria .
the turgor pressure creates no torques on the cell , and for this reason need not be taken into account throughput the analysis . to show this , consider an integral over the closed red cross - sectional contour surface sketched in fig .
the force and torque due to these forces must vanish , since the material inside is in mechanical equilibrium . in the analysis of cell wall deformations , one has to find the torque exerted on the cell wall by external forces .
the contribution to this torque by the turgor pressure around point @xmath49 , for example , is exactly that associated with the above contour , except for the missing additional torque associated with the flat surface inside the cell ( denoted by the dashed line in the figure ) .
the forces there , however , give rise to a negligible torque , since they have no lever arm .
we conclude that the torque due to turgor pressure around any point along the cell would vanish .
the only effect of the pressure is thus to create large stresses @xmath47 and @xmath14 in the cell wall .
we will now estimate the rate of bending due to the asymmetric synthesis of cell wall as a response to an external stress ( due to the viscous flow ) .
this asymmetry gives rise to an additional bending mechanism , which is irreversible , since it involves remodeling of the peptidoglycan mesh .
we will assume that we are still in a regime where the bending is relatively small , so we can use the previous results for the flow around a straight cylinder , and for the forces on a slightly deformed rod .
for simplicity of the analysis , let us start by assuming that the young modulus is large enough such that elastic bending is negligible ; we shall see that even for @xmath91 the plastic bending is finite . in the next section
we will explain how the calculation can be extended to take both elastic and irreversible plastic deformations into account .
consider a cross - section of the bacterium , at a distance @xmath15 away from the end of the microchannel , as illustrated in fig .
the _ two - dimensional _ stress due to the turgor pressure @xmath30 is independent of @xmath19 , and given by @xmath92 .
( [ stress ] ) shows that the additional stress @xmath14 due to the viscous drag is positive for @xmath93 , corresponding to the stretched part of the rod , and negative otherwise .
the maximal relative change in the stress occurs at the points @xmath94 , at @xmath76 ( corresponding to the end of the microchannel ) and equals : / _ = . upon inserting the numbers for a relatively long filamentous bacterium of length 10 @xmath1 m
, we find a ratio of order unity .
thus , the flow can induce significant changes in the stress , for long cells .
we shall now invoke the formalism of dislocation - mediated growth developed in ref .
@xcite , which hinges on the fact that new material is inserted circumferentially via the rotating strand - extension centres .
the @xmath14 stress induces a peach - koehler force on an edge dislocation with a burgers vector in the @xmath15 direction , acting in a tangential direction ( in the @xmath95 plane ) , @xmath96 , as discussed in the main text .
this force acts in addition to the always positive stress @xmath97 due to turgor pressure .
tensile stresses enhance growth while compressive stresses inhibit it . under the assumption that the additional stresses are small compared to that due to turgor pressure
, we can taylor expand the dependence of local growth rate on stress , and find : g /g = f [ _ / ^0 _ ] , [ taylor ] where the coefficient @xmath98 multiplying the excess stress @xmath14 due to bending is expected to be of order unity . upon inserting the additional stress when a force @xmath22 per unit length is exerted on the cell , we find that : g /g , [ assymetry ] ( where we take @xmath99 ) .
this differential growth , accumulated over time , will turn out to give a non - negligible plastic deflection even when @xmath100 , due to the large aspect ratio of the filamentous bacterium .
the bending of the cell will be such that the additional strains corresponding to the bending will compensate those due to the differential growth : locally , the ratio of arcs @xmath101 in fig . 3 should correspond to the relative amount of material incorporated into the cell wall .
let us denote the _ integrated _ differential growth arising eq .
( [ assymetry ] ) by @xmath102 .
the radius - of - curvature @xmath103 resulting from the differential grows ( see fig .
3 ) obeys : l / dl ( dl_2-dl_1)/dl_1= 2r / r , [ growth_curvature ] leading to a curvature @xmath104 .
the angle @xmath105 resulting from this differential growth obeys @xmath106 , hence : d= l / dl .
indeed , we find that the relative growth asymmetry of @xmath107 is enhanced by the geometrical factor of @xmath108 . for a long cell with a constant differential growth
, this factor would correspond to half the aspect ratio .
note , however , that this argument gives the _ local _ curvature due to differential growth , which in our experiments is non - constant along the length of the cell . these results can be also obtained via differential geometry , by considering the gaussian curvature associated with the non - uniform metric due to differential growth .
we thus expect that as the bacterium grows to length @xmath17 , the angle associated with the plastic bending ( as long as it is small ) should scale as : ~ .
[ scaling_plastic]upon comparing this angle to that of the elastic deformations ( eq . [ elastic_bend ] ) , we find that both elastic and plastic effects have the same scaling with @xmath17 , allowing for a quantification of their relative importance in terms of a single dimensionless parameter : .
[ plastic_vs_elastic ] for large enough @xmath31 , the plastic effect dominates over the elastic one . with estimated values for _ e. coli _ , we find that @xmath29 is of order unity , implying that elastic and plastic effects are of the same order of magnitude .
thus , one has to solve the coupled problem , without separating the discussion into an elastic and plastic part , as we did so far . for bacteria with stiffer cell walls ,
we expect the plastic effect to dominate over the elastic one .
.summary of parameters for calculating @xmath29 for _ e. coli _ and _ b. subtilis_. @xmath109 refers to the present work . [
cols="<,^,^,^,^",options="header " , ] finally , we note that when going from eq .
( [ assymetry ] ) , giving the instantanous relative growth , to the integrated asymmetry @xmath102 , one should also take into account the effect of a finite processivity , which in the absence of differential growth would lead to an exponential decay of the curvature @xcite .
the accumulated _ relative _ differential growth @xmath110 obeys : f = -+ , [ integrate ] where @xmath111 is the growth rate ( i.e. , the doubling time equals @xmath112 ) , the first term corresponds to the dilution " effect due to the finite procreativity ( taken as infinite in the above equation , for simplicity ) , and the second term corresponds to eq .
( [ taylor ] ) .
the straightening described in the main text ( see also fig .
3 of the main text and supplementary movie s2 ) is in sharp contrast to the results of ref .
@xcite , where curved , arc - shaped cells taken out of microfluidic channels do not seem to straighten as they grow , but rather , maintain a self - similar shape with a slowly decreasing radius - of - curvature as they become larger .
in fact , the latter is precisely what is expected from circumferential insertions with long processivity @xcite : circumferential insertions would dilute " the assymtery due to differential growth and thus lower the curvature , yet since the cell grows longer the shape can be shown to be self - similar . taking into account a finite processivity of insertions would make the decay of the curvature slower @xcite . therefore ,
if processive insertions alone are responsible for straightening , during continued growth the tip should move towards the bending direction ( @xmath113 ) .
in contrast , the cells after the partial snap back , however , actually move towards ( @xmath114 ) the channel the bent cell is embedded in ( fig .
3b of the main text ) .
note that due to the turgor pressure , the stress on the cell wall during the straightening process is non - uniform , and one should also take the coupling of the stress distribution to the growth . however , when a cell is deformed , the stress component along the bacterium s long axis ( @xmath14 ) remains approximately constant , while it is the circumferential component of the stress tensor that is modified , in a manner proportional to the curvature of the bent cell @xcite .
there are no indications that this stress component is coupled to the growth process .
we therefore conclude from our experiments that another mechanism is at play , which makes the cell straightening significantly faster in the absence of mechanical forces .
one possibility is a curvature - sensing mechanism that has been discussed in the context of protein localization in bacteria @xcite .
a second mechanism is the appearance of residual stresses due to the differential growth in the first stage of the experiment : the combination of turgor pressure and the differential growth can lead to residual stresses in the cell wall , which remain even after the flow is turned off , and are in general non - uniform .
we leave the study of this intriguing phenomenon to future work .
in the main text we demonstrate that plastic deformation of growing cells results from the application of flow - generated hydrodynamic force over extended periods of time .
we further demonstrate that , in the time after the flow is stopped , the cells straighten as they grow .
we attribute this behavior to a dislocation - mediated growth mechanism in which anisotropic stresses give rise to varying cell - wall synthesis rates which leads to plastic deformation .
a remaining possibility , however , is that this plastic deformation is disconnected from growth and the cell wall passively reorganizes ( deforms ) in response to the anisotropic stresses . to eliminate this mechanism we ran two separate controls .
first , we show that non - growing cells do not plastically deform after extended application of flow ( fig .
second , we demonstrate that cells grown in ( and plastically deformed by ) flow do not straighten over time ( fig
. 9b ) . taken together and in contrast to fig . 3 of the main text ,
these experiments indicate that the plastic deformation quantified in this paper is connected to growth of the cell , consistent with a dislocation - mediated growth mechanism .
we have shown above that filamentous bacteria will have significant elastic as well as irreversible bending at a length @xmath115 @xmath1 m . in this section
we explain how both of the effects can be quantitatively combined to calculate the shape of the cell at a given point . for the elastic effect ,
the radius - of - curvature at a given point is proportional to the strain at that point . for the irreversible effect ,
it is proportional to the non - uniformity in insertion rates on the two sides of the bacterium ( relative to the flow direction ) .
since both effects turn out to be of the same order - of - magnitude for the experimentally relevant case , we have to consider both contributions simultaneously . upon repeating the previous arguments
, we find that : in eq .
( [ taylor ] ) .
the elastic effect is maximal just outside the microfluidic channel , where the stresses are maximal , and vanishes at the stress - free tip of the bacterium . the plastic effect , on the other hand , relies on accumulation of differential growth over time , and therefore vanishes both at the stress - free tip and at the end of the microfluidic channel , which contains `` fresh '' cell wall with no asymmetry ,
having just come out of the microfluidic channel.,width=8 ] we can now quantitatively analyze the experimental scenario : at every point in time we calculate the resulting torque @xmath80 due to the flow .
this torque gives rise to an elastic curvature ( as quantified in eq .
( [ c_e ] ) ) , which is an instantaneous effect .
it will also give rise to a non - uniform insertion rate , which , _ integrated over time _
( [ integrate ] ) ) , will give rise to the irreversible component of the curvature ( as quantified in eq .
( [ c_i ] ) ) .
note that a given point on the cell elongates exponentially in time , and that when calculating the accumulated asymmetry @xmath116 at a given point we have to follow the path of the point over time , and integrate the contribution of the stresses at different points in space and time , associated with the history " of that part of the cell wall .
for example , the part of the bacterium which is close to the end of the side channels had no stress on it while it was in the channel , and for this reason accumulated no asymmetry : it will have no radius - of - curvature associated with the irreversible effect .
it will have , however , the _ maximal _ radius - of - curvature associated with the elastic contribution , since the torque at that point is maximal .
8 shows the relative contributions of both effects , for realistic experimental parameters .
yao , x , jericho , m , pink , d , & beveridge , t. ( 1999 ) thickness and elasticity of gram - negative murein sacculi measured by atomic force microscopy thickness and elasticity of gram - negative murein sacculi measured by atomic force microscopy . .
maass , s , sievers , s , zhlke , d , kuzinski , j , sappa , p. k , muntel , j , hessling , b , bernhardt , j , sietmann , r , vlker , u , hecker , m , & becher , d. ( 2011 ) efficient , global - scale quantification of absolute protein amounts by integration of targeted mass spectrometry and two - dimensional gel - based proteomics .
, 26772684 .
sliusarenko , o , cabeen , m. t , wolgemuth , c. w , jacobs - wagner , c , & emonet , t. ( 2010 ) processivity of peptidoglycan synthesis provides a built - in mechanism for the robustness of straight - rod cell morphology . | cell walls define a cell shape in bacteria . they are rigid to resist large internal pressures , but remarkably plastic to adapt to a wide range of external forces and geometric constraints .
currently , it is unknown how bacteria maintain their shape . in this work ,
we develop experimental and theoretical approaches and show that mechanical stresses regulate bacterial cell - wall growth . by applying a precisely controllable hydrodynamic force to growing rod - shaped _
escherichia coli _ and _ bacillus subtilis _ cells , we demonstrate that the cells can exhibit two fundamentally different modes of deformation .
the cells behave like elastic rods when subjected to transient forces , but deform plastically when significant cell wall synthesis occurs while the force is applied .
the deformed cells always recover their shape .
the experimental results are in quantitative agreement with the predictions of the theory of dislocation - mediated growth .
in particular , we find that a single dimensionless parameter , which depends on a combination of independently measured physical properties of the cell , can describe the cell s responses under various experimental conditions .
these findings provide insight into how living cells robustly maintain their shape under varying physical environments .
biological systems exhibit many properties rarely found in condensed matter physics , which are often caused by _ growth_. when coupled to mechanical forces , growth can drive a wide range of cellular phenomena such as regulation of the eukaryotic cell morphology by actin networks @xcite , collective behavior in tissues @xcite , cell differentiation @xcite , and shape and division of yeast and plant cells @xcite . of fundamental interest as well as practical importance
is understanding the relationship between growth , form , and structure of bacterial cell walls @xcite .
bacterial cell walls define a cell morphology and hold the large internal ( turgor ) pressure .
many antibiotics target them to efficiently hamper cell growth and reproduction . as
such , cell walls and their synthesis have been the subject of extensive biochemical @xcite and biophysical @xcite studies in the context of cell growth @xcite , cell shape @xcite and cell division @xcite . despite the long history @xcite
, however , we are still far from being able to predict the shape or dimensions of any cells from first principles based on the information obtained from studies so far .
recent experimental work shed new insights in this regard .
for example , bacteria can significantly deform when grown with constraints @xcite and yet are able to recover their native shape @xcite .
however , the mechanism underlying deformation and recovery , as well as the cues which regulate cell - wall growth , have not been well understood .
we have developed combined experimental and theoretical methods to directly address how mechanical stresses are involved in regulation of cell wall growth .
our experimental approach is illustrated in fig .
1 .
rod - shaped _
eschericia coli _ or _ bacillus subtilis _ cells are inserted into snuggly fitting micro - channels where they are grown in a controlled environment .
filamention is induced in the cells by suppressing division , see the materials and methods section for further details .
the filamentous cell is subjected to precisely calibrated hydrodynamic forces .
this simple approach has two notable advantages to previous methods : ( 1 ) we can directly probe the mechanical properties and responses of the cell walls _ non - invasively _ for a wide temporal range ( from @xmath0 1 second to over one hour ) ( 2 ) we can achieve force scales several orders of magnitude larger than what is possible by optical traps @xcite or atomic force microscopy @xcite . using this approach ,
we demonstrate that the cells can exhibit two fundamentally different modes of deformations , and recover from them .
the first mode is _ elastic _ in that the cell recovers its shape as soon as the external force is removed .
the other mode is _ plastic _ and requires growth for morphological recovery .
the plastic deformations are due to _ differential growth _ , resulting from a coupling to mechanical stresses varying along the cell walls .
we also provide a theoretical framework that explains our experimental findings quantitatively .
the basic idea is that external forces are transduced to the activity of the cell - wall growth machinery . to explain shape deformations , we calculated the effect of force - transduction by extending the theory of dislocation - mediated growth of bacterial cell walls @xcite .
we found that , for pulse - like forces that are transient , the cell walls respond elastically in agreement with @xcite .
in contrast , if the duration of applied forces is comparable or longer than the timescale of cell - wall growth , the force - transduction causes differential growth of the cell walls and plastic deformations as seen in @xcite .
@xmath1 m .
the cross section of the micro - channels is square ( approximately 1.2 @xmath1 m x 1.2 @xmath1 m . ) and the cells fit snuggly .
cell division can be blocked by expressing sula proteins that inhibit formation of the ( ftsz ) constriction ring , so that we can induce large deformations and recovery of the cell for a long period .
the cells were grown typically up to 50 @xmath1 m during measurements .
controlled hydrodynamic forces are exerted on the cells by the viscous drag of the growth media ( si text ) .
[ fig : device],width=8 ] |
the thermodynamics of self - gravitating systems is a fascinating subject .
it started with antonov ( 1962 ) s discovery that , when a self - gravitating system is confined within a box of radius @xmath2 , no maximum entropy state can exist below a certain critical energy @xmath3 .
this intriguing result was further discussed by lynden - bell & wood ( 1968 ) who conjectured that for @xmath4 the system would collapse and overheat .
this is called `` gravothermal catastrophe '' or `` antonov instability '' .
lynden - bell & wood have related this phenomenon to the very particular property of self - gravitating systems to possess _ negative _ specific heats .
the gravothermal catastrophe picture has been confirmed by sophisticated numerical simulations ( larson 1970 , cohn 1980 , lynden - bell & eggleton 1980 ) and is expected to play a crucial role in the evolution of globular clusters .
it is found that the collapse proceeds self - similarly ( with power law behaviors ) and that the central density becomes infinite in a finite time .
this instability has been known as `` core collapse '' and many globular clusters have probably experienced core collapse ( binney & tremaine 1987 ) . in the case of dense clusters of compact stars ( neutron stars or stellar mass black holes ) ,
the gravothermal catastrophe can lead to the formation of supermassive black holes of the right size to explain quasars and agns ( shapiro & teukolsky 1995 ) .
statistical mechanics is also relevant for collisionless stellar systems ( e.g. , elliptical galaxies , dark matter , ... ) undergoing a violent relaxation ( lynden - bell 1967 , chavanis _ et al .
_ 1996 , chavanis & sommeria 1998 , chavanis 1998a,2001a ) .
in particular , the inner regions of elliptical galaxies are close to isothermal and this is an important ingredient to understand de vaucouleurs @xmath5 law ( hjorth & madsen 1993 ) . on a theoretical point of view , the stability of isothermal spheres has been first investigated by katz ( 1978 ) with a very powerful method extending poincar s theory of linear series of equilibrium .
he found that instability sets in precisely at the point of minimum energy .
this stability analysis was reconsidered by padmanabhan ( 1989 ) who studied the sign of the second variations of entropy and reduced the problem of stability to an eigenvalue equation .
this leads to the same stability limit as katz but the method of padmanabhan provides in addition the form of the perturbation that induces instability at the critical point .
it is found that this perturbation presents a `` core - halo '' structure .
the analysis of padmanabhan ( 1989 ) was performed in the microcanonical ensemble in which the energy is fixed .
the microcanonical ensemble is probably the most relevant for studying stellar systems like elliptical galaxies or globular clusters ( binney & tremaine 1987 ) .
indeed , apart from a slow evaporation , these systems can be assumed isolated so the evolution conserves energy @xmath6 and mass @xmath7 .
in addition , from the viewpoint of statistical mechanics , only the microcanonical ensemble is rigorously justified for non extensive systems , as discussed in the review of padmanabhan ( 1990 ) . however , it is always possible to define formally a canonical ensemble or a grand canonical ensemble in which the temperature is fixed instead of the energy . as suggested by de vega _
( 1996a , 1996b ) these ensembles may be suitable for describing the cold interstellar medium where the temperature is imposed by the cosmic background radiation at @xmath8 in the outer parts of galaxies , devoid of any star and heating sources ( pfenninger _ et al .
_ 1994 , pfenninger & combes 1994 ) .
in particular , by working out the statistical mechanics of the self - gravitating gas , de vega _ et al . _
( 1996a , 1996b ) have shown that self - gravity can provide a dynamical mechanism to produce the fractal structure of the interstellar medium .
they used the same approach to explain the fractal structure of the universe ( de vega _ et al .
_ 1998 ) , assuming that galaxies have reached a quasi - thermodynamical equilibrium like in the early work of saslaw & hamilton ( 1984 ) . for self - gravitating systems
, it is well known that the thermodynamical ensembles do not coincide in the whole range of parameters ( padmanabhan 1990 ) . using toy models ,
lynden - bell & lynden - bell ( 1977 ) and padmanabhan ( 1990 ) demonstrated that the region of negative specific heats allowed in the microcanonical ensemble is replaced by a phase transition in the canonical ensemble .
this phase transition separates a dilute `` gaseous '' phase from a dense `` collapsed '' phase .
since these toy models are not very realistic , the self - gravitating gas was also studied in a meanfield approximation . in this viewpoint ,
an isothermal sphere is stable if and only if it is a local maximum of an appropriate thermodynamical potential ( the entropy in the microcanonical ensemble and the free energy in the canonical ensemble ) . as expected from physical grounds ,
phase transition occurs when the gaseous sphere ceases to be a local maximum of this potential and becomes a saddle point . in this paper
, we investigate the stability of isothermal gaseous spheres in the canonical ensemble by studying the sign of the second variations of the free energy .
our analysis is a direct extension of padmanabhan ( 1989 ) s approach in the microcanonical ensemble .
the two studies therefore provide a unified description of the stability of isothermal spheres in the meanfield approximation in terms of thermodynamical potentials . for a long time , the thermodynamics of self - gravitating systems was considered exclusively in a meanfield approach ( or with toy models ) .
however , recently , de vega & sanchez ( 2001a , b ) have developed a rigorous statistical mechanics of self - gravitating systems by using field theoretical methods . in particular , they evidenced the existence of a thermodynamic limit in which the number of particles @xmath9 and the volume @xmath10 go to infinity keeping @xmath11 fixed .
this very unusual thermodynamic limit proves to be appropriate to non extensive systems . by using monte carlo simulations and analytical calculations
, they showed that the meanfield approximation correctly describes the thermodynamic limit except near the critical points where a phase transition occurs .
they also _ derived _ the equation of state of a self - gravitating gas instead of assuming it , as is done in the meanfield treatments .
therefore , their work fully justifies the studies of previous authors and specifies their range of validity .
this paper is organized as follows . in section [ sec_thermo ]
, we introduce a meanfield description of the system in the canonical ensemble and show that critical points of free energy @xmath12 at fixed temperature @xmath13 and mass @xmath7 correspond to isothermal spheres like those studied in the context of stellar structure ( chandraskhar 1942 ) .
we show that there is no global maximum of free energy .
there is not even a local maximum of free energy in an unbounded domain : unbounded isothermal spheres have an infinite mass !
we restrict therefore our analysis to the case of self - gravitating systems confined within a spherical box of radius @xmath2 . in this case
, there exists _ local _ maxima of @xmath12 ( metastable states ) if the normalized temperature @xmath14 is less than @xmath15 and the density contrast @xmath16 .
critical points of free energy with density contrast @xmath17 are unstable saddle points . for @xmath18 , there are not even critical points of free energy : in that case , the system is expected to undergo a phase transition and collapse .
this `` isothermal collapse '' is the counterpart of the `` gravothermal catastrophe '' in the microcanonical ensemble ( see figs .
[ gravotherme]-[isocollapse ] ) . in section [ sec_pad ]
, we study the sign of the second variations of free energy by using the methods of padmanabhan ( 1989 ) introduced in the microcanonical ensemble .
we show analytically that instability sets in precisely at the point of minimum temperature in agreement with the theorem of katz ( 1978 ) .
the perturbation that induces instability at this point is calculated explicitly ; it has _ not _ a `` core - halo '' structure contrary to what happens in the microcanonical ensemble . in section [ sec_dyn ] , we study jeans type gravitational instability of isothermal gaseous spheres described by navier - stokes equations .
the introduction of a container removes the problems associated with an infinite homogeneous medium and avoids the jeans swindle .
we show analytically the equivalence between dynamical stability and thermodynamical stability and the fact that the stability of isothermal gas spheres does not depend on the viscosity .
this confirms the findings of semelin _
( 2001 ) who used numerical methods or approximations .
we also give a simpler derivation of the geometric hierarchy of scales inducing instability discovered by these authors using sophisticated renormalization group technics ( semelin _ et al .
this provides a more illuminating interpretation of their results . in section [ sec_frag ] ,
we make speculations about the fragmentation and the fractal structure of an isothermal self - gravitating gas .
we distinguish between the jeans length @xmath19 defined with the mean density and the king s length @xmath20 defined with the central density .
if we fix the jeans length , instability occurs for @xmath21 and is marked by the _ absence _ of critical point of free energy ( i.e. , hydrostatic equilibrium ) above this threshhold . in that case
the system is expected to collapse without fragmenting .
if we fix the king s length ( or core size ) , a first instability occurs for @xmath22 . above this threshold
critical points of free energy _ still _ exist but they are unstable saddle points .
secondary instabilities occur at larger box radii that asymptotically follow a geometric progression @xmath23^{n } l_{k}$ ] .
the density profiles that trigger these high order modes of instability are calculated analytically .
they present more and more oscillations whose zeros also follow a geometric progression @xmath24^{n}r$ ] .
the profile that destabilizes the singular isothermal sphere has an infinite number of nodes !
each oscillation can be interpreted as a ` germ ' in the langage of phase transition and the above picture suggests that the system will fragment into a series of ` clumps ' .
it is expected that these ` clumps ' will evolve by achieving higher and higher density contrasts , and finally fragment in turn into substructures .
this yields a hierarchy of structures fitting one into each other in a self - similar way .
this picture is given further support by the fact that _ both _ the domain sizes inducing instability and the zeros of the perturbation profile in each domain follow a geometric progression with the same ratio .
this double - geometric progression may explain in a natural way the fractal structure of a self - gravitating gas like the interstellar medium and the large scale structures of the universe .
this gives further support to the interpretation of de vega _ et al . _
( 1996a , 1996b , 1998 ) who first realized the importance played by self - gravity in building a fractal distribution of matter .
consider a system of @xmath9 particles , each of mass @xmath25 , interacting via newtonian gravity .
the particles can be galaxies , stars , atoms , etc ... we assume that the system is non rotating and non expanding . let @xmath26 denote the distribution function of the system , i.e. @xmath27 gives the mass of particles whose position and velocity are in the cell @xmath28 at time @xmath29 .
the integral of @xmath30 over the velocity determines the spatial density @xmath31 on the other hand , in the meanfield approximation , the total mass and the total energy of the system can be expressed as @xmath32 @xmath33 where @xmath34 is the kinetic energy and @xmath35 the potential energy .
the gravitational potential @xmath36 is related to the star density by the newton - poisson equation @xmath37 finally , the boltzmann entropy is given by the standard formula @xmath38 which can be obtained by counting the number of microstates corresponding to a given macrostate and taking the logarithm of this number ( ogorodnikov 1965 ) .
we shall work in the canonical ensemble in which the temperature is fixed , allowing the energy to fluctuate . in that case ,
the relevant thermodynamical potential is the massieu function related to the helmholtz free energy @xmath39 by @xmath40 .
hence @xmath41 at equilibrium , the system is expected to be in a state that maximizes the massieu function ( [ j ] ) for a fixed total mass @xmath7 ( in the following we shall call @xmath12 the free energy , although this is only the free energy up to a negative proportionality factor ) .
following padmanabhan ( 1989 ) s procedure , we start to maximize the free energy @xmath12 for a given density field @xmath42 .
this yields the maxwell - boltzmann distribution @xmath43 which is a _
global _ maximum of @xmath12 with the previous constraint . substituting this optimal distribution function in eqs .
( [ e])-([bentropy ] ) , we can express the energy and the entropy in terms of the spatial density in the form @xmath44 @xmath45 we can now determine the variations of @xmath12 around a given density profile @xmath42 . to second order in the expansion
we get @xmath46 introducing a lagrange multiplier @xmath47 to satisfy the conservation of mass , the condition that @xmath12 is an extremum is written ( to first order ) @xmath48 this condition must be satisfied for any variations @xmath49 .
this yields the boltzmann distribution @xmath50 where we have set @xmath51 therefore , the boltzmann distribution ( [ rhob ] ) is a _
critical _ point of free energy .
this does not insure , however , that it is a _
maximum _ of @xmath12 .
it is not even clear that the extremum problem leading to eq .
( [ rhob ] ) has a solution .
indeed , the gravitational potential that appears in eq .
( [ rhob ] ) must be determined self - consistently by solving the mean field equation @xmath52 obtained by substituting the density ( [ rhob ] ) in the poisson equation ( [ poisson ] ) , and relating the lagrange multiplier @xmath53 to the constraint @xmath7 .
as we shall see , this problem does not always have a solution .
when a solution exists , we must consider the sign of the second order variations @xmath54 to determine whether it is a maximum or not .
it is easy to show that there is no global maximum of free energy . to prove this result
, we just need to consider a homogeneous sphere of mass @xmath7 and radius @xmath2 .
the total energy of this sphere is @xmath55 and its free energy @xmath56 from the above formula it is clear that @xmath12 goes to @xmath57 when we spread the density to infinity ( @xmath58 ) or when we contract the system to a point ( @xmath59 ) .
since the mass is conserved in this process no global maximum of free energy can exist .
in addition , there is not even a local maximum of free energy in an unbounded domain : unbounded isothermal spheres have an infinite mass .
for non rotating systems , the equilibrium states are expected to be spherically symmetric . in that case , the boltzmann - poisson equation ( [ mf ] ) can be rewritten @xmath60 this equation has been studied extensively in the context of isothermal gaseous spheres ( chandrasekhar 1942 ) .
we can check by direct substitution that the distribution @xmath61 is an exact solution of eq .
( [ mfr ] ) known as the _ singular isothermal sphere _ ( binney & tremaine 1987 ) . since @xmath62 at large distances , the total mass of the system @xmath63 is infinite !
more generally , we can show that any solution of the meanfield equation ( [ mfr ] ) behaves like the singular sphere as @xmath64 and has therefore an infinite mass ( chandrasekhar 1942 ) .
we shall avoid the infinite mass problem by confining artificially the system within a spherical box of radius @xmath2 .
it is only under this simplifying assumption that a rigorous thermodynamics of self - gravitating systems can be carried out .
physically , the maximum cut - off determines the scale at which other processes intervene to limitate the spatial extent of the system .
it is clear from eq .
( [ jmax ] ) that the introduction of an upper cut - off does not remove the absence of global maximum of free energy ( the system can always increase @xmath12 by collapsing to a point ) .
however , the presence of a wall at radius @xmath2 allows the existence of _ local _ maxima of free energy ( metastable states ) .
the microcanonical and canonical ensembles yield the same critical points , i.e. the critical points of entropy at fixed mass and energy and the critical points of free energy at fixed mass and temperature coincide .
only the onset of instability will differ from one ensemble to the other .
the thermodynamical parameters for bounded isothermal spheres in the meanfield approximation have been calculated by lynden - bell & wood ( 1968 ) and we shall directly use their results . to that purpose , we introduce the function @xmath65 where @xmath66 is the gravitational potential at @xmath67 .
then , the density field can be written @xmath68 where @xmath69 is the central density . introducing the notation @xmath70 ,
the boltzmann - poisson equation ( [ mfr ] ) reduces to the standard emden form ( chandrasekhar 1942 ) @xmath71 eq .
( [ emden ] ) has a simple analytic solution , the singular sphere @xmath72 the regular solutions of eq .
( [ emden ] ) must satisfy the boundary conditions @xmath73 when @xmath74 .
these solutions must be computed numerically .
however , their asymptotic behaviors are well - known @xmath75 @xmath76 the curve ( [ ab2 ] ) intersects the singular solution ( [ sing2 ] ) infinitely often at points that asymptotically increase geometrically in the ratio @xmath77 .
this property will have important physical consequences in section [ sec_frag ] . in the case of bounded isothermal systems , the solutions of eq .
( [ emden ] ) are terminated by the box at different radii given by @xmath78 .
lynden - bell & wood ( 1968 ) show that the reduced temperature and reduced energy can be expressed in terms of @xmath47 by @xmath79 @xmath80 for each value of the normalized inverse temperature @xmath81 , we can solve eq .
( [ eta ] ) to get @xmath47 . substituting the result in eq .
( [ lambda ] ) , we deduce the corresponding value of the normalized energy @xmath82 .
we can then determine the equilibrium diagram @xmath83 ( figs .
[ gravotherme]-[isocollapse ] ) .
the critical points of entropy or free energy form a spiral .
this curve is parametrized by @xmath47 that goes from @xmath84 to @xmath57 as we spiral inward . instead of the parameter @xmath47
, it may be more relevant to introduce the density contrast @xmath85 which gives a more physical parametrization of the solutions . for @xmath86
, we can use the asymptotic behavior @xmath87 of the potential at the origin [ eq .
( [ ab1 ] ) ] and we find @xmath88 , @xmath89 and @xmath90 .
this corresponds to high temperatures @xmath91 . in this case , self - gravity is negligible with respect to thermal motion and the system behaves like an ordinary gas with uniform density .
we check , by eliminating @xmath47 from the two foregoing relations , that @xmath92 .
this is indeed the equation of state for an ideal gas without self - gravity .
when we proceed along the spiral , the density contrast increases monotonically and goes to @xmath57 at the center of the spiral . using the asymptotic expansion @xmath93 of the potential at large distances [ eq .
( [ ab2 ] ) ] , we find @xmath94 , @xmath95 and @xmath96
. we can check by a direct calculation that the ending point @xmath97 of the spiral corresponds to the singular sphere ( [ singular ] )
. it will be convenient in the following to introduce the milne variables @xmath98 defined by ( chandrasekhar 1942 ) : @xmath99 taking the logarithmic derivative of @xmath100 and @xmath101 with respect to @xmath102 and using eq .
( [ emden ] ) , we get @xmath103 @xmath104 taking the ratio of the foregoing equations , we find that the variables @xmath100 and @xmath101 are related to each other by a first order differential equation @xmath105 therefore , by using the milne variables , the degree of the meanfield equation ( [ emden ] ) has been reduced from two to one . as discussed extensively by chandrasekhar ( 1942 ) , this property is related to the homology invariance of the solutions of the emden equation .
the solution curve in the @xmath98 plane is well known and is ploted in fig .
[ uvcrit ] .
the curve is parametrized by @xmath102 .
starting from the point @xmath106 corresponding to @xmath74 , the solution curve spirals indefinitely around the point @xmath107 , corresponding to the singular sphere , as @xmath102 tends to infinity .
all isothermal spheres must necessarily lie on this curve . for
bounded isothermal spheres , @xmath102 must be terminated at the box radius @xmath47 . clearly , the spiral behavior of the @xmath98 curve ( and also the curve of figs . [ gravotherme]-[isocollapse ] ) can be ascribed to the oscillating behavior of the solution ( [ ab2 ] ) as @xmath108 .
it turns out that the normalized temperature and energy can be expressed very simply in terms of the values of @xmath100 and @xmath101 at the normalized box radius @xmath47 . indeed , writing @xmath109 and @xmath110 and using eqs .
( [ eta])-([lambda ] ) , we get @xmath111 @xmath112 the relation ( [ uvlambda ] ) was previously noticed by padmanabhan ( 1989 ) .
the intersection between the lines defined by eqs .
( [ uvlambda])-([uvbeta ] ) and the spiral in the @xmath98 plane determines the value of @xmath47 corresponding to a given energy or temperature . as noticed by padmanabhan , for @xmath113 there is no intersection .
thus , in the microcanonical ensemble , no isothermal sphere can exist if @xmath114 .
this result can also be read from fig .
[ gravotherme ] and was discovered by antonov ( 1962 ) .
for sufficiently small energies , the system is expected to collapse and overheat in agreement with the `` gravothermal catastrophe '' picture ( lynden - bell & wood 1968 ) . similarly , considering eq .
( [ uvbeta ] ) , we find that there is no intersection for @xmath115 ( where @xmath116 corresponds to @xmath117 ) .
this result can also be read from fig .
[ isocollapse ] . in the canonical ensemble
, a gaseous sphere is expected to collapse below a critical temperature @xmath118 ( lynden - bell & wood 1968 ) .
more precisely , it is possible to prove the following stability result : when a self - gravitating system is confined within a box and maintained at a constant temperature @xmath13 , local maxima of free energy exist only for @xmath119 ; they have a density contrast @xmath120 ( and @xmath121 ) .
critical points of free energy with density contrast @xmath122 ( and @xmath123 ) are unstable saddle points . for @xmath124 , there are no critical points of free energy . the stability of the solutions can be deduced from the topology of the @xmath83 curve by using the method of katz ( 1978 ) who has extended poincar s theory of linear series of equilibrium .
the parameter conjugate to the free energy with respect to the inverse temperature @xmath125 is @xmath126 .
then , if we plot @xmath6 as a function of @xmath125 ( we just need to rotate fig .
[ isocollapse ] by @xmath127 ) we have the following results : ( i ) a change of stability can occur only at a limit point where @xmath125 is an extremum ( @xmath128 infinite ) ( ii ) a mode of stability is lost when the curve rotates clockwise and gained otherwise .
now , we know that for @xmath13 sufficiently large the solutions are stable because , in this limit , self - gravity is negligible and the system behaves like an ordinary gas . from point ( i )
we conclude that the solutions with @xmath120 are stable . as the curve spirals inward for @xmath122 , more and more modes of stability
are lost . in this respect , the singular sphere , at the end of the spiral , is the most unstable solution .
it can be noted that instability sets in precisely when the specific heat @xmath129 becomes negative ( in the range @xmath130 ) .
general considerations indicate that negative specific heats are forbidden in the canonical ensemble ( padmanabhan 1990 ) . by contrast
, thermally isolated systems can have negative specific heats : in the microcanonical ensemble , the solutions with @xmath130 are stable ( katz 1978 , padmanabhan 1989 ) .
this clearly demonstrates that the statistical ensembles do not coincide for self - gravitating systems : the region of negative specific heats in the microcanonical ensemble is replaced by a phase transition ( an `` isothermal collapse '' ) in the canonical ensemble ( padmanabhan 1990 ) .
it is at the verge of this phase transition that the meanfield approximation ceases to be valid as demonstrated by de vega _ et al . _
( 2001a , 2001b ) . in this section ,
we study some analogies with the theory of critical phenomena .
further analogies are discussed in chavanis _
( 2001 ) where a simple dynamical model for studying phase transitions in self - gravitating systems is proposed .
the specific heat can be written @xmath131 using eqs .
( [ uvlambda])([uvbeta])([uv1])([uv2 ] ) , we easily get @xmath132 @xmath133 therefore , the specific heat is determined as a function of @xmath47 by the expression @xmath134 of course , @xmath135 if @xmath136 and @xmath137 if @xmath138
. on the other hand , for the ideal gas @xmath139 , we recover the well - known result @xmath140 .
the specific heat is ploted as a function of @xmath47 on fig .
[ capacite ] . at the critical point @xmath141
at which @xmath142 diverges , we write @xmath143 and expand @xmath144 and @xmath145 in terms of @xmath146 , using eqs .
( [ crit2])-([crit3 ] ) and @xmath147 , @xmath148 .
to lowest order , we have @xmath149 @xmath150 where @xmath151 is the value of @xmath145 at @xmath152 . eliminating @xmath146 from these expressions
, we find the relation between @xmath144 and @xmath145 close to the critical point @xmath153 then , using eq .
( [ crit1 ] ) , we find explicitely @xmath154 the specific heat diverges as a function of the energy with an exponent @xmath155 and as a function of the temperature with an exponent @xmath156 ( see figs .
[ phencrit]-[ceta ] ) .
it becomes _ negative _ for @xmath157 .
this region of negative specific heats is allowed in the microcanonical ensemble but not in the canonical ensemble .
it is also of interest to determine the behavior of the central density @xmath69 near the critical point ( c. sire , private communication ) .
the central density is related to @xmath47 by the relation @xmath78 or , alternatively , @xmath158 to first order in @xmath146 , we have @xmath159 where @xmath160 is the value of the central density at the critical point @xmath161 .
using eqs .
( [ crit5])-([crit6 ] ) , we get @xmath162 quite remarkably , this stationary result can also be derived from dynamical considerations ( see chavanis _
note also that , using eqs .
( [ contraste])([uv ] ) , we find @xmath163 , which implies @xmath164 .
the stability of the gaseous spheres can be studied by extending padmanabhan ( 1989 ) s stability analysis to the canonical ensemble .
this will provide the profiles of the perturbations that induce instability .
a critical point of free energy is a local maximum if , and only if , the second variations @xmath165 are negative for any variation @xmath49 that conserves mass to first order .
this conservation law imposes @xmath166 we shall restrict ourselves to spherically symmetric perturbations since only spherically symmetric perturbations can induce instability for non rotating systems .
following padmanabhan ( 1989 ) , we introduce the variable @xmath167 defined by @xmath168 physically , @xmath167 represents the mass perturbation @xmath169 within the sphere of radius @xmath170 .
the boundary conditions on @xmath167 are thus @xmath171 substituting the foregoing expression for @xmath172 in eq .
( [ s1 ] ) , we obtain @xmath173 integrating by parts and using the boundary conditions ( [ s4 ] ) , we get @xmath174 then , using gauss theorem in the form @xmath175 we find @xmath176 or , equivalently , @xmath177 the second variations of free energy can be positive ( implying instability ) only if the differential operator which occurs in the integral has positive eigenvalues .
we need therefore to consider the eigenvalue problem @xmath178 with @xmath179 .
if all the eigenvalues @xmath180 are negative , then the critical point is a _
maximum _ of free energy .
if at least one eigenvalue is positive , the critical point is an unstable saddle point .
the point of marginal stability @xmath141 is determined by the condition that the largest eigenvalue is equal to zero .
we thus have to solve the differential equation @xmath181 with @xmath182 .
note that this equation corresponds to that found by padmanabhan ( 1989 ) in the microcanonical ensemble with @xmath183 . introducing the dimensionless variables of section [ sec_diagram ] , we can rewrite the foregoing equation in the form @xmath184 with @xmath185 . as shown by padmanabhan ( 1989 ) , this equation can be solved without solving explicitly the emden equation ( [ emden ] ) .
let us introduce the differential operator @xmath186 using eq .
( [ emden ] ) , it is readily established that @xmath187 similarly , we find @xmath188 these identities suggest to seek a solution of eq .
( [ s12 ] ) in the form @xmath189 where @xmath190 and @xmath191 are constants . substituting the foregoing expression for @xmath192 in the differential equation ( [ s12 ] ) , we find that @xmath193 .
the function @xmath192 can therefore be reexpressed as @xmath194 the boundary condition @xmath195 is automatically satisfied ( we can show furthermore that eq .
( [ s17 ] ) is the only solution consistent with this boundary condition ) .
the condition @xmath196 will determine the point on the spiral of fig .
[ isocollapse ] at which the gaseous spheres become unstable ( i.e. , the point of marginal stability ) . from eq .
( [ s17 ] ) , we get @xmath197 this can be expressed in terms of the milne variables ( [ uv ] ) as @xmath198 considering eq .
( [ uv3 ] ) , we see that the condition ( [ s19 ] ) determines the points for which @xmath101 is extremum ( @xmath199 ) . since @xmath200 according to eq .
( [ uvbeta ] ) , we deduce that a change of stability occurs each time that the temperature is extremum ( see fig .
[ etaalpha ] ) in complete agreement with the theorem of katz ( 1978 ) .
in particular , the first instability ( for which the largest eigenvalue @xmath180 becomes positive ) corresponds to @xmath201 ( for this value @xmath202 ) .
we show in the next section that the secondary instabilities occur at values of @xmath47 that follow asymptotically a geometric progression .
the profiles of the perturbations that induce these instabilities are discussed in section [ sec_profile ] .
a mode of stability is lost each time that @xmath203 .
now , it is easy to have an analytical estimate of @xmath204 for large values of @xmath102 . introducing the asymptotic behavior ( [ ab2 ] ) of @xmath205 in the milne variable @xmath100 defined by eq .
( [ uv ] ) we get for @xmath108 : @xmath206 therefore , the condition @xmath207 corresponds to values of @xmath47 satisfying @xmath208 or equivalently @xmath209 this determines a succession of values that follow the geometric progression @xmath210^{n } \qquad ( n\rightarrow + \infty,\ { \rm integer } ) .
\label{g4}\ ] ] at these points a new eigenvalue becomes positive implying secondary instabilities .
clearly , this geometric progression can be traced back to the ` curious ' asymptotic behavior of the solutions governing the isothermal gas sphere ( see eq .
( [ ab2 ] ) ) which intersects the singular solution ( [ sing2 ] ) infinitely often at points that asymptotically increase geometrically in the ratio @xmath211 .
a mode of stability is lost each time that @xmath47 achieves one of the values given by eq .
( [ g4 ] ) .
if one prefers , the same criterion can be expressed in terms of the density contrast @xmath212 . using eq .
( [ contraste ] ) and the asymptotic expansion ( [ ab2 ] ) we get @xmath213 for @xmath214 .
hence , the result ( [ g4 ] ) is translated in @xmath215^{n } \qquad ( n\rightarrow + \infty,\ { \rm integer } ) .
\label{g6}\ ] ] in this section , we study the form of the perturbations that trigger instability at the critical points . following the method of padmanabhan ( 1989 ) , it is possible to describe the behavior of these perturbation profiles without numerical work . indeed , according to eq .
( [ s3 ] ) , the eigenfunction associated with the eigenvalue @xmath216 can be written @xmath217 where @xmath192 is given by eq .
( [ s17 ] ) . substituting this result in eq .
( [ p1 ] ) and simplifying the derivative with the aid of eq .
( [ emden ] ) we obtain @xmath218 introducing the milne variable @xmath101 defined by eq . (
[ uv ] ) , we can rewrite the foregoing equation in the form @xmath219 the density perturbation @xmath49 becomes zero at the point(s ) @xmath220 such that @xmath221 . the number of zeros is therefore given by the number of intersections between the spiral in the @xmath98 plane and the line @xmath222 ( see fig . [ uvosci ] ) . for the first mode of instability @xmath223 , we need to follow the spiral until the first extremum of @xmath101 , which corresponds to the point at which the box terminates ( recall that @xmath224 , see section [ sec_marginal ] ) . in that case
there is only one intersection @xmath225 with the line @xmath222 .
therefore , the first mode of instability does _ not _ present a `` core - halo '' structure ( fig .
[ dnsn - xb2.51760387xi ] ) , contrary to what happens in the microcanonical ensemble ( padmanabhan 1989 ) . for the second mode of instability ( @xmath226 )
, we need to follow the spiral until the second extremum of @xmath101 .
the intersection with the line @xmath222 now determines two zeros @xmath225 and @xmath227 , the first one being the same as for the first mode of instability .
the profile of the second mode of instability therefore has a `` core - halo '' structure ( fig . [ dnsn - x.b1.84269667xi ] ) .
when we spiral inward , more and more intersections are obtained .
therefore the high order modes of instability present more and more oscillations .
for these modes , it is easy to determine the asymptotic positions of their zeros . from eqs .
( [ uv ] ) and ( [ ab2 ] ) , we have for @xmath108 : @xmath228 thus , substituting @xmath222 in eq .
( [ p4 ] ) , we get @xmath229 therefore , the zeros follow asymptotically the geometric progression ( titius - bode law ) .
this is recognized to be an effect of scale invariance ( graner & dubrulle 1994 , dubrulle & graner 1994 ) .
there is , however , no real connexion with the present work and planets may have formed quite differently ( see chavanis ( 2000 ) and references therein ) . ]
@xmath230^{i } \qquad ( i\rightarrow + \infty , i \
{ \rm integer } ) .
\label{p5}\ ] ] note that the perturbation that destabilizes the singular sphere ( at the end of the spiral ) has an _ infinite _ number of oscillations !
we shall now investigate the dynamical stability of isothermal spheres described by navier - stokes equations and compare the results with the thermodynamical approach .
this problem was previously addressed by yabushita ( 1968 ) in the isobaric ensemble and by semelin _
( 2001 ) in the canonical ensemble .
we shall prove analytically and with no approximation the equivalence between dynamical and thermodynamical stability and the fact that the stability of isothermal spheres does not depend on the viscosity .
we shall also describe the behavior of the velocity profiles at the critical points .
the equations of the problem are the equation of continuity , the equation of motion and the poisson equation @xmath231 @xmath232 @xmath233 they must be completed by an equation of state that we take of the form @xmath234 where @xmath13 is constant .
we also recall that we work within a box of radius @xmath2 .
for astrophysical fluids , the reynolds numbers are so huge that we can neglect the viscosity in the navier - stokes equation
. we shall take therefore @xmath235 in most of the calculations but we will also prove that viscosity does not change the onset of instability . clearly , the stationary solutions of eqs .
( [ f1])([f2])([f3 ] ) correspond to the isothermal gaseous spheres studied in the previous sections .
it should be recalled that the condition of mechanical equilibrium corresponds to the balance between the pressure force and the gravitational force . for an isothermal gas , this yields @xmath236
we now consider a small perturbation around a stationary solution and write @xmath237 @xmath238 where the bar refers to the stationary solution ( in the following we shall drop the bar ) .
the linearized equations for the perturbations are @xmath239 @xmath240 @xmath241 since we work with exact stationary solutions of the fluid equations , that are inhomogeneous , we do not have to advocate the jeans swindle ( binney & tremaine 1987 ) . in this sense ,
our approach is more rigorous than the classical treatment of jeans starting from an infinite and homogeneous medium . in the following ,
we restrict ourselves to radial perturbations .
it is known that non - radial perturbations do not lead to new instabilities . writing the time dependance of the perturbation in the form @xmath242 , @xmath243 , ... , the equations of the problem become @xmath244 @xmath245 @xmath246 introducing the notation ( [ s3 ] ) and using the gauss theorem ( [ s7 ] ) we see that the poisson equation ( [ d11 ] ) is automatically satisfied .
the continuity equation ( [ d10 ] ) becomes @xmath247 this equation is readily integrated . using the boundary condition @xmath248 , we get @xmath249 substituting this result back into eq .
( [ d9 ] ) , we obtain @xmath250 using the condition of mechanical equilibrium ( [ f4 ] ) , the foregoing equation can be rewritten @xmath251 this is similar to the eigenvalue equation ( [ s10 ] ) associated with the second variations of the free energy . in particular they coincide for marginal stability @xmath216 .
therefore , dynamical and thermodynamical instability occur at the same point in the series of equilibrium .
this was not obvious _ a priori _ since the navier - stokes equations without viscosity conserve the free energy .
taking into account a finite viscosity , we obtain instead of eq .
( [ d15 ] ) , @xmath252 for @xmath216 the viscous term cancels out .
therefore , viscosity does not change the onset of instability , nor does it alter the profile of the perturbation that triggers the instability .
the profiles of the density perturbation have already been discussed in section [ sec_profile ] .
the profiles of the velocity perturbation at the critical points are given by eq .
( [ d13 ] ) . in order to work with dimensionless variables ,
we introduce the speed of sound @xmath253 note that @xmath254 can be written @xmath255 where @xmath256 is a time scale comparable with the dynamical time ( binney & tremaine 1987 ) .
then @xmath257 where we have set @xmath258 . using eqs . (
[ s17])-([uv ] ) , we find @xmath259 we assume that we are just at the onset of the instability ( @xmath260 ) so that eq .
( [ d18 ] ) is applicable with @xmath261 ( the velocity profile at marginal stability , i.e. @xmath262 , is simply @xmath263 ) . since @xmath264
, the velocity always vanishes at the center of the sphere @xmath265 .
the other zero(s ) are determined by the condition @xmath266 .
since the critical points of instability are also characterized by @xmath267 , see section [ sec_marginal ] , we deduce that the velocity perturbation always vanishes at the wall : @xmath268 .
this result is to be expected since the mass does not leave the sphere .
the other zeros are determined by the intersections between the spiral in the @xmath98 plane and the line @xmath117 ( see fig .
[ uvosci ] ) .
they correspond precisely to the values of @xmath47 at which a new mode of instability occurs , i.e. @xmath269 .
it is therefore straightforward to determine the number of zeros in the velocity profile .
for the first instability there are only two zeros at @xmath74 and @xmath270 .
for the second instability there are three zeros at @xmath74 , @xmath270 and @xmath271 .
for the @xmath272 instability there are @xmath273 zeros @xmath74 , @xmath270 , @xmath271 , ... , @xmath274 .
the exact velocity profiles given by eq .
( [ d18 ] ) are displayed on figs .
[ deltav1]-[deltav2 ] for @xmath275 .
they are in agreement with the numerical results of semelin _
( 2001 ) . for high order modes of instability ,
the zeros follow asymptotically the geometric progression ( [ g4 ] ) . returning to dimensional variables ,
the zeros of the @xmath272 mode of instability are located at @xmath67 and @xmath276 .
we shall now express the previous results in a more physical form and make some speculations about the fragmentation and the fractal structure of an isothermal self - gravitating gas .
introducing the king s radius ( binney & tremaine 1987 ) @xmath277 the parameter @xmath47 defined in section [ sec_diagram ] can be rewritten @xmath278 the king s radius gives a good estimate of the core radius of an isothermal sphere . in the following , we shall consider that the core radius @xmath20 is fixed and that the domain size @xmath2 is increased .
clearly , this amounts to increasing the parameter @xmath47 along the spiral . from the study of section [ sec_marginal ] , we find that the onset of instability corresponds to @xmath279 for larger radii there are still critical points of free energy but they are more and more unstable . according to eq .
( [ g4 ] ) , a new mode of instability occurs at radii that follow asymptotically the geometric progression @xmath280^{n } l_{k}. \label{g12}\ ] ] this result is clearly related to that of semelin _
( 1999 ) who first noticed the appearance of a hierarchy of scales connected to the instability of isothermal gaseous spheres . however , the connection with their work is not straightforward .
( 1999 ) approximate the regular solutions of eq .
( [ mfr ] ) by the analytical profile ( [ singular ] ) and introduce an arbitrary short cut - off distance @xmath281 to avoid the central singularity .
by contrast , there is no cut - off at small scales in our theory since we consider regular spheres that are exact solutions of the emden equation ( although we do not have analytical expressions for them ) . in this sense , there are no approximation nor indetermination in our theory .
in fact , an `` effective '' cut - off is played by the king s length ( or core radius ) but it is not arbitrary and depends on physical parameters , the central density @xmath69 and the temperature @xmath13 .
for @xmath282 , the density profile behaves like @xmath283 so that regular gaseous spheres can indeed be approximated by a @xmath283 profile plus a cut - off @xmath281 at small scales as done by semelin _
our approach suggests to take @xmath281 as the king s length .
we have therefore given a new derivation of their result which is considerably simpler and does not introduce sophisticated renormalization group technics .
it also avoids the introduction of an arbitrary cut - off .
this provides , hopefully , an easier interpretation of their intriguing results .
according to section [ sec_profile ] , if we increase the domain size along the series of equilibrium ( i.e. maintaining a fixed core radius ) smaller and smaller regions become unstable . the modes that trigger these instabilities present numerous oscillations whose
zeros also follow a geometric progression .
these oscillations can be regarded as sort of ` germs ' leading to the formation of ` clumps ' preceding the fragmentation of the isothermal gas .
it is expected that these ` clumps ' will evolve by achieving higher and higher density contrasts , and finally fragment in turn into substructures .
this yields a hierarchy of structures fitting one into each other in a self - similar way .
this picture is given further support by the fact that _ both _ the domain sizes that induce instability and the zeros of the perturbation in a given domain follow a geometric progression with the same ratio ( see eqs .
( [ g12])([p5 ] ) ) .
this property may explain in a quite natural fashion the fractal structure that a self - gravitating medium ( e.g. , the interstellar medium , the large scale structures of the universe , ... ) can build under certain conditions .
the hierarchy of scales that appears in an isothermal gas is due intrinsically to the oscillating nature of the equilibrium phase diagram ( see figs . [ gravotherme],[isocollapse],[uvcrit ] ) .
this curious behavior has been known for a long time in the context of stellar structure ( chandrasekhar 1942 ) but it had not been apparently related to the geometric progression of scales inducing instability .
we can also introduce another length scale defined with the average density @xmath284 of the system instead of the central density .
this is the jeans length ( binney & tremaine 1987 ) @xmath285 the reduced temperature @xmath144 can be expressed in terms of the jeans length as ( de vega & sanchez 2001a ) @xmath286 in the following , we shall consider that the jeans length @xmath19 is fixed and that the domain size @xmath2 is increased .
clearly , this amounts to increasing the parameter @xmath144 along the vertical axis of fig .
[ isocollapse ] . using the results of section [ sec_katz ]
, we find that the gaseous sphere becomes unstable when @xmath287 for larger radii the system will collapse because there are _ no _ critical point of free energy ( i.e. hydrostatic equilibrium ) . in that case
there are clearly no secondary instabilities .
indeed , when the condition ( [ g15 ] ) is realized we leave the spiral and loose the geometric hierarchy of scales associated with its turning points .
since the mode of instability at the critical point does not oscillate ( see section [ sec_profile ] ) , we expect that the system will collapse _ without _ fragmenting , probably like in the study of penston ( 1969 ) .
the fragmentation of the isothermal self - gravitating gas and the fractal structure that it can generate were previously discussed by de vega _ et al . _ ( 1996a , 1996b , 1998 ) and we here improve and complete their physical arguments .
in particular , the analytical calculation of the modes of instablity ( without approximation ) , the geometric progression of _ both _ the domain sizes and the ` clumps ' sizes , and the precise physical picture that emerges from these results are new . in addition , we give arguments explaining why the instability of a self - gravitating gas does not always lead to a fractal structure ( indeed globular clusters or elliptical galaxies , for example , do not show a hierarchical structure ) .
in our model , the selection of the regime is related to how the size of the domain compares with the jeans length or the king s length . if @xmath288 , we expect a collapse without fragmentation but if @xmath289 ( with a sufficiently large ratio ) the medium is expected to fragment into several ` clumps ' . the distinction between the jeans scale and the king s scale
is therefore essential in the present context and had not been noticed previously . in particular
it is not the jeans lengths but the king s lengths that follow a geometric progression [ see formula ( [ g12 ] ) ] contrary to the claim of semelin _
these two scales only coincide for a perfect gas with no self - gravity that is homogeneous . indeed , for @xmath86
, we have @xmath290 . by contrast , for @xmath214 ( corresponding to the singular sphere ) we get @xmath291 and @xmath292 ( we have used the asymptotic expansions of section [ sec_diagram ] ) . to appreciate the difference between the two length scales , it is convenient to introduce a dimensionless number @xmath293 from the above formulae , we have the equivalent expressions @xmath294 for @xmath86 ( perfect gas ) , @xmath295 and for @xmath214 ( singular sphere ) , @xmath296 . therefore the number @xmath297 quantifies the inhomogeneity of the self - gravitating gas . the evolution of @xmath298 along the spiral is ploted on fig .
[ n ] . from eq .
( [ g17 ] ) , @xmath297 is just the ratio between the central density and the average density to the power @xmath299 .
in the first part of the paper , we have investigated the stability of isothermal spheres in the canonical ensemble using a thermodynamical approach .
this is a natural extension of the work of padmanabhan ( 1989 ) in the microcanonical ensemble .
as is well known , there is no hydrostatic equilibrium for an isothermal gas below a critical temperature @xmath118 ( lynden - bell & wood 1968 ) .
this corresponds to a situation in which the domain size becomes comparable with the jeans length , i.e. @xmath300 . in that case
, we expect the system to collapse . by analyzing the second variations of free energy
, we have found that the point of minimum temperature is precisely the point at which the free energy ceases to be a local maximum and becomes a saddle point in the series of equilibrium .
this vindicates the theorem of katz ( 1978 ) and provides in addition the form of the perturbation that induces instability .
contrary to the microcanonical ensemble , it has not a `` core - halo '' structure , nor multiple oscillations .
we expect therefore the system to collapse without fragmenting . on the other hand , if we start from isothermal gaseous spheres with high density contrasts @xmath122 ( this corresponds to @xmath289 ) , such spheres are unstable saddle points and
probably fragment into small ` clumps ' ( although they would be said to be stable by a naive jeans argument since @xmath301 ) .
the number of clumps , presumably associated with the number of zeros in the mode of instability , increases with the density contrast or , equivalently , as the domain size becomes larger and larger with respect to the king s radius .
the distinction between the jeans length and the king s length is essential in the present context and had not been noticed previously .
the jeans length is related to the parameter @xmath144 determining whether an hydrostatic equilibrium state exists or not ( for an isothermal sphere ) while the king s length is related to the parameter @xmath47 parametrizing the series of equilibrium .
they are very different parameters . in the second part of the paper ,
we have extended jeans instability criterion to the case of inhomogeneous bounded gaseous spheres described by navier - stokes equations .
this treatment avoids the well - known difficulties associated with an infinite and homogeneous medium ( jeans swindle ) .
quite remarkably , we have provided an elegant analytical solution of this more general problem without making any approximation .
now , the question that naturally emerges is what happens to a gravitationally unstable gaseous configuration .
the collapse of isothermal gas spheres has been investigated by penston ( 1969 ) using navier - stokes equations ( without viscosity ) .
he found an analytical solution for which the collapse is self - similar and develops a finite time singularity ( i.e. , the central density becomes infinite in a finite time ) .
this only _ partially _ answers the above question .
indeed , his solution is just a particular solution of the fluid equations and other solutions ( e.g. , fragmentation ) may be possible . in particular , it is not clear whether his solution is stable with respect to non spherically symmetric perturbations . in addition , penston ( 1969 ) works in an infinite domain for which the isothermal spheres have infinite mass and are always unstable .
therefore , he could not evidence a phase transition between an equilibrium `` gaseous '' state and a `` collapsed '' state , depending on the value of the temperature .
the thermodynamics of self - gravitating systems is so intriguing that we have been tempted to explore this problem anew from a dynamical point of view ( chavanis _ et al .
we have introduced a numerical algorithm in the form of a relaxation equation ( chavanis 1996 , 2001a ; chavanis _ et al .
_ 1996 ) constructed so as to increase monotonically entropy at fixed energy in the microcanonical ensemble and free energy at fixed temperature in the canonical ensemble . with this numerical algorithm , we have covered the whole bifurcation diagram in parameter space and we have checked , by an independant method , the stability limits of katz ( 1978 ) . we have verified that the density profile that induces instability at the critical point has a `` core - halo '' structure in the microcanonical ensemble but not in the canonical ensemble , in agreement with the study of padmanabhan ( 1989 ) and the present study .
we have also checked that the number of oscillations in the perturbation profile increases when we start from unstable isothermal spheres with high density contrasts . with this algorithm ,
we have explored the `` bassin of attraction '' of the stable isothermal spheres .
since they only are _ local _ maxima of entropy or free energy , an initial condition can either relax towards a metastable equilibrium state or collapse , depending on its _
topology_. hence , the control parameter @xmath145 or @xmath144 is not sufficient to characterize completely the final state of the system even in the case when an equilibrium exist .
this depends whether the initial condition lies in the `` bassin of attraction '' of the equilibrium solution or not .
the complete characterization of this bassin of attraction is difficult but we have given some examples . with our numerical algorithm , it will be possible to include rotation in order to compute entropy maxima that are not spherically symmetric and determine the influence of angular momentum conservation on the collapse of a rotating self - gravitating system ( chavanis & rosier , in preparation ) .
some interesting works have been done in that direction ( lagoute & longaretti 1996 , laliena 1999 , fliegans & gross 2001 and lynden - bell 2000 ) but the complete description of rotating isothermal gaseous spheres is still in its infancy stage .
it is also of first interest to test numerically the speculations that we have made concerning the development of the instability , i.e. if the system will collapse as a whole or if it will fragment and break into substructures .
these two regimes are clearly separated , at least in the theoretical model that we have considered here ( antonov model ) . we have suggested that if we start from a solution of the emden equation with a high density contrast , it should break in a series of ` clumps ' associated with the oscillations of the density perturbation that we have calculated in the linear regime . obviously ,
if we want to follow the nonlinear evolution correctly , we must relax the hypothesis of spherical symmetry . in this paper
, we have assumed that the particles can be treated by classical mechanics and that they are point - like .
the extension of our work to general relativity ( for , e.g. , neutron stars ) is almost straightforward ( chavanis 2001b ) .
we find that there is no hydrostatic equilibrium if the system size is smaller than a multiple of the schwarzschild radius .
we also show analytically that the point of smallest radius corresponds precisely to the point at which the series of equilibrium becomes unstable .
secondary modes of instability are found ; they follow a geometric progression like in the newtonian case but with a different ratio .
when the speed of light goes to @xmath57 , we recover the results of the present study .
it is seen that relativistic effects _ favour _ gravitational instability , i.e. instability occurs sooner than in the newtonian limit . on the other hand , the case of self - gravitating fermions enclosed within a box has been considered by chavanis & sommeria ( 1998 ) .
their study was formulated in the context of the `` violent relaxation '' of collisionless stellar systems introduced by lynden - bell ( 1967 ) but the results also apply to quantum particles such as massive neutrinos in dark matter models ( ingrosso _ et al .
_ 1992 ) or degenerate electrons in white dwarf stars ( chandrasekhar 1959 ) .
when degeneracy is accounted for , there exists a _
entropy maximum for all values of energy ( chavanis & sommeria 1998 , robert 1998 ) .
it has a `` core - halo '' structure with a degenerate core and a dilute atmosphere . depending on the degeneracy parameter
, there can be a `` gravothermal catastrophe '' at @xmath3 but the core ceases to shrink when it becomes degenerate . considering a classical gas with a short distance cut - off
@xmath302 essentially leads to the same results ( see , e.g. , padmanabhan 1990 , follana & laliena 2000 ) . for astrophysical purposes
, it is still a matter of debate to decide whether collisionless stellar systems like elliptical galaxies are degenerate ( in the sense of lynden - bell ) or not .
since degeneracy can stabilize the system without changing its overall structure at large radii , we have suggested that degeneracy could play a role in galactic nuclei ( chavanis & sommeria 1998 ) .
the recent simulations of leeuwin and athanassoula ( 2000 ) and the theoretical model of stiavelli ( 1998 ) seem to go in that direction : if the nucleus of elliptical galaxies contains a ( primordial ) black hole , degeneracy must be taken into account and can explain the cusps observed in the center of galaxies .
this form of degeneracy is also relevant for massive neutrinos in dark matter models where it competes with quantum degeneracy ( kull _ et al .
_ 1996 ) .
all together , these results suggest that the statistical mechanics and the thermodynamics of self - gravitating systems can have quite relevant astrophysical applications at different scales of the universe : stars ( chandraskhar 1959 ) , interstellar medium ( de vega _ et al .
_ 1996a , 1996b ) , globular clusters ( lynden - bell & wood 1968 ) , elliptical galaxies ( lynden - bell 1967 , hjorth & madsen 1993 , chavanis & sommeria 1998 ) , dark matter ( ingrosso _ et al .
_ 1992 , kull _ et al .
_ 1996 ) , cosmology ( saslaw & hamilton 1984 , de vega _ et al . _
1998) ... the same ideas of statistical mechanics have been introduced in two - dimensional turbulence to explain the formation and maintenance of coherent vortices , like jupiter s great red spot , which are common features of large - scale geophysical and astrophysical flows ( see bouchet & sommeria 2000 , bouchet _ et al .
_ 2001 and references therein ) .
the analogy between the statistical mechanics of two - dimensional vortices and stellar systems has been described extensively by chavanis ( 1996 , 1998b , 2001c ) .
this analogy concerns not only the equilibrium state ( the formation of large - scale structures ) but also the relaxation towards equilibrium ( chavanis _ et al .
_ 1996 , chavanis 2001d ) and the statistics of fluctuations ( chavanis & sire 2000 ) .
these ideas may also have applications in the context of planet formation where large - scale vortices present in the solar nebula could efficiently trap dust particles to form the planetesimals and the planets ( barge & sommeria 1995 , tanga _ et al .
_ 1996 , bracco _ et al .
_ 1999 , chavanis 2000 ) .
accordingly , the statistical mechanics of these nonlinear media seems to be able to account for the fascinating process of self - organization in nature .
i acknowledge interesting discussions with l. athanassoula , f. combes , o. fliegans , j. hjorth , h. kandrup , v. laliena , f. leeuwin , d. lynden - bell , t. padmanabhan , r. robert , b. semelin and j. sommeria on these exciting problems of thermodynamics .
this work was initiated during my stay at the institute for theoretical physics , santa barbara , during the program on hydrodynamical and astrophysical turbulence ( february - june 2000 ) .
this research was supported in part by the national science foundation under grant no .
phy94 - 07194 . | we investigate the stability of bounded self - gravitating systems in the canonical ensemble by using a thermodynamical approach .
our study extends the earlier work of padmanabhan [ apj supp .
71 , 651 ( 1989 ) ] in the microcanonical ensemble . by studying the second variations of the free energy
, we find that instability sets in precisely at the point of minimum temperature in agreement with the theorem of katz [ mnras 183 , 765 ( 1978 ) ] .
the perturbation that induces instability at this point is calculated explicitly ; it has _ not _ a `` core - halo '' structure contrary to what happens in the microcanonical ensemble .
we also study jeans type gravitational instability of isothermal gaseous spheres described by navier - stokes equations .
the introduction of a container and the consideration of an inhomogeneous distribution of matter avoids the jeans swindle .
we show analytically the equivalence between dynamical stability and thermodynamical stability and the fact that the stability of isothermal gas spheres does not depend on the viscosity .
this confirms the findings of semelin _
et al . _
[ phys . rev .
d 63 084005 ( 2001 ) ] who used numerical methods or approximations .
we also give a simpler derivation of the geometric hierarchy of scales inducing instability discovered by these authors .
the density profiles that trigger these instabilities are calculated analytically ; high order modes of instability present numerous oscillations that also follow a geometric progression .
this suggests that the system will fragment in a series of ` clumps ' and that these ` clumps ' will themselves fragment in substructures .
the fact that both the domain sizes leading to instability and the ` clumps ' sizes within a domain follow a geometric progression with the same ratio suggests a fractal - like behavior .
this gives further support to the interpretation of de vega et al .
[ nature , 383 , 56 ( 1996 ) ] .
@xmath0 laboratoire de physique quantique , universit paul sabatier , + 118 route de narbonne 31062 toulouse , france .
+ @xmath1 institute for theoretical physics , university of california , santa barbara , california .
+ e - mail : [email protected] ; web : http://w3-phystheo.ups-tlse.fr/chavanis/ + |
in many practices , spectral problems are faced for differential equations which have discontinuous coefficient and discontinuity conditions in interval ( @xcite-@xcite ) .
these problems generally emerge in physics , mechanics and geophysics in non - homogeneous and discontinuous environments .
we consider a heat problem in a rod which is composed of materials having different densities . in the initial time , let the temperature is given arbitrary .
let be the temperature is zero in one end of the rod and the heat is isolated at the other end of the rod . in this case
the heat flow in non - homogeneous rod is expressed with the following boundary problem : @xmath0 @xmath1 where @xmath2 , @xmath3 are physical parameters and have specific properties .
for instance , @xmath2 defines the density of the material and piecewise - continuous function . applying the method of separation of variables to this problem
, we get the spectral problem below : @xmath4 @xmath5 here @xmath6 is a real - valued function , @xmath2 piecewise - continuous function the following : @xmath7 @xmath8 is spectral parameter and @xmath9 . when @xmath10 or @xmath11 , that is , in continuous case , the solution of inverse problem is given in @xcite-@xcite .
the spectral properties of sturm - liouville opeator with discontinuos coefficient in different boundary conditions are examined in @xcite-@xcite . in this study ,
the main equation is obtained which has an important role in solution of inverse problem for boundary value problem and according to spectral data , the uniqueness of solution of inverse problem is proved .
similar problems are examined for the equation ( [ 1 ] ) with different boundary conditions in @xcite .
it was proved ( see @xcite ) , that the solution @xmath12 of the equation ( [ 1 ] ) with initial conditions @xmath13 @xmath14 can be represented as @xmath15 where @xmath16 belongs to the space @xmath17 for each fixed @xmath18 $ ] and is related with the coefficient @xmath19 of the equation ( [ 1 ] ) by the formula:@xmath20@xmath21is the solution of ( [ 1 ] ) when @xmath22@xmath23it is similarly shown in @xcite , @xcite that the roots of the equation @xmath24 have the following form @xmath25where @xmath26 are the eigenvalues of problem ( [ 1 ] ) , ( [ 2 ] ) when @xmath27 , @xmath28 is a bounded sequence , @xmath29 and norming constants : @xmath30
for each fixed @xmath18 $ ] the kernel @xmath16 from the representation ( [ 4 ] ) satisfies the following linear functional integral equation @xmath31@xmath32where@xmath33@xmath34@xmath26 are eigenvalues and @xmath35 are norming constants of the boundary value problem ( [ 1 ] ) , ( [ 2 ] ) when @xmath36 from ( [ 4 ] ) we have@xmath37it follows from ( [ 4 ] ) and ( [ 11 ] ) that@xmath38@xmath39@xmath40@xmath41@xmath42@xmath43@xmath44@xmath45using the last two equalities , we obtain@xmath46@xmath40@xmath47@xmath42@xmath48or@xmath49where@xmath50@xmath51@xmath52@xmath53@xmath54it is easily found by using ( [ 9 ] ) and ( [ 10])@xmath55let @xmath56 be an absolutely continuous function , @xmath57 then using expansion formula ( see @xcite),@xmath58@xmath59using ( [ 13 ] ) we have:@xmath60@xmath61@xmath62@xmath63we obtain uniformly on @xmath18$]@xmath64@xmath65@xmath66it follows from ( [ 6 ] ) that @xmath67 taking into account ( [ 16 ] ) and ( [ 13 ] ) , we get @xmath68 @xmath69@xmath70@xmath71@xmath72@xmath73@xmath74@xmath75@xmath76@xmath77@xmath78@xmath79@xmath80@xmath81substituting @xmath82 and @xmath83 we obtain @xmath84@xmath85@xmath86since @xmath87 when @xmath88 , we have @xmath89@xmath90@xmath91thus , uniformly on @xmath18:$]@xmath92@xmath93using ( [ 9 ] ) , uniformly on @xmath18$]@xmath94@xmath95@xmath96@xmath97@xmath96@xmath98using the residue theorem and the formula @xmath99 ( see @xcite ) , where @xmath100 is the solution of ( [ 1 ] ) with initial condition @xmath101 @xmath102 and @xmath103 is the characteristic function of ( [ 1])-([3 ] ) , @xmath104 we calculate@xmath105@xmath106@xmath107@xmath108 dt=\]]@xmath109@xmath110@xmath111@xmath112@xmath113where @xmath114 since ( see @xcite)@xmath115(@xmath116 @xmath117 is a sufficiently small positive number ) and according to lemma 1.3.1 from @xcite@xmath118from the equality ( [ 19 ] ) we get @xmath119multiplying both sides of ( [ 12 ] ) by @xmath120 integrating from @xmath121 to @xmath122 , tending to limit when @xmath123 and using ( [ 14 ] ) , ( [ 15 ] ) , ( [ 17 ] ) , ( [ 18 ] ) and ( [ 20 ] ) we have @xmath124@xmath125since @xmath56 can be chosen arbitrarily , we obtain @xmath126@xmath127
for each fixed @xmath18 $ ] main equation ( [ 8 ] ) has a unique solution @xmath128 .
we show that for each fixed @xmath129 the equation ( [ 8 ] ) is equivalent to the equation of the form @xmath130 where @xmath131 is a completely continuous operator , @xmath132 is an identity operator in the space @xmath133 .
( when @xmath134 this fact is obvious . )
when @xmath129 rewrite ( [ 8 ] ) as @xmath135where@xmath136@xmath137 it is sufficient to prove that @xmath138 is invertible , i.e. has a bounded inverse in @xmath17 .
consider the equation @xmath139
@xmath140 i.e. @xmath141from here it is easily to obtain @xmath142we show that @xmath143 in fact , @xmath144@xmath145@xmath146@xmath147@xmath148@xmath149we put @xmath150 , when @xmath151 .
then @xmath152 so the operator @xmath138 is invertible in @xmath17 . then according to theorem 3 from @xcite ( see p. 275 ) it is sufficient to prove that the equation @xmath153@xmath154has only trivial solution @xmath155 let @xmath156 be a non - trivial solution of ( [ 22 ] ) .
then @xmath157 @xmath158 @xmath159 from ( [ 9 ] ) we have @xmath157 @xmath160 @xmath161 @xmath162 @xmath163 using ( [ 7 ] ) and ( [ 16 ] ) we obtain @xmath157 @xmath164 @xmath165 @xmath166 @xmath167 @xmath168 @xmath167 @xmath166 @xmath169 @xmath168 @xmath170 @xmath171 @xmath172 @xmath173 @xmath174 @xmath175 @xmath174 @xmath173 @xmath176 @xmath175 @xmath177 substituting @xmath178 in third , fourth , ninth , and tenth double integrals and @xmath179 in fifth , sixth , eleventh and twelfth double integrals we get @xmath157@xmath164@xmath165@xmath180@xmath181@xmath182@xmath181@xmath183@xmath184@xmath185@xmath186@xmath171@xmath187@xmath188@xmath189@xmath190@xmath191@xmath190@xmath192@xmath193@xmath194@xmath195from which we have @xmath157@xmath196@xmath197@xmath198@xmath197@xmath199@xmath200@xmath201@xmath202@xmath203@xmath204@xmath205@xmath204@xmath206@xmath207@xmath208@xmath209thus we obtain @xmath157@xmath210@xmath211using the parseval s equality @xmath212for the function @xmath213we have @xmath214or @xmath215since the system @xmath216 is compete in @xmath217 we have @xmath218i.e .
@xmath219 where the operator @xmath138 is defined by ( [ 21 ] ) . from invertibility of @xmath138 in @xmath220
we get @xmath221 let @xmath222 and @xmath223 be two boundary value problems and@xmath224then @xmath225.\ ] ] according to ( [ 9 ] ) and ( [ 10 ] ) @xmath226 and @xmath227 . then from the main equation ( [ 8 ] ) , we have @xmath228 .
it follows from ( [ 5 ] ) that @xmath229 @xmath18 $ ] .
using @xcite , we can transform the main equation ( [ 8 ] ) to the following equation : @xmath230 where @xmath231 and @xmath232 we assume that @xmath233 from the formula ( [ 24 ] ) , we obtain @xmath234 substituting ( [ 25 ] ) into the main equation ( [ 23 ] ) we obtain @xmath235,\ ] ] @xmath236 where @xmath237 substituting ( [ 26 ] ) into ( [ 27 ] ) we obtain @xmath238 using the ( [ 6 ] ) , we calculate the integral into last equation . then , we have @xmath239 where @xmath240 @xmath241+\ ] ] @xmath242.\ ] ] from ( [ 28 ] ) , @xmath243^{-1 } \quad , \quad 0<x < a \\
\varphi_0(x,\lambda_1)\left[1+\frac{1}{\pi}\phi(x,\lambda_1)\right]^{-1 } \quad , \quad a < x<\pi ,
\end{array } \right . \label{29}\ ] ] substituting ( [ 29 ] ) into ( [ 26 ] ) we get @xmath244^{-1 } \quad , \quad 0<x < a \\
-\frac{1}{\pi}\rho(t)\varphi_0(t,\lambda_1)\varphi_0(x,\lambda_1)\left[1+\frac{1}{\pi}\phi(x , lambda_1)\right]^{-1 } \quad , \quad a < x<\pi , \end{array}\right . \label{30}\ ] ] where @xmath245 and @xmath246 .
thus , we obtain the solution of main equation ( [ 8 ] ) .
if we use the formula ( [ 5 ] ) then , we obtain the potential @xmath19 .
this work is supported by the scientific and technological research council of turkey ( tubitak ) .
anderssen rs .
the effect of discontinuities in destiny and shear velocity on the asymptotic overtone structure of torsional eigenfrequencies of the earth .
, geophys .
j. r. astr . soc . ,
1977 , 50 , 303 - 309 .
aliev ba , yakubov ys .
solvability of boundary value problems for second - order elliptic differential - operator equations with a spectral parameter and with a discontinuous coefficient at the highest derivative .
differential equations , 2014 , 50(4 ) , 464475 .
altinisik n , kadakal m , mukhtarov o. eigenvalues and eigenfunctions of discontinuous sturm - liouville problems with eigenparameter dependent boundary conditions .
acta math . hung . , 2004 , 102(12 ) , 159175 . | in this work , a boundary value problem for sturm - liouville operator with discontinuous coefficient is examined .
the main equation is obtained which has an important role in solution of inverse problem for boundary value problem and uniqueness of its solution is proved .
uniqueness theorem for the solution of the inverse problem is given .
[ multiblock footnote omitted ] [ multiblock footnote omitted ] |
compound magnetic structures formed by a ferromagnet ( fm ) in close contact with an antiferromagnet ( afm ) show interesting proximity effects which are enhanced when going to the nanoscale range @xcite . among them , one of the most studied phenomenon is the displacement of the hysteresis loops along the field axis observed in layered fm / afm systems and core / shell nanoparticles when measured after cooling in a magnetic field , an effect that has been termed exchange bias ( eb ) after its first observation by meiklejohn and bean in co particles @xcite . in spite of the intensive research developed in the last decades , there are still many open issues to be understood and no clear - cut model has been able to account for all the observed phenomenology @xcite .
several recent experiments based on spectrocopic techniques @xcite have focused on the observation of the magnetization reversal at the fm / afm interface in different layered systems since it is widely accepted that the magnitude of the exchange bias field is related to the existence of uncompensated moments at the interface
. however , in nanoparticles with fm core / afm shell structure , details about the core / shell interface can not be accessed by these techniques and only indirect information about the magnetic order at the inerface can be gained by magnetization measurements . in this study , we present the results of monte carlo simulations of hysteresis loops of an individual core / shell nanoparticles , which , apart from the usual shift in the field axis direction , display asymmetry between the positive and negative field branches and also shifts in the vertical direction as observed in some experimental systems @xcite . by inspection of the spin configurations during field cycling
, we will show how this phenomenolgy is related to the peculiar interplay between exchange coupling at the core / shell interface and the magnetic order of its uncompensated spins induced after the field cooling process .
the details of model for the core / shell particle on which simulations are based were already presented in our previous studies @xcite , where the details about the simulation protocol used for the monte carlo method can also be found .
the simulations are based on the following hamiltonian : @xmath0 where @xmath1 are classical heisenberg spins of unit magnitude placed at the nodes of a sc lattice . in the first term , the value of the n.n .
exchange constants @xmath2 depends on the spins belonging to different particle regions . at the core , @xmath2 is fm and
has been fixed to @xmath3 k. af spins at the shell have @xmath4 .
the exchange coupling at the interface [ defined as those spinson the core ( shell ) with at least one neighbor on the shell(core ) ] is @xmath5 .
hysteresis loops obtained for a particle with total radius @xmath6a , surface shell thickness @xmath7a and @xmath8 , obtained after cooling from above the nel temperature of the af shell in a field @xmath9 are shown in fig .
[ fig_1 ] ( circles)for three values of @xmath5 . as can be clearly seen in the upper panels of the figure
, the loops display an increasing shift towards the negative field direction with increasing values of @xmath5 , demonstrating that the observance of eb is related to the af coupling of the shell spins the core spins at the interface . a detailed inspection of the microscopic configurations attained after the fc process @xcite revealed that , inspite of the af character of the coupling between the shell spins and also between the interfacial spins , a net magnetization component along the field direction is induced at the shell interface due to the geometrical symmetry breaking and the alignment of groups of spins into the field direction .
this net magnetic moment generates local fields on the core spins that point into the same direction as the external field , causing the shift of the hysteresis loops .
this is further supported by the observation that the hysteresis loops are shifted by the same amount but towards the positive field axis when obtained after cooling in a field applied in a direction negative with respect to the measuring field ( see for example the dashed lines in fig [ fig_1 ] . for @xmath10 ) .
( color online ) upper panels display the hysteresis loops for a particle with radius @xmath11 obtained after fc down to @xmath12 in a field @xmath9 ( circles ) and @xmath13 ( dashed lines ) for three values of the exchange coupling constant @xmath14 at the core / shell interface .
lower panels show the average magnetization projection of the core spins along the field axis @xmath15 ( squares ) and the hysteresis loops for the component of the magnetization transverse to the field direction @xmath16 ( circles ) . ] in addition , a clear asymmetry between the upper and lower loop branches developes with increasing values of the interface coupling .
this feature can be more clearly seen by looking at the average absolute value of the magnetization projection along the field axis through the reversal process , @xmath17 displayed in the lower panels of fig .
[ fig_1 ] for the core spins .
the origin of the loop asymmetry can be traced by monitoring the values the so - called overlap @xmath18 and link overlap @xmath19 functions along the hysteresis loops , that are a generalization of similar quantities commonly used in the spin - glass literature @xcite and that are defined as @xmath20 and @xmath21 , where in @xmath19 the summation is over nearest neighbors and @xmath22 is a normalization factor that counts the number of bonds .
an example of the field dependence of these overlaps computed only for the interfacial spins is shown in fig .
[ fig_2 ] , where we have separated the contribution of the shell and core spins . a departure of @xmath23 from unity
is known to be proportional to the surface of reversed domains formed at field @xmath24 and , therefore , @xmath23 is sensible to the existence of domain walls .
the sharp decrease of @xmath23 for core spins and the symmetry of the peak around the negative coercive field indicates uniform reversal .
however , the progressive reduction of @xmath23 along the ascending branch and the asymmetry of the peak around the positive coercive field indicates the formation of domain walls at the particle core that sweep the particle during reversal .
the function @xmath25 measures differences of the spin configuration at field @xmath24 with respect to the one attained after fc .
therefore , the decrease of @xmath26 for the interface shell spins when reducing the magnetic field indicates the existence of a fraction of shell spins that reverse dragged by core spins , while the constancy of @xmath26 in the ascending branch reveals the existence of spins pinned during core reversal .
clearly correlated to the observation of eb and the loop asymmetry , the loops experience a shift along the vertical ( @xmath29 ) axis which increases with @xmath14 , as reflected in fig .
[ fig_1 ] by the difference of the @xmath29 values in the high field region of the two loop branches or at the remanence points . the field dependence of magnetization transverse to the field direction @xmath16 ( circles in the lower panels of fig .
[ fig_1 ] ) , indicates that @xmath16 has higher values for the descending loop branch that in the ascending branch , and that @xmath16 increases with @xmath14 .
snapshots of the spin configurations at the remanence points shown in fig .
[ fig_3 ] , show the existence of a higher amount of core spins with transverse orientation near the interface at the lower branch ( fig .
[ fig_3]b , d ) than at the upper branch .
this is in agreementwith the results of some experiments of oxidized particles where this vertical shift was also observed @xcite .
our simulation results above , indicate that the microscopic origin of the vertical shift is the due to the different reversal mechanisms on the two loop branches due to the existence of uncompensated pinned moments at the core / shell interface that faciltate the nucleation of non - uniform magnetic structures during the ascending field branch of the loops . we acknowledge h. katzgraber for helpful comments on overlap functions and cesca and cepba under coordination of c@xmath30 for computer facilities . this work has been supported by the spanish meyc through the mat2003 - 01124 project and the generalitat de catalunya through the 2005sgr00969 cirit project .
h. ohldag _ et al .
87 ( 2001 ) 247201 ; h. ohldag _ et al .
91 ( 2003 ) 017203 ; p. kappenberger _
et al . _ , _ ibid . _ 91 ( 2003 ) 267202 ; p. blomqvist _
_ , _ ibid .
_ 94 ( 2005 ) 107203 ; s. roy _ et al .
_ , _ ibid .
_ 95 ( 2005 ) 047201 ; v. k. valev _
_ , _ ibid .
_ 96 ( 2006 ) 067206 ; e. arenholz
_ et al . _ ,
88 ( 2006 ) 072503
. r. k. zheng _
et al . _ ,
b 69 ( 2004 ) 214431 ; l. del bianco _ et al .
_ , _ ibid .
_ 70 ( 2004 ) 052401 ; j. b. tracy _
et al . _ , _ ibid . _ 72 ( 2005 ) 064404 ; e. passamani _ et al . _ , j. magn .
magn . mat .
299 ( 2006 ) 11 . | we present monte carlo simulations of hysteresis loops of a model of a magnetic nanoparticle with a ferromagnetic core and an antiferromegnetic shell with varying values of the core / shell interface exchange coupling which aim to clarify the microscopic origin of exchange bias observed experimentally .
we have found loops shifts in the field direction as well as displacements along the magnetization axis that increase in magnitude when increasing the interfacial exchange coupling .
ovelap functions computed from the spin configurations along the loops have been computed to explain the origin and magnitude of these features microscopically . , magnetic nanoparticles ; exchange bias ; monte carlo simulation ; magnetic hysteresis 75.60.-d ; 05.10.ln ; 75.50.tt ; 75.60.jk |
the averaged quantities can be obtained in two different ways in magnetohydrodynamics .
the first way is to solve 3d mhd equations and then average the results .
the second way is to solve some system of equations on averages .
combination of numerical simulations and averaged theory brings phenomenology that can describe observations or experimental data .
the problem of spherically symmetric accretion takes its origin from bondi s work @xcite .
he presented idealized hydrodynamic solution with accretion rate @xmath0 however , magnetic field @xmath1 always exists in the real systems .
even small seed @xmath1 amplifies in spherical infall and becomes dynamically important @xcite .
magnetic field inhibits accretion @xcite .
none of many theories has reasonably calculated the magnetic field evolution and how it influences dynamics .
these theories have some common pitfalls .
first of all , the direction of magnetic field is usually defined .
secondly , the magnetic field strength is prescribed by thermal equipartition assumption . in third , dynamical effect of magnetic field
is calculated with conventional magnetic energy and pressure .
all these inaccuracies can be eliminated . in section 2[section_method ] i develop a model that abandons equipartition prescription , calculates the magnetic field direction and strength and employs the correct equations of magnetized fluid dynamics . in section 3[results ] i show this accretion pattern to be in qualitative agreement with sgr a * spectrum models .
i discuss my assumptions in section 4 [ discussion ] .
reasonable turbulence evolution model is the key difference of my method .
i build an averaged turbulence theory that corresponds to numerical simulations .
i start with the model of isotropic turbulence that is consistent with simulations of collisional mhd in three regimes .
those regimes are decaying hydrodynamic turbulence , decaying mhd turbulence and dynamo action .
i introduce effective isotropization of magnetic field in 3d model .
isotropization is taken to have a timescale of the order of dissipation timescale that is a fraction @xmath2 of the alfven wave crossing time @xmath3 common misconception exists about the dynamical influence of magnetic field .
neither magnetic energy nor magnetic pressure can represent @xmath1 in dynamics .
correct averaged euler and energy equations were derived in @xcite for radial magnetic field .
magnetic force @xmath4 $ ] can be averaged over the solid angle with proper combination of @xmath5 i extend the derivation to random magnetic field without preferred direction .
dynamical effect of magnetic helicity @xcite is also investigated .
i neglect radiative and mechanical transport processes .
the derived set of equations requires some modifications and boundary conditions to be applicable to the real astrophysical systems .
i add external energy input to turbulence to balance dissipative processes in the outer flow .
the outer turbulence is taken to be isotropic and has magnetization @xmath6 transonic smooth solution is chosen as possessing the highest accretion rate as in @xcite .
[ fig1 ] , title="fig : " ]
[ fig2 ] with radius.,title="fig : " ] the results of my calculations confirm some known facts about spherical magnetized accretion , agree with the results of numerical simulations and have some previously unidentified features .
initially isotropic magnetic field exhibits strong anisotropy with larger radial field @xmath7 perpendicular magnetic field @xmath8 is dynamically unimportant in the inner accretion region fig[fig1 ] . because magnetic field dissipates , infall onto the black hole can proceed @xcite .
turbulence is supported by external driving in the outer flow regions , but internal driving due to freezing - in amplification takes over in the inner flow fig[fig2 ] .
magnetization of the flow increases in the inner region with decreasing radius consistently with simulations @xcite .
density profile appears to be @xmath9 that is different from traditional adaf scaling @xmath10 @xcite .
thus the idea of self - similar behavior is not supported .
compared to non - magnetized accretion , infall rate is 2 - 5 times smaller depending on outer magnetization . in turn ,
gas density is 2 - 5 times smaller in the region close to the black hole , where synchrotron radiation emerges @xcite .
sgr a * produces relatively weak synchrotron @xcite .
so , either gas density @xmath11 or electron temperature @xmath12 or magnetic field @xmath13 are small in the inner flow or combination of factors works .
thus low gas density in magnetized model is in qualitative agreement with the results of modelling the spectrum .
flow is convectively stable on average in the model of moving blobs , where dissipation heat is released homogeneously in volume .
moving blobs are in radial and perpendicular pressure equilibriums .
they are governed by the same equations as the medium .
the presented accretion study self - consistently treats turbulence in the averaged model .
this model introduces many weak assumptions instead of few strong ones .
i take dissipation rate to be that of collisional mhd simulations .
but flow in question is rather in collisionless regime .
observations of collisionless flares in solar corona @xcite gives dissipation rate @xmath14 times smaller than in collisional simulations @xcite .
however , flares in solar corona may represent a large - scale reconnection event rather than developed turbulence .
it is unclear which dissipation rate is more realistic for accretion .
magnetic field presents another caveat .
magnetic field lines should close , or @xmath15 should hold .
radial field is much larger than perpendicular in the inner region .
therefore , characteristic radial scale of the flow is much larger than perpendicular .
if radial turbulence scale is larger than radius , freezing - in condition does not hold anymore .
matter can freely slip along radial field lines into the black hole .
if matter slips already at the sonic point , the accretion rate should be higher than calculated .
some other assumptions are more likely to be valid .
diffusion should be weak because of high mach number that approaches unity at large radius .
magnetic helicity was found to play very small dynamical role .
only when the initial turbulence is highly helical , magnetic helicity conservation may lead to smaller accretion rate .
neglect of radiative cooling is justified a posteriori .
line cooling time is about @xmath14 times larger that inflow time from outer boundary .
9 d. biskamp , _ magnetohydrodynamic turbulence _ , cambridge university press , oxford , 2003 h. bondi , _ mon . not .
soc . _ * 112 * , 195 ( 1952 ) .
i. igumenshchev , _ astrophys .
j. _ * 649 * , 361 ( 2006 ) l. d. landau , e. m. lifshitz , l. p. pitaevskii , _ electrodynamics of continuous media _ , pergamon press , oxford , 1984 j. b. noglik , r. w. walsh , j. ireland , _ astron . &
_ * 441 * , 353 ( 2005 ) r. narayan , i. yi , r. mahadevan , _ nature _ * 374 * , 623 ( 1995 ) e. t. scharlemann , _ astrophys . j. _ * 272 * , 279 ( 1983 ) a. a. schekochihin et al . _
j. _ * 276 * , 612 ( 2004 ) v. f. schwartzman , _
soviet astronomer _ * 15 * , 377 ( 1971 ) k. r. sreenivasan , _ phys . of fluids
_ * 7 * , 2778 ( 1995 ) | transonic accretion flow with self - consistent treatment of the magnetic field is presented .
my website http://www.cfa.harvard.edu/~rshcherb/. address = harvard university , astronomy department , 60 , garden st , cambridge , usa |
bilayer quantum hall systems at total landau level filling factor @xmath0 @xcite allow for novel interlayer coherent phases .
these phases have attracted a great deal of theoretical and experimental attention @xcite over the last sixteen years , dating back to a seminal paper by halperin @xcite , in which the multicomponent generalization of the laughlin wavefunction was first considered in a rather general context . in particular , there is strong experimental evidence @xcite and a compelling theoretical basis @xcite to believe that a spin - polarized bilayer @xmath5 quantum hall system would have a novel spontaneous interlayer coherent incompressible phase for small values of the interlayer separation @xmath2 , even in the absence of interlayer tunneling .
( we consider only the situation without any interlayer tunneling in this paper our considerations also apply to the physical situation with weak interlayer tunneling .
the situatiion with strong interlayer tunneling is trivial by virtue of the tunneling - induced symmetric - antisymmetric single - particle tunneling gap which leads to the usual @xmath0 quantum hall state in the symmetric band . ) in the limit @xmath6 , however , one expects two decoupled layers , each with @xmath5 , and hence no quantized hall state .
the phase diagram for this incompressible to incompressible quantum phase transition in @xmath5 bilayer systems has been studied extensively in the literature @xcite , but we still do not have a complete qualitative understanding of the detailed nature of this transition .
in particular , one does not know how different kinds of incompressible ( and compressible ) phases compete as system parameters ( e.g. @xmath2 ) are tuned , and how to distinguish among possible competing incompressible phases . in this paper
, we revisit this issue by arguing that , in principle , there are several interesting and nontrivial quantum hall phases in the @xmath0 bilayer system which could be systematically probed via interlayer drag experiments carried out at various values of the interlayer separation @xmath2 . among the more interesting quantum hall phases
are the so - called paired hall states , which have been extensively studied theoretically @xcite . in these states ,
the composite fermions form an incompressible superconducting paired state in which two electrons composite fermions bind into effective cooper pairs which then condense into an incompressible ground state analogous to the bcs state .
the well - studied moore - read pfaffian state @xcite is a spin - polarized version of such a paired hall state for a single layer system .
bilayer paired hall states have been discussed earlier in the literature in the context of @xmath5 ( and @xmath7 systems ) , but no definitive idea has emerged regarding their experimental observability or their relation to the more intensively studied ( and robust ) @xmath4 state @xcite .
the main purpose of the current paper is to critically discuss the possible existence of a paired @xmath5 bilayer hall state which , we argue , is distinguishable from the better - studied @xmath4 incompressible state ( as well as from compressible states ) through interlayer drag experiments .
given the great current interest @xcite in the phyiscs of @xmath5 bilayer systems , we believe that the results presented in this paper could shed considerable light on the nature of the possible quantum phase transitions in bilayer systems . in sections ii , iii , and iv of this paper , we consider the possible bilayer @xmath5 quantum hall phases in the parameter regimes @xmath8 ( section ii ) , @xmath9 ( section iii ) , and @xmath10 ( section iv ) , where @xmath11 is the magnetic length , which sets the scale for intralayer correlations .
we argue that the likely ground states in these three regimes are , respectively , compressible ( fermi - liquid - like ) states ( @xmath8 ) , paired hall states ( @xmath9 ) , and @xmath4 states ( @xmath10 ) .
we conclude in section v with a critical discussion of the various drag resistivities which we argue can , in principle , distinguish among these phases and could be used to study bilayer quantum phase transitions .
let us consider a bi - layer system in which each layer has filling factor @xmath13 in the limit that the layer separation @xmath2 is much larger than the typical interparticle spacing , which is of the order of the magnetic length @xmath14 . as a starting point , we will model the system by two almost independent fermi - liquid - like compressible states , one in each layer .
there are two alternative and complementary descriptions of compressible states at even - denominator filling fractions .
we will briefly recapitulate some features of both , as they will inform the following discussion of paired states .
one description of a fermi - liquid - like compressible state at @xmath13 is based on the lowest landau level wavefunction @xcite : @xmath15 where @xmath16 has the matrix elements @xmath17 . here , @xmath18 is the position of electron @xmath19 .
the @xmath20s are parameters which are chosen so that the total energy of the system is minimized .
we will discuss their interpretation below .
@xmath21 projects states into the lowest landau level ; it has the following action @xcite @xmath22 where @xmath23 are the guiding center coordinates of the electrons .
the corresponding wavefunction of the double - layer system at @xmath0 can be written as @xmath24 where @xmath25 label the two layers .
the lowest landau level constraint and fermi statistics displace the electrons from the correlation holes
i.e. zeroes of the wavefunction or , equivalently , vortices represented by the @xmath26 factor in the wavefunction .
the @xmath27 factor is necessary to ensure fermi statistics ; it is a displacement operator @xcite because @xmath28 acts as @xmath29 in the lowest landau level .
thus the composite fermion made of an electron and two correlation holes has a dipolar structure . using these ideas ,
the system can be described as a collection of dipolar ` composite fermions ' in which each dipolar fermion consists of an electron and the corresponding correlation holes @xcite . to each composite fermion
we assign a @xmath20 which is equal in magnitude ( in units of @xmath14 ) and perpendicular to the displacement between the electron and its correlation holes . by fermi statistics ,
the @xmath20s must be distinct .
the structure of a dipolar composite fermion is given schematically in fig.1 .
the energy of a composite fermion increases with @xmath20 , so the ground state is a filled fermi sea in @xmath30-space . in the long wavelength limit
, the total energy of these dipolar composite fermions can be approximately written as @xmath31 with effective mass @xmath32 which is determined by the interaction potential @xcite .
an alternative formulation springs from the observation that an electron may be represented by a ` composite fermion ' together with a chern - simons gauge field which attaches two fictitious flux quanta to each fermion @xcite .
this representation is mathematically equivalent to the original one , but it naturally suggests a new approximation .
the composite fermions see zero average magnetic field due to the cancellation between the external magnetic field and the average fictitious magnetic field coming from the fictitious flux quanta .
consequently , the system can be described as an almost free ( composite ) fermion system in zero effective magnetic field . in this approach ,
the fictitious flux quanta are introduced to represent the phase winding of the electron wavefunction around correlation holes associated with the positions of other electrons . the @xmath33 $ ] factor in ( [ eqn : rr_wavefcn ] )
can be interpreted as the wavefunction of the almost free composite fermion system .
@xcite then @xmath20s can be regarded as the `` kinetic momenta '' of the composite fermions . in the long - wavelength , low - frequency limit ,
the two formulations are equivalent .
we will now implement the latter formulation in a double - layer system , and show that the compressible state has a strong pairing instability @xcite .
we introduce two composite fermion fields @xmath35 and two chern - simons gauge fields @xmath36 , @xmath37 .
the hamiltonian is @xmath38 with the constraints @xmath39 . @xmath40
if the filling factor of each layer is @xmath13 .
here @xmath41 is the density disturbance measured from the average value @xmath42 .
the interaction potential is given by @xmath43 and @xmath44 where @xmath45 is the dielectric constant .
it is convenient to change the gauge field variables to @xmath46 .
then @xmath47 can be rewritten as @xmath48\end{aligned}\ ] ] with @xmath49 $ ] . from ( [ eqn : h_0 ] )
, we see that @xmath50 and @xmath51 have the same gauge `` charges '' for @xmath52 , but opposite gauge `` charges '' for @xmath53 . as a result
, there will be an attractive interaction between the composite fermions in different layers via @xmath54 and a repulsive interaction via @xmath55 .
composite fermions in the same layer have repulsive interactions . as a result of coulomb intractions ,
the attractive interaction mediated by the @xmath54 gauge field dominates in the low energy limit and there exists a pairing instability between the composite fermions in different layers .
this result can be understood in physical terms as follows .
the @xmath54 and @xmath55 fields represent antisymmetric and symmetric density fluctuations . in the presence of coulomb interactions , symmetric density fluctuations
are highly suppressed , but antisymmetric density fluctuations can still be large . as a result , the dynamic density fluctuations in the antisymmetric channel becomes more important in the low energy limit and lead to a pairing instability .
this pairing instability has a natural explanation in the dipolar composite fermion picture @xcite .
let us take a dipolar composite fermion in layer @xmath56 with wavevector @xmath57 and a dipolar composite fermion in layer @xmath58 with wavevector @xmath59 . as seen in fig.2 , this configuration can lower the interlayer coulomb energy because the electron in layer @xmath56 and the vortex in layer @xmath58 can sit on top of each other and vice versa .
= 1.4 in this analysis predicts that , at least in principle , the fermi - liquid - like compressible state is always unstable to pairing . in practice , the pairing gap will be small in the limit @xmath60 and easily destroyed by disorder . as a result , we expect the pairing instability discussed above to be relevant for @xmath61 not too small . when @xmath61 become small , on the other hand , the starting point of two decoupled compressible states is no longer sensible , and we take a different starting point , as described in later sections . the wavefunction of the paired quantum hall state constructed in this way can be written as @xmath62 where @xmath63 \cr & \equiv & { \cal a } [ f(z_1 , z_2 ; \uparrow , \downarrow ) f(z_3 , z_4 ; \uparrow , \downarrow ) \cdots ] \ .
\end{aligned}\ ] ] and@xmath64 is the pair wavefunction which depends on the symmetry of the pairing order parameter .
$ ] denotes the pfaffian , which is defined in the second line , with @xmath66 $ ] denoting the antisymmetrized product .
notice that @xmath67 can be regarded as the product of the wavefunction of the paired composite fermions and that of the ( 2,2,0 ) bosonic laughlin quantum hall state .
it is not immediately clear what choice of @xmath64 is most favorable energetically . in the chern - simons theory of @xcite ,
there is a pairing instability in all angular momentum channels @xcite . in a modified chern - simons theory ,
it was claimed that the leading instability occurs in the @xmath68-wave channel @xcite .
these approximate calculations do not necessairly capture the detailed energetics which determines the pairing symmetry .
hence , we will not enter into a discussion of energetics , but limit ourselves to a discussion of the simplest ( and , therefore , likeliest ) possibilities .
the simplest possibility is @xmath69 pairing , @xmath70 \prod_{i > j } ( z^{\uparrow}_i - z^{\uparrow}_j)^2 \prod_{k > l } ( z^{\downarrow}_k - z^{\downarrow}_l)^2 \ .\ ] ] using the cauchy identity @xcite , @xmath71 this can be rewritten as @xmath72 thus @xmath67 is the @xmath73 state if one takes the @xmath74 pairing .
this wavefunction is well - behaved in the long distance limit , but has a short distance singularity . in the presence of landau - level mixing ,
the short distance part of the wavefunction can be modified without changing the structure of the wavefunction in the long distance limit .
@xmath75\,\times \cr & & { \hskip 1 cm } \prod_{i > j } ( z^{\uparrow}_i - z^{\uparrow}_j)^2 \prod_{k >
l } ( z^{\downarrow}_k - z^{\downarrow}_l)^2 \ .\end{aligned}\ ] ] here , @xmath76 and @xmath77 as @xmath78 . in realistic systems , where landau - level mixing is substantial
, @xmath79 could be a good candidate for the paired quantum hall state represented by @xmath67 .
it is natural to assume that @xmath67 does not have an `` interlayer josephson effect '' because there is no gapless neutral mode in the system , in contrast to the case of the @xmath4 state @xcite .
we will show this later by direct calculation .
another possibility for the pair wavefunction @xmath80 is the exponentially decaying function @xmath81 with a correlation length @xmath82 .
this would correspond to a `` strong '' pairing state while the previous choice of @xmath83 corresponds to a `` weak '' pairing state in the terminology of of read and green @xcite .
finally , the pairing wavefunction could have @xmath84-wave symmetry , which would entail a pairing form similar to ( [ eqn : other_pairing ] ) , but without the @xmath85 factor .
when the layer separation becomes sufficiently small , the interlayer coulomb interaction can be comparable to or even larger than the intralayer coulomb interaction . in this case
, it should be more advantageous to first form an `` interlayer '' dipolar object which consists of an electron in one layer and two vortices in the other layer , then form a paired state of these `` interlayer '' composite fermions , as shown in fig.3 . in the chern - simons formulation , this corresponds to the situation in which the electron in one layer can only see fictitious flux in the other layer .
the appropriate chern - simons constraint equation are : @xmath87 as in the previous case , we can form symmetric and antisymmetric combinations of the gauge fields @xmath88 and @xmath89 . again
, the antisymmetric combination mediates an attractive interaction .
the wavefunction of the corresponding paired quantum hall state has the following form .
@xmath90 where @xmath91\ ] ] and @xmath92 is the appropriate pair wavefunction .
notice that the wavefunction @xmath93 can be regarded as the product of the wavefunction of a paired state of the composite fermions and that of the ( 0,0,2 ) bosonic laughlin quantum hall state .
= 1.4 in however , this line of thinking appears to conflict with the conventional wisdom that a bi - layer quantum hall system at @xmath94 is described by the @xmath4 state for @xmath95 @xcite .
fortunately , the @xmath4 state , @xmath96 can be rewritten in the form @xmath97 \prod^{n}_{i , j=1 } ( z^{\uparrow}_i - z^{\downarrow}_j)^2 \ .\ ] ] using the cauchy identity . in other words
, @xmath98 is the correct choice for @xmath95 .
notice that this form of @xmath99 corresponds to the ( pseudo-)spin triplet @xmath69 pairing order parameter for the `` interlayer '' composite fermions . from this point of view , it is natural to have the same pairing symmetry ( [ eqn:(3,3,-1 ) ] ) for @xmath100 but for `` intralayer '' rather than `` interlayer '' composite fermions .
in the preceeding sections , we have described three possible states of a bi - layer system at @xmath94 . as one decreases the ratio of @xmath61 , we expect the following sequence of phase transitions @xmath101 the transition between @xmath102 and @xmath79 is disorder - driven since @xmath102 should never be the ground state in the presence of disorder .
hence , we expect the transition to be second - order . on the other hand ,
the transition between @xmath79 and @xmath103 is likely to be first order because the nature of the underlying quasiparticles changes from the `` intralayer '' composite fermions to the `` interlayer '' composite fermions and this change is not likely to be continuous .
more formally , the change of the wavefunction requires a flux attachment transformation without changing the total filling factor which does not usually occur continuously . of course , disorder will eventually round the transition and make it second order .
since the first order transition may be obscured by disorder , it is important to discuss experiments which can distinguish the @xmath79 and @xmath103 states . here , we propose coulomb drag experiments in which the longitudinal , @xmath104 and hall , @xmath105 , drag resistivities are used to distinguish the different phases . these may be calculated in chern - simons theory from @xmath106 where @xmath107 , @xmath108 , @xmath109 is the antisymmetric tensor , and @xmath110 in the compressible and @xmath73 states ( `` intralayer '' composite fermions ) while @xmath111 in the @xmath4 state ( `` interlayer '' composite fermions ) .
first , consider the longitudinal drag resistivity . in the compressible state , if we neglect gauge field fluctuations , @xmath112 , so @xmath113 . including these fluctuations
, it vanishes as @xmath114 at low temperatures @xcite . in @xmath103 and @xmath79 , @xmath115 and ,
hence , @xmath116 vanish at zero temperature and are activated at low temperatures .
now , let us consider the hall drag resistivity . in the compressible state , both terms on the right - hand - side of ( [ eqn : drag ] ) vanish , so the hall drag resistivity vanishes . in the @xmath73 state
, @xmath117 is that of a @xmath69 superconductor , which has vanishing charge resistivity ( since it is a superconductor ) but quantized spin hall resistivity . in other words , @xmath118 , @xmath119 , @xmath120 .
consequently , @xmath121 , and @xmath122 . in the @xmath4 state ,
@xmath117 is identical , but @xmath123 is different , so @xmath124 , and @xmath125 .
we conclude by summarizing our results .
we have shown that the @xmath5 ( @xmath0 ) bilayer quantum hall system ( in the absence of interlayer tunneling ) is likely to have as its ground state a novel paired hall state ( possibly of @xmath68-wave symmetry ) for intermediate layer separations @xmath126 , which gives way to the usual @xmath4 state for smaller layer separations ( @xmath10 ) , and to compressible fermi - liquid - type states ( two decoupled halperin - lee - read @xcite @xmath5 layers ) for large layer separations ( @xmath127 ) .
we argue that the quantum phase transitions separating the paired states from the @xmath4 and bilayer halperin - lee - read states can be experimentally studied via the measurement of various components of interlayer drag resistivities .
we would like to thank l. balents and , especially , n. read for discussion .
ybk , cn , and sds would like to thank the aspen center for physics for hospitality .
this work was supported by the nsf under grant numbers dmr-9983783 ( ybk ) and dmr-9983544 ( cn ) , the a.p .
sloan foundation ( ybk and cn ) , the harvard society of fellows ( ed ) , and the onr ( sds ) .
see , for example , the articles by j. p. eisenstein , s. m. girvin and a. h. macdonald in _ perspectives in quantum hall effects _ , editied by s. das sarma and a. pinczuk ( wiley , new york , 1997 ) ; and references therein .
n. e. bonesteel , phys .
b * 48 * , 11484 ( 1993 ) ; n. e. bonesteel , i. a. macdonald , and c. nayak , phys . rev . lett . * 77 * , 3009 ( 1996 ) ; m. greiter , f. wilczek , and x .-
wen , phys .
lett . * 66 * , 3205 ( 1991 ) . | we consider three states which are likely to be realized in bi - layer quantum hall systems at total landau level filling fraction @xmath0 .
two of these states may be understood as paired states .
one can occur as an instability of a compressible state in the large @xmath1 limit , where @xmath2 and @xmath3 are the interlayer distance and magnetic length .
the other is the @xmath4 state , which is expected in the small @xmath1 limit .
we propose that the longitudinal and hall drag resistivities can be used to distinguish these different quantum hall states .
the drag resistivities may be computed in a straightforward manner using the paired state formulation . 2 |
dynamical friction ( hereafter , df ) is an important physical process for a wide variety of dynamics , e.g. stellar dynamics , galaxy and planetary formation etc .
that s why it is in great demand to seek deeper understanding of the nature of df . as a historical milestone ,
@xcite has succeeded in formulation of the effect of df . however , several approximations were unavoidable in the derivation of the formula .
the chandrasekhar formula was derived under the assumptions of uniform density and isotropic velocity field .
moreover , interaction between a perturber and field particles was supposed to be integration of countless two - body interactions and orbital periodicity of all particles was not taken into account . by efforts of many numerical studies , current astrophysicists have realized that the chandrasekhar formula is not perfect and loses accuracy of the estimation in some cases ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in this paper , i concentrate on the suppression of df on a globular cluster ( hereafter , gc ) in a cored halo of dwarf galaxies @xcite .
however , the discussion in this paper can be generalized to other astronomical bodies by scaling physical units .
the drag force by df is negligibly weak for gcs in large systems , like our galaxy .
in contrast , in small systems like dwarf galaxies , the drag force is far too strong ( see * ? ? ? * ) .
thus , the gcs in dwarfs are presumed to lose their orbital energy and fall into the galactic centre by strong friction force in time - scale of the order of @xmath0 gyr @xcite .
nevertheless , these gcs still do exist in many dwarf galaxies even now and are as old as the age of universe ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this problem is referred to as ` _ _ the dynamical friction problem _ _ ' .
however , a solution of the problem has been proposed .
if dwarf galaxies have a cored dark matter halo which has a constant density region in its centre , the df against the halo is weakened considerably , and gcs should be able to survive beyond the age of the universe . by an analytical approach using the chandrasekhar formula
, @xcite discovered that a king - model halo can weaken the df in the core region .
recently , by n - body simulations , @xcite , @xcite and @xcite confirmed complete cessation of df in a core region of haloes .
thus , although the chandrasekhar formula indicates gradual shrinkage of the gc orbit , the n - body simulations demonstrate complete defunctionalization of df .
this discrepancy between analytic and numerical studies ( the chandrasekhar formula and n - body simulations ) still remains . besides , the cessation of df is a topic not restricted to dwarf galaxies and gcs .
@xcite has discussed the same core - stalling of df on binary black hole against a cored stellar cluster and suggested that fluctuation of the halo potential weakens the df in the core .
my goals in this paper are to clarify hidden physics in the nature of df and the mechanism of the df suppression in cored density profiles by n - body simulations .
constant density distribution is called , in other words , ` harmonic oscillator potential ' , which means that all particles in the potential have the same orbital period . in this sense ,
harmonic potentials are expected to induce complex effects in non - linear manner and be beyond the applicable domain of the chandrasekhar formula .
it has been known for a long time that df shows curious behaviors in constant densities .
@xcite analytically argued that a constant density core causes energy feedback on a perturber by dynamical resonance in positive or negative sense depending on the density distribution . in the same context as @xcite
, @xcite indicated that density response of the density field to a perturber orbiting in the core region can exert torque on the perturber and drive the orbital motion .
moreover , @xcite also analytically argued that in the harmonic potentials the perturber acts as a catalyst for the evolution of the energy distribution of the field particles by the orbital resonance . by the way , the cored dark haloes are inconsistent with results of cosmological n - body simulations ( e.g. * ? ? ?
* ; * ? ? ?
* ) . however , among small scale galaxies , many observations of low surface brightness ( lsb ) galaxies have confirmed the cored dark haloes ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and refernces therein ) , and nearby dwarf galaxies also favor the cored haloes @xcite .
in addition , numerical simulations including baryonic physics also predict core creation by bursty supernovae in the centre of dwarf haloes in cosmological context @xcite and isolated models @xcite . this paper is organized as follows : i will explain my simulation method and present my simulation results in 2 . in
3 , i will show analyses of the data generated by my simulations . in
4 , i will summarize my results and present discussion .
my numerical simulation is a typical pure n - body simulation ( no gas component ) .
i use barnes - hut modified tree - code @xcite , setting an open angle of @xmath1 .
a special - purpose calculator for collisionless n - body simulations , grape-7 , is used with the tree algorithm @xcite .
the leapfrog time - integrator is adopted with the shared - timestep method .
the total number of timesteps is 11841 for the whole of a simulated period which corresponds to 10 gyrs in the real time - scale .
in this paper , i use burkert profile @xcite as a cored halo model of dwarf galaxies , which is motivated from observations of lsb galaxies @xcite and numerical simulations of formation of dwarf galaxies @xcite , @xmath2 @xmath3 is the effective core density , @xmath4 is the core radius which defines the constant density region . in my simulation , i set @xmath5 , @xmath6 kpc , and the virial mass of the halo @xmath7 , the virial radius @xmath8 kpc .
eq.[burkert ] estimates the mass enclosed inside 300 pc to be @xmath9 .
this value is consistent with the result of @xcite which has suggested that all dwarfs have the same mass scale , @xmath10 .
i add an exponentially decaying envelope to prevent instability at the outer region caused by the artificial cut - off radius , @xmath11 @xcite , @xmath12 where @xmath13 and @xmath14 .
i set @xmath15 . with this envelope ,
total halo mass enclosed in @xmath16 pc becomes @xmath17 .
velocity dispersion is given by the solution of jeans equations , @xmath18 where @xmath19 is the anisotropy parameter .
in this paper , i assume the isotropic velocity state in the halo , setting @xmath20 ( @xmath21 ) .
the velocity distribution is determined by local maxwellian approximation , @xmath22 where @xmath23 is a probability distribution function of velocities @xcite . by the assumptions of isotropic and maxwellian velocity distribution ,
i satisfy the suppositions underlying the derivation of the chandrasekhar formula , although @xcite has reported slight instability in the burkert profile realized by the local maxwellian approximation .
i conduct n - body simulations by @xmath24 particles to resolve the halo .
all halo particles have the same mass , @xmath25 , with a softening length of @xmath26 pc . before operating actual simulations
, the halo model needs to be relaxed to some extent .
i run a simulation without gcs for 2 gyrs . after the relaxation ,
i employ the relaxed halo as the initial state ( @xmath27 ) of the halo . in my simulation ,
the gc is represented by a point mass with @xmath28 .
the softening length of the gc particle is set to @xmath29 pc .
initial orbital radius of the gc is set to 750 pc .
the initial orbit is circular , and i employ a single gc in my simulation .
multi - gc cases are also interesting , but have been investigated by @xcite and @xcite , which did nt indicate any large differences from the single gc cases .
if the gc is resolved by many particles , some additional effects may play important roles : mass - loss , dynamical heating , etc ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
besides , stellar components in a dwarf galaxy can enhance the df of gcs @xcite . however ,
in this paper , i concentrate on investigations of fundamental physics of df , and do not consider such effects .
.,width=317 ] the time - evolution of orbital radius of the gc is plotted in fig.[fig1 ] . as a comparison
, i also plot an analytic calculation by the chandrasekhar formula , @xmath30 , \label{chandra}\ ] ] @xmath31 where @xmath32 .
@xmath33 is circular velocity of the gc as a function of radius .
@xmath34 is the gravitational constant . for the coulomb logarithm
, i set @xmath35 , which is determined by the best - fitting to the result of simulation for @xmath36 gyr . if i take the original definition , @xmath37 , where @xmath38 are the maximum and the minimum impact parameter . in the derivation of eq.[drdt ] , the assumption of circular orbit is used @xcite . as seen from fig.[fig1 ] , in the n - body simulation , the orbital shrinkage stops after the gc entered into the constant density region ( @xmath39 ) . on the other hand , the chandrasekhar formula fails to reproduce the n - body result in the core region , in spite of the excellent fit outside the core . in this paper ,
i define @xmath40 gyr as ` _ _ exerting df phase _ _ ' , and @xmath41 gyr as ` _ _ suppressed df phase _ _ ' . and
[ envelope ] ) and in the n - body simulation for t=0 and 10 gyr.,width=317 ] the density profiles of the halo are plotted in fig.[fig2 ] . even in the end - state of the simulation ,
the density profile is stable .
the energy conservation rate of the system , @xmath42 , is @xmath43 .
in the n - body simulation , for @xmath44 gyr , df is clearly exerted on the gc , leading to orbital shrinkage .
the chandrasekhar formula with @xmath35 succeeds in estimation of the effect in this phase .
firstly , i calculate dynamical energies of all particles .
the specific energy of the @xmath45th particle is @xmath46 where @xmath47 mean position vectors of the @xmath48th field particles and the gc , respectively
. diagram .
the ordinate indicates an amount of increase of dynamical energy of each particle between 2 and 3 gyr : @xmath49 .
the abscissa indicates energy before the change : @xmath50 .
the values are normalized by @xmath51 which is absolute value of the potential energy at @xmath52 analytically estimated .
colour - bar labels mass in a bin , @xmath53 .
black colour means no mass . in this diagram , i separate the particles into three groups ; positive - horn ( the red square ) , negative - horn ( the green square ) , the others.,width=317 ] next , i calculate increase of @xmath54 at intervals of 1 gyrs . in fig.[fig3 ] , i show a distribution of all field particles in @xmath55-@xmath56 diagram .
@xmath56 corresponds to the increase of @xmath54 between 2 and 3 gyr . in this figure ,
i can find a notable feature . at a certain energy , @xmath57
, there are some particles of which energies changed largely .
this feature looks like two _ horns_. thus , i define @xmath58 and @xmath59 ( inside the red square ) as _ positive - horn _ ( p - horn ) , @xmath58 and @xmath60 ( inside the green square ) as _ negative - horn _
( n - horn ) .
the numbers of p- and n - horn particles in this plot are 1,513 ( @xmath61 in total ) and 2,028 ( @xmath62 in total ) , respectively .
i set x - axis to the direction of the gc from the galactic centre .
z - axis coincides with the gc angular momentum vector .
y - axis lies in the gc orbital plane .
the gc is rotating counter - clockwise in the x - y plane.,width=340 ] in fig.[fig4 ] , i show positions of the p- and n - horn particles at t=2 gyr .
the particles are projected on the gc orbital plane . by calculating dot products of angular momentum vectors of the gc and the horn particles ,
i confirm that most of the particles plotted have positive dot products at t=2 gyr : prograde rotations with the gc ( 1,194 of 1,513 p - horn and 1,734 of 2,028 n - horn particles have prograde rotations ) .
thus , i find that most of the p - horn particles are orbiting behind the gc , whereas , the n - horn particles are orbiting ahead of the gc . moreover , since most of these particles have short distances from the gc and the prograde rotations , i can ascribe the horn - feature in fig.[fig3 ] to the similarity between orbits of the horn particles and the gc .
these particles strongly interact with the gc .
conversely , the gc ought to be affected by these particles .
i examine orbital motions of the horn particles . in fig.[rp_t2
] , i illustrate typical orbits of the p- and n - horn particles in the coordinate rotating with the gc .
the p - horn particle in this figure approaches the gc from behind .
the orbit expands when catching - up , and finally retreats from the gc after the orbital expansion .
this orbital expansion corresponds to the energy increase seen in fig.[fig3 ] . on the other hand ,
the orbit of the n - horn particle is the reverse , comes close to the gc ( in inertia coordinate , the gc is catching up to the n - horn particle ) , shrinks the orbital radius , and runs away from the gc . as a result , the energy of the n - horn particle decreases .
the mechanism of these orbital behaviors is simple . under density gradients ,
generally , inner orbits have shorter orbital periods , on the other hand , outer ones have longer periods .
the p - horn particle is pulled forward by gravity of the gc , and gains energy .
however , this energy increase causes the orbital expansion and makes angular velocity slow . in the case of the n - horn particle , the particle is pulled backward and decrease the energy . this energy decrease results in the orbital shrinkage and faster angular velocity .
this phenomenon , that the particle slows down in azimuth when pulled forward and speeds up when held back , has come to be called ` the donkey effect ' @xcite .
the orbits in fig.[rp_t2 ] imply that higher energy particles lose orbital energy , on the other hand , lower energy particles gain energy .
this behavior can be confirmed in fig.[fig3 ] , which indicates that the two horns lean slightly from top - left to bottom - right .
@xcite has also investigated this kind of orbital motion ( trojan horseshoe orbits ) in n - body systems , which look like the orbits in fig.[rp_t2 ] ( see * ? ? ? * fig.7 ) .
it is worthy of special mention that these orbits exhibit quite different behavior from two - body interaction that is assumed in the derivation of the chandrasekhar formula .
this orbital behavior of the horn particles is presumed to be caused by many - body interaction , rather than two - body interaction ( see , appendix a ) .
means energy injection into the gc , decrease means energy absorption from the gc.,width=317 ] to probe the energy transfer from the horn particles to the gc , i rerun the same simulation while calculating mechanical work on the gc , @xmath63 this integration means cumulation of the integrand on each timestep .
@xmath64 represents net acceleration exerted on the gc by a specific particle group ( e.g. the p- , n - horn or the other particles ) .
i compute energy transfers on the gc in the same simulation and show the result in fig.[fig5 ] .
though the horn particles were defined by energy change during 2 - 3 gyr , i draw the energy transfers during t=1.5 - 3.5 gyr in fig.[fig5 ] ; setting @xmath65 gyr in eq.[de ] . in agreement with a naive expectation
, i find that the p - horn particles absorb energy from the gc and the n - horn particles inject energy into the gc .
it is notable that the behaviors of the injection and absorption are monotonic .
the transfers are sharp in 2 - 3 gyr , slow in @xmath66 gyr and @xmath67 gyr .
this implies that the energy transfers are transient .
but setting @xmath68 gyr in eq.[de].,width=317 ] the energy transfers in fig.[fig5 ] do nt completely balance ; however , the energy absorption by the p - horn particles slightly surpasses the injection by the n - horn .
i compare the energy transfer on the gc by both of the p- and n - horn particles with that by the other particles in fig.[fig6 ] .
the figure indicates that the other particles also steal energy from the gc .
this energy loss of the gc seems to be caused by df against the other field particles except the horn particles .
interestingly , the energy loss caused by the horn particles is comparable to that by the other particles , accounting for a half of actual df by all particles .
the dotted line labelled as ` chandrasekhar formula ' is calculated by substituting eq.[chandra ] for @xmath64 in eq.[de ] using the data of gc orbit in the simulation .
the df from all particles is well fitted by the chandrasekhar formula with @xmath35 this result implies that actual df can be separated into two modes .
the first mode is the well - known conventional df based on the concept of @xcite .
this is a df that is attributed to random encounters with a myriad of field particles in an isotropic velocity distribution . on the other hand , unlike the first mode , the cause of the second mode is a small number of particles which are in many - body interaction ; the horn particles .
the horn particles have only 0.0354 per cent of the total mass of the system .
however , keep in mind that the criterion for defining the horn particles is somewhat arbitrary .
moreover , the result in fig.[fig6 ] does nt necessarily mean that df caused by ` _ _ true _ _ ' many - body interacting particles accounts for a half of the actual df . even in two - body interaction ,
if a field particle has a similar velocity vector to @xmath69 and a small impact parameter , the particle causes strong interaction and large energy transfer .
there would be no clear - cut way to extract only the true many - body interacting particles .
however , it is sure that the horn particles defined in fig.[fig3 ] include many many - body interacting particles . for @xmath70 gyr
, the df does nt seem to act on the gc to shrink the orbit .
this phenomenon itself has already been known .
@xcite found that this core - stalling of df can be explained by the chandrasekhar formula .
however , their estimation was based on the assumption of completely constant density in the core . without this assumption , the chandrasekhar formula predicts slow but certain orbital shrinkage @xcite , as seen in fig.[fig1 ] . on the other hand ,
the n - body simulations indicate the complete suppression of the df in the cored region . the chandrasekhar formula is no longer viable in this phase .
@xcite has suggested that the mechanism of the df suppression is anisotropic velocity distribution induced by the gc .
they attributed the df suppression to increase of number of particles in ` co - rotating state ' , in which more field particles rotate with the gc in prograde direction than initial isotropic state .
@xcite , however , revealed that the increase of co - rotating particles is marginal and anisotropy induced by the gc is very weak .
besides , @xcite suggested that fluctuation of the halo potential weakens the df in the core .
but , they supposed the case of a black hole and a stellar cored structure , the perturber being so heavy that their theory is not applicable to the case of a dwarf galaxy and gcs .
, but the differential is taken between 7 and 8 gyr : @xmath71 .
criteria for defining p - horn ( n - horn ) particles is @xmath72 and @xmath59 ( @xmath60).,width=317 ] gyr.,width=317 ] i execute the same analyses as in the previous subsection . in fig.[t7-t8_contour ] , i redefine p- and n - horn particles by the same way .
the numbers of the p- and n - horn particles are 613 and 834 , respectively .
positions of these horn particles at t=7 gyr are plotted in fig.[t7-t8_positions ] . as well as the exerting df phase , i confirm that most of the particles plotted have prograde rotations with the gc ( 399 of 613 p - horn and 761 of 834 n - horn particles ) . in fig.[rp_t7 ] , i illustrate typical orbits of p- and n - horn particle .
this figure indicates very complex orbits .
both particles come close to the gc from the front , come inside the gc orbital radius , rotate around the gc , expand the orbits and finally go away behind the gc .
it seems as if the particles dodge the gc .
although these particles are respectively defined as a p- and a n - horn particle , these orbits are identical .
regardless of the complexity , by considering absence of the donkey effect in the constant density region , the following explanation can be made . in constant densities ,
orbital period becomes independent of orbital radius . in coming close to the gc
, the particles are orbiting slightly outside the core with slower angular velocities than the gc . by the gravity from the gc ,
the particles are pulled backward and lose energy .
the particles sink toward the centre by the energy loss .
however , by the absence of the donkey effect in the core region , the angular velocities of the particles are kept constant even in the inner radii .
after getting behind the gc , the particles are pulled forward by the gc and gain energy .
as a result , the orbits of the particles expand again and are exiled from the core .
this orbit is also not in two - body interaction but in many - body interaction . in appendix
a , i will discuss about this orbit in restricted three - body problem .
gyr in this plot).,width=317 ] in fig.[et_horns_t7-t8 ] , i show cumulative energy transfers into the gc . in consistency with the complexity of the orbits ,
the mechanical works on the gc also show very intricate features .
the transfers are no longer monotonic , unlike in the exerting df phase ( fig.[fig5 ] ) .
the n - horn particles continue to inject energy until @xmath73 gyr , but absorb energy from the gc afterwards .
meanwhile , the p - horn particles inject a little energy at first , but largely absorb later .
thus , the total transfer from all horn particles indicates an intricate waving behavior .
although these particles are defined by @xmath74 , the particles largely inject and absorb energy in @xmath75 gyr and @xmath76 gyr .
these behaviors implies not transient but long - term interaction between the particles and the gc .
.the numbers of particles defined as horn particles in each time - intervals .
the top and bottom tables are 1gyr - interval and 0.5gyr - interval , respectively .
the leftmost column indicates time - period in which i calculate @xmath56 to define the horn particles .
the bottommost row is the number of particles for which all particles listed above are gathered .
because some particles overlap , the number of particle gathered is not identical to simple sum of rightmost columns . [ cols="<,^,^,>",options="header " , ] [ list_1gyr ] to be 1 gyr . for comparison , an analytical estimation by the chandrasekhar formula
is also plotted , but it is reduced by half from the naive estimation by eq.[chandra].,width=317 ] from the discussion above , in the suppressed df phase , it is clear that the energy transfers from some particles have long duration . in this case
, i could nt extract all many - body interacting particles successfully . in order to overcome this difficulty , i screen particles in t=5 - 6 , 6 - 7 , 7 - 8 , 8 - 9 , 9 - 10 gyr by calculating @xmath56 .
if a particle is defined as a horn particle even only once in the five time - periods , i regard the particle as a many - body interacting particle in all time - periods .
i gather all many - body interacting particles in this five time - periods .
the numbers of particles listed up are shown in the top of table [ list_1gyr ] . while operating the same simulation starting from t=5 gyr , i compute energy transfers into the gc from the _ gathered _ horn particles and the others .
the result is shown in fig.[et_1gyr - interval ] . in this figure ,
the cumulative energy transfer from the many - body interacting particles gradually increases . on the other hand ,
the other particles absorb energy from the gc .
these cumulative transfers , however , indicate periodic flapping .
this periodicity appears just in 1 gyr period , implying an artificial effect caused by setting the time - interval to calculate @xmath56 to be 1 gyr . , but using 0.5gyr - interval to extract the gathered horn particles.,width=317 ] therefore , i follow the same procedure to calculate @xmath56 with 0.5gyr - interval . the numbers of particles listed up are shown in the bottom of table [ list_1gyr ] .
i show cumulative energy transfers in fig.[et_0.5gyr - interval ] with an analytical estimation by chandrasekhar formula for the sake of comparison .
however , the analytical estimation is halved , according to the result of 3.1 .
the periodic flapping in fig.[et_1gyr - interval ] disappears .
moreover , the energy injection from the horn particles are monotonic for @xmath77 gyr .
the energy absorption by the other particles is also monotonic .
interestingly , the behavior of the energy transfer from the horn particles is largely different from that in the exerting df phase ; the horn , many - body interacting , particles continue to inject energy into the gc in the suppressed df phase . on the other hand ,
the other particles absorb energy from the gc as well as in the exerting df phase .
the halved analytic formula provides excellent fit to the absorption for @xmath77 gyr though there is a slight deviation at @xmath78 gyr .
these influences from horn and the other particles are canceled out and the gc orbit can remain on a constant orbital radius as if the df on the gc vanished .
by the n - body simulation in this paper , i conclude as follows . in the exerting df phase , the chandrasekhar formula seems good at estimating the actual df .
nevertheless , quantitatively , a very small number of particles accounts for a non - negligible fraction of the actual df although they occupy less than 0.1 per cent of the halo mass .
these particles are in many - body interaction , which have quite different orbits from two - body interaction .
the presence of such many - body interacting particles disagrees with the concept of @xcite .
however , in his formula , uncertainty of determining @xmath79 would conceal the contribution from these particles . in the suppressed df phase , the many - body interacting particles alter direction of energy transfer and feed energy into the perturber .
on the other hand , the other particles exert df , the orthodox df , ceaselessly . in this phase , as a result , these energy transfers cancel out with each other .
the orbit of the perturber is thus stabilized .
i suggest that this could be a possible mechanism of the suppressed df in a constant density structure .
although my results indicate a failure of the chandrasekhar formula , this is not surprising .
the derivation of the formula has assumed an unrealistic circumstance : uniform density , isotropic velocity field and non - periodic straight orbit .
moreover , a perturber in a real galaxy has to rotate around center of the galaxy .
therefore , from the very beginning it is unreasonable to apply the chandrasekhar formula to all kinds of density distributions .
@xcite and @xcite have analytically suggested energy feedback on a perturber by density response of field .
especially , figures in @xcite are very intelligible schematically .
he indicated shapes of density wakes . in his results , outside core region ,
the density wake is asymmetric and exerts drag force on the perturber . on the other hand , inside the core ,
the wake become symmetric and the drag force disappears .
the symmetric shape is similar to the distribution of the horn particles in my fig.[t7-t8_positions ] although the horn particles in my simulation is selected by energy response , not density .
however , in my simulation , the symmetric density wake indicated in @xcite can not be observed in the suppressed df phase .
besides , the n - body simulations of @xcite , almost the same simulations as mine , also do nt indicate the symmetric density wake , but asymmetric wake ( see fig.4 of @xcite ) even inside the core region .
@xcite also showed similar asymmetric desity wake in their fig.5 .
this discrepancy may or may not be caused by low resolution in our simulations .
also , @xcite has suggested that the field particles on circular orbits lose or gain energy according to the relative phase to the orbit of the perturber while the perturber plays the role of a catalyst to modify the distribution function of the field particles and induces anisotropy .
the result of this paper indicates that not only two - body interactions but also many - body interactions are responsible for the actual df .
the many - body interacting particles , horn particles , are very small in number and co - rotating with the perturber .
moreover , the resistance to the df by this corrective effect is also feasible in shallow cusp density profiles although the shallow cusp may evolve into a constant density ( see , appendix a ) .
this implies sensitivity of the actual df to velocity field of the halo models .
i employed in this paper the isotropic assumption , @xmath80 in eq.[jeans ] , as the most fundamental model .
however , many numerical simulations have suggested dependence of @xmath19 on radius in dwarf and other galaxies @xcite .
these results imply that actual df against a _ live _ halo would depend sensitively on the distribution functions of the haloes .
in addition , there are some generalised applications of the chandrasekhar formula ; against aspherical distributions @xcite , variable @xmath79 @xcite , df on spherical bodies @xcite , in a gaseous medium(e.g . * ? ? ? * ; * ? ? ?
* ) , etc .
the results in this paper alert to thoughtless usages of the chandrasekhar formula .
the author is financially supported by research fellowships of the japan society for the promotion of science ( jsps ) for young scientists . the numerical simulations reported here
were carried out on muv ( mitaka underground vineyard ) grape systems kindly made available by cfca ( center for computational astrophysics ) at national astronomical observatory of japan .
the numerical code was based on a software distributed on the website of joshua e. barnes ( http://ifa.hawaii.edu/~barnes/software.html ) .
i thank masafumi noguchi for his fascinating and helpful discussion , masaki iwasawa and yusuke tsukamoto for innovative advices .
it is pleasure to thank martin d. weinberg for his stimulating comments , which drastically improved the description given in this paper and my understanding of galactic dynamics .
i thank the anonymous reviewer for important suggestions that helped improve the paper .
angus g. w. , diaferio a. , 2009 , mnras , 396 , 887 barnes j. , 1990 , j. comput . phys .
, 87 , 161 barnes j. , hut p. , 1986
, nat , 324 , 446 bekki k. , 2010 , mnras , 401 , 2753 binney j. , 1977 , mnras , 181 , 735 binny j. , tremaine s. , 2008 , galactic dynamics second edition .
princeton univ . press , princeton boily c. m. , padmanabhan t. , paiement a. , 2008 , mnras , 383 , 1619 buonanno r. , corsi c. e. , castellani m. , marconi g. , pecci f. f. , zinn r. , 1999 , aj , 118 , 1671 buonanno r. , corsi c. e. , zinn r. , pecci f. f. , hardy e. , suntzeff n. b. , 1998 , apj , 501 , 33 burkert a. , 1995 , apj , 447 , 25 capuzzo - dolcetta r. , miocchi p. , 2008
, apj , 681 , 1136 chandrasekhar s. , 1943 , apj , 97 , 255 cowsik r. , wagoner k. , berti e. , sircar a. , 2009 , apj , 699 , 1389 w. j. g. , 2010 , adv .
astron . , 2010 , 789293 de blok w. j. g. , mcgaugh s. s. , bosma a. , rubin v. c. , 2001 , apj , 552 , 23 durrell p. r. , harris w. e. , geisler d. , pudritz r. e. , 1996 , aj , 112
, 972 esquivel o. , fuchs b. , 2007 , mnras , 378 , 1191 fujii m. , funato y. , makino j. , 2006 , pasj , 58 , 743 fujii m. , iwasawa m. , funato y. , makino j. , 2009 , apj , 695 , 1421 gilmore g. , wilkinson m. i. , wyse r. f. g. , kleyna j. t. , koch a. , evans n. w. , grebel e. k. , 2007 , apj , 663 , 948 t. , moore b. , read j. i. , stadel j. , 2010 , apj , 725 , 1707 goerdt t. , moore b. , read j. i. , stadel j. , zemp m. , 2006 , mnras , 368 , 1073 f. , et al .
, 2010 , nat , 463 , 203 greco c. , et al . , 2007 , apj , 670 , 332 gualandris a. , merritt d. , 2008 , apj , 678 , 780 hashimoto y. , funato y. , makino j. , 2003 , apj , 582 , 196 hernandez x. , gilmore g. , 1998 , mnras , 297 , 517 hernquist l. , 1993 , apj , 86 , 389 inoue s. , 2009 , mnras , 397 , 709 jiang i .-
g . , binney j. , 2000 , mnras , 314 , 468 kim h. , kim w .-
, snchez - salcedo f. j. , 2008 , apj , 679 , 33 n. , chen d. , 2009 , research in astronomy and astrophysics , 9 , 1173 w. , fu y. , deng z. , huang j. , 2005 , pasj , 57 , 541 lotz j. m. , miller b. w. , ferguson h. c. , 2004 , apj , 613 , 262 lynden - bell d. , kalnajs a. j. , 1972 , mnras , 157 , 1 mackey a. d. , gilmore g. f. , 2003 , mnras , 340 , 175 makino j. , 1991 , pasj , 43 , 621 mashchenko s. , couchman h. m. p. , wadsley j. , 2006 , nat , 442 , 539 mashchenko s. , wadsley j. , couchman h. m. p. , 2008
, sci , 319 , 174 miller b. w. , lotz j. m. , ferguson h. c. , stiavelli m. , whitmore b. c. , 1998 , apj , 508 , 133 miocchi p. , capuzzo - dolcetta r. , di matteo p. , vicari a. , 2006 , apj , 644 , 940 navarro j. f. , frenk c. s. , white s. d. m. , 1997 , apj , 490 , 493 oh k. s. , lin d. n. c. , richer h. b. , 2000 , apj , 531 , 727 s. , brook c. , governato f. , brinks e. , mayer l. , de blok w. j. g. , brooks a. , walter f. , 2010 , arxiv 1011.2777 ostriker e. c. , 1999 , apj , 513 , 252 s. , grebel e. k. , berczik p. , spurzem r. , dehnen w. , 2010 , a&a , 514 , 47 pesce e. , capuzzo - dolcetta r. , vietri m. , 1992 , mnras , 254 , 466 read j. i. , goerdt t. , moore b. , pontzen a. p. , stadal j. , 2006 , mnras , 373 , 1451 salucci p. , burkert a. , 2000 , apj , 537 , 9 snchez - salcedo f. j. , reyes - iturbide j. , hernandez x. , 2006 , mnras , 370 , 1829 springel v. , et al . , 2008 , mnras , 391 , 1685 springel v. , white d. m. , 1999 , mnras , 307 , 162 strader j. , brodie j. p. , forbes d. a. , beasley m. a. , huchra j. p. , 2003
, aj , 125 , 1291 strigari l. e. , bullock j. s. , kaplinghat m. , simon j. d. , geha m. , willman b. , walker m. g. , 2008 , nat , 454 , 1096 tremaine s. , 1976 , apj , 203 , 345 tremaine s. , weinberg m. d. , 1984 , mnras , 209 , 729 van albada t. s. , 1982 , mnras , 201 , 939 vesperini e. , 2000 , mnras , 318 , 841 vesperini e. , 2001 , mnras , 322 , 247 m. d. , 1989 , mnras , 239 , 549
i discuss here why and how the orbits indicated in fig.[rp_t7 ] appear in the suppressed df phase . since the orbit of the perturber is nearly circular in my simulation and density field is stable , i expect that circular restricted three - body problem would be a valid method to investigate the orbits in the core region . here , i assume that the gc is orbiting at a separation of @xmath81 from the galactic center in a circular orbit .
let the enclosed mass of the halo , @xmath82 .
i set the coordinate center to the center of mass of the system .
then , the galactic center and the gc are located at @xmath83 $ ] and @xmath84 $ ] , respectively .
then , i calculate effective potential at @xmath85 $ ] in the coordinate rotating with the gc , @xmath86 , \label{jacobi}\ ] ] where @xmath87 , @xmath88 .
the first and second term are gravitational potential of the halo density distribution and the gc , respectively .
the third term is centrifugal potential . .
colour - bar labels the energy , @xmath89 .
the coordinate center is the center of mass of the system .
the dot - dashed , solid and dashed white lines indicate equipotential surfaces of lagrangian points l1 , l2 and
l3 , respectively . the green line indicates an isosurface on which the orbits indicated in fig.[rp_t7 ] are possible.,width=377 ] in fig.[mas ] , i show a contour plot of the effective potential in the case of that the core radius is set to @xmath90 , which corresponds to the suppressed df phase in my simulation .
i also plot equipotential surfaces of lagrangian points , l1 , l2 and l3 .
the green line indicates a equipotential surface between the effective potentials at l2 and l3 : @xmath91 and @xmath92 .
this green line runs through around the gc just like the horn - particle orbits in fig.[rp_t7 ] .
if field particles have a jacobi integral higher than @xmath91 and lower than @xmath92 , some of these can orbit along this green line .
i consider the particles shown in fig.[rp_t7 ] to be examples of this kind of particle .
special attention should be paid to the fact that the effective potential at l2 ( the solid white line ) is lower than that at l3 ( the dashed white line ) .
this height relation between @xmath91 and @xmath92 permits appearance of orbits like fig.[rp_t7 ] . on the other hand , in cases of cuspy density distribution , @xmath91 is generally higher than @xmath92 ( see below ) . in fig.[tadpole
] , i show a contour plot of the effective potential in the case where the core radius is set to @xmath93 , which corresponds to the case of cuspy density distribution : the exerting df phase . in this case , @xmath91 is higher than @xmath92
. then , the green lines in this case can _ not _ run across between potential minima , _
i.e. _ , no orbit can run through between the gc and the galactic center since such a route is completely prohibited .
the green lines in fig.[tadpole ] correspond to the horn - particle orbits in fig.[rp_t2 ] . .
the dot - dashed , solid and dashed white lines indicate equipotential surfaces of lagrangian points l1 , l2 and
l3 , respectively . the green line indicates an isosurface on which the orbits indicated in fig.[rp_t2 ] are possible.,width=377 ] in the case of the cored density , if the gc is orbiting in even inner region of the core , much more particles can become horn - particles shown as the green line in fig.[mas ] .
therefore , force against df becomes strong toward inner region , at a certain radius , the force balances with the df on the gc .
this means that the df suppression is not attributed anisotropic velocity state induced by the gc rotation : co - rotating state suggested by @xcite .
@xcite has shown that such a co - rotating anisotropic velocity state is very marginal .
if the appearance of horn - particles is the true mechanism of df suppresion , even though the velocity distribution is isotropic , core - stalling of df is possible .
although the horn - particles i suggest are co - rotating particlles in a sense , but _ there is no need to induce the anisotropy _ in the velocity field .
this is consistent with the marginally anisotropic velocity distribution found by @xcite .
next , i address the cases of shallow cusp density distibutions by means of the circular restricted three - body problem .
some studies have concluded that actual dark matter profile of dwarf galaxies would have a shallow inner cusp with slopes between 0.3 - 0.6 @xcite . in order to investigate the cases of shallow cusps ,
i use another density profile , @xmath94^{(\beta-\gamma)/\alpha}},\ ] ] where @xmath95 and @xmath96 .
i set the parameters , @xmath97 , @xmath98 ( outer slope ) and @xmath99 or @xmath100 ( inner slope ) .
if @xmath97 , @xmath98 and @xmath101 , this profile becomes very similar to burkert profile . in this profile ,
i assume that the gc with the mass of @xmath102 is orbiting at a separation of @xmath81 from the galactic center in a circular orbit .
as i discussed above , some field particles can be in the special orbit ( fig.[rp_t7 ] ) if the effective potential on l3 point is higher than that on l2 .
this fact implies that all i have to do is to calculate the effective potential on l3 and l2 .
then , i plot the effective potential along x - axis ( the line of the gc and the halo centre ) .
fig.[a1 ] shows the effective potential in the case of @xmath99 . in this case , the effective potential on l3 is higher than that on l2 .
i expect that some horn particles in the special orbit would be existent
. then , according to the main discussion of the paper , these horn particles could counter and weaken df on the gc .
i show the effective potential in the case of @xmath103 in fig.[a2 ] .
the position of the gc , the scale radius , @xmath4 , and other parameters are the same .
the effective potential on l3 is still higher than that on l2 .
however , the difference between these is very small in comparison with the previous case
. only a minority of particles can be in the special orbit if @xmath103 . in this case
, the horn particles may be unable to balance against the df on the gc . from the discussion above
, steeper inner slope of the density profile allows fewer particles to orbit in many - body interaction manner like fig.[rp_t7 ] .
however , in this discussion , it is very difficult or impossible to investigate how many particles in this orbit are required to cancel out the df on the gc .
after all , detailed investigation needs to operate n - body simulations .
however , as @xcite has concluded , n - body simulation suffers from the decline of the inner cusp .
then , i can not conclude what value of @xmath104 is the criterion of the df stalling in ( nearly ) cored density profiles .
since steeper ( shallower ) cusp allows fewer ( more ) horn particles , the number of particles required to observe this kind of orbits in n - body simulations would depend on the inner slope . in the cases of steeper cusps , to operate n - body simulation needs more particles . | dynamical friction is a fundamental and important phenomenon in astrophysics .
the chandrasekhar formula is a well - known analytical estimation of the effect .
however , current astrophysicists have realized that the formula is not correct in some cases because of several approximations dared in the formulation and/or complex non - linearities in the real universe .
for example , it has been indicated that the dynamical friction does nt work in cored density profiles ( constant density in the central region ) despite that the chandrasekhar formula predicts drag force even in the constant densities . in the former half of this paper ,
i discuss by n - body simulations that many - body interactions are also important in actual dynamical friction though derivation of the chandrasekhar formula is based on the assumption of two - body interaction . in the simulation ,
the many - body interactions are caused by a very small number of field particles co - rotating with a perturber .
however , the contribution from the many - body interactions accounts for a non - negligible fraction of the actual dynamical friction . in the latter half ,
i discuss why the cored profiles suppress the dynamical friction .
one possible explanation is that corrective effect of the many - body interactions drive orbital motion of the perturber .
the cessation of dynamical friction by this corrective effect would be feasible even in shallow cusp density profiles although the shallow cusp may evolve into a constant density .
[ firstpage ] methods : n - body simulations galaxies : dwarf galaxies : haloes galaxies : kinematics and dynamics galaxies : star clusters . |
upcoming surveys of galaxies and hi will probe increasingly large scales , approaching and even exceeding the hubble scale at the survey redshifts . on these cosmological scales ,
surveys can in principle provide the best constraints on dark energy and modified gravity models and will be able to test general relativity itself .
furthermore , constraints on primordial non - gaussianity from large - scale surveys of the matter distribution will be competitive with cmb constraints .
however , in order to realise the potential of these surveys , we need to ensure that we are using a correct analysis , i.e. a general relativistic analysis , on cosmological scales .
there are two fundamental issues underlying the gr analysis .
* we need to correctly identify the galaxy overdensity @xmath0 that is _ observed _ on the past light cone .
the overdensity @xmath1 defined in different gauges gives the same results on sub - hubble scales , but leads to different results on large scales and this remains true even if we use gauge - invariant definitions of @xmath1 .
the observed @xmath0 is necessarily gauge - invariant , and is unique .
* we need to account for all the distortions arising from observing on the past light cone , including redshift distortions ( with all gr effects included ) and volume distortions .
these gr effects come in to the _ measured _ 2-point correlation function in redshift space @xmath2 .
we provide a new representation of this function which is relatively simple to calculate , and which takes into account all gr and wide - angle contributions .
our fully general relativistic wide - angle formalism recovers and generalizes previous work in the newtonian plane - parallel ( flat - sky ) @xcite and newtonian wide - angle @xcite cases . the gr analysis of the matter power spectrum @xcite leads to corrections on cosmological scales of the standard newtonian analysis ( which is accurate on small scales ) .
the observed galaxy overdensity is a function of the observed direction @xmath3 and redshift @xmath4 .
it may be expressed in any chosen gauge .
we use synchronous - comoving gauge , in which @xmath5 dx^idx^j\big\}.\ ] ] in @xmath6cdm , we have @xmath7 @xcite ( a prime denotes @xmath8 )
. we can write the observed overdensity at observed redshift @xmath4 and in the unit direction @xmath3 as @xmath9 here @xmath10 is a local term ( i.e. evaluated at the source ) which includes the galaxy density perturbation , the redshift distortion and the change in volume entailed by the redshift perturbation .
@xmath11 is the weak lensing convergence integral along the line of sight , and @xmath12 is a time delay integral along the line sight . in the gauge ( [ sync ] ) , we have @xcite @xmath13\left ( \partial_\parallel e ' + e '' \right ) \nonumber\\ & & - \frac{(1 + { z})}{h } \partial^2_\parallel e ' -\frac{2 } { { \chi}}\left(1-\mathcal{q}\right)\left(\chi { \cal r}+e'\right)\ ; , \\ \label{delk } { \delta}_\kappa & = & \left(1-\mathcal{q}\right)\nabla^2_\perp \int_0^ { { \chi } } d\tilde\chi \left ( { \chi}-\tilde\chi\right ) \frac { { \chi}}{\tilde\chi}\left(e''-{\cal r } \right)\;\\ \label{deli } { \delta}_i & = & -\frac{2 } { { \chi}}\left(1-\mathcal{q}\right ) \int_0^ { { \chi } } d\tilde\chi\left(e''-{\cal r } \right ) \nonumber \\ & & + \left[b_e - \left(1 + 2\mathcal{q}\right)\frac{(1 + { z})}{h}\frac{dh}{dz}-\frac{2 } { { \chi}}\left(1-\mathcal{q}\right)\frac{(1 + { z})}{h}\right ] \int_0^ { { \chi } } d\tilde\chi e'''.\end{aligned}\ ] ] here @xmath14 is the comoving distance , @xmath15 is the bias and @xmath16 \over d z},\ ] ] where @xmath17 is the background number density .
( we have changed some of the notation in @xcite . )
the directional derivatives are defined as @xmath18 the term @xmath19 encodes the magnification bias , which arises from the perturbation to the flux of a galaxy , relative to a galaxy at the same observed redshift in the unperturbed universe ( see @xcite ) . the local term @xmath20 contains the newtonian local terms , and in addition some gr corrections .
the line of sight term @xmath21 is a pure gr correction .
the lensing term @xmath22 is the same as in the newtonian analysis .
note that in @xmath23 we do not drop the radial derivative term @xmath24 which gives a negligible contribution to ( [ delk ] ) on small scales , but should not be neglected on large scales .
we can relate the metric perturbations to the matter density contrast in synchronous gauge , removing the residual gauge ambiguity .
taking into account that @xmath25 , we get @xmath26 here @xmath27 is the matter density and @xmath28 is the growth rate , @xmath29 where @xmath30 is the growing mode of @xmath31 .
the correlation function can be decomposed in spherical harmonics as @xcite @xmath32 however , this turns out to be computationally expensive to implement for wide - angle correlations , and we follow the alternative decomposition used in previous analyses based on a newtonian approach @xcite .
( the formula for the correlation function in redshift space , including wide - angle effects , was first derived in @xcite , without gr corrections . )
this alternative expands the redshift space correlation function using tripolar spherical harmonics , with the basis functions @xmath33^{1/2}\ !
\sum_{m_1,m_2,m } \left ( \begin{array } { ccc } \ell_1 & \ell_2 & l \\ m_1 & m_2 & m \end{array } \right ) y_{\ell_1 m_1 } ( { { { \bf n}}}_1 ) y_{\ell_2 m_2 } ( { { { \bf n}}}_2 ) y_{l m } ( { { { \bf n}}}_{12 } ) , \nonumber \\ & & -\ell_1 \leq m_1 \leq \ell_1\ ; , \quad -\ell_2 \leq m_2 \leq \ell_2\ ; , \quad -l \leq m \leq l\ , . \label{tripo}\end{aligned}\ ] ] here @xmath34 is the wigner 3@xmath35 symbol . because our triangles are closed , only two independent @xmath36 s appear .
the expansion for @xmath37 may then be written in redshift space as a sum over these functions with @xmath38 in the range @xmath39 ( see @xcite for details ) . in the conventional expansion of @xmath37 into @xmath40 s ,
the @xmath36 represents the angular momentum of the correlation function , and is summed to infinity . by contrast , in the tripolar expansion , the @xmath36 s are representative of the radial derivatives @xmath41 in the two directions @xmath42 and @xmath43
. consequently , the sums involved in the tripolar expansion are no longer infinite , making explicit computations in redshift space comparatively simple .
the positions of two galaxies and their separation are given by ( see fig .
[ lcone ] ) @xmath44 the overdensity in the form ( [ delta_g ] ) leads to the cross correlation functions in redshift space : @xmath45
first , we define the spherical transform of the matter overdensity @xcite @xmath46 where @xmath47 is a legendre polynomial .
then by , @xmath48 can be be decomposed as @xmath49 where @xmath50\frac{(1+z)}{h(z)}\right ] \ ; , \\
\label{beta } \beta(z ) & = & \frac{f(z)}{b(z)}\ ; , \\
\gamma(z ) & = & \frac{h(z)}{(1+z ) } \left\{\frac{h(z)}{(1+z ) } \left[\beta(z ) -\frac{3}{2}\frac{\omega_m ( z)}{b(z ) } \right]b_e(z ) + \frac{3}{2}\frac{h(z)}{(1+z ) } \beta(z ) \big [ \omega_m ( z ) - 2 \big ] \right .
\nonumber \\ & & \left .
-\frac{3}{2 } \frac{h(z)}{(1+z ) } \frac{\omega_m ( z)}{b(z ) } \left[1 - 4\mathcal{q}(z ) + \frac{3}{2}\omega_m ( z)\right ]
+ \frac{3}{\chi(z ) } \big[1-\mathcal{q}(z)\big]\frac{\omega_m ( z)}{b(z ) } \right\ } .
\label{gamma}\end{aligned}\ ] ] here @xmath51 is a generalization of the newtonian expression , @xmath52 has the same form as in the newtonian analysis and @xmath53 is a new term arising from gr corrections : see fig .
[ abgms ] below . in order to decompose @xmath54
, we need the correlator of ( [ a - tensor ] ) : @xmath55 where @xmath56 is defined by @xmath57 in terms of the primordial power spectrum and the transfer function , we have @xmath58 expanding @xmath59 and @xmath60 in spherical harmonics and applying the gaunt integral @xcite , we obtain @xmath61 this is an expansion in the tripolar basis functions , with coefficients @xmath62 finally , we arrive at the tripolar decomposition of @xmath54 in the most general ( gr wide - angle ) case : @xmath63 by , @xmath64 so that @xmath65 ; by we also have @xmath66 .
the @xmath67 coefficients for @xmath54 and the other @xmath68 ( see below ) are given in appendix [ bco ] .
these coefficients involve the functions and that contain gr corrections . in order to compare the gr and newtonian cases , we set @xmath69 for simplicity .
we rewrite ( [ alpha ] ) as @xmath70 + \frac{d \ln{n_g}}{d \chi}+ \frac{2}{\chi},\end{aligned}\ ] ] where @xmath71 is the comoving galaxy number density .
then for small redshift ( i.e. @xmath72 ) , we recover the newtonian limit of @xmath51 @xcite : @xmath73 in the same limit , shows that @xmath74 \frac{d \ln{n_g}}{d \chi } + \frac{2}{\chi } \right\ } , \label{gamma3}\end{aligned}\ ] ] which is nonzero . however , to recover the newtonian limit , we have to take into account ( [ xiln ] ) when @xmath75
. then we find : @xmath76 so that @xmath53 drops out of @xmath54 in the newtonian approximation .
+ the remaining @xmath68 all involve integrals along the lines of sight .
the spherical transforms of @xmath77 are @xmath78 , \label{dkexp } \\
\delta_i ( { \bf
n } , z)&= & b(z)\int^{\chi}{d\tilde\chi } \ ; \mu(z,\tilde{z})\mathcal{a}^2_0(\tilde{\bf x},\tilde{z } ) , \label{diexp}\end{aligned}\ ] ] where @xmath79}{b ( { z } ) } \omega_m ( \tilde z ) , \label{sigma}\\ \mu ( z,\tilde z ) & \equiv & 3 \frac{h^2(\tilde{z})}{(1+\tilde{z})^2}\frac{\omega_m ( \tilde{z})}{b ( { z})}\left\{\frac{2 } { { \chi } } \big[1-\mathcal{q}(z ) \big ] \right .
\nonumber \\ & & \left . - \frac{h(\tilde{z})}{(1+\tilde{z } ) } \big[f(\tilde{z})-1\big ] \left [ b_e ( { z } ) - \big[1 + 2\mathcal{q}(z)\big ] + \frac{3}{2 } \omega_m ( { z})- \frac{2 } { { \chi}}\big[1-\mathcal{q } ( z ) \big ] \frac{(1 + { z})}{h ( { z})}\right ] \right\ } , \label{mu}\end{aligned}\ ] ] and @xmath80 .
these functions are illustrated in fig .
[ sigmu ] : @xmath81 has the same form as in newtonian analysis and @xmath82 corresponds to gr terms that vanish in the newtonian limit . for the lensing - lensing correlation
, we find @xmath83 and for the @xmath84 correlation @xmath85 the integration variables @xmath86 are given by @xmath87}.\label{kk - ii2}\end{aligned}\ ] ] similarly , we find : @xmath88 where @xmath89}.\end{aligned}\ ] ] note that the remaining @xmath68 follow from the symmetry in .
( see appendix [ bco ] for the explicit expressions . )
here we consider the gr wide - angle correlation function @xmath90 in various limits and then in the general case . for simplicity , and
since it does not affect the comparisons , we assume @xmath69 . for small angle and small galaxy separation , we have the plane - parallel or flat - sky limit , i.e. @xmath42 and @xmath43 are almost parallel @xcite . using the properties of the wigner coefficients and spherical functions , we get @xcite @xmath91 with @xmath92 ( @xmath93 ) , @xmath94 is given by @xmath95 . from ( [ eq : xi_ss ] ) we obtain the expansion of the redshift space correlation function in monopole , quadrupole and hexadecapole terms : @xmath96 \xi_0^{2 } ( { \chi}_{11 } ; { z}_1,{z}_1 ) \nonumber \\ & & ~+ \gamma_1 ^ 2 \xi_0^{4 } ( { \chi}_{11 } ; { z}_1,{z}_1 ) \bigg\ } \mathcal{p}_{0 } ( { { { \bf n}}}_1 \cdot { { { \bf n}}}_{11 } ) + \left[- 4 \beta_1 \left ( \frac{1}{3}+ \frac{1}{7}\beta_1\right)\xi_2^{0 } ( { \chi}_{11 } ; { z}_1,{z}_1 ) \right.\nonumber \\ & & ~+ \frac{2}{3 } \beta_1 \left ( 2 \gamma_1- \frac{\beta_1\alpha_1 ^ 2}{\chi_1 ^ 2}\right)\xi_2^{2 } ( { \chi}_{11 } ; { z}_1,{z}_1)\bigg ] \mathcal{p}_{2 } ( { { { \bf n}}}_1 \cdot { { { \bf n}}}_{11 } ) + \frac{8}{35 } \beta_1 ^ 2 \xi_4^{0 } ( { \chi}_{11 } ; { z}_1,{z}_1 ) \mathcal{p}_{4 } ( { { { \bf n}}}_1 \cdot { { { \bf n}}}_{11 } ) . \label{ppxiss } \end{aligned}\ ] ] ( this is consistent with @xcite . ) in , we can divide the terms into 3 categories : those including only @xmath52 , those with @xmath51 , and those with @xmath53 .
the terms with @xmath52 are the ones considered in the standard kaiser analysis , and if we consider only them we will recover the classical results ( see e.g. @xcite ) .
the terms including @xmath51 are responsible for the `` mode - coupling '' effects , and they arise in the jacobian relating real- to redshift - space ( for details , in newtonian analysis , see e.g. @xcite ) .
these terms are usually ignored when using the plane - parallel approximation .
note that they vanish if the comoving radial distribution of galaxies is constant .
finally , the terms with @xmath53 are the gr corrections . in fig .
[ pplxi ] , we illustrate the gr corrections in to the newtonian monopole and quadrupole of @xmath54 . for galaxy separations up to @xmath97mpc , the gr corrections to the quadrupole remain negligible while the monopole is corrected by up to 2%
. note that in the newtonian limit , a monopole and a quadrupole term proportional to @xmath98 survive : @xmath99 see also @xcite .
these terms are usually omitted from the standard flat - sky analyses @xcite ( for details see @xcite ) .
the remaining correlation functions @xmath68 , which have not before been explicitly computed in the plane - parallel limit , are presented in appendix [ ppl ] .
for simplicity we consider the angular correlation , with @xmath100 .
we rewrite @xmath101 as @xmath102 where we impose a large - scale cutoff @xmath103 , which we take as @xmath104 .
we take @xmath105 . in this case , for @xmath106 , we have @xmath107 , and @xmath108 .
the second integral , for @xmath109 , can be approximated as @xmath110}{\gamma[(l+2)/2]}\;.\end{aligned}\ ] ] for @xmath111 , the integral vanishes because @xmath112 .
also , we can take @xmath113 since @xmath114 .
then , for the spectral index @xmath115 , we obtain the analytical approximation : @xmath116 & { \rm for}~~ l>0 . \end{array } \right .
\end{aligned}\ ] ] note that , when @xmath117 and @xmath111 , we require an infrared ( ir ) cutoff , @xmath118 , since @xmath119 becomes power - law divergent .
( if @xmath120 , there is a logarithmic divergence . )
the ir cutoff appears only in the terms of the correlation function that contain @xmath121 with @xmath122 .
( in this case @xmath123 . )
therefore they add only an overall additive normalization to @xmath37 which is unobservable , and so the ir cutoff actually conveys no information .
in addition , for large @xmath124 , the slope of @xmath101 is less steep when @xmath125 increases . in other words , on large scales , @xmath126 are bigger when they contain gr corrections .
( the same is true for the other @xmath127 . ) defining @xmath128 as the angular separation of two galaxies , we have @xmath129^{1/2}\chi$ ] .
we assume that @xmath130 , where @xmath131 is a suitable positive constant .
then by and , in the large - scale limit @xmath132 , we have @xmath133,\\ \gamma & \to & -3\frac{\omega_m(z)}{b(z)\chi_h^2(z)}\left[3\omega_m(z)+8 - 2f(z)\right]\;.\end{aligned}\ ] ] finally , here we show the gr corrections to the full wide - angle correlation function . for illustrative purposes
, we split the effects along the line of sight ( @xmath134 ) , and transverse to the line of sight ( @xmath135 ) .
the line of sight case corresponds to a configuration where one galaxy is much further than the other and the angular separation @xmath128 is small . in the transverse case ,
the two galaxies have the same redshift and @xmath128 is not small . in fig .
[ waxi ] we plot the gr corrections for the two examples : * @xmath54 , as a function of @xmath136 , computed for @xmath137 and @xmath138 .
* @xmath54 , as a function of @xmath139 , computed for @xmath140 . along with the correlation
function corrections , we plot the comoving galaxy separation , in order to have a clearer picture of the scales involved . in these computations we assumed a constant bias @xmath141 and @xmath17 from the universal mass function prescription , see @xcite .
the plots of the gr corrections show the ratio of the correlation function computed using the gr formalism with the one in the newtonian case .
wide - angle effects are included in both cases , the only difference being the inclusion or not of terms deriving from @xmath53 .
we decided to analyze two different cases , both for uniformity with previous literature and for physical reasons : pairs of galaxies ( almost ) along , and across the line of sight .
the first case ( left panel ) corresponds to pairs that have a large difference in redshift and experience the maximum of the redshift - space distortions effect , while the second one ( right panel ) corresponds to pairs of galaxies with the same redshift , and for which the redshift space distortion effect is minimum . in a following companion paper
we will present a detailed analysis of all the gr effects and their dependencies on scale , angular separation , bias and radial distribution of sources for all the terms presented in this work , along with predictions for observing these effects with future surveys .
we derived a fully general relativistic expression for correlation functions in the wide - angle case , that will need to be used in future surveys ( such as ska and euclid ) which measure galaxy correlations on very large scales .
our formalism recovers and generalizes previous results in the plane - parallel ( flat - sky ) and newtonian approximations .
we presented new results for the gr - corrected plane - parallel case .
the gr corrections to the newtonian monopole of @xmath54 , for galaxy separations up to @xmath97mpc , are up to 2% ( see fig .
[ pplxi ] ) .
stronger gr corrections arise in the wide - angle case .
we showed via illustrative examples that gr corrections on large scales ( @xmath142gpc separation ) can in principle be of order 510% , as shown in fig .
[ waxi ] . of course
, the observability of these effects is severely degraded by cosmic variance .
future large - volume surveys will reduce the cosmic variance but the problem can in fact be removed if we are able to observe multiple tracers of the underlying matter distribution , as shown in @xcite . in a companion paper we will present a detailed analysis of the predicted gr corrections for large - volume surveys , taking into account cosmic variance .
* acknowledgments : * + db and rm are supported by the south african square kilometre array project .
rm is supported by the stfc ( uk ) ( grant no .
st / h002774/1 ) .
rm and cc are supported by the national research foundation ( nrf , south africa ) .
db , rm and cc are supported by a royal society ( uk)/ nrf ( sa ) exchange grant .
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 .
for ( [ eq : xi_ss ] ) : @xmath143 where @xmath144 , @xmath145 and @xmath146 . for ( [ eq : xi_kk ] ) :
@xmath147 where @xmath148 . for ( [ eq : xi_ii ] ) : @xmath149 where @xmath150 . for ( [ eq : xi_si ] ) : @xmath151 for ( [ eq : xi_sk ] ) : @xmath152 for ( [ eq : xi_ki ] )
: @xmath153 the remaining @xmath68 are @xmath154 where @xmath155},\end{aligned}\ ] ] and the corresponding @xmath67 coefficients are @xmath156 @xmath157 @xmath158
@xmath54 is given in .
the remaining gr @xmath68 have not before been given in the plane - parallel limit .
@xmath164 correlation : @xmath165 \mathcal{p}_{0 } ( { { { \bf n}}}_1 \cdot { { \bf n } } _ { { 1}1 ' } ) \nonumber \\ & & ~- \frac{\beta_1 \alpha_1}{\chi_1 } \xi_1^{3 } ( \chi _ { { 1 } 1 ' } ; { z}_{1 } , z_{1 ' } ) \mathcal{p}_{1 } ( { { { \bf n}}}_1 \cdot { { \bf n } } _ { { 1}1 ' } ) + \frac{2}{3 } \beta_1 \xi_2^{2 } ( \chi _ { { 1 } 1 ' } ; { z}_{1 } , z_{1 ' } ) \mathcal{p}_{2 } ( { { { \bf n}}}_1 \cdot { { \bf
n } } _ { { 1}1 ' } ) \bigg\}\ ; , \end{aligned}\ ] ] where we assumed @xmath166
. then @xmath167 @xmath168 correlation : @xmath169 \mathcal{p}_{0 } ( { { { \bf n}}}_1 \cdot { { \bf n}}_{1'1 } ) \nonumber \\ & & ~+ \frac{\beta_1 \alpha_1}{\chi_1}\xi_1^{3 } ( \chi_{1'1 } ; { z}_{1 ' } , z_{1 } ) \mathcal{p}_{1 } ( { { { \bf n}}}_1 \cdot { { \bf n}}_{1 ' 1 } ) + \frac{2}{3 } \beta_1 \xi_2^{2 } ( \chi_{1'1 } ; { z}_{1 ' } , z_{1 } ) \mathcal{p}_{2 } ( { { { \bf n}}}_1 \cdot { { \bf n}}_{1'1 } ) \bigg\}\ ; , \end{aligned}\ ] ] where we defined @xmath170 .
then @xmath171 @xmath172 correlation : @xmath173 \mathcal{p}_{0 } ( { { { \bf n}}}_1 \cdot { { \bf
n } } _ { { 1}1 ' } ) \nonumber \\ & + & \bigg [ 3 \left(\frac{1}{\chi_{1'}}+\frac{3}{5\,\chi_{1'}}\beta_1 + \frac{1}{5}\frac{\beta_1\alpha_1 } { \chi_1}\right ) \xi_1^{1 } ( \chi _ { { 1 } 1 ' } ; { z}_{1 } , z_{1 ' } ) - \frac{3}{\chi_{1 ' } } \gamma_1\xi_1^{3 } ( \chi _ { { 1 } 1 ' } ; { z}_{1 } , z_{1 ' } ) \bigg ] \mathcal{p}_{1 } ( { { { \bf n}}}_1 \cdot { { \bf n } } _ { { 1}1 ' } ) \nonumber \\ & + & \bigg [ \left(1-\frac{1}{7}\beta_1\right ) \xi_2^{0 } ( \chi _ { { 1 } 1 ' } ; { z}_{1 } , z_{1 ' } ) + \left(2\frac{\beta_1 \alpha_1}{\chi_{1'}\chi_1}-\gamma_1 \right ) \xi_2^{2 } ( \chi _ { { 1 } 1 ' } ; { z}_{1 } , z_{1 ' } ) \bigg ] \mathcal{p}_{2 } ( { { { \bf n}}}_1 \cdot { { \bf
n } } _ { { 1}1 ' } ) \nonumber \\ & + & \frac{3}{5}\beta_1\left(\frac{\alpha_1}{\chi_1}-\frac{2}{\chi_{1'}}\right ) \xi_3^{1 } ( \chi _ { { 1 } 1 ' } ; { z}_{1 } , z_{1 ' } ) \mathcal{p}_{3 } ( { { { \bf n}}}_1 \cdot { { \bf n } } _ { { 1}1 ' } ) - \frac{12}{35}\beta_1\xi_4^{0 } ( \chi _ { { 1 } 1 ' } ; { z}_{1 } , z_{1 ' } ) \mathcal{p}_{4 } ( { { { \bf n}}}_1
\cdot { { \bf n } } _ { { 1}1'})\bigg\}\;. \end{aligned}\ ] ] @xmath174 correlation : @xmath175 \mathcal{p}_{0 } ( { { { \bf n}}}_1 \cdot { { \bf n}}_{1 ' 1 } ) \nonumber \\ & - & \bigg [ 3 \left(\frac{1}{\chi_{1'}}+\frac{3}{5\,\chi_{1'}}\beta_1 + \frac{1}{5}\frac{\beta_1 \alpha_1 } { \chi_{1}}\right ) \xi_1^{1 } ( \chi_{1 ' 1 } ; z_{1 ' } , { z}_{1 } ) - \frac{3}{\chi_{1 ' } } \gamma_1 \xi_1^{(3 ) } ( \chi_{1 ' { 1 } } ; z_{1 ' } , { z}_{1 } ) \bigg ] \mathcal{p}_{1 } ( { { { \bf n}}}_1 \cdot { { \bf n}}_{1'1 } ) \nonumber \\ & + & \bigg [ \left(1-\frac{1}{7}\beta_1 \right ) \xi_2^{0 } ( \chi_{1 ' { 1 } } ; z_{1 ' } , { z}_{1 } ) + \left(2\frac{\beta_1 \alpha_1}{\chi_{1'}\chi_{1}}-\gamma_1 \right ) \xi_2^{2 } ( \chi_{1'1 } ; z_{1 ' } , { z}_{1 } ) \bigg ] \mathcal{p}_{2 } ( { { { \bf n}}}_1 \cdot { { \bf n}}_{1'1 } ) \nonumber \\ & - & \frac{3}{5}\beta_1\left(\frac{\alpha_1}{\chi_{1}}-\frac{2}{\chi_{1'}}\right ) \xi_3^{(1 ) } ( \chi_{1'1 } ; z_{1 ' } , { z}_{1 } ) \mathcal{p}_{3 } ( { { { \bf n}}}_1 \cdot { { \bf n}}_{1'1 } ) - \frac{12}{35}\beta_1\xi_4^{0 } ( \chi _ { 1'1 } ; z_{1 ' } , { z}_{1 } ) \mathcal{p}_{4 } ( { { { \bf n}}}_1 \cdot { { \bf n}}_{1'1})\bigg\}\;. \end{aligned}\ ] ] note that @xmath180 @xmath181 , and @xmath182 unless @xmath183 .
this is connected to the fact that we have already fixed galaxy @xmath184 and galaxy @xmath185 _ a priori _ and by the definition of @xmath186 , when we impose the plane - parallel limit .
indeed the second subscript @xmath184 is the second galaxy and not the first one i.e. in the flat - sky limit @xmath187 .
n. kaiser , mon . not .
soc . * 227 * 1 ( 1987 ) a. j. s. hamilton , astro - ph/9708102 .
a. s. szalay , t. matsubara and s. d. landy , astro - ph/9712007 .
s. bharadwaj , astrophys .
j. * 516 * ( 1999 ) 507 [ astro - ph/9812274 ] . t. matsubara , astro - ph/9908056 .
i. szapudi , astrophys .
j. * 614 * 51 ( 2004 ) [ astro - ph/0404477 ] .
p. papai and i. szapudi , arxiv:0802.2940 .
a. raccanelli , l. samushia and w. j. percival , arxiv:1006.1652 .
l. samushia , w. j. percival and a. raccanelli , arxiv:1102.1014 .
j. yoo , m. zaldarriaga and l. hernquist , phys .
d * 81 * 123006 ( 2010 ) [ arxiv:1005.0847 ] .
j. yoo , phys .
d * 82 * 083508 ( 2010 ) [ arxiv:1009.3021 ] .
n. bartolo , s. matarrese and a. riotto , jcap * 1104 * ( 2011 ) 011 c. bonvin and r. durrer , phys .
d * 84 * , 063505 ( 2011 ) [ arxiv:1105.5280 ] .
a. challinor and a. lewis , phys .
d * 84 * , 043516 ( 2011 ) [ arxiv:1105.5292 ] .
m. bruni , r. crittenden , k. koyama , r. maartens , c. pitrou and d. wands , phys .
rev . d*85 * 041301 ( 2012 ) [ arxiv:1106.3999 ] .
t. baldauf , u. seljak , l. senatore and m. zaldarriaga , arxiv:1106.5507 .
d. jeong , f. schmidt and c. m. hirata , phys .
d * 85 * , 023504 ( 2012 ) [ arxiv:1107.5427 ] .
j. yoo , n. hamaus , u. seljak and m. zaldarriaga , arxiv:1109.0998 .
f. schmidt and d. jeong , arxiv:1204.3625 .
d. jeong and f. schmidt , arxiv:1205.1512 .
f. schmidt and d. jeong , arxiv:1205.1514 . c. p. ma and e. bertschinger , astrophys .
j. * 455 * 7 ( 1995 ) [ arxiv : astro - ph/9506072 ] .
s. matarrese , s. mollerach and m. bruni , phys .
d * 58 * 043504 ( 1998 ) [ arxiv : astro - ph/9707278 ] .
d. a. varshalovich , a. n. moskalev and v. k. khersonsky , _ quantum theory of angular momentum : irreducible tensors , spherical harmonics , vector coupling coefficients , 3nj symbols _ ( world scientific , singapore , 1988 ) . | we extend previous analyses of wide - angle correlations in the galaxy power spectrum in redshift space to include all general relativistic effects .
these general relativistic corrections to the standard approach become important on large scales and at high redshifts , and they lead to new terms in the wide - angle correlations . we show that in principle the new terms can produce corrections of nearly 10% on gpc scales over the usual newtonian approximation .
general relativistic corrections will be important for future large - volume surveys such as ska and euclid , although the problem of cosmic variance will present a challenge in observing this . |
if @xmath2 is a measurable transformation of a lebesgue measure space @xmath3 to itself , that does not preserve the measure @xmath4 , one can study the invariant measures of @xmath2 and compare them to @xmath4 .
a especially interesting case is when @xmath2 is non - singular with respect to @xmath4 ( in the sense that @xmath5 iff @xmath6 ) , but nevertheless there exist no @xmath1-finite invariant measure which is absolutely continuous with respect to @xmath4 .
such maps @xmath2 are called of _ type iii _ ( with respect to the measure ) .
their existence was conjectured by halmos @xcite and established by ornstein @xcite .
other examples were given later ; let us cite a few ( when not specified , the relevant measure is riemannian ) : * piecewise linear homeomorphisms of the circle , by herman @xcite ; * @xmath7-diffeomorphisms of the circle , by katznelson @xcite ; * a @xmath7 non - invertible map of the @xmath8-torus , by hawkins and silva @xcite ; * the full shift on @xmath8 symbols , with respect to some product measure , by hamachi @xcite . * a @xmath0 expanding map of the circle ( constructed using hamachi s example ) , by bruin and hawkins @xcite .
recall that @xmath9 expanding maps have absolutely continuous invariant probability measures , so the regularity of the example of bruin and hawkins is essentially sharp .
the question of whether the absence of aci@xmath1 is actually a generic ( in the usual topological sense ) phenomenon for @xmath0 expanding maps of the circle seems to have been first posed by quas @xcite .
later investigations @xcite indicated that the known methods failed to decide the question either way .
it was also known that @xmath0-generic maps do behave `` pathologically '' in some respects ( they have no absolutely continuous invariant _
measure @xcite ) , but not in others ( they are ergodic and conservative with respect to lebesgue measure @xcite , and they possess a unique physical measure @xcite ) . in this paper we show that the type iii property is indeed @xmath0-generic for expanding maps of the circle .
we also mention that the non - existence of _ finite _ invariant measures that are absolutely continuous with respect to riemannian measure was shown to be a generic property among ( expanding or not ) @xmath0 maps of compact manifolds of any dimension
see @xcite .
of course , it is natural to ask whether the result of the present paper is still true for expanding maps on higher dimension .
it is not clear whether our methods can be extended .
concerning non - necessarily expanding maps of a compact manifold , there are @xmath0-open sets of transformations that do have some absolutely continuous @xmath1-finite invariant measure ( maps with a sink , for instance ) . in this regard , we ask whether a @xmath0-generic map has no absolutely continuous @xmath1-finite invariant measure which is conservative ( all w.r.t .
riemannian measure ) .
we will show that this is true at least for one - dimensional maps , see corollary [ besta ] .
let us now give the precise statements .
let @xmath10 be the circle .
let @xmath11 be the set of all @xmath0 maps @xmath12 which are expanding , i.e. , @xmath13 we endow the set @xmath11 with the @xmath0 topology .
let @xmath14 denote the lebesgue measure on @xmath15 normalized so that @xmath16 .
we say that a @xmath1-finite measure on @xmath15 is an _
aci@xmath1 _ for a map @xmath12 if it is absolutely continuous with respect to @xmath14 and it is @xmath2-invariant . our main result is : [ t.main ] there exists a residual set @xmath17 such that if @xmath18 then @xmath2 has no aci@xmath1 .
[ besta ] let @xmath19 be either the circle @xmath15 or the compact interval @xmath20 $ ] .
there exists a residual set @xmath21 of the space of @xmath22 such that if @xmath23 then @xmath2 has no conservative aci@xmath1 .
hyperbolic maps form an open and dense subset @xmath24 of @xmath22 by @xcite .
( see also @xcite for the recent extension to higher regularity.)the map that associates to @xmath25 its non - wandering set @xmath26 is upper semi - continuous in the hausdorff topology .
moreover , if @xmath27 then @xmath28 unless @xmath29 and @xmath2 is expanding , see @xcite .
it follows that generically , either @xmath28 or @xmath30 . in the first case
, @xmath2 can not have a conservative aci@xmath1 ( since any conservative measure must be supported on the non - wandering set ) . in the second case , generically there is no aci@xmath1 at all , by theorem [ t.main ] .
let @xmath31 be the subset of @xmath11 consisting of maps @xmath2 satisfying @xmath32 and such that holds with @xmath33 .
we will actually prove : [ t.reduced ] there exists a residual set @xmath34 such that if @xmath35 then @xmath2 has no aci@xmath1 .
let us show that theorem [ t.reduced ] implies theorem [ t.main ] .
let @xmath30 , and let @xmath36 be a fixed point of @xmath2 .
then for @xmath37 in a small open neighborhood @xmath38 of @xmath2 in @xmath11 , there exists a unique fixed point @xmath39 of @xmath37 near @xmath36 , moreover this fixed point depends continuously on @xmath37 .
let @xmath40 be such that @xmath41 for every @xmath42 .
define @xmath43 where @xmath44 is the degree of @xmath2 .
let @xmath45 be the orientation preserving @xmath0 diffeomorphism of @xmath15 such that @xmath46 and @xmath47 .
then @xmath45 depends continuously on @xmath48 .
define a map @xmath49 . for any @xmath42 ,
writing @xmath50 , we have @xmath51 hence @xmath52 .
shrinking @xmath38 to a smaller open neighborhood of @xmath2 if necessary , we see that @xmath53 belongs to @xmath31 for every @xmath54 .
consider the mapping @xmath55 ; it is clearly continuous .
it is also open : for any @xmath56 close to @xmath57 , @xmath58 is close to @xmath2 and is mapped by @xmath59 to @xmath60 .
the preimage of @xmath61 under @xmath59 is thus a residual subset of @xmath38 , which contains only maps which have no aci@xmath1 .
let @xmath62 be the set of lipschitz local homeomorphisms @xmath12 , such that @xmath32 and @xmath63 we consider @xmath62 endowed with the topology induced from the lipschitz metric : @xmath64 we also let @xmath65 the distortion of the restriction of some iterate of @xmath2 to some interval is @xmath66 clearly , if @xmath67 is 1 - 1 onto @xmath68 then for every measurable @xmath69 of positive measure , @xmath70 we say that @xmath71 is _ ergodic _ with respect to @xmath14 if every measurable set @xmath19 such that @xmath72 satisfies @xmath73 or @xmath74 .
we say that @xmath71 is _ conservative _ with respect to @xmath14 if every measurable set @xmath19 such that @xmath75 satisfies @xmath73 .
it is easy to see that @xmath71 is ergodic and conservative with respect to @xmath14 if and only if for every measurable set @xmath19 which is forward invariant ( that is , @xmath76 ) we have @xmath73 or @xmath74 .
if @xmath71 and @xmath77 is a measurable set , then we denote @xmath78 the set of points in @xmath19 that return to @xmath19 by forward iteration by @xmath2 .
it is easy to see that if @xmath2 is conservative then @xmath79 .
we denote by @xmath80 the first return map .
[ l.preserve ] if @xmath81 is any @xmath2-invariant @xmath1-finite measure then @xmath82 for all measurable @xmath83 .
we can assume @xmath84 .
for @xmath85 , let @xmath86 then @xmath87 . since @xmath88
, we have @xmath89 therefore @xmath90 .
let @xmath71 .
the points in @xmath91 divide the circle into @xmath92 open intervals , which are called _ markov intervals of order @xmath93_. the image of a markov interval of order @xmath93 is a markov interval of order @xmath94 . if @xmath68 is a markov interval of order @xmath93 then @xmath95 is a 1 - 1 map onto @xmath96 .
if @xmath68 is a markov interval and @xmath97 is the first return map to @xmath68 , then there exist disjoint ( markov ) intervals @xmath98 such that @xmath99 is an homeomorphism onto @xmath68 for each @xmath100 .
if @xmath2 is conservative with respect to lebesgue measure then the intervals @xmath101 cover @xmath68 @xmath14-mod @xmath102 .
let us say that a map @xmath71 belongs to @xmath103 if there exists @xmath40 such that for every markov interval @xmath68 of order @xmath93 , @xmath104 is linear .
[ l.eplisdense ] the set @xmath103 is dense in @xmath31 in the lipschitz metric . given @xmath105 and @xmath106 , we define a map @xmath107 as follows : for each markov interval @xmath68 of order @xmath93 for @xmath2 , let @xmath108 map @xmath68 onto @xmath109 linearly , and so that @xmath108 equals @xmath2 in the boundary of @xmath68 .
clearly , @xmath110 .
we claim that @xmath111 in the lipschitz metric .
notice that the lengths of markov intervals @xmath68 of order @xmath93 go uniformly to @xmath102 as @xmath112 : in fact , @xmath113 . since @xmath2 is @xmath0 , for every @xmath114 we can choose @xmath115 such that in each markov interval @xmath68 of order @xmath116 , @xmath117
. then @xmath118 .
[ l.linearisergodic ] every @xmath119 is ergodic and conservative with respect to lebesgue measure .
given @xmath119 , let @xmath93 be such @xmath2 is linear on markov intervals of order @xmath93 .
let @xmath19 be a forward invariant set of positive lebesgue measure .
notice that if @xmath68 is a markov interval of order @xmath120 , then @xmath121 is linear onto some markov interval of order @xmath93 , and @xmath122 is onto @xmath96 with distortion bounded by @xmath123 . by the lebesgue density points theorem , for almost every @xmath124 , @xmath125 , where @xmath126 is the markov interval of order @xmath127 containing @xmath128 . applying @xmath129 and using the bound on the distortion
, we see that @xmath130 as @xmath131 .
so @xmath74 and the result follows .
[ d.distorted ] let @xmath68 be an interval and @xmath132 non - singular ( w.r.t .
the lebesgue measure ) map .
then @xmath133 is called _ distorted _ if there exists a measurable set @xmath134 such that @xmath135 , and @xmath136 on @xmath137 ( where @xmath138 is lebesgue measure on @xmath68 ) .
recall that @xmath139 [ d.good ] if @xmath114 and @xmath71 then @xmath2 is called _
@xmath140-good _ if there exists a family @xmath141 of markov intervals ( possibly of different orders ) , all of length at most @xmath140 , such that @xmath142 and for every @xmath143 , the first return map @xmath144 is distorted . .
in this example , @xmath137 is an interval .
there should be infinitely many branches.,width=151 ] we can now state the three key technical results of this paper .
[ p.criterium ] if @xmath145 is @xmath140-good for every @xmath114 then @xmath2 has no aci@xmath1 .
the next proposition says the condition of being @xmath140-good is open in two senses .
[ p.open ] let @xmath146 be @xmath140-good for some @xmath114 .
then : 1 .
there exists @xmath147 such that if a map @xmath148 satisfies @xmath149 then it is @xmath140-good .
2 . assuming that @xmath150 , there exists @xmath151 such that if @xmath152 satisfies @xmath153 then @xmath37 is @xmath140-good .
[ p.dense ] for any @xmath114 , the set of maps that are @xmath140-good is dense in @xmath31 with the lipschitz metric .
let us first see how to conclude theorem [ t.reduced ] ( and hence , by the reduction , theorem [ t.main ] ) from the three propositions .
the set @xmath154 of @xmath145 which are @xmath140-good is @xmath0-open , by part ( ii ) of proposition [ p.open ] , and @xmath0-dense , by proposition [ p.dense ] .
so @xmath155 is a residual set of @xmath31 , and by proposition [ p.criterium ] it consists of maps which do not have an aci@xmath1 .
we now prove propositions [ p.criterium ] and [ p.open ] , and leave the harder proof of proposition [ p.dense ] ( where part ( i ) in proposition [ p.open ] is used ) for the next section .
assume that for all @xmath114 , @xmath30 is @xmath140-good ; let @xmath156 be the corresponding family of intervals as in definition [ d.good ] .
assume @xmath2 has an aci@xmath1 @xmath81 , and let @xmath157 be its density .
let @xmath158 be such that the set @xmath159 has positive lebesgue measure .
by a density point argument , there exists an interval @xmath160 such that @xmath161 .
since @xmath97 is distorted , there exists @xmath134 such that @xmath162 and @xmath163 on @xmath137
. let @xmath164 .
we have @xmath165 and @xmath166 moreover , @xmath167 \\
> c [ 2 \ , m(y ) - .1 \ , m(i ) ] \ge c [ 2 \times .3 - .1 ] m(i ) = .5 c \ , m(i ) > \mu(y).\end{gathered}\ ] ] this contradicts lemma [ l.preserve ] .
let @xmath141 be the family of markov intervals as in definition [ d.good ] ; clearly we can assume it is finite , say , @xmath168 .
let @xmath169 be the set that gets enlarged under @xmath170 according to definition [ d.distorted ] .
let @xmath171 , @xmath172 , be the connected components of the domain of @xmath173 , and let @xmath174 be such that @xmath175 . then @xmath176 slightly reducing the sets @xmath177 ( still keeping @xmath178 ) , we can find @xmath179 such that @xmath180 where @xmath181 is some fixed positive number .
also let @xmath182 .
if @xmath37 is another map in @xmath62 which is @xmath183-close to @xmath2 , then for each interval @xmath184 there is an interval @xmath185 which is markov for @xmath37 and is close to @xmath186 .
clearly if the @xmath183-distance between @xmath37 and @xmath2 is sufficiently small then each @xmath185 has length @xmath187 , and their union has measure @xmath188 .
further , for each @xmath185 there exist intervals @xmath189 , @xmath190 , which are close to @xmath191 and such that @xmath192 . with these notations fixed ,
we complete the proofs of the two parts of the proposition separately .
_ part ( i ) : _ let @xmath148 so that the set @xmath193 has @xmath194 , where how small @xmath195 needs to be will become clear along the way . first , notice that the @xmath183-distance between @xmath37 and @xmath2 is small ( in fact , less than @xmath196 ) .
so we can define intervals @xmath185 , @xmath189 , for @xmath197 and @xmath190 as explained above .
let @xmath198 .
then @xmath199 is small .
define @xmath200 .
then @xmath201 is small : at most @xmath202 .
so if @xmath195 is small enough then @xmath203 .
moreover , if @xmath204 then @xmath205 .
this shows that @xmath206 is distorted for each @xmath207 and accordingly that @xmath37 is @xmath140-good . _
part ( ii ) : _ now assume @xmath2 is @xmath0 and @xmath37 is @xmath208-@xmath0-close to @xmath2 .
again we can define intervals @xmath185 , @xmath189 , for @xmath197 and @xmath190 .
let @xmath209 . by taking a small @xmath208 ,
we guarantee that @xmath210 is @xmath0-close to @xmath211 for @xmath212 , and therefore @xmath213 so the @xmath206 are distorted and and @xmath37 is @xmath140-good .
we remark that with a little more effort it is possible to improve simultaneously the two parts of proposition [ p.open ] , showing that being @xmath140-good is an open condition in @xmath62 in the _ bounded variation metric _ @xmath214 .
let @xmath215 and @xmath114 ; we will show that there exists a @xmath140-good map @xmath216 such that @xmath217 . for simplicity
, we will assume that @xmath2 is orientation - preserving .
the proof can be easily adapted to cover the general case .
let @xmath220 be large ( to be specified later ) .
fix a markov interval @xmath221 such that the sets @xmath221 , @xmath222 , , @xmath223 are disjoint , and their union has lebesgue measure less than @xmath140 . let @xmath224 be the collection of ( markov ) subintervals of @xmath221 that are sent onto @xmath221 by @xmath225 .
let @xmath226 . notice that for any @xmath227 , @xmath228 , @xmath229 and @xmath230 is linear .
* @xmath57 equals @xmath2 outside @xmath237 .
* let @xmath238 .
consider all sequences of intervals @xmath239 with @xmath240 and @xmath241 .
let @xmath242 .
then @xmath243 . let @xmath244 .
we define @xmath245 for @xmath246 as the unique orientation preserving homeomorphism onto @xmath247 whose restriction to each connected component of @xmath248 is linear and such that @xmath249 . let @xmath250 .
it is the homeomorphism of @xmath96 depicted in figure [ f.q ] . 1 .
either @xmath256 is a singularity for @xmath2 in this case it is contained in @xmath257 ; 2 . or
@xmath256 belongs to @xmath258 for some @xmath246 ; 3 . or @xmath256 belongs to @xmath259 for some @xmath246 .
if @xmath266 is not linear then @xmath267 intersects a singularity @xmath256 of @xmath57 for some @xmath268 . since @xmath269 belongs to @xmath261 , there are at least three different @xmath270 such that @xmath271
. thus @xmath256 can not be of the type ( i ) above : indeed , the set @xmath272 is forward invariant for both @xmath2 and @xmath57 , and it does not intersect @xmath221 .
@xmath256 can not be of the type ( ii ) : the first iterate of @xmath256 that belongs to @xmath221 belongs indeed to @xmath273 , and subsequent iterates do not enter @xmath221 again .
@xmath256 can not be a singularity of type ( iii ) : the first iterate of @xmath256 that belongs to @xmath221 also belongs to @xmath274 , so the second iterate that belongs to @xmath221 is @xmath275 , and the subsequent iterates lie outside @xmath221 .
so there can be no such singularity , and the result follows .
take @xmath277 .
let @xmath278 be such that @xmath279 .
then @xmath280 is linear , and hence @xmath281 .
this implies the assertion of the lemma for @xmath282 and @xmath283 .
now , if @xmath284 and and @xmath285 then , by lemma [ l.g ] , @xmath286 is linear , so the assertion also follows .
assume that @xmath288 has zero lebesgue measure . by a density point argument
, we can take a @xmath57-markov interval @xmath260 such that @xmath289 is close to @xmath290 .
any forward - image @xmath291 can not be contained in the set @xmath292 . by the markov property , @xmath293 for @xmath294 .
since @xmath57 equals @xmath2 outside @xmath295 , @xmath296 is linear . in particular
, @xmath297 it follows that @xmath298 is close to @xmath290 .
this shows that the assumption @xmath299 can not be true . since @xmath2 is conservative ( lemma [ l.linearisergodic ] ) , @xmath14-almost every point in @xmath221 returns to @xmath221 by forward iterates of @xmath2 .
it follows that the same is true for @xmath57 .
so the intervals in @xmath253 cover @xmath221 @xmath14-mod @xmath102 .
moreover , these intervals have lengths at most @xmath300 . therefore we can find @xmath260 and @xmath301 such that @xmath285 and @xmath289 is arbitrarily close to @xmath290 . by lemma [ l.distortion ] , @xmath302 is arbitrarily close to @xmath290 .
it follows that @xmath303 .
so @xmath304 has full @xmath14-measure on the circle .
we have to show that for each @xmath143 , @xmath314 is distorted . for this
, it is enough to prove that ( where @xmath315 is the collection of intervals @xmath316 that are sent onto @xmath68 by @xmath314 ) : we claim that @xmath324 is linear .
this is clear if @xmath325 .
notice that @xmath326 is an element of @xmath224 : indeed , @xmath327 takes @xmath326 onto @xmath221 .
this implies that @xmath328 is an element of @xmath329 . by lemma [ l.g ]
, @xmath330 is linear , therefore @xmath324 is linear , as claimed . for @xmath77 and @xmath333 ,
let @xmath334 be the @xmath181-neighborhood of @xmath19 .
let @xmath335 be the ( already described ) set of singularities of @xmath57 ; then @xmath335 is closed in @xmath15 and @xmath336 . in particular , @xmath337 as @xmath338 .
let @xmath339 be the lift of @xmath57 satisfying @xmath340 .
for @xmath341 , let @xmath342 then @xmath343 is the lift of some @xmath0 map @xmath344 such that for every @xmath42 , @xmath345 ( recall @xmath346 is the original @xmath0 map . ) hence @xmath347 for every @xmath42 if @xmath348 is sufficiently small .
moreover , since @xmath57 is linear on connected components of @xmath349 , @xmath350 equals @xmath57 outside the @xmath348-neighborhood of @xmath335 .
unfortunately , @xmath350 does not necessarily fix @xmath102 . to remedy that ,
take a family of @xmath0 diffeomorphisms @xmath351 parameterized by @xmath352 , such that @xmath353 define @xmath354 .
if @xmath348 is small then @xmath355 and @xmath356 .
also , @xmath357 . by part ( i ) of proposition [ p.open ] , @xmath358 is @xmath140-good provided @xmath348 is small enough .
this concludes the proof of proposition [ p.dense ] .
this research was partially conducted during the period a.a
. served as a clay research fellow .
the paper was written while j.b .
was visiting impa , partially supported by procad / capes project . j.b .
thanks philippe thieullen for providing several references and for stimulating conversations . | we show that a @xmath0-generic expanding map of the circle has no absolutely continuous invariant @xmath1-finite measure . |
the kibble
zurek mechanism@xcite ( kzm ) aims at the description of kinetically governed phase transitions induced by rapid decrease of the temperature .
kzm may be didactically expressed as the division of the transition into three stages , with underlying postulates for each stage . during the initial stage
the system is in a high symmetry phase ; it explores lower symmetry states due to thermal fluctuations , but is symmetric on the average .
the system is able to follow the change of the temperature and may be considered to remain in thermal equilibrium . during the intermediate stage the critical temperature is crossed . during this stage
the system is sluggish , so that its final state will be very similar to the initial state . in the final stage
several lower symmetry states are long lived ; the selected state will have symmetry properties that are determined by the state that was randomly occupied by the system when it entered the second stage .
kzm has been tested by many experiments and simulations , and various degrees of agreement have been obtained
. we will be interested in the case of systems with loop topology.@xcite in this paper we consider a superconducting loop cooled at a fast rate from slightly above to slightly below its critical temperature , in the absence of applied magnetic field , and investigate the possibilities for spontaneous emergence of a permanent supercurrent .
the first postulate specifies the temperature at which the system becomes sluggish : the crossover from the initial to the intermediate stage occurs when @xmath4 . here
@xmath5 is the reduced distance from the critical temperature , @xmath6 is the derivative of @xmath7 with respect to time , and @xmath8 is the relaxation time of the system .
the following postulates are formulated here for the case of a system with loop geometry , described by a complex order parameter .
the moment at which @xmath9 will be taken as @xmath10 and the moment at which the system passes from the initial to the intermediate stage will be denoted as @xmath11 .
[ p2 ] we denote by @xmath12 the coherence length of the order parameter at @xmath11 and by @xmath13 the perimeter of the loop . the behavior of the system can be estimated by enviewing it as if it were divided into @xmath14 pieces , such that in each piece the phase of the order parameter is uniform , and such that there is no correlation between the phases in different pieces .
let us denote by @xmath15 the phase difference between regions @xmath16 and @xmath17 .
it follows from postulate [ p2 ] that the variance of this phase difference is @xmath18 .
let @xmath19 be the phase difference accumulated around the loop .
since @xmath20 for @xmath21 , if we ignore the constraint that @xmath22 has to be an integer multiple of @xmath23 , we obtain _ c^2=^2/3 .
[ vartetc ] it turns out that the constraint has no effect . taking the @xmath15 s uncorrelated for @xmath24 and picking @xmath25 as the phase difference with the smallest absolute value such that @xmath26 is integer , eq .
( [ vartetc ] ) is recovered for @xmath27 .
@xmath28 is the winding number of the order parameter , and in view of eq .
( [ vartetc ] ) obeys n^2=/12 .
[ varn ] the winding number can be regarded as a topological invariant that remains unchanged for @xmath29 .
this postulate is far from plausible .
although @xmath28 is a discrete variable , there is no impediment for a continuous passage of the order parameter to a different winding number , provided that it vanishes at some point .
the idea behind postulate 3 is that during the period @xmath30 the system is sluggish and its order parameter practically does not change , and after @xmath31 there is an energy barrier between states with different winding number .
this statement would be true if at @xmath31 the order parameter were close to local equilibrium , which most probably will not be the case .
since the order parameter is random and nearly vanishes at @xmath11 , the same is true at @xmath31 , and we have no good reason to expect a considerable energy barrier for the passage to a different winding number .
postulate 3 has been amended in later publications ( e.g. ref . ) .
the revised claim is that ( i ) there is a time @xmath32 at which the order parameter shoots up sharply ; ( ii ) @xmath28 remains constant after @xmath32 ; ( iii ) @xmath33 and @xmath32 is of the order of @xmath34 ; ( iv ) since the coherence lengths are the same for @xmath35 and @xmath36 , it follows from ( i)(iii ) that the scaling properties of the system will be the same as if postulate 3 were true . most of the literature on kzm studies the scaling of the density of defects ( the winding number in our case ) with the rate at which the system is cooled . for this purpose ,
postulate 1 is of central importance . in the present study we will first consider the case in which the system is cooled at an ideally fast rate and
will check the applicability of postulates 2 and 3 ; verification of postulate 1 will be postponed to sec .
a simple approach for the description of the dynamics of a superconducting sample is the time - dependent ginzburg landau model ( tdgl ) , with the addition of langevin terms that bring thermal fluctuations into account ( e.g. ref . ) ; in the case of 1d systems , simplifications are possible.@xcite we assume that the magnetic field induced by the current around the loop is negligible and write the ginzburg
landau energy as g=_c wds , [ energy ] where @xmath37 is the mass of an electron pair , @xmath38 is the arc length and the integral covers the loop , @xmath39 is the coherence length at @xmath40 , @xmath41 is the order parameter , @xmath42 is the tangential component of the electromagnetic vector potential , @xmath43 is the quantum of flux , @xmath44 is the cross section of the loop and @xmath45 , where @xmath46 is the electron charge , @xmath47 is the ginzburg landau parameter and @xmath48 the speed of light . the usual coherence length in the ginzburg landau model is @xmath49 .
we assume that @xmath41 and @xmath42 only depend on time and @xmath38 , and not on the lateral position .
@xmath42 is required in the energy functional in order to take account of the fluctuations of the electric field . in the numerical procedure ,
the loop is divided into @xmath50 segments and the integral is approximated by the sum@xcite @xmath51w_j \ ; , \label{discrete}\end{aligned}\ ] ] where the segment @xmath52 is identified with the segment 1 , and the evolution of @xmath53 and @xmath54 during a period of time @xmath55 is given by @xmath56 here @xmath57 , @xmath58 , @xmath59 is the diffusion coefficient and @xmath60 is the conductivity , @xmath61 , @xmath62 and @xmath63 are random numbers with gaussian distribution , zero average and variances @xmath64 , @xmath65 , where @xmath66 is the boltzmann constant .
the material parameters in the model can be evaluated in terms of @xmath0 , the fermi wavevector @xmath67 , the mean free path @xmath68 and the electron density @xmath69 . using bcs , dirty limit and free electron gas approximations , ^2(0)=,^2=0.021,d= , = .
[ bcs ] tdgl is not expected to provide a quantitative description of the superconductor dynamics in the entire range to which we will apply it . nevertheless , it is a self - consistent model from the point of view of statistical mechanics , so that its predictions have at least theoretical value .
an important characteristic length is@xcite @xmath70 . for @xmath71
, @xmath72 is the order of the length over which @xmath41 is not expected to vanish for a fluctuation such that @xmath73 . in order to have a 1d situation ,
@xmath72 and @xmath74 should be larger than the linewith and the thickness . in the absence of significant magnetic fields ,
the magnetic penetration depth @xmath75 is unimportant .
we note that for the present model the relaxation time does not diverge at @xmath0 .
first , relaxation of the electromagnetic potential is insensitive to the distance to @xmath0 , and indirectly drives the phase of @xmath41 .
moreover , for @xmath9 the remaining terms in eq .
( [ energy ] ) lead to a residual relaxation time of the order of @xmath76 .
since this postulate is intended to be applied at a moment such that the system is still able to follow the temperature variation , in this section we consider systems that are allowed to reach thermal equilibrium at some fixed reduced temperature @xmath77 , and check the result predicted by eq .
( [ varn ] ) .
we studied rings of length @xmath78 cm and electron density @xmath79@xmath80 , divided into 300 computational cells .
the system evolved during a lapse of time @xmath81 , in steps of @xmath82 . at the end of this evolution
the winding number @xmath83 was evaluated .
for each temperature above ( respectively below ) @xmath0 this procedure was repeated 800 ( respectively 400 ) times .
the evolution time seemed to be sufficient , since there was no significant correlation beween the initial and the final winding number ; besides , except for temperatures considerably below @xmath0 and unrealistic initial distributions , longer runs lead to essentially the same results .
we studied both the case @xmath84 and the case @xmath85 .
the initial distribution of the order parameter was taken from a quadratic approximation , as described in the appendix .
figure [ uniform ] shows our results .
we observe that ( i ) @xmath86 is a universal function of @xmath87 and @xmath88 ; the rhombs and the circles in the figure , that were obtained for samples with different parameters but the same value of @xmath88 , lie along the same curve ; ( ii ) for @xmath89 , @xmath86 is surprisingly close to @xmath90 ; ( iii ) for @xmath84 and @xmath91 , @xmath86 does not depend significantly on @xmath88 ; ( iv ) as the temperature decreases below @xmath0 , @xmath86 decreases until the probability for @xmath92 becomes negligible . the fact that below @xmath0 @xmath86 decreases as @xmath93 increases indicates that @xmath94 is not necessarily the characteristic length over which the phase remains essentially unchanged .
the relevant length ought to be the correlation length , which can be defined as follows : we define the autocorrelation function @xmath95\rangle/\langle |\psi |^2\rangle$ ] , and then the correlation length @xmath96 is obtained from a fit @xmath97 .
@xmath96 and @xmath98 are not equivalent concepts for two reasons .
first , @xmath98 arises from a competition between the stiffness and the condensation energy , whereas in the case of @xmath96 also the thermal energy and the quartic term enter the competition .
more fundamentally , @xmath98 is a healing length , which is relevant when @xmath41 has to obey constraints , e.g. , it has to vanish at every vortex ; in the problem we are considering , the only constraint is periodicity . the lower inset in fig .
[ sk ] shows the temperature dependence of @xmath99 , scaled by @xmath88 ( for two samples ) , and the higher inset compares the temperature dependences of @xmath99 and @xmath98 . sufficiently above @xmath0 , @xmath100 . near @xmath0
, @xmath99 does not diverge ; @xmath101 is of the order of @xmath88 .
contrary to @xmath98 , @xmath99 is a monotonic function and rises sharply as the temperature decreases below @xmath0 .
if @xmath96 is not significantly smaller than @xmath13 , then the autocorrelation is influenced by the connectivity of the loop and @xmath102 is not well fitted by an exponential function .
the main panel in fig .
[ sk ] shows @xmath86 as a function of @xmath103 . within our numerical accuracy
we obtain a universal line for both samples , which is fitted , both below and above @xmath0 , by @xmath104 , i.e. , @xmath105 is the effective length over which the phase is essentially uniform .
we now study a loop with @xmath106 weak links .
for a situation in which the phase is fairly uniform between weak links but uncorrelated across them , we could argue that the present case is equivalent to that of eq .
( [ varn ] ) , with @xmath107 .
we consider a temperature that is above @xmath0 and very close to it . between weak links we took @xmath108@xmath109 , @xmath110 cm , @xmath111k , @xmath112 ; the other parameters
were taken as in the previous section .
the weak links had lengths @xmath113 , had a central part with @xmath114 ( i.e. , had lower @xmath0 ) , and had shorter mean free paths .
the outer parts of the weak links had positive values of @xmath35 , chosen to compensate the influence that the weak links would have on the rest of the loop due to proximity .
this time we took @xmath115 . in the evaluation of the winding number , the weak links were skipped . the dashed line in fig .
[ weak ] corresponds to the ideal case @xmath107 .
taking into account that the weak links are not perfect would lead to @xmath116 , where @xmath117 is some opacity coefficient .
" we also have to take into account that , according to the previous section , along the length @xmath118 of the strong part of the loop , @xmath119 divisions into different regions effectively occur . therefore , instead of eq .
( [ varn ] ) we expect n^2=c/12_+(p-_w/_)_w/12 .
[ varnw ] the moderately weak links in fig .
[ weak ] had length @xmath120 cm , mean free path @xmath121 cm , and the region with low @xmath0 had length @xmath122 cm ; for these weak links we obtained @xmath123 . the weaker links in the figure had length @xmath124 cm , mean free path @xmath125 cm , and the region with low @xmath0 had length @xmath126 cm ; for these weak links we obtained @xmath127 .
the equilibrium distributions that were obtained in sec . [ vp2 ] will now be taken as initial distributions , and we will study under what conditions the winding number remains unchanged if the system is _ instantaneously _ cooled to a new temperature @xmath128 and left at this temperature for a considerable time . since @xmath129 may be shorter than the linewidth or the thickness of the loop , our first concern should be whether the system can still be treated as 1d . following our findings in sec .
[ unif ] , we may expect that the relevant requirement is @xmath130 .
we will assume here that the 1d treatment is still applicable ; deviations from 1d will be studied in the following section .
it is known that in the absence of magnetic flux a state with winding number @xmath28 such that @xmath131 is locally unstable ( e.g. ref . ) , so that postulate 3 necessarily fails in this situation . in view of postulate 2 and eq .
( [ varn ] ) , this situation can be avoided provided that @xmath132 .
we should require that no further changes of the winding number occur after a state close to a local minimum of @xmath133 is reached .
for this effect , the energy barrier for the passage to another winding number should be significantly larger than @xmath134 .
it follows from expression ( [ energy ] ) that , near a local minimum , the energy per unit length is @xmath135 . in order to change the winding number ,
the order parameter has to vanish at some point and the smallest possible energy increment involved is obtained when the order parameter becomes small in a region of length @xmath136 . from here
, @xmath137 has to be significantly larger than @xmath134 , a condition equivalent to the requirement that @xmath138 be significanly less than @xmath72 .
provided that at time 0 ( i ) the sample is described by an equilibrium distribution with temperature @xmath139 , ( ii ) its winding number is @xmath92 and ( iii ) the temperature is instantaneously lowered to @xmath140 , we define the conservation probability " @xmath141 by @xmath142 the condition of invariance is required in a strict sense , i.e. , @xmath143 for all @xmath144 . according to postulate 3 ,
@xmath141 ought to be 1 . the markers in fig . [ conserven23 ]
show the conservation probability for the sample represented by squares in fig .
[ uniform ] , as a function of the initial reduced temperature @xmath35 . in the case of the black markers , @xmath145 , so that the final energy barriers are expected to be considerably larger than @xmath134 .
we note that @xmath141 is substantially less than 1 .
it also appears that , as the initial temperature increases above @xmath0 , @xmath141 decreases exponentially down to practically zero within a very narrow range .
the red markers are for higher @xmath146 [ @xmath147 .
it would seem that the energy barriers are still safely high , but the conservation probability for the case @xmath148 is quite lower than in the case @xmath149 . in order to estimate the amount by which @xmath141 decreases due to thermal jumps over the energy barriers ,
we repeated some simulations with thermal fluctuations turned off at @xmath10 ( blue and purple markers ) ; the absence of fluctuations leads to conservation probabilities that are larger by about @xmath150 .
we therefore conclude that thermal fluctuations are not the main reason for the change of the winding number ; @xmath28 usually changes just because the initial order parameter is not close to a local minimum of @xmath133 , and is not within the basin of attraction of the local minimum that has winding number equal to @xmath151 .
the lower conservation probabilities for small @xmath152 could be due to insufficiently fast growth of the order parameter .
figure [ conswt ] shows the dependence of @xmath141 on @xmath44 ( blue and green markers ) and on @xmath0 ( red and black markers ) , for a range somewhat larger than an order of magnitude that includes the sample studied in fig .
[ conserven23 ] , with initial temperature close to @xmath0 . as a general trend
, we may regard @xmath141 as independent of @xmath0 and proportional to @xmath153 . extrapolating this result
we would obtain @xmath154 for @xmath155@xmath109 .
close to the parameters in fig .
[ conserven23 ] , @xmath141 is also roughly proportional to @xmath156 and to @xmath157 ( not shown ) .
in addition to the general trends , we have also noticed ( but not systematically investigated ) a non monotonic detailed structure in the dependence of @xmath141 on @xmath35 and on the sample parameters . this dependence may be due to the availability of additional channels when the energy landscape varies . at the moment that the sample is ( instantaneously )
quenched , the order parameter may or may not be within the basin of attraction of the local minimum of the energy with the same winding number .
if it is ( respectively , is not ) , we may regard this situation as a sort of metastable " ( respectively unstable " ) state .
we may expect that unstable states decay very fast , so that after waiting a short time the winding number could indeed be conserved . with these heuristic ideas in mind ,
we define [ under the same conditions as eq .
( [ p ] ) ] the restricted conservation probability " @xmath158 by @xmath159 a weak version of postulate 3 ( wp3 ) is then : _ for an appropriate short time @xmath160 , @xmath161_. figure [ pr ] shows the restricted conservation probability as a function of @xmath35 for several samples and choices for @xmath160 .
the initial temperatures were taken above @xmath0 .
the black markers are for @xmath162 and the colored markers for shorter waiting times .
the squares are for the same sample as the squares in fig .
[ uniform ] ; all the other markers are for samples in which one of the parameters was changed by a factor 10 .
figure [ pr ] indicates that wp3 is more pervasive than could be expected from a non quantitative statement ; except for the sample with small cross section , @xmath163 is indistinguishable from 1 in the entire studied range .
note that @xmath160 for the black markers is smaller by three orders of magnitude than the following period of time during which the winding number is monitored for possible changes at the final temperature .
let us model the cross section of the loop as a rectangle of length @xmath164 ( i.e. , @xmath164 is the width of the loop ) .
how small @xmath164 has to be in order to justify the 1d treatment ?
it is usually assumed that @xmath164 has to be smaller than the coherence length and the pearl length
. these would be the appropriate requirements if there were a significant magnetic field and the temperature were sufficiently far from @xmath0 ; we claim that in the present problem the relevant requirement is @xmath165 , where @xmath96 is the correlation length defined in sec .
[ unif ] .
it might be argued that @xmath96 was evaluated under the assumption that the system is 1d , so that adopting the same correlation length for the lateral direction is a sort of circular reasoning . as a test for our claim
, we consider a simple model in which the lateral direction is taken into account .
we divide the loop into two halves , each having length @xmath13 and width @xmath166 .
we denote the order parameters in the respective halves by @xmath167 and @xmath168 and approximate the lateral component of the gradient of @xmath41 by @xmath169/\delta$ ] . ignoring the lateral component of the electromagnetic potential , which amounts to ignoring johnson noise in the lateral direction ,
the energy becomes @xmath170\right .
\nonumber \\ & & \left .
-\frac{8}{\delta^2}\left(\psi_{(1)}\psi_{(2)}^*+\psi_{(2)}\psi_{(1)}^*\right)\right\}wds \;. \label{gq2d}\end{aligned}\ ] ] let us write @xmath171 and consider the average @xmath172\rangle$ ] ( statistical average , for arbitrary @xmath38 ) . if @xmath172\rangle\approx 1 $ ] , it means that the phase variation is practically the same in each of the halves of the loop and the 1d treatment is justified ; if @xmath172\rangle\approx 0 $ ] , there is no correlation between the phase variations in each of the halves , and we have a qualitatively different problem .
we note that the energy @xmath173 has the same structure as that of a two - band superconductor , and a procedure for the evaluation of @xmath172\rangle$ ] is available.@xcite figure [ deltacos ] shows the temperature dependence of @xmath172\rangle$ ] for the sample represented by circles in fig .
[ uniform ] .
when we take @xmath174 ( filled symbols ) , there are temperature ranges where @xmath172\rangle\approx 0 $ ] , indicating that a loop of width @xmath87 can not be treated as 1d ; there are also ranges where @xmath172\rangle\approx 1 $ ] , indicating that wider samples could still be treated as 1d .
on the other hand , taking @xmath175 ( empty red symbols ) leads to intermediate correlations for all the considered cases , showing that this is the verge of the 1d behavior .
there is a moderate tendency for a decrease of @xmath172\rangle$ ] for @xmath175 as the temperature is lowered , but , in view of our results in the lower inset in fig .
[ sk ] , we still conclude that the admissible width for 1d treatment does not decrease as temperature decreases below @xmath0 .
it follows that if the order parameter is initially 1d , thermalization at @xmath146 will not take the system away from one dimensionality , even if @xmath176 . in the previous sections we entirely neglected the magnetic field induced by the current @xmath177 that circulates around the loop . for a quasi 1d loop
, the influence of this magnetic field can be taken into account by means of a self inductance @xmath178 that gives rise to a magnetic flux @xmath179 enclosed by the loop .
self inductance restricts the rate at which current can vary in a circuit to the characteristic time @xmath180 . since usually a change in winding number is accompanied by a change in the current , our results in fig .
[ pr ] lead us to naively expect that self inductances such that @xmath181 could result in larger conservation probabilities .
to logarithmic accuracy , @xmath182 . from here and eq .
( [ bcs ] ) , @xmath183 .
for the squares in fig .
[ pr ] @xmath184 , indicating that for quasi 1d samples it would be rather difficult to have @xmath185 significantly larger than @xmath186 .
we now study the evolution of the flux during a step between @xmath187 and @xmath188 .
we write the flux as @xmath189=lci$ ] and determine the current by @xmath190 , where @xmath191 and @xmath192 are the normal and the superconducting current at segment @xmath193 .
the normal current is given by @xmath194/c\delta t$ ] and the superconducting current can be taken as @xmath195 , where @xmath196\psi ( s)$ ] is the gauge invariant order parameter . from here
we can obtain _
j=1^n a_j(t+t)=_j=1^n .
[ fluxev ] the flux enclosed by the loop acts a constraint on the values of @xmath54 for each individual segment .
it should be noted that the winding number is that of @xmath41 , not that of @xmath197 .
figure [ taul ] shows the influence of the induced flux on the conservation probabilities for the samples represented by the squares and by the rhombs in fig .
[ pr ] , in the range @xmath198 .
we found that @xmath141 does increase with @xmath185 , but this increase is moderate and @xmath141 remains far from 1 .
there is a limited region of steep increase when @xmath185 becomes of the order of @xmath186 ; the increase in this region is larger for the sample with lower @xmath0 .
we finally test postulate 1 . for this purpose
we let the sample equilibrate at a temperature considerably above @xmath0 , then quench through @xmath0 at a uniform rate @xmath199 , and finally stabilize at @xmath200 .
denoting @xmath201 , postulate 1 becomes ( ) /||=_q .
[ epkova ] assuming scalings of the form @xmath202 and @xmath203 , eq .
( [ epkova ] ) leads to n^2(c/12_0)(_0/_q)^/(1+z ) .
[ kzn2 ] the initial temperatures were taken in the range @xmath204 , with @xmath205 .
the initial values of @xmath206 that we took are safely larger than the estimated values of @xmath207 .
the region of temperatures covered in this section is much wider than that of the previous sections , and we therefore used variable values for the time step @xmath208 and for the number of segments @xmath50 . for @xmath208 we typically took @xmath209/d$ ] .
@xmath50 was initially larger than 300 , but when the autocorrelation @xmath210 was steadily greater than 0.9 , @xmath50 was reduced .
equilibration times were typically @xmath211 , and changing these times by a factor of 3 had no appreciable influence on our results .
our results are shown by black symbols in fig .
[ post1 ] .
these results indicate that @xmath212 is a universal function of @xmath213 . for mean field ,
the critical exponents are @xmath214 and @xmath215 , so that the exponent in eq .
( [ kzn2 ] ) is 1/4 .
we have therefore fitted the rightmost part of the graph to the form @xmath216 . identifying @xmath217 with @xmath39
, this fit corresponds to @xmath218 in eq .
( [ kzn2 ] ) . at the leftmost part of the graph
, @xmath219 seems to be a weaker function of @xmath1 .
the fit in this region corresponds to @xmath220 .
the crossover occurs at @xmath221 .
weaking of the dependence of @xmath219 on @xmath1 should actually be expected for sufficiently fast quench since , as we learned in the previous sections , even in the limit of instantaneous quench , there is a considerable loss " of metastable states .
we note that in contrast to the previous sections , where our study focused on the region @xmath222 , the relevant time unit in fig .
[ post1 ] is @xmath223 rather than @xmath76 .
as discussed in ref . , in a superconducting loop there is a competition between the fluctuations of the order parameter and those of the gauge field , and it is not obvious in general which of them will dominate . if gauge fields are to play a dominant role , the infinite range and the speed with which different sections of the loop can influence one another could prove significant .
we should clarify that , in the case of the 1d loop we are considering , the induced magnetic flux for winding number @xmath28 is much less than the fluxoid @xmath224 .
moreover , except for the case of sec .
[ infll ] , the magnetic field is completely neglected .
the effect of the electromagnetic potential @xmath42 is not the creation of magnetic fluctuations in space , but rather of fluctuations of the electric field along the loop . in this section we set @xmath225 in eq .
( [ energy ] ) , and proceed exactly as before .
the variances of the winding number for equilibrium distributions are still described by fig .
[ uniform ] and @xmath226 close to @xmath0 .
the probabilities for decays of states with non zero winding number are still fairly described by fig .
[ pr ] . the blue symbols in fig .
[ post1 ] are our results for @xmath212 as a function of @xmath227 . as in the case of sec .
[ rate ] , we obtain a universal function , independent of the loop parameters . the blue line is a fit to the form @xmath216 . as opposed to the case of sec
. [ rate ] , the increase of @xmath212 with the quenching rate does not seem to saturate for @xmath228 , suggesting that the fluctuations of the electromagnetic field are responsible for this saturation trend . in the absence of electromagnetic fluctuations the final winding numbers are larger than in their presence .
this means that these fluctuations are more effective on destroying metastable states than on creating them .
we have analyzed the probabilities for final states with permanent currents when a 1d superconducting loop is quenched through its critical temperature in the absence of magnetic flux .
the analysis was based on the time - dependent ginzburg landau model with thermal fluctuations .
the predictions of the kibble zurek mechanism ( kzm ) are obtained in the appropriate limits . sufficiently above @xmath0 , for coherence lengths @xmath74 that are short compared to @xmath72 , the phase of the order parameter remains essentially constant over lengths comparable to @xmath74 . a large fraction of the topological defects " that are present immediately after the temperature is abruptly lowered below @xmath0 decay during a lapse of time of the order of @xmath229 , but most metastable states that survive this stage conserve their winding number in the following .
the requirement of a sort of incubation time " before the winding number becomes locked was also noticed in ref . .
equation ( [ kzn2 ] ) is obeyed for @xmath230 , with the mean field critical exponents @xmath214 and @xmath215 . for shorter quenching times , the increase of @xmath219 with the quenching rate is somewhat slower . when fluctuations of the gauge field are ignored , the values of @xmath219 increase , and eq . ( [ kzn2 ] ) is obeyed over a wider range of quenching rates .
we have not investigated the influence of the self inductance on the dependence of @xmath219 on the quenching rate .
this research was supported by the israel science foundation , grant no .
numeric evaluations were performed using computer facilities of the technion
israel institute of technology .
i am grateful to arnab das , francesco tafuri , vladimir zhuravlev and wojciech zurek for their answers to my inquiries .
we consider a uniform loop and distinguish between the cases above and below the critical temperature . in this case
an undemanding approximation can be obtained by neglecting @xmath231 and @xmath232 in eq .
( [ discrete ] ) .
decomposing into fourier components , @xmath233 , we can write g=(re^2 _ k+im^2 _
k)e_k , [ gm ] with e_k= ( -+^2 ) .
[ em ] from here , the equilibrium distributions for @xmath234 and @xmath235 are gaussian , with variance @xmath236 . in this case
it will be more realistic to assume that the order parameter is close to the ground state , i.e. , _
j=+u_j+iv_j , [ uiv ] where @xmath237 and @xmath238 are real and small . substituting eq .
( [ uiv ] ) into eq .
( [ discrete ] ) and keeping terms up to quadratic in @xmath239 and @xmath240 , we are left with g = constant+u_j^2 + .
we introduce now the decomposition @xmath241 , @xmath242 , with @xmath243 and @xmath244 ; in the decomposition for @xmath238 we do not include a constant term , since it would just lead to multiplication of the order parameter by a uniform phase . taking @xmath50 even , the free energy becomes @xmath245\end{aligned}\ ] ] with @xmath246 from here , the equilibrium distributions for @xmath247 , @xmath248 , @xmath249 and @xmath250 are gaussian , with variances given by the appropriate @xmath251 .
b. kibble , j. phys .
a * 9 * , 1387 ( 1976 ) ; phys .
rep . * 67 * , 183 ( 1980 ) ; phys . today * 60 * , 47 ( 2007 ) .
w. h. zurek , nature ( london ) * 317 * , 505 ( 1985 ) ; acta phys . pol .
b * 24 * , 1301 ( 1993 ) .
w. h. zurek , phys . rep .
* 276 * , 177 ( 1996 ) .
laguna and w. h. zurek , phys .
lett . * 78 * , 2519 ( 1997 ) .
r. carmi , e. polturak , and g. koren , phys .
lett . * 84 * , 4966 ( 2000 ) .
m. ghinovker , b. ya .
shapiro and i. shapiro , europhys .
lett . * 53 * , 240 ( 2001 ) .
j. r. kirtley , c.c .
tsuei , and f. tafuri , phys . rev .
lett . * 90 * , 257001 ( 2003 ) .
j. berger , physica c * 468 * , 294 ( 2008 ) .
r. monaco , j. mygind , r.j .
rivers , and v.p .
koshelets , phys .
b * 80 * , 180501 ( 2009 ) and references therein .
a. das , j. sabbatini , and w.h .
zurek , sci . rep . * 2 * , 352 ( 2012 ) .
y. enomoto and k. okada , j. phys . : condens .
matter * 9 * , 10203 ( 1997 ) .
j. berger , phys .
b * 75 * , 184522 ( 2007 ) ; j. phys . :
* 23 * 225701 ( 2011 ) .
j. berger and j. rubinstein , siam j. appl . math . * 58 * , 103 ( 1998 ) .
j. berger and m. v. miloevi , phys .
b * 84 * , 214515 ( 2011 ) . | we study the final distribution of the winding numbers in a 1d superconducting ring that is quenched through its critical temperature in the absence of magnetic flux .
the study is conducted using the stochastic time - dependent ginzburg landau model , and the results are compared with the kibble zurek mechanism ( kzm ) .
the assumptions of kzm are formulated and checked as three separate postulates .
we find a characteristic length and characteristic times for the processes we study .
besides the case of uniform rings , we examined the case of rings with several weak links . for temperatures close or below @xmath0
, the coherence length does not characterize the correlation length . in order to regard the winding number as a conserved quantity , it is necessary to allow for a short lapse of time during which unstable configurations decay .
we found criteria for the validity of the 1d treatment . the is no lower bound for final temperatures that permit 1d treatment . for moderate quenching times
@xmath1 , the variance of the winding number obeys the scaling @xmath2 , as predicted by kzm in the case of mean field models ; for @xmath3 , the dependence is weaker .
we also studied the behavior of the system when fluctuations of the gauge field are suppressed , and obtained that the scaling @xmath2 is obeyed over a wider range . |
the phase diagram of the cuprates , like , e.g. , @xmath1 ( lsco ) , is characterized by three distinct regions which are reached by varying the doping @xmath2 : ( i ) the overdoped region , where the system displays a markedly metallic character , which can be reasonably described within the standard fermi - liquid picture ; ( ii ) the region around optimal doping , where the superconducting critical temperature @xmath3 is maximum , and no other energy scale besides the temperature @xmath4 is seen in various experiments , so that the metallic phase above @xmath3 strongly deviates from the fermi - liquid behavior ; ( iii ) the underdoped region where , on the contrary , there is an abundance of energy scales , and an even stronger violation of the fermi - liquid behavior .
these are related to a doping - dependent temperature @xmath5 , below which a suppression of the density of states at the fermi level is revealed in static measurements and to a doping - dependent temperature @xmath0 below which various experiments reveal a pseudogap @xmath6 .
@xmath0 strongly depends on the characteristic time scale of the probe used to measure the pseudogap properties . other energy scales of the underdoped phase are the superconducting gap @xmath7 and the critical temperature @xmath3 . the generic partition of the phase diagram into three qualitatively quite distinct regions with markedly different behaviors and the absence of any energy scale but the temperature in the intermediate ( optimally doped ) region , suggest the presence of a quantum critical point ( qcp ) near optimal doping @xcite .
if this is the case , the identification of the optimally doped region with the region where quantum critical fluctuations are present naturally accounts for the peculiar non - fermi - liquid behavior .
various proposals have been made to specify the broken symmetry phase to be associated with the underdoped region , ranging from a circulating - current phase @xcite , to a change in symmetry of the superconducting order parameter @xcite .
the scenario that we have been proposing along the years @xcite is related to the occurrence of an instability for charge ordering ( co ) , along a line @xmath8 ending in the qcp near optimal doping @xcite .
this proposal of a ( dynamical ) co instability finds support from neutron scattering experiments in ne - doped lsco @xcite and @xmath9 systems @xcite , where charge domain walls ( stripes ) have been detected . in this paper , by extending previous results , we account for the various pseudogap - formation temperatures and give a reason why formed stripes may be present in some classes of cuprates and be absent in other classes .
we also address the issue of the anomalous isotopic effects of @xmath3 and @xmath0 in the cuprates and provide a novel mechanism based on the relevance of co fluctuations .
the above co scenario found one of its possible realizations within a minimal model of strongly correlated electrons interacting with phonons , i.e. , the one - band hubbard - holstein ( hh ) model in the presence of long - range coulomb forces @xcite . in this model
there is a hopping term @xmath10 representing the electron kinetic energy , a ( large ) on - site hubbard repulsion @xmath11 , representing the strong electron - electron ( e - e ) correlation , a coupling @xmath12 between a local lattice distortion and the local electron density , a phonon frequency @xmath13 , characterizing the relevant lattice distortion , and a long - range e - e coulomb repulsion @xmath14 .
the simplification with respect to models taking into account the complex structure of the cuprates was adopted for two main reasons : first of all it makes the model manageable despite the formal complications related to the treatment of strong e - e correlations via the slave - boson large-@xmath15 expansion formalism .
moreover it allows to concentrate on the relevant aspects providing both a non - fermi - liquid behavior and a strong pairing mechanism .
although the relevant features related to quantum criticality should be generic , different cuprates should of course show different aspects related to their specific structure .
two major results were obtained within the hh model @xcite .
first of all , at mean field , one obtains the @xmath16 phase diagram reported in fig .
1 with a line of the electron - phonon ( e - ph ) coupling @xmath12 as a function of the doping @xmath2 . this line marks a second - order transition for the onset of co characterized by an order parameter @xmath17 representing the microscopic charge modulation at a wave vector @xmath18 @xcite .
each point on the curve @xmath19 is a qcp , which is the @xmath16 end point of a critical line @xmath20 .
secondly , charge collective fluctuations mediate a doping- and temperature - dependent singular scattering among the quasiparticles @xmath21 where @xmath22 is a residual repulsion , @xmath23 is the strength of the singular interaction , @xmath24 is a characteristic time scale of the charge fluctuations , and @xmath25 is the inverse square correlation length , which measures the distance from criticality .
this interaction diverges at @xmath26 and at a critical wave vector @xmath18 when @xmath25 vanishes .
this occurs at @xmath16 when the doping is reduced down to a mean - field critical value @xmath27 , and at finite @xmath4 , when approaching the mean - field critical line @xmath20 . as we discuss in sec .
ii b , the co fluctuations beyond mean field shift the mean - field critical line @xmath28 to a lower value @xmath8 , ending in the fluctuation - corrected qcp at @xmath29 ( @xmath30 ) , and modify accordingly the expression for the coherence length @xmath31 .
the singular part of the interaction ( [ gamma ] ) ( with @xmath32 if fluctuations are taken into account ) is attractive both in the particle - particle and in the particle - hole channels .
therefore , the quasiparticles feel an increasingly strong attraction by approaching the co critical line . in the particle - hole channel
, this interaction can produce a pseudogap due to the incipient co. at the same time in the particle - particle channel , the strong attraction can lead to pair formation even in the absence of phase coherence @xcite .
therefore the pseudogap - formation temperature @xmath33 closely tracks the underlying co transition line .
we point out that the assumption of an infinitely large hubbard repulsion @xmath11 and the large-@xmath15 expansion do not allow for a correct description of the spin degrees of freedom @xcite .
however , at large but finite @xmath11 the above scenario is not expected to drastically change , except that the modulation of the charge profile within the co state should favor antiferromagnetic ordering in the charge - poor regions .
this state with charge and ( enslaved ) spin modulation mimics the stripe phase observed in the cuprates , when approaching it from the large - doping side , where charge degrees of freedom play a major role .
enslaved nearly critical antiferromagnetic fluctuations with a characteristic wave vector @xmath34 are expected to mediate an interaction similar to eq .
( [ gamma ] ) in the particle - particle channel , which becomes repulsive in the cooper channel .
this additional contribution is relevant to properly reproduce the main features of the single - particle spectra @xcite .
the @xmath35-wave symmetry of the superconducting order parameter , which is already favored by the residual repulsive term in eq .
( [ gamma ] ) @xcite , is further stabilized by the antiferromagnetic fluctuations . the separation between the homogeneous phase and the co phase at @xmath16 , reported in fig .
1 , is marked by a line having a minimum value @xmath36 , below which no co ( or stripes ) could be formed as a static phase . of course strong quantum fluctuations mediating the singular scattering ( [ gamma ] ) , which leads to pairing and non - fermi - liquid behavior in the normal phase ,
would be present both above and below the quantum critical line .
the crucial point in obtaining the co instability is that strong electronic correlations reduce the effect of the homogenizing electron kinetic - energy term , thus favoring the possibility of an instability induced by a residual interaction , which in our model is provided by the e - ph interaction .
the relevant role of the lattice in driving the instability is , indeed , strongly suggested by the peculiar isotopic effect of @xmath0 and @xmath3 @xcite and by recent exafs experiments @xcite , which identified interesting features associated to the lattice structure .
specifically a microstrain @xmath37 of the cu - o bonds was measured in terms of the deviation of the cu - o distance with respect to a reference distance @xmath38 , which introduces a mismatch in the lattice between the cuo@xmath39 layers and the rock - salt layers .
actually there is a close resemblance between the experimental @xmath40 and the theoretical @xmath41 phase diagrams indicating that the cu - o microstrain and the e - ph coupling @xmath12 of the simplified hh model are strictly related .
from the microscopic point of view it is indeed quite natural that a lattice contraction in the cuo@xmath39 planes can enhance the effective coupling between the electrons and the ions .
therefore the comparison between experiments and theory allows to draw the conclusion that a one - to - one monotonic relation @xmath42 is likely to exist .
once this general framework is settled , one can move to identify more detailed possible scenarios .
in particular the order of the homogeneous - metal to charge - ordered - metal transition is of obvious relevance .
the simplest possibility is that in the real materials this transition is of the second order . in this case , at @xmath16 , the different microstrains determined by the rock - salt layers directly correspond to different e - ph couplings and are reflected in the similar phase diagrams .
of course this simple theoretical picture is not necessarily realized in the quite complex real materials where i ) anharmonic effects and/or ii ) additional non - ordering fields can partially or entirely transform the second - order transition into a first - order one . in this last case ,
the above scenario maintains its validity , as long as this transition is _ weakly _ first - order .
we stressed in several papers @xcite that the actual onset of a fully developed co phase is competing with local or coherent pair formation , which modifies the fermionic spectrum stabilizing the system against the electronic co transition .
there is the complementary possibility that the charge - ordered state can never be reached because @xmath12 is below @xmath36 .
the corresponding observable quantity identified by the experiments in ref .
@xcite is the critical microstrain @xmath43 , below which no stripe phase can be observed .
therefore the cuprates can deviate from co criticality not only by @xmath44 in the ( overdoped ) quantum - disordered region , and by @xmath4 or @xmath45 in the quantum - critical or underdoped region , but also because the microstrain @xmath37 is smaller or larger than @xmath43 , thereby tending to a homogeneous or an inhomogeneous phase respectively .
as long as @xmath46 is not too large and @xmath4 is finite , for @xmath47 the quantum - critical region , with strong dynamical fluctuations , can be reached in both cases . according to the proposed mechanism of pairing mediated by critical fluctuations ,
the stronger are these fluctuations and the larger is @xmath3 . indeed in the experiment the mercury compound hg1212 , having @xmath48 , has the the largest @xmath3 , while hg1201 ( with @xmath49 ) and bi2212 or lsco ( with @xmath50 ) have a lower @xmath3 .
moreover , since @xmath49 in hg1201 , we expect no @xmath8 , and therefore no @xmath0 , for this material . in this section
we calculate the temperature for the onset of the co instability within the mean - field approximation , @xmath28 , and its corrected value , @xmath8 , in the presence of the leading fluctuations beyond mean field . for establishing the onset of the co phase coming from high - doping and high - temperature regime
, the direct involvement of the spin degrees of freedom can be safely ignored .
for @xmath51 the co fluctuations become substantial , leading to a reduction of the quasiparticle density of states .
we therefore identify our mean - field line @xmath28 with the weak - pseudogap crossover line @xmath52 observed in knight - shift , transport , and static susceptibility measurements @xcite , as the incipient reduction of the quasiparticle density of states .
the fluctuation - corrected critical line @xmath8 [ @xmath53 , where the correlation length @xmath31 diverges , and the scattering ( [ gamma ] ) becomes effectively singular , is instead closely tracked by the pseudogap crossover line @xmath33 , according to the discussion of sec .
all the complicated formal structure of the quasiparticle scattering , mediated by phonons and by the coulomb interaction within the slave - boson approach , is , at the end , simply represented , at least near criticality , by a rpa resummation @xmath54 $ ] of an effective static interaction @xmath55 @xcite .
here @xmath56 is the fermionic polarization bubble , @xmath57 is the dimensionless e - ph coupling , and @xmath58 is the residual short - range repulsion between quasiparticles , and @xmath59 is the fourier transform of the e - e coulomb repulsion . for the correspondence of @xmath60 with the parameters of the original hh model see ref .
the mean - field co instability condition , which occurs for reasonable values @xmath61 , is @xmath62 . at @xmath16
this determines @xmath18 and the position of the mean - field qcp @xmath27 . by expanding @xmath63 near the instability at @xmath16 , we find @xmath64 . at @xmath65 ,
considering the @xmath4 dependence of the bare polarization bubble @xmath56 , the instability condition determines the mean - field critical line @xmath28 ( the solid - line curves in fig .
2 ) , which ends at @xmath27 .
we evaluate @xmath28 by taking standard quasiparticle ( i.e. , dressed by the slave bosons ) band parameters ( @xmath66 mev , @xmath67 mev , leading to @xmath68 mev and @xmath69 mev in the residual repulsion @xmath70 ) .
@xmath71 mev and @xmath72 mev are adjusted to match with the extrapolated experimental values of @xmath52 for lsco @xcite ( @xmath73 ) with the mean - field qcp @xmath27 . without any further adjustment of the parameters
the curve @xmath28 agrees remarkably well with the experimental data for @xmath52 .
similar parameters are taken for bi2212 to fit the data in refs .
@xcite , with @xmath71 ev and @xmath74 ev . due to the strong anisotropy of the layered cuprates , the contribution of the fluctuations related to any instability is significant @xcite .
specifically the co fluctuations which mediate the critical effective interaction , eq .
( [ gamma ] ) , can be included in the polarization bubble @xmath56 ( entering the instability condition ) via the diagrams of fig .
3 , leading to corrections beyond rpa . from the explicit evaluation of these diagrams
we find the self - consistent correction to the inverse square correlation length @xmath75 where @xmath76 are the bosonic matsubara frequencies , and the coupling constant @xmath77 .
since the phonons are responsible for the instability , @xmath13 appears as the frequency cut - off @xcite .
the fluctuation - corrected instability condition is @xmath78 .
as it stands , eq .
( [ mass1 ] ) is written for the two - dimensional cu - o planes . at @xmath16 ,
( [ mass1 ] ) with @xmath78 leads to a finite shift of the two - dimensional ( 2d ) qcp @xmath79 ( see fig .
2 ) . at finite @xmath4
, purely 2d fluctuations suppress the transition .
however , by considering the more realistic anisotropic 3d character of the critical fluctuations , we obtain a finite fluctuation - corrected transition line @xmath8 , which is reported as the lowest dashed line in fig
thus , the inclusion of fluctuations brings the critical line from temperatures of the order of typical electronic energies ( @xmath80 ) down to much lower temperatures @xmath81 of the order of the observed @xmath0 s . the spread in the measured values of @xmath0 depending on the typical frequency of the experimental probe @xmath82 ( see ,
e.g. , ref .
@xcite ) suggests , however , within our identification of @xmath81 with @xmath0 , that the co instability may be `` dynamical '' . in this case , the critical condition @xmath78 is replaced by @xmath83 in the self - consistency condition , eq .
( [ mass1 ] ) .
two examples are reported in fig.2 for @xmath84 mev ( dot - dashed line in figs .
2(a , b ) , typical of neutron scattering experiments , and @xmath85ev ( second dashed line from bottom in fig .
2(a ) , typical of static experiments ( nqr , nmr ) .
the corresponding experimental data for @xmath0 in lsco are also reported for comparison .
the coupling @xmath23 between the charge fluctuations and the quasiparticles , which is the only parameter for which an a priori estimate is difficult , is fixed by matching the position of the fluctuation - corrected qcp with the @xmath16 extrapolation of the experimental values of @xmath0 .
we used @xmath86 ev@xmath87 and @xmath88 ev@xmath87 for lsco and bi2212 respectively , and @xmath89 ev for both .
note that , similarly to the case of @xmath28 , the agreement between the calculated @xmath8 and the experimental points for @xmath0 is obtained without adjustment of other parameters .
in the present framework , any mechanism shifting the position of the qcp and the line @xmath8 is mirrored by a corresponding shift of the superconducting critical line @xmath90 and of the pseudogap formation temperature @xmath33 .
this is the case of the shift induced by isotopic substitution , and we want to connect our results with the observation of isotopic effects on @xmath3 @xcite and on @xmath0 @xcite . as shown in sec .
ii a , the mean - field weak - pseudogap crossover temperature @xmath91 depends on the phonon frequency only through the dimensionless e - ph coupling @xmath92 , which is known to display no isotopic dependence .
therefore @xmath93 , and consequently @xmath5 and the portion of @xmath90 near and above @xmath52 , are not expected to display isotope effects . on the other hand , the co fluctuations , as shown above , crucially involve @xmath13 , and new physical effects can be induced by the isotopic substitutions , on fluctuation - dependent physical quantities , such as @xmath29 and @xmath8 . since @xmath13 decreases with increasing ionic mass , the effect of fluctuations is reduced , and @xmath29 and @xmath8 remain closer to their mean - field values . in fig .
4 , we report the line @xmath8 , calculated via eq .
( [ mass1 ] ) , with the same parameters used in sec .
ii b to fit the @xmath0 data of lsco , together with its isotopic shift calculated for @xmath94 substitution ( i.e. , for a five percent reduction of @xmath13 ) .
4 . calculated effect of the isotopic change @xmath95 ( i.e. @xmath67 mev and @xmath96 mev ) on @xmath0 in lsco , both for a fast- and a slow - probe measurement .
the inferred shift on @xmath3 is also reported , although hardly visible on this scale .
contrary to standard theories based on co pseudogap @xcite , the isotopic shift is positive ( i.e. anomalous ) for @xmath97 .
moreover , if the slope of @xmath8 is large , a small isotopic shift in @xmath29 can result in a substantial shift in @xmath97 , i.e. , the steeper is @xmath0 , the larger is the isotopic effect . since the slope of @xmath8 increases by increasing @xmath82 , whereas the qcp is unshifted , the isotope effect on @xmath97 is enhanced , and we have the general trend that faster probes should detect a larger isotope effect on @xmath0 . although we can not account for the near - absent or negative isotope effect on @xmath0 within the almost static probes in @xmath98 @xcite , this general trend is in qualitative agreement with a much stronger effect observed in the isostructural @xmath99 ( hbco-124 ) with fast neutron scattering @xcite .
indeed , this fast - probe experiment should be represented by the curve @xmath8 corresponding to @xmath84 mev in fig .
4 . the huge isotope effect on @xmath0 observed in hbco-124 suggests that the curve @xmath8 in this system is steeper than in lsco .
as far as @xmath3 in the underdoped region is concerned , we can only infer some expected consequences .
assuming that the shift in @xmath29 produces a rigid shift of @xmath90 along the doping axis , the isotopic effect for @xmath3 is negative ( i.e. normal ) upon reducing @xmath13 , and much smaller than for @xmath0 ( the @xmath90 curve being much flatter ) , in agreement with long - standing experiments @xcite .
this large difference in the isotope effect for @xmath3 and @xmath0 , @xmath100 , is indeed experimentally observed in hbco-124 , and reported in ref .
@xcite , where a `` striking similarity between isotopic substitution and underdoping with respect to both @xmath3 and @xmath0 '' is pointed out .
although we are not aware of any systematic analysis of the doping dependencies of @xmath3 , @xmath0 and their isotopic shifts in hbco-124 , allowing for a strict comparison , this observation finds its natural interpretation within our theory , where the isotopic substitution increases @xmath29 and is therefore nearly equivalent to underdoping . in the strongly underdoped materials , where substantial isotope effects in the penetration depth @xcite and in the x - ray absorption @xcite have been observed , our results can provide only qualitative indications , as other effects ( magnetic , polaronic , lattice - pinning ) not included in our
hh model become relevant .
in conclusion we have revisited the co scenario , and explicitly evaluated the mean - field and the fluctuation - corrected critical lines , @xmath28 and @xmath8 , for the onset of the charge inhomogeneous phase , which should represent the stripe phase of the cuprates , when present .
the nearly singular effective interaction among the electrons mediated by the co fluctuations near the instability , suggests the natural identification of the experimentally observed weak - pseudogap crossover line @xmath52 with @xmath28 , and of the pseudogap formation temperature @xmath0 with @xmath8 .
we obtain in this way a good fitting of the experimental data , and by assuming that the co transition has a dynamical character , we account for the spread of the measurements of @xmath33 , depending on the characteristic time scale of the experimental probe .
the existence of the co transition , depending on the strength of the e - ph coupling , as produced by our zero - temperature phase diagram , can give an answer to the puzzling question why features of formed stripes are present in some cuprates and absent in others .
the co instability condition at mean - field level depends only on the dimensionless e - ph coupling which is not changed by isotopic substitution .
therefore , in our model , @xmath28 should not undergo any isotopic shift . on the contrary , @xmath8 , and therefore @xmath0 ,
are strongly modified by the effect of co fluctuations , which in our approach is proportional to the phonon frequency @xmath13 . within our model ,
therefore , @xmath0 shows a novel isotope effect produced by charge fluctuations , with a positive shift , which depends on the slope of @xmath8 .
it results , therefore , that the isotope effect is stronger when detected with faster probes .
we also infer that @xmath90 should have no isotope effect in the optimal and overdoped regimes , and a small negative effect in the underdoped region , which increases upon underdoping . in this new isotope effect
, the phonon do not appear directly as mediators of pairing , but are indirectly involved , via the co fluctuations , and qualitatively explain the complex behavior of @xmath0 and @xmath3 upon isotopic substitution .
0.5truecm m. gutmann , _ et al .
_ , cond - mat/0009141 .
evidences of a qcp around optimal doping are listed in , e.g. , c. castellani , _ et al .
_ , z. phys .
b * 103 * , 137 ( 1997 ) . j. l. tallon and j. w. loram , _ physica _ c * 349 * , 53 ( 2001 ) g. boebinger , _ et al . _ ,
lett . * 77 * , 5417 ( 1996 ) . c. panagopoulos , _ et al .
_ , cond - mat/0002239 .
d. l. feng , _ et al .
_ , science * 289 * , 277 ( 2000 ) .
c. m. varma , phys .
14554 ( 1997 ) and references therein . m. vojta , y. zhang , and s. sachdev , phys .
lett . * 85 * , 4940 ( 2000 ) .
s. caprara , _ et al .
_ , physica b * 280 * , 196 ( 2000 ) ; c. di castro , _ et al .
_ , physica c * 341 - 348 * , 1715 ( 2000 ) . c. castellani , c. di castro , and m. grilli , phys .
lett . * 75 * , 4650 ( 1995 ) .
f. becca , _ et al .
_ , phys . rev .
b * 54 * , 12 443 ( 1996 ) . s. andergassen , s. caprara , c. di castro , and m. grilli , phys .
lett . * 87 * , 056401 ( 2001 ) .
j. tranquada , _ et al .
_ , nature * 375 * , 561 ( 1995 ) .
h. a. mook , _ et al .
_ , nature * 404 * , 729 ( 2000 ) . in the framework of refs
@xcite the co instability is a second - order transition leading to a gradual increase of the charge modulation .
it is only when one enters deeply inside the ( locally ) ordered phase that anharmonic distortions of the charge profile may arise from the enhanced interactions with the spin and the lattice degrees of freedom .
this may lead to the stripe formation .
phase coherence may be established at a much lower temperature , see , e.g. , a. perali , _ et al .
b * 62 * , r9295 ( 2000 ) . for the same reason , the low - doping region in fig .
1 is only indicative .
s. caprara , _ et al .
b * 59 * , 14980 ( 1999 ) .
n. l. saini , _ et al .
lett . * 79 * , 3467 ( 1997 ) .
a. perali , _ et al .
b * 54 * , 16216 ( 1996 ) .
a. bianconi , _ et al .
_ , j. phys .
: condens .
matter * 12 * , 10655 ( 2000 ) .
this feature is the natural consequence of the damping of the charge fluctuations due to the creation of particle - hole pairs at arbitrarily low energy when @xmath101 joins two `` hot '' points on the fermi surface .
g. seibold , _ et al .
j. b * 13 * , 87 ( 2000 ) .
h. ding , _ et al .
_ , j. phys .
solids * 59 * , 1888 ( 1998 ) .
k. ishida , _ et al .
b * 58 * , r5960 ( 1998 )
. a. j. millis , phys .
b * 48 * , 7183 ( 1993 ) .
r. s. markiewicz , physica c * 169 * , 63 ( 1990 ) .
j. p. franck , in _ physical properties of high temperature superconductors iv _ , edited by d. m. ginsberg ( world scientific , singapore , 1994 ) , p. 189 .
m. k. crawford , _ et al .
b * 41 * , 282 ( 1990 ) . f. raffa , _ et al . _ ,
* 81 * , 5912 ( 1998 ) .
g. v. m. williams , _ et al .
b * 61 * , r9257 ( 2000 ) .
d. rubio temprano , _ et al .
rev . lett .
* 84 * , 1990 ( 2000 ) .
i. eremin , _ et al .
b * 56 * , 11305 ( 1997 ) .
j. hofer , _ et al .
lett . * 84 * , 4192 ( 2000 ) .
a. lanzara , _ et al .
_ , j. phys .
: condens .
matter * 11 * , l541 ( 1999 ) . | within the hubbard - holstein model with long - range coulomb forces , we revisit the charge - ordering scenario for the superconducting cuprates and account for the presence or the absence of a formed stripe phase in different classes of cuprates .
we also evaluate the mean - field and the fluctuation - corrected critical lines for charge ordering and we relate them with the various pseudogap crossover lines occurring in the cuprates and we discuss a mechanism for their peculiar isotopic dependence . considering the dynamical nature of the charge - ordering transition , we explain the spread of @xmath0 and of its isotopic shift , obtained with experimental probes with different characteristic time scales .
pacs : 71.10.-w , 74.72.-h , 71.45.lr , 71.28.+d 2 |
stars of horizontal - branch nature are important objects in studies of the older stellar populations and in studies of galactic structure in relation with formation theories for the milky way .
their virtue lies in three properties .
first , they have rather well defined absolute magnitudes and their distances are therefore easy to determine .
second , their nature is such that they are easy to discover , in particular at higher galactic latitudes .
third , the stellar evolution leading to such stars is in principle relatively well understood .
the horizontal branch ( hb ) stellar phase is the core helium burning late stage of evolution of originally low to medium mass stars . during the red giant phase
the stars have lost mass such that a helium core of @xmath1 m@xmath2 , surrounded by an outer shell of hydrogen gas , remains .
for very thin shells , @xmath3 m@xmath2 , the stars are rather blue and of spectral class sdb ( subdwarf b ) , for ever thicker shells their atmospheres are cooler , leading to spectral types such as horizontal branch b and a ( hbb and hba ) and to the red hb ( rhb ) stars with @xmath4 m@xmath2 . between the hba and rhb stars
lies the pulsational instability strip with the rr lyrae stars .
( for the systematics of the spectral classification see de
boer et al . 1997c . )
hba and hbb stars are often called blue hb ( bhb ) stars , the even hotter sdb and sdo stars are also known as extended or extreme hb ( ehb ) stars .
bhb and ehb stars can be easily found based on their blueness , the rr lyr stars due to their variability . of the bhb stars ,
the hbas are easily identified , as their physical parameters differ from main sequence a stars , while hbbs may be confused with main sequence b stars . however , at higher galactical latitudes a or b main sequence stars are very rare .
rhb stars can easily be confused with subgiants and giants because they lie in the same region of the hrd . to trace the structure of the ( local part of the ) milky way one employs several techniques ( see also the review by majewski 1993 ) .
the classical method is to perform star counts ( see , e.g. , bahcall & soneira 1984 ) .
selecting stars based on their proper motion allows , if their radial velocities and distances are also determined , to study the true kinematics of the stars ( see , e.g. , carney et al .
1996 and papers cited there ) . basing oneself on proper motions one
naturally studies the general group of more nearby stars extended by true high - velocity stars .
another method is to sample distances and radial velocities of a specific set of stars , such as bhb or rr lyr stars , to investigate these parameters in a statistical manner ( see , e.g. , kinman et al .
1994 , 1996 ) .
a more specific method is to observe statistically complete samples of stars of a special type in several directions to derive scale heights ( for sdb stars see e.g. , heber 1986 , moehler et al . 1990 ,
theissen et al .
1993 ) , or scale lengths in the milky way .
a further possibility is to go beyond the present kinematic parameters by calculating orbits based on distances , radial velocities , and proper motions .
this method has been used for high proper motion stars ( carney et al .
1996 ) , other dwarf stars ( schuster & allen 1997 ) , and for globular clusters ( dauphole et al .
also the orbits of sdb stars have been investigated ( colin et al .
1994 , de boer et al .
the latter study showed that most of the sdb stars have orbits staying fairly well inside the milky way disk , indicating that the sdb stars are not generally part of the halo population . in the present study we have attempted to perform a similar analysis for hba and hbb stars ( for short : hba / b stars ) .
our sample consists of the local hb stars which were observed by the hipparcos satellite .
these are the hb - like stars with the most accurate spatial and kinematic data available to date .
however , only for a few hba / b stars are the parallaxes accurate enough to calculate reliable distances ( de boer et al .
1997b ) . for the other stars
the distance still must be derived from photometry .
here one needs to know the absolute magnitude of field horizontal branch stars . especially since the publication of the hipparcos catalogue a lot of effort has gone into fixing this value .
however , this has not yet led to total agreement . for a review of various approaches to solving this problem
we refer to de boer ( 1999 ) and popowski & gould ( 1999 ) .
an important parameter in these studies is the metallicity of the stars , as it is generally thought to be correlated with age . for dwarf stars metallicities
can be estimated using photometric indices or spectroscopy ( see the summary by majewski 1993 ) . for hb stars
this is , unfortunately , not a trustworthy method .
the atmospheres of many hb stars have most probably been altered chemically with respect to the original composition . gravitational settling of heavy elements in the sdb / o star and possibly hbb
star atmospheres leads to a present lower content of elements like he , while levitation of heavy elements leads to atmospheres with enhanced abundances of certain elements like fe or au as found in several field horizontal branch stars , e.g. feige 86 ( bonifacio et al . 1995 ) .
levitation must also be the explanation for the high metal abundances in blue hb stars in m 13 ( behr et al .
1999 ) and ngc 6752 ( moehler et al .
1999 ) finally uncovered to explain deviant flux distributions near the balmer jump of globular cluster blue hb stars ( grundahl et al .
therefore , original metallicities ( as well as the original masses ) are no longer accessible quantities .
determining the kinematic properties can help deciding which of the hb stars are intrinsically more metal poor and which are more metal rich , and hence of somewhat younger origin .
the main subjects of our study are the hba / b stars which are located in the colour magnitude diagram on the horizontal branch between the rr - lyrae stars and the hot subdwarfs
. unfortunately the main sequence crosses the hb at the hbb region , so that hbb stars can be confused with normal b stars .
therefore we have only few hbb stars in our sample .
we mainly focus on hb stars with temperatures lower than 10000 k which lie above the main sequence .
2 deals with the data neccessary for our study . in sect .
2.3 we determine the absolute magnitudes and distances of the hba / b and sdb / o stars with the method of _ auto - calibration _ using the shape of the hb defined by the stars with the best hipparcos parallaxes . in sect .
3 we discuss the kinematical behaviour of the hba / b and sdb / o stars and make comparisons with the results of de boer et al .
( 1997a ) . to further explore a possible trend in kinematics of stars along the hb we investigate ( sect .
4 ) the orbits of a sample of rr - lyrae stars . [ cols= " < , > , > , > , > , > , < , > , > , < " , ] + to determine the parameters @xmath5 , the maximum height reached above the galactic plane and @xmath6 and @xmath7 , the apo- and perigalactic distances , we calculated the orbits over 10 gyr .
this for certain does not give true orbits as the orbits are probably altered in time by heating processes .
however this long timespan allows to better show the area the orbit can occupy in the meridional plane ( see fig .
2 ) . as in de boer et al . ( 1997a ) , we also calculated the eccentricity @xmath8 of the orbit , given by @xmath9 and the normalised @xmath10-extent , @xmath11 , given by @xmath12 the parameter @xmath11 is more relevant than @xmath5 , since it accounts for the effect of diminished gravitational potential at larger galactocentric distance @xmath13 . to assign a star to a population
often the @xmath14-velocities and their dispersions are used , as well as the orbital velocity @xmath15 . for stars near the sun ( small @xmath16 )
, the @xmath17 velocity is nearly the same as @xmath15 .
however , for stars further away from the sun s azimuth , @xmath15 becomes a linear combination of @xmath18 and @xmath17
. therefore @xmath15 should be preferred . in order to make comparisons with results from other studies , we use both @xmath14 and @xmath15 .
we calculated the errors of the velocity components and the orbital velocity using monte carlo simulations of gaussian distributions to vary the input parameters within their errors as described by odenkirchen ( 1991 ) .
this is neccesary rather than just calculating errors using gauss error propagation laws because the parameters are significantly correlated . for the error calculation we used the software of odenkirchen ( priv .
the proper motion errors were taken from the hipparcos catalogue .
the errors of the distances were calculated from the error in absolute magnitude as derived in sect .
we took the errors of the radial velocities as published in the respective articles .
for those radial velocities of wilson ( 1953 ) and evans ( 1967 ) having quality mark `` e '' , meaning the error is larger than 10 kms@xmath19 , we used 15 kms@xmath19 as error .
this is justified as can be seen by comparison of these values with those of other studies .
generally the error in the velocity components is less than 10 kms@xmath19 .
only a few stars have somewhat larger errors , the largest error in @xmath15 being 12 kms@xmath19 . for the hba / b stars the typical value of @xmath20 is about 7 kms@xmath19 , for the on average closer
sdb / o stars @xmath20 is 1 to 2 kms@xmath19 .
we estimated errors for @xmath11 , @xmath8 , @xmath21 and @xmath22 because they have not been used individually in the interpretation .
moreover the larger values of @xmath11 are very sensitive to small variations in the shape of the orbit .
this especially applies to stars having chaotic orbits .
variations in the input distance modulus showed that the resulting variations in all of these quantities except @xmath11 are relatively small in most cases . for a discussion of overall effects on a sample see de boer et al .
( 1997a ) .
the orbits of the hba / b stars show a large variety of shapes .
nearly all of the cooler hba stars have a small perigalactic distance ( @xmath23 3 kpc ) and the most extreme case , hd 86986 , reaches a perigalactic distance of only 0.4 kpc .
the single exception is hd 117880 , which has a @xmath22 of nearly 4 kpc .
four stars have truly chaotic orbits , the rest has boxy type orbits , but some of these show signs of chaotic behaviour as well .
hd 79813 has an orbit staying very close to the disc , while hd 117880 orbits nearly perpendicular to the galactic plane . on the whole about half of our stars have orbits which are chaotic or show signs of that .
this agrees quite well with the results of schuster & allen ( 1997 ) who analysed a sample of local halo subdwarfs .
most of the stars have apogalactic distances of @xmath24 8 to 11 kpc , just one star ( hd 86986 ) goes well beyond .
the reason for this clumping in @xmath21 is not physical but due to selection effects .
stars with @xmath25 kpc never venture into the observable zone ( at least observable by hipparcos ) . on the other hand
the probability of finding the stars is greatest when they are near their major turning point , @xmath22 .
so it is clear that the mean @xmath21 , as well as to a lesser extent the eccentricity , are affected by selection effects .
stars belonging to the thin disk would have orbits with very small eccentricities and @xmath11 values ( solar values : @xmath8= 0.09 , @xmath11= 0.001 , see de boer et al .
1997a ) , while thick disk stars would have larger values on average .
halo stars have generally orbits with large eccentricities while their @xmath11 show a large range .
the eccentricities of the hba star orbits are very large , ranging from 0.5 to nearly 1.0 , the values for @xmath11 vary by a huge amount , from 0.04 ( hd 78913 ) to 5 ( hd 117880 ) . the stars bd + 36 2242 and feige 86 are exceptions , their values for both parameters are more appropriate for disk objects .
we note that these two stars are the hottest of the hba / b sample .
the kinematics of the four hbb stars from schmidt ( 1996 ) show overall behaviour similar to that of bd + 36 2242 and feige 86 ( fig .
all of these are hotter than 11000 k , the @xmath26 of bd + 36 2242 .
the star hd 117880 features an orbit somewhat dissimilar from the others . while its @xmath11 is very high
, its eccentricity is by far the lowest of the sample of hba stars . # 1#21.0 the hba stars ( @xmath26 @xmath27 10,000
k ) have a mean orbital velocity of @xmath15 = 17 kms@xmath19 , lagging about 200 kms@xmath19 behind the local standard of rest
. however , the velocity dispersions are large : 102 , 53 and 95 kms@xmath19 in @xmath18 , @xmath17 , @xmath28 respectively .
this shows that there are many stars with a non disk - like kinematical behaviour in the sample of hba / b stars .
they therefore belong to the galactic halo population rather than to the disk .
the orbital velocities of the hba stars in the sample do not have a gaussian distribution , as one might have expected .
instead , they seem to have a somewhat flatter distribution ( see fig .
about 75% of our stars have prograde velocities , four stars have retrograde orbits
. however the exact distribution can not be studied reliably due to the limited number of stars at disposal .
both the analysis of the kinematic properties and the shapes of the orbits imply that the hba / b stars mostly are members of the galactic halo population
. however , there seems to be a difference in kinematics and hence population membership between the cooler and the hotter stars .
stars cooler than about 10,000 k have low orbital velocities and a large spread in @xmath11 .
in contrast to this are the hotter stars whose kinematics and orbits are consistent with those of disk objects .
the hbb stars of schmidt ( 1996 ) which are all hotter than 10,000 k behave like sdb stars , too .
the sample of sdb / o stars show classical disk behaviour : their mean orbital velocity is @xmath29 kms@xmath19 , meaning a negligible asymmetric drift .
the @xmath17 velocity dispersion ( which is also the dispersion in @xmath15 , because the stars are in the solar vicinity ) is relatively small , similar to that of old thin disk orbits , while the dispersion in @xmath18 is much larger , fitting to thick disk values .
the dispersion @xmath30 is somewhere in between .
these values are quite similar to those of the sdb star sample of de boer et al .
( 1997a ) .
until now no population of _ field _ sdb stars with halo kinematics has been found . yet , hot subdwarfs of the horizontal branches of halo globular clusters are , of course , well known ( see e.g. moehler et al .
1997 ) . given the results above there seems to be a trend in the kinematics of star types along the blue part of the horizontal branch ( see fig .
3 ) . the sdb / o stars have disklike orbits .
the same probably applies to the hbb stars hotter than about 10,000 k , though the statistics are rather poor for this part of the hb .
in contrast to that stand the cooler hba stars which have much smaller orbital velocities , large orbital eccentricities and large ranges of @xmath11 , thus showing a behaviour fitting more to halo than to disk objects .
this result suggests to analyse the kinematics of the adjoining cooler stars of the hb , the rr lyraes . #
recently , martin & morrison ( 1998 ) carried out an investigation of the kinematics of rr lyrae stars which is mainly based on the study of layden ( 1994 ) . for our analysis we will use only those stars having hipparcos data .
six hipparcos stars were excluded because they have a proper motion error larger than 5 mas / yr . the rr lyrae stars present the observational difficulty in that they are variables with both @xmath17 and @xmath31 changing continously . for most of the sample
we were able to take the mean magnitudes from layden ( 1994 ) .
for the remaining stars we derived the intensity - mean magnitudes with help of the formula given by fitch et al .
( 1966 ) and revised by barnes & hawley ( 1986 ) which is the same method as used by layden ( 1994 ) , using the photometric data of bookmeyer et al .
the layden photometry was dereddened using the burstein & heiles ( 1982 ) reddening maps . for later steps in this study
it is necessary to know the mean @xmath31 of the rr lyrae stars . as the colour curves of the stars are quite similar to the brightness curves , with the star being bluest when it is near maximum brightness , we took the same formula as we used to calculate the mean magnitude .
this is not entirely correct but gives @xmath31 close to the actual one . for six stars we did not have the appropiate light curve data ,
so we could not determine the mean @xmath31 for them .
therefore only 26 rr lyraes are shown in fig .
3 . as the rr - lyrae stars are in most cases fainter and therefore farther away than our hba / b stars they have a rather large @xmath32 . for this reason we used the absolute magnitude derived in sect . 2.3 to calculate the distances for these stars .
we thus have ignored the effects of metallicity on @xmath33 for individual stars . also possible
evolutionary effects on @xmath33 ( see clement & shelton 1999 ) have been ignored , an aspect groenewegen & salaris ( 1999 ) did not consider in their determination of the rr lyrae @xmath33 either .
since we study the orbits of the rr lyrae as a sample these limitations will not affect our conclusions . for most rr lyrae stars we took the radial velocities from the sources mentioned in sect .
2.5 , supplemented by radial velocities from layden ( 1994 ) .
the metallicities of the rr - lyrae stars were taken from layden ( 1994 ) as far as possible .
a few values come from layden et al .
( 1996 ) and preston ( 1959 ) .
we calculated the orbits for the rr lyrae stars in the same manner as for the hba / b and sdb stars .
the rr - lyrae stars show a spread in kinematical behaviour wider than that of the hba / b stars .
many stars have orbits similar to those of the hba stars , others show disklike orbits with orbital velocities in the vicinity of 200 kms@xmath19 . of the halo rr - lyrae stars many have perigalactic distances smaller than 1 kpc , as we also found for the hba stars .
the rr - lyrae stars have orbital velocities typically spanning the entire range found for disk and halo stars ( see fig .
three members of our sample of rr lyr stars have orbits shaped somewhat different from those of the rest of halo orbits , looking similar to that of hd 117880 . in fig .
3 we have sorted the rr lyr stars according to their metallicity using different plot symbols . the stars with an [ fe / h]@xmath34 dex have high @xmath15 like disk stars .
the stars with lower metallicities are more evenly distributed in @xmath15 .
there are several stars with disk - like kinematics with a very low metallicity as low as [ fe / h]@xmath35 dex ( see table 4 ) .
the study of the spatial distribution of hb stars involves , unfortunately , several selection effects .
the general aspects have been reviewed by majewski ( 1993 ) and will not be repeated in detail here . yet , for each stellar type discussed in this paper a few comments are in place .
hba stars have in most cases been identified from photometry , notably because of a larger than normal balmer jump .
this larger jump is mostly due to lower metallicity of the stellar atmosphere .
if the atmospheric metallicity is identical to the original one , then the criterion favours intrinsically metal poor stars , which are presumably the older ones . however , also stars starting with a little more mass than the sun and thus of solar composition will become hb stars and , when as old as the sun , by now are solar metallicity hb stars .
if they were of hba type , they would not have been recognized in photometry of the balmer jump .
such stars would be underrepresented in our sample .
the hba stars considered here come from all galactic latitudes , so that selection effects due to galactic latitude are not to be expected .
however stars which have orbits going far away from the disk are always underrepresented , as their fraction of time near the disk ( and hence being observable ) is much smaller than for those which do not go far from the disk .
rr lyraes , being variables , are not prone to such selection effects .
most of them are identified solely by their variability .
metallicity or high velocity are generally not used as criteria for the identification for rr lyrae stars . for a discussion of selection effects due to galactic latitude
we refer to martin & morrison ( 1998 ) , as our sample is a subsample of theirs .
the sdb / o stars were identified in surveys for quasars , e.g. the pg catalogue ( green et al .
1986 ) or hamburger quasar survey ( hagen et al . 1995 ) .
this means their blue colour is the criterion , rather than proper motion , radial velocity or metallicity .
therefore we do not expect a selection bias towards metal poor halo stars .
moreover , de boer et al .
( 1997a ) showed that the sdb / o stars observed now near the sun come from widely differing locations in the milky way .
as these catalogues only map objects which are somewhat away from the galactic plane , they miss the majority of stars with solar type orbits .
sdo stars may be confused with pagb stars descending down the hrd towards the white dwarf regime .
the hbb stars of schmidt ( 1996 ) are also taken from the pg catalogue , so that there should not be noticeable selection effects , either .
however hbb stars and main sequence stars have similar physical properties such as log@xmath36 , so that there may be confusion with the latter .
apart from this the selection effects mentioned for the sdb / o stars apply to the hbb stars , too .
finally , some words concerning the distribution of distances of the different samples are in place .
generally , if one deals with stars having different absolute magnitudes , as in our case when the sdbs are several magnitudes fainter than the hbas , one gets samples with different mean distances .
the intrinsically fainter stars are on average much nearer than the brighter stars , if the two groups have similar apparent magnitudes .
this means that the spatial regions sampled differ depending on the absolute magnitude of the stars .
this would imply that the sdb sample is biased towards disk stars as we do not sample them far enough from the galactic plane where there may be a higher concentration of halo stars than further in .
this is however not the case . as we include some of the results of de boer et al .
( 1997a ) which come from a completely different source , namely mostly from the pg - catalogue ( green et al .
1986 ) dealing with significantly fainter stars , the pg stars actually have on average larger distances than any of our hba stars .
for this reason we do not expect that the difference in kinematics arises from the distribution of the distances in the samples .
the _ sdb stars _ have in general rather disk - like orbits and kinematical properties .
the ones analysed here ( table 4 ) show the same behaviour as those from the large sample of sdb stars investigated previously ( de boer et al .
1997a ) .
the ( hotter ) _ hbb stars _ behave rather like the sdbs with orbits of disk - like characteristics .
however , such stars are difficult to recognize and our sample is small .
a much larger sample may show a larger variation in kinematics .
the _ hba stars _ have mostly halo orbits ( mean @xmath37 kms@xmath19 ) .
this is very similar to the value at which most other studies concerning metal poor stars in the solar neighbourhood arrive ( see table 2 of kinman 1995 ) .
however , the known sample may be observationally skewed toward stars with low atmospheric metallicity ( large balmer jump ) .
the _ rr - lyrae stars _ have orbits spanning a large range in orbital parameters , too
. however , a trend seems to be present with metallicity .
the metal poor stars have halo orbits similar to those of the hba stars with rather low orbital velocities of less than 100 kms@xmath19 , and large @xmath8 and @xmath11 .
the metal rich stars on the other hand have rather disk - like kinematical characteristics .
a similar distribution of metallicities and orbital velocities was also found in the studies of chen ( 1999 ) and martin & morrison ( 1998 ) .
although there are a few rr lyraes having high orbital velocities ( @xmath38 kms@xmath19 ) and clearly disk - like orbits ( some of which are very metal poor ) , hba stars with such characteristics are not found in our sample . on the other hand
no rr lyraes with [ fe / h]@xmath34 dex with halo - like orbits or kinematics are present .
this means that a high metallicity for a rr lyr star is a good indicator that it is a disk star .
however , a low metallicity does not mean that a star neccessarily belongs to the halo .
since the sdb stars ( and possibly the hbb stars ) have disk - like orbits , these stars must be part of a relatively younger , more metal rich group among the hb stars .
majewski ( 1993 ) uses the expression ` intermediate population ii ' , other authors use the words ` thick ' or ` extended disk ' .
in addition to the disk - nature of their orbits , the vertical distribution is consistent with a scale height of the order of 1 kpc ( villeneuve et al .
1995 , de boer et al .
since the amount of metals in their atmospheres may have been altered by diffusion it is not possible to estimate the true age from the metallicity .
the hba stars have really halo orbits .
this must mean they belong to a very old population .
their atmospheric metal content is low indeed , the determinations showing a large scatter per star and from star to star ranging between @xmath39 and @xmath40 dex .
however , metal rich hba stars which are known to exist in star clusters ( see peterson & green 1998 ) , would likely be underrepresented in the sample .
if the halo contains mostly old stars , like globular cluster stars , then the resulting halo hb stars should occupy the hb in ranges related with metallicity as with the globular clusters ( see renzini 1983 ) .
the very metal poor ones ( [ @xmath41 @xmath42 dex ) would be hb stars of hbb and hba nature as well as rr lyrae , the ones of intermediate metallicity ( [ @xmath41 @xmath43 dex ) would be very blue down to sdb like , and the metal rich ones ( [ @xmath41 @xmath44 dex ) would be rhb stars , perhaps including some rr lyrae .
this behaviour may also explain the existence of the two oosterhof groups ( see van albada & baker 1973 or lee et al .
1990 ) of rr lyrae , since only the very metal poor and the relatively metal rich globular clusters contain rr lyrae .
evolutionary changes of the hb stars may also affect the location on the hb ( sweigart 1987 , clement & shelton 1999 ) . however , sdb stars with halo kinematics have not been found ( de boer et al .
instead , they have only disk orbits .
this must mean that the stars which originally formed in the halo had an initial mass , a metallicity and a red giant mass loss such that rr lyrae and hba stars were the end product , and not sdb stars .
as for the rr lyrae stars , they show a wide range in kinematic behaviour , more or less in line with the atmospheric metal content .
the actual metallicity did not bias the identification of these stars , since they are selected based on variability .
one tends to divide the rr lyrae sample into metal poor and metal rich rr lyrae ( see layden 1994 ) .
here we recall that in the hb stars the contents of heavier elements in their atmospheres may be altered ( see sect .
the rr lyrae stars with the continuous upheaval of the pulsation may stimulate mixing so that their atmospheres probably show the true metallicity . thus , for rr lyraes the metallicity may be used as a general population tracer .
the observed range of metallicities would mean that there are old as well as younger rr lyraes .
old rr lyrae must be very metal poor and should have halo orbits .
the majority of the rr lyrae included in our analysis fit these parameters .
there are , however , a substantial number of rr lyr stars in our sample with disk - like kinematics but low metallicities , in several cases as low as @xmath40 dex .
the origin of this group of stars , dubbed the ` metal weak thick disk ' , is still unknown ( see martin & morrison 1998 for a discussion ) .
young ( or younger ) rr lyrae should be relatively metal rich and have disk orbits .
the investigated sample contains such stars .
these objects should have an age , main - sequence mass , metallicity and rgb mass loss such that rr lyrae emerge , i.e. hb stars with a thicker hydrogen shell .
they are , being relatively metal rich , also of slightly different @xmath33 than the metal poor and old ones .
in fact , they are fainter and their distances should be based on the appropriate brightness - metallicity relation .
the dependence is , however , feeble and amounts to just 0.1 mag for 0.5 dex .
we tested how serious ignoring this effect is on the derived orbits by reducing the rr lyr star distances by 10 % .
it does not lead to a change of significance in the histogram of fig.4 .
our orbit studies allow to see a trend in the kinematics of the field hb stars along the horizontal branch .
this appears to give us access to the structure of the milky way and its halo as well as information about possible formation scenarios .
the trends related with age and history could only be found using the kinematics , since it has become clear that the atmospheric metallicity in hb - like stars has no relation to the one of the main sequence progenitor .
the location of the stars on the hb must be a complicated function of age , main - sequence mass , initial metallicity , and mass loss on the rgb . for the hb - like stars of today indications for the age
can be determined from the present kinematic parameters . only detailed models for metallicity dependent stellar evolution from main sequence through the rg phase with mass loss
should , in comparison with the observables of horizontal branch stars , eventually be able to retrieve the true origin of the hb stars . | horizontal branch ( hb ) stars in the field of the milky way can be used as tracers for the study of early stages of the evolution of our galaxy . since the age of individual hb stars
is not known a priori , we have studied the kinematics of a sample of field hb stars measured with hipparcos to look for signs of age and population nature .
our sample comprises 14 hba , 2 hbb and 5 sdb / o stars .
we found that the kinematics of the hba stars is very different from that of the sdb / o stars ( including those from an earlier study ) .
the hba stars have low orbital velocities , some are even on retrograde orbits .
their orbits have large eccentricities and in many cases reach large distances above the galactic plane .
in contrast , the sdb / o stars show disk - like orbital characteristics .
the few hbb stars ( with @xmath0 k ) in our sample seem to have kinematics similar to that of the sdb / o stars . in order to see if there is a trend among the hb stars in their kinematics , we investigated also rr lyrae stars measured with hipparcos .
here we found a mixed kinematical behaviour , which was already known from previous studies .
some rr lyrae stars have disk - like orbits ( most of these being metal rich ) but the majority has halo - like orbits , very similar to those of our hba stars .
since the atmospheres of most types of hb stars do not reflect original metallicities any more the kinematics is the only aspect left to study the origin and population membership of these stars .
thus , the clear trend found in kinematics of stars along the hb , which is also a sequence in stellar mass , shows that the different kinds of field hb stars arose from stars having different origins in age and , e.g. , metallicity or mass loss rate . |
conventional ( @xmath0 ) josephson junctions ( jjs ) in the ground state , have zero @xmath5 , the gauge invariant phase difference of the superconducting order parameters on each side of the barrier . however , in some cases a jj may have a ground state with @xmath5 equal to @xmath6 ( @xmath1jj ) .
the possibility of a @xmath1jj was first pointed out theoretically in the tunneling through an oxide layer with magnetic impurities in sis junctions , because of negative josephson coupling @xcite .
negative coupling also arises in jjs with ferromagnetic intermediate layers @xcite . in that case
the state of the jj , i.e. , a @xmath0state or a @xmath1state depends sensitively on the sample design and the temperature @xcite .
another realization of @xmath1jjs became possible due to the @xmath7wave symmetry of the order parameter in high@xmath8 superconductors .
thus , @xmath1jjs were demonstrated using twist - tilt , and @xmath9 tilt @xmath10 grain boundary jjs @xcite .
very recently , a new technique of tailoring the barrier of superconductor - insulator - ferromagnet - superconductor ( low @xmath11 ) junctions can be used for accurate control of the critical current densities of @xmath1jjs ( as well as @xmath0jjs and @xmath12 jjs ) @xcite .
all these junctions are characterized by an intrinsic phase shift of @xmath6 in the current - phase relation or , in other words , an effectivelly negative critical current . one - dimensional ( 1d ) discrete arrays of parallel biased @xmath0jjs for which both dimensions are smaller than the josephson length @xmath13 ( i.e. , small jjs ) , have been studied extensively in recent years @xcite .
such systems represent an experimental realization of the spatially discrete sine - gordon ( sg ) lattice , whose dynamic behaviour has been modelled with the discrete sine - gordon ( dsg ) equation .
that equation is formally equivalent to the frenkel - kontorova model , which describes the motion of a chain of interacting particles subjected to an external on - site sinusoidal potential @xcite .
experiments and numerical simulations have shown that , even with large discreteness , the dynamics of a localized kink ( fluxon ) in the sg lattice exhibit some features of solitonic nature , close to the properties of the continuous sg solitons @xcite .
however , discreteness also introduces a number of aspects in fluxon dynamics which have no counterparts in the continuum system .
the experimental investigation of discrete jj arrays is a relevant issue in superconducting electronics because they are the basis of the so called phase - mode logic @xcite and of the rapid single - flux - quantum ( rsfq ) circuits @xcite , for the construction of superconducting quantum interference filters ( sqifs ) @xcite , as well as for developing high frequency mixers @xcite , or variable inductors and tunable filters @xcite .
naturally , the next step in this field was to consider jj arrays comprising @xmath1jjs .
the usage of @xmath1jjs as passive @xmath1shifters in rsfq circuits was recently suggested in reference @xcite , and realized in @xcite .
moreover , recent studies ( both theoretical and experimental ) of jj arrays with @xmath1jjs have shown some novel features arising out of the interplay between @xmath0 and @xmath1jjs @xcite .
for instance , when a @xmath14 kink is introduced in a hypothetical parallel array with alternating @xmath0 and @xmath1jjs @xcite , it will break up into two separate @xmath1kinks .
this effect is reffered to as `` fractionalization of a flux quantum '' and , surprisingly , it also appears in long continuous jjs with alternating @xmath15 and @xmath6 facets @xcite . in the limit of very short , randomly alternating @xmath15 and @xmath6 facets ,
that problem has been studied in references @xcite .
given that @xmath1jjs can now be formed in a variety of ways , a parallel array of alternating @xmath0 and @xmath1jjs could , in principle , be made in the lab .
it is worth noting that parallel jj arrays consisting of both @xmath0 and @xmath1 jjs were also studied as a model of long , faceted , high@xmath8 bicrystal jjs with high misorientation angle of the bicrystalline film ( @xmath16 ) @xcite . in the present work ,
we investigate numerically the fluxon dynamics in a 1d , highly discrete array of alternating @xmath0 and @xmath1jjs .
we are mainly concerned with linear arrays , with open ended boundary conditions , and different ( even or odd ) total number of jjs . in the next section ,
we describe the array model and typical ground states of such arrays , while in section 3 we present numerically obtained , typical current - voltage characteristics where zero field - like ( zf - like ) steps appear .
we finish in section 4 with the conclusions . and
@xmath1jjs in parallel .
solid lines show the superconducting electrodes , while the crosses and the crosses - in - box show the @xmath0 and @xmath1jjs , respectively . ]
consider a 1d array of alternatingly @xmath0 and @xmath1jjs , which are connected in parallel via superconducting leads , as shown schematically in figure 1 .
the hamiltonian function of this system in the simplest approximation , where all mutual inductances between different cells are neglected , is given by @xmath17 where @xmath18 is the phase difference over the @xmath19th jj , @xmath20 is the critical current of the @xmath19th jj , @xmath21 is the dissipation coefficient due to quasiparticles crossing the barriers , @xmath22 is the coupling strength between nearest neighbouring jjs , with @xmath23 being the discreteness parameter , @xmath24 is the applied constant ( dc ) bias current assumed to be injected and extracted uniformly ( see figure 1 ) , and @xmath25 is the canonical variable conjugate to @xmath18 . for @xmath26
that variable reduces to the ( normalized ) voltage - phase relation , @xmath27 , for the @xmath19th jj .
since @xmath28 depends explicitly on time @xmath29 , it is obvious that it is not a constant of the motion .
however , it is interesting that the system in consideration belongs in a certain class of systems having time - dependent hamiltonians , which may be useful when studying stability issues of specific solutions . for @xmath26 and @xmath30
we get the simpler hamiltonian @xmath31 which has been used ( with @xmath32 ) to describe a harmonic chain of particles subjected to an external sinusoidal potential @xcite . in this context , the variable @xmath18 represents the displacement of a particle from its equilibrium position , and the three terms in the earlier hamiltonian @xmath33 are interpreted as the kinetic energy of the particles , the harmonic ( elastic ) interaction of the nearest neighbouring particles in the chain , and the interaction of the chain with an external on - site sinusoidal potential , respectively . if the ratio @xmath34 of the coefficients of the elastic coupling energy to the on - site potential energy is less than unity , then the system is considered to be highly discrete . in respect
to kink propagation , the system is highly discrete if the kink width is of the order of lattice spacing @xmath23 .
the @xmath19dependent critical current @xmath20 is given in compact form by @xmath35 , \end{aligned}\ ] ] where @xmath36 with @xmath37 ( @xmath38 ) being the critical current of the @xmath0 ( @xmath1 ) jjs . from equations ( [ 3 ] ) and ( [ 4 ] ) we have that @xmath39 and @xmath40 for odd and even @xmath41 , respectively .
note that @xmath20 may differ both in sign and magnitude for the @xmath0 and @xmath1jjs . from the hamiltonian function given in equation ( [ 1 ] )
we get the equations @xmath42 vs plaquette number @xmath41 , for ( a ) a periodic array ( @xmath43 ) ;
( b ) an open - ended array with even @xmath44 ( @xmath43 ) ; ( c ) an open - ended array with odd @xmath44 ( @xmath45 ) . @xmath46 , and @xmath47 . ] for a finite array with @xmath44 total number of jjs , @xmath48 .
the earlier equations may also be obtained by applying kirchhoff laws in the equivalent electrical circuit @xcite , where each single jj is described by the resistively and capacitively shunted junction ( rcsj ) model . in this model
the jj consists of a capacitive branch , a resistive branch , and a superconducting branch , all connected in parallel .
equations ( [ 5 ] ) are written in the standard normalization form , i.e. @xmath49 is normalized to @xmath50 with @xmath51 being the spatial separation of neighbouring jjs , the temporal variable @xmath29 to the inverse plasma frequency @xmath52 , and @xmath53 is normalized to @xmath37 . then , @xmath54 in equations ( [ 5 ] ) , and @xmath38 actually represents the ratio @xmath55 .
the voltage @xmath56 across the @xmath19th jj is normalized so that @xmath57 where @xmath58 , and @xmath59 is the flux quantum ( with @xmath60 and @xmath61 being the planck s constant and the electron charge , respectively ) .
equation ( [ 5 ] ) is analogous to that deduced in reference @xcite in the context of analysis of zig - zag arrays @xcite .
below we deal with arrays both with periodic and open ends , i.e. , annular and linear finite arrays , respectively .
then , in both cases , the governing equations ( [ 5 ] ) remain the same for @xmath62 .
the same equations are also valid at the end points @xmath63 and @xmath64 , if we define artificial phases @xmath65 and @xmath66 such that @xmath67 for periodic boundary conditions ( e.g. , @xcite ) , and @xmath68 for open - ended boundary conditions .
the integer @xmath69 in equation ( [ 66 ] ) is the number of @xmath14 fluxons trapped in the annular array .
equations ( [ 5 ] ) are integrated with a standard fourth order runge - kutta algorithm with fixed time - stepping ( typically @xmath70 ) . in figure 2
we plot the calculated normalized flux @xmath71 in each elementary cell ( plaquette ) as a function of the junction number @xmath41 , for three different arrays in their ground state : a periodic array ( figure 2(a ) ) , and open - ended arrays with even and odd @xmath44 ( figures 2(b ) and 2(c ) , respectively ) . in the former case the flux clearly alternates between two values of opposite sign and the same magnitude , the latter depending on the coupling constant @xmath34 , indicating that the flux lattice is actually periodic with a period of two elementary cells .
we can thus define a new extented unit cell , with respect to the flux periodicity , which consists of two elementary cells .
then , we can evaluate the ( normalized ) average flux in each of the extended cells , @xmath72 , where @xmath73 numbers the extended cell composed of the ( @xmath74)th and @xmath75th elementary cells .
clearly , @xmath76 vs @xmath77 is a constant which equals to zero .
the same remarks also hold for the open - ended arrays , both with even and odd @xmath44 , relatively far from the end jjs .
for obtaining the current - voltage ( @xmath78 ) characteristics for an open - ended array , we initialize the system with @xmath79 and @xmath6 for @xmath41 odd and even , respectively , and @xmath80 for any @xmath41 at @xmath81 and @xmath30 .
( the overdot denotes derivation with respect to the temporal variable . ) then equations ( [ 5 ] ) are integrated forward in time up to @xmath82 , by which time the system has reached a steady state .
we then calculate the time averaged voltage @xmath83 across each jj by averaging over an additional @xmath84 dimensionless time - units , i.e. , @xmath85 , where @xmath86 .
the voltage of the array @xmath87 is obtained by averaging over all jjs , i.e. , @xmath88 .
the dc bias @xmath24 is then increased by @xmath89 and the calculation is repeated , assuming as initial condition the steady state achieved in the previous calculation .
we continue this procedure increasing @xmath24 by @xmath89 until @xmath90 , then decrease @xmath24 with same steps back to zero , obtaining the @xmath78 characteristics shown in figures 3 and 4 for even @xmath44(=100 ) and odd @xmath44(=101 ) , respectively ( @xmath91 for the curves in the main figures and @xmath92 for the curves in the insets ) .
large hysteresis is observed in the high - voltage part ( @xmath93 ) of all characteristics in both figures , even though the damping coefficient is rather high ( @xmath94 ) .
small hysteretic effects are also apparent in the low - voltage parts ( not shown for clarity ) .
apparently , the high - voltage characteristics in figures 3 and 4 for same @xmath34 are practically the same . the solutions on the high voltage branches have been discussed in detail in reference @xcite , and it seems that they are not significantly affected by the choice of boundary conditions . here , we focus on the the low - voltage parts of those characteristics , which exhibit significant quantitative differences for odd and even @xmath44 . that can be seen by comparing the insets of figures 3 and 4 , which display a magnified portion of the corresponding low - voltage characteristics seen in the main figures .
there we can see the appearence of current steps , similar with those observed in @xmath0jj arrays ( see for example references @xcite ) .
those steps are the discrete counterparts of the zero - field steps ( zfss ) appearing in the @xmath78 characteristics of continuous , long @xmath0jj , due to fluxon resonant motion . in this case
the propagating fluxons are @xmath4kinks and their multitudes , while single @xmath1kinks are unstable . in a continuous , long @xmath0jj , with equivalent normalized length @xmath95 ,
the normalized voltage at maximum fluxon velocity @xmath96 of the first zfs would be @xmath97 for @xmath47 .
thus , the first half - integer zfs ( zfs1/2 ) would have @xmath98 , while the second integer zfs ( zfs2 ) and second half - integer zfs ( zfs3/2 ) would have @xmath99 and @xmath100 , repectively .
the fluxon width decreases ( i.e. , the fluxon size contracts ) with increasing velocity .
thus , in the discrete system , a fluxon can not reach the maximum velocity of its continuous counterpart , since its contracted size can not be smaller than @xmath23 . in each of the insets in figures 3 and 4
there are two zfs - like branches with @xmath47 , with maximum @xmath87 ( defined at the top of the steps ) equal to @xmath101 , @xmath102 , and @xmath103 , @xmath104 , respectively .
thus , the steps in the inset of figure 3 can be identified with the discrete counterparts of half - integer zfss ( namely with zfs@xmath105 and zfs@xmath106 ) , while those in the inset of figure 4 with the discrete counterparts of integer zfss ( zfs@xmath107 and zfs@xmath108 ) .
inspection of the time evolution of the phases ( not shown ) reveals that the zfs@xmath105-like branch for even @xmath44(=100 ) is due to a @xmath1kink propagating in the array , on a staggered phase background . in figure 5
we show the instantaneous profile of the normalized flux as a function of the plaquette number @xmath41 , and @xmath76 as a function of the number of the extended cell @xmath77 , for @xmath109 .
the antiferromagnetic background in the flux ( figure 5(a ) ) obscures the picture , but its proper elimination with the doubling of the unit cell along with the definition of @xmath76 makes things more clear ( figure 5(b ) ) .
then , in figure 5b we see clearly the discrete @xmath1kink in a form which is the discrete counterpart of the spatial phase derivative in continuous systems .
also apparent they are small oscillations behind the @xmath1kink , a phenomenon known from @xmath0jj arrays @xcite , where moving @xmath4kinks excite linear oscillating modes of the array ( cherenkov radiation ) . those oscillations , however , get strongly damped and die out fast away from the kink . in figure 6 the same quantities for an array with the same parameters but odd @xmath44(=101 ) are plotted for @xmath109 on the zfs@xmath107-like branch . here
we always have a @xmath1kink and a @xmath1antikink pair .
cherenkov radiation is also apparent here ( figure 6(b ) ) behind both the propagating @xmath1kinks .
inserting one ( or more ) @xmath4kink(s ) in both the even and odd @xmath44 arrays , we get higher order zfs - like branches . the zfs@xmath106- and zfs@xmath108-like branches for even and odd @xmath44 ,
respectively , are shown in the insets of figure 3 and 4 , respectively ( only for @xmath47 ) .
those branches were obtained by inserting one @xmath4kink in the arrays .
again , inspection of the phases reveals that the @xmath4kink is fractionalized into two closely spaced but clearly separated @xmath1kinks which , however , are not identical .
the snapshot in figure 7 shows the normalized flux as a function of the plaquette number @xmath41 , and @xmath76 as a function of the number of the extended cell @xmath77 , for @xmath109 on the zfs@xmath106-like branch in the inset of figure 3 ( @xmath43 , even ) .
the fractionalized @xmath4kink at the right of figure 7(b ) , now coexists with the single @xmath1kink in the array . the corresponding figure ( figure 8) for the zfs@xmath108-like branch in the inset of figure 4 ( @xmath45 , odd ) , shows the fractionalized @xmath4kink along with a @xmath1kink - @xmath1antikink pair .
the direction of motion of the fractionalized @xmath4kinks depends on their polarity when they are inserted in the array .
these kinks , either the single one or the ones obtained by fractionalization , behave in many aspects like the usual @xmath4kinks .
they get reflected by the ends changing their polarity , and interact completelly elastically , passing the one through the other without any apparent shape change . although not shown in figures 3 and 4 , the depinning current @xmath110 related to the peierls - nabarro potential , which the kink should overcome in order to start moving , is practically the same for @xmath44 even and odd .
it increases with increasing discreteness ( decreasing @xmath34 ) , having the values @xmath111 and @xmath112 for @xmath113 and @xmath114 , respectively .
the fractionalized @xmath4kinks shown for a specific illustrative case in the figures 7(b ) and 8(b ) , are closely spaced and they move together in the same direction without changing their distance of separation .
thus , they essentially form a single bound state which resembles the bunched fluxon state observed in @xmath0jj arrays @xcite .
such a state results from strong interactions between kinks , which is mediated by their cherenkov `` tails '' . as a result
, the cherenkov radiation behind the bunched state is strongly suppressed , which is exactly what we observe in our figures 7(b ) and 8(b ) . in annular @xmath12 jj arrays with both even and odd @xmath44 , which satisfy the boundary conditions ( [ 66 ] )
, the effect of fractionalization of @xmath4fluxons also appears @xcite . in that case
, @xmath4fluxons can be inserted in the array either by setting @xmath69 to a nonzero integer value or by an appropriate initial condition . for both even and odd @xmath44 and a nonzero dc bias @xmath24 ( e.g. , @xmath109 ) ,
there is a number of @xmath1kinks and @xmath1antikinks moving along the array such that the total flux is always an integer as required by the boundary conditions .
they appear either as @xmath1kink - @xmath1antikink pairs moving in opposite directions , or as two bunched @xmath1kinks looking very much like those in figures 7 and 8 .
however , we did not find single @xmath1kinks in this case , so that half - integer zf - like branches do not appear .
thus , it seems that half - integer zf - like branches appear only for arrays with even @xmath44 satisfying the boundary conditions ( [ 6 ] ) , i.e. , for linear arrays .
this is probably due to the amount of the net total flux in a @xmath12 jj array in its ground state . for arrays satisfying periodic boundary conditions
without inserted flux , with either odd or even @xmath44 , the ground state is always arranged such that the total flux is equal to zero .
apparently , this is not always the case for open - ended boundary conditions . for odd @xmath44
there is an even number of elementary cells in the array , so that the contributions from all cells exactly cancel , resulting in zero total flux .
however , for even @xmath44 there is an odd number of elementary cells in the array , and the total flux attains a nonzero value ( approximatelly equal to @xmath115 in figure 2(b ) ) . with increasing bias current above the peirls - nabarro barrier , that ground state becomes unstable against the formation of a @xmath1kink .
dynamic simulations for such @xmath12 jj arrays with periodic boundary conditions have been performed in reference @xcite , where it was shown that @xmath4kinks fractionalize into a pair of @xmath1kinks in certain regions of @xmath116 space .
however , zf - like steps on the current - voltage characteristics are not discussed .
although a @xmath12 jj array is still hypothetical , it can be possibly realized with the existing technology .
moreover , it has been argued @xcite that bicrystal grain boundary jjs with @xmath9 misorientations angle , form intrinsically an array of alternatingly @xmath15 and @xmath6 contacts . in this context , a discrete array of alternating @xmath0 and @xmath1jjs has been used as a model to study the dynamics as well as the magnetic field dependence of such jjs @xcite .
in conclusion , we caclulated numerically the @xmath78 characteristics for a 1d discrete , planar array of small , parallel connected alternating @xmath0 and @xmath1jjs , which exhibit current steps in the low - voltage part . for even @xmath44 ,
the maximum voltage of the steps ( evaluated at the top of each step ) is approximately two times smaller than the maximum voltage of the steps for odd @xmath44 .
comparison with the continuous long jj , helps in identifying those steps as half - integer and integer zfs - like steps for even and odd @xmath44 , respectively .
inspection of the phases and the fluxes reveals that in both cases the system supports @xmath1kinks , either a single @xmath1kink or a @xmath1kink - @xmath1antikink pair for even and odd @xmath44 , respectively , which can propagate in the array .
moreover , any @xmath4kink inserted in the array is fractionalized into two , generally different @xmath1kinks , which propagate together in the array along with the single @xmath1kink or the @xmath1kink - @xmath1antikink pair , for even and odd @xmath44 , respectively , giving rise to higher order half - integer and integer zfs - like branches . in the continuum
, half - integer zfss have been observed experimentally in not very long @xmath12 jjs due to the hopping of the semifluxons @xcite .
similar zfs - like branches can be found in rather wide parameter range .
the case with @xmath47 , for which the system is already highly discrete , was chosen only for better illustration of the half - integer zfss .
those steps can also be found even for low damping ( i.e. , @xmath117 ) , as well as for higher damping ( i.e. , @xmath118 ) , which is more appropriate for high - temperature superconductor jj arrays @xcite . also , they may exist even for slightly different critical currents ( i.e. , for @xmath119 for even @xmath44 ) .
dynamic states such as those described above could in principle be detected by low - temperature scanning electron microscopy ( ltsem ) imaging @xcite .
the author thanks zoran radovi for many useful discussions on @xmath6 junctions . | the dynamics of a one - dimensional , highly discrete , linear array of alternating @xmath0 and @xmath1 josephson junctions is studied numerically , under constant bias current at zero magnetic field .
the calculated current - voltage characteristics exhibit half - integer and integer zero - field - like steps for even and odd total number of junctions , respectively .
inspection of the instantaneous phases reveals that , in the former case , single @xmath1kink excitations ( discrete semi - fluxons ) are supported , whose propagation in the array gives rise to the @xmath2step , while in the latter case , a pair of @xmath1kink @xmath1antikink appears , whose propagation gives rise to the @xmath3step .
when additional @xmath4kinks are inserted in the array , they are subjected to fractionalization , transforming themselves into two closely spaced @xmath1kinks .
as they propagate in the array along with the single @xmath1kink or the @xmath1kink - @xmath1antikink pair , they give rise to higher half - integer or integer zero - field - like steps , respectively . |
nonlinear filters are important tools for dynamical data assimilation with applications in a variety of research areas , including biology @xcite , mathematical finance @xcite , signal processing @xcite , image processing @xcite , and multi - target tracking @xcite . to put it succinctly , nonlinear filtering is an extension of the bayesian framework to the estimation and prediction of nonlinear stochastic dynamics . in this effort
, we consider the following nonlinear filtering model @xmath0 where @xmath1 and @xmath2 are two nonlinear functions , @xmath3 and @xmath4 are the stochastic state and observation processes , respectively , @xmath5 is a random vector representing the uncertainty in @xmath6 , and @xmath7 denotes the random measurement error in @xmath8 . in the discrete setting , the nonlinear filtering model in takes the form @xmath9 where @xmath10 and @xmath11 are mutually independent white noises .
let @xmath12 denote the @xmath13 filed generated by the observational data up to the step @xmath14 .
the goal of nonlinear filtering is to find the posterior probability density function ( pdf ) of the state @xmath15 , given the observation data @xmath16 , so as to compute the quantity of interest ( qoi ) , given by @xmath17 = \inf\left\ { \mathbb{e } [ | \phi({x}_k ) - z |^2 ] ; z \in \mathcal{z}_k \right\},\ ] ] where @xmath18 is a test function , and @xmath19 denotes the space of all @xmath16-measurable and square integrable random variables .
tremendous efforts have been made to solve nonlinear filtering problems in the last few decades .
two of the well - known bayesian filters are extended kalman filters ( ekfs ) @xcite , and particle filters @xcite .
the key ingredient of the ekfs is the linearization of both @xmath1 and @xmath2 in , so that the standard kalman filter can be applied directly .
thus , if the nonlinearity of the state and the observation systems is not severe , then the ekfs can provide efficient and reasonable inferences about the state , otherwise , the performance of the ekfs can be very poor . for particle filters , the central theme is to approximate the desired posterior pdf of the state by the empirical distribution of a set of adaptively selected random samples ( referred to as `` particles '' ) .
the particle filter method is essentially a sequential monte carlo approach , which requires no assumption on the linearity of the underlying system . as such , with sufficiently large number of particles
, it is capable of providing an accurate approximation of the posterior pdf for a highly nonlinear filtering problems .
however , there are some fundamental issues concerning the efficiency and robustness of particle filters @xcite . for example , since the empirical pdf is constructed based on particles with equal weights after resampling , the particle filter still needs a lot of samples in order to accurately approximate the target distribution . to overcome such a disadvantage , the authors proposed a new nonlinear filter named implicit filter " @xcite .
this approach adopts the framework of bayesian filtering , which has two stages at each time step , i.e. , prediction and update . at the prediction stage
, we estimate the prior pdf @xmath20 of the future state @xmath21 given the current available observation information @xmath22 ; at the update stage , we update the prior pdf by assimilating the newly received data @xmath23 to obtain the estimate of the posterior pdf @xmath24 . the implicit filter is distinguished from the particle filters by the use of interpolatory approximations to the prior and posterior pdfs .
specifically , in the particle filter , @xmath20 is approximated by _
explicitly _ propagating the samples of the current state @xmath25 through the nonlinear state equation @xmath26 , and constructing the empirical pdf of @xmath27 . in the implicit filter
, the interpolation of @xmath20 requires its function values at a set of grid points of the future state @xmath28 . under the condition that @xmath29
, we solve _ implicitly _ the state equation @xmath30 given a set of monte carlo samples of @xmath31 , so that the value of @xmath32 , at the grid point of @xmath33 , can be estimated by averaging the function values of @xmath34 at all the solutions of the state equation . as an implicit scheme , the implicit filter has a stabilizing effect which provides more accurate numerical approximations to the solution of the nonlinear filtering problem than the particle filter method @xcite .
the main challenge of the implicit filter method is that the conditional pdf of the nonlinear filtering solution is estimated at grid points .
as such the method suffers the so called `` the curse of dimensionality '' when the dimension of the state variable is high .
in addition , the efficiency of the method may be significantly reduced when the domain of the pdf is unbounded . in this paper
, we propose to construct a meshfree implicit filter algorithm to alleviate the aforementioned challenges .
motivated by the particle filter method , we first generate a set of random particles and propagate these particles through the system model and use these particles to replace the grid points in the state space .
after that we generate other necessary points through the shepard s method which constructs the interpolant by the weighted average of the values on state points @xcite . in order to prevent particle degeneracy in the generation of random state points
, we introduce a resample step in the particle propagation .
in addition we choose state points according to the system state , which make them adaptively located in the high probability region of the pdf of state . in this way , we solve the nonlinear filtering problem in a relatively small region in the state space at each time step and approximate the solution on a set of meshfree state points distributed adaptively to the desired pdf of the state .
furthermore , since we approximate the pdf as a function on each state point , instead of using state points themselves to describe the empirical distribution , the implicit filter algorithm requires much fewer points than the particle filter method to depict the pdf of the state .
the rest of this paper is organized as follows . in [ bf ] , we introduce the mathematical framework of the bayesian optimal filter . in
[ algorithm ] , we construct meshfree implicit algorithm . in
[ sec : ex ] , we demonstrate the efficiency and accuracy of our algorithm through numerical experiments . finally , [ sec : con ] contains conclusions and directions for the future research .
for @xmath35 , let @xmath36 and @xmath37 denote the @xmath13 fields generated by @xmath38 and @xmath39 , respectively . for @xmath40
, we use @xmath41 to represent a realization of the random variable @xmath28 , and define @xmath42 for notational simplicity .
it is easy to see that the dynamical model in is markovian in the sense that @xmath43 we also know that the measurements are conditionally independent given @xmath41 , i.e. , @xmath44 the bayesian optimal filter constructs the conditional distribution @xmath45 recursively in two stages : prediction stage and update stage .
for @xmath46 , assume that @xmath47 is given .
in the prediction stage @xmath48 is evaluated through the chapman - kolmogorov formula : @xmath49 in the update stage , the prior pdf obtained in is used to obtain the posterior pdf @xmath50 via the bayes formula : @xmath51
in this section , we construct the meshfree implicit filter algorithm .
the algorithm is based the implicit filter algorithm on grid points @xcite . the implicit filter algorithm introduced in @xcite
is developed from the general framework of the bayesian optimal filter discussed above , in which the primary computational challenge is the numerical approximation of the term @xmath52 in . for @xmath53 ,
the goal of this stage is to approximate the prior distribution @xmath48 of the state @xmath28 , given the posterior distribution @xmath54 of the state @xmath55 . due to the the fact that @xmath56 = \int_{\mathbb{r}^{r } } p(x_k | x_{k-1 } , w_{k-1 } ) \cdot p(w_{k-1 } ) d w_{k-1 } , \ ] ] the prior pdf @xmath57 derived in identity can be rewritten as @xmath58 p(x_{k-1 } | y_{1:k-1 } ) d x_{k-1 } , \end{aligned}\ ] ] where @xmath59 $ ] represents the expectation with respect to the white noise @xmath31 , and the pdf @xmath60 is @xmath61 with @xmath62 for any @xmath63 and @xmath64 . the definition in can be viewed as a generalization of the dirac delta function in the space @xmath65 , where the mass is located according to the state equation @xmath66 .
note that the estimation of requires an approximation to the expectation @xmath67 $ ] . to this end , we first draw @xmath68 independent samples @xmath69 of the white noise @xmath31 , and define an approximation to @xmath60 as @xmath70 with @xmath71 which is essentially a restriction of @xmath72 in the subset @xmath69
. therefore , the expectation @xmath73 $ ] in can be approximated by @xmath74 & \approx \mathbb{e}_w\left[\pi^m(x_k | x_{k-1 } , w_{k-1})\right ] , \\ & = \sum_{j=1}^m \int_{\mathbb{r}^r } \delta_{w_{k-1}^{j}}(x_k| x_{k-1 } , w_{k-1 } ) p(w_{k-1})dw_{k-1}. \end{aligned}\ ] ] to construct an interpolation of @xmath75 , the next step is to approximate @xmath75 at a point set @xmath76 with @xmath77 . by substituting @xmath78 into - , we have @xmath79 p(x_{k-1 } | y_{1:k-1 } ) d x_{k-1 } + \mathcal{r}_{k | k-1}^i , \label{integ}\ ] ] where @xmath80 p(x_{k-1 } | y_{1:k-1 } ) d x_{k-1}$ ] is the approximation error .
then , by further fixing @xmath81 , the location of the mass of @xmath82 in the space of @xmath83 , denoted by @xmath84 , can be obtained by _
implicitly _ solving the state equation @xmath85 which is the reason we named the approach the implicit filter . now substituting @xmath86 into , and using the same sample set @xmath87 as above to approximate the integral on the right hand side of
, we obtain @xmath88 & = \sum_{j=1}^m \left({\frac}{1}{m}\sum_{j'=1}^m \delta_{w_{k-1}^{j}}(x_k^i| x_{k-1 } , w_{k-1}^{j'})\right)\\ & = { \frac}{1}{m}\sum_{j=1}^m \delta_{w_{k-1}^{j}}\left(x_k^i| x_{k-1 } , w_{k-1}^{j}\right ) , \end{aligned}\ ] ] then replacing @xmath89 $ ] in with , we have @xmath90 p(x_{k-1 } | y_{1:k-1 } ) d x_{k-1 } + \mathcal{r}_{k | k-1}^i\\ & = { \frac}{1}{m}\sum^m_{j=1 } p \left ( x^{i , j}_{k-1 } \big| y_{1:k-1 } \right ) + \mathcal{r}_{k | k-1}^i , \end{aligned}\ ] ] where @xmath91 is the value of @xmath92 at @xmath93 . neglecting the error term @xmath94 in
, we obtain the following iterative numerical scheme for constructing an approximation , denoted by @xmath95 , of the prior pdf @xmath96 , i.e. , @xmath97 in our previous work @xcite , the subsets @xmath98 , for @xmath99 , were defined by a full tensor product mesh , denoted by @xmath100 on a @xmath101-dimensional hyper - cube @xmath102 \times \cdots \times [ a_d , b_d]$ ] , where @xmath103 , is a uniform partition of the interval @xmath104 $ ] with @xmath105 grid points .
it is simple to implement but has several significant disadvantages .
first , at each time step , one needs to approximate the prior pdf @xmath106 at a total of @xmath107 grid points which grows exponentially as the dimension @xmath101 increases .
this is also known as `` the curse of dimensionality '' . on the other hand ,
since the construction of @xmath108 is not informed by the target pdf , the domain @xmath102 \times \cdots \times [ a_d , b_d]$ ] needs to be defined sufficiently large , so as to capture the statistically significant region of the pdf .
this may lead to a great waste of computation effort in the low probability region of @xmath106 .
to alleviate such disadvantages , we propose to develop a distribution - informed meshfree interpolation approach to efficiently approximate the prior pdf .
the central idea of the generation of random points for the state variable is to build a set of points , denoted by @xmath98 , according to the state distribution . to begin with ,
we generate @xmath109 of @xmath110 random samples from the initial pdf @xmath111 of the initial state : @xmath112 if the initial pdf @xmath111 is close to the true state distribution , it s obvious that our random state points are more concentrated near the target state . for @xmath113 ,
we propagate points @xmath114 to @xmath115 through the state equation : @xmath116 where @xmath117 are @xmath110 random samples according to the pdf of @xmath31 . denote @xmath118 and approximate the conditional pdf @xmath119 on @xmath98 with the scheme given by . in this way , the random points in @xmath98 move according to the state model .
as opposed to particle filter methods , which use the number of particles to represent empirical distributions and require a large number of particles to follow the state distribution , in the implicit filter method we provide an approximation of the value of the pdf at each state point .
therefore , much fewer points are needed to describe the state pdf and the random state points are not necessary to accurately follow the state distribution . by incorporating the new data @xmath23 , we update the prior pdf @xmath48 at each grid point @xmath86 , using the bayesian formula , to obtain @xmath120 where @xmath121 is given in , @xmath122 is the normalization factor , and @xmath123 is the approximation error . by neglecting the error term @xmath124 in @xmath125 , we obtain the following iterative numerical scheme for the update stage on @xmath98 , i.e. , @xmath126 where @xmath127 is desired the approximation of the posterior pdf @xmath128 .
next , we use interpolation methods to construct the approximation @xmath129 of @xmath130 from values @xmath131 via @xmath132 where @xmath133 is the set of basis functions .
since the state points in @xmath98 are generated randomly in the meshfree framework , standard polynomial interpolation @xcite is unstable due to the uncontrollable lebesgue constant .
instead , we propose to use the shepard s method @xcite , which is an efficient meshfree interpolation technique , to construct the interpolant @xmath129 .
the basic idea of the shepard s method is to use the weighted average of @xmath134 in the interpolating approximation .
specifically , for a given point @xmath135 , we re - order the points in @xmath136 by the distances to @xmath137 to get a sequence @xmath138 such that @xmath139 where @xmath140 is the euclidean norm in @xmath141 .
then , for a pre - chosen integer @xmath142 we use the first @xmath142 values in @xmath143 to approximate @xmath129 as follows @xmath144 where the weight @xmath145 is defined by @xmath146 note that @xmath147 . from
, we have @xmath148 where @xmath149 is the error of the shepard s interpolation .
we assume that @xmath130 has bounded first order derivative . for each pair
@xmath150 and @xmath151 the approximation error @xmath152 is controlled by the distance @xmath153 and the derivative @xmath154 , where @xmath155 is a point between @xmath137 and @xmath156 .
it is reasonable to assume that in high probability region of the derivative @xmath154 is large .
it s worth pointing out that the random state points generated in this algorithm are concentrated in the high probability region .
thus , if @xmath137 lies in the high probability region , the distance @xmath153 is small , which balances the error brought by the large derivative . on the other hand , if @xmath137 lies in the low probability region , although the distance @xmath153 is relatively large , the approximation error is still small due to the small value of the derivative @xmath154 .
similar to the particle filter method , the above random state points generation suffers from the degeneracy problem for long term simulations , especially for high - dimensional problems .
after several time steps , the probability density tends to concentrate on a few points which dramatically reduces the number of effective sample points in @xmath98 . in this work ,
we propose an occasional resampling procedure to address these problems and rejuvenate the random points cloud . at the time step @xmath157
, the resampling procedure takes place after we obtain @xmath158 , in order to remove the degenerated points in @xmath159 using the information provided by @xmath158 .
specifically , the first step is to develop a degeneracy metric to determine the necessity of doing resampling . to this end , we define the following degenerated subset @xmath160 , @xmath161 where @xmath162 is a user - defined threshold .
we also define @xmath163 to be the index set of @xmath164 .
then , the degeneracy of @xmath159 can be measured by the ratio @xmath165 $ ] , where @xmath166 denotes the number of points in a set .
if the ratio is smaller than a threshold @xmath167 $ ] , then we will skip the resampling step and propagate @xmath159 to get @xmath136 ; otherwise , the set @xmath159 is considered degenerated , and the resampling procedure is needed . in resampling , instead of propagating @xmath168 to @xmath169 , we aim at constructing an intermediate point set , denoted by @xmath170 and propagate @xmath171 through the state model to obtain @xmath136 . according to the definition of @xmath164 in
, we consider the state points in @xmath172 are in the statistically significant region of @xmath173 , so that we first put those points in @xmath171 , i.e. , @xmath174 for the state points in @xmath164 , we replace them by generating new samples from @xmath175 using the importance sampling @xcite , i.e. , @xmath176 as a result , the resampling procedure helps us remove the state points with low probabilities , and makes the state point set @xmath136 concentrated in the high probability region of the posterior pdf @xmath175 at each time step .
finally , we summarize the entire meshfree implicit filter algorithm introduced in [ prediction]-[resampling ] in algorithm 1 below .
p0.95 : _ the meshfree implicit filter algorithm _ + 1.1 [ algorithm2 ] : set the number of samples @xmath68 for estimating @xmath59 $ ] , the number of state points @xmath110 , the resampling thresholds @xmath177 and @xmath178 compute the ratio @xmath179 propagate @xmath159 through the state model to obtain @xmath136 resample and construct the intermediate state set @xmath171 propagate @xmath171 through the state model to obtain @xmath136 : solve @xmath180 using , at each point in @xmath98 : solve @xmath181 using and +
in this section , we present two numerical examples to examine the performance of our meshfree implicit filter method . in example 1
, we use a two dimensional nonlinear filtering problem to show the distributions of the random points @xmath98 . in example 2 , we solve a three dimensional bearing - only tracking problem , which is a six dimensional nonlinear filtering problem . for this higher dimensional problem , we compare the accuracy and efficiency of our meshfree implicit filter method with the extended kalman filter and the particle filter . in this example , we consider the two dimensional noise perturbed tumoral growth model @xcite @xmath182 where @xmath183 is a two dimensional standard brownian motion and @xmath184 .
the state process @xmath185 is a two dimensional vector , @xmath186 is defined as @xmath187 and @xmath188 here , @xmath189 models the gompertzian growth rate of the tumor and @xmath190 gives the degree of vascularization of the tumor which is also called `` angiogenic capacity '' . to approximate the state variables ,
we discretize the dynamic system in time and obtain a discrete state model @xmath191 here , @xmath192 is a two dimensional zero mean gaussian white noise process with covariance @xmath193 , where @xmath194 is the @xmath195 identity matrix and @xmath196 is the time partition stepsize .
the measurement of the state model is given by @xmath197 where @xmath198 is a two dimensional zero mean gaussian white noise process with covariance @xmath199 , @xmath194 is a @xmath195 identity matrix and @xmath200 . in the numerical experiment ,
we use uniform time partition with stepsize @xmath201 and simulate the state process for @xmath202 with initial state @xmath203 and parameters @xmath204 , @xmath205 , @xmath206 . at time step @xmath207 , we initialize the prior pdf @xmath111 by @xmath208 , where @xmath209 and @xmath210 [ scatter_1 ] [ scatter_2 ] + [ scatter_3 ] [ scatter_10 ] + [ scatter_20 ] [ scatter_40 ] + in figure [ scatter_0 ] , we plot @xmath211 random samples generated from the initial pdf @xmath111 , which are our initial random points @xmath212 .
figure [ 2d_grids ] illustrates the behavior of random state points @xmath98 at time steps @xmath213 , respectively . in figure
[ 2d_grids ] , the blue dots in each figure plot the random state points obtained by using the dynamic state points generation method introduced in section [ algorithm ] and the red cross in each figure gives the true state @xmath214 at the corresponding time step . from the figures we can see that all the points are moving according to the state model and
are concentrated around the true state . to present the accuracy of the algorithm
, we show the simulation of the tumoral growth states in figure [ 2d_simulation ] .
the black curves are the true @xmath215 and @xmath216 coordinate values of the tumoral growth states , respectively .
the blue curves show the simulated states obtained by using the meshfree implicit filter method .
[ 2d_simulation_x1 ] [ 2d_simulation_x2 ] in this example , we study a six dimensional target tracking problem . in figure
[ model_6d ] , the target , denoted by the red line , moves in the three dimensional space and two platforms on the ground , denoted by pentagons , take angular observations of the moving target .
the state process @xmath217 is described by the following dynamic model @xmath218 where @xmath219 describes the position of the moving target which is controlled by parameters @xmath220 .
the system noise @xmath221 is a zero mean gaussian white noise process with covariance @xmath222 , @xmath194 is the @xmath223 identity matrix and @xmath196 is a given time period , @xmath224 is a constant vector and @xmath1 is given by @xmath225 the measurements @xmath226 of the state process from the two locations are given by @xmath227 where @xmath198 is a 4 dimensional zero mean gaussian white noise process with covariance @xmath199 , @xmath194 is a @xmath228 identity matrix , @xmath229 , @xmath230 and @xmath231 are locations of two observers .
[ com_6d_x1 ] [ com_6d_x2 ] + [ com_6d_x3 ] [ com_6d_x4 ] + [ com_6d_x5 ] [ com_6d_x6 ] + we choose @xmath232 , @xmath233 , @xmath234 .
also , we assume that platforms are located at @xmath235 , @xmath236 and the initial sate is given by a gaussian @xmath237 where @xmath238 and @xmath239 the target will be observed over the time period @xmath240 . in the numerical experiments ,
we compare the performance of our meshfree implicit filter with the extended kalman filter and the particle filter .
in particular , we compare the estimated mean values of the states process along each dimension in figure [ 6d_state_comparison ] . in the particle filter method , we choose @xmath241 particles . in the meshfree implicit filter method
, we choose the number of state points to be @xmath242 and the number of random samples in the implicit filter monte carlo simulation to be @xmath243 .
the black curves in figure [ 6d_state_comparison ] show the real states process along each direction , the green curves give the estimated means obtained by the extended kalman filter method , the red curves give the estimated means obtained by the particle filter method , and the blue curves give the estimated means obtained by the meshfree implicit filter .
we also plot the @xmath244 error @xmath245 corresponding to all three methods in figure [ 6d_l2 ] .
as we can see from figure [ 6d_state_comparison ] and [ 6d_l2 ] , the implicit filter and the particle filter are much more accurate than the extended kalman filter and the implicit filter is the most accurate approximation in this experiment . to further compare the efficiency between the meshfree implicit filter and the particle filter , we repeat the above experiment over @xmath246 realizations and show the average cpu time and the corresponding global root mean square error @xmath247 defined by @xmath248 where @xmath249 is the @xmath244 error of the @xmath250-realization at time step @xmath14 . in table [ efficiency ] , we can see that with @xmath241 particles , the cpu time of the particle filter method is comparable to that of the implicit filter with @xmath251 random state points , but the global rmse of the particle filter is more than doubled the rmse of the implicit filter . with @xmath252 particles , the particle filter method achieves an accuracy comparable to the implicit filter , but at a significantly higher cost .
.example 2 : efficiency comparison [ cols="^,^,^",options="header " , ]
in this work , we proposed an efficient meshfree implicit filter algorithm by evaluating the conditional pdf on meshfree points in the state space .
these meshfree points are chosen adaptively according to the system state evolution .
we also apply shepard s method as the meshfree interpolation method to compute interplants with random state points . in order to address the degeneracy of the random points , we use importance sampling method to construct a resample step .
numerical examples demonstrate the effectiveness and efficiency of our algorithm . in the future
, we plan to perform a rigorous numerical analysis for the meshfree implicit filter algorithm . | in this paper , we propose a meshfree approximation method for the implicit filter developed in @xcite , which is a novel numerical algorithm for nonlinear filtering problems .
the implicit filter approximates conditional distributions in the optimal filter over a deterministic state space grid and is developed from samples of the current state obtained by solving the state equation implicitly .
the purpose of the meshfree approximation is to improve the efficiency of the implicit filter in moderately high - dimensional problems .
the construction of the algorithm includes generation of random state space points and a meshfree interpolation method .
numerical experiments show the effectiveness and efficiency of our algorithm .
nonlinear filtering , implicit algorithm , meshfree approximation , shepard s method |
the set - up we consider is the following .
let @xmath0 be a symplectic manifold ( we assume our manifolds to be paracompact ) , @xmath1 and @xmath2 two poisson manifolds , and @xmath3 and @xmath4 two surjective submersive poisson maps . in remark [ rmk : subversive ]
we describe the effect of replacing the condition submersion " by open " .
let @xmath5 denote the algebra of the pull - backs of smooth functions on @xmath6 , that is , @xmath7 since @xmath8 and @xmath9 are poisson it follows that @xmath10 and @xmath11 are poisson subalgebras of @xmath12 .
if @xmath13 is open , we write @xmath14 for the algebra @xmath15 this is a poisson subalgebra of @xmath16 . for a subset @xmath17
we write @xmath18 for the centralizer of @xmath19 with respect to the poisson structure on @xmath20 , that is , @xmath21 where @xmath22 is the restriction of the poisson bracket to @xmath23
. note that in the following two definitions @xmath8 and @xmath9 are assumed to be poisson maps but not necessarily submersions .
[ regular dual pairs definitions ] consider the diagram ( 9,5.5 ) ( 4.9,5)(0,0)@xmath0 ( 1,0.5)(0,0)@xmath1 ( 9,0.5)(0,0)@xmath2 ( 5,4.5)(1,-1)3 ( 4.5,4.5)(-1,-1)3 ( 2,3.5)(0,0)@xmath24 ( 7.3,3.5)(0,0)@xmath9 * the diagram forms a * _ _ * howe ( h ) dual pair if the poisson subalgebras @xmath10 and @xmath11 centralize each other : @xmath25 * the diagram forms a * _ _ * lie - weinstein ( lw ) dual pair when @xmath26 and @xmath27 are symplectically orthogonal distributions .
that is , for each @xmath28 , @xmath29 in each case , the dual pair is * _ _ * regular when the maps @xmath30 are assumed to be surjective submersions ; otherwise they are * _ _ * singular dual pairs .
this notion of singular dual pair is less general than that in @xcite .
notice that for a regular lie - weinstein dual pair , the dimensions of @xmath31 and @xmath32 sum to the dimension of @xmath33 .
actually , we note that if the manifold @xmath33 is lindelf or paracompact as a topological space then the lie - weinstein condition can not hold unless the dual pair is regular .
this is because the lw condition implies that the two maps @xmath34 and @xmath35 are of complementary rank ( the ranks sum to @xmath36 ) and , by the lower semicontinuity of the rank of a smooth map , the maps must both be of constant rank
. since they are surjective they must be submersions ( by sard s theorem ) .
we emphasize that the definition of a lie - weinstein dual pair is local , while that of a howe dual pair is global .
however , the latter definition can be localized as follows .
the diagram above forms a * _ _ * local howe ( lh ) dual pair if for each @xmath37 and each neighbourhood @xmath38 of @xmath39 there is a neighbourhood @xmath23 of @xmath39 with @xmath40 such that the algebras @xmath41 and @xmath42 centralize each other in @xmath16 .
if in addition , @xmath8 and @xmath9 are surjective submersions , then the local howe dual pair is said to be * _ _ * regular ; otherwise it is * _ _ * singular .
the notion of howe dual pair has its origins in the study of group representations arising in quantum mechanics ( see for instance @xcite , and references therein ) and it appears for the first time in the context of poisson geometry in @xcite . the definition of lie - weinstein dual pair can be traced back to @xcite and , in its modern formulation , is due to @xcite .
examples of dual pairs arising in classical mechanics can be found in @xcite , and references therein .
further details on dual pairs can also be found in @xcite .
the relationships between the three notions of regular dual pair can be summed up in the following two results .
[ prop : local ] the two local notions of regular dual pair , that is , lie - weinstein and local howe , are equivalent . in remark [ rmk
: subversive ] we provide an example showing that this result no longer holds if the regularity is dropped .
[ thm : local - global ] 1 .
if a regular howe dual pair is such that the poisson manifolds @xmath31 and @xmath32 are of complementary dimension , that is , @xmath43 , then it forms a regular local howe dual pair .
if a regular local howe dual pair is such that the fibers of @xmath8 and @xmath9 are connected then it is a regular howe dual pair . before giving the proofs ( in section [ sec : proofs ] ) ,
we give two examples showing the necessity of the hypotheses .
[ ex : lh not h ] this example ( suggested to us by andrea giacobbe ) shows that the hypothesis of connected fibers in the passage from local to global in the theorem above is necessary .
let @xmath44 be the 2-torus considered as a symplectic manifold with the area form @xmath45 .
consider the diagram @xmath46 with @xmath47 .
the fibers of @xmath35 have two connected components .
it is easy to see that this forms a lie - weinstein dual pair ( and hence a local howe dual pair ) but not a howe dual pair .
indeed , the function @xmath48 belongs to @xmath49 but not to @xmath11 . in the example below and in subsequent proofs ,
we use the following notation . on the symplectic manifold @xmath50
, we write the poisson tensor as @xmath51 . if @xmath52 then the hamiltonian vector field @xmath53 is defined by @xmath54 .
the poisson tensor , defined by @xmath55 = \langle \mathbf{d}g , x_h \rangle = \omega(x_g,\,x_h)$ ] , induces the vector bundle morphism over the identity @xmath56 given by @xmath57 .
[ ex : howe not l - w ] this example shows that without the dimension hypothesis , a regular howe dual pair need not be locally howe ( nor lie - weinstein ) .
let @xmath58 and @xmath59 be linearly independent over @xmath60 .
define on @xmath33 the symplectic structure @xmath61 whose poisson tensor @xmath62 is given by @xmath63 let @xmath64 be the projection onto the @xmath65 factor .
the hamiltonian vector field associated to the function @xmath66 is @xmath67 then the diagram @xmath68 is a regular howe pair but clearly not a lie - weinstein dual pair and hence not a regular local howe dual pair . in order to see that it is a regular howe dual pair let @xmath69 .
the trajectories of the vector field @xmath70 on @xmath33 are irrational windings which are dense in the fibers of @xmath71 . since @xmath72 is invariant under this hamiltonian flow it must be constant on the fibers of @xmath66 and hence @xmath73 , as required .
[ rmk : subversive ] * 1 . *
if one merely assumes the maps @xmath30 to be open rather than submersions , then one can still pass from local howe to global , provided of course the fibers are connected ( see the proof in section [ sec : proofs ] ) .
* 2 . * as was already observed , if the lie - weinstein condition is satisfied then the two poisson maps @xmath30 are of constant rank .
any example of a local howe dual pair which is not lie - weinstein must of course be singular ( the maps can not be submersions ) and not of constant rank .
a simple example is obtained by putting @xmath74 with coordinates @xmath75 and its usual symplectic form @xmath76 .
let @xmath77 be given by @xmath78 ( inner product ) and @xmath79 be given by @xmath80 ( outer product ! ) .
in coordinates , @xmath81 the fibers of both @xmath8 and @xmath9 are connected and the image @xmath82 is a cone in @xmath83 .
we claim that @xmath84 forms a singular howe dual pair .
it is clearly not lie - weinstein at the origin , as @xmath85 and @xmath86 both vanish at that point . that it is a howe pair follows from the paper of karshon and lerman @xcite , since @xmath8 is the orbit map for the @xmath87 action on @xmath88 and @xmath9 its momentum map .
note that in @xcite it is shown that momentum maps of representations are @xmath89-open , although we do not know whether @xmath89-openness is sufficient to be able to pass from local to global in the singular case .
define @xmath90 ( for @xmath91 ) , which are two subbundles of @xmath92 .
since the maps @xmath30 are submersions , we have for each @xmath37 @xmath93 furthermore , since @xmath94 it follows that @xmath95 we establish the equivalence of the two local notions of regular dual pair .
recall that @xmath8 and @xmath9 are assumed to be surjective submersions .
[ [ lh - rightarrow - lw ] ] lh @xmath96 lw : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we wish to show that @xmath97 , which we do by double inclusion .
first we show @xmath98 .
let @xmath99 be coordinates on @xmath32 .
then by the local howe condition , @xmath100 implies @xmath101 so that @xmath102 , by ( [ eq : komega ] ) .
for the converse inclusion , @xmath103 , let @xmath104 and @xmath105 . then there is a function @xmath106 such that @xmath107 .
then for @xmath108 we have @xmath109 , so that @xmath110 .
since @xmath111 is a subbundle of @xmath112 and hence locally trivial , we can choose @xmath106 on a neighbourhood @xmath38 of @xmath39 so that at each @xmath113 , @xmath114
. thus @xmath115 by the local howe condition , this implies there is a sub - neighbourhood @xmath116 of @xmath39 such that @xmath117 .
consequently , @xmath118 is a section of @xmath119 over @xmath120 and , since @xmath107 , it follows that @xmath121 .
[ [ lw - rightarrow - lh ] ] lw @xmath96 lh : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we choose @xmath23 such that both submersions @xmath122 have connected fibers ( this can be done using lemma [ lem : connected ] , by choosing any function @xmath123 on @xmath33 with a non - degenerate local minimum at @xmath39 ) .
we prove the equality @xmath124 , again by double inclusion .
let @xmath125 .
since @xmath126 , for any @xmath127 and , by hypothesis , @xmath128 , we have @xmath129 where the last equality follows from ( [ eq : annihilator ] ) .
consequently , for an arbitrary @xmath130 , we conclude @xmath131 which implies that @xmath132 .
conversely , let @xmath133 for some @xmath23 . in order to prove that @xmath134 we start by showing that it is locally constant on the fibers of @xmath35 . indeed , since @xmath135 is a surjective submersion and the diagram @xmath136 forms a lie - weinstein dual pair , for any @xmath28 we have @xmath137 this equality , together with ( [ eq : annihilator ] ) , guarantees that any vector @xmath138 tangent at @xmath39 to the fiber @xmath139 can be written as @xmath140 , for some @xmath130 .
hence , @xmath141 since both @xmath37 and @xmath142 are arbitrary and , by the choice of @xmath23 , the fibers of @xmath143 are connected , this equality guarantees that the function @xmath144 is constant on the fibers of @xmath135 .
this implies that there exists a unique function @xmath145 that satisfies the equality @xmath146 . since @xmath106 is smooth and @xmath147 a submersion it follows that @xmath148 is smooth and hence @xmath125 . the equality @xmath149 is proved analogously .
@xmath150 [ [ global - rightarrow - local ] ] global @xmath96 local : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let @xmath28 be an arbitrary point . we will now show that the hypotheses in the statement imply that @xmath151 . the inclusion @xmath152 follows from the global howe hypothesis and the equalities ( [ eq : komega ] ) and ( [ eq : annihilator ] ) . indeed , @xmath153 the converse inclusion follows immediately from the hypothesis on the dimensions and the submersiveness of @xmath8 and @xmath135 .
indeed , @xmath154 and hence @xmath155 , as required .
[ [ local - rightarrow - global ] ] local @xmath96 global : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + using the symmetry of the statement with respect to the exchange of @xmath156 with @xmath35 it suffices to show that , for instance , @xmath157 .
notice that the proof below only requires the poisson maps @xmath30 to be open onto @xmath6 , rather than submersions , and that the @xmath6 are not even required to be manifolds ( in which case the proof would hold with an appropriate algebraic definition of smooth function on @xmath158 ) .
first , it is easy to show that given @xmath159 and @xmath160 then @xmath161 .
indeed , let @xmath37 and let @xmath23 be a neighbourhood of @xmath39 on which the local howe condition holds .
then , @xmath162 and similarly @xmath163 and hence , @xmath164 where @xmath165 is the restriction of the poisson bracket to @xmath23 .
since @xmath37 , @xmath166 , and @xmath167 are arbitrary , it follows that @xmath168 second , let @xmath169 ; one needs to show that @xmath170 . for each @xmath37
there is a neighbourhood @xmath171 on which the local howe hypothesis is valid .
this provides a cover of @xmath33 , from which we can extract a locally finite subcover @xmath172 . for each @xmath173
, we claim that @xmath174 .
it then follows by hypothesis that @xmath175 , which allows us to write @xmath176 for some @xmath177 .
our second claim is that there is a function @xmath178 such that @xmath179 .
the result then follows as @xmath180 .
we now establish the two claims . the first claim is proved by contradiction .
suppose that there exists a function @xmath181 and @xmath182 such that @xmath183 .
let @xmath184 be an open neighbourhood of @xmath185 such that @xmath186 .
then there is an extension @xmath187 of @xmath188 .
since @xmath189 it follows that @xmath190 contradicting . as to the second claim , notice first that ( [ pull back gives us another ] ) implies that @xmath72 is locally constant along the fibers of @xmath35 .
since by hypothesis these fibers are connected , @xmath72 is constant on the fibers of @xmath35 and @xmath191 is therefore well defined .
moreover , @xmath191 coincides with @xmath192 on the open sets of the form @xmath193 and so is smooth .
[ lem : connected ] let @xmath23 be a manifold and @xmath123 be a smooth real - valued function with a non - degenerate local minimum at @xmath194 , and suppose @xmath195 .
let @xmath196 be a submersion in a neighbourhood of @xmath197 .
then for @xmath198 sufficiently small the level sets of @xmath66 restricted to @xmath199 are diffeomorphic to an open ball , where @xmath199 is the connected component of @xmath200 containing @xmath197 .
write @xmath201 .
we wish to show that when it is nonempty , @xmath202 is diffeomorphic to an open ball . first , choose coordinates near @xmath197 and @xmath203 such that @xmath204 . since the restriction of @xmath123 to @xmath205 has a non - degenerate minimum at @xmath206 , by the splitting lemma ( or morse lemma with parameters , see e.g. @xcite ) there is a neighbourhood @xmath207 of @xmath197 on which one can change coordinates by @xmath208 in such a way that @xmath209 , where @xmath210 is a positive definite quadratic form and @xmath72 is a smooth function .
choose @xmath211 sufficiently small so that @xmath212 is contained in this neighbourhood .
in these coordinates , for @xmath213 and for each @xmath214 , @xmath215 it is clear that for each @xmath216 this set is either empty or diffeomorphic to an open ball .
there are a number of theorems in the literature which are stated with the hypothesis that a given set - up is a howe dual pair , while the proof uses the lie - weinstein property .
the most famous of these are probably the symplectic leaves correspondence theorem @xcite and weinstein s theorem on transverse poisson structures ( * ? ? ?
* theorem 8.1 ) . here
we give an example showing that these theorems fail if one does not assume a local ( lie - weinstein ) hypothesis .
of course , by the theorem above the hypothesis that @xmath31 and @xmath32 are of complementary dimension is also sufficient .
let @xmath217 with coordinates @xmath218 , where @xmath219 , @xmath220 , and @xmath221 .
let @xmath222 and @xmath223 be given by @xmath224 we endow @xmath33 with a poisson structure whose poisson tensor can be written in block form as @xmath225.\ ] ] we assume that the entries of @xmath226 are independent over @xmath60 , excluding those forced to vanish .
an argument along similar lines to that for example [ ex : howe not l - w ] shows that @xmath227 forms a regular howe dual pair .
however , since the dimensions of @xmath228 and @xmath229 differ by an odd number , it is clear that they can not have the property of having anti - isomorphic transverse poisson structures , up to a product with a symplectic factor .
indeed , the poisson structure on @xmath32 is of course trivial , while that on @xmath31 is of rank 2 .
this example also shows that in the absence of the local lie - weinstein hypothesis , the correspondence between the symplectic leaves of two poisson manifolds in duality may fail . in order to see this take @xmath230
this is a symplectic leaf in @xmath65 , and if the symplectic leaves correspondence theorem were valid then @xmath231 would be a symplectic leaf in @xmath232 .
this is obviously impossible as @xmath233 is of odd dimension .
* acknowledgments .
* we thank ana cannas da silva , eugene lerman , rui loja fernandes , and alan weinstein for patiently answering our questions and for several illuminating discussions .
this research was partially supported by the european commission and the swiss federal government through funding for the research training network _
mechanics and symmetry in europe _ ( masie ) . the support of the swiss national science foundation is also acknowledged .
marsden , j.e . and weinstein , a. [ 1983 ] coadjoint orbits , vortices , and clebsch variables for incompressible fluids .
_ proc . of the conference on order in chaos ,
_ los alamos , 1982 .
( a. scott ed . ) .
_ physica d _ , * 7*(13 ) , 305323 . | in this note we clarify the relationship between the local and global definitions of dual pairs in poisson geometry .
it turns out that these are not equivalent . for the passage from local to global
one needs a connected fiber hypothesis ( this is well known ) , while the converse requires a dimension condition ( which appears not to be known ) .
we also provide examples illustrating the necessity of the extra conditions . |
in a @xmath8cdm universe , the inner regions of massive galaxies like the milky way gain a large fraction of their mass through tidal disruption and accretion of a large number of low - mass fragments ( e.g van den bosch , tormen & giocoli 2004 ) .
the fossil records of these merging processes may be observable nowadays in form of long tidal streams or large - scale stellar substructures around the parent spiral galaxies .
numerical galaxy formation simulations show that dynamical friction brings sub - structures from outer halo regions to the neighbourhood of the parent galaxies disks .
if haloes are flattened ( oblate ) and their axi - symmetry plane is that of the disk , orbits of non - polar satellite galaxies tend to become co - planar and circularise with time ( pearrubia , kroupa & just 2004 ) . in this scenario , tidal debris of several disrupted satellite galaxies
might have contributed to the formation of the stellar disk if they followed nearly circular orbits with a low orbital inclination at late times of their evolution ( navarro 2004 and references therein ) .
the milky way is an important laboratory in which the predictions of this cosmological scenario can be tested . in the last decade ,
large - scale surveys have proved the existence of tidal streams ( sgr : ibata et al .
1994 ; monoceros : newberg et al . 2002 ) in our galaxy , providing strong observational evidence that disruption of dwarf satellites contributes to the assembly of some components of our galaxy .
these tidal streams offer an unique opportunity to study accretion events in considerable detail using the chemical , kinematic and spatial distribution of tidal stream s stars , which can be directly compared against predictions of n - body simulations of merging events ( e.g law , johnston & majewski 2004 ) .
recently , the sloan digitized sky survey team reported the discovery of a coherent ring - like structure at low galactic latitude spanning about 100 degrees in the sky ( newberg et al .
2002 ; yanny et al .
follow - up observations ( ibata et al . 2003 ) found that this structure of low metallicity stars surrounds the galactic disk at galactocentric distances from @xmath9 15 kpc to @xmath10 kpc .
different groups proposed two different scenarios to explain the nature of this feature in the outer galactic disk ( see helmi et al .
2003 ) : i ) a new tidal stream ( yanny et al .
2003 ) ; and ii ) a stellar ring resulting from perturbations in the disk similar to the remnants of ancient warps ( ibata et al .
2003 ) . tracing this structure with 2mass m giant stars , rocha - pinto et al .
( 2003 ) concluded that its structural characteristics are consistent with the interpretation of this structure as the fossil of a merging dwarf galaxy , similar to the sgr dwarf galaxy ( ibata et al .
1994 ) , in the galactic plane .
interestingly , frinchaboy et al . ( 2004 ) and crane et al . (
2003 ) also suggested some nearby galactic open and globular clusters with coordinated heliocentric radial velocities , indicating a possible common origin with the tidal stream . unlike the sagittarius dwarf galaxy
, the monoceros stream has been detected prior to locating the main body of the parent galaxy .
the location is still controversial . from the analysis of the 2mass catalogue , martin et al .
( 2004a ) reported a strong elliptical - shaped stellar over - density in the constellation of canis major , which is coincident in position and radial velocity with a small group of 4 globular clusters ( see sec .
[ sec : cluster ] ) . using a fairly simple model , they identify the cma over - density as the main body of the progenitor dwarf galaxy of the monoceros stream ( named canis major dwarf ) .
bellazzini et al .
( 2004 ) presented colour - magnitude diagrams in the surroundings of the cma location , concluding that the system is situated at 8@xmath11 1 kpc from the sun and it is composed by a metal - rich , intermediate - age population .
momany et al .
( 2004 ) , however , comparing proper motions and radial velocities of canis major m - giant stars , obtain that the overdensity in this region mimics the thick disk kinematics . moreover , they find that the star counts in that region are those expected in the standard milky way model if taking into account the warp and flare of the disk .
however , a deep colour - magnitude diagram of the center of the canis majoris over - density by martnez - delgado et al .
( 2004 ) shows a very well - defined main - sequence consistent with a limited extent in distance , confirming that this stellar population is associated with a distinct , possibly still bound stellar system with properties ( surface brightness , absolute magnitude , stellar content ) compatible with those of local group dwarf spheroidal galaxies . in base of these last results
, the canis major overdensity will be referred to the cma dwarf galaxy in this work . in this paper
we present the results of an extensive search of possible disruption scenarios of satellite galaxies to explore : a ) how many of the over - density signatures can be attributed to a single stream and b ) to constrain its progenitor orbit .
with an objective criterion we select those orbits that reproduce the spatial and kinematical distribution of candidate debris . beyond exploring the possible progenitor location ,
the results will : ( i ) help to carry out new surveys in different regions of the sky and ( ii ) constrain the thick disk formation history through the distribution evolution of stripped stars .
this contribution has been organised as follows : in section [ sec : obs ] we compile the observational data available so far .
section [ sec : tools ] outlines the galaxy and satellite models that we use in our investigation .
we also describe our semi - analytic orbit algorithm and the results of the orbital parameter survey .
we obtain the distribution of debris via the n - body algorithm sketched in section [ sec : nbody ] . in section [ sec : res ] we compare the best - fit orbit with the observational data , whereas in section [ sec : dis ] we comment different implications of our results .
for the present work , we have attempted a comprehensive compilation of observational constraints on the geometry and kinematics of the monoceros stream . given the great deal of activity in detecting tidal streams in general , given the diversity of available data sets , and given in particular the great interest in the monoceros stream spawned by the original discovery ,
the constraints are both numerous and inhomogeneous .
typically , detections of overdensities have resulted from photometry , finding an axcess of main - sequence stars or m - giants at a given apparent magnitude .
depending on the tracer population , this approach provides distance estimates of differing quality .
radial velocities are available for a sub - set of stars within a sub - set of directions with detected overdensities . in principle , also the chemical composition of stars can serve as a constraint , as it can be expected to vary continuously along the stream , presumably with decreasing metallicity in the most loosely bound ( and hence first lost ) material .
finally , proper motions ( which typically can be measured to @xmath12 3 - 4 mas / year ) can provide some rough contraints ; their precision in physical units is , however , only @xmath13 , where @xmath14 is the distance from the sun and 4.74 is the factor that coverts [ kpc mas / yr ] into [ km / s ] .
we have compiled both published constraints and those available to us , but still in the process of publication .
a few constraints were derived specifically for the present paper .
we have summarized the constraints , their nature ( photometric , kinematic , etc .. ) and their sources in table [ tab : obser ] . in following plots
we use different symbols in order to distinguish between different data sources . in order to simplify our figures
we have made an exception for those data with available radial velocities , which we have been plotted everywhere with _
full triangles_. @xmath15}$ ] & @xmath16}$ ] & @xmath17}$ ] & symbol + yanny et al .
( 2002 ) ; & @xmath18 $ ] , @xmath19 $ ] & cmd & yes & no & 4 & full triangle + ibata et al .
( 2003 ) & @xmath20 $ ] , @xmath21 $ ] & cmd@xmath22}$ ] & no & no & 14 & open triangle + crane et al .
( 2003 ) & @xmath23 $ ] , @xmath24 $ ] & m - giant & yes & yes@xmath25}$ ] & 58 & full triangle + newberg et al .
( 2004 ) & @xmath26 $ ] , @xmath27 $ ] & cmd@xmath22}$ ] & no & no & 22 & open box + rocha - pinto et al .
( 2004 ) & @xmath28 $ ] , @xmath29 $ ] & m - giant@xmath30}$ ] & yes & no & 31 & full box + [ tab : obser ] additionaly , we have proper motions of confirmed stream star members selected from the radial velocity compilation by crane et al .
proper motions are determined by combining recalibrated usno - b1.0 positions ( monet et al .
2003 ) with those from the sloan digital sky survey of the sky .
drift - scan imaging in the five sdss bandpasses ( @xmath31 ) ( fukugita et al . 1996 , gunn et al . 1998 , hogg et al . 2001 ) is processed through data reduction pipelines to measure photometric and astrometric properties ( stoughton et al .
2002 , smith et al .
2002 , pier et al . 2003 , abazajian et al . 2004 ) and to identify targets for spectroscopic followup . ]
catalog , as detailed in munn et al .
( 2004 ) . proper motion of our target stars are given in table [ tab : obs ] .
it is necessary to clarify that the measurements of proper motions have not been included in the orbital fit and will be used only to discriminate the sense of motion of the progenitor s orbit .
* @xmath32 * & * @xmath33 * + ( deg . ) & ( deg . ) & ( kpc ) & ( km / s ) & ( mas / yr ) & ( mas / yr ) + 186.382 & 23.910 & 11.2 & @xmath34 & 2.6 & -1.9 + 189.316 & 23.251 & 12.6 & @xmath35 & 0.6 & -4.4 + 186.894 & 24.181 & 10.0 & @xmath36 & -5.5 & -13.9 + 189.741 & 23.345 & 12.7 & @xmath37 & 6.0 & -1.0 + 198.778 & 25.063 & 11.9 & @xmath38 & -1.2 & 2.6 + 224.108 & 21.344 & 10.9 & @xmath39 & 1.0 & -6.3 + 178.371 & 36.786 & 10.4 & @xmath40 & 6.1 & 1.5 + 186.992 & 38.816 & 12.3 & @xmath41 & 0.4 & 3.0 + 221.989 & 29.900 & 12.3 & @xmath42 & -3.3 & -6.2 + 223.126 & 32.262 & 11.8 & @xmath43 & -3.5 & 1.9 + [ tab : obs ] in addition , we are faced with the problem of which constraints ( on which candidate sections of the monoceros stream ) to include in our modelling . guided by the goal of finding the largest number of stream portions , that may have arisen from a single disruption event , we have used an iterative procedure . starting with the original stream detections ( newberg et al .
2002.,yanny et al . 2003 ) and the comprehensive kinematic survey of m - stars ( crane et al .
2003 ) spanning 100@xmath44 in the sky .
initial modelling of these constraints made it clear , that other , independently found , overdensities , most likely are also part of the stream .
in particular , the tri / and stream ( majewski et al .
2004 ) , a more distant metal - poor stellar stream , showed a location in the sky and a radial velocity curve fairly similar to the predictions of our first - iteration model .
therefore , the available data ( including radial velocities from rocha - pinto et al . 2004 ) on the tri / and stream were then included as inputs in a second iteration of our survey of the best - candidate orbit to better constrain its properties and reduce the number of possible scenarios .
finally , we explicitly test whether the canis majority dwarf is likely part of of the stream , presumably the parent of the tidal debris .
we do this by omitting cma in the dirst modelling , and then comparing its orbit with that of our model for a possible monoceros stream progenitor .
in this section we outline how we perform the orbital analysis of dwarf satellites and the method used to fit numerical orbits to the observational data .
+ we first describe a milky way model that matches the mass distribution of our galaxy .
subsequently , we discuss the parameter space that we must cover in order to determine the orbital characteristics of a possible monoceros stream progenitor .
lastly , we present the method used to constrain the orbital properties of a possible monoceros tidal stream progenitor , which can be divided in two steps : ( i ) satellite orbits are calculated using a well - tested semi - analytic algorithm in order to perform a survey of our large parameter space .
subsequently , the orbit collection has been compared against the observational sample to determine the parameter sets that provide the best fits .
( ii ) once the orbital constraints are known , we carry out n - body simulations to analyze the spatial and kinematical distribution of debris .
our dynamical model for the milky way follows hernquist ( 1993 ) , where : ( i ) the disk is exponential in the radial direction and isothermal in the vertical direction ; ( ii ) the bulge is described by a hernquist model ( hernquist 1990 ) with spherical symmetry ; ( iii ) the halo follows a flattened , non - singular isothermal profile with given core and cut - off radius .
+ for the density distributions of the disk we take @xmath45 where @xmath46 is the disk mass , @xmath47 kpc is the vertical thickness , and @xmath48 kpc is the exponential scale length in the radial direction .
the mass profile decays exponentially with r and is composed of isothermal sheets along the vertical direction .
velocities are assumed to have a gaussian distribution .
for the bulge we adopt the spherical hernquist profile ( hernquist 1990 ) , @xmath49 where @xmath50 is the bulge mass and @xmath51 kpc is the spherical scale length .
this analytical profile fits the de vaucouleurs law ( de vaucouleurs 1948 ) .
the velocity field is constructed from the jeans equations by assuming isotropic gaussian velocity distributions at each radial distance ( hernquist 1993 ) .
we use a non - singular isothermal profile for the dark matter halo ( dmh ) , @xmath52 where @xmath53 in cylindrical coordinates .
@xmath54 is the halo density flattening , @xmath55 the dmh mass , @xmath56 kpc the cut - off radius , @xmath57 kpc the core radius , and @xmath58\}^{-1 } = \\
\nonumber 1 + \sqrt{\pi}\beta + ( \pi -2 ) \beta^2 + o(\beta^3 ) \end{aligned}\ ] ] where @xmath59 in our calculations . to construct the flattened ( oblate ) dmhs ,
a non - homologous transformation is applied to ( [ eqn : rho_h ] ) to achieve the desired axis ratio @xmath54 while preserving the central density . in order to minimise computational time
when constructing flattened dmhs with embedded bulges and disks , we apply a highly efficient technique using multipole potential expansions to tailor the local velocity ellipsoid to the required morphology ( boily , kroupa & pearrubia 2001 ) .
the algorithm to add together individual components in a single galaxy is adapted from hernquist ( 1993 ) .
this code scales linearly with particle number and hence we can construct flattened dmhs with @xmath60 particles in a short computational time .
the specific parameters are chosen to reproduce the observed rotational curve of the milky way .
the only free parameter of the galaxy model that we explore in our calculations is the halo axis - ratio ( @xmath54 ) .
the dwarf satellite model follows a king profile ( king 1966 ) with central potential @xmath61 and concentration @xmath62 , where @xmath63 are the tidal and king radii , respectively .
since we are assuming that the tidal debris represent multiple wraps ( and thus , multiple perigalacticon passages ) of the tidal stream originating from one disrupting satellite , we restrict the tidal radius of the progenitor dwarf galaxy in each model with the following criteria : ( i ) the tidal radius must be small enough to prevent galaxy disruption by the first peri - galacticon passage and ( ii ) the tidal radius must be large enough to induce a progressive mass loss that leads to the formation of a tidal stream .
we have selected the satellite tidal radius to match the jacobi limit ( section 7.3 of binney & tremaine 1986 ) at the first peri - galacticon , which is a simple method to achieve slow mass loss along the orbit . given the milky way and satellite mass profiles ,
three aspects determine the subsequent orbit evolution : ( i ) the flattening of the milky way halo @xmath54 ; ( ii ) the mass of the satellite , determining the degree of dynamical friction and ( iii ) the initial orbital geometry .
the corresponding free parameters are outlined in table [ tab : param ] .
we have performed 151,200 simulations by means of our semi - analytic code ( see below ) in order to fit the observational data .
each orbit is repeated with 12 different azimuthal angles .
this turns out to be necessary since dynamical friction prevents an orbit from filling up its phase space surface which , therefore , `` breaks '' the axi - symmetry of the problem . due to the high efficiency of the semi - analytic code , the resolution achieved in the parameter space is constrained by the limted number of available observational data rather than by cpu limitations . more densely - sampled parameter surveys would not provide stronger constraints , owing to the large degeneracy seen already in this sparse sampling of parameters space ( see section [ sec : res_sa ] ) .
@xmath64 ( kpc)&@xmath65 & @xmath66 ( deg . ) & @xmath67 ( deg . ) & @xmath68 & + range & @xmath69 $ ] & [ 17.5,80.5 ] & [ 0.,0.7 ] & [ 5.0,45.0 ] & [ 0.0,360.0 ] & [ 0.5,1 ] & + & & & & [ 135.0,175.0 ] & & & + precision & @xmath70 & @xmath71 & @xmath72 & @xmath73 & @xmath74 & @xmath72 & + # values & 3 & 10 & 7 & 10 & 12 & 6 & + & @xmath75 & @xmath64 ( kpc ) & @xmath65 & @xmath66 ( deg . )
& @xmath67 ( deg . ) & @xmath68 & @xmath76 ( kpc ) + prograde & & & & & & & + _ pro1 _ & @xmath77 & 22.8 & 0.1 & 25.0 & 300.0 & 0.6 & 4.4 + _ pro2 _ & @xmath77 & 22.8 & 0.1 & 25.0 & 150.0 & 0.8 & 4.6 + _ pro3 _ & @xmath77 & 22.8 & 0.1 & 25.0 & 120.0 & 0.7 & 4.7 + retrograde & & & & & & & + _ ret1 _ & @xmath78 & 22.8 & 0.5 & 165.0 & 180.0 & 0.6 & 5.6 + _
ret2 _ & @xmath78 & 22.8 & 0.5 & 155.0 & 210.0 & 0.6 & 5.7 + _ ret3 _ & @xmath78 & 22.8 & 0.5 & 165.0 & 210.0 & 0.5 & 5.7 + exploring a large number of initial conditions and accounting for the time - dependent dynamical friction and satellite disruption can only be carried out in a reasonable time by using semi - analytic algorithms .
we have used that proposed by pearrubia ( 2003 ) which provides , for the galaxy density profile outlined in section [ sec : models ] , the evolution of @xmath79 , where @xmath79 is the satellite centre - of - mass position , and the dwarf satellite mass @xmath80 .
this code has been tested against n - body calculations for a large spectrum of orbital parameters and satellite masses ( just & pearrubia 2004 ) as well as halo flattenings ( pearrubia , just & kroupa 2004 ) , showing that @xmath81 kpc for time integrations of 3 gyr , once the coulomb logarithm is fit to n - body simulations ( the best - fit values being @xmath82 for the halo and @xmath83 for the disk ) . in order to compare our collection of orbits to the observational data we assume that the distribution of stream stars can be reproduced by a single stream within a given number of wraps .
equivalently , we can define the number of wraps as the time interval that a point - mass particle needs to cover the phase space defined by the observational data , @xmath84 , which _ a priori _ is an unknown quantity that we have estimated from our n - body simulations to lie between @xmath85 , where @xmath86 is one dynamical period at @xmath87kpc from the galaxy centre ; @xmath88 km s@xmath89 .
the maximum integration time of semi - analytic orbits is @xmath90 gyr .
the fitting algorithm that we use is the following : + ( i ) each orbit is divided into @xmath91 overlapping segments of extension @xmath84 . following the above discusion
, we carry out the fit for stream extensions of @xmath92 gyr ( i.e @xmath93 ) .
the number of segments for a single orbit is @xmath94 , respectively , i.e , 0.6 gyr for @xmath95 and 1.2 gyr for @xmath96 . ] .
+ ( ii ) for a given segment we calculate @xmath97 , where @xmath98 is the heliocentric radial velocity and @xmath99 determines the relative weights of kinematical and spatial constraints ( @xmath100 kpc km@xmath89 s ) .
the sub - indexes @xmath101 take the values @xmath102 , where @xmath103 is the number of observational data , @xmath104 , and @xmath105 is the number of semi - analytic points in a segment . subsequently , for each observational value @xmath66 we look for the semi - analytic point that leads to the smallest value of @xmath106 , obtaining a set of @xmath103 semi - analytic points where the @xmath107values find a minimum .
the average value of @xmath108 in a given segment is @xmath109 .
+ ( iii ) the calculation is repeated for the rest of segments .
+ ( iv ) the @xmath108-values of each segment are sorted , obtaining the time interval of the orbit where the best fit occurs and the minimum @xmath108 .
we must note that this method fits observational to semi - analytic points , and not _ vice versa _ , which allows us to fit more than one orbital period to observational points located in a small region of the space ( see section [ sec : res_sense ] ) .
+ ( v ) the process repeats for the whole orbit collection ( regarding that each orbit corresponds to one point of our parameter space ) .
the semi - analytic code reproduces remarkably well the motion of a dwarf satellite centre - of - mass and its mass evolution .
however , as piatek & pryor ( 1995 ) showed , the mass loss process itself is fairly complicated and difficult to implement in semi - analytic algorithms . in order to describe the distribution of stripped stars in the galaxy
we perform n - body simulations from the best - fitting orbits found using the semi - analytic code .
we carry out self - consistent n - body calculations with superbox , a particle - mesh code ( see felhauer et al .
2000 ) , which calculates the gravitational potential in three boxes centred at the disk , bulge , halo and at the dwarf satellite , each box with @xmath110 grid - cells .
we refer the reader to pearrubia ( 2003 ) and pearrubia , kroupa & boily ( 2002 ) for a detailed description of the code parameters . here
we merely comment that , after fixing the time - step to 0.65 myr , we obtain a conservation of total energy and total angular momentum of around 1% .
the number of particles of each sub - system is : @xmath111 ( disk ) , @xmath112 ( bulge ) , @xmath113 ( dark matter halo ) and @xmath114 ( dwarf galaxy ) . as in the case of semi - analytic calculations , we evolve n - body satellites approximately 3 gyr .
three factors induce lead us to select this integration time : ( i ) to reproduce semi - analytic calculations thoroughly , ( ii ) to minimise feedback effects from the host galaxy affecting real dwarf satellites that are difficult to implement in n - body realizations of the milky way ( such as the disk s spiral arms , over - density regions , warps ... etc ) .
those effects can be treated like small corrections to our galaxy potential and , thus , neglected in short - time orbit calculations . and
( iii ) to approximate the orbits of tidal stream stars as the orbit of the main system .
when obtaining the progenitor s main orbital properties from semi - analytic calculations , one assumes implicitly that the escaping particles follow the main system s orbit during a given time interval .
this assumption only holds , therefore , for a limited number of orbital periods .
in fig . [ fig : aj ] we plot the results of our semi - analytic fit to observational data for two different stream extensions , @xmath115 and @xmath116 , and for prograde and retrograde orbits , as the sense of rotation has not yet been unambiguously determined .
this figure shows that the best - fitting parameter values do not depend on the exact value of @xmath84
. however , larger segments do alter the value of the minima and the smoothness of the curves by increasing the ratio @xmath117 ( i.e , in a given segment , more semi - analytic points can be fit to the same number of observational constraints ) , improving the quality of the fit but , at the same time , leading to less pronounced minima , which hardens the selection of the best - fitting parameter sample .
+ in lower - right panel we show the minimum value of @xmath108 as a function of the orbit s initial eccentricity .
the minima are located at @xmath118 for prograde orbits and @xmath119 for retrograde ones . as we see in the lower - left panel , the initial heliocentric distance that leads to best fits independent of the orbital sense of motion
is @xmath120 kpc .
+ in the upper - left panel we show the dependence of @xmath121 on the inclination angle with respect to the disk plane ( note that the orbital inclination of retrograde orbits is @xmath122 with this notation ) .
we find that the orbital inclination of retrograde orbits can not be accurately determined .
these orbits reach the solar circle which , therefore , reduces the number of passages by the galactic anticenter ( where most of the observational data are located ) . as a result ,
retrograde solutions can not account for the large vertical dispersion of observational points ( see fig .
[ fig : retsa ] ) .
in contrast , prograde orbits present a well - marked minimum at @xmath123 .
+ the absence of a defined vertical structure of debris also leads to a degenerate value of the halo flattening , as we see in the upper - right panel . for prograde orbits ,
the halo axis - ratio that leads to the best solution is for @xmath124 , although similar @xmath121 have been found for @xmath125 .
no information about the halo flattening can be obtained from the fit if orbits are retrograde .
as one would expect there are a large number of orbits that lead to similar values of @xmath108 .
that is far from surprising taking into account that : ( i ) the region of the galaxy where the monoceros stream has been detected is relatively small ( @xmath126 $ ] kpc , @xmath127 , in galactocentric coordinates ) , which represents a small interval of the orbit and ( ii ) the stream presents a large dispersion in the @xmath128 direction ( perpendicular to the disk plane ) with a poorly defined structure ( see fig . [
fig : prosa ] ) . in table
[ tab : param ] we summarize the fit results and show the three best - fitting retrograde and prograde orbits .
the values of @xmath121 clearly indicate that retrograde orbits lead to considerable worse fits @xmath129 .
our fit technique also provides a coarse estimate of the progenitor s initial mass , which we estimate to be around @xmath130 if it moves on a prograde orbit and @xmath131 if the orbit is retrograde .
we remark that this result is fairly approximate ( see precision estimates in table [ tab : param ] ) , since the orbital decay within 3 gyr is very low for prograde orbits and practically negligable for retrograde ones ( note that the effect of the dwarf satellite mass on the semi - analytic orbit calculation occurs through dynamical friction ) .
the poor sensitivity of the result to the satellite mass justifies our early choice of a coarse grid in satellite masses . in fig .
[ fig : prosa ] we compare the semi - analytic satellite orbits with the observational data used for the fit .
we only show the orbit at the time interval when the best fit to observations occurs ( see sec .
[ sec : fit_sa ] ) .
the best prograde orbits stay relatively far from the galaxy centre , @xmath132 $ ] kpc , and has a peri - center at @xmath133 .
different values of halo flattening and initial azimuthal angle match the observed heliocentric distances , radial velocities and projected positions of debris with a similar accuracy , leading to multiply denerate solutions . in fig .
[ fig : retsa ] we illustrate our best - fitting retrograde orbits .
those orbits have an apo - galacticon at @xmath133 , where most of observational points are located .
they reach the solar circle , moving within @xmath134 $ ] kpc .
since orbits in the galactic potential follow rosettes , the retrograde orbits pass the galactic anticenter considerably less frequently than the prograde ones .
the satellite tidal radius was selected to match that of the galaxy at the peri - centre , therefore , we find that for prograde satellites @xmath135 kpc whereas for retrograde ones @xmath136 kpc , where @xmath137 are the king and the tidal radius , respectively .
note that our retrograde satellites are approximately 3.3 times smaller than the progrades in order to prevent tidal disruption by the first peri - galacticon passages .
in this section we analyze the resulting kinematical and spatial distribution of debris from the best - fitting orbits obtained by our n - body algorithm . in fig .
[ fig : proper ] we plot the projection of the pro and retrograde orbits in the x - y plane , the radial velocity curve and proper motions ( in galactocentric longitudinal and latitudinal components ) obtained from n - body realizations of our best - fit models and compare them to the observational data outlined in section [ sec : obs ] .
we have integrated the prograde orbit 2.99 gyr in order to reproduce the observed projected distribution of debris ( see section [ sec : mb ] ) whereas the retrograde one was evolved only 1.9 gyr , until it was close to total disruption . even imposing the initial satellite tidal radius to match the jacobi limit at the first peri - galaction distance ( which leads to a retrograde satellite approximately 3.3 times smaller than the prograde one )
is not sufficient to assure longer survival times , likely due to the enhanced mass loss induced by disk and bulge tidal shocks . as commented in sec .
[ sec : res_sa ] , the slope of the radial velocity curve nearby the anti - center @xmath138 can be well reproduced either by low eccentricity , prograde orbits with peri - galacticon at @xmath133 or by high eccentricity , retrograde orbits with apo - galacticon at @xmath133 .
hence , the information provided by the radial velocity curve is not sufficient to determine the rotational sense of motion of the tidal stream progenitor .
we must remark on the notably low observational velocity dispersion , which is better reproduced by the prograde model , as one would expect from semi - analytical results ( see middle panel of fig .
[ fig : prosa ] ) . proper motions in the latitudinal direction do not break the degeneracy between the two different orbital senses of motion , either , since both models predict a similar range of @xmath33 values , consistent with the observations taking into account the large error bars .
in contrast , proper motion constraints in the galactic plane direction , @xmath32 , settle the sense of rotation to be prograde ( fig .
[ fig : proper ] , lower panel ) .
the prograde model predicts proper motions in that region of the order of @xmath139 mas / yr , whereas for the retrograde model we find a much larger azimuthal angular velocity @xmath140 mas / yr due to the larger relative velocity between the sun and the dwarf galaxy . in order
to conclusively differenciate the orbital sense of motion , we use proper motions of 10 stars in the stream derived by munn et al .
( 2004 ) . comparing these measurements with the theoretical predictions we find that , except for two stars , observed values of @xmath32 are remarkably well reproduced by the prograde model ( in magnitude as well as in sign ) .
we have transformed @xmath141 into cartesian velocities in the galaxy frame ( eq . [ eqn : vel ] ) in order to determine the components of angular momentum and the orbital inclination @xmath142 , where @xmath143 ( bottom panel are positive definite quantities with large errors , which leads to a biased estimate unless some statistical correction is applied .
we have corrected them by using the technique of wardle & kronenberg ( 1974 ) . ] ) . in fig .
[ fig : angmom ] we plot the distribution of n - body particles ( models _ pro1 _ and _ ret1 _ ) and observational points in the angular momentum plane @xmath144 ( upper panel ) .
tidal stream particles are located in well - defined regions determined by the main orbital inclination ( @xmath145 , @xmath146 for prograde and retrograde orbits , respectively ) and by the main eccentricity ( which decreases for decreasing values of @xmath147 ) . in the bottom panel
we plot the orbital inclination as a function of galactocentric longitude ( @xmath148 prograde orbits , @xmath149 retrograde orbits ) which shows that the monoceros stream progenitor likely follows a prograde orbit .
unfortunately , the observational errors are too large to provide an estimation of the main orbital inclination from these stars . + it is interesting to note that simply by measuring kinematical properties of stream stars ( radial velocities plus proper motions ) we would be able to determine , not only the sense of motion of the progenitor system , but also its orbital eccentricity and inclination if those measurements were accurate enough . as fig .
[ fig : angmom ] shows , the kinematical properties of tidal stream particles are similar to those of the progenitor during several periods . as a result
, the spatial distribution of debris is an approximate representation of the dwarf galaxy orbit . in fig .
[ fig : proper ] we show the projection in the galactic x - y plane of dwarf galaxy n - body particles for models _ pro1 _ and _ ret1 _ , observing that : + ( i ) stripped particles from the prograde satellite are located preferentially at two galactocentric distances , @xmath150kpc and @xmath151kpc , building two nearly concentric `` rings '' in space . that peculiar spacial distribution forms due to 1 ) the low orbital eccentricity and 2 ) the anisotropic mass loss . as piatek & pryor ( 1995 ) found ,
disrupting dwarf galaxies become elongated due to the action of tidal fields .
stripped particles escape preferentially parallel to the major axis which , at the same time , is oriented perpendicular to the density gradient vector . due to this process , particles belonging to the trailing and the leading tail move , respectively , outwards and inwards with respect to the main body orbit .
this process can be clearly seen in top - left panel , the satellite centre - of - density being located at @xmath152 kpc and moving clockwise .
the retrograde orbit forms a rosette with apo - galacticon at 22 kpc and peri - galacticon at 7.3 kpc , also providing a reasonably good fit to the monoceros star detections but failing to reproduce the gap between the two tails observed in the tri / and regions .
there are some facts that point to the possible survival of the monoceros stream s progenitor .
stars escaping from a tidally disrupting system follow the main system orbit for some time ( the exact evolution of their orbits depend on several parameters ) , building up what has been defined as `` tidal tails '' .
stars that were lost several orbital periods ago follow orbits in the host galaxy potential nearly independent of their parent system .
these particles would spread in a large volume of space ( no over - density signature ) with no velocity gradient .
helmi et al .
( 2003 ) find that only careful measurements of the phase - space structure might indicate whether they belonged originally to a disrupted body .
according to their predictions , the spatial location of debris forming a typical tidal stream structure as well as their well - defined radial velocity curve indicate that the tidal tails that we observe are possibly _ young _ and , therefore , either ( i ) the main body has not been yet completely destroyed or ( ii ) the disruption occurred recently .
as shown in section [ sec : res_sa ] , the range of distances and galactocentric latitudes where parts of the stream are detected provides robust values of the orbital inclination and eccentricity of the progenitor .
however , the exact vertical and radial distribution of debris depends on the location of the main body ( or , equivalently , the time when we stop the n - body simulation ) .
we have two main constraints to fix the progenitor location : ( i ) geometrical distribution of debris and ( ii ) different metalicities observed in different regions . 0.3 cm ( i ) * geometrical constraint * : + detections obtained so far show that the monoceros stream forms a complex vertical structure from the disk plane , if all detections belong to a single tidal stream . as shown in the bottom panel of fig .
[ fig : pro ] , such structure can plausibly arise from a single disrupted satellite . different studies ( majewski et al . 2004 , newberg et al .
2004 ) have reported the presence of two parts of the stream at @xmath153 , both located at similar projected position but at different heliocentric distances .
observations close to the disk plane do not show presence of the distant tail ( @xmath154 $ ] ) , likely due to dust abortion , up to @xmath155 where it is again detected .
we remark that the close tail has been detected only at negative latitudes ( @xmath156 $ ] ) .
+ we mainly use observational points at @xmath157 to determine the progenitor position , since the range of distances of detections at @xmath158 $ ] is considerably smaller which , therefore , provides a weaker constraint .
looking at this longitudinal range we observe that , if the main body is placed at @xmath159 ( integration time @xmath160 gyr ) we find no distant tail at @xmath161 ( only leading tail particles moving in the closest `` ring '' can be found in this region ) , whereas if placing the main body at @xmath162 ( integration time @xmath163 gyr ) we find a luminous close tail at @xmath164 , not present in observations .
therefore , we obtain comparable fits to observations if the main body is located in the range @xmath165 , which corresponds to integration times from 2.66 gyr to 3.04 gyr , respectively ( note that prograde orbits move with @xmath166 ) .
0.3 cm ( ii ) * metallicity constraint * : + detections of the distant tail at @xmath167 show metalicities considerably lower than those of the close tail in the same longitudinal range ( rocha - pinto et al .
although we have no information on the initial metallicity gradient of the progenitor , we assume that : 1 ) stars with originally high binding energy are those with the highest metallicity , as star formation nd metallicity enrichment happen more intensely in the center of dwarf galaxies ( pagel & edmunds 1981 , harbeck et al .
2001 ) and 2 ) those processes generating or enhancing any metallicity gradient have stopped , or are very slow , in the last 3 gyr . + under those assumptions , the actual distribution of stream metalicities is related to the age of the tidal tails , since low metallicity stars with low binding energy are stripped out by the action of tidal forces more rapidly than stars with initially high binding energy and high metallicity , which will be preferentialy located in recently stripped tails . + matching the metallicity observed in different regions of the stream reduces the range of possible progenitor locations obtained from geometrical and kinematical constraints .
we have compared relative metalicities between the distant and the close tails within @xmath167 . in fig .
[ fig : metal ] we plot the distribution of stream particles as a function of their initial binding energy for two different positions of the main body ( equivalently , two final integration times ) .
solid lines show the initial distribution ( once the satellite is numerically relaxed ) , dotted and dashed lines the distribution of particles with @xmath168 kpc ( `` inner ring '' ) and @xmath169 kpc ( `` outer ring '' ) , respectively .
+ the upper panel shows the result of placing the satellite at @xmath170 .
we can see that the number of particles with low initial binding energy ( which one would expect to be metal - poor ) is larger in the close tail , contradicting observations .
if we integrate a longer time , so that the main body locates at @xmath171 , the number of particles in the distant tail with initial low binding energies is clearly larger than in the close tail . in this case
, the outer tail has in average lower metallicity than the close one .
it is interesting to note that the ideal mass loss process , in which mass shells are progressively removed , is fairly approximate .
shell crossing occurs during the evolution of the satellite , likely due to the action of galactic tidal forces and shocks . as a result
, we observe the presence of particles with initially high binding energy in both stream tails .
this fact might explain observations of high and low metallicity stars in same fields ( e.g rocha - pinto et al .
although some mixing occurs , fig . [ fig : metal ] shows that the number of particles with initial high binding energy that remain in the main system is larger than in the tidal streams . in fig .
[ fig : pro ] we plot the heliocentric distance ( upper panel ) and galactocentric latitude ( middle panel ) of debris as a function of galactocentric longitude .
lower panels show the distribution of particles around the main system position .
the combination of geometrical and metallicity constraints fixes the main body at @xmath172 deg .
, @xmath173 deg . , corresponding to an integration time of @xmath174 gyr .
the heliocentric distance is approximately @xmath175 kpc .
we remark that this result is not definitive .
unfortunately , the available data are not yet sufficient to provide strong constrains on the main system location .
the progenitor position presented here is fairly approximate and is based on the assumption of an initial metallicity gradient of the progenitor .
we expect that future detections , mapping larger areas of the sky , will provide stronger geometrical constraints and reduce the number of possible scenarios .
in this section we discuss the possible identification of the monoceros stream progenitor in canis major ( martin et al .
2004a ) , by comparing its orbital properties to those of our model _ pro1_. we also analyze the orbits of three globular clusters with measured proper motions thought to be associated to cma dwarf .
lastly , we briefly summarise and discuss alternative explanations for the monoceros tidal stream that can be found in the literature . in fig .
[ fig : pertri ] we show the location of the recent detected tri / and tidal streams ( full squares , rocha - pinto et al .
2004 ) against the debris distribution of model _ pro1 _ ( grey ) .
we also show those detections used to constrain the orbital parameters . the location and radial velocity of cma over - density region
is shown by a large open circle .
the prograde model suggests the recently discovered stream in triangulus / andromeda as natural part of the monoceros stream , both fitting accurately to the modeled kinematics and spatial distribution of debris .
the observed tri / and streams ( @xmath176 $ ] ) appear as a connection between the monoceros stream ( @xmath177 $ ] ) and the new sdss and ibata detections at @xmath178 $ ] , justifying the inclusion of these data in our fit , as we discuss in section [ sec : obs ] .
martin et al .
( 2004a ) identified an elliptical over - density of m - giant stars in the canis major region with properties suggestive of a disrupted dwarf galaxy : ( i ) standard models of the milky way can not account for the large number of red giants in that region of our galaxy and ( ii ) the low dispersion of the radial velocity distribution ( approximately 20 - 25 km / s ) is unexpected for a disk population . from the number of m - giant stars
, these authors estimate a mass of @xmath179 , similar to the sgr dwarf galaxy .
the location in galactocentric coordinates is @xmath180,@xmath181 , with no data for @xmath182 due to dust abortion . in a second paper , martine et al .
( 2004b ) find that the heliocentric distance of this systems is @xmath183 kpc , with the maximum surface density being located at @xmath184 and @xmath185 .
they also measure a the radial velocity of @xmath186 km / s . adopting the selection criteria of bellazzini
et al . ( 2004 ) and martin et al . ( 2004a ) to select m - giant stars from the cma field , momany et al . ( 2004 ) measure the following proper motions @xmath187 mas / yr . martin et al . ( 2004a ) concluded that cma is a satellite galaxy undergoing tidal disruption and , due to its apparent similar position , they suggested that it probably is the remnant of the monoceros stream progenitor . here
we compare the orbital properties of cma to our best fit of the monoceros stream progenitor in order to analyze a possible common origin of both systems .
we must remark that our comparison is still fairly preliminar since it goes beyond the original scope of this paper .
we have integrated the orbit of cma back in time in order to compare its kinematical properties with those of our model .
the velocity of cma in cartesian coordinates with origin in the galactic center was obtained using the following expression : @xmath188 where @xmath189 km / s ( binney & merrifield 1998 ) is the solar velocity , @xmath14 the heliocentric distance and @xmath190 a conversion factor from ( kpc mas / yr ) to ( km / s ) .
the velocity vector is ( -145.4,193.2,-4.2 ) km / s .
the resulting cma s orbit has low inclination ( @xmath191 ) and is nearly circular @xmath6 , as argued by momany et al .
[ fig : imu ] shows that the value of @xmath66 is strongly sensitive to the errors in proper motions .
taking into account the errors estimated by these authors we obtain that , within @xmath192 error , the orbital inclination of cma lies in the range @xmath193 $ ] if fixing the radial velocity to @xmath186 km / s and the heliocentric distance to 7.2 kpc . comparing this result with the best - fitting orbits that we present in this work
one can see that : + ( i ) the orbital eccentricity is fairly similar to the one we find ( model _ pro1 _ predicts @xmath194 ) .
+ ( ii ) the orbital inclination of cma is too low to account for the large vertical dispersion of stream stars .
this result is , however , not conclusive since a similar orbit to that of model _ pro1 _ lies within @xmath195 error .
more thorough measurements of proper motions are necessary to clarify the orbit of the cma dwarf .
+ ( iii ) the geometrical and metallicity constrains that we can impose with the available observational data fix the final position of our best - fitting model at @xmath196 , @xmath197 and a distance to the sun of @xmath198 kpc .
cma appears in a similar direction as the main system of model _
pro1 _ but 7 kpc closer .
if cma proves to be the monoceros stream progenitor , a way to reconcile the close distance of the cma with the distant stream detections might be found by increasing the progenitor mass .
a simple estimation of the decay rate induced by dynamical friction on nearly circular orbits is @xmath199 ( eq .
7 - 25 of binney & tremaine 1986 ) . for the model
_ pro1 _ @xmath200 kpc after @xmath201 gyr . imposing @xmath202 kpc in 3
gyr leads to an initial mass of the progenitor of approximately @xmath203 , i.e @xmath204 .
however , it is unclear whether the tidal debris of such a massive satellite would also reproduce the observed distribution . in order to check this assumption , new investigations adopting the cma dwarf as the monoceros stream progenitor appear necessary , which goes beyond the scope of this paper .
we have also analyzed the distribution of radial velocities and proper motions in different regions of the stream in order to contrast them to the recently available observational values .
+ as we can observe in fig .
[ fig : pertri ] , our model predicts a hot tail overlapping the progenitor remnants .
this tail was stripped out approximately 0.6 gyr ago and belong to the trailing tail ( i.e to the `` inner ring '' ) . in upper panel of fig .
[ fig : vr ] we plot the radial velocity distribution ( dotted line ) of 50 particles located in the neighbourhood of the main system s projected position and compare it to that of the center of our satellite model ( full line , note that each distribution is separately normalized to the number of particles in the sample ) .
this figure shows than the velocity distribution of both tails can be clearly differentiated , with peaks at @xmath205 km / s ( cold tail ) and @xmath206 km / s ( hot tail ) .
although some particles of the underlying tail are included in the velocity distribution of the satellite centre , one can not appreciate a second peak because the surface density of the hot tail is much lower that of the satellite centre .
in lower panel of this figure , we repeat the calculations in a crossing tail region , where the surface densities of both tails are comparable . in this case , one observes clearly two distributions of radial velocities ( those from the distant and the close tail ) in the same particle sample .
+ the observed low - dispersion bimodality can not be reproduced by standard models of the milky way and is likely a proof of a stream detection , as we show here .
intriguingly , martin et al .
( 2004b ) report a bimodal velocity distribution in cma stars located at the center of the cma dwarf . measuring the radial velocity of 27 m - giant stars in one degree radius around cma position they observe two peaks in the radial velocity at @xmath207 km / s ( 10 stars ) and @xmath208 km / s ( 17 stars ) with very low dispersion ( around 5 km / s and 11 km / s , respectively ) .
these values are around 20% larger than those of model _
pro1_. extrapolating the resulting distribution of our model to their detections suggests that in the cma region two stream tails are overlapping .
moreover , the fact martin et al .
( 2004b ) measure similar number of stars in both peaks indicates that their surface density must be similar ( if we assume that the number of m - giants in a given stream is proportional to its surface density ) .
this , however , contradicts the presence of a dwarf galaxy in canis major since the surface density at the center of these objects is several orders of magnitude larger than in the tidal streams .
that paradox might be solved if their small sample of m - giant stars was strongly contaminated with stars that belong to the hot tail surrounding cma . in that case , the proper motions provided by these authors might be also affected . in right column of fig .
[ fig : vr ] we show the distribution of proper motions in the regions indicated above . as in the distribution of radial velocities ,
different stream parts lead to well differentiated curves .
looking at the region where the dwarf remnants are located , we can observe that the main system shows a fairly narrow distribution , with maxima at @xmath209 mas / yr , whereas the distribution of the hot , overlapping stream tail is centred at @xmath210 mas / yr . thereferore , including hot tail stars in the dwarf remanant sample ( in a significant proportion ) would lead to a smaller @xmath211 and a larger @xmath212 , which results in a lower orbital inclination ( see fig .
[ fig : imu ] ) .
frinchaboy et al .
( 2004 ) have collected a set ( 15 ) of globular and open cluster that show a trend in their radial velocity curve as well as in spatial location which may indicate that those systems were stripped from a satellite in disruption process .
martin et al .
( 2004a ) claimed that four of those clusters ( ngc 1851 , 1904 , 2298 and 2808 ) belong to a globular cluster system associated with the cma dwarf , arguing for the possible detection of the monoceros stream s progenitor in canis major .
we have earlier shown that radial velocities and positions do not provide sufficient information to distinguish between pro- and retrograde orbital motions . in order to break that degeneracy ,
accurate measurements of proper motions are needed . in this section
we test the possible association of those clusters with measured proper motions ( ngc 1851 , 1904 and 2298 from dinescu , girard & van altena et al .
1999 ) with the cma dwarf .
additionally , we also discuss the possible common origin of six `` possibly associated clusters '' suggested by frinchaboy et al .
( 2004 ) by contrasting locations and radial velocities with the predictions of our best - fitting model .
unfortunately , no proper motions are available for those systems , so that the results shall not be conclusive . in fig .
[ fig : cum ] we plot the x - y projection of model _
( upper - right panel ) and clusters listed in the upper - left panel .
middle - upper panel shows the radial velocity , middle - lower and lower panels plot proper motions in the longitudinal ( @xmath32 ) and latitudinal ( @xmath33 ) directions , respectively .
this figure shows that : 0.3 cm ( i ) * cma clusters * : + none of these globular clusters present kinematical properties consistent with those of the cma dwarf .
+ firstly , ngc 1851 and ngc 1904 have radial velocities of approximately 320 km / s and 208 km / s whereas that of the cma dwarf is approximately 109 km / s ( martin et al .
after integrating the orbit we find that both globular clusters move on orbits with much higher eccentricities , @xmath213 and @xmath214 , respectively ( in agreement with dinescu , girard & van altena et al .
1999 , who use a a similar galaxy potential ) . on the other hand ,
the orbital eccentricity of escaping particles remains fairly similar to that of the main body ( see fig . [
fig : angmom ] ) , leading to a well - defined radial velocity curve .
if one assumes that this holds for stripped globular clusters , we come to the conclusion that ngc 1851 and 1904 are unlikely associated with cma .
additionaly , the orbital inclinations of those systems are considerably higher than that of cma
. + the third globular cluster with measured proper motions , ngc 2298 , follows a retrograde , highly eccentric orbit ( @xmath215 , dinescu , girard & van altena et al .
1999 ) , arguing against any association with cma and the other clusters .
+ unfortunately , there are no proper motions available for ngc 2208 and , therefore , insufficient information to allow a determination of its possible association with cma
. 0.3 cm \(ii ) * other clusters * : + in upper - middle panel we can see that these clusters present a trend in radial velocities shown by frinchaboy et al .
( 2004 ) consistent with our model .
their projected positions ( upper panel ) also appear to form a `` stream '' in space .
searching through the galactic globular cluster sample provides some additional , plausable candidates to be associated with the monoceros stream : rup 106 , ic 4499 and ngc 6779 , from a comparison of their projected position , distance and radial velocity with the predictions of our model . unfortunately , whether or not these clusters are associated to the monoceros stream can not be asserted in absence of proper motions .
the main selection criteria used to discriminate between milky way stars and stars of an `` external '' origin were : ( i ) stream stars are observed to be in overdense regions not predicted by our standard milky way model and ( ii ) the turn - off stars appear blue , old and metal - poor , characteristics similar to those of the sagittarius dwarf galaxy stars ( newberg et al .
those conditions by themselves do not exclude other explanations for the observations .
+ here would like to comment briefly on two possible origins of the observed tidal streams : ( i ) thick disk and ( iii ) a satellite disruption .
in section [ sec : res ] we have shown that the stream detections can be reproduced by a disrupting satellite on a prograde , low - inclination , nearly circular orbit .
the kinematical as well as spatial distribution of debris resemble what one would expect for a thick disk stars ( see fig .
[ fig : proper ] ) except for one point : it is difficult to reconcile the large vertical dispersion ( @xmath216 kpc@xmath217kpc ) and the large distances ( @xmath218kpc from the galaxy centre ) of proposed stream stars with a disc - like distribution . to clarify this point ,
we have carried out a simple experiment : we have calculated the probability of finding thick disk stars at the monoceros stream location .
the number of stars between @xmath219 and @xmath220 , where @xmath219 is apparent magnitude can be written as @xmath221 where @xmath14 is the distance of a star to the sun .
we have used the relationship @xmath222 to convert distances into apparent magnitudes .
the probability of finding a star between @xmath14 and @xmath223 , in the range of solid angles @xmath224 is @xmath225 where @xmath226 is the thick disk density distribution in the solar frame . combining both equations
one has that the probability function is @xmath227 where @xmath228 is some normalization constant .
the fraction of thick disk stars in a given solid angle is therefore @xmath229 as we can see , @xmath230 is zero for @xmath231 and @xmath232 so that , for a given solid angle , there is a distance where the probability finds a maximum . in fig .
[ fig : prob ] we plot the maximum like - hood iso - contour of observing thick disk stars ( strong full line ) in the galactic anticenter direction ( eq .
[ eqn : p ] ) .
the sun is placed at @xmath233 .
full lines show the iso - contours of the thick disk fraction of particles as a function of their position in cylindrical coordinates when looking in the galactic anticenter direction ( eq .
[ eqn : n ] ) .
dotted lines show the iso - contours if looking @xmath234 away ( i.e , @xmath235 or @xmath236 ) in order to take into account the range of galactic longitudes where the stream has been observed . in both cases
we plot the 10% , 50% and 90% contours .
as we can see , the galactic anticenter is the direction for which the number of distant thick disk stars is maximum . for this plot ,
we have used the thick disk model proposed by chen et al .
( 2001 ) , which follows exponential profiles in the planar and vertical directions with scale - lengths of 3.5 kpc and 0.75 kpc , respectively . +
this figure shows that stream detections presented in this work are unlikely related to the thick disk population because ( i ) they are located in a narrow range of co - planar distances ( @xmath237 ) , as one would expect for debris from a disrupting galaxy in a nearly circular orbit ; ( ii ) they present a large vertical dispersion , of approximately 20 kpc which can not be reconciled with a disc - like structure and ( iii ) they have been detected in positions where the fraction of thick disk stars is @xmath23810% of the total number of stars in that direction .
the only doubtful data might be found in the colour - magnitude detections at @xmath239 kpc , which lie within 90% fraction of thick disk stars .
these points correspond to what our model identifies as the `` close ring '' , which presents a higher metallicity than more distant detections . + it is interesting to note that stars stripped from the monoceros stream progenitor have not yet contributed to the thick disk population .
the decay rate of the dwarf galaxy appears slower ( around 3 kpc in 3 gyr ) than the mass loss rate ( 50% in the same time period ) , so that the dwarf galaxy will be likely detstroyed before reaching the inner regions of the thick disc . the mass loss process , however , depends on the initial structural parameters of the dwarf , which were fixed _
ad hoc _ in this work .
more accurate measurements of the stream surface brightness will provide better estimations of its survival time .
in this paper , we have used a combined semi - analytic / n - body technique to explore the nature and origin of the monoceros tidal stream . this method has allowed us to explore a large parameter space in a systematic way , with the goal of constraining the orbital properties of the monoceros stream .
+ we have found that the available observational data at the present day is sufficient to robustly constrain some aspects of the monoceros stream history and progenitor . in particular : ( i ) the heliocentric distance range at which the stream is observed plus the heliocentric radial velocity curve determines an orbital eccentricity of @xmath240 , ( ii ) the range of galactic latitudes indicates an orbital inclination of @xmath241 from the disk plane and , lastly , ( iii ) proper motions of stream stars are only compatible with a prograde orbit .
we found that it is considerably less straightforward to predict the main body position through numerical calculations .
the range of distances , the radial velocity curve and the vertical extension of debris are relatively insensitive to the final location of the satellite galaxy , since they define a given volume in phase - space that the progenitor fills up after several wraps .
yet , the projected positions of stream tails are time - dependent reflecting , therefore , the position of the main system or , equivalently , the final integration time of our n - body model .
+ besides the uncertainty in the satellite remnant position , the lack of detections at low galactic latitudes ( owing to disk absorption within @xmath242 ) as well as for a complete range of galactic longitudes makes difficult to constrain the location of the main system . owing to these limitations , our model matches the geometrical and kinematical distribution of debris if the main system remnants locate within the range @xmath165 , corresponding to integration times from 2.66 to 3.04 gyr .
+ on the other hand , observations show that close detections of the stream have higher metalicities than the distant ones , which provides an additional constraint to our model .
we have used this constraint by assuming that the progenitor s stellar metalicities are related to their initial binding energy .
in particular , we assume that stars with low binding energies move , in average , in the outer satellite regions and , therefore , should present lower metalicities than those with high binding energies . by comparing the initial binding energy ( metallicity ) distribution with observations for the range of locations commented above ,
we find that the metallicity gradient between the distant and the close detections can only be reproduced for integration times longer than 2.9 gyr . in that case
, the main system location ( @xmath243 , @xmath244 ) is similar to that of the canis major dwarf , with a heliocentric distance of @xmath175 kpc .
the treatment employed to determine the distribution of metalicities in the stream is , however , approximate , because have no way of knowing the initial metallicity gradient in the stream progenitor .
moreover , we implicitly assume that detections with different metalicities reflect an initial property of the progenitor .
other scenarios are possible and , for example , we can not discard that streams with different metalicities at different distances belong to different progenitors .
the best - fitting model predicts a halo axis - ratio of @xmath124 .
however , the poorly - constrained selection function of observations and the incomplete area coverage leads to degenerated solutions for different values of @xmath54 .
in particular , we find that similar fits can be obtained with halo axis - ratios @xmath245 $ ] .
the model we present here is , therefore , far from being definitive .
future detections , mapping larger areas of the sky , will provide more constrains on the progenitor s orbit and a better determination of parameters listed in table [ tab : param ] .
new detections are also necessary to constrain the progenitor location by means of numerical models . with the recently available kinematical data of the canis major dwarf , which martin et al .
( 2004a ) claim to be the progenitor of the monoceros stream , we have integrated the orbit backwards in time , comparing its orbital properties against those of our model .
we find contradictory points in favour and against that suggestion : * the orbital eccentricity of cma ( @xmath246 ) and orbital sense of motion ( prograde ) are consistent with those of our model . *
the cma s projected location ( @xmath184 , @xmath247 ) is consistent with the geometrical and metallicity constrains imposed by the monoceros stream stars . * the orbital inclination ( @xmath7 deg . )
is consistent with that of our model within @xmath195 errors from proper motions . * the radial velocity of cma ( @xmath248 km / s ) is also similar to that of our model ( @xmath249 km / s ) and presents a low dispersion incompatible with thick disk features .
* the orbits of globular clusters ngc 1851 , 1904 and 2298 , which bellazzini , ferraro & ibata ( 2003 ) and martin et al .
( 2004a ) claim to be an evidence that the cma over - density region is the remnant of the monoceros stream progenitor , are inconsistent with the orbit of cma , since ngc 1851 and ngc 1904 move on highly eccentric orbits and ngc 2298 on a retrograde , eccentric one . * the bimodality in the velocity distribution of cma s m - giant stars ( martin et al .
2004b ) can be reproduced by our model in regions where the projected positions of two stream tails overlap .
that bimodality in @xmath250 can not be observed in the central part of our satellite model since the surface density is several orders of magnitude larger than that of overlapping stream tails .
this fact seems to point out to a possible contamination of background giant stars in the sample used to measure proper motions and the radial velocity of the main system . as a result
, the orbital inclination might have been underestimated . *
the main body of our best - fit model s progenitor is @xmath251 kpc from the sun twice as distant as the observed canis majoris stellar structure .
since the mass of our model is poorly constrained , larger mass values can not be rejected . in particular ,
dynamical friction would drive our satellite model down to 7.2 kpc in 3 gyr for an initial mass of approximately @xmath252 , whereas our best - fitting model has an approximate initial mass of @xmath253 .
it remains unclear , however , whether the resulting distribution of debris from such a massive satellite undergoing tidal disruption would also reproduce the observations .
unfortunately , with the results obtained in this work we can not unambigously determine whether the orbit of the cma dwarf is consistent with that of the monoceros stream progenitor , partly due to the uncertainty in the available observational data .
more strict criteria to select stars that belong to the cma dwarf appear necessary to obtain reliable measurements of the kinematical properties of this system .
we want to thank dana i. dinescu for helping us to analyse the information obtained from proper motions .
we also thank h. rocha - pinto and j. d. crane for giving us access to their data before being published .
funding for the creation and distribution of the sdss archive has been provided by the alfred p. sloan foundation , the participating institutions , the national aeronautics and space administration , the national science foundation , the u.s .
department of energy , the japanese monbukagakusho , and the max planck society .
the sdss web site is ( ` http://www.sdss.org/ ` ) .
the sdss is managed by the astrophysical research consortium ( arc ) for the participating institutions .
the participating institutions are the university of chicago , fermilab , the institute for advanced study , the japan participation group , the johns hopkins university , los alamos national laboratory , the max - planck - institute for astronomy ( mpia ) , the max - planck - institute for astrophysics ( mpa ) , new mexico state university , university of pittsburgh , princeton university , the united states naval observatory , and the university of washington .
abazajian , k. , et al .
2003 , , 126 , 2081 bellazzini m. , ibata r. , monaco l. , martin n. , irwin m.j . , lewis g.f . , 2004 ,
mnras.tmp , 499b binney j. , tremaine s. , 1987 , galactic dynamics .
princeton university press , princeton , new jersey binney j. , merrifield m. , 1998 , galactic astronomy .
princeton university press , princeton , new jersey boily c.m .
, kroupa p. , pearrubia j. , 2001 , newa , 6 , 27 chen b. et al .
, 2001 , apj , 553 , 184 crane j.d .
et al . , 2003 ,
apj , 594 , l119 dinescu d.i . ,
girard t.m . , van altena w.f . , 1999 ,
apj , 117 , 1792 fellhauer m. , kroupa p. , baumgardt h. , bien r. , boily c. m. , spurzem r. , wassmer n. , 2000 , newa , 5 , 305 frinchaboy p.m. et al . , 2004 , apj , 602 , l21 fukugita , m. , ichikawa , t. , gunn , j. e. , doi , m. , shimasaku , k. , & schneider , d. p. 1996
, , 111 , 1748 gunn , j.e .
1998 , , 116 , 3040 harbeck d. , grebel e. k. , holtzman j. , guhathakurta p. , brandner w. , geisler d. , sarajedini a. , dolphin a. , hurley - keller d. , mateo m. , 2001 , aj , 122 , 3092 hernquist l. , 1990 , apj , 356 , 359 hernquist l. , 1993 , apjs , 86 , 389 helmi a. , j.f .
navarro , a. meza , m. steinmetz , v.r .
eke , 2003 , apj , 592l , 25 hogg , d.w . , finkbeiner , d.p .
, schlegel , d.j . , and gunn , j.e .
2001 , , 122 , 2129 ibata r. , gilmore g. , irwin m. j. , 1994 , natur . , 370 , 194 ibata r.,irwin m. j. , lewis g. f. , ferguson a. m. n. , tanvir n. , 2003 , mnras , 340 , 21l just a. , pearrubia j. , 2004 , a&a accepted king i.r . , 1966 , aj , 71 , 65 kinman t.d . ,
saha a. , pier j.r . , 2004 ,
apj , 605 , l25 law d.r . , johnston k.v . ,
majewski s.r . , 2004 , apj accepted , astro - ph/0407566 martin n.f . ,
ibata r.a . , bellazzini m. , irwin m.j . ,
lewis g.f . ,
dehnen w. , 2004a , mnras , 348 , 12 martin n.f . , ibata r.a . , conn b.c . ,
lewis g.f . , bellazzini m. , irwin m.j . ,
mcconnachie a.w . ,
2004b , mnras submitted , astro - ph/0407391 martnez - delgado d. , butler d. , rix h.w .
, franco y.i . , penrrubia j. 2004 , apj , submitted mateo m.l . , 1998 , ara&a , 36 , 435 momany y. , zaggia s.r . , bonifacio p. , piotto g. , de angeli f. , bdin l.r .
, carraro g. , 2004 , astro - ph/0405526 monet , d. g. , et al .
2003 , , 125 , 984 munn , j. a. , et al .
2004 , , 127 , 3034 navarro j. , 2004 , astro - ph/0405497 , paper to be read at the penetrating bars through masks of cosmic dust conference in south africa newberg h.j . ,
2002 , apj , 569 , 245 newberg h.j . ,
2004 , in preparation pagel b.e.j . , edmunds m.g . , 1981 , ara&a , 19 , 77 pearrubia j. , kroupa p. , boily c.m . ,
2002 , mnras , 333 , 779 pearrubia j. , 2003 , phd thesis , universitt heidelberg , germany .
pearrubia j. , just a. , kroupa p. , 2004 ,
mnras , 349 , 747 piatek s. , pryor c. , 1995 , aj , 109 , 1071 pier , j.r . , munn , j.a . ,
hindsley , r.b , hennessy , g.s . ,
kent , s.m . ,
lupton , r.h . , & ivezic , z. 2003 , , 125 , 1559 rocha - pinto h. , majewski , skrutskie , crane , 2003 , apj , 594 , l115 smith , j.a .
2002 , , 123 , 2121 stoughton , c. et al .
2002 , , 123 , 485 van den bosch f.c . , tormen g. , giocoli c. , 2004 , mnras submitted , astro - ph/0409201 de vaucouleurs g. , 1948 , annales dastrophysique , 11 , 247 velzquez h. , white s.d.m . , 1999 , mnras , 304 , 254 wardle j.f.c . ,
kronenberg p.p . , 1974 , apj , 194 , 249 yanny b. , et al .
, 2003 , apj , 588 , 824 york , d.g .
2000 , , 120 , 1579 | we have compiled an extensive dataset on potential parts of the monoceros tidal stream and performed an exhaustive survey of dwarf galaxy semi - analytic orbits in order to constrain its orbital properties .
the best - fit orbits are subsequently realized as self - consistent @xmath0-body simulations in order to reproduce the spatial and velocity distribution of satellite debris .
we find that all kinematic and geometric constraints can be fit by a single stream allowing for multiple wraps . the orbital eccentricity and inclination of the progenitor
are strongly constrained to be @xmath1 and @xmath2 .
ten new estimates of proper motions from the sloan digital sky survey ( sdss ) clearly exclude all retrograde orbits .
particles lost by the satellite populate two , nearly concentric rings naturally explaining the detection of stream stars at both 68kpc ( ibata et al .
2003 , newberg et al . 2003 ) and 1218 kpc ( the tri / and stream ; rocha - pinto et al .
2003 ) from the sun .
we have attempted to predict the present location of the monoceros stream progenitor using different information : ( i ) the kinematical and spatial distribution of detections and ( ii ) the different mean metallicity in the innner and the outer rings . due to the lack of observational data in the whole range of galactic latitudes ,
the geometrical / kinematical constraints leads to a wide range of possible locations . by associating older parts of the model stream with lower metallicity parts of the observed data ,
we argue in favor of a current location of @xmath3 . , with a distance to the sun @xmath415 kpc . the mass of the progenitor has been poorly constrained due to the slow orbital decay .
similar fits have been obtained for masses @xmath5 . + we have analyzed to possible common origin of the canis major dwarf and the monoceros stream .
canis major dwarf moves on a prograde , nearly circular orbit ( @xmath6 ) in the milky way disk ( @xmath7 deg . ) .
this orbital inclination is too low to account for the large vertical dispersion of stream stars .
however , the bimodal distribution of radial velocities in the central region found by martin et al .
( 2004b ) probably indicates that their selection criteria to indentifying dwarf stars lead to a contamination of background stars . in that case
, the kinematical data outlined above might result to an underestimated orbital inclination .
lastly , the distance estimation to canis major dwarf is around a factor two smaller than that obtained from our model .
unfortunately , the possible identification of the monoceros stream progenitor in canis major remains unclear . |
massive stars ( m@xmath118 ) earlier than about b2.5v dominate the cosmic production of ionizing photons and stellar wind momentum before terminating in nature s most energetic explosive events , leaving behind neutron stars and black holes .
formation scenarios for these explosions and subsequent compact stellar remnants require the existence of a close stellar companion . in x - ray binaries
the massive star becomes a neutron star accreting from an evolved star .
mergers of compact objects ( neutron star - neutron star or black hole - neutron star ) produce gravitational waves ( gw ) that should be detected by the immanent generation of gw experiments @xcite . `` runaway '' or high - velocity stars also owe their extreme kinematics to hard interactions with massive binary systems @xcite . in particular , stars in the range 825 that have
close companions may become the progenitors of type ibc supernovae when the h envelope of the more massive star is stripped during a phase of common envelope evolution @xcite .
close binaries may even produce all type ibc supernova if single h - poor wolf - rayet stars collapse to become black holes without producing a supernova @xcite .
furthermore , supernovae of type ic and @xmath12-ray bursts appear to happen simultaneously , suggesting a connection between the two types of events that may have their origins in massive binary progenitors @xcite .
close massive binaries also produce the population of low - mass x - ray binary systems through evolution of the more massive star to a supernova , production of a neutron star or black hole , and then the subsequent evolution of the lower - mass companion that becomes the mass donor @xcite .
the cygnus ob2 radial velocity survey @xcite is an optical spectroscopic survey of 128 photometrically selected o and early - b stars comprising an unbiased ( with respect to binarity ) sample within the core region of the nearby cygnus ob2 association ( @xmath101.4 kpc ; @xcite ) .
this sample contains 45 stars of spectral type o and 83 stars of spectral type b. luminosity class v ( i.e. , unevolved ) stars account for 91 of the 128 .
the survey is designed to statistically measure the binary properties of massive stars within a common formation environment at a similar age .
although massive star formation in cyg ob2 may not be coeval , studies suggest an age of 34 myr , young enough that the most massive stars are still present and the majority of these stars are still on the main sequence @xcite . papers i vi in this series ( @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ) describe the survey and document orbital parameters for 25 massive binary systems known from the inception of the survey in 1999 through the 2011 observing season .
in particular , paper v @xcite provides a high - level overview of the survey s constraints on statistics of massive binary stars by use of a monte carlo analysis .
stars down to spectral type @xmath10b2.5v are included in the survey because they are expected to dominate the population of supernova progenitors , given typical initial mass functions .
other large studies of binarity among massive stars include the spectroscopic surveys of @xcite , @xcite , and @xcite as well as the imaging surveys of @xcite and @xcite .
in this contribution we report the observations ( section 2 ) and orbital solutions for 22 new single - lined spectroscopic binaries and one new double - lined binary ( section 3 ) .
these data contribute to the growing census of massive binary statistics in a complete sample of massive stars defined in paper i. this increasingly complete compilation of orbital periods , eccentricities , and mass ratios provides a rich dataset to constrain our understanding of massive star formation , evolution through binary channels , and massive star end states .
section 3 also contains a list of 16 stars that show minimal velocity variations after extensive observation and six stars all supergiants that exhibit irregular variations presumed to originate from atmospheric pulsations .
an appendix provides an update on orbital solutions for six stars from papers ii and iii for consistency with the most recent cyg ob2 analyses .
we conclude by tabulating orbital parameters for the 48 known binary ( and several triple ) systems and conducting a high - level analysis of the distribution of orbital parameters ( section 4 ) .
nomenclature of objects discussed herein follows the mt91 # # # notation of @xcite , with `` s # '' indicating the numeration of @xcite and `` cpr2002-a # # '' or `` b # # '' for that of @xcite .
observational methodologies and data reduction procedures closely follow those described in @xcite and earlier papers .
paper i describes instrumental setups and dates of observation for observing runs using keck+hires spectrograph ( 19992001 ) , lick+hamilton echelle spectrograph ( 19992000 ) , and wiyn+hydra spectrograph ( 20012008 ) .
results presented here principally use spectra obtained at the wyoming infrared observatory ( wiro ) 2.3 meter telescope+longslit optical spectrograph using a 2000 l mm@xmath13 grating in first order during the 20112013 observing seasons , with a small number of observations obtained in 2014 may .
resolutions of r@xmath144500 were achieved using a 12 slit .
multiple 600 s spectra of each target were obtained over the wavelength range 54006700 .
data from wiro in 2010 and earlier used either a 600 l mm@xmath13 grating in second order or an 1800 l mm@xmath13 grating in first order to achieve spectral resolutions of r@xmath142500 and 4000 , respectively .
these observations are more fully described in earlier papers in this series .
routine cuar lamp exposures provided wavelength solutions having a typical rms of 0.0120.030 .
spectra were combined and transformed into the heliocentric reference frame , yielding a final signal - to - noise ratio ( snr ) of 60200 in the vicinity of the 5876 line used to measure radial velocities .
the ensemble of spectra spanning hours to years were shifted by small amounts ( 2 8) so that the interstellar @xmath15 5889/96 lines align at the mean velocity across all observations .
this corrects for small changes in the wavelength solution zero point across epochs .
data from keck and wiyn did not cover the lines so these spectra potentially have systematic velocity zero point differences compared to the wiro data .
however , our analysis of constant - velocity stars in the survey shows good agreement , at the level of 36 , between the velocities obtained from keck , wiyn , and wiro spectra .
nevertheless , solutions are obtained solely from wiro data , where possible .
when necessary , data from keck and/or wiyn are included in the solutions .
data from lick were generally of lower quality and are not used .
radial velocities were measured by fitting gaussian profiles to the 5876 photospheric line ( 4471 for the keck and wiyn spectra ) adopting a rest wavelength of 5875.69 ( 4471.55 for keck and wiyn spectra ) .
constant - velocity stars showed good agreement ( 36 ) between velocities from the various telescopes and instruments , lending confidence to the solutions that involve all three sources of data .
our fitting code fixes the gaussian width and depth to be the mean determined from all the spectra , after rejecting outliers , and it solves for the best - fitting line center and its uncertainty . since the primary goal of the cygnus ob2 radial velocity survey is to obtain orbital parameters for massive binaries , a goal that requires good _ relative _ radial velocities , we did not observe radial velocity standard stars , and consequently the absolute space velocities reported are likely to be accurate to @xmath1035 .
nevertheless , there is good agreement between the mean velocities of our cyg ob2 ob star sample ( @xmath16=-15.6 ; @xmath17=8.2 ) and results from other workers ( @xmath18 - 18 ; n. wright , private communication ) .
table [ vel.tab ] records the heliocentric julian date , orbital phase , the heliocentric radial velocity , the velocity uncertainty , and the observed - minus - computed ( o - c ) velocity for each measurement based on the orbital solutions that follow .
all systems , with the exception of mt91 646 , are single - lined spectroscopic binaries so that only one velocity is reported for each epoch .
mt91 268 is a probable triple system exhibiting two periodicities , so both velocity components appear in table [ vel.tab ] .
figure [ color ] shows a three - color view of the cygnus ob2 region with 4.5 , 8.0 , and 24 @xmath19 m mosaic images from the _ spitzer cygnus - x legacy survey _ @xcite in blue , green , and red , respectively .
white points depict massive stars in the cygnus ob2 radial velocity survey parent sample , while magenta points mark known binary or higher - order systems . the 5 pc bar at lower left marks the linear scale at the adopted distance of 1.4 kpc .
binaries are apparently distributed across the face of the association without preference for radial distance from center .
we analyzed the radial velocity power spectrum for each object to select likely periods and then examined the folded velocity curve for periods corresponding to the strongest peaks . in most cases
the strongest peak yielded a clear , unambiguous period and a convincing phased velocity curve .
secondary peaks and possible aliases could be eliminated by visual inspection owing to the much larger dispersion in the data at any given phase .
we used the binary orbital solution package `` binary '' by d. gudehusgudehus / binary.html ] with these initial period estimates and the radial velocity data to solve for the full suite of orbital parameters and associated uncertainties .
tables [ solutions1.tab ] through [ solutions5.tab ] compile these best - fitting parameters and uncertainties for each object .
listed within the table are the period in days ( @xmath4 ) , eccentricity of the orbit ( @xmath20 ) , longitude of periastron in degrees ( @xmath21 ) , systemic radial velocity ( @xmath12 ) , epoch of periastron ( @xmath22 ) , primary velocity semi - amplitude ( @xmath23 ) and ( if applicable ) secondary velocity semi - amplitude ( @xmath24 ) , spectral classifications from this survey ( s.c.@xmath25 & s.c.@xmath26 , if available ) , estimates of the inclination ( @xmath27 ) , the adopted primary stellar mass ( @xmath28 ) and ( if applicable ) secondary mass ( @xmath29 ) , mass ratio ( @xmath30 ) , semi - major axis ( @xmath31 ) , and reduced chi squared values of the best fitting solution .
tables [ solutions1.tab ] through [ solutions5.tab ] list orbital parameters for 22 new systems reported in this work .
masses for ob stars are taken from @xcite for o stars and @xcite for early b stars .
upper and lower limits on the inclinations are obtained , in most cases , by adopting 90 and the lowest inclination compatible with the absence of secondary spectral features ( i.e. , where the mass of the secondary approaches that of the primary ) . _
mt91 021_@xcite classified the mt91 021 primary ( v=13.74 ; @xcite ) as a probable single - lined binary system on the basis of four data points . using a combination of three wiyn spectra
we revise the spectral type to b1.5v by comparison in the @xcite spectral atlas .
table [ vel.tab ] lists the 14 radial velocity measurements from wiro spanning 2010 august 2013 september
. figure [ sol021 ] shows the best - fitting orbital solution ( solid curve ) and folded velocity data ( points with error bars ) .
table [ solutions1.tab ] summarizes the full suite of orbital parameters .
if we adopt a mass of 12 for the primary ( mt91 021a ) , an inclination of @xmath27=85 implies a secondary ( mt91 021b ) mass of 2.2 , while @xmath27=15 requires the secondary ( mt91 021b ) mass to approach that of the primary .
this system is interesting for having such a high eccentricity despite its short period .
however , the eccentricity is strongly dependent on just a few data points , and a solution with fixed eccentricity of zero still yields a similar @xmath32 value .
finally , we note that the reduced @xmath32 of the fit ( 3.2 ) is large .
the o - c residuals appear to show evidence for periodic variation at @xmath104.3 days with an amplitude near 15 , suggesting the possibility of this being a triple system .
this would make the tertiary , mt91 021c , a probable solar mass star with a minimum mass of 0.5 .
additional measurements will be needed to confirm the periodicity and amplitude of a second velocity component in this single - lined system .
inclusion of older ( but less precise ) wiyn ( 2 spectra ) and keck ( 2 spectra ) measurements strengthen the evidence for more than one velocity component in mt91 021 . _
mt91 187_the b1 v primary of this v=13.24 system was not initially detected as a probable binary on the basis of five measurements @xcite .
the present dataset consisting of two keck spectra , four wiyn spectra , and 19 wiro spectra spanning 19992013 ( table [ vel.tab ] ) reveals this systems to be a @[email protected] d binary with @[email protected] .
figure [ sol187 ] shows the best - fitting orbital solution ( solid line ) and folded velocity curve data ( points with error bars ) .
this is one of the lowest amplitude systems yet detected in the survey .
the velocity variability is consistent with a binary ( reduced @xmath32=0.82 for 12 degrees of freedom ) but not a constant - velocity system ( reduced @xmath32=1.8 for 19 degrees of freedom , yielding a probability of @xmath342% ) .
the eccentricity is not well constrained at @[email protected] .
an alternative period of @xmath356.1 d is also possible , but less probable based on the power spectrum and the higher eccentricity required in the best fit .
there are no known eclipses from this system , so the inclination is essentially unconstrained . if we adopt a mass of 14 for the b1v primary , an inclination of @xmath27=85 implies a secondary ( mt91 187b ) mass of 0.43 , while @xmath27=3 allows a secondary mass approaching that of the primary .
given the probable low - mass secondary and extreme mass ratio , this system is a candidate for being a progenitor of a low - mass x - ray binary system if the system remains bound once the primary reaches its end - state as a neutron star . _
mt91 202_this @xmath36=14.40 binary is listed as an sb1 by @xcite on the basis of five data points , and they type the primary as b2v .
we find an orbital period of @[email protected] d with no credible aliases and an amplitude of @[email protected] km s@xmath13 using 17 measurements from wiro .
the eccentricity is @[email protected] .
adopting 11 for the primary yields @xmath29=2.0 if @xmath27=90 . the lack of known eclipses does not provide useful constraints on the inclination given the relatively long period .
the mass of the secondary ( mt91 202b ) approaches that of the primary if @xmath27=15 .
these rough mass constraints mean that the secondary spectral type probably lies in the range a to mid - b , assuming a main - sequence star .
although there are hints of line width variations that may suggest a double - lined system , the low snr of our spectra on this faint star coupled with the broad linewidth and small velocity amplitude means that no attempt was made to separate components .
figure [ sol202 ] shows the best - fitting orbital solution and folded velocity data . _
mt91 234_this v=13.25 system is dominated by its b1.5v ( revised from b2v in paper i ) primary star that exhibits long - term radial velocity variations .
the solution required a combination of data spanning the whole range of the cygnus ob2 radial velocity survey from 1999 october using keck+hires ( two measurements ) through 2001 august2008 june using wiyn+hydra ( 10 measurements ) , and 20112013 october using wiro ( 14 measurements ) .
figure [ sol234 ] shows the best - fitting orbital solution and folded velocity data .
symbol types designate data from keck ( triangles ) , wiyn ( diamonds ) , and wiro ( squares ) .
the period of [email protected] years makes this the longest period system yet uncovered in cyg ob2 .
the eccentricity is small ( [email protected] ) and consistent with zero .
a velocity amplitude of @[email protected] is large for such a long - period system and implies a fairly massive companion . adopting a primary mass of 12 yields a minimum secondary ( mt91 234b ) mass of 11 for @xmath37 .
this suggests that the system is seen nearly edge - on and that the secondary is also a b star . _
mt91 241_a b1.5v ( revised here from b2v in paper i ) primary star , the 11 wiro data combined with three keck and five wiyn measurements show that mt91 241a has a period of 671@xmath332 d ( 1.83 yr ) and semi - amplitude [email protected] with eccentricity of [email protected] .
figure [ sol241 ] shows the best - fitting orbital solution and folded velocity data . adopting a primary mass of 12 , the implied secondary mass ( mt91 241b ) is 5.2 for @xmath27=90 .
the very narrow lines show no sign of a second component , so it seems likely that the secondary star mass is closer to this lower limit than to the primary mass , implying a mass ratio in the range [email protected] . _
mt91 268_a b2v ( revised from b2.5v in paper i ) , this single - lined shows unambiguous evidence for being a triple system .
the initial single - component solution at a period near 33.2 d displayed systematic residuals with a period near 5.0 days .
we used these initial guesses to fit a joint orbital solution for two velocity components using two data from keck , five data from wiyn , and 15 data from wiro between 1999 and 2013 .
our best - fitting solution consists of the first component having @[email protected] d , @[email protected] , @[email protected] , and the second component having @[email protected] d , @[email protected] , and @[email protected] .
figures [ sol268a ] and [ sol268b ] display the orbital solution and folded velocity data for each component .
the existence of several outlier measurements in both plots is consistent with the hypothesis of this being a quadruple system . adopting 11 for the mass of the primary star mt91 268a ,
the implied minimum masses for the unseen companions are @xmath452.9 for the 33.2-day component 1 ( mt91 268b ) and @xmath460.7 for the 5.08-day component 2 ( mt91 268c ) .
this allows that the companions are probable a
mid - b and g
b dwarfs , respectively . given the moderate eccentricity , short period , and extreme mass ratio , this may be a dynamically unstable system . _
mt91 292_with a period of [email protected] d , this b2v system lies just above the 714 day period gap noticed by @xcite .
the velocity semi - amplitude of @xmath23= [email protected] is well - defined by the combination of 1 keck data point , six wiyn measurements , and 13 wiro measurements spanning nearly the entire length of the 20002013 survey period .
adopting a primary mass of 11 , an inclination of @xmath27=90 implies a minimal secondary mass for mt292b of @xmath29=1.6 .
the secondary mass approaches that of the primary if @xmath4712 .
figure [ sol292 ] shows the best - fitting orbital solution and folded velocity data . _
mt91 295_fourteen measurements from wiro during 2012 and 2013 indicate that this b1.5v ( revised from b2v in paper i ) has @[email protected] d and @[email protected] .
this is an unusually large eccentricity for such a short - period system , but solutions with zero eccentricity yielded substantially poorer fits .
the semi - amplitude of @[email protected] is small but detectable given the typical uncertainty of 3.9 .
the power spectrum has multiple peaks consistent with this being a triple system having components with periods of 4.67 d and 1.68 d and velocity semi - amplitudes near 46 .
the two - component solution provides a somewhat better fit , but the number of free parameters approaches the number of data points . in either case , this is a short - period multiple system .
figure [ sol295 ] shows the best - fitting _ single - component _ orbital solution and folded velocity data .
adopting 12 for the mass of the primary , the implied lower limit on the secondary ( mt91 295b ) mass is @xmath29=0.3 .
the inclination is unconstrained , as @xmath29 approaches @xmath28 for @xmath27=3 . _
mt91 336_this narrow - lined b2v ( revised slightly from b3iii in paper i ) has one observation from keck , seven from wiyn , and four from wiro spanning 1999 2014 .
the twelve data points yield a period of [email protected] with amplitude @[email protected] and eccentricity of @[email protected] .
there is a possible alias near 4.24 days .
figure [ sol336 ] shows the best - fitting orbital solution and folded velocity data . adopting 11 for the mass of the primary , the implied lower limit on the secondary ( mt91 336b ) mass is @xmath29=0.3 .
the inclination is unconstrained , as @xmath29 approaches @xmath28 for @xmath27=3 . _
mt91 339_with a velocity amplitude of just @[email protected] this o8v primary has one of the smallest amplitudes of any in our survey .
the narrow , deep he lines allow for a high level of precision , leading to typical velocity uncertainties of @xmath102 on the 34 wiro spectra spanning 20082013 .
the orbital period is @[email protected] d with eccentricity [email protected] .
a monte carlo simulation wherein the julian dates are shuffled randomly among the observed velocities shows that in only 9% of the iterations does the peak in the power spectrum exceed the observed one .
we conclude that the probability of a real periodicity is high .
figure [ sol339 ] shows the best - fitting orbital solution and folded velocity data .
the orbital inclination is unconstrained , allowing for secondary ( mt91 339b ) masses from 0.5 to 21 . _
mt91 378_the v=13.49 b0v
primary shows a velocity amplitude of @[email protected] and a period of [email protected] d. figure [ sol378 ] shows the best - fitting orbital solution and folded velocity data .
although there is some hint of variable line widths suggestive of a double - lined system , the limited snr of the spectra preclude attempts at spectral deblending using the 19 data obtained at wiro between 2010 and 2013 .
for an adopted primary mass of 18 , an inclination of @xmath4890 yields @xmath29=4.1 .
the secondary mass approaches the primary mass if @xmath27=17 .
hence , the secondary ( mt91 378b ) is limited to be a b spectral type , assuming it is on the main sequence . _
mt91 390_with a velocity amplitude of @[email protected] , this o8v primary is among the lowest - amplitude systems observed in the survey .
the period of @[email protected] d using 17 radial velocity measurements from wiro between 2011 june and 2013 august places mt91 390 among the conspicuous group of short - period binaries . there are no known eclipses , so the inclination is unconstrained . for @xmath27=90
the implied secondary mass is @xmath29=0.32 , while for @xmath27=2 the secondary ( mt91 390b ) mass approaches that of the primary .
figure [ sol390 ] shows the best - fitting orbital solution and folded velocity data . _
403_this b1v primary exhibits a velocity semi - amplitude of @[email protected] and a period of @[email protected] d using 18 measurements from wiro . with
no known eclipses , there are no inclination constraints .
the adopted primary mass of 14 leads to a secondary mass of m@xmath26=4.1 for @xmath27=90 and @xmath29=14 for @xmath27=23 .
these limits dictate that the secondary ( mt91 403b ) is a probable b star .
figure [ sol403 ] shows the best - fitting orbital solution and folded velocity curve . _
mt91 417b ( schulte # 22b)_mt91 417 is a visual double consisting of mt91 417a ( o3i ; the northwest component ) and mt91 417b ( o6v ; the southeast component ) at a separation of 15 ( @xmath102100 au projected separation at 1.4 kpc ) at pa@xmath250 .
we obtained 14 spectra of mt91 417b at wiro in good seeing conditions between 2013 august 6 and 2014 may 26 with the 12 slit oriented perpendicular to the two stars .
one additional measurement from wiro on 2008 june 26 was used .
mt91 417b exhibits variability with a period of @[email protected] d and @[email protected] .
figure [ sol417b ] shows the best - fitting orbital solution and folded velocity curve .
the implied secondary ( mt91 417b - b ) minimum mass is 1.8 for @xmath27=90 .
the secondary mass approaches that of the primary for @xmath27=5 .
the velocity curve is not well - sampled at all phases so the derived parameters are particularly uncertain . _
mt91 448_a period of @[email protected] d for this o6v primary places it among the shortest period systems in cyg ob2 .
the 18 wiro data spanning 20112013 yield a secure velocity semi - amplitude of [email protected] and a low eccentricity of @[email protected] .
there are no known eclipses .
an inclination of @xmath27=90 yields a minimum secondary ( mt91 448b ) mass @xmath29=2.1 , while the secondary mass approaches that of the primary for @xmath27=6 .
figure [ sol448 ] shows the best - fitting orbital solution and folded velocity curve data . _
mt91 473_this o8.5v has three measurements from keck ( 19992000 ) , nine from wiyn ( 20002008 ) and 24 from wiro ( 20102013 ) .
the power spectrum is complex , showing multiple peaks on timescales of 2000 d to 1.9 d. the dominant peak at 1700 d yields a folded velocity curve that appear consistent with the long - term trends observed in an unphased velocity curve .
the wiro data alone suggest a period in this same vicinity .
a monte carlo simulation wherein the julian dates are shuffled randomly among the observed velocities shows that in only 1.3% of the iterations does the peak in the power spectrum exceed the observed one .
we conclude that the observed periodicity is highly likely to be real .
our best - fitting orbital solution gives a period of @xmath4=1687@xmath3351 d and amplitude @[email protected] .
figure [ sol473 ] shows the best - fitting orbital solution and folded velocity data .
smaller peaks in the power spectrum suggest the possibility of short - period variations on the @xmath102 d timescale , and a couple of the keck hires spectra suggest a possibility of a double - lined system .
we fit single gaussian curves to the @xmath494471 lines in the keck and wiyn spectra and the @xmath495876 lines in the wiro spectra .
the fwhm of the lines varies over the range 3.7 4.4 with typical uncertainties of 0.25 . by applying a power spectrum analysis to the fwhm values
, we find a peak at 1.846 d , constituting evidence for a probable double - lined system with twice this period , 3.692 d. the spectra have insufficient snr and spectral resolution to attempt a deblending , but we conclude that mt91 473 likely consists of at least three stars . in this scenario
, components mt91 473a and mt91 473b would constitute a tight binary with a period of about 3.6 d and similar spectral types near o8.5v .
the inclination for this pair must be low , near 10 , such that the line splitting does not exceed the observed values .
mt91 473c is the unseen companion responsible for the radial velocity curve shown in figure [ sol473 ] .
adopting masses of 19 for each of components a and b implies a mass of 5.0 for @xmath27=90 , meaning that the secondary must be at least a mid - b star . for inclinations
as small as 11 the mass of component c could approach that of a+b . _
mt91 485 _
mt91 485 is an o8v showing long - term radial velocity variations with @xmath4=4066@xmath3345 d and semi - amplitude @[email protected] based on two keck measurements , nine wiyn measurements , and 17 wiro measurements covering the period 2001 august through 2013 october .
we found it necessary to fix the eccentricity during fitting to avoid extremely large values .
we estimate @[email protected] by manual experimentation .
figure [ sol485 ] shows the best - fitting orbital solution and folded velocity curve data .
the phase light curve is essentially the same as the unphased light curve , given that the period of 11.1 years is nearly as long as the duration of our survey .
adopting 21 for the primary mass , the inclination of @xmath27=90 leads to a secondary mass of m@xmath26=11.4 .
@xmath29 approaches @xmath28 for @xmath27=40 .
the secondary , mt91 485b , must be more massive than an early - b dwarf star .
mt91 485 is an example of a highly eccentric system that could easily be missed in a radial velocity survey if it were not for fortuitous phase coverage . _
mt91 555_one of the long - term radial velocity variables in the survey , mt91 555 ( o8v ) has a period @[email protected] d ( 3.50 years ) and semi - amplitude @[email protected] .
figure [ sol555 ] shows the best - fitting orbital solution and folded velocity data using 1 datum from keck , 7 from wiyn , and 20 from wiro . adopting a primary mass of 21 yields a minimum secondary mass @xmath29=10.3 , meaning that the secondary ( mt91 555b ) is constrained to be about b2v or earlier . for @xmath27=34
the secondary mass approaches that of the primary . _
mt91 561_a b2v , the primary star of mt91 561 shows a period of @[email protected] d and a velocity semi - amplitude of [email protected] using 17 wiro data points between 2007 and 2013 .
figure [ sol561 ] shows the best - fitting orbital solution and folded velocity curve . for an adopted primary mass of 11 ,
the secondary mass is @xmath29=3.3 if @xmath27=90 .
@xmath29 approaches @xmath28 when @xmath27=23 .
this constrains the secondary ( mt91 561b ) to be a least as massive as a late b main sequence star , and the mass ratio lies in the range @xmath30=0.301 . _
mt91 588_this b0v has a moderately eccentric orbit with @[email protected] d , @[email protected] , and a velocity semi - amplitude of [email protected] using one measurement from keck , four from wiyn , and 16 from wiro over the period 19992013 .
figure [ sol588 ] shows the best - fitting orbital solution and folded velocity data . adopting a primary mass of 18 yields a minimum secondary mass @xmath29=3.2 , meaning that the secondary ( mt91 588b ) is constrained to be an early a star or earlier . for @xmath27=15
the secondary mass approaches that of the primary .
the sampling of the velocity curve is such that aliases of 34.9 d , 69.8 d , 104 d , 174 d are possible but less likely . _
mt91 601_the two keck , one wiyn and 28 wiro data of this o9.5iii reveal a period of [email protected] d and @[email protected] .
owing to the incomplete phase coverage , we found it necessary to fix the eccentricity ( at 0.67 ) to avoid extremely large eccentricities that tended to result during the fitting process .
manual experimentation suggests an eccentricity uncertainty of @xmath100.2 .
figure [ sol601 ] shows the best - fitting orbital solution and folded velocity curve . adopting a primary mass of 21 yields a minimum secondary mass @xmath29=4.1 , meaning that the secondary ( mt91 601b ) is likely a b or o main sequence star .
the reduced @xmath32 of the fit is unusually large , at 3.4 .
we examined the o - c residuals for signs of periodicity , but no strong peaks were seen .
we conclude that this evolved primary star exhibits irregular photospheric variations in addition to the identified periodic modulation . _
mt91 646_classified as b1.5v in both @xcite and @xcite , spectra of this system reveal a variable line width indicating an sb2 .
in several of our 19 wiro spectra the lines are sufficiently separated to see that the line depths and widths are similar , having a fwhm near 2.9 .
we fit the most deblended of the 19 wiro spectra with two - component fixed - width gaussian profiles to measure velocities for each component .
the solution yields a period of @[email protected] d with velocity semi - amplitudes @[email protected] and @[email protected] .
this implies a mass ratio of @[email protected] .
figure [ sol646 ] shows the best - fitting double - lined orbital solution .
the mass function implies a lower limit on the system mass of 12.5 , consistent with a b1v and b1.5v seen at a high inclination . as such ,
mt91 646 is one of the most `` twinlike '' systems among the cyg ob2 sample . _
mt91 745_this o7v has two observations from keck , eight from wiyn , and 22 from wiro over 19992013 , yielding a period @[email protected] d , semi - amplitude @[email protected] , and eccentricity @[email protected] .
figure [ sol745 ] shows the best - fitting orbital solution and folded velocity curve .
adopting a primary mass of 25 , the implied lower limit to the secondary mass is @xmath29=4.0 for @xmath27=90 .
@xmath29 approaches @xmath28 for @xmath27=15 .
thus , the secondary ( mt91 745b ) is at least the mass of a b dwarf star .
@xcite noted that some of their o - star sample exhibited irregular rather than periodic velocity variations .
this phenomenon is quite common in massive stars , especially evolved stars , and has been attributed to line profile variations caused by non - radial pulsations or clumps in the stellar winds @xcite .
we note here six objects that have more than 18 data measurements where significant velocity variations at the level of 2530 are observed , but no strong periodicity is evident .
table [ random.tab ] lists the heliocentric julian dates , velocities , and velocity uncertainties of each measurement . _
mt91 138_the two keck data , eight wiyn data , and 22 wiro data between 1999 and 2008 show that this o8i displays a low level of irregular variability . a long - term
, eccentric binary can not be ruled out . _
mt91 417a ( schulte # 22a_this o3if shows variations at the level of about 25 on the basis of 20 wiro data from 2008 and 2014 . our lack of long - term data prevents us from ruling out the possibility of longer term orbital variations .
we can only say that the observed variations appear random on the basis of the data presented in table [ random.tab ] .
because of the very weak lines in this very hot star , velocities are measured from @xmath495411 assuming a rest wavelength of 5411.45 .
should not be present in such a hot star , but light from the psf wings of mt91 417b may contaminate some of our spectra given the close 2 separation . _
mt91 457 _ one keck , four wiyn , and 11 wiro data on this o3if during 19992008 show that this star exhibits non - periodic variations at the level of @xmath1025 . _
mt91 483_this o5i / iii shows low - amplitude ( 15 ) variations but exhibits power on many scales from days to months using three data from keck , three from wiyn , and 25 data from wiro over 1999 - 2013 .
we show , using a monte carlo analysis which randomly shuffles the dates of observation among the observed velocities , that the peaks near 124 days and five days have at least a 0.32 probability of occurring by chance . because of the weak lines in this very hot star
, velocities are measured from @xmath495411 assuming a rest wavelength of 5411.45 . _
mt91 556_mt91 556 was observed 40 times between 2008 and 2013 at wiro with an average velocity uncertainty of 4 and variations of up to 20 .
although velocity variations are observed in this b1i , there is no dominant periodicity and they are consistent with random atmospheric fluctuations . _
mt91 632_mt691 32 was observed 30 times between 2008 and 2011 at wiro with an average velocity uncertainty of 3 .
although velocity variations are observed in this o9i at the level of 25 , there is no dominant periodicity and they are consistent with random atmospheric fluctuations .
we report here 16 stars having at least 12 velocity measurements spanning at least five years that show no evidence for velocity variations .
these stars are candidates for single stars , stars with very low - mass companions , or systems seen at very low inclination angles . among this list
there may be undetected binaries with long periods and/or highly eccentric orbits .
additionally , systems with periods near multiples of one year undergoing periastron during december april when cygnus is minimally observable have a low probability of being detected in this survey . for each non - variable system
we report the mean velocity , the rms velocity dispersion , and the mean velocity uncertainty . _
mt91 005_this o6v was observed twice at keck , three times at wiyn , and ten times with wiro over the period 19992013 .
the radial velocities have a mean systemic velocity of @xmath505.5 , an rms of 2.1 and mean uncertainties of 2.5 .
the very narrow lines ( 1.9 fwhm ) suggest that the rotational axis may be nearly parallel to the line of sight . _
mt91 020_twelve measurements at wiro , four from keck , and two from wiyn spanning 19992011 show no evidence for variability , with a mean velocity of @xmath503.2 , an rms of 4.3 , and mean uncertainties of 4.2 . _
mt91 083_with 22 measurements at three observatories spanning 19992008 , this b1i star shows no evidence for variability .
the mean systemic velocity is @xmath501.3 , with an rms of 1.7 and mean uncertainties of 2.4 in the nine wiro data , one keck datum , and 7 wiyn data . _
mt91 217_having two measurements from keck , seven from wiyn , and 17 from wiro spanning 19992011 , this o9v star shows no evidence for variability .
the mean systemic velocity is @xmath5011.1 , with an rms of 2.4 and mean uncertainties of 3.0 . _
mt91 227_two data from keck , six from wiyn , and 26 from wiro over the period 19992011 , are remarkably constant for this o9v .
the mean velocity is @xmath503.8 with an rms of 5.5 and mean uncertainties of 5.1 . _
mt91 259_one keck , six wiyn , and seven wiro data over the period 19992011 show a mean velocity of @xmath5014.0 with an rms of 1.3 and mean uncertainties of 2.1 . _
mt91 299_this o7v has two data from keck , six from wiyn , and 37 from wiro between 1999 and 2013 .
the mean velocity is @xmath5014.4 with an rms of 6.0 and a typical velocity uncertainty of 5.5 .
this object shows some evidence for a periodicity near 405 days , but the amplitude is low compared to the observational uncertainties , and a convincing orbital solution was not obtained . _
mt91 317_this o8v has two data from keck and 11 from wiyn between 1999 and 2007 .
the mean velocity is @xmath504.2 with an rms of 4.4 and a typical velocity uncertainty of 5.5 .
this object shows some evidence for a periodicity near 405 days , but the amplitude is low compared to the observational uncertainties , and a convincing orbital solution was not obtained . _
mt91 376_on the basis of two keck and 10 wiyn data over the period 19992006 this o8v shows a mean velocity of @xmath5023.7 and rms of 6.2 for a mean velocity uncertainty of 7.5 . _
mt91 455 _ this o8v has two data from keck , 3 from wiyn , and 20 from wiro between 1999 and 2013 .
the mean velocity is -12.2 with and rms of 3.3 and a mean velocity uncertainty of 3.6 . _
mt91 462 _ this o7iii has 12 data from from wiro between 2008 and 2014 .
the mean velocity is -11.9 with and rms of 2.8 and a mean velocity uncertainty of 4.1 . _
mt91 470_two measurements from keck , five from wiyn , and nine from wiro over the the period 19992013 show no evidence for velocity variations beyond the uncertainties .
the mean velocity is @xmath5021.3 with an rms of 6.9 and a mean velocity uncertainty of 5.2 . _
mt91 480_we measured both and velocities for this broad - lined o7v using 4 wiyn and 14 wiro data between 2001 and 2013 , finding no compelling evidence for velocity variations .
the line profiles are broad and appear to vary without evidence of periodicity . the mean velocity is @xmath5016.2 with an rms of 13.3 and a mean velocity uncertainty of 11.7 . _
mt91 507_two keck , two wiyn , and 14 wiro data between 1999 and 2013 yield a mean velocity of @xmath508.9 with an rms of 5.1 and a mean velocity uncertainty of 6.5 for this o9v star . _
mt91 534_three keck , two wiyn , and 14 wiro data between 1999 and 2013 yield a mean velocity of @xmath505.0 with an rms of 3.8 and a mean velocity uncertainty of 3.1 for this o8.5v star . _
mt91 611_two keck and 14 wiyn data between 1999 and 2008 yield a mean velocity of @xmath5023.1 with an rms of 3.7 and a mean velocity uncertainty of 3.9 for this o7v .
the cygnus ob2 radial velocity survey was designed to produce a complete census of massive binaries drawn from a photometrically selected sample , namely that of @xcite .
the parent sample contained 150 stars
the 146 listed in table 1 of paper i plus mt91 267 ( inadvertently excluded initially but reported as a binary in paper vi ) , mt91 417b ( not initially recognized as a close companion to mt91 417a ) , schulte # 3 and schulte # 73 . of these 150 ,
six were found to be probable emission - line objects ( mt91 186 , 213 , 488 , 522 , 650 , 793 ) and excluded from further observation because radial velocities were considered unreliable .
another 16 objects were determined to have spectral types later than b3 and excluded from further observation ( mt91 170 , 189 , 196 , 222 , 239 , 271 , 273 , 427 , 444 , 453 , 459 , 493 , 539 , 554 , 576 , 641 ) .
this leaves 128 objects in the `` unbiased '' survey sample that we deem to be the complete photometrically selected targets . in the above sections we identified 6 systems ( 4.6% ) , all supergiants , as exhibiting irregular velocity variation and 16 objects ( 12.5% ) as consistent with having no intrinsic velocity variability .
forty - five objects of the 128 in the complete unbiased sample ( 35.1% ) have orbital solutions presented either in this work or previous papers in this series .
this leaves 63 systems ( about half ) that have insufficient data to make a reliable determination . the vast majority of these are b1b2.5 primaries , although a few o stars remain in this category because there are fewer than 12 spectra of suitable quality .
the average number of spectra among the 62 `` indeterminate '' systems is five .
none have evidence for significant velocity variations .
we expect that most of these will turn out to be constant - velocity objects , but we also anticipate discovery of a handful of new binaries in future observing seasons .
we expect most of these to be long - period and/or high - eccentricity systems that have eluded detection because of limited data .
nevertheless , the completeness analysis in section 3.5 leads us to conclude that the vast majority of the detectable binaries in cyg ob2 now have complete orbital solutions , as published herein .
we estimated the completeness of the survey as a function of orbital period using a monte carlo analysis identical to that described in @xcite .
our code generates populations of binary systems with periods 1 5000 days having mass ratios and eccentricities described by power laws : prob(@xmath30 ) @xmath51 @xmath30@xmath52 , and prob(@xmath20 ) @xmath51 @xmath20@xmath53 .
the nominal power - law exponents are @xmath54 and @xmath55 , as inferred , in part , by @xcite and strengthened by the additional data in this work .
a random inclination is assigned to each system before being sampled at the actual dates and times of the survey observations .
a nominal detection threshold of 7 is used to determine whether the system would be observed as a binary in the survey .
this is lower than the more conservative value of 15 used in @xcite , but it better represents the sensitivity of the data to radial velocity variability given the typical resolutions of @xmath564500 ( 60 fwhm ) achieved in the majority of wiro spectra .
velocity precisions of better than 0.1 of the spectral fwhm are typically attained by centroiding on the strong @xmath495876 feature .
precisions are 815 in some objects where rotation broadens the line profile . upon careful examination ,
the detection threshold is really a complex function of the signal - to - noise ratio of the spectra , the spectral type of the star , and the breadth of the spectral features ( i.e. , rotational velocity ) ; nevertheless , the adopted value of 7 is generally applicable as a survey average , while 15 is a conservative value .
figure [ complete ] displays the completeness as a function of orbital period , where completeness is defined as the ratio of detected binaries to all binaries of the same period in the synthesized population .
different line styles depict different values for the power law exponent @xmath12 and different choices for the detection velocity threshold .
completeness exceeds 80% for systems with @xmath1100 days in the case of a 7 detection threshold and is only @xmath1010% lower for the conservative threshold of 15 . at @xmath4=1000 d
we find that 4060% of binaries are still detected .
the smooth decline in completeness with orbital period reflects the difficulty in detecting wide , long - period systems . while the completeness estimates are only as valid as the input power - law assumptions for the mass ratio and period distributions , the conclusion that the vast majority of systems with periods less than a few hundred days have been detected is a robust one .
completeness estimates beyond about 2000 d should be regarded as highly uncertain .
table [ bigtable.tab ] summarizes the spectroscopic binary type ( sb1/sb2 ) , spectral types for the components , orbital period , primary semi - amplitude , eccentricity , and mass ratio constraints for all 48 massive multiple systems now known in cygnus ob2 .
twenty three are new results from this work , twenty are reported in previous papers in this series , and five stem from other published works .
the final two columns provide references to published orbital solutions and additional notes on each system .
the table contains six stars not initially included in table 1 of @xcite among the unbiased list survey targets .
mt91 267 and mt91 417b , explained in section 3.4 constitute two of these .
the double - lined system cpr2002 b17 was reported and analyzed by @xcite .
schulte # 5 has been known as a binary since @xcite .
cpr2002 a36 and a45 were added later after being reported as suspected binaries . in summary ,
schulte # 5 , cpr2002 a36 , cpr2002 a45 , and cpr2002 b17 should not be included in the `` unbiased '' sample used to estimate binary fractions because they were targeted only after evidence of their binary nature became known .
the compilation of 48 targets in table [ bigtable.tab ] includes 16 sb2 and 32 sb1 systems .
fourteen contain at least one evolved component .
eight are probable triple systems based on either radial velocity data or high - angular - resolution imaging ( see discussion in section 3 ) .
eight are known to show eclipses .
figure [ peq ] plots eccentricity ( top panel ) and mass ratio ( lower panel ) versus orbital period for all 48 massive binaries . filled circles mark sb2 systems and open circles denote sb1 systems from cyg ob2 .
pluses show the o - star binaries from @xcite .
error bars in the lower panel illustrate the range of mass ratios allowed by the available constraints on orbital inclination .
statistically , systems will preferentially lie toward the lower end of this range since it is more likely to observe the orbital plane edge - on than face - on .
the overall trend in the top panel has been noted elsewhere .
the shortest - period systems have low eccentricities while longer - period systems have larger eccentricities .
any trend in the lower panel is less obvious .
mass ratios span the the range from near 1.0 for some of the shortest - period systems to about 0.1 , the approximate lower limit detectable by the velocity precision of the survey .
long - period systems generally have weak constraints on the eccentricity and mass ratio as a result of incomplete phase coverage .
the two samples plotted here are affected by different target selection , sampling , and sensitivity biases , so appropriate caution should be exercised in attempts to statistically compare them .
nevertheless , the distribution of points among the two samples is similar .
figure [ edist ] shows the observed distribution of orbital eccentricities for the full sample of 48 cyg ob2 binaries .
the left - hand y - axis labels and the plotted points shows the cumulative fraction .
the right - hand y - axis and plotted histogram shows the number of objects in each eccentricity bin .
the distribution is approximately uniform ( prob(@xmath20)@xmath57 ) between @xmath58 and @xmath590.6 .
a two - sided kolmogorov - smirnov ( k - s ) test indicates that the eccentricities have a 75% chance of being drawn from the a uniform distribution between the limits @xmath8 .
we also employed the anderson - darling ( a - d ) two - sample statistic which is more sensitive to differences between two populations when differences in the the cumulative distributions are both positive and negative across the sample range , or when these differences occur near the ends of the distributions .
the a - d statistic concurs that the probability of the eccentricities being drawn from a uniform distribution between @xmath8 is 75% .
however , over the full eccentricity range there is an overabundance of @xmath60 systems compared to a uniform distribution .
the paucity of points at high eccentricity may partly be an observational bias against detecting highly eccentric systems , and it may partly reflect a real dearth of such loosely bound , easily disrupted systems .
our completeness analysis above assumes two different values , 0 and @xmath500.6 ( the nominal value from * ? ? ?
* ) , for @xmath12 , the power law exponent describing the eccentricity distribution .
completeness is shown to be insensitive to @xmath12 .
nevertheless , a few long - period , high - eccentricity systems have likely evaded detection with the current dataset .
unless there are an disproportionate number of high-@xmath20 systems in cyg ob2 , the @xmath20 distribution can be described as approximately uniform out to periods of @xmath101000 days .
the distribution of mass ratios among our sample can be recovered , statistically , using monte carlo methods , by assigning randomly chosen inclinations on the unit sphere . in most cases
an upper limit on the inclination is estimated by either the observed eclipse profile or the lack of observed eclipses ; for long - period systems the upper limit on the inclination is effectively 90 .
a lower limit on the inclination is obtained by letting the mass of the secondary approach that of the primary ; the lack of observed spectral features from a secondary star is a weak constraint on this lower bound . for each system
we use the measured orbital period , adopted primary mass , observed velocity amplitude , and a random inclination between the observationally imposed limits to solve for the mass of the secondary star .
the mean value of @xmath30 is then computed after 1000 such iterations for each system .
figure [ qdist ] shows the resulting histogram of mass ratios .
the overall distribution is approximately uniform .
however , incompleteness appreciably affects the bins at @xmath61 , so the true number of binaries in these bins is highly uncertain .
given the large uncertainties on any individual mass ratio value , we elect not to pursue a statistical analysis of these data . in principle
we could estimate the completeness factor at @xmath62 given assumptions about the functional form of the underlying secondary mass distribution and period distribution ( e.g. , by adopting a power law distribution , following @xcite ) .
however , as we will show below , the true distribution of orbital parameters , in particular the periods , may not be a single power law .
figure [ pcum1 ] displays the observed cumulative distribution of orbital periods ( in @xmath63 @xmath4 ) for the full sample of 48 massive ( b3 and earlier ) systems in cyg ob2 ( filled circles ) , o stars from six galactic open clusters ( * ? ? ?
* plusses ) , and bright o stars drawn from the general galactic population from ( * ? ? ?
* open squares ) .
fiducial marks near the top of the plot indicate the period in days .
other annotations mark the corresponding semi - major axis for the given period , assuming a total system mass of 30 .
the @xcite sample has been normalized by a factor of 1.5 to allow better comparison with the other two surveys in the short - period limit where all three are highly complete .
both the @xcite sample and the cyg ob2 sample are reasonably complete to periods of several hundred days , while the @xcite sample lacks such longer - period systems a consequence of the limited observing campaign .
a constant slope in this plot corresponds to a uniform distribution in @xmath63 @xmath4 .
a steeper slope means a larger number of detected binaries per @xmath63 @xmath4 interval than a shallower slope .
the observed cumulative period distributions for the three samples appear remarkably similar at short periods , rising rapidly with similar slope from a short - period limit of about 1.4 days to near 7 days .
an a - d test shows that the cyg ob2 and @xcite observed distributions have a 91% chance of being drawn from the same parent populations in this range , increasing to 97% if the complete instead of the unbiased cyg ob2 sample is used .
all three samples show an abundance of short - period binaries , noted previously by @xcite .
all three samples also show a flattening of the slope beginning at about 6 days .
the cyg ob2 sample exhibits the most pronounced flattening between 6 and 14 days , indicating a paucity of systems in this range , as noted previously by @xcite .
there is then an upturn between 14 and @xmath1045 days , with a slope approximately matching that of the very short - period systems .
the @xcite and @xcite samples , by contrast , show lesser degrees of flattening longward of six days , with the former being steeper than the latter .
the limited number of data points precludes any strong statement regarding the similarity of the three samples in this restricted range , or the reality of the change in slope near six days .
both the cyg ob2 and the @xcite samples flatten or exhibit a break near 45 d. the @xcite sample has become considerably incomplete at these periods and is not considered further .
both samples exhibit similar slopes out to periods of several thousand days where both become significantly incomplete . @xcite and
@xcite considered the possibility that period distribution could be characterized by a double - pik ( i.e. , uniform in @xmath64 ; * ? ? ?
* ) distribution with a break near 10 days , however they found that such a description did not provide a better fit to the data .
corrections for observational bias ( i.e. , incompleteness ) become important at periods greater than several hundred days , and we address these in the context of the cyg ob2 sample below . a two - sample anderson - darling test shows that the cyg ob2 and the @xcite observed cumulative distributions are individually consistent with uniform ( in @xmath63 p given by @xmath65 ) between 1.4 and 45 days at the 42% and 45% levels , respectively ; the cyg ob2 and @xcite samples becomes consistent with uniform at the 86% and 62% level for periods @xmath3425 days .
the cyg ob2 and @xcite samples are consistent with each other at the 91% levels for @xmath1 25 d but only 19% for @xmath1 45 d. the cyg ob2 and the @xcite samples are consistent with each other at the 40% and 5% levels for upper period limits of 25 d and 45 d , respectively .
taken together , these statistics indicate that the period distributions in the cyg ob2 and @xcite surveys are probably consistent with each other @xmath66 d , but not at @xmath67 d and are not consistent with a uniform distribution even at the shortest periods . when all three surveys are combined , the probability that the combined sample is consistent with uniform drops to 14% and 21% for upper period limits of 25 d and 45 d , respectively .
we interpret this to be evidence for structure in the period distribution that becomes apparent with sufficiently large numbers .
thus , the period distribution does not appear to be scale - free ; it is suggestive of features imposed by physical phenomena occurring during the formation and/or evolution of massive systems .
the most obvious feature in the period distribution at @xmath68 d is the flattening of the slope near 6 days seen in all three datasets .
@xcite characterized this structure as a possible excess of 36 d systems accompanied by a deficit of 714 d systems , observed here as a flattening of the slope of the cumulative distribution over that range in all three samples .
this possible feature seen in the cyg ob2 distribution is not as obvious in the other two data sets .
incompleteness in @xcite sample , though not explicitly quantified in that work , probably begins in the 1530 d range and could easily mask such a signal if it were present .
by contrast , the completeness of the cyg ob2 survey is @xmath1190% for @xmath4@xmath3410 d. the reality of such a signal in the @xcite sample is uncertain owing to the small number of points between 6 d and 45 d ( 10 systems ) , even though the sample is likely to be similarly complete .
we conclude that the cyg ob2 period data is not convincingly consistent with uniform ( a single power law of slope zero ) at @xmath4@xmath3445 d. there is evidence for structure in the form of a 36 d period excess and the 614 d period `` deficit '' that becomes more pronounced when the three samples are combined . in the period range
60 4000 d the cyg ob2 and @xcite samples , as observed , are consistent with uniform at the 94% and 55% levels .
the two samples are consistent with each other at the 99% level . by contrast , over the full period range 1.4 5000 days , both samples are inconsistent with uniform , having a @xmath340.01% chance of being drawn from such a distribution .
the single @xmath65 power law is a poor description of the observed period distribution over the full range , but observational biases against detection of long - period systems must be considered before strong conclusions may be drawn .
figure [ pcum2 ] ( lower panel ) shows the observed cumulative period distribution for the unbiased sample of 45 systems as observed ( filled symbols ) and after correction for completeness ( open symbols ) as described by completeness curve in figure [ complete ] using the solid line representing the 15 ( most conservative ) detection threshold .
the y - axis is now scaled to indicate the fraction of the total number of stars in the unbiased cyg ob2 sample .
the correction from observed to underlying period distribution is performed by tabulating the cumulative incompleteness ( 1-completeness ) at the location of each observed system , n , starting from the shortest periods , and inserting an additional `` system '' when the cumulative incompleteness reaches an integer value ; systems are added at periods halfway between the period of system n and n-1 .
this can result in the appearance of two or more systems plotted at the same orbital period when the completeness is low ( i.e. , for long periods ) .
open symbols in figure [ pcum2 ] trace a curve similar to the filled symbols until the correction for completeness becomes a large factor beyond about 2000 days .
the difference between the filled and open symbols is small for periods less than a few hundred days but approaches 20% at the 5000-day limit of the survey .
the open symbols display an apparent change in slope near 45 days , indicating that the putative break in the cumulative distribution is not likely to be caused by observational biases . under the adopted prescriptions for the mass ratio distribution ( i.e. , uniform ) and the eccentricity distribution ( i.e. , uniform ) adopted in the monte carlo modeling
, this figure shows that the binary fraction reaches 30% by @xmath4=45 d. fifty - five percent of the systems are binaries with periods less than 5000 days .
this is nearly identical to the binary fraction of 51% computed for o stars in the large magellanic cloud @xcite and may indicate that the binarity characteristics of massive stars are insensitive to metallicity .
an extrapolation of the quasi - linear trend defined by systems with @xmath5100 days would suggest a binary fraction near 70% for systems with periods less than @xmath69 days . on the other hand , a naive extrapolation of the much steeper slope at @xmath145 days would reach a binary fraction of 100% near @xmath69 days .
we used the a - d test to quantify the probability of a change in slope of the completeness - corrected cumulative period distribution by comparing it to a hypothetical uniform ( i.e. , linear ) distribution in a moving window of encompassing seven observed systems .
window widths of five to nine do not appreciably change the results .
figure [ pcum2 ] ( upper panel ) plots the probability that the completeness - corrected period distribution is consistent with uniform .
the probability drops below 1% ( 2.6@xmath70 , shown by the dotted horizontal line ) in the vicinity of the hypothesized 45-day break .
the low probability near six days and 14 days supports the hypothesis of a change slope near these locations as well .
the similarity of the cyg ob2 and the @xcite observed cumulative distribution slopes above 45 days in figure [ pcum1 ] is noteworthy as is the apparent flattening near 45 days ( @xmath20.9 a.u ) .
the shallower slope above 45 d may indicate either a decreasing binary fraction for systems with semi major axes of @xmath111 a.u .
alternatively , it may indicate a change in the distribution of secondary star masses toward a preference for lower masses .
we interpret the evidence for structure in the cumulative period distribution as an indication that the processes responsible for setting massive binary separations are not scale - free .
nevertheless , we investigated whether a single power law might at least approximate the data .
the solid curve in the lower panel of figure [ pcum2 ] is a best - fit single @xmath71 power law to the completeness - corrected data over the range 1.42000 d. the completeness correction is highly uncertain beyond 2000 d , so we do not consider this regime in the fit .
an a - d test indicates a 24% probability that the data is drawn from such a single power - law distribution .
this probability does not constitute strong evidence in favor of this particular power law , especially given that the results are sensitive to the magnitude of the completeness corrections required for long - period systems . nevertheless , the @xmath3 power law is not an unreasonable _ approximation _ over 3.3 orders of magnitude in period ( 1.4 days to 2000 days ) , even if it does not describe the fine details of the distribution at a statistically compelling level of significance .
we conclude by noting that the agreement between the observed orbital period distributions of the cyg ob2 sample , where ob stars formed across an extent of nearly 50 pc , compared to the @xcite sample from several young galactic clusters , suggests that the large - scale physical environment does not determine the orbital parameters of massive binaries formed therein .
this stands in contrast to suggestions that the binary fraction depends on local stellar density , being lower in the densest regions @xcite .
rather , the small - scale physics of cloud collapse and fragmentation on the sub - a.u . scale appear more important in establishing the high fraction of binaries among massive stars and their mass ratio distribution which clearly shows that massive stars preferentially have massive companions .
the results of the massive star surveys stand in stark contrast to those from radial velocity surveys of solar - type stars where the distribution of periods is log - normal with a mean of 180 years @xcite .
the inflections in the cumulative period distribution near 0.2 a.u . and near 0.9 a.u .
are suggestive of key physical size scales beyond which the formation of close massive binaries becomes less efficient .
the addition of 23 new massive binary orbital solutions presented herein raises the total number known in the cygnus ob2 association to 48 .
this is the largest collection of complete solutions obtained within a single environment and constitutes a substantial fraction of all early type orbits published to date .
at least 35% of massive systems in cyg ob2 are binary or higher - order multiples , with several more long - period or highly eccentric systems expected to be discovered .
the number of unobserved high - inclination or extreme - mass - ratio systems has been modeled statistically .
these computations suggests that the true binary fractions are near 55% for periods
@xmath725000 d , nearly identical to the results from o stars in the lmc , suggesting a degree of metallicity independence .
extrapolation to longer periods suggests binary fractions in the range 6580% for periods @xmath73 d. the distribution of orbital periods is described as approximately uniform in @xmath0 between the short - period limit of 1.2 d and @xmath245 d , but there is evidence for structure in this distribution , with an excess of 36 d systems and a possible dearth of 714 d systems , along with a break near 45 days . between 45 d and the long - period survey limit of @xmath105000 d
the distribution is also approximately uniform , but with either a lower binary fraction or a population of secondary masses favoring lower - mass companions compared to the 145 d subsample .
no single power - law distribution describes the period data at a statistically significant level .
a single power law of slope @xmath71 also provides a rough description of the cumulative period distribution between 1.4 and @xmath102000 d if structure in the distribution is ignored .
the statistical power of even the present completeness - corrected sample has not yielded a convincing power - law or other analytic description of the present - day binary period distribution .
it seems likely that evolutionary effects such as period migration and stellar mergers have altered the primordial distributions to the ones observed as the present - day distributions in the 34 myr old cygnus ob2 association and other young o - star cluster .
such evolutionary effects may occur during the pre - main - sequences phases , rendering any simple primordial distribution unobservable even in principle .
one goal of this work is to provide a robust measure of massive binary parameters to help inform stellar population synthesis models , including those used to predict cosmic rates of high - energy phenomena such as supernova and @xmath12-ray bursts .
it is our hope that such models may now be grounded more solidly in the data .
this compilation may also prove helpful in guiding models of massive star formation and evolution by providing observational constraints on the binary frequency and distribution of periods and mass ratios among early - type systems .
we expect that there are several more discoverable binary ( and multiple ) systems among the survey targets , but this present compilation likely contains the vast majority of systems discoverable using radial velocity techniques .
insightful comments from an expert anonymous referee greatly improved this manuscript .
we thank our long - term collaborator chris fryer for continued encouragement , in particular for a timely conversation after a colloquium at uc santa cruz in 1998 that launched the cygnus ob2 radial velocity survey .
we acknowledge continued support from the national science foundation through research experience for undergraduates ( reu ) program grant ast 10 - 63146 , through grant ast 03 - 07778 , and grant ast 09 - 08239 . the wyoming nasa space grant consortium provided student support through grant # nnx10a095h .
we thank tyler ellis , sarah eftekharzadeh , ian ewing , and garrett long for assistance with observing at wiro .
we are grateful to wiro staff james weger and jerry bucher whose diligent support work enabled this program to obtain spectra on more than 290 nights at the wyoming infrared observatory 2.3 m telescope .
generous allocations of telescope time at the lick and keck telescopes through the university of california and at the wiyn 3.5 m telescope through noao and the university of wisconsin made possible such a large observational project . some of the data presented herein were obtained at the w.m .
keck observatory , which is operated as a scientific partnership among the california institute of technology , the university of california and the national aeronautics and space administration .
the observatory was made possible by the generous financial support of the w.m .
keck foundation .
the wiyn observatory is a joint facility of the university of wisconsin - madison , indiana university , yale university , and the national optical astronomy observatory .
_ facilities : _
, , blaauw , a. 1961 , ban , 15 , 265 chini , r. , hoffmeister , v. h. , nasseri , a. , stahl , o. , & zinnecker , h. 2012 , , 424 , 1925 comern , f. , pasquali , a. , rodighiero , g. , et al .
2002 , , 389 , 874 contreras , m. e. , rodriguez , l. f. , tapia , m. , cardini , d. , emanuele , a. , badiali , m. , & persi , p. 1997 , apj , 488 , 153 de becker , m. , rauw , g. , & manfroid , j. 2004 , a&a , 424 , l39 dominik , m. , belczynski , k. , fryer , c. , holz , d. e. , berti , e. , bulik , t. , mandel , i. , & oshaughnessy , r. 2013 , arxiv:1308.1546 drilling , j. s. , & landolt , a. u. 2000 , in astrophysical quantities , ed .
a. n. cox ( 4th ed .
; new york ; springer ) , 381 duquennoy , a. , & mayor , m. 1991 , , 248 , 485 eldridge , j. j. , fraser , m. , smartt , s. j. , maund , j. r. , & crockett , r. m. 2013 , , 436 , 774 feigelson , e. d. & babu , g. j. , 2012 , `` modern statistical methods for astronomy with r applications '' , cambridge univ press ( chpt 3 ) fullerton , a. w. , gies , d. r. , & bolton , c. t. 1996 , , 103 , 475 fryer , c. l. , mazzali , p. a. , prochaska , j. , et al .
2007 , , 119 , 1211 garmany , c. d. , conti , p. s. , & massey , p. 1980
, , 242 , 1063 gies , d. r. , & bolton , c. t. 1986 , apjs , 61 , 419 gonzlez , j. f. , & levato , h. 2006 , a&a , 448 , 283 hall , d. s. 1974 , aca , 24 , 69 hanson , m. m. 2003 , apj , 597 , 957 hoogerwerf , r. , de bruijne , j. h. j. , & de zeeuw , p. t. 2001 , , 365 , 49 hora , j. , adams , j. , allen , l. , et al .
2007 , spitzer proposal , 40184 hubeny , i. , & lanz , t. 1995 , , 439 , 875 hunter , i. , lennon , d. j. , dufton , p. l. , et al .
2008 , , 479 , 541 kiminki , d. c. 2010 , ph.d .
thesis , university of wyoming kiminki , d. c. , et al .
2007 , apj , 664 , 1120 ( paper i ) kiminki , d. c. , mcswain , m. v. , & kobulnicky , h. a. 2008 , apj , 679 , 1478 ( paper ii ) kiminki , d. c. , kobulnicky , h. a. , gilbert , i. , bird , s. , chunev , g. 2009 , aj , 137 , 4608 ( paper iii ) kiminki , d. c. , kobulnicky , h. a. , ewing , i. , et al .
2012 , , 747 , 41 ( paper iv ) kiminki , d. c. , & kobulnicky , h. a. 2012 , apj , 751 , 4 ( paper v ) kobulnicky , h. a. , & fryer , c. l. 2007 , , 670 , 747 kobulnicky , h. a. , smullen , r. a. , kiminki , d. c. , et al .
2012 , , 756 , 50 ( paper vi ) kobulnicky , h. a. , kiminki , d. c. , et al . 2014 , , in prep kouwenhoven , m. b. n. , brown , a. g. a. , portegies zwart , s. f. , & kaper , l. 2007 , , 474 , 77 lanz , t. , & hubeny , i. 2003 , apjs , 146 , 417 markwardt , c. b. 2009 , astronomical data analysis software and systems xviii , 411 , 251 martins , f. , schaerer , d. , & hillier , d. j. 2005 , , 436 , 1049 mason , b. d. , hartkopf , w. i. , gies , d. r. , henry , t. j. , & helsel , j. w. 2009 , , 137 , 3358 massey , p. , & thompson , a. b. 1991 , aj , 101 , 1408 miczaika , g. r. 1953 , pasp , 65 , 141 naz , y. , de becker , m. , rauw , g. , & barbieri , c. 2008 , a&a , 483 , 543 naz , y. , mahy , l. , damerdji , y. , et al . 2012 , , 546 , a37 nomoto , k. i. , iwamoto , k. , & suzuki , t. 1995 , , 256 , 173 pik , e. j. 1924 , tartu obs .
, 25 otero , s. 2008a , open european journal on variable stars , 83 , 1 otero , s. 2008b , open european journal on variable stars , 91 , 1 pigulski , a. , & kolaczkowski , z. 1998 , mnras , 298 , 753 rauw , g. , vreux , j. m. , & bohannan , b. 1999 , apj 517 , 416 rios , l. y. , & degioia - eastwood , k. 2004 , baas , 205 , no .
09.05 romano , g. 1969 , mmsai , 40 , 375 sana , h. , & evans , c. j. 2011 , iau symposium , 272 , 474 sana , h. , gosset , e. , naz y. , rauw , g. , & linder , n. 2008 , mnras , 386 , 447 sana , h. , de mink , s. e. , de koter , a. , et al .
2012 , science , 337 , 444 scappini , f. , cecchi - pestellini , c. , casu , s. , & olberg , m. 2007 , , 466 , 243 sana , h. , de koter , a. , de mink , s. e. , et al .
2013 , , 550 , a107 schulte , d. h. 1958 , aj , 128 , 41 smartt , s. j. , eldridge , j. j. , crockett , r. m. , & maund , j. r. 2009 , , 395 , 1409 smith , r. j. , longmore , s. , & bonnell , i. 2009 , , 400 , 1775 smith , n. 2014 , arxiv.org/abs/1402.1237 stroud , v. e. , clark , j.s . , negueruela , i. , roche , p. , & norton , a.j .
2009 , a&a , 511 , 84 van den heuvel , e. p. j. 1983 , accretion - driven stellar x - ray sources , 303 vogt , s. s. , & penrod , g. d. 1983 , , 275 , 661 walborn , n. r. 1973 , apj , 180 , l35 walborn , n. r. & fitzpatrick , e. l. 1990 , 102 , 379 wilson , o. c. 1948 , pasp , 60 , 385 wilson , o. c. , & abt , a. 1951 , apj , 144 , 477 woosley , s. e. , & bloom , j. s. 2006 , , 44 , 507 wozniak , p. r. , et al .
2004 , aj , 127 , 2436 , northern sky variability survey : public data release zinnecker , h. , yorke , h. w. 2007 , ara&a , 45 , 481
in this appendix we briefly re - analyze systems originally presented in papers ii and iii for purposes of self - consistency , using the methods employed in this work and incorporating any new spectra , when available .
most of the solutions are essentially unchanged , but results are presented in a manner consistent with the solutions reported in papers iv , vi , and in this contribution . in three cases ( mt91 252 , s 73 , cpr2002 a45 ) , additional data allowed us to discover that the original solutions were aliases and to present a better solution . in this work , we disentangled the component spectra of double - lined binaries using the method of @xcite .
one of the strengths of this method is that the radial velocities can be refined via cross correlation after each iteration ( i.e , cross - correlating the the residual spectra with the resultant component spectrum as the template ) .
the cross - correlated velocities generally have smaller uncertainties and utilize more lines , leading to a better measurement of the true stellar velocity and the binary systemic velocity .
_ mt91 059_table [ velupdated1.tab ] lists the updated ephemeris and table [ solutionsupdated1.tab ] provides refined orbital elements for the single - lined binary mt91 059 originally presented in paper ii .
figure [ sol059 ] displays the best - fitting solution and folded velocity data .
the original period of 4.85 d is changed only slightly , but we note here the large reduced @xmath74 of 4.3 that indicates either an additional velocity component or photosspheric variability . attempts to identify periodicities in the observed - minus - computed velocities in table [ velupdated1.tab ] did not yield convincing evidence for an additional velocity component .
hints of emission in the cores of the @xmath494471 lines seeing in our two high - resolution spectra from keck support the possibility of extra - photospheric variability , perhaps as part of an accretion stream in this close ( @xmath100.18 a.u . )
binary . _
mt91 145_table [ velupdated1.tab ] lists the updated ephemeris and table [ solutionsupdated1.tab ] provides refined orbital elements for this single - lined o9ii star originally presented in paper iii .
the orbital elements are only slightly revised from the original published estimate .
figure [ sol145 ] displays the best - fitting solution and folded velocity data . _
mt91 252_we identified this system as an sb2 in paper ii but could only perform a limited analysis owing to the small number of spectra and low snr of the data . using the technique described in this paper , we included 17 epochs of wiro spectra and a keck+hires spectrum to compute the solution listed in table [ solutionsupdated2.tab ] .
the previously estimated period of 1819 days has been revised to @xmath75 days .
the power spectra for both components showed an alias at @xmath1011.4 days .
however , the folded velocity curve showed significantly higher scatter and was ruled out .
figure [ sol252 ] shows the computed solution .
no revisions to the estimated spectral classifications are provided . by adopting a mass of 1114 for the b12v components ,
the period and velocity amplitudes imply an inclination between 30 and 40 degrees . _
mt91 258_originally presented as an sb1 in paper ii , the updated ephemeris for this 14.658 d o8v appears in table [ velupdated1.tab ] and the orbital parameters appear in table [ solutionsupdated1.tab ] .
figure [ sol258 ] displays the best - fitting solution and folded velocity curve .
we attribute the unusually high reduced @xmath32 value to velocity zero point differences between data from different observatories used in the solution . _
mt91 372_identified as a 2.2 d binary in paper iii , this eclipsing double - lined system is analyzed in greater detail in conjunction with the eclipsing binary distance to cygnus ob2 in a forthcoming work @xcite .
_ schulte # 73_this system was identified as a 17.28 day sb2 in paper iii .
velocities remeasured with the technique described in papers iv , vi , and this work were used as the initial guesses for the @xcite method of spectral deconvolution and cross correlation for all wiro data . owing to the limited phase coverage of the wiyn+hydra spectra , we did not apply the @xcite method to these data .
however , since the 5876 velocities show a strong linear correlation with the final cross - correlation velocities , we used this correlation to correct the wiyn 4471 velocities .
the final wiro+wiyn data yielded a period 34.88 days .
the power spectra showed additional strong signals at @xmath1010 days , @xmath1017 days , @xmath1057 days , and @xmath1067 days .
however , these were ruled out based on the large scatter in the corresponding folded velocity curves .
the updated solution is provided in table [ solutionsupdated2.tab ] and shown in figure [ sols73 ] .
we slightly revise the spectral classifications , based on equivalent width ratios of 5411 to 5876 , as o8.5iii : and o9.0iii : . using the theoretical masses from @xcite ,
these classifications indicate that the system has an inclination near @xmath1040 . _
cpr2002 a45_also identified as an sb2 in paper iii , this system was previously listed with a period of 2.884 days and an uncharacteristically high eccentricity of 0.273 . using the @xcite spectral deconvolution method and some additional observations from wiro ,
the revised radial velocities indicate a period near half the original solution . both components power spectra show a singular strong signal at 1.5020 days .
the final combined solution yields a period of @xmath76 d and a considerably smaller eccentricity of @xmath77 .
the revised solution is listed in table [ solutionsupdated2.tab ] and shown in figure [ sola45 ] .
we retain the primary spectral classification of b0.5v , but owing to the new , higher mass ratio of [email protected] , we revise the secondary spectral classification to b1v b2v . based on the theoretical masses for stars of this type , we estimate an inclination between 40 and 45 .
lcrrr ptabularcr ptabularcr ptabularcr 55418.648 & 0.367 & -37.9 & 18.0 & -0.20 + 55489.731 & 0.695 & 1.9 & 17.9 & -4.77 + 55780.694 & 0.691 & 11.6 & 6.4 & 5.47 + 55791.720 & 0.828 & 20.8 & 4.2 & -1.34 + 55805.681 & 0.267 & -41.1 & 6.1 & 7.09 + 56209.582 & 0.906 & 18.7 & 5.3 & -1.94 + 56215.552 & 0.522 & -27.4 & 5.4 & -9.83 + 56446.902 & 0.372 & -38.0 & 8.7 & -1.02 + 56461.683 & 0.896 & 48.6 & 10.6 & 26.77 + 56463.804 & 0.115 & -51.8 & 6.9 & -4.22 + 56466.758 & 0.419 & -29.4 & 10.1 & 1.77 + 56470.764 & 0.832 & 8.8 & 6.2 & -13.63 + 56472.721 & 0.034 & -21.3 & 8.7 & 2.25 + 56566.728 & 0.726 & 32.1 & 7.3 & 21.15 + ptabularcr ptabularcr ptabularcr 51467.858 & 0.166 & -11.7 & 2.1 & -0.85 + 51805.908 & 0.149 & -8.8 & 4.1 & 1.55 + 52146.686 & 0.333 & -17.3 & 4.1 & -4.50 + 52162.651 & 0.513 & -14.6 & 2.9 & -1.96 + 53338.617 & 0.419 & -15.7 & 5.0 & -2.83 + 53989.652 & 0.532 & -13.4 & 4.2 & -0.84 + 54696.894 & 0.799 & -8.3 & 2.3 & 0.89 + 55791.858 & 0.720 & -14.7 & 2.3 & -3.85 + 55805.801 & 0.750 & -11.8 & 3.3 & -1.54 + 55836.692 & 0.033 & -1.7 & 3.6 & 0.87 + 56211.700 & 0.747 & -8.6 & 3.2 & 1.73 + 56215.635 & 0.038 & -4.0 & 3.3 & -0.97 + 56494.713 & 0.662 & -9.6 & 3.0 & 1.96 + 56446.784 & 0.120 & -8.0 & 4.3 & 1.28 + 56459.737 & 0.077 & -7.6 & 1.9 & -0.91 + 56462.891 & 0.311 & -12.9 & 2.2 & -0.17 + 56463.903 & 0.385 & -14.2 & 5.4 & -1.32 + 56470.884 & 0.901 & -6.5 & 2.4 & -1.43 + 56472.752 & 0.039 & -2.1 & 2.8 & 1.14 + 56490.729 & 0.368 & -5.9 & 3.2 & 7.00 + 56493.744 & 0.591 & -11.9 & 3.3 & 0.26 + 56494.745 & 0.665 & -9.6 & 3.0 & 1.94 + 56495.752 & 0.739 & -5.1 & 6.2 & 5.44 + 56496.901 & 0.824 & -8.7 & 4.2 & -0.20 + 56497.906 & 0.898 & -2.9 & 2.8 & 2.32 + ptabularcr ptabularcr ptabularcr 54696.928 & 0.880 & -15.9 & 16.5 & -0.13 + 54697.883 & 0.902 & -18.9 & 15.4 & 0.51 + 55781.847 & 0.065 & -33.9 & 8.8 & -0.90 + 55836.786 & 0.340 & 12.1 & 21.8 & 16.12 + 56247.641 & 0.878 & -7.7 & 11.0 & 7.73 + 56251.586 & 0.969 & -32.8 & 8.9 & -3.17 + 56446.855 & 0.502 & -6.2 & 15.2 & -10.45 + 56461.717 & 0.847 & -21.2 & 11.2 & -10.42 + 56464.770 & 0.918 & -23.3 & 15.6 & -1.28 + 56469.713 & 0.033 & -25.5 & 16.2 & 8.37 + 56489.726 & 0.497 & 7.2 & 12.4 & 3.12 + 56491.691 & 0.543 & 7.1 & 19.8 & 2.02 + 56504.789 & 0.847 & -7.4 & 11.8 & 3.39 + 56533.762 & 0.520 & 4.4 & 13.4 & -0.24 + 56566.612 & 0.282 & -6.8 & 14.2 & 2.35 + 56588.593 & 0.792 & 0.3 & 16.9 & 4.35 + 56607.643 & 0.235 & -18.7 & 11.9 & -4.63 + ptabularcr ptabularcr ptabularcr 51467.855 & 0.288 & -16.8 & 6.4 & -2.74 + 51805.911 & 0.356 & -33.8 & 7.4 & -13.41 + 52146.686 & 0.424 & -20.5 & 5.5 & 5.30 + 52161.793 & 0.427 & -22.3 & 4.8 & 3.71 + 52162.851 & 0.427 & -24.3 & 5.3 & 1.73 + 53338.617 & 0.663 & -40.0 & 5.3 & -7.75 + 53340.576 & 0.663 & -33.9 & 4.1 & -1.66 + 53989.652 & 0.793 & -13.4 & 4.7 & 11.04 + 53990.848 & 0.793 & -18.3 & 8.4 & 6.12 + 54286.904 & 0.852 & -28.5 & 5.3 & -10.53 + 54628.880 & 0.921 & -15.5 & 7.1 & -6.06 + 54630.852 & 0.921 & -5.4 & 4.9 & 3.99 + 55713.928 & 0.138 & -12.0 & 10.6 & -11.18 + 55717.858 & 0.139 & -0.8 & 3.9 & 0.02 + 55737.828 & 0.143 & -3.6 & 5.7 & -2.50 + 55741.587 & 0.144 & -1.3 & 5.5 & -0.12 + 55766.847 & 0.149 & 1.8 & 3.6 & 3.27 + 55781.912 & 0.152 & -3.3 & 8.3 & -1.62 + 55782.854 & 0.152 & -0.2 & 4.7 & 1.48 + 55795.453 & 0.154 & -4.1 & 4.0 & -2.23 + 56118.661 & 0.219 & -8.5 & 4.5 & -1.24 + 56166.369 & 0.229 & -6.5 & 5.2 & 1.70 + 56178.604 & 0.231 & -8.2 & 5.8 & 0.21 + 56460.868 & 0.287 & -13.7 & 5.7 & 0.33 + 56464.894 & 0.288 & -19.4 & 4.8 & -5.29 + 56468.901 & 0.289 & -12.8 & 5.8 & 1.34 + ptabularcr ptabularcr ptabularcr 51364.067 & 0.035 & 10.0 & 2.1 & 0.72 + 51466.739 & 0.188 & -12.0 & 1.4 & -0.71 + 51806.950 & 0.695 & -26.7 & 4.1 & 0.66 + 52161.793 & 0.224 & -7.3 & 4.9 & 7.16 + 53340.576 & 0.980 & 3.0 & 3.3 & -1.70 + 53989.774 & 0.948 & -4.4 & 5.6 & -0.73 + 54286.901 & 0.390 & -19.9 & 9.1 & 3.42 + 54628.748 & 0.900 & 4.0 & 7.8 & 18.22 + 56107.562 & 0.103 & -0.4 & 2.6 & 0.01 + 56240.279 & 0.301 & -22.4 & 2.2 & -2.93 + 56258.544 & 0.328 & -18.9 & 2.0 & 1.95 + 56445.897 & 0.607 & -27.1 & 2.2 & 0.47 + 56463.763 & 0.634 & -28.7 & 2.6 & -1.04 + 56471.705 & 0.646 & -27.7 & 2.5 & -0.07 + 56472.675 & 0.647 & -25.5 & 4.2 & 2.16 + 56475.891 & 0.652 & -26.8 & 2.6 & 0.79 + 56555.611 & 0.771 & -29.1 & 2.8 & -3.41 + 56590.618 & 0.823 & -22.8 & 1.8 & 0.19 + 56608.615 & 0.850 & -21.4 & 2.2 & -0.65 + ptabularcr ptabularcr
ptabularcr 51364.087 & 0.872 & -68.0 & 3.1 & -4.41 + 51466.729 & 0.952 & -73.0 & 4.1 & -11.75 + 52146.829 & 0.359 & -10.8 & 8.1 & -9.03 + 52161.650 & 0.803 & -63.7 & 6.1 & -9.59 + 52162.828 & 0.839 & -46.9 & 8.1 & 12.39 + 54628.880 & 0.835 & -38.8 & 10.1 & 20.03 + 54630.852 & 0.895 & -67.3 & 14.1 & -1.65 + 56107.937 & 0.216 & 25.0 & 5.1 & 13.99 + 56178.806 & 0.343 & 10.7 & 3.8 & 10.82 + 56445.758 & 0.353 & -9.3 & 6.5 & -8.14 + 56469.761 & 0.073 & -2.3 & 5.5 & -7.41 + 56470.718 & 0.102 & -0.3 & 4.4 & -11.65 + 56490.802 & 0.704 & -46.9 & 4.6 & -6.79 + 56491.810 & 0.735 & -43.8 & 5.1 & 0.42 + 56493.854 & 0.796 & -26.7 & 5.0 & 26.32 + 56494.745 & 0.823 & -43.1 & 5.6 & 13.94 + 56505.868 & 0.156 & 2.0 & 5.5 & -11.93 + 56533.824 & 0.995 & -36.5 & 5.4 & 2.89 + 56552.632 & 0.560 & -32.3 & 4.2 & -9.76 + 56574.679 & 0.221 & 23.9 & 4.0 & 13.26 + 56589.679 & 0.671 & -30.8 & 4.3 & 4.97 + 56608.547 & 0.237 & -25.3 & 7.2 & -34.73 + ptabularcr ptabularcr ptabularcr 51364.087 & 0.658 & -13.5 & 3.1 & -2.54 + 51466.729 & 0.857 & -14.8 & 4.1 & 0.48 + 52146.829 & 0.690 & -7.9 & 8.1 & 4.09 + 52161.650 & 0.606 & -16.6 & 6.1 & -7.48 + 52162.828 & 0.838 & 4.4 & 8.1 & 19.84 + 54628.880 & 0.119 & 13.8 & 10.1 & -2.93 + 54630.852 & 0.507 & -9.2 & 14.1 & -3.87 + 56107.937 & 0.175 & 19.3 & 5.1 & 6.88 + 56178.806 & 0.121 & 14.1 & 3.8 & -2.51 + 56445.758 & 0.653 & -4.8 & 6.5 & 5.92 + 56469.761 & 0.376 & 0.0 & 5.5 & -0.29 + 56470.718 & 0.564 & -4.0 & 4.4 & 3.54 + 56490.802 & 0.517 & -8.4 & 4.6 & -2.65 + 56491.810 & 0.715 & -1.8 & 5.1 & 11.02 + 56493.854 & 0.117 & 22.6 & 5.0 & 5.69 + 56494.745 & 0.293 & 9.4 & 5.6 & 4.74 + 56505.868 & 0.481 & -5.3 & 5.5 & -0.94 + 56533.824 & 0.983 & 0.7 & 5.4 & -2.55 + 56552.632 & 0.684 & -9.0 & 4.2 & 2.76 + 56574.679 & 0.022 & 18.7 & 4.0 & 3.45 + 56589.679 & 0.974 & 4.1 & 4.3 & 3.71 + 56608.547 & 0.687 & -29.6 & 7.2 & -17.70 + ptabularcr ptabularcr ptabularcr 51805.904 & 0.838 & -35.0 & 4.1 & -7.11 + 52146.686 & 0.847 & -23.0 & 8.1 & 6.38 + 52161.793 & 0.867 & -25.2 & 10.1 & 7.73 + 52162.651 & 0.925 & -35.9 & 9.9 & 10.24 + 53340.576 & 0.458 & -10.5 & 12.8 & -3.24 + 53989.774 & 0.291 & -11.0 & 12.1 & -1.59 + 54403.777 & 0.244 & -24.8 & 10.6 & -13.19 + 54409.683 & 0.643 & -12.2 & 5.3 & -0.82 + 56118.870 & 0.046 & -48.0 & 5.4 & -1.62 + 56219.688 & 0.853 & -27.6 & 3.7 & 2.78 + 56443.749 & 0.982 & -55.0 & 6.3 & 2.47 + 56464.729 & 0.398 & 21.7 & 18.6 & 28.99 + 56496.764 & 0.561 & -11.0 & 6.6 & -2.27 + 56497.801 & 0.631 & -14.1 & 5.7 & -3.19 + 56500.729 & 0.829 & -25.0 & 3.9 & 1.56 + 56502.770 & 0.967 & -62.6 & 8.5 & -6.82 + 56504.743 & 0.100 & -26.3 & 5.3 & 4.15 + 56505.810 & 0.172 & -19.6 & 6.4 & -1.78 + 56555.657 & 0.538 & -6.4 & 3.3 & 1.84 + 56560.782 & 0.884 & -40.9 & 4.7 & -4.68 + ptabularcr ptabularcr ptabularcr 56118.832 & 0.998 & -28.9 & 3.3 & -0.17 + 56240.645 & 0.458 & -17.9 & 3.3 & -1.38 + 56447.807 & 0.574 & -6.5 & 4.7 & 7.55 + 56465.847 & 0.899 & -18.9 & 4.5 & 0.52 + 56471.784 & 0.309 & -18.3 & 3.6 & 3.03 + 56489.874 & 0.655 & -10.3 & 4.4 & 2.83 + 56496.690 & 0.422 & -20.4 & 3.9 & -2.91 + 56502.730 & 0.875 & -18.4 & 4.8 & -0.72 + 56504.830 & 0.727 & -14.7 & 4.7 & -1.55 + 56555.786 & 0.417 & -13.2 & 4.3 & 4.39 + 56589.731 & 0.200 & -28.0 & 4.0 & -1.68 + 56590.565 & 0.539 & -19.5 & 3.2 & -4.79 + 56599.638 & 0.223 & -24.9 & 3.8 & 0.31 + 56623.557 & 0.935 & -23.4 & 4.4 & -0.67 + ptabularcr ptabularcr ptabularcr 51364.116 & 0.849 & -20.6 ( 2.2 ) & 0.95 + 52146.820 & 0.362 & -1.6 ( 4.8 ) & 4.62 + 52161.650 & 0.629 & -13.3 ( 4.6 ) & -1.67 + 52162.820 & 0.202 & -8.2 ( 4.7 ) & 0.98 + 54629.860 & 0.017 & -18.0 ( 4.6 ) & 3.10 + 54630.740 & 0.448 & -9.6 ( 2.5 ) & -2.71 + 54632.740 & 0.428 & -1.2 ( 4.5 ) & 5.43 + 54633.850 & 0.972 & -23.2 ( 4.6 ) & -0.11 + 56623.601 & 0.922 & -25.6 ( 2.3 ) & -2.04 + 56798.913 & 0.822 & -16.2 ( 6.2 ) & 4.17 + 56799.837 & 0.275 & -7.4 ( 2.7 ) & -0.41 + 56804.884 & 0.748 & -15.1 ( 4.7 ) & 1.68 + ptabularcr ptabularcr
ptabularcr 54641.903 & 0.475 & -10.7 & 1.2 & 1.10 + 54642.890 & 0.501 & -14.0 & 4.1 & -1.90 + 54643.757 & 0.524 & -14.7 & 4.0 & -2.22 + 54644.826 & 0.552 & -14.0 & 3.7 & -1.10 + 54647.860 & 0.632 & -16.3 & 3.0 & -2.28 + 54669.803 & 0.212 & -10.0 & 2.0 & -0.17 + 54671.884 & 0.267 & -9.6 & 1.9 & 0.30 + 54672.834 & 0.292 & -10.9 & 1.3 & -0.93 + 54673.889 & 0.320 & -10.1 & 1.5 & 0.01 + 54674.914 & 0.347 & -10.4 & 1.9 & -0.01 + 56118.932 & 0.483 & -12.5 & 1.9 & -0.61 + 56179.923 & 0.093 & -10.9 & 2.5 & 0.53 + 56209.798 & 0.882 & -15.6 & 1.2 & 0.99 + 56211.649 & 0.931 & -18.0 & 1.7 & -1.96 + 56438.789 & 0.930 & -16.8 & 5.6 & -0.66 + 56443.890 & 0.065 & -13.6 & 2.8 & -1.37 + 56444.745 & 0.087 & -9.3 & 2.9 & 2.31 + 56445.742 & 0.114 & -10.0 & 2.7 & 0.96 + 56446.934 & 0.145 & -9.6 & 2.3 & 0.83 + 56447.930 & 0.171 & -11.2 & 2.5 & -1.11 + 56457.798 & 0.432 & -11.2 & 2.1 & 0.04 + 56458.700 & 0.456 & -10.4 & 2.5 & 1.18 + 56475.835 & 0.908 & -15.2 & 2.0 & 1.22 + 56486.814 & 0.198 & -11.8 & 2.9 & -1.90 + 56491.785 & 0.330 & -9.5 & 2.8 & 0.76 + 56493.726 & 0.381 & -11.5 & 2.6 & -0.86 + 56494.672 & 0.406 & -10.5 & 2.4 & 0.46 + 56559.699 & 0.123 & -11.0 & 2.2 & -0.24 + 56566.758 & 0.310 & -10.0 & 2.4 & 0.11 + 56567.769 & 0.336 & -6.8 & 3.6 & 3.49 + 56574.633 & 0.518 & -13.3 & 2.0 & -0.94 + 56585.607 & 0.807 & -16.9 & 3.2 & -0.51 + 56588.682 & 0.889 & -18.3 & 2.2 & -1.72 + 56590.548 & 0.938 & -15.8 & 2.0 & 0.17 + ptabularcr ptabularcr ptabularcr 55424.755 & 0.861 & -38.2 & 31.8 & -16.80 + 55491.691 & 0.138 & 22.4 & 27.3 & -17.51 + 55779.871 & 0.938 & -12.5 & 10.2 & -12.36 + 55790.594 & 0.303 & 14.1 & 5.6 & -2.92 + 55866.476 & 0.883 & -9.9 & 6.8 & 6.64 + 56076.633 & 0.030 & 33.2 & 7.0 & 2.70 + 56080.493 & 0.162 & 36.2 & 7.3 & -1.82 + 56092.490 & 0.570 & -23.3 & 7.2 & -2.07 + 56093.492 & 0.604 & -24.1 & 9.2 & 0.51 + 56097.520 & 0.741 & -32.1 & 5.8 & -0.12 + 56463.835 & 0.198 & 36.2 & 7.5 & 2.49 + 56465.733 & 0.263 & 25.3 & 7.1 & 1.61 + 56469.682 & 0.397 & 10.0 & 8.0 & 8.30 + 56470.827 & 0.436 & -4.8 & 6.2 & -0.63 + 56472.788 & 0.503 & -16.5 & 6.1 & -3.20 + 56489.839 & 0.082 & 37.2 & 8.4 & -2.22 + 56491.859 & 0.151 & 34.1 & 7.3 & -4.89 + 56518.751 & 0.066 & 40.4 & 7.1 & 2.87 + 56519.753 & 0.100 & 46.1 & 8.7 & 5.62 + ptabularcr ptabularcr ptabularcr 55715.888 & 0.670 & -18.7 & 5.2 & 0.73 + 55718.899 & 0.321 & -13.5 & 2.7 & -0.97 + 55768.806 & 0.111 & -16.2 & 3.7 & -0.50 + 55782.732 & 0.122 & -17.8 & 4.8 & -2.48 + 55796.785 & 0.160 & -14.6 & 5.9 & -0.53 + 55805.923 & 0.136 & -13.3 & 6.3 & 1.58 + 56211.817 & 0.893 & -23.0 & 4.5 & 0.08 + 56212.744 & 0.093 & -14.5 & 3.4 & 1.96 + 56240.753 & 0.149 & -14.2 & 4.4 & 0.20 + 56251.637 & 0.502 & -16.5 & 2.8 & -1.37 + 56494.867 & 0.090 & -17.9 & 3.7 & -1.31 + 56496.807 & 0.510 & -14.0 & 6.7 & 1.32 + 56497.841 & 0.733 & -21.1 & 3.9 & -0.08 + 56502.855 & 0.817 & -29.3 & 9.0 & -6.48 + 56504.902 & 0.260 & -10.9 & 4.1 & 1.60 + 56505.702 & 0.433 & -11.2 & 4.4 & 2.64 + 56533.900 & 0.530 & -15.0 & 4.8 & 0.75 + ptabularcr ptabularcr ptabularcr 55714.893 & 0.479 & -8.1 & 4.7 & -2.87 + 55717.890 & 0.659 & -33.6 & 4.0 & 4.45 + 55727.901 & 0.261 & 30.0 & 5.0 & 0.59 + 55738.896 & 0.922 & -62.0 & 4.2 & -5.76 + 55740.551 & 0.021 & -2.9 & 3.5 & 4.59 + 55756.824 & 0.999 & -19.6 & 5.4 & 0.94 + 55796.877 & 0.407 & 2.1 & 10.4 & -5.26 + 55832.549 & 0.551 & -18.7 & 4.2 & -0.63 + 55855.693 & 0.942 & -46.5 & 4.2 & 3.27 + 55857.646 & 0.059 & 7.7 & 3.8 & -4.60 + 55906.555 & 0.999 & -22.6 & 3.4 & -1.73 + 55911.533 & 0.298 & 28.4 & 3.8 & 3.85 + 56076.514 & 0.214 & 33.7 & 6.5 & -0.25 + 56080.524 & 0.455 & -6.4 & 5.4 & -5.43 + 56092.534 & 0.177 & 35.4 & 6.0 & 0.05 + 56093.552 & 0.238 & 33.3 & 6.1 & 1.43 + 56179.471 & 0.402 & 5.0 & 5.8 & -3.23 + 56205.353 & 0.958 & -41.8 & 5.2 & 1.25 + ptabularcr ptabularcr ptabularcr 54672.823 & 0.194 & -10.1 ( 7.8 ) & 2.41 + 56511.750 & 0.528 & -19.9 ( 6.8 ) & -1.87 + 56512.762 & 0.554 & -18.9 ( 3.6 ) & 0.23 + 56519.663 & 0.736 & -31.2 ( 4.9 ) & -4.39 + 56533.734 & 0.106 & -17.8 ( 5.1 ) & -1.29 + 56534.756 & 0.132 & -13.4 ( 5.3 ) & 1.47 + 56556.732 & 0.710 & -25.0 ( 3.8 ) & 0.73 + 56574.608 & 0.180 & -10.8 ( 4.0 ) & 2.07 + 56582.586 & 0.390 & -15.3 ( 4.4 ) & -1.65 + 56589.552 & 0.573 & -19.1 ( 4.0 ) & 0.69 + 56590.676 & 0.602 & -22.8 ( 4.0 ) & -1.72 + 56606.553 & 0.020 & -20.9 ( 3.8 ) & 2.97 + 56626.550 & 0.545 & -22.7 ( 3.9 ) & -3.94 + 56795.922 & 0.997 & -23.4 ( 10.0 ) & 2.37 + 56804.923 & 0.233 & -12.6 ( 6.1 ) & -0.71 + ptabularcr ptabularcr ptabularcr 55766.875 & 0.552 & -33.9 & 6.1 & 0.08 + 55781.721 & 0.235 & -8.0 & 5.6 & 4.25 + 55832.632 & 0.293 & -11.6 & 10.8 & 9.80 + 55834.666 & 0.935 & 16.8 & 6.6 & -0.49 + 55847.701 & 0.046 & 17.2 & 5.8 & -0.44 + 55876.617 & 0.167 & -6.9 & 6.2 & -6.88 + 56077.852 & 0.640 & -29.6 & 4.1 & -2.51 + 56080.797 & 0.569 & -28.5 & 5.4 & 4.56 + 56092.806 & 0.356 & -30.3 & 6.2 & -1.32 + 56093.840 & 0.683 & -12.3 & 6.8 & 9.59 + 56097.807 & 0.934 & 18.3 & 4.7 & 1.07 + 56179.749 & 0.780 & -9.5 & 5.4 & -3.06 + 56446.824 & 0.020 & 17.2 & 6.2 & -2.13 + 56447.900 & 0.359 & -26.2 & 5.7 & 3.07 + 56458.767 & 0.787 & -10.2 & 6.6 & -5.00 + 56463.867 & 0.396 & -43.0 & 8.8 & -10.72 + 56469.840 & 0.280 & -20.1 & 4.6 & -0.64 + 56470.796 & 0.581 & -29.9 & 5.3 & 2.40 + ptabularcr ptabularcr ptabularcr 51365.110 & 0.925 & -10.1 & 5.1 & 3.51 + 51467.882 & 0.986 & -12.1 & 5.1 & 0.69 + 51805.789 & 0.186 & -4.1 & 5.1 & -4.29 + 52161.793 & 0.397 & -3.0 & 5.1 & -0.71 + 53340.576 & 0.095 & 3.0 & 7.1 & 2.68 + 53989.770 & 0.480 & -8.0 & 7.1 & -4.64 + 54286.904 & 0.656 & -12.0 & 7.1 & -6.02 + 54628.748 & 0.859 & -4.0 & 8.1 & 6.95 + 54629.860 & 0.860 & -14.0 & 9.1 & -3.03 + 54630.739 & 0.860 & -14.0 & 5.1 & -3.01 + 54631.850 & 0.861 & -10.0 & 6.1 & 1.01 + 54633.850 & 0.862 & -7.0 & 8.1 & 4.06 + 55420.926 & 0.328 & 1.8 & 10.4 & 3.30 + 55461.790 & 0.353 & 7.4 & 10.3 & 9.11 + 55490.761 & 0.370 & -2.5 & 10.6 & -0.56 + 55717.919 & 0.504 & -0.9 & 5.9 & 2.74 + 55740.894 & 0.518 & -0.3 & 5.8 & 3.54 + 55790.919 & 0.548 & -4.8 & 7.0 & -0.48 + 55794.758 & 0.550 & -11.4 & 9.4 & -7.09 + 55834.790 & 0.574 & -3.7 & 3.8 & 0.99 + 55847.762 & 0.581 & -2.6 & 4.2 & 2.22 + 55855.760 & 0.586 & -5.3 & 4.1 & -0.42 + 55856.545 & 0.587 & -10.8 & 5.3 & -5.92 + 55866.778 & 0.593 & 6.2 & 6.3 & 11.15 + 56128.911 & 0.748 & -9.8 & 6.1 & -2.05 + 56190.867 & 0.785 & -9.0 & 5.4 & -0.33 + 56205.661 & 0.794 & -14.7 & 5.7 & -5.81 + 56457.907 & 0.943 & -16.0 & 4.7 & -1.77 + 56472.825 & 0.952 & -7.6 & 5.7 & 6.87 + 56486.779 & 0.960 & -28.3 & 7.4 & -13.87 + 56489.822 & 0.962 & -13.3 & 6.7 & 1.15 + 56490.852 & 0.963 & -18.2 & 5.2 & -3.81 + 56491.892 & 0.963 & -13.2 & 6.0 & 1.26 + 56493.777 & 0.964 & -8.7 & 5.8 & 5.67 + 56494.689 & 0.965 & -14.7 & 5.3 & -0.29 + 56495.909 & 0.966 & -15.2 & 7.5 & -0.89 + ptabularcr ptabularcr ptabularcr 51467.878 & 0.813 & -23.1 & 4.0 & -8.26 + 51805.804 & 0.896 & -17.2 & 4.2 & -4.82 + 52146.829 & 0.980 & 2.5 & 5.2 & -1.17 + 52161.650 & 0.983 & 5.5 & 5.5 & -0.26 + 52162.820 & 0.984 & 8.5 & 5.1 & 2.57 + 53339.570 & 0.273 & -12.3 & 4.7 & 1.93 + 53989.870 & 0.433 & -12.2 & 5.0 & 3.35 + 54285.920 & 0.506 & -23.3 & 4.8 & -7.47 + 54628.700 & 0.590 & -16.2 & 7.4 & -0.24 + 54630.739 & 0.590 & -17.8 & 5.4 & -1.84 + 54631.850 & 0.591 & -12.2 & 5.6 & 3.76 + 55714.847 & 0.857 & -9.5 & 3.8 & 4.38 + 55715.919 & 0.857 & -16.7 & 3.5 & -2.81 + 55718.762 & 0.858 & -12.8 & 2.7 & 1.12 + 55738.873 & 0.863 & -11.6 & 3.8 & 2.16 + 55740.574 & 0.863 & -13.4 & 2.8 & 0.32 + 55757.764 & 0.868 & -13.0 & 2.9 & 0.56 + 55768.894 & 0.870 & -11.5 & 4.2 & 2.02 + 55781.819 & 0.874 & -11.6 & 2.9 & 1.72 + 55834.836 & 0.887 & -12.2 & 2.4 & 0.66 + 55866.756 & 0.894 & -10.5 & 3.6 & 1.96 + 56076.828 & 0.946 & -7.9 & 4.5 & -0.49 + 56080.853 & 0.947 & -8.1 & 4.1 & -0.91 + 56092.861 & 0.950 & -7.9 & 3.9 & -1.20 + 56097.882 & 0.951 & -7.1 & 4.1 & -0.64 + 56460.731 & 0.040 & -0.2 & 4.0 & -0.03 + 56471.928 & 0.043 & -1.5 & 3.7 & -0.61 + 56590.763 & 0.072 & -5.9 & 4.3 & 0.29 + ptabularcr ptabularcr ptabularcr 51805.820 & 0.775 & -21.0 & 4.1 & -2.37 + 52146.686 & 0.042 & 6.3 & 10.1 & -2.92 + 52161.793 & 0.053 & 4.4 & 13.1 & -1.58 + 52162.651 & 0.054 & 16.2 & 16.1 & 10.41 + 53338.617 & 0.973 & 21.9 & 9.1 & 8.95 + 53340.576 & 0.975 & 16.7 & 9.1 & 3.41 + 53989.652 & 0.482 & -25.2 & 12.1 & -0.72 + 54628.880 & 0.981 & -4.5 & 14.1 & -19.13 + 55718.875 & 0.833 & -11.1 & 5.1 & 3.13 + 55740.870 & 0.851 & -10.5 & 4.6 & 1.98 + 55768.789 & 0.872 & -9.7 & 4.4 & -0.06 + 55782.825 & 0.883 & -2.9 & 5.0 & 5.09 + 55791.940 & 0.890 & -7.2 & 4.3 & -0.43 + 55794.842 & 0.893 & -12.1 & 6.3 & -5.69 + 55847.678 & 0.934 & 3.4 & 5.1 & 0.74 + 56118.744 & 0.146 & -12.4 & 6.2 & -0.30 + 56179.888 & 0.194 & -22.6 & 6.5 & -6.02 + 56191.831 & 0.203 & -12.3 & 6.4 & 5.02 + 56460.777 & 0.413 & -23.4 & 6.7 & 0.64 + 56465.764 & 0.417 & -22.7 & 6.2 & 1.41 + 56468.815 & 0.419 & -30.8 & 7.0 & -6.64 + 56518.789 & 0.459 & -7.9 & 9.1 & 16.49 + 56552.756 & 0.485 & -31.0 & 5.4 & -6.47 + 56566.807 & 0.496 & -7.5 & 8.3 & 16.98 + 56567.714 & 0.497 & -20.1 & 12.1 & 4.38 + 56574.804 & 0.502 & -21.2 & 5.6 & 3.28 + 56588.656 & 0.513 & -22.5 & 5.4 & 1.96 + 56608.668 & 0.529 & -23.9 & 4.4 & 0.56 + ptabularcr ptabularcr ptabularcr 54403.790 & 0.752 & -31.7 & 9.2 & -1.61 + 54406.679 & 0.824 & -15.5 & 8.5 & 0.27 + 55377.898 & 0.050 & 41.5 & 21.3 & 15.08 + 55426.768 & 0.269 & 12.5 & 21.6 & 6.35 + 55457.652 & 0.039 & 72.0 & 25.9 & 46.35 + 55470.731 & 0.365 & -4.2 & 25.5 & 9.70 + 55758.813 & 0.551 & -37.6 & 7.4 & 3.19 + 55790.941 & 0.353 & -15.3 & 6.8 & -4.09 + 56166.884 & 0.730 & -29.2 & 9.7 & 4.33 + 56179.680 & 0.049 & 6.5 & 10.7 & -19.82 + 56445.866 & 0.689 & -48.3 & 11.2 & -9.75 + 56461.898 & 0.089 & 32.6 & 9.3 & 4.92 + 56465.792 & 0.186 & 20.9 & 9.2 & 0.36 + 56469.808 & 0.286 & 4.9 & 10.8 & 2.21 + 56475.715 & 0.433 & -27.3 & 9.7 & -0.68 + 56490.877 & 0.811 & -16.1 & 10.3 & 2.47 + 56494.909 & 0.912 & 7.1 & 11.8 & 2.61 + ptabularcr ptabularcr ptabularcr 51805.819 & 0.648 & -11.3 & 5.0 & 8.16 + 52146.686 & 0.038 & -18.9 & 5.7 & -2.00 + 52162.651 & 0.103 & 4.3 & 5.5 & 12.31 + 53989.652 & 0.558 & -19.1 & 5.1 & -2.85 + 53990.848 & 0.562 & -16.3 & 10.1 & 0.12 + 56165.921 & 0.437 & -13.3 & 6.1 & -0.66 + 56218.697 & 0.652 & -24.2 & 6.0 & -4.55 + 56258.636 & 0.815 & -31.2 & 6.8 & -3.46 + 56460.754 & 0.640 & -18.1 & 5.2 & 1.09 + 56464.704 & 0.656 & -21.3 & 5.4 & -1.52 + 56534.801 & 0.942 & -35.8 & 5.8 & -0.03 + 56556.854 & 0.032 & -12.9 & 5.2 & 5.71 + 56559.742 & 0.043 & -15.8 & 4.1 & -0.14 + 56560.699 & 0.047 & -17.5 & 4.3 & -2.73 + 56566.830 & 0.072 & -17.1 & 4.4 & -6.43 + 56574.574 & 0.104 & -4.6 & 3.9 & 3.39 + 56582.616 & 0.137 & -6.5 & 4.2 & 0.46 + 56587.554 & 0.157 & -2.2 & 5.8 & 4.58 + 56588.566 & 0.161 & -10.0 & 3.5 & -3.25 + 56590.729 & 0.170 & -9.7 & 3.9 & -2.91 + 56607.576 & 0.239 & -7.5 & 4.8 & 0.25 + ptabularcr ptabularcr ptabularcr 51467.916 & 0.681 & -10.8 & 3.1 & 1.81 + 51805.746 & 0.343 & -11.0 & 3.1 & 6.70 + 52146.686 & 0.012 & -1.0 & 3.8 & 0.48 + 54643.857 & 0.906 & 0.2 & 4.7 & 2.84 + 54644.926 & 0.908 & -0.4 & 5.1 & 1.95 + 54647.882 & 0.914 & 1.6 & 4.6 & 3.38 + 54669.940 & 0.957 & 4.7 & 2.4 & 0.74 + 54672.765 & 0.963 & -0.3 & 2.1 & -5.16 + 54673.770 & 0.965 & 4.4 & 2.2 & -0.75 + 54674.896 & 0.967 & 11.1 & 2.1 & 5.69 + 56076.678 & 0.715 & -5.6 & 2.7 & 6.24 + 56080.915 & 0.723 & -11.8 & 2.5 & -0.24 + 56092.907 & 0.746 & -8.5 & 2.5 & 2.44 + 56098.779 & 0.758 & -17.4 & 2.2 & -6.82 + 56110.915 & 0.782 & -11.8 & 2.5 & -2.04 + 56118.713 & 0.797 & -9.0 & 3.1 & 0.19 + 56128.951 & 0.817 & -3.0 & 3.0 & 5.40 + 56166.976 & 0.892 & -2.7 & 4.0 & 1.20 + 56179.913 & 0.917 & -5.6 & 2.4 & -4.11 + 56212.580 & 0.981 & 6.7 & 2.2 & 0.01 + 56222.715 & 0.001 & 2.6 & 1.8 & -0.68 + 56440.951 & 0.429 & -18.8 & 2.9 & -2.04 + 56460.688 & 0.467 & -26.9 & 2.8 & -10.70 + 56462.832 & 0.471 & -25.2 & 2.5 & -9.02 + 56491.924 & 0.528 & -13.9 & 3.4 & 1.47 + 56497.897 & 0.540 & -15.2 & 2.8 & -0.03 + 56552.875 & 0.648 & -12.0 & 2.3 & 1.28 + 56559.771 & 0.661 & -8.3 & 2.5 & 4.77 + 56590.754 & 0.722 & -9.8 & 2.5 & 1.81 + 56622.557 & 0.784 & -5.6 & 2.2 & 4.09 + 56795.895 & 0.124 & 2.8 & 2.2 & 21.40 + ptabularcr ptabularcr ptabularcr 56191.706 & 0.278 & -72.7 ( 4.8 ) & 0.8 + 56461.866 & 0.701 & 6.4 ( 3.7 ) & -6.7 + 56464.862 & 0.761 & 31.1 ( 4.3 ) & 5.0 + 56467.795 & 0.820 & 42.4 ( 10.2 ) & 5.9 + 56469.872 & 0.862 & 56.6 ( 6.1 ) & 16.4 + 56495.871 & 0.383 & -65.1 ( 8.9 ) & -10.3 + 56496.732 & 0.401 & -64.6 ( 5.2 ) & -13.2 + 56500.887 & 0.485 & -44.4 ( 14.1 ) & -9.9 + 56504.870 & 0.564 & 2.7 ( 7.1 ) & 20.0 + 56505.735 & 0.582 & 0.6 ( 4.4 ) & 14.1 + 56552.682 & 0.524 & -56.9 ( 10.5 ) & -30.9 + 56556.801 & 0.607 & 12.4 ( 5.2 ) & 20.5 + 56560.740 & 0.686 & 12.1 ( 8.3 ) & 2.4 + 56566.777 & 0.807 & 33.9 ( 3.0 ) & -0.7 + 56574.649 & 0.965 & 11.9 ( 2.1 ) & -1.2 + 56582.648 & 0.126 & -62.3 ( 5.2 ) & 16.5 + 56589.771 & 0.269 & -71.5 ( 2.9 ) & 3.4 + 56590.695 & 0.287 & -70.7 ( 4.0 ) & 1.4 + 56599.736 & 0.469 & -57.7 ( 7.6 ) & -19.9 + ptabularcr ptabularcr
ptabularcr 56191.706 & 0.278 & 39.3 ( 5.9 ) & 0.4 + 56461.866 & 0.701 & -65.8 ( 4.6 ) & 7.2 + 56464.862 & 0.761 & -81.5 ( 5.4 ) & 8.3 + 56467.795 & 0.820 & -96.2 ( 12.7 ) & 7.0 + 56469.872 & 0.862 & -98.8 ( 7.6 ) & 9.2 + 56495.871 & 0.383 & 14.7 ( 11.1 ) & -0.1 + 56496.732 & 0.401 & 22.7 ( 6.5 ) & 12.2 + 56500.887 & 0.485 & 13.9 ( 17.7 ) & 25.4 + 56504.870 & 0.564 & -55.0 ( 8.9 ) & -21.4 + 56505.735 & 0.582 & -61.6 ( 5.4 ) & -23.1 + 56552.682 & 0.524 & 1.7 ( 13.1 ) & 24.0 + 56556.801 & 0.607 & -60.2 ( 6.5 ) & -14.5 + 56560.740 & 0.686 & -74.9 ( 10.4 ) & -6.3 + 56566.777 & 0.807 & -105.4 ( 3.7 ) & -4.6 + 56574.649 & 0.965 & -77.6 ( 2.6 ) & -4.6 + 56582.648 & 0.126 & 50.1 ( 6.5 ) & 4.2 + 56589.771 & 0.269 & 30.9 ( 3.6 ) & -9.8 + 56590.695 & 0.287 & 36.1 ( 5.0 ) & -1.0 + 56599.736 & 0.469 & 5.5 ( 9.5 ) & 12.7 + ptabularcr ptabularcr ptabularcr 56166.691 & 0.507 & -16.7 & 7.2 & 0.67 + 56211.440 & 0.803 & -12.5 & 3.7 & -2.52 + 56218.487 & 0.849 & -14.1 & 5.8 & -7.27 + 56457.883 & 0.432 & -16.1 & 12.3 & 1.80 + 56460.712 & 0.451 & -21.6 & 5.2 & -3.82 + 56462.843 & 0.465 & -20.3 & 3.8 & -2.56 + 56475.700 & 0.550 & -20.5 & 5.3 & -3.60 + 56500.921 & 0.717 & -8.3 & 5.0 & 5.29 + 56504.690 & 0.742 & -10.8 & 5.6 & 1.98 + 56505.680 & 0.749 & -13.7 & 4.9 & -1.18 + 56511.662 & 0.788 & -13.0 & 5.1 & -2.21 + 56518.815 & 0.835 & -10.1 & 4.9 & -2.27 + 56552.859 & 0.061 & -6.0 & 4.6 & -3.31 + 56556.872 & 0.087 & -3.3 & 4.1 & 4.61 + 56559.781 & 0.106 & -11.0 & 3.8 & -0.50 + 56560.621 & 0.112 & -12.0 & 3.9 & -0.95 + 56571.604 & 0.185 & -17.2 & 5.2 & -1.52 + 56585.544 & 0.277 & -15.7 & 4.5 & 1.87 + 56588.724 & 0.298 & -15.6 & 3.8 & 2.23 + 56599.768 & 0.371 & -20.8 & 3.9 & -2.73 + 56606.580 & 0.416 & -14.0 & 3.7 & 4.00 + 56608.599 & 0.429 & -15.2 & 4.3 & 2.69 + 51467.889 & 0.430 & -21.4 & 4.5 & -3.47 + 51805.782 & 0.665 & -14.6 & 4.7 & 0.38 + 52146.829 & 0.920 & 4.5 & 5.4 & 2.25 + 52161.650 & 0.018 & 11.3 & 3.7 & -0.20 + 52162.828 & 0.026 & 1.3 & 6.0 & -7.32 + 53339.572 & 0.809 & -7.1 & 4.6 & 2.55 + 53340.576 & 0.815 & -1.8 & 4.6 & 7.45 + 53989.774 & 0.109 & -9.3 & 4.7 & 1.48 + 54285.779 & 0.067 & -12.9 & 6.0 & -8.82 + 54286.904 & 0.074 & -9.5 & 9.1 & -3.85 + lrrrrr @xmath4 ( days ) & [email protected] & [email protected] & [email protected] & 4996@xmath33330 & 671@xmath332 + @xmath20 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + @xmath21 ( deg ) & 81@xmath3331 & 15@xmath3319 & 160@xmath3340 & 328@xmath3354 & 338@xmath336 + @xmath12 ( ) & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + @xmath22 ( hjd-2,400,000 ) & [email protected] & [email protected] & [email protected] & 55024@xmath33722 & 125161@xmath33189 + @xmath78 ( ) & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + @xmath79 ( ) & & & & & + s. c.@xmath25 & b1.5v & b1v & b2v & b1.5v & b1.5v + s. c.@xmath26 & mid - late b ?
& ... & b , a & b ? & early b + @xmath27 ( degrees ) & 1685 & 385 & 1590 & @xmath1090 & @xmath1090 + @xmath80 ( adopted ; ) & 12 & 14 & 11 & 12 & 12 + @xmath81 ( ) & 112 & 14 0.43 & 11 2 & 17 : & 114.6 + @xmath30 & 10.18 & 1 0.03 & 10.18 & 12 ? & 0.80.42 + @xmath31 ( au ) & @xmath100.25 & @xmath82 & @xmath100.60 & @xmath1016 & @xmath104 + @xmath83 & 3.22 & 0.69 & 0.33 & 1.1 & 1.0 + lrrrrr @xmath4 ( days ) & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + @xmath20 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + @xmath21 ( deg ) & 256@xmath3313 & 294@xmath3328 & 191@xmath3313 & 137@xmath3346 & 216@xmath3351 + @xmath12 ( ) & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + @xmath22 ( hjd-2,400,000 ) & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + @xmath78 ( ) & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + @xmath79 ( ) & & & & & + s. c.@xmath25 & b2v & b2v & b2v & b2v & b2v + s. c.@xmath26 & b a & b g & b f & b k & b
m + @xmath27 ( degrees ) & 3390 & 3390 & 1290 & 390 & 390 + @xmath80 ( adopted ; ) & 9 & 9 & 11 & 11 & 11 + @xmath81 ( ) & 93.9 & 93.9 & 111.6 & 11 - 0.3 & 116 + @xmath30 & 10.43 & 10.43 & 10.14 & 10.026 & 0.550.85 + @xmath31 ( au ) & @xmath100.49 & @xmath100.49 & @xmath100.32 & @xmath100.08 & @xmath100.07 + @xmath83 & 1.9 & 1.9 & 0.95 & 0.92 & 0.74 + lrrrrr @xmath4 ( days ) & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + @xmath20 & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + @xmath21 ( deg ) & 251@xmath3332 & 299@xmath3314 & 240@xmath3352 & 260@xmath339 & 244@xmath3331 + @xmath12 ( ) & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + @xmath22 ( hjd-2,400,000 ) & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + @xmath78 ( ) & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + @xmath79 ( ) & & & & & + s. c.@xmath25 + & b0v & o8v & b1v & o6v + s. c.@xmath26 & o k & b : & o k & b : & o6v - a + @xmath27 ( degrees ) & 290 & 1990 & 290 & 2390 & 590 + @xmath80 ( adopted ; ) & 21 & 18 & 21 & 14 & 31 + @xmath81 ( ) & 210.5 & 174.1 & 210.32 & 14 4.0 & 311.8 + @xmath30 & 10.02 & 10.23 & 1 - 0.015 & 10.29 & 10.06 + @xmath31 ( au ) & @xmath100.7 & @xmath100.55 & @xmath100.15 & @xmath100.35 & @xmath100.7 + @xmath83 & 0.32 & 0.48 & 0.17 & 0.85 & 0.25 + lrrrrr @xmath4 ( days ) & [email protected] & 1687@xmath3351 & 4066@xmath3345 & [email protected] & [email protected] + @xmath20 & [email protected] & [email protected] & 0.75 ( fixed ) & [email protected] & [email protected] + @xmath21 ( deg ) & 3@xmath3334 & 255@xmath3328 & 352@xmath339 & 2@xmath3313 & 325@xmath33118 + @xmath12 ( ) & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + @xmath22 ( hjd-2,400,000 ) & [email protected] & 64990@xmath33438 & 52229@xmath3329 & 57208@xmath3349 & [email protected] + @xmath78 ( ) & [email protected] & [email protected] & [email protected] & [email protected] & [email protected] + @xmath79 ( ) & & & & & + s. c.@xmath25 & o6v & o8.5v+o9v : & o8v & o8v & b2v + s. c.@xmath26 & o6v - a & o b & o b & o b2 & o - a1 + @xmath27 ( degrees ) & 690 & 1290 & 4090 & 3490 & 2490 + @xmath80 ( adopted ; ) & 31 & 19 + 19 & 21 & 21 & 11 + @xmath81 ( ) & 312.0 & 395 & 2111 & 2110 & 113.3 + @xmath30 & 10.06 & 10.12 & 10.54 & 10.49 & 10.30 + @xmath31 ( au ) & @xmath100.14 & @xmath1010 & @xmath1016 & @xmath107.5 & @xmath100.6 + @xmath83 & 0.87 & 0.61 & 0.58 & 0.94 & 0.84 + lrrrr @xmath4 ( days ) & [email protected] & [email protected] & [email protected] & [email protected] + @xmath20 & [email protected] & 0.67 ( fixed ) & [email protected] & [email protected] + @xmath21 ( deg ) & 242@xmath3316 & 41@xmath3320 & 260@xmath338 & 15@xmath3310 + @xmath12 ( ) & [email protected] & [email protected] & [email protected] & [email protected] + @xmath22 ( hjd-2,400,000 ) & [email protected] & [email protected] & [email protected] & [email protected] + @xmath78 ( ) & [email protected] & [email protected] & [email protected] & [email protected] + @xmath79 ( ) & & & [email protected] & + s. c.@xmath25 & b0v & o9.5iii & b1v & o7v + s. c.@xmath26 & b a & o b & b1.5v & a b + @xmath27 ( degrees ) & 1590 & 1890 & 7590 & 1590 + @xmath80 ( adopted ; ) & 18 & 21 & 14 & 25 + @xmath81 ( ) & 183.2 & 214.1 & @xmath1012 & 254.0 + @xmath30 & 10.18 & 10.20 & [email protected] & 10.16 + @xmath31 ( au ) & @xmath102.2 & @xmath103.7 & @xmath100.6 & @xmath101.8 + @xmath83 & 1.0 & 3.4 & 4.5 & 0.60 + lrr ptabularcr ptabularcr ptabularcr 52146.686 & -14.1 & 5.2 + 52161.793 & -8.8 & 3.5 + 52162.651 & -5.2 & 6.6 + 53338.617 & -31.6 & 9.1 + 53989.652 & -6.6 & 5.3 + 53990.848 & -1.6 & 4.1 + 54285.925 & -23.4 & 4.8 + 54286.904 & -19.6 & 3.9 + 54341.887 & -17.1 & 3.5 + 54342.777 & -16.5 & 3.7 + 54343.746 & -15.6 & 3.6 + 54344.775 & -22.4 & 4.2 + 54345.790 & -17.1 & 3.5 + 54347.789 & -21.9 & 4.4 + 54348.812 & -20.6 & 3.7 + 54399.696 & -24.8 & 8.1 + 54401.649 & -22.4 & 7.9 + 54402.719 & -21.0 & 3.4 + 54641.844 & -22.2 & 1.8 + 54643.718 & -22.5 & 5.4 + 54644.789 & -16.4 & 3.9 + 54647.828 & -19.5 & 4.3 + 54669.831 & -19.6 & 2.7 + 54747.682 & -23.4 & 3.6 + 54748.641 & -22.8 & 2.3 + 54753.648 & -37.1 & 3.1 + 54754.717 & -22.4 & 1.7 + 54755.763 & -23.1 & 2.1 + 54756.690 & -22.4 & 4.0 + 54757.747 & -29.3 & 3.9 + ptabularcr ptabularcr ptabularcr 54643.768 & -12.4 & 6.0 + 54644.844 & -3.8 & 6.1 + 54669.787 & -2.9 & 6.7 + 54672.812 & -4.3 & 5.9 + 56511.725 & -22.8 & 3.9 + 56512.739 & -17.7 & 2.1 + 56519.639 & -21.1 & 9.6 + 56533.721 & -16.9 & 8.3 + 56534.739 & -8.7 & 5.0 + 56556.722 & -28.8 & 3.1 + 56574.599 & -12.3 & 5.4 + 56582.571 & -12.5 & 5.5 + 56588.703 & -21.2 & 11.0 + 56589.54 & -18.0 & 6.5 + 56590.665 & -24.2 & 3.3 + 56599.693 & -21.9 & 7.4 + 56607.561 & -22.4 & 6.6 + 56626.536 & -3.8 & 8.1 + 56798.871 & 14.6 & 11.5 + 56804.946 & -7 & 8.9 + ptabularcr ptabularcr ptabularcr 51381.776 & -9.2 & 2.5 + 51382.947 &
-11.4 & 2.4 + 51411.719 & -12.9 & 2.4 + 51412.762 & -19.1 & 2.0 + 51736.792 & -18.2 & 2.6 + 51737.871 & -7.9 & 1.3 + 52146.829 & -12.0 & 2.0 + 52161.650 & -12.5 & 3.0 + 52162.828 & -9.6 & 2.9 + 53339.572 & -14.7 & 3.0 + 54642.776 & -23.0 & 4.3 + 54643.792 & -12.5 & 3.0 + 54644.736 & -10.0 & 3.2 + 54645.779 & -1.2 & 3.6 + 54646.755 & -15.7 & 3.0 + 54647.726 & -12.9 & 1.6 + 54648.830 & -10.7 & 3.1 + 54669.770 & -26.4 & 5.2 + 54672.734 & -23.6 & 2.8 + 54673.738 & -11.8 & 2.3 + 54674.770 & -16.3 & 5.8 + ptabularcr ptabularcr ptabularcr 51365.094 & -8.1 & 1.5 + 51467.700 & -24.0 & 2.2 + 51805.727 & 7.3 & 1.7 + 52146.829 & -1.8 & 5.0 + 52162.828 & 4.5 & 4.4 + 53339.572 & -15.0 & 5.7 + 54285.779 & 6.7 & 8.1 + 54642.793 & -7.3 & 3.1 + 54644.721 & 6.2 & 3.7 + 54647.873 & -3.9 & 2.7 + 54669.894 & -7.3 & 2.8 + 54672.673 & -5.5 & 4.2 + 54673.819 & -6.5 & 3.6 + 54674.704 & 0.4 & 3.5 + 56438.944 & -11.0 & 2.0 + 56457.720 & -21.4 & 2.4 + 56460.945 & -5.3 & 3.8 + 56462.751 & -6.0 & 2.1 + 56466.921 & -15.5 & 2.5 + 56486.860 & -4.1 & 3.8 + 56493.800 & 0.8 & 2.4 + 56494.899 & -1.8 & 3.2 + 56495.706 & -9.0 & 4.0 + 56496.680 & -4.3 & 2.8 + 56520.906 & -2.3 & 4.5 + 56533.953 & -3.9 & 3.1 + 56556.883 & -18.0 & 4.1 + 56561.636 & -15.8 & 3.3 + 56571.636 & -4.1 & 7.3 + 56574.827 & -4.7 & 2.8 + ptabularcr ptabularcr ptabularcr 54643.850 & 2.5 & 5.2 + 54644.908 & -12.5 & 5.7 + 54645.874 & -12.3 & 5.4 + 54647.792 & -12.0 & 5.0 + 54671.914 & -3.7 & 4.5 + 54672.755 & -3.8 & 4.8 + 54673.759 & -5.1 & 4.4 + 54674.829 & 9.9 & 4.8 + 54747.789 & -17.4 & 3.3 + 54748.685 & -9.6 & 3.5 + 54749.582 & -7.8 & 3.5 + 54753.624 & -11.2 & 3.0 + 54754.754 & 4.1 & 4.3 + 54755.655 & 1.7 & 5.8 + 54756.677 & -8.2 & 3.5 + 55726.901 & -7.8 & 2.7 + 55738.919 & 17.8 & 5.1 + 55741.732 & -4.9 & 6.1 + 55779.962 & -13.2 & 3.0 + 55791.962 & -6.3 & 3.6 + 55794.878 & -4.2 & 7.5 + 55834.592 & -17.6 & 4.1 + 55847.807 & -27.4 & 2.7 + 55855.665 & -7.8 & 10.8 + 55866.534 & -0.3 & 2.5 + 55901.662 & -1.2 & 3.2 + 56118.721 & -2.9 & 3.2 + 56205.506 & -10.8 & 4.2 + 56209.837 & -16.3 & 3.8 + 56212.651 & -3.6 & 3.4 + 56219.824 & -17.9 & 3.1 + 56444.745 & -11.9 & 4.2 + 56445.936 & 3.0 & 3.8 + 56457.688 & 1.8 & 3.3 + 56460.696 & -8.9 & 3.4 + 56462.768 & -12.5 & 4.2 + 56464.925 & -10.9 & 4.5 + 56465.927 & -11.1 & 6.2 + 56468.880 & -6.0 & 3.6 + 56470.917 & -10.6 & 3.0 + 56472.889 & -10.8 & 3.3 + 56475.750 & 2.6 & 2.9 + ptabularcr ptabularcr ptabularcr 54642.809 & -8.4 & 4.8 + 54643.873 & -8.1 & 3.9 + 54644.713 & -6.4 & 4.7 + 54645.759 & 2.0 & 4.0 + 54646.738 & -17.3 & 3.7 + 54647.709 & -10.6 & 3.2 + 54648.810 & -14.1 & 3.5 + 54669.956 & -11.2 & 3.4 + 54671.710 & -14.6 & 3.2 + 54672.654 & -8.1 & 1.7 + 54673.787 & -3.6 & 2.6 + 54674.666 & -10.0 & 2.6 + 54675.777 & -0.6 & 2.8 + 54746.768 & 17.9 & 2.7 + 54747.630 & 4.7 & 2.6 + 54748.617 & -19.5 & 2.0 + 54748.805 & -15.5 & 1.9 + 54749.573 & -8.0 & 2.9 + 54753.599 & 1.8 & 2.1 + 54754.602 & -8.5 & 2.3 + 54755.642 & -0.0 & 2.6 + 54756.662 & 3.8 & 2.1 + 54757.639 & -9.4 & 2.3 + 54758.702 & -9.2 & 2.4 + 55010.732 & -12.2 & 1.6 + 55011.632 & 8.7 & 1.6 + 55012.637 & -2.5 & 2.4 + 55013.635 & -10.1 & 2.2 + 56212.561 & -0.0 & 3.5 + 56258.623 & -10.8 & 3.9 + lcccccclc mt91 021 & sb1 & b1.5v + b + ? & [email protected] & [email protected] & [email protected] & 0.181 & 0 & a + mt91 059 & sb1 & o8v + b & [email protected] & [email protected] & [email protected] & 0.230.67 & 2,0 & b + mt91 070 & sb1 & o9iii + b & 6.19 yr & 9@xmath331 & [email protected] & 0.251 & 1 & + mt91 103 & sb2 & b1v + b2v : & [email protected] & [email protected] & [email protected] & [email protected] & 1 & + mt91 145 & sb1 & o9iii + mid b & [email protected] & [email protected] & [email protected] & 0.230.63 & 3,0 & c + mt91 174 & sb1 & b2iii + g b & [email protected] & 9@xmath331 & [email protected] & 0.030.9 & 1 & + mt91 187 & sb1 & b1v + g b & [email protected] & [email protected] & [email protected] & 0.0181 & 0 & d + mt91 202 & sb1 & b2v + a
b & [email protected] & [email protected] & [email protected] & 0.181 & 0 & + mt91 234 & sb1 & b1.5v + early b & 4996@xmath33330 & [email protected] & [email protected] & 0.91 & 0 & + mt91 241 & sb1 & b1.5v + early b & 671@xmath332 & [email protected] & [email protected] & 0.420.8 & 0 & + mt91 252 & sb2 & b2iii + b1v & [email protected] & 70@xmath332 & [email protected] & [email protected] & 2,0 & b + mt91 258 & sb1 & o8v + b & [email protected] & [email protected] & [email protected] & 0.180.89 & 2,0 & b + mt91 267 & sb1 & o7.5iii
i + o / b & [email protected] & 24@xmath332 & [email protected] & 0.091 & 1 & + mt91 268c1 & sb1 & b2v + b & [email protected] & [email protected] & [email protected] & 0.431 & 0 & a + mt91 268c2 & sb1 & b2v + b g & [email protected] & [email protected] & [email protected] & 0.061 & 0 & a + mt91 292 & sb1 & b2v + b f & [email protected] & [email protected] & [email protected] & 0.141 & 0 & + mt91 295 & sb1 & b1.5v + g b & [email protected] & [email protected] & [email protected] & 0.031 & 0 & a ? + mt91 311 & sb2 & b2v + b3v & [email protected] & 88@xmath335 & [email protected] & [email protected] & 4 & + mt91 336 & sb1 & b2v + b m & [email protected] & [email protected] & [email protected] & 0.550.85 & 0 & + mt91 339 & sb1 & o8v + o g & [email protected] & [email protected] & [email protected] & 0.021 & 0 & + mt91 372 & sb2/ea & b0v + b2:v & [email protected] & [email protected] & [email protected] & [email protected] & 3,5,0 & e + mt91 378 & sb1 & b0v + b : & [email protected] & [email protected] & [email protected] & 0.231 & 0 & + mt91 390 & sb1 & o8v + g o & [email protected] & [email protected] & [email protected] & 0.0151 & 0 & + mt91 403 & sb1 & b1v + b & [email protected] & [email protected] & [email protected] & 0.291 & 0 & + mt91 417b & sb1 & o6v + a
o + mt91 417a & [email protected] & [email protected] & [email protected] & 0.061 & 0 & + mt91 421 & sb1/ea & o9v + b9v a0v & 4.161 & & & @xmath100.160.19 & 6 & + mt91 429 & sb1/ea & b0v + ( b3v+b6v ) & [email protected] & 140@xmath3315 & [email protected] & 0.057 : & 4,6 & a + mt91 431 ( s9 ) & sb2 & o5i + o4iii & [email protected] & [email protected] & [email protected] & [email protected] & 18,19 & + mt91 448 & sb1 & o6v + early a o & [email protected] & [email protected] & [email protected] & 0.061 & 0 & + mt91 465 ( s8a ) & sb2 & o5.5i + o6 : & 21.908 ( fixed ) & [email protected] & [email protected] & [email protected] & 16,17 & + mt91 473 ( s8d ) & sb1 & o8.5v + o9v + a2v o & 1687@xmath3351 & [email protected] & [email protected] & 0.121 & 0 & a + mt91 485 & sb1 & o8v + early b o & 4066@xmath3345 & [email protected] & [email protected] & 0.541 & 0 & + mt91 555 & sb1 & o8v + b2o & [email protected] & [email protected] & [email protected] & 0.491 & 0 & + mt91 561 & sb1 & b2v + b o & [email protected] & [email protected] & [email protected] & 0.301 & 0 & + mt91 588 & sb1 & b0v + early a b & [email protected] & [email protected] & [email protected] & 0.181 & 0 & + mt91 601 & sb1 & b0iab + b o & [email protected] & [email protected] & [email protected] & 0.201 & 0 & f + mt91 605 & sb2 & b1v + b1 : & [email protected] & 44@xmath333 & [email protected] & [email protected] & 4 & + mt91 646 & sb2 & b1v + b1.5v & [email protected] & [email protected] & [email protected] & [email protected] & 0 & + mt91 696 & sb2/ew & o9.5v + b0v & [email protected] & 250@xmath3310 & 0 ( fixed ) & [email protected] & 4,7 & a , e + mt91 720 & sb2 & b0b1 + b1b2 & [email protected] & 173@xmath335 & [email protected] & [email protected] & 2,4 & + mt91 734 ( s11 ) & sb1 & o5i + o / early b & [email protected] & 26@xmath331 & [email protected] & 0.15 1 & 1 & + mt91 745 & sb1 & o7v + b o & [email protected] & [email protected] & [email protected] & 0.161 & 0 & + mt91 771 & sb2 & o7v + o9v & [email protected] & 139@xmath335 & [email protected] & [email protected] & 2,4 & + schulte # 3 & sb2/ea : & o6iv : + o9iii & [email protected] & 96@xmath335 & [email protected] & [email protected] & 2,8 & a , e + schulte # 5 & sb2/eb & o7ianfp + ofpe / wn9 & 6.6 ( fixed ) & [email protected] & 0.0 ( fixed ) & [email protected] & 9,10,11,12 & a + & & ( + b0v : ) & [email protected] & & [email protected] & 0.061 & 13,14,15 & + schulte # 73 & sb2 & o8iii + o8iii & [email protected] & 84@xmath331 & [email protected] & [email protected] & 3 & + cpr2002 a36 & sb2/ea & b0ib + b0iii & [email protected] & 150@xmath336 & 0 ( fixed ) & [email protected] & 3,20,21 & e + cpr2002 a45 & sb2 & b0.5v + b1v : b2v : & [email protected] & 138@xmath333 & [email protected] & [email protected] & 3,21 & + cpr2002 b17 & sb2/eb : & o7 : + o9 : & [email protected] & 257@xmath337 & 0 ( fixed ) & 0.75 ( fixed ) & 22,23 & + lcrrr ptabularcr ptabularcr ptabularcr 51467.913 & 0.294 & 37.0 & 5.2 & 5.52 + 51805.750 & 0.918 & -124.7 & 5.4 & -16.85 + 52146.697 & 0.182 & 7.5 & 5.1 & 3.36 + 52162.650 & 0.470 & 27.7 & 5.3 & 7.63 + 53338.617 & 0.819 & -87.0 & 4.2 & 7.59 + 53570.693 & 0.646 & -16.0 & 9.7 & 16.08 + 53573.675 & 0.261 & 27.7 & 11.8 & 0.98 + 53633.634 & 0.618 & -34.6 & 4.3 & -12.84 + 53634.676 & 0.832 & -82.8 & 9.4 & 15.38 + 53636.753 & 0.260 & 31.7 & 6.8 & 5.08 + 53657.660 & 0.569 & 10.8 & 4.1 & 16.50 + 53903.839 & 0.303 & 9.8 & 14.5 & -22.42 + 53904.708 & 0.482 & -28.7 & 9.3 & -46.15 + 53905.821 & 0.711 & -65.2 & 6.5 & -8.51 + 53906.717 & 0.896 & -118.0 & 11.7 & -9.98 + 53907.769 & 0.113 & -34.3 & 16.6 & -5.98 + 53907.864 & 0.132 & -33.3 & 8.9 & -15.20 + 53932.831 & 0.278 & 39.7 & 9.2 & 10.28 + 53932.839 & 0.279 & 46.5 & 20.2 & 16.85 + 53935.775 & 0.885 & -97.4 & 16.5 & 9.90 + 53987.695 & 0.584 & -25.2 & 14.1 & -14.63 + 53988.695 & 0.791 & -97.2 & 9.1 & -11.43 + 53989.652 & 0.988 & -87.2 & 4.3 & 5.40 + 53990.848 & 0.234 & 15.2 & 5.0 & -5.73 + 54285.925 & 0.045 & -55.4 & 7.0 & 10.35 + ptabularcr ptabularcr ptabularcr 51381.906 & 0.869 & 10.7 & 2.1 & 1.16 + 51383.775 & 0.943 & -16.6 & 2.5 & 0.48 + 51467.903 & 0.291 & -29.0 & 1.1 & 0.20 + 51736.897 & 0.997 & -47.9 & 2.9 & -0.14 + 51737.961 & 0.039 & -64.9 & 3.9 & -0.61 + 52146.686 & 0.306 & -27.1 & 1.8 & -0.49 + 52161.793 & 0.908 & -2.8 & 1.6 & -1.49 + 52162.650 & 0.942 & -17.1 & 3.2 & -0.72 + 53338.617 & 0.744 & 16.1 & 2.0 & -1.62 + 53340.576 & 0.822 & 10.1 & 1.6 & -5.79 + 53570.879 & 0.988 & -46.4 & 4.2 & -3.53 + 53632.836 & 0.454 & -8.4 & 3.5 & -2.74 + 53633.714 & 0.489 & -14.1 & 4.2 & -12.50 + 53636.865 & 0.614 & 13.6 & 5.3 & 3.07 + 53657.691 & 0.443 & -4.7 & 8.0 & 2.27 + 53987.694 & 0.577 & 6.8 & 2.1 & -0.53 + 53988.740 & 0.619 & 11.9 & 1.9 & 1.02 + 53989.651 & 0.655 & 15.5 & 1.8 & 1.94 + 53990.848 & 0.703 & 20.5 & 2.7 & 4.19 + 54285.925 & 0.446 & -2.6 & 3.2 & 3.97 + 54286.655 & 0.475 & -3.3 & 2.5 & -0.17 + 54287.721 & 0.518 & 6.0 & 2.9 & 4.46 + 54397.705 & 0.895 & 6.6 & 1.5 & 3.78 + 54399.685 & 0.974 & -36.7 & 1.6 & -2.05 + 54401.662 & 0.053 & -66.2 & 1.2 & 0.38 + 54402.711 & 0.094 & -66.7 & 1.2 & -0.36 + ptabularcr ptabularcr ptabularcr 51365.058 & 0.607 & -63.3 & 1.4 & -0.6 + 54341.855 & 0.038 & 74.4 & 3.7 & -0.4 + 54342.841 & 0.141 & 60.7 & 4.4 & 4.4 + 54343.789 & 0.240 & 17.4 & 3.3 & 0.4 + 54345.845 & 0.455 & -43.1 & 4.4 & 1.2 + 54346.880 & 0.563 & -56.1 & 4.7 & 3.6 + 54348.783 & 0.762 & -49.6 & 3.8 & 3.2 + 54410.722 & 0.242 & 18.8 & 4.1 & 2.8 + 54696.705 & 0.162 & 44.5 & 4.0 & -3.7 + 54697.690 & 0.265 & 14.9 & 3.1 & 7.2 + 54698.826 & 0.384 & -30.7 & 5.1 & -2.0 + 54700.838 & 0.594 & -55.8 & 4.1 & 6.2 + 54701.798 & 0.695 & -55.0 & 4.1 & 7.0 + 54747.741 & 0.501 & -62.1 & 5.5 & -10.0 + 54748.781 & 0.610 & -62.0 & 4.3 & 0.8 + 54755.716 & 0.335 & -27.3 & 3.5 & -11.7 + 54757.701 & 0.543 & -57.7 & 5.0 & -0.1 + 55866.304 & 0.524 & -61.2 & 5.8 & -5.9 + ptabularcr ptabularcr ptabularcr 51365.058 & 0.607 & 55.0 & 2.3 & 0.9 + 54341.855 & 0.038 & -120.7 & 6.2 & -15.6 + 54342.841 & 0.141 & -61.5 & 7.4 & 22.4 + 54343.789 & 0.240 & -42.1 & 5.6 & -3.9 + 54345.845 & 0.455 & 27.8 & 7.3 & -5.1 + 54346.880 & 0.563 & 52.2 & 7.9 & 1.6 + 54348.783 & 0.762 & 48.6 & 6.4 & 6.0 + 54410.722 & 0.242 & -40.7 & 7.0 & -3.6 + 54696.705 & 0.162 & -60.2 & 6.5 & 14.1 + 54697.690 & 0.265 & -41.0 & 5.1 & -13.5 + 54700.838 & 0.594 & 49.8 & 6.9 & -3.5 + 54701.798 & 0.695 & 50.6 & 6.8 & -2.8 + 54747.741 & 0.501 & 34.2 & 9.4 & -7.7 + 54748.781 & 0.610 & 50.1 & 7.1 & -4.2 + 54757.701 & 0.543 & 47.4 & 8.2 & -0.8 + ptabularcr ptabularcr ptabularcr 51467.915 & 0.647 & 2.1 & 2.8 & 16.48 + 51805.748 & 0.694 & -10.0 & 1.7 & -5.46 + 52146.686 & 0.953 & 21.4 & 6.6 & -6.22 + 52161.793 & 0.983 & 24.5 & 3.8 & 1.63 + 52162.651 & 0.042 & 8.3 & 4.1 & 0.38 + 53338.617 & 0.266 & -47.1 & 3.5 & -0.89 + 53340.576 & 0.400 & -49.6 & 2.8 & -1.28 + 53987.695 & 0.547 & -37.2 & 4.6 & -4.74 + 53988.741 & 0.618 & -15.8 & 4.7 & 4.21 + 53989.652 & 0.680 & -10.3 & 4.7 & -2.89 + 53989.774 & 0.689 & -2.5 & 4.1 & 3.14 + 53990.848 & 0.762 & 10.1 & 5.1 & -0.00 + 54342.873 & 0.777 & 6.4 & 3.5 & -6.85 + 54343.759 & 0.838 & 31.9 & 2.6 & 7.81 + 54345.811 & 0.978 & 17.9 & 3.8 & -6.04 + 54347.802 & 0.113 & -14.6 & 2.6 & 0.01 + 54285.925 & 0.892 & 28.3 & 3.7 & -1.25 + 54286.656 & 0.942 & 32.8 & 9.1 & 4.12 + 54287.721 & 0.015 & 14.6 & 8.9 & -1.03 + ptabularcr ptabularcr ptabularcr 54628.748 & 0.991 & -76.3 & 9.2 & -3.9 + 54629.866 & 0.023 & -108.9 & 10.5 & -6.3 + 54630.739 & 0.048 & -106.4 & 4.7 & 3.7 + 54631.855 & 0.080 & -95.6 & 8.2 & 8.5 + 54633.859 & 0.137 & -77.2 & 5.1 & 0.6 + 54628.880 & 0.994 & -76.7 & 3.9 & 0.3 + 54629.759 & 0.020 & -107.0 & 25.8 & -6.4 + 54630.852 & 0.051 & -112.9 & 3.0 & -2.7 + 54633.741 & 0.134 & -75.4 & 9.9 & 4.0 + 54674.865 & 0.313 & -14.0 & 3.2 & -3.4 + 54675.729 & 0.338 & 3.3 & 7.4 & 7.2 + 54696.744 & 0.940 & -9.1 & 3.0 & -0.9 + 54696.955 & 0.946 & -19.6 & 5.0 & -4.5 + 54697.763 & 0.969 & -39.0 & 4.1 & 5.3 + 54697.841 & 0.972 & -55.6 & 9.1 & -8.3 + 54697.913 & 0.974 & -40.9 & 4.4 & 9.2 + 54698.705 & 0.996 & -75.0 & 4.3 & 4.0 + 54698.905 & 0.002 & -86.7 & 4.6 & -1.5 + 54699.713 & 0.025 & -102.0 & 5.5 & 1.9 + 54699.923 & 0.031 & -108.2 & 5.5 & -1.6 + 54700.713 & 0.054 & -108.6 & 4.1 & 1.5 + 54701.718 & 0.083 & -104.6 & 5.1 & -1.5 + 54701.934 & 0.089 & -100.6 & 7.1 & 0.1 + 54724.680 & 0.741 & 61.6 & 4.9 & 4.3 + 54724.800 & 0.744 & 63.9 & 6.3 & 6.5 + 54725.685 & 0.770 & 60.0 & 4.5 & 2.6 + 54725.794 & 0.773 & 58.3 & 4.4 & 0.9 + 54726.694 & 0.799 & 56.1 & 4.3 & 0.0 + 54726.817 & 0.802 & 56.6 & 4.5 & 0.8 + 54729.653 & 0.884 & 34.9 & 5.3 & -0.6 + 54729.834 & 0.889 & 36.4 & 4.7 & 3.6 + ptabularcr ptabularcr ptabularcr 54628.748 & 0.991 & 68.8 & 5.8 & 0.3 + 54630.739 & 0.048 & 99.5 & 3.0 & -6.4 + 54631.855 & 0.080 & 96.5 & 5.3 & -3.5 + 54633.859 & 0.137 & 78.0 & 3.3 & 4.2 + 54628.880 & 0.995 & 72.6 & 2.5 & -0.4 + 54630.852 & 0.051 & 100.7 & 1.9 & -5.3 + 54633.741 & 0.134 & 75.0 & 6.4 & -0.5 + 54674.865 & 0.313 & -5.9 & 3.3 & -13.0 + 54675.729 & 0.338 & -6.6 & 6.4 & -7.0 + 54696.744 & 0.940 & 1.3 & 2.8 & -3.4 + 54696.955 & 0.946 & 12.3 & 3.9 & 0.8 + 54697.763 & 0.969 & 38.4 & 3.4 & -2.2 + 54697.841 & 0.972 & 49.6 & 7.7 & 6.0 + 54697.913 & 0.974 & 46.6 & 4.1 & 0.2 + 54698.705 & 0.996 & 79.9 & 3.3 & 4.9 + 54698.905 & 0.002 & 86.8 & 3.5 & 5.6 + 54699.713 & 0.025 & 104.0 & 4.2 & 4.3 + 54699.923 & 0.031 & 112.7 & 4.4 & 10.2 + 54700.713 & 0.054 & 108.8 & 4.1 & 2.9 + 54701.718 & 0.083 & 106.6 & 4.1 & 7.6 + 54701.934 & 0.089 & 104.5 & 5.8 & 7.9 + 54724.680 & 0.741 & -61.1 & 4.9 & -0.7 + 54724.800 & 0.744 & -62.7 & 5.1 & -2.3 + 54725.685 & 0.770 & -60.4 & 4.2 & 0.1 + 54725.794 & 0.773 & -58.5 & 4.6 & 1.9 + 54726.694 & 0.799 & -61.3 & 3.6 & -2.1 + 54726.817 & 0.802 & -56.7 & 4.5 & 2.1 + 54729.653 & 0.884 & -36.3 & 5.3 & 2.3 + 54729.834 & 0.889 & -33.2 & 3.7 & 2.8 + ptabularcr ptabularcr ptabularcr 54403.658 & 0.437 & 127.7 & 5.6 & -4.6 + 54403.738 & 0.491 & 139.8 & 7.8 & -4.1 + 54405.714 & 0.806 & -60.0 & 9.1 & -5.0 + 54405.737 & 0.822 & -67.5 & 8.0 & -1.3 + 54406.642 & 0.424 & 112.5 & 10.3 & -14.3 + 54406.652 & 0.431 & 106.3 & 10.2 & -23.5 + 54406.770 & 0.509 & 156.0 & 10.7 & 12.4 + 54409.739 & 0.486 & 130.1 & 10.1 & -13.5 + 54410.675 & 0.110 & -104.7 & 5.4 & -0.8 + 54641.836 & 0.031 & -133.5 & 6.6 & -4.0 + 54642.901 & 0.740 & -2.2 & 7.3 & -1.3 + 54643.707 & 0.277 & 6.9 & 8.1 & -11.9 + 54647.805 & 0.005 & -150.0 & 5.5 & -18.1 + 54672.895 & 0.712 & 9.5 & 4.7 & -14.2 + 54724.704 & 0.210 & -16.2 & 5.9 & 21.9 + 54724.776 & 0.258 & 1.2 & 4.7 & -0.8 + 54725.713 & 0.881 & -89.0 & 4.7 & 13.5 + 54729.700 & 0.536 & 149.5 & 5.6 & 10.5 + 54746.635 & 0.813 & -46.5 & 8.1 & 13.2 + 54747.583 & 0.444 & 132.0 & 6.4 & -2.7 + 54748.591 & 0.115 & -108.8 & 7.9 & -7.6 + 54754.767 & 0.227 & -25.0 & 6.2 & -1.4 + ptabularcr ptabularcr ptabularcr 54403.658 & 0.437 & -194.2 & 9.8 & -7.9 + 54403.738 & 0.491 & -185.2 & 10.9 & 17.3 + 54406.642 & 0.424 & -195.4 & 17.3 & -16.7 + 54406.652 & 0.431 & -202.9 & 16.9 & -20.2 + 54406.770 & 0.509 & -179.1 & 16.1 & 22.9 + 54409.739 & 0.486 & -220.0 & 15.8 & -17.9 + 54410.675 & 0.110 & 143.1 & 9.7 & -1.1 + 54641.836 & 0.031 & 196.2 & 12.8 & 16.3 + 54642.901 & 0.740 & -5.9 & 12.3 & -5.9 + 54643.707 & 0.277 & 4.4 & 16.4 & 31.9 + 54647.805 & 0.005 & 178.3 & 15.9 & -5.0 + 54672.895 & 0.712 & 1.8 & 10.4 & 36.1 + 54724.704 & 0.210 & 29.1 & 11.3 & -23.1 + 54724.776 & 0.258 & 15.2 & 10.0 & 19.2 + 54725.713 & 0.881 & 145.1 & 9.6 & 2.8 + 54726.719 & 0.551 & -188.6 & 10.4 & 0.9 + 54726.791 & 0.599 & -162.1 & 13.9 & -4.7 + 54729.700 & 0.536 & -192.7 & 12.0 & 3.1 + 54746.635 & 0.813 & 68.3 & 14.5 & -14.1 + 54747.583 & 0.444 & -191.9 & 11.3 & -2.2 + 54748.591 & 0.115 & 161.1 & 17.6 & 20.7 + 54754.767 & 0.227 & 11.1 & 13.8 & -20.8 + lrrr @xmath4 ( days ) & [email protected] & [email protected] & [email protected] + @xmath20 & [email protected] & [email protected] & [email protected] + @xmath21 ( deg ) & 225@xmath3322 & 125@xmath333 & 43@xmath3328 + @xmath12 ( ) & [email protected] & [email protected] & [email protected] + @xmath22 ( hjd-2,400,000 ) & [email protected] & [email protected] & [email protected] + @xmath78 ( ) & [email protected] & [email protected] & [email protected] + @xmath79 ( ) & & & + s. c.@xmath25 & 08v & o9iii & o8v + s. c.@xmath26 & o midb & o midb & o
b + @xmath27 ( degrees ) & 1990 & 1990 & 1590 + @xmath80 ( adopted ; ) & 21 & 23 & 21 + @xmath81 ( ) & 215 & 235.3 & 213.9 + @xmath30 & 10.23 & 10.23 & 10.19 + @xmath31 ( au ) & @xmath100.18 & @xmath100.59 & @xmath100.3 + @xmath83 & 4.3 & 2.1 & 4.9 + lrrr @xmath4 ( days ) & [email protected] & [email protected] & [email protected] + @xmath20 & [email protected] & [email protected] & [email protected] + @xmath21 ( deg ) & 326@xmath338 & 133@xmath332 & 181@xmath3325 + @xmath12 ( ) & -9@xmath331 & [email protected] & -1@xmath332 + @xmath22 ( hjd-2,400,000 ) & [email protected] & [email protected] & 72825@xmath336 + @xmath78 ( ) & 70@xmath332 & 84@xmath331 & 138@xmath333 + @xmath79 ( ) & 81@xmath334 & 83@xmath331 & 193@xmath336 + s. c.@xmath25 & b12v & o8.5iii : & b0.5v + s. c.@xmath26 & b12v & o9.0iii : & b12v + @xmath27 ( degrees ) & 3040 & @xmath1040 & 4045 + @xmath85 ( ) & [email protected] & [email protected] & [email protected] + @xmath86 ( ) & [email protected] & [email protected] & [email protected] + @xmath30 & [email protected] & [email protected] & [email protected] + @xmath31 ( /rsun ) & [email protected] & [email protected] & [email protected] + @xmath83 & 2.3 & 1.3 & 2.7 + | we analyze orbital solutions for 48 massive multiple - star systems in the cygnus ob2 association , 23 of which are newly presented here , to find that the observed distribution of orbital periods is approximately uniform in @xmath0 for @xmath145 d , but it is not scale - free .
inflections in the cumulative distribution near 6 d , 14 , d , and 45 d , suggest key physical scales of @xmath20.2 , @xmath20.4 , and @xmath21 a.u . where yet - to - be - identified phenomena create distinct features .
no single power law provides a statistically compelling prescription , but if features are ignored , a power law with exponent @xmath3 provides a crude approximation over @xmath4=1.4
2000 d , as does a piece - wise linear function with a break near 45 d. the cumulative period distribution flattens at @xmath545 d , even after correction for completeness , indicating either a lower binary fraction or a shift toward low - mass companions . a high degree of similarity ( 91% likelihood ) between the cyg ob2 period distribution and that of other surveys suggests that the binary properties at @xmath625 d are determined by local physics of disk / clump fragmentation and are relatively insensitive to environmental and evolutionary factors .
fully 30% of the unbiased parent sample is a binary with period @xmath145 d. completeness corrections imply a binary fraction near 55% for @xmath15000 d. the observed distribution of mass ratios @xmath7 is consistent with uniform , while the observed distribution of eccentricities @xmath8 is consistent with uniform plus an excess of @xmath9 systems .
we identify six stars , all supergiants , that exhibit aperiodic velocity variations of @xmath1030 attributed to atmospheric fluctuations . |
the recent observation of ultracold co in the inner disk of m31 by allen et al .
( 1995 ) , and the inference that ultracold co exists in some spiral arm dustlanes where neither co nor hi have been detected in spite of a large dust extinction ( m83 : allen , atherton & tilanus 1985 ; tilanus & allen 1993 ; m51 : tilanus & allen 1989 ; m100 : rand 1995 ; m81 : allen et al .
1997 ) lead us to wonder about the form of the initial mass function ( imf ) for any stars that are born at such low temperatures . here
we propose that star formation in ultracold gas is biased towards brown dwarfs , and we show that these objects , along with the stars that are likely to form with them , could have escaped detection up to now because they do not heat the gas enough to raise the temperature to normal levels , nor do they appear as stellar objects in existing images . in the solar neighborhood ,
brown dwarfs are so uncommon that the imf can not continue to rise below about @xmath16 m@xmath6 ( zuckerman & becklin 1992 ; pound & blitz 1993 , 1995 ; reid & gazis 1997a , b ) .
the number of possible brown dwarfs is consistent with an extension of the imf flattening seen between @xmath17 m@xmath6 and 0.4 m@xmath6 ( reid 1998 )
. the relative number of brown dwarfs could be much higher in the ultracold gas of m31 , however .
the detection of numerous brown dwarfs there would be an important confirmation of recent imf models .
the imf is observed to be remarkably uniform from region to region in any one galaxy ( see reviews in massey 1998 ; elmegreen 1999a ) , covering star formation that spans a factor of @xmath18 in density ( massey & hunter 1998 ; luhman & rieke 1998 ) and a factor of @xmath19 in metallicity ( freedman 1985 ; massey , johnson & degioia - eastwood 1995 ) .
it is also about the same in different hii regions in various galaxies ( bresolin & kennicutt 1997 ) , and in many different galaxies on average ( kennicutt , tamblyn & congdon 1994 ) , as indicated by the equivalent widths of hydrogen emission lines .
detailed studies of color magnitude diagrams in the lmc and local dwarf galaxies give the same imf too ( greggio et al .
1993 ; marconi et al . 1995 ; holtzman et al . 1997 ; grillmair et al .
1998 ) .
there is apparently some uniformity in the imf with time also , because stars with a wide range of ages in our galaxy all have about the same function , as suggested by halo stars ( nissen et al .
1994 ) , and globular cluster stars ( de marchi & paresce 1997 ) .
similarly , a nearly universal imf was found from abundance ratios ( e.g. , fe / o , reflecting the ratio of low mass to high mass supernova processing ) in qso damped ly@xmath20 ( lu et al .
1996 ) and ly@xmath20 forest ( wyse 1998 ) lines , the intracluster medium ( renzini et al . 1993 ; wyse 1997 , 1998 ; but see loewenstein & mushotzky 1996 ) , and elliptical galaxies ( see review in wyse 1998 ) .
an imf biased towards high - mass stars has been suggested for starburst regions , based on the ratio of luminous to dynamical mass ( rieke et al .
1980 , 1993 ; kronberg , biermann , & schwab 1985 ; wright et al . 1988 ) , galactic evolution models ( doane & matthews 1993 ) , spectroscopic line ratios ( doyon , joseph , & wright 1994 ; smith et al . 1995 ) , and infrared excesses ( smith , herter & haynes 1998 ) .
however , a lower extinction correction for m82 makes the imf there normal ( devereux 1989 ; satyapal et al .
1995 , 1997 ) , and more recent evolutionary models ( schaerer 1996 ) , multiwavelength spectroscopy and broad - band infrared photometry ( calzetti 1997 ) , and emission line spectroscopy ( stasiska & leitherer 1996 ) give normal imfs too .
large imf shifts in starburst galaxies should also produce unobserved red populations of stars after the turnoff age reaches the stellar lifetime at the truncation mass ( charlot et al .
1993 ) , and too high an oxygen abundance ( wang & silk 1993 ) .
an imf shift towards lower mass stars has been reported for the extreme field by massey et al .
however , this result could also come from a normal imf in each star - forming region if massive stars form preferentially in high mass clouds and stop further star formation when they do ( elmegreen 1999b ) .
the extreme field imf can not be typical , because the imfs in most clusters and associations are about the same as the galaxy - integrated imfs .
there are also imf dips , gaps , and a @xmath21 cluster - to - cluster variation in the power law slope ( scalo 1998 ) , but such variations are expected statistically given the small numbers of stars that are usually included in cluster studies ( elmegreen 1999b ) .
there are few theoretical predictions of imf variations .
fabian ( 1994 ) predicted that the imf would be biased towards low mass stars at the cores of galaxy cluster cooling flows because of the low jeans mass that results from the expected high pressures there .
larson ( 1999 ) proposed that a top - heavy imf in the early universe could explain the g - dwarf problem , the high temperature and high metal abundance of intracluster gas , and the large luminosities of young elliptical galaxies .
neither of these predictions have been directly confirmed , however . in what follows
, we discuss the possible dependence of the minimum stellar mass on cloud temperature and pressure .
first we discuss the minimum mass on general terms , and then we apply the results to ultracold clouds in m31 .
a lower mass limit is required for star formation because cloud pieces in pre - stellar clouds extend down to masses much smaller than the smallest stellar mass . clumps with masses as low as @xmath22 m@xmath6 have been observed in great abundance as part of the normal , power - law clump mass function in the polaris spur ( heithausen et al .
1998 ; kramer et al . 1998 ) .
this means that the stellar mass range is only a small part of the total mass range for cloud clumps , and that there must be some physical process which limits the stellar mass at the low end .
one possibility is that the minimum stellar mass is proportional to the thermal jeans mass , @xmath23 for temperature @xmath1 and cloud - core pressure @xmath2 . in this model ,
cloud pieces much smaller than @xmath24 are not likely to become stars because they are not strongly self - gravitating ( larson 1992 ; elmegreen 1997 , 1999b )
. the mass of a star that actually forms in such a minimum unstable cloud piece can be smaller than @xmath24 , perhaps by a factor of 10 , because not all of the gas goes into stars , and because the resulting stellar system could be binary , in which case the stars would have to share the clump mass .
for this reason , @xmath24 , or some factor of order unity times @xmath24 , is identified with the mass at the break point in the imf , where the power law first becomes flat , not with the minimum mass of a star .
this is because the model predicts only a fundamental change in the physical properties of cloud clumps at this mass , but does not specify how the stars actually form inside the clumps . indeed , stellar masses are often observed to continue down to at least one - quarter of the mass of the break point , but such stars are not as common as would be expected from an extrapolation of the salpeter power - law function . the expression for @xmath24 is the bonner - ebert condition for stability of a non - magnetic , pressure - bounded isothermal sphere . magnetic forces increase this critical mass ( e.g. , mouschovias & spitzer 1976 ) , so they should not be included in the minimum value .
also , the mass - to - flux ratio varies in a turbulent cloud as a result of random compressions and magnetic diffusion , so regions with large mass - to - flux ratios will develop spontaneously and systematically over time , suggesting again that equation ( [ eq : mj ] ) , without magnetic fields , is the proper condition ( see also nakano 1998 ) .
the numerator in @xmath24 contains the thermal temperature , rather than the turbulent rms speed , because turbulence increases the critical mass , and because the smallest stars will generally form on such small scales that the relative turbulent speed is less than the sound speed anyway .
the pressure in the expression for @xmath24 is the total pressure at the boundary of the isothermal sphere .
this pressure is not well defined for a real cloud because it comes partly from turbulent interclump motions and partly from thermal pressure in the interclump and overlying media ( e.g. , hunter & fleck 1982 ; ballesteros - paredes , vazquez - semadeni & scalo 1999 ) . the value of it for any particular cloud is not well defined either because pressure varies from place to place .
nevertheless , the total pressure is much more uniform in a cloud than either the turbulent or the thermal pressures alone , which interchange roles during transient compressions , and also more uniform than the density because of the larson ( 1981 ) scaling relations , which make @xmath25 constant for density @xmath26 and rms speed @xmath27 . in taurus - like clouds , @xmath28 k@xmath29 , in gmcs , @xmath30 k@xmath29 , and in cluster - forming gmc cores , @xmath31 k@xmath29 .
these differences in @xmath2 suggest that the imf may differ as well , which means that the lowest - mass stars might be smaller in gmcs than in taurus - like clouds .
any such differences would not be large , however , because of the square root dependence of @xmath24 on @xmath2 .
small imf differences would not have been recognized either .
most stars form in gmcs under similar conditions , so their local imfs are similar .
where the conditions differ , as in taurus , there are too few stars to get an accurate imf .
in addition , the expected decrease in the mass at the peak of the imf for dense gmc cores is offset by stellar mass segregation ( a concentration of massive stars near the center ) and by a slightly increased temperature from nearby ob stars .
strong variations in @xmath24 are not expected on a galactic scale either .
the numerator in @xmath24 is roughly proportional to the cloud cooling rate and the denominator is roughly proportional to the background galactic heating rate from stars and cosmic rays ( elmegreen 1997 ; 1999b ) .
thus molecular clouds in equilibrium will all have about the same @xmath24 , as long as the galactic mass - to - light ratio is about the same ( expected variations in the imf with m / l are discussed in elmegreen 1999c ) .
the transition from the power law part of the imf to the low - mass turnover should contain important information about the star formation process , unlike the power - law part itself , which theory suggests contains only information about hierarchical cloud structure , i.e. , about the initial conditions for star formation .
the mathematical form for this transition , i.e. , the way in which the lower mass limit actually affects the shape of the imf , can not be determined from theory yet .
we have approximated this form simply by writing a probability @xmath32 that a cloud piece fails to form a star on a time scale comparable to its internal crossing time .
the functional form of this probability is unknown ( cf .
[ eq : imffun ] ) , so the exact value of the minimum critical mass can not be known yet either . differences of a factor of three in the definition for @xmath24 can easily be compensated by differences in the form of @xmath32 . nevertheless the concept that thermal pressure in a star - forming cloud ultimately limits the mass of a star is an essential part of the model , and leads to a characteristic mass at the low end of the imf that should scale with @xmath33 .
the lower mass limit comes from another source in an alternative imf model , which follows from ideas by larson ( 1982 ) and shu , adams , & lisano ( 1987 ) , namely , that protostellar winds limit the accretion onto a star and thereby set the final stellar mass at a value close to the lower limit necessary to drive such a wind ( nakano , hasegawa , & norman 1995 ; adams & fatuzzo 1996 ) .
the wind - limited mass is not the same as the deuterium - burning limit itself , which is about 0.018 m@xmath6 ( dantona & mazzitelli 1994 ) .
the lower stellar mass in this theory comes from a combination of the deuterium limit and the accretion rate , @xmath34 , for sound speed @xmath35 ( shu , adams , & lisano 1987 ) .
higher accretion rates require stronger winds and more massive stars before the accretion stops .
this sensitivity to accretion rate makes the lower mass limit depend on temperature in an analogous way to @xmath24 , so that lower temperature clouds are likely to produce lower mass stars in the wind - limited accretion model too .
the exact form of this temperature dependence and whether there is a pressure dependence are not known .
an important difference between these two models is that the random - sampling model assumes that stellar mass is determined mostly by the pre - stellar clump mass .
then the imf comes simply from random selection of these clumps in a certain order ( densest ones first ) .
there is , in fact , growing evidence that the stellar mass is determined by the pre - stellar clump mass ( casoli et al . 1986 ; myers , ladd , & fuller 1991 ; myers & fuller 1993 ; pound & blitz 1993 , 1995 ; motte , andr , & neri 1998 ; testi & sargent 1998 ) , and not by wind - limited accretion in a large reservoir .
the assumption of stellar mass proportional to clump mass is reasonable if even a modest fraction , e.g. , @xmath36% , of a clump s mass goes into the stars it forms , because then the total variability in possible stellar mass for a given clump mass is much less than the total mass range of the imf , which is a factor of 1000 . in this case
most of the imf has to come from clump mass variations and not from variations in the star - to - clump mass fraction .
moreover , even if a random and small fraction of a clump s mass gets into a star and the rest goes away in a wind or dispersed disk , the stellar mass will still be proportional to the clump mass and the imf will be unchanged ( elmegreen 1999b ) if the average of this fraction is about constant . given these differences between the two imf models , the primary way to distinguish between them is to determine the fraction of the clump mass that goes into stars . if this fraction is either constant or relatively large ( larger than the inverse of the total mass range for stars )
, then the imf must be tracing the pre - stellar clump spectrum in some way .
imf models with a minimum stellar mass that depends on the thermal temperature and pressure in the star - forming cloud predict that ultracold gas in the inner disk of m31 ( allen et al .
1995 ) should produce a great abundance of brown dwarfs .
this gas is located in the inner disk of m31 , in the midst of the large , dark dust features studied by hodge ( 1980 ) .
the surface density of stars is comparable to or higher there than it is locally because it is at a smaller galactocentric radius than the sun , and the surface density of the dark co clouds is higher than the local average too .
thus the pressure in the cloud cores , and perhaps in the average interstellar medium at this radius too , is larger than it is locally , by the product of the ratios of these two surface densities .
this pressure change lowers @xmath24 by about a factor of 3 compared to the value of @xmath24 for local star formation in molecular clouds .
usually such a pressure increase is accompanied by a larger stellar volume emissivity and cosmic ray flux in these inner disks , both of which increase @xmath1 along with @xmath2 .
but in m31 , the stellar blue surface brightness and the non - thermal radio continuum flux are lower than in the inner milky way , probably as a result of the earlier hubble type for m31 .
this implies that radiative and cosmic ray heating rates are low , so the molecular gas equilibrates at a lower temperature , around @xmath37 k ( allen et al .
1995 ; loinard & allen 1998 ) . in that case
, there is a factor of @xmath38 drop in the temperature of potential star - forming material for the ultracold co gas in m31 , and , when combined with the expected higher pressure , a resulting net decrease in @xmath24 by a factor of @xmath39 .
this gives a value of @xmath40 in the inner disk dark clouds of m31 .
if the pressure is the same in the cloud cores as it is locally , then @xmath41 m@xmath6 . observation of a significant population of @xmath42 m@xmath6 stars and brown dwarfs in the inner dark clouds of m31 would therefore support theories of the imf based on either the thermal jeans mass limit or the thermal accretion rate limit for the lowest mass star .
the random sampling model also predicts that the slope of the imf above the thermal jeans mass should be independent of @xmath1 , @xmath2 and other cloud properties since clouds presumably partition themselves in a standard way by turbulence ( for a discussion of cloud structure and turbulence , see falgarone , phillips & walker 1991 ) .
we expect the same salpeter slope , namely , @xmath43 on a @xmath44 plot , for stars with @xmath45 m@xmath6 in the m31 inner dark clouds as we find in the intermediate to high mass parts of the imf measured locally .
a simulation showing this independence between the high - mass slope and @xmath24 was shown by figure 4 in elmegreen ( 1997 ) .
thus _ the random sampling model predicts that the imf in ultracold clouds should be exactly shifted towards lower mass without any change in the slope of the power - law part . _
the observed imf can be approximated by the function @xmath46 for constant @xmath47 and for @xmath20 in the range from 1 to 2 ; @xmath48 for an imf that flattens at low mass , and @xmath49 for an imf that turns over at low mass .
there are only a few observations yet of a low mass turnover on a @xmath44 plot ( reid & gazis 1997 ; hillenbrand 1997 ; nota et al . 1998 ) , but numerous observations show a flattening . note that there are usually many stars in this flattened region , and that the minimum stellar mass can be as low as one - quarter the mass at the turnover point . in the model , @xmath20 appears only in the probability for failure to form a star ( @xmath50 ) .
equation ( [ eq : imffun ] ) can be integrated to give the total stellar mass in various mass ranges .
if @xmath11 m@xmath6 and the brown dwarf limit is @xmath51 m@xmath6 , then 0.479 of all the mass @xmath9 will be between @xmath24 and @xmath52 .
_ approximately half the stellar - like mass down to the peak in the imf would be in the form of brown dwarfs . _
the formation of a sufficiently large density of brown dwarfs in an ultracold co cloud could raise the gas temperature and make it more `` normal . ''
the energy input required to heat a molecular cloud to @xmath19k is about @xmath53 erg @xmath54 s@xmath55 for molecular density @xmath56 ( neufeld , lepp & melnick 1995 ) .
this corresponds to a ratio of luminosity to mass equal to @xmath57 erg s@xmath55 g@xmath55 .
if the ratio of the stellar luminosity that heats the gas divided by the gas mass , @xmath58 , from brown dwarfs and other young stars is close to this , then the gas would not be supercold .
a typical @xmath59 for pre - main sequence brown dwarfs is @xmath17 in these units ( dantona & mazzitelli 1994 ) , so if the star formation efficiency is 1% and a high fraction of the stellar radiation heats the gas , then the stellar luminosity per unit total mass would be 0.01 erg s@xmath55 g@xmath55 , which is 16 times larger than the ratio needed to heat the cloud to 10k .
most embedded stellar radiation does not go into the gas directly , though , it goes into the dust which radiates it away in the ir .
the dust luminosity scales approximately with dust temperature as @xmath60 ( e.g. , hollenbach & mckee 1979 ) , and a typical cold dust temperature in m31 is @xmath61 k ( haas et al .
if the dust and gas are thermally coupled in ultracold co clouds , then the dust luminosity has to be less in @xmath38k clouds than it is locally by a factor of about @xmath62 .
the background radiation field in m31 is brighter than @xmath63 times the local value , so the dust and gas can not be well coupled in the m31 ultracold co clouds .
this means that the average gas density in these clouds is much less than @xmath64 @xmath54 .
the density will be higher than this in star - forming cores , but these are shielded from outside light by dust , so the core dust temperature should be low there .
the likely range for the average cloud density can be estimated from the inclination - corrected column density of @xmath65 m@xmath6 pc@xmath66 ( loinard & allen 1998 ) . considering a typical cloud projected size of @xmath65 pc ,
the cloud thicknesses are probably somewhere between @xmath17 pc if they are thin shells , and @xmath65 pc , the thickness of the galaxy .
this puts the average cloud density between 1 and 100 m@xmath6 pc@xmath67 , which corresponds to a molecular hydrogen density of 151500 @xmath54 .
this is consistent with the result of loinard , allen & lequeux ( 1995 ) , who estimated a density of @xmath65 @xmath54 from co line ratios .
thus the average density is indeed too low for thermal coupling between the gas and dust . a better way to determine
if embedded brown dwarfs can significantly heat ultracold co is to compare the summed luminosities of these stars to the incident and embedded luminosity from field stars .
if the brown dwarf luminosity in the clouds is less than the total absorbed field star power , then the clouds would not show any excess emission in either dust or gas from the embedded young stars .
the total field star power received by a spherical cloud of radius @xmath68 is @xmath69 for field star volume emissivity @xmath70 and average pathlength @xmath71 for field star radiation .
the first contribution in the parenthesis is from external field stars , and the second is from internal field stars .
the total luminosity of embedded young stars with volume emissivity @xmath72 is @xmath73 the ratio @xmath74 has to be large for the cloud temperature to increase significantly as a result of embedded star heating .
this ratio gives a critical ratio of volume emissivities @xmath75 for embedded star heating ; here we have taken an external path length equal to @xmath76 times the cloud size , considering this as the ratio of cloud mean free path to size , or the inverse of the volume filling factor of these ultracold co clouds .
the volume emissivity of embedded stars is @xmath77 for average stellar luminosity - to - mass ratio @xmath58 , star formation efficiency @xmath78 ( the ratio of the star mass to the total cloud mass ) , and gas density @xmath26 . to find the average @xmath58 as a function of time
, we integrate the pre - main sequence stellar luminosities over the stellar mass function , @xmath79 using the alexander + rogers & iglesias , cm model for @xmath80 in dantona & mazzitelli ( 1994 ) and @xmath81 from equation [ eq : imffun ] .
these pre - main sequence models are only for @xmath82 m@xmath6 , so our results are valid only in this limit .
equation [ eq : lmburst ] gives @xmath58 for a time @xmath83 after a short burst of star formation . to get @xmath58 from continuous star formation that has lasted for a time @xmath83
, we use : @xmath84 for star formation rate by number , @xmath85 , which is proportional to @xmath86 for a uniform star formation rate .
the lower limits to these integrals are taken to be @xmath24 instead of the absolute minimum stellar mass , because the form of the imf is not well known below the peak and because lower mass brown dwarfs should not contribute much to the luminosity anyway .
the burst and continuous @xmath58 are shown in figure 1 as functions of time since the burst and as functions of age , respectively , for @xmath11 m@xmath6 , and for mass ranges 0.021 and 0.022.5 m@xmath6 . for continuous star formation , @xmath87 l@xmath6/ m@xmath88 erg s@xmath55 g@xmath55 for this imf . for the larger mass range , the burst @xmath58 levels off because massive stars dominate the light and reach the main sequence after @xmath89 years .
the volume emissivity of field stars is @xmath90 . for sb galaxies like m31 , @xmath91 l@xmath6/m@xmath6 ( roberts & haynes 1994 ) .
thus the condition for significant young star heating is @xmath92 the background stellar density in the inner m31 disk is several m@xmath6 pc@xmath67 , and the density in the ultracold co clouds is in the range 1100 m@xmath6 pc@xmath67 , given above , so @xmath93 0.550 .
this is comparable to the critical density ratio for young star heating only if @xmath94 .
_ we conclude that a brown dwarf+stellar population with @xmath95 m@xmath6 would not heat the ultracold co clouds in m31 noticeably if the star - to - cloud mass ratio is less than 10% . _
another way to get a limit on how many brown dwarfs and other young stars might be present is from a limit on the mass of the most massive star that can be associated with these clouds .
the largest stellar mass , @xmath96 , that is likely to come from an imf with a total stellar mass @xmath97 is given by @xmath98 figure 2 ( solid lines , left axis ) shows the total stellar and brown dwarf masses versus @xmath96 ( the two lines nearly overlap ) .
if the largest pre - main sequence or stellar mass that can be hidden in and around the ultracold co clouds in m31 is @xmath99 m@xmath6 , then the total mass of all the brown dwarfs+stars has to be less than @xmath100 m@xmath6 .
figure 2 ( dashed lines , right axis ) shows the numbers of brown dwarfs ( upper dashed line ) and stars as functions of @xmath96 . with @xmath101 m@xmath6 ,
_ there could be @xmath102 brown dwarfs and @xmath103 stars less massive than 2.5 @xmath104 in each of the large ultracold co clouds in m31_.
this number of @xmath102 hidden brown dwarfs in ultracold co clouds may seem large compared to what has been found so far in the solar neighborhood , but the associated number of normal stars and the total mass in the form of stars or brown dwarfs is remarkably small for such a cloud with an estimated @xmath89 m@xmath6 of gas ( loinard & allen 1998 ) .
if the efficiency of star+brown dwarf formation is anything like it is locally , namely @xmath105% , then so many stars should have formed that the imf would have sampled out far enough into the high mass tail to produce o - type stars , which would be seen easily .
this implies that star formation is unusually inefficient in the m31 ultracold co clouds .
one possible explanation for this is that the clouds are much less dense than the excitation densities given by loinard & allen ( 1998 ) .
these authors found , on the basis of excited - state co line ratios , that the gas is optically thick in the 3 - 2 , 2 - 1 , and 1 - 0 @xmath106co lines . in that case , the density required for excitation is less by the factor @xmath107 for opacity @xmath108 than the density that makes the collision rate equal to the spontaneous transition rate @xmath47 .
thus the molecular density can be @xmath19 times less than their estimate , and the clouds much less strongly self - gravitating .
if the clouds are like galactic translucent clouds , they may form no stars at all ( e.g. , hearty et al .
1999 ) .
we can use the example provided by m31 to estimate the importance of brown dwarf formation generally .
suppose the efficiency of brown dwarf+star formation in ultracold gas is the upper limit given by the m31 clouds , which is @xmath109 in @xmath110 m@xmath6 clouds to avoid stars more massive than 2 m@xmath6 .
suppose also that half the stellar - like mass goes into brown dwarfs and that the ages of the ultracold clouds are at least @xmath111 years , which is a modest fraction of the shear time in the inner disk m31 .
then the brown dwarf formation rate is @xmath112 m@xmath6 my@xmath55 per @xmath110 m@xmath6 of ultracold gas .
this implies a gas consumption time that is very large , @xmath113 years , which means hardly any conversion of gas into stars or brown dwarfs in a hubble time . to get a significant mass in brown dwarfs , the efficiency of star formation has to be larger than this by a factor of @xmath65 , but then the resulting luminosity would heat an ultracold cloud to normal temperatures , shutting off brown dwarf production in favor of regular stars .
the formation of brown dwarfs in ultracold gas requires a negligible mass of normal stars so the gas remains cold .
this would occur naturally in the lowest mass clouds from an ensemble of clouds if the stars formed randomly .
for example , if the maximum total mass that can form in a cloud before a massive star is likely to heat the gas is @xmath65 m@xmath6 , from figure 2 , and if the efficiency is large , 10% , then the clouds that make brown dwarfs have to be smaller than @xmath114 m@xmath6 .
such small clouds are rarely isolated , however , and nearby gas can still make massive stars that provide heat . thus is it unlikely brown dwarfs can form with a significant total mass if the imf is simply shifted towards a lower peak mass with a salpeter spectrum above this .
an upper mass cutoff in star formation , which is not present in the theory , would seem to be necessary if brown dwarfs are ever found to dominate the mass in a region .
the ultracold clouds in m31 provide a good test for theories of the imf . because of their low temperatures and normal - to - high pressures
, these clouds should have a thermal jeans mass of only @xmath115 m@xmath6 , making the imf shift towards lower mass with the same slope at higher mass . in that case
, ultracold clouds should produce half of their stellar - like mass in the form of brown dwarfs .
if the total number of such dwarfs per cloud is @xmath116 , then they and their accompanying h - burning stars would not significantly heat the cloud , nor be visible in existing surveys . a deep k - band search for brown dwarfs in ultracold gas
would be necessary to see them .
considering the luminosities of such pre - main sequence stars found locally , which is @xmath117 mag dereddened ( luhman et al .
1998 ) , the k band luminosity of such a star in m31 would be @xmath118 mag plus extinction .
adams , f.c . , & fatuzzo , m. 1996 , apj , 464 , 256 allen , r. j. , atherton , p.d . , & tilanus , r. p. j. 1985 , nature , 319 , 296 allen , r. j. , le bourlot , j. , lequeux , j. , pineau des forets , g. , & roueff , e. 1995 , apj , 444 , 157 allen , r. j. , knapen , j. h. , bohlin , r. , & stecher , t. p. 1997 ,
apj , 487 , 171 ballesteros - paredes , j. , vzquez - semadeni , e. , & scalo , j. 1999 , apj , in press bresolin , f. , & kennicutt , r. c. , jr .
1997 , aj , 113 , 975 calzetti , d. 1997 , aj , 113 , 162 casoli , f. , dupraz , c. , gerin , m. , combes , f. , & boulanger , f. 1986 , a&a , 169 , 281 charlot , s. , ferrari , f. , matthews , g. j. , & silk , j. 1993 , apj , 419 , l57 dantona , f. , & mazzitelli , i. 1994 , apjs , 90 , 467 de marchi , g. , & paresce , f. 1997 , apj , 476 , l19 devereux , n. a. , 1989 , apj , 346 , 126 doane , j.s . , & matthews , w.g . 1993 ,
apj , 419 , 573 doyon , r. , joseph , r.d . , & wright , g.s .
1994 , apj , 421 , 101 elmegreen , b.g .
1997 , apj , 486 , 944 elmegreen , b. g. 1999a , in unsolved problems in stellar evolution , ed . m. livio , cambridge : cambridge univ .
press , in press elmegreen , b. g. 1999b , apj , 515 , in press elmegreen , b.g .
1999c , in evolution of galaxies on cosmological timescales , ed . j.e .
beckman and t.j .
mahoney , asp conference series , san francisco , in press .
fabian , a.c .
1994 , araa , 32 , 277 falgarone , e. , phillips , t. g. , & walker , c. k. 1991 , apj , 378 , 186 freedman , w.l . 1985 , apj , 299 , 74 greggio , l. , marconi , g. , tosi , m. , & focardi , p. 1993
, aj , 105 , 894 grillmair , c.j . , et al .
1998 , aj , 115 , 144 haas , m. , lemke , d. , stickel , m. , hippelein , h. , kunkel , m. , herbstmeier , u. , & mattila , k. 1998 , a&a , 338 , l33 hearty , t. , magnani , l. , caillault , j.p . ,
neuhuser , r. , schmitt , j.h.m.m . , & stauffer , j. 1999 .
a&a , 341 , 163 heithausen , a. , bensch , f. , stutzki , j. , falgarone , e. & panis , j. f. 1998 , a&a , 331 , 65 hillenbrand , l.a .
1997 , aj , 113 , 1733 hodge , p.w . 1980 , aj , 85 , 376 hollenbach , d. , & mckee , c.f . 1979 ,
apjs , 41 , 555 holtzman , j. a. et al .
1997 , aj,113 , 656 hunter , j. h. , jr . , & fleck , r. c. , jr .
1982 , apj , 256 , 505 kennicutt , r.c . , jr . , tamblyn , p. , & congdon , c.w .
1994 , apj , 435 , 22 kramer , c. , stutzki , j. , rhrig , r. , & corneliussen , u. 1998 , a&a , 329 , 249 kronberg , p. p. , biermann , p. , & schwab , f. r. 1985 , apj , 291 , 693 larson , r.b .
1981 , mnras , 214 , 379 larson , r.b .
1982 , mnras , 200 , 159 larson , r. b. 1992 , mnras , 256 , 641 larson , r.b . 1999 ,
mnras , in press .
loinard , l. , allen , r.j . , & lequeux , j. 1996 , a&a , 310 , 93 loinard , l. , & allen , r.j .
1998 , apj , 499 , 227 loewenstein , m. , & mushotzky , r. f. 1996 , apj , 466 , 695 lu , l. , sargent , w. , churchill , c. , & vogt , s. 1996 , apjs , 107 , 475 luhman , k. l. , & rieke , g. h. 1998 , apj , 497 , 354 luhman , k. l. , rieke , g. h. , lada , c. j. , & lada , e. a. 1998 , apj , 508 , 347 massey , p. 1998 , in the stellar initial mass function , ed . g. gilmore , i. parry , & s. ryan , cambridge : cambridge university press , p. 17
massey , p. , & hunter , d.a .
1998 , apj , 493 , 180 massey , p. , lang , c.c .
, degioia - eastwood , k. , & garmany , c.d .
1995 , apj , 438 , 188 marconi , g. , tosi , m. , greggio , l. , & focardi , p. 1995 , aj , 109 , 173 motte , f. , andr , p. , & nerti , r. 1998 , a&a , 336 , 150 mouschovias , t.c . , & spitzer , l.jr .
1976 , apj , 210 , 326 myers , p.c . ,
ladd , e.f .
, & fuller , g.a .
1991 , apj , 372 , l95 myers , p. c. , & fuller , g. a. 1993 , apj , 396 , 631 nakano , t. 1998 , apj , 494 , 587 nakano , t. , hasegawa , t. , & norman , c. 1995 , apj , 450 , 183 neufeld , d.a .
, lepp , s. , & melnick , g.j . , 1995 ,
apjs , 100 , 132 nissen , p. , gustafsson , b. , edvardsson , b. , & gilmore , g. 1994 , a&a , 285 , 440 nota , a. , sirianni , m. , leitherer , c. , de marchi , g. , & clampin , m. 1998 , baas , 192 , 84.01 pound , m.w . , & blitz , l. 1993 , apj , 418 , 328 pound , m.w . , & blitz , l. 1995 , apj , 444 , 270 rand , r.j .
1995 , aj , 109 , 2444 rieke , g.h . ,
lebofsky , m.j . ,
thompson , r.i . , low , f.j . , & tokunaga , a.t .
1980 , apj , 238 , 24 rieke , g.h . ,
loken , k. , rieke , m.j . , tamblyn , p. 1993 , aj , 412 , 99 reid , i. n. 1998 , in the stellar initial mass function , eds . g. gilmore , i. parry , & s. ryan , san francisco : asp , p. 121
reid , i.n , & gazis , j.e .
1997a , aj , 113 , 2246 reid , i.n , & gazis , j.e .
1997b , aj , 114 , 1992 renzini , a. , ciotti , l. , dercole , a. , & pellegrini , s. 1993 , apj , 419 , 52 roberts , m. , & haynes , m. 1994 .
araa , 32 , 115 satyapal , s. , et al .
1995 , apj , 448 , 611 satyapal , s. , et al . 1997 ,
apj , 483 , 148 scalo , j.s .
1998 , in the stellar initial mass function , ed .
g. gilmore , i. parry , & s. ryan , cambridge : cambridge university press , p. 201 schaerer , d. 1996 , apj , 467 , l17 shu , f.h . ,
adams , f.c . , & lizano , s. 1987 , ara&a , 25 , 23 smith , d.a .
, herter , t. , haynes , m.p . ,
beichman , c.a .
, & gautier , t.n .
, iii 1995 , apj , 439 , 623 smith , d. a. , herter , t. , & haynes , m. p. 1998
, apj , 494 , 150 stasiska , g . , & leitherer , c. 1996 , apjs , 107 , 427 tilanus , r.p.j . , & allen , r.j .
1989 , apj , 339 , l57 tilanus , r.p.j . ,
& allen , r.j .
1993 , a&a , 273 , 707 wang , b. , & silk , j. 1993 , apj , 406 , 580 wright , g.s . ,
joseph , r.d .
, robertson , n.a .
, james , p.a . , and meikle , w.p.s .
1988 , mnras , 233 , 1 wyse , r. f. g. 1997 , apj , 490 , l69 wyse , r.f.g .
1998 , in the stellar initial mass function , ed .
g. gilmore , i. parry , & s. ryan , cambridge : cambridge university press , p.89 zuckerman , b. , & becklin , e.e .
1992 , apj , 386 , 260 | a recent model for the stellar initial mass function ( imf ) , in which the stellar masses are randomly sampled down to the thermal jeans mass from hierarchically structured pre - stellar clouds , predicts that regions of ultra - cold co gas , such as those recently found in nearby galaxies by allen and collaborators , should make an abundance of brown dwarfs with relatively few normal stars .
this result comes from the low value of the thermal jeans mass , which scales as @xmath0 for temperature @xmath1 and pressure @xmath2 , considering that the hierarchical cloud model always gives the salpeter imf slope above this lower mass limit .
the ultracold co clouds in the inner disk of m31 have @xmath3k and pressures that are probably @xmath4 higher than in the solar neighborhood .
this gives a mass at the peak of the imf equal to @xmath5 m@xmath6 , well below the brown dwarf limit of @xmath7 m@xmath6 . using a functional approximation to the imf given by @xmath8 ^ 2}\right ) m^{-1.35}d\log m$ ] for @xmath9 , which fits the local imf for the expected value of @xmath10 m@xmath6
, an imf with @xmath11 m@xmath6 in m31 has 50% of the mass and 90% of the objects below the brown dwarf limit .
the brightest of the brown dwarfs in m31 should have an apparent , extinction - corrected k - band magnitude of @xmath12 mag in their pre - main sequence phase .
for typical star - formation efficiencies of @xmath13% , brown dwarfs and any associated stars up to @xmath14 m@xmath6 should not heat the gas noticeably , but if the imf continues up to arbitrarily high masses , then the star formation efficiency has to be @xmath15 to avoid heating from massive stars .
keywords : stars : brown dwarfs stars : mass function galaxies : stellar content |
many processes in chemistry and biology involve catalysis by transition metal centers . these include systems ranging from ubiquitous catalysts of organic synthesis to the active sites of most enzymes .
unfortunately , the electronic structure of such systems is often _ strongly correlated _ , which makes the theoretical study of their nature and reactivity difficult .
strong correlation is characterized by the necessity of considering more than one slater determinant to describe the electronic wavefunction _ even qualitatively_. this property renders standard methods like kohn - sham density functional theory ( dft ) or single - reference wave function methods like mp2 or ccsd(t ) unreliable , if not straight out inapplicable . to describe such systems , multi - reference ( mr ) methods
are needed . however , currently available mr methods still have various defects , which frequently force researchers into uneasy compromises regarding active space size , quantitative accuracy , reliability , intruder states , or the need to strongly truncate the studied system to make computational expense manageable . we here present new variants of multi - reference linearized coupled cluster ( mrlcc2)@xcite and @xmath1-electron valence perturbation theory ( nevpt2),@xcite which greatly alleviate these problems . by judiciously combining internal contraction ( ic)@xcite and matrix product state perturbation theory ( mpspt),@xcite we arrive at mr methods capable of efficiently and accurately describing large and complex strongly correlated systems
_ quantitatively _
( i.e. , including dynamic correlation , _ vide infra _ ) if combined with a suitable active space treatment , such as provided by dmrg reference functions . to quantitatively describe a strongly correlated system , both dynamic correlation and static correlation
must be accounted for .
however , both problems have rather different demands on the wave function ansatz . to account for static correlation
, a wave function ansatz needs high flexibility in many - body configuration space , but only low flexibility in one particle space a small number of active orbitals is normally sufficient .
this problem is commonly treated with brute force full configuration interaction ( fci),@xcite which is practical for up to approximately 16 active orbitals .
however , recently significant progress has been made with alternative methods to fci ; in particular , due to developments in the density matrix renormalization group ( dmrg ) algorithm , fci quantum monte carlo ( fciqmc),@xcite and selected configuration interaction approaches@xcite , static correlation can now be treated accurately in much larger active spaces .
we here focus on dmrg,@xcite a variational method which minimizes the energy of a wavefunction parametrized as a matrix product state ( mps).@xcite dmrg can handle active spaces of around 3040 orbitals , and in some cases even up to 100@xcite . however , by themselves the mentioned methods are not efficient for obtaining quantitative accuracy and for this dynamic correlation must also be calculated . to account for dynamic correlation ,
a wave function ansatz capable of efficiently handling a _ large _ set of ( virtual ) orbitals is required , while the effective many - body structure of the wave function can be comparatively simple and is amendable to convenient parameterization .
an efficient approach for this is to treat excitations out of a qualitatively correct , strongly correlated reference function .
methods of this class are hierarchies based on truncated multireference configuration interaction ( mrci)@xcite , various flavors of perturbation theory@xcite and coupled cluster theory.@xcite however , the established methods of these classes are based on ci treatments of the static correlation , which either makes them cumbersome to apply ( ras , gas)@xcite , or limits them to small active spaces ( cas ) .
some alternative approaches fall outside this static - dynamic partitioning framework ( e.g. , methods based on geminals@xcite , jastrow factors@xcite or reduced density matrices@xcite ) ; while promising , they have so far not matured into routinely applied methods .
for these reasons , a major challenge in recent years has been finding ways of combining the powerful dmrg approach to static correlation with one of the techniques more suitable for dynamic correlation ( sec .
[ sec : previouswork ] ) .
it was only recently realized that second - order perturbation theories ( pt2s ) can be formulated with matrix product states ( mps ) by recasting them into an effective optimization problem.@xcite this approach , called matrix product state perturbation theory ( mpspt ) , converges to the fully uncontracted version of a pt2 , and can be performed with relatively minor modifications to an existing dmrg program.@xcite mpspt can treat complex zeroth order hamiltonians @xmath2 involving two - body interactions ; this allows formulating pt2 methods which are much more accurate than one - body-@xmath2-based pt2s such as mp2 or caspt2 .
we recently introduced multi - reference linearized coupled cluster ( mrlcc ) as one such method.@xcite while formally a pt , the third order mrlcc3 was shown to be extremely accurate ( comparable to mrci+q ) , stable with no intruder state problems , exactly size - consistent , and to a large extent able to compensate for sub - optimal reference states.@xcite however , the previous implementation of mrlcc is computationally still rather expensive ; this is a result of parameterizing the perturbed state using an mps , which can not efficiently treat large virtual orbital spaces . here
we solve this problem by parameterizing the perturbed wave function using a mixture of celani - werner - style partial internal contraction ( ic)@xcite and mpspt ; and limiting ourselves to calculating only the second order correction to the energies .
the resulting method can effectively handle large virtual spaces , while at the same time being applicable to systems with 30 active orbitals and more . as a side product
, we also obtain an efficient variant of nevpt2 , which can be applied to systems with 1000 virtual orbitals and 30 active orbitals at the same time without the need for invoking the strong - contraction approximation .
the article is organized as follows : sec .
[ sec : previouswork ] discusses mr methods of describing dynamic correlation , including their benefits and drawbacks , and previous work aiming to make them compatible with dmrg .
[ sec : ti ] introduces the nevpt2 and mrlcc2 theories , and covers our approach to their efficient implementation with ic and mpspt .
[ sec : vb ] covers accuracy tests and benchmark applications ; in particular , applications to first row dimers , cr@xmath0 , pentacene and oxo - mn(salen ) , are shown , and timing tests for mrlcc2 and nevpt2 with different active spaces and basis set sizes are presented .
[ sec : c ] covers concluding remarks and outlook for future research .
the naive treatment of dynamical correlation in mr problems is extremely challenging,@xcite but a major breakthrough was made with the introduction of internal contraction ( ic ) by meyer.@xcite ic , now widely used in mr theories@xcite aggressively truncates the many body basis and reduces the cost of the calculation from exponential to merely polynomial in the number of active space orbitals .
the difficulty with ic is that the working equations become extremely cumbersome and can often only be derived using domain specific computer algebra systems .
the working equations can be simplified by treating the active orbitals and the core orbitals on an equal footing as was done in the mrci program developed by werner - knowles ( wk scheme)@xcite . but large efficiency gains can be achieved by recognizing that density matrices should only contain active space indices .
this was done in the case of caspt2 by celani and werner ( cw scheme)@xcite , and much later for mrci by shamasundar et al@xcite .
the straightforward use of ic in perturbation theory or truncated configuration interaction with dmrg is complicated by the fact that fourth ( and sometimes even higher ) order reduced density matrices ( rdm ) in the active space are required .
calculation and storage of such high order rdms is limited to an active space of around 25 orbitals beyond which it become prohibitively expensive@xcite . to circumvent this difficulty , some researches have resorted to approximating the higher body rdm by reconstructing its disconnected part by antisymmetric multiplication of lower body rdm and ignoring the density cumulant which is the connected part@xcite .
the rdm reconstruction can be performed by setting the three body and four body cumulants to zero as was done for n - electron valence perturbation theory ( nevpt2)@xcite ; which resulted in severe numerical problems rendering the theory virtually unusable .
canonical transformation ( ct ) theory@xcite also sets three and higher body cumulants to zero and is known to suffer from _ intruder _ state problems ( the ct intruder states are different than the ones in perturbation theory ) .
a much milder approximation is to use the exact three body cumulants but setting the four body cumulants to zero@xcite .
this increases the numerical stability considerably but artifacts still remain that have to be eliminated by using level shifts .
the underlying difficulty in neglecting cumulants beyond a certain rank is that , while the cumulants decay exponentially rapidly with the rank in weak correlation , they either decay very slowly or not at all in highly multireference situations@xcite . in this work
we follow a different route , by using _ partial _ internal contraction .
all terms that require 4-rdms are treated fully uncontracted , and the rest of the terms , requiring three body or lower rdms are treated using ic .
this partitioning is in fact also used in the mrci of shamasundar et al.@xcite .
the reason this is a practical route is as follows : all terms which are difficult to treat using ic and require 4-rdms ( ic states with three active space indices and one virtual or core index ) are relatively easily treated using uncontracted theories , because they involve only one core or virtual orbital , and are effectively singly excited states . however , if these uncontracted terms are represented using wave functions , as done in previous work,@xcite their number _ still _ increases exponentially in the number of active orbitals . for this reason , we here treat such terms with mpspt,@xcite which does not have this problem
this combination of ic and mpspt is the key development of the current work , which makes efficient mrlcc2 and nevpt2 methods accessible to the large active spaces treatable with dmrg .
we will now discuss its theory and implementation .
the molecular orbitals are divided into three subsets : the core orbitals indexed by @xmath3 , which are always doubly occupied in the reference wavefunction @xmath4 ( casscf or cas - ci ) ; the active orbitals @xmath5 , which can have any occupancy in @xmath4 ; and the virtual orbitals @xmath6 , which are unoccupied in @xmath4 .
general orbitals ( i.e. , the union of the three subsets ) are indexed by @xmath7 .
we work exclusively with spin free quantities , by integrating out the spin degrees of freedom .
thus , the hamiltonian is @xmath8 here and in the following , repeated indices imply contraction , operators are written with a hat ( e.g. @xmath9 ) and numerical tensors are written without a hat ( e.g. @xmath10 ) .
@xmath11 are the one electron integrals , @xmath12 are the two electron integrals , and @xmath13 are the spin - free single- and double excitation operators , in which @xmath14 denote the spin degrees of freedom . in the rayleigh - schroedinger perturbation theory , the zeroth order wavefunction @xmath4 must be an eigenfunction of the zeroth order hamiltonian @xmath2 .
however , for a given @xmath4 , this requirement does _ not _ uniquely determine @xmath2 , and in general each viable choice of @xmath2 will lead to a different perturbation theory .
for example , the casscf ( or cas - ci ) wavefunction is , at the same time , an eigenfunction of the fock operator , dyall s hamiltonian,@xcite and fink s hamiltonian;@xcite choosing these as @xmath2 leads to caspt2 theory , nevpt2 theory , and mrlcc2 theory , respectively , all of which have vastly different properties . in this work
we focus on the dyall- and fink- hamiltonians , which are described below .
these have previously been shown to lead to accurate multi - reference perturbation methods which do not suffer from the intruder state problems plaguing caspt2.@xcite concretely , dyall s hamiltonian@xcite is given by @xmath16 where @xmath17 is a fock matrix .
@xmath17 contains modified one body integrals , approximating the interaction with core- and active electrons as a mean - field .
it is defined as @xmath18 where @xmath19 denotes the 1-rdm . when using casscf canonical orbitals , @xmath17 becomes diagonal , and the diagonal elements can be viewed as orbital energies . a diagonal fock matrix ( implicitly assumed in the work of angeli et al.@xcite ) simplifies nevpt2 theory . however , in the present work we avoid this assumption , and the expressions derived here are more general .
fink s hamiltonian@xcite is @xmath20 where @xmath21 signifies that all terms which would change the total number of electrons in the core , active or virtual spaces are removed .
an interesting property of fink s hamiltonian is that the zeroth order energy is equal to the energy of the reference wavefunction , and the first order energy is zero .
this is not true in nevpt2 theory ; similarly to mller - plesset perturbation theory ( mp2 ) , in nevpt2 the reference energy is given by the sum of the zeroth and first order energies , neither of which vanish . in perturbation theory , the first order correction ( @xmath23 ) to the reference wavefunction ( @xmath24 ) is determined by solving the linear equation @xmath25 in which @xmath26 is the perturbation , and @xmath27 is a projector . to define the representation of @xmath22 ,
we first split the full @xmath1-electron fock - space into configuration spaces characterized by a unique occupation pattern @xmath28 .
the three integers hereby denote the change in the _ total _ number of electrons in core ( @xmath29 ) , active ( @xmath30 ) , and virtual ( @xmath31 ) orbitals relative to @xmath32 . with a two - body perturbation @xmath33
, eight such spaces can be reached by @xmath34 , and are therefore required in the expansion of @xmath22 ( eq . ) .
these spaces are listed in table [ tab : eight ] , and their representation will be discussed next .
neither dyall s nor fink s hamiltonian ( @xmath15 ) change the number of electrons in the three orbital spaces ; consequently , both are block - diagonal with respect to the eight configuration spaces . therefore eq . can be solved in each space independently .
.the eight classes of perturber states reachable by applying a two - body perturbation @xmath33 to the zeroth order state @xmath4 , and which therefore contribute to @xmath22 .
the classes are formed by changing the total number of electrons in core , active , and virtual orbitals , relative to the occupation pattern of @xmath4 ( indicated by positive / negative numbers ) .
column 5 and 6 denote the basis of the many - body states and the order of the active - space rdm needed , _ if _ the space were parameterized using internal contraction ( ic ) .
mpspt denotes matrix product state perturbation theory . in the methods discussed here ,
the contribuitions to @xmath22 within spaces i viii can be solved for independently of each other . [ cols="<,^,^,^,^,^,^ " , ] the results of our calculations are presented in table [ tab : salen ] , where it can be seen that the singlet - triplet splitting not only depends on the dynamical correlation but also depends quite sensitively on the size of the basis set .
interestingly the dmrg - scf calculations with vdz basis set predicts the triplet as being more stable than the singlet by 1.1 me@xmath35 , which is in contradiction to the results obtained by stein and reiher where they obtained singlet as being more stable than the triplet by 0.6 me@xmath35 .
we do nt expect our dmrg - scf calculations performed with m=2000 ( or their calculations performed with m=1000 ) to be accurate enough to resolve such small differences .
further there is a strong likelihood that the ground state triplet is nearly degenerate with the first excited triplet state@xcite which can easily explain the difference in the excitation energies .
the splitting calculated using dmrg - scf seems to converge with basis set when going from vtz to vqz basis set .
the natural orbitals of the singlet state calculated using the vdz basis set are shown in figure [ fig : natorbs ] . the shape and the occupancy of the natural orbitals
do not change substantially when the basis set is changed , signifying that the qualitative nature of the singlet state is captured already at the vdz basis set . comparing
the occupation number of the singlet and triplet natural orbitals shows that one of the electrons from the doubly occupied @xmath36 orbital in the singlet state gets excited to the @xmath37 orbital in the triplet state .
this transition is in agreement with other multireference calculations .
orbital gets excited to the empty @xmath37 orbital which is shown schematically with the red arrow .
the occupation numbers of the @xmath36 orbital and @xmath37 orbitals change from 1.96 and 0.16 in the singlet state to 1.00 and 1.04 ( shown in red ) in the triplet state respectively.,scaledwidth=50.0% ] the dynamical correlation is quite sensitive to the basis set , with vdz basis set tends to strongly stabilize the singlet state in sharp contrast to the results calculated using the vtz and vqz basis set .
the agreement between the nevpt2 results with the vtz and vqz basis set is quite good and nevpt2 results at the vtz basis set are in good agreement with the mrlcc2 results with the same basis set .
( we were unable to perform mrlcc2 calculations with vqz basis set because of the size of the external space . ) from our work it is clear that the difference between the energy of the singlet and triplet states is relatively small and is more or less within the error bars of the methods being used .
an interesting question is that of the reaction mechanism on the different potential energy surfaces , which can be further explored by calculating the barrier heights of the transition states using mrlcc2 and nevpt2 methods ; this will be the topic of a future publication .
here we have shown that celani - werner scheme of internal contraction can be used to perform nevpt2 and mrlcc2 calculations for problems where the reference wavefunction is given as an mps following a dmrg - ci or a dmrg - scf calculation .
this program can be used to perform nevpt2 calculations for active spaces of 30 orbitals or more and with about 1000 virtual orbitals and mrlcc calculations with similar active spaces and about 500 virtual orbitals .
this lends us the ability to calculate dynamical correlations on organometallic clusters where so far only static correlation have been calculated using dmrg@xcite .
our results show that we can improve the accuracy of the mrlcc theory by calculating the third order energy corrections .
the third order energies can be calculated by evaluating @xmath38 , where @xmath39 is the @xmath40 perturber class of the first order state .
there are eight different perturber classes and the resulting 28 different matrix elements need to be calculated to evaluate the third order energy . out of these , all the ones where both the @xmath40 and @xmath41 perturber classes are parametrized using internal contraction can be easily calculated using a maximum of 3-rdms .
the matrix elements , where the @xmath40 perturber class in calculated using mpspt and the @xmath41 perturber class is calculated using internal contraction , can not be calculated in a straightforward way and will be the topic of a subsequent paper .
the size of problems treatable with this method can be significantly extended by using explicit correlation ( f12 methods ) and local correlations using pair natural orbitals ( pno ) .
it has already been demonstrated that by performing dmrg calculations on an effective hamiltonian obtained by canonical transcorrelation of yanai@xcite , spectroscopic accuracy can be obtained for the be@xmath0 dimer@xcite .
the canonical transcorrelation method@xcite although universal ( much like torheyden s post f12 correction scheme@xcite ) , has a severe memory bottleneck and can only be used for relatively small basis sets@xcite . a more efficient route for using f12-geminals in our method
will be the approach used by ten - no@xcite and later by shiozaki et al.@xcite for multireference problems .
pno were used in the multireference problems for the first time by fink et al.@xcite , and after a long lull pno based methods have seen a huge revival due to work of neese et al.@xcite .
our next project will aim to make use of these recent developments by extending this method by combining it with f12-methods , pnos . .
legeza , r. m. noack , j. slyom , l. tincani , _ computational many - particle physics _ , h. fehske , r. schneider , a. weie , eds .
( springer berlin heidelberg , 2008 ) , vol . 739 of _ lecture notes in physics _ , pp . 653664 . | we present two efficient and intruder - free methods for treating dynamic correlation on top of general multi - configuration reference wave functions including such as obtained by the density matrix renormalization group ( dmrg ) with large active spaces .
the new methods are the second order variant of the recently proposed multi - reference linearized coupled cluster method ( mrlcc ) [ s. sharma , a. alavi , j. chem .
phys .
143 , 102815 ( 2015 ) ] , and of n - electron valence perturbation theory ( nevpt2 ) , with expected accuracies similar to mrci+q and ( at least ) caspt2 , respectively .
great efficiency gains are realized by representing the first - order wave function with a combination of internal contraction ( ic ) and matrix product state perturbation theory ( mpspt ) . with this combination ,
only third order reduced density matrices ( rdms ) are required .
thus , we obviate the need for calculating ( or estimating ) rdms of fourth or higher order ; these had so far posed a severe bottleneck for dynamic correlation treatments involving the large active spaces accessible to dmrg . using several benchmark systems , including first and second row containing small molecules , cr@xmath0 , pentacene and oxo - mn(salen ) , we shown that active spaces containing at least 30 orbitals can be treated using this method . on a single node , mrlcc2 and
nevpt2 calculations can be performed with over 550 and 1100 virtual orbitals , respectively .
we also critically examine the errors incurred due to the three sources of errors introduced in the present implementation - calculating second order instead of third order energy corrections , use of internal contraction and approximations made in the reference wavefunction due to dmrg . |
the basic problem in numerical probability is to _ optimize _ some way or another the computation by a monte carlo simulation of a real quantity @xmath0 known by a probabilistic representation @xmath1\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}\ ] ] where @xmath2 is a random vector having values in a banach space @xmath3 and @xmath4 is a borel function ( and @xmath5 is square integrable ) .
the space @xmath3 is @xmath6 but can also be a functional space of paths of a process @xmath7}$ ] .
however , in this introduction section , we will first focus on the finite dimensional case @xmath8 .
assume that @xmath9 has an absolutely continuous distribution @xmath10 ( @xmath11 denotes the lebesgue measure on @xmath12 ) and that @xmath13 with @xmath14 ( otherwise the expectation is clearly @xmath15 and the problem is meaningless ) .
furthermore we assume that the probability density @xmath16 is _ everywhere positive _ on @xmath6 .
the paradigm of importance sampling applied to a parametrized family of distributions is the following : consider the family of absolutely continuous probability distributions @xmath17 , @xmath18 , such that @xmath19 , @xmath20-@xmath21 .
one may assume without loss of generality that @xmath22 is an open non empty connected subset of @xmath23 containing @xmath15 so that @xmath24 .
in fact we will assume throughout the paper that @xmath25 .
then for any @xmath6-valued random variable @xmath26 with distribution @xmath27 , we have @xmath28\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } = { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } f({x^{(\theta)}})\frac{p({x^{(\theta)}})}{p_\theta({x^{(\theta ) } } ) } { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}.\ ] ] among all these random variables having the same expectation @xmath29\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}$ ] , the one with the lowest variance is the one with the lowest quadratic norm : minimizing the variance amounts to finding the parameter @xmath30 solution ( if any ) to the following minimization problem @xmath31 where , for every @xmath32 , @xmath33\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } = { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } f^2(x ) \frac{p(x)}{p_\theta(x ) } { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } \le + \infty.\ ] ] a typical situation is importance sampling by mean translation in a finite dimensional gaussian framework _
i.e. _ @xmath34\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}.\ ] ] then the second equality in ( [ v = e ] ) is simply the cameron - martin formula .
this specific framework is very important for applications , especially in finance , and was the starting point of the new interest for recursive importance sampling procedures , mainly initiated by arouna in @xcite ( see further on ) .
in fact , as long as variance reduction is concerned , one can consider a more general framework without extra effort . as a matter of fact , if the distributions @xmath35 satisfy @xmath36 and @xmath37 satisfies @xmath38\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } < + \infty$ ] for every @xmath32 , then ( see proposition [ prop_intro ] below ) , the function @xmath39 is finite , convex , goes to infinity at infinity . as
a consequence @xmath40 _ is non empty_. assumption @xmath41 can be localized by by considering that one the two conditions holds only on a borel set @xmath42 of @xmath6 such that @xmath43 . if @xmath44 is _ strictly _
@xmath45-concave for every @xmath46 in a borel set @xmath47 such that @xmath48\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}>0 $ ] , then @xmath39 is strictly convex and @xmath40 is reduced to a single @xmath49 .
these results follow from the second representation of @xmath39 as an expectation in ( [ v = e ] ) which is obtained by a second change of probability ( the reverse one ) . for notational convenience
we will temporarily assume that @xmath50 in this introduction section , although our main result needs no such restriction .
a classical procedure to approximate @xmath51 is the so - called robbins - monro algorithm .
this is a recursive stochastic algorithm ( see ( [ rm ] ) below ) which can be seen as a stochastic counterpart of deterministic recursive zero search procedures like the newton - raphson one .
it can be formally implemented provided the gradient of the ( convex ) target function @xmath39 admits a representation as an expectation . since we have no _
a priori _ knowledge about the regularity of @xmath37 ( is smooth enough alternative approaches have been developed based on some large deviation estimates which provide a good approximation of @xmath51 by deterministic optimization methods ( see @xcite ) . ] ) and do not wish to have any , we are naturally lead to _ formally _ differentiate the second representation of @xmath39 in ( [ v = e ] ) to obtain a representation of @xmath52 as @xmath53\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}.\ ] ] then , if we consider the function @xmath54 such that @xmath55 naturally defined by ( [ grad1 ] ) , the derived robbins - monro procedure writes @xmath56 with @xmath57 a _ step _ sequence decreasing to 0 ( at an appropriate rate ) , @xmath58 a sequence of i.i.d . random variables with distribution @xmath59 . to establish the convergence of a robbins - monro procedure to @xmath60
requires seemingly not so stringent assumptions .
we mean by that : not so different from those needed in a deterministic framework .
however , one of them turns out to be quite restrictive for our purpose : the sub - linear growth assumption in quadratic mean @xmath61 which is the stochastic counterpart of the classical non - explosion condition needed in a deterministic framework . in practice , this condition is almost never satisfied in our framework due to the behaviour of the term @xmath62 as @xmath63 goes to infinity .
the origin of recursive importance sampling as briefly described above goes back to kushner and has recently been brought back to light in a gaussian framework by arouna in @xcite .
however , as confirmed by the numerical experiments carried out by several authors ( @xcite ) , the regular robbins - monro procedure ( [ rm ] ) does suffer from a structural instability coming from the violation of ( [ nec ] ) .
this phenomenon is quite similar to the behaviour of the explicit discretization schemes of an @xmath64 @xmath65 when @xmath66 has a super - linear growth at infinity .
furthermore , in a probabilistic framework no implicit scheme " can be devised in general . then the only way out _ mutatis mutandis _ is to kill the procedure when it comes close to explosion and to restart it with a smaller step sequence .
formally , this can be described as some repeated projections or truncations when the algorithm leaves a slowly growing compact set waiting for stabilization which is shown to occur @xmath67 .
then , the algorithm behaves like a regular robbins - monro procedure .
this is the so - called projection la chen " avatar of the robbins - monro algorithm , introduced by chen in @xcite and then investigated by several authors ( see @xmath68 @xcite ) formally , repeated projections la chen " can be written as follows : @xmath69 where @xmath70 denotes the projection on the convex compact @xmath71 ( @xmath72 is increasing to @xmath6 as @xmath73 ) . in @xcite
is established a a central limit theorem for this version of the recursive variance reduction procedure . some extensions to non gaussian framework
have been carried out by arouna in his phd thesis ( with some applications to reliability ) and more recently to the marginal distributions of a lvy processes by kawai in @xcite .
however , convergence occurs for this procedure after a long stabilization phase " provided that the sequence of compact sets have been specified in an appropriate way .
this specification turns out to be a rather sensitive phase of the tuning " of the algorithm to be combined with that of the step sequence . in this paper , we show that as soon as the growth of @xmath37 at infinity can be explicitly controlled , it is always possible to design a regular robbins - monro algorithm which @xmath67 converges to a variance minimizer @xmath51 with no risk of explosion ( and subsequently no need of repeated projections ) . to this end
the key is to introduce a _
third _ change of probability in order to control the term @xmath62 . in a gaussian framework
this amounts to switching the parameter @xmath63 from the density @xmath16 to the function @xmath37 by a third mean translation .
this of course corresponds to a new function @xmath74 but can also be interpreted _ a posteriori _ as a way to introduce an _ adaptive _ step sequence ( in the spirit of @xcite ) . in terms of formal importance sampling , we introduce a new positive density @xmath75 ( everywhere positive on @xmath76 ) so that the gradient writes @xmath77\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } = { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } \widetilde h_v(\theta,\widetilde { x^{(\theta ) } } ) { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}},\ ] ] where @xmath78 .
the weight " @xmath79 may seem complicated but the rle of the density @xmath75 is to control the critical term @xmath80 by a ( deterministic ) quantity only depending on @xmath63 .
then we can replace @xmath81 by a function @xmath82 in the above robbins - monro procedure ( [ rm ] ) where @xmath83 is a positive function used to control the behaviour of @xmath84 for large values of @xmath46 ( note that @xmath85\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } = 0 { \ifcase 2\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \bigl\ { } { \e\ ! { } { { \ifcase 2\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } h(.,\widetilde { x^{(\theta ) } } ) { \ifcase 2\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } = 0 { \ifcase 2\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \bigl\ { } { \e\ ! { } { { \ifcase 2\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } h(.,\widetilde { x^{(\theta ) } } ) { \ifcase 2\or ]
\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } = 0 { \ifcase 2\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \biggl\ { } { \e\ ! { } { { \ifcase 2\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } h(.,\widetilde { x^{(\theta ) } } ) { \ifcase 2\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } = 0 { \ifcase 2\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \biggl\ { } { \e\ ! { } { { \ifcase 2\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } h(.,\widetilde { x^{(\theta ) } } ) { \ifcase 2\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } = 0 { \ifcase 2\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \left\ { } { \e\ ! { } { { \ifcase 2\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } h(.,\widetilde { x^{(\theta ) } } ) { \ifcase 2\or ]
\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } = 0 { \ifcase 2\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\fi } } = { { \ifcase 6\or \ { } \nabla v = 0 { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \bigl\ { } \nabla v = 0 { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \bigl\ { } \nabla v = 0 { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \biggl\ { } \nabla v = 0 { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \biggl\ { } \nabla v = 0 { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \left\ { } \nabla v = 0 { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\fi}}$ ] ) .
we will first illustrate this paradigm in a finite dimensional setting with parametrized importance sampling procedures : the mean translation and the esscher transform which coincide for gaussian vectors on which a special emphasis will be put .
both cases correspond to a specific choice of @xmath75 which significantly simplifies the expression of the weight . as a second step
, we will deal with an infinite dimensional setting ( path - dependent diffusion like processes ) where we will rely on the girsanov transform to play the role of mean translator . to be more precise ,
we want now to compute @xmath86\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}$ ] where @xmath9 is a path - dependent diffusion process and @xmath37 is a functional defined on the space @xmath87,\r^d)$ ] of continuous functions defined on @xmath88 $ ] .
we consider a @xmath89-dimensional it process @xmath7}$ ] solution of the path - dependent sde @xmath90 where @xmath91}$ ] is a @xmath92-dimensional standard brownian motion , @xmath93}$ ] is the stopped process at time @xmath94 , @xmath95\times { \cal c}([0,t],\r^d)\to \r^d$ ] and @xmath96\times { \cal c}([0,t],\r^d)\to { \cal m}(d , q)$ ] are lipschitz with respect to the @xmath97 on the space @xmath98,\r^d)$ ] and continuous in @xmath99\times { \cal c}([0,t],\r^d)$ ] ( see @xcite for more details about these path - dependent sde s ) .
let @xmath100 be a fixed borel bounded functional on @xmath87,\r^d)$ ] with values in @xmath101 ( where @xmath102 is a free integral parameter ) . then
a girsanov transform yields that for every @xmath103,\r^p)$ ] , @xmath104\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } = { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } f({x^{(\theta ) } } ) e^{-\int_0^t\ ! { { \ifcase 1\or \langle\or \bigl\langle\or \bigl\langle\or \biggl\langle\or \biggl\langle\or \left\langle\fi } \varphi({x^{(\theta),s } } ) \theta(s ) , \operatorname{d\!}w_s { \ifcase 1\or \rangle\or \bigr\rangle\or \bigr\rangle\or \biggr\rangle\or \biggr\rangle\or \right\rangle\fi } } - \frac{1}{2 } { { { \ifcase 6\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \varphi({x^{(\theta)}}{}^{,.})\theta { \ifcase 6\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}_{\scriptscriptstyle{\!l^2_{t , q}}}}}^2 } { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}\ ] ] where @xmath26 is the solution to @xmath105 . the functional to be minimized is now @xmath106\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}},\qquad \theta\!\in l^2_{t , p}.\ ] ] in practice we will only minimize @xmath39 over a _ finite dimensional subspace _ of @xmath107 . the paper is organized as follows .
section [ dimfinie ] is devoted to the finite dimensional setting where we recall the main tool including a slight extension of the robbins - monro theorem in the subsection [ argmin - target ] and the gaussian case investigated in @xcite is revisited to emphasize the new aspects of our algorithm in the subsection [ gaussian_case ] .
in section 2 we successively investigate the translation for log - concave distributions probability and the esscher transform . in section [ diffusion ]
we introduce a functional version of our algorithm based on the girsanov theorem to deal the sde . in section 4
we provide some comments on the practical implementation and in section [ numerical ] some numerical experiments are carried out on some option pricing problems .
@xmath108 we will denote by @xmath109 the fact that a symmetric matrix @xmath110 is positive definite .
@xmath111 will denote the canonical euclidean norm on @xmath112 and @xmath113 will denote the canonical inner product .
@xmath108 the real constant @xmath114 denotes a positive real constant that may vary from line to line .
@xmath108 @xmath115 if @xmath116 is an @xmath117-valued ( class of ) borel function(s ) .
[ prop_intro ] suppose holds .
+ then the function @xmath39 defined by ( [ v = e ] ) is convex and @xmath119 . as a consequence @xmath120 * proof .
* by the change of probability @xmath121 we have @xmath122\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}$ ] .
let @xmath46 fixed in @xmath6 .
the function @xmath123 is concave , hence @xmath124 is convex so that , owing to the young inequality , the function @xmath125 is convex since it is non - negative . to prove that @xmath39 tends to infinity as @xmath126 goes to infinity , we consider two cases : * if @xmath127 for every @xmath128 , the result is trivial by fatou s lemma . *
if @xmath129 for every @xmath128 , we apply the reverse hlder inequality with conjugate exponents @xmath130 to obtain @xmath131\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}^3 { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } { { \ifcase 6\or ( \or \bigl(\or \bigl(\or \biggl(\or \biggl(\or \left(\fi } \frac{p(x)}{p_{\theta/2}(x ) } { \ifcase 6\or ) \or \bigr)\or \bigr)\or \biggr)\or \biggr)\or \right)\fi}}^{-1 } { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}^{-2 } , \\ & \ge { \e\ ! { } { { \ifcase 5\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } f^{2/3}(x ) { { \ifcase 4\or ( \or \bigl(\or \bigl(\or \biggl(\or \biggl(\or \left(\fi } \frac{p^2_{\theta/2}(x)}{p(x)p_\theta(x ) } { \ifcase 4\or ) \or \bigr)\or \bigr)\or \biggr)\or \biggr)\or \right)\fi}}^{\frac{1}{3 } } { \ifcase 5\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}^3,\end{aligned}\ ] ] ( @xmath16 and @xmath132 are probability density functions ) .
one concludes again by fatou s lemma . @xmath133 the set @xmath134 , or to be precise
, the random vectors taking values in @xmath134 will the target(s ) of our new algorithm .
if @xmath39 is strictly convex , @xmath68 if @xmath135\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } > 0,\ ] ] then @xmath50 .
nevertheless this will not be necessary owing to the combination of the two results that follow .
[ un]let @xmath136 be a convex differentiable function , then @xmath137 furthermore , if @xmath138 is nonempty , it is a convex closed set ( which coincide with @xmath139 ) and @xmath140 a sufficient ( but in no case necessary ) condition for a nonnegative convex function @xmath141 to attain a minimum is that @xmath142 .
now we pass to the statement of the convergence theorem on which we will rely throughout the paper .
it is a slight variant of the regular robbins - monro procedure whose proof is rejected in an annex .
[ thmrz ] ( extended robbins - monro theorem ) let @xmath143 a borel function and @xmath9 an @xmath6-valued random vector such that @xmath144\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}<+\infty$ ] for every @xmath145 . then set @xmath146\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}.\ ] ]
suppose that the function @xmath66 is continuous and that @xmath147 satisfies @xmath148 let @xmath149 be a sequence of gain parameters satisfying @xmath150 suppose that @xmath151\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } \le c(1 + { { \ifcase 1\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \theta { \ifcase 1\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}}^2)\ ] ] ( which implies @xmath152 ) .
let @xmath153 be an i.i.d .
sequence of random vectors having the distribution of @xmath9 , a random vector @xmath154 , independent of @xmath153 satisfying @xmath155\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } < + \infty$ ] , all defined on the same probability space @xmath156 .
then , the recursive procedure defined by @xmath157 satisfies : @xmath158 the convergence also holds in @xmath159 , @xmath160 .
the proof is postponed to the appendix at the end of the paper .
the natural way to apply this theorem for our purpose is the following : * step 1 : we will show that the convex function @xmath39 in ( [ v = e ] ) is differentiable with a gradient @xmath52 having a representation as an expectation formally given @xmath161\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}$ ] .
* step 2 : then set @xmath162 where @xmath163 is a ( strictly ) positive function on @xmath23 . as a matter of fact , with the notations of the above theorem @xmath164 so that @xmath165 and is satisfied ( set @xmath166 in lemma [ un ] ) . * step 3 : specify in an appropriate way the function @xmath167 so that the linear quadratic growth assumption is satisfied .
this is the sensitive point that will lead us to modify the structure more deeply by finding a new representation of @xmath52 as an expectation not directly based on the local gradient @xmath168 .
the gaussian is the framework of @xcite .
it is also a kind of introduction to the infinite dimensional diffusion setting investigated in section [ diffusion ] . in the gaussian case ,
the natural importance sampling density is the translation of the gaussian density : @xmath169 for @xmath170 ( @xmath171 @xmath172 ) .
we have @xmath173 the assumption is clearly satisfied by the gaussian density , and we assume that @xmath37 satisfies @xmath174\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } < + \infty$ ] so that @xmath39 is well defined . in @xcite
, arouna considers the function @xmath54 defined by @xmath175 it is clear that the condition is not satisfied even if we simplify this function by @xmath176 ( which does not modify the problem ) .
when @xmath5 have finite moments of any order , a naive way to control directly @xmath177 by an explicit deterministic function of @xmath63 ( in order to rescale it ) is to proceed as follows : one derives from hlder inequality that for every couple @xmath178 , @xmath179 of conjugate exponents @xmath180 setting @xmath181 and @xmath182 , yields @xmath183 then , @xmath184 satisfies the condition and theoretically the standard robbins - monro algorithm implemented with @xmath185 @xmath67 converges and no projection nor truncation is needed .
numerically , the solution is not satisfactory because the correcting factor @xmath186 goes to zero much too fast as @xmath63 goes to infinity : if at any iteration at the beginning of the procedure @xmath187 is sent too far " , then it is frozen instantly .
if @xmath188 is too small it will simply not prevent explosion .
the tuning of @xmath188 becomes quite demanding and payoff dependent .
this is in complete contradiction with our aim of _ a self - controlled variance reducer_. a more robust approach needs to be developed .
on the other hand this kind of behaviour suggests that we are not in the right asymptotics to control @xmath177 .
note however that when @xmath37 is bounded with a compact support , then one can set @xmath189 and the above approach provides an efficient answer to our problem .
we consider the density @xmath190 by , we have @xmath191\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}},\ ] ] with @xmath192 , _
i.e. _ @xmath193 .
since @xmath16 is the gaussian density , we have @xmath194 . as a consequence ,
the function @xmath195 defined by @xmath196 provides a representation @xmath197\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}$ ] of the gradient @xmath52 .
as soon as @xmath37 is bounded , this function satisfies the condition .
otherwise , we note that thanks to this new change of variable the parameter @xmath63 lies now _ inside _ the payoff function @xmath37 and that the _ exponential term _ has disappeared from the expectation .
if we have an _ a priori _ control on the function @xmath198 as @xmath199 goes to infinity , say @xmath200 then we can consider the function @xmath201 which satisfies @xmath202 the resulting robbins - monro algorithm reads @xmath203 we no longer to tune the correcting factor and one verifies on simulations that it does not suffer from freezing in general . in case of a too dissymmetric function @xmath37 this may still happen but a _ self - controlled _ variant is proposed in section [ translation ] below to completely get rid of this effect ( which can not be compared to an explosion ) .
we consider importance sampling by mean translation , namely we set @xmath204 for @xmath170 . in this section
we assume that @xmath205 so that holds .
moreover , we make the following additional assumption on the probability density @xmath16 @xmath206 \ ; \mbox { such that } \;\left\ { \begin{array}{ll}(i ) & \frac{\left|\nabla p\right|}{p}(x)=o(|x|^{a-1 } ) \ ; \mbox { as } \ ; |x|\to \infty \\ ( ii)&\exists \delta>0,\ ; \log p(x)+\delta |x|^a\;\mbox { is convex . }
\end{array}\right.\ ] ] first we will use ( [ v = e ] ) to differentiate @xmath39 since suppose and are satisfied and the function @xmath37 satisfies @xmath207\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } < + \infty \quad\text { and } \quad \forall\ , c>0,\quad { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } f^2(x)e^{c|x|^{a-1 } } { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } < + \infty.\ ] ] then @xmath39 is finite and differentiable on @xmath6 with a gradient given by @xmath208\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}\\ & = { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } f^2(x-\theta)\frac{p^2(x-\theta)}{p(x)p(x-2\theta)}\frac{\nabla p(x-2\theta)}{p(x-2\theta ) } { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}.\label{gradv2 } \end{aligned}\ ] ] * proof . * the formal differentiation to get ( [ gradv1 ] ) from ( [ v = e ] ) is obvious .
so it remains to check the domination property for @xmath63 lying inside a compact set .
let @xmath209 and @xmath145 .
the @xmath45-concavity of @xmath16 implies that @xmath210 so that @xmath211 using the assumption yields , for every @xmath212 , @xmath213 to derive the second expression for the gradient , we proceed as follows : an elementary change of variable shows that @xmath214\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}.\cqfd\end{aligned}\ ] ] the second change of variable ( in ) has been processed to withdraw the parameter @xmath63 from the possible non smooth function @xmath37 to make possible the differentiation of @xmath39 ( since @xmath16 is smooth ) .
the second expression results form a _
third _ translation of the variable in order to plug back the parameter @xmath63 into the function @xmath37 which in common applications has a known controlled growth rate at infinity .
this last statement may look strange at a first glance since @xmath63 appears in the weight " term of the expectation that involves the probability density @xmath16 .
however , when @xmath215 , this term can be controlled easily since it reduces to @xmath216 the following lemma shows that , more generally in our strongly unimodal setting , if and are satisfied , this
weight " can always be controlled by a deterministic function of @xmath63 .
[ lemmetec ] if holds , then there exists two real constants @xmath217 , @xmath47 such that @xmath218 * proof .
* let @xmath219 be the convex function defined on @xmath6 by @xmath220 .
then , for every @xmath221 , @xmath222 note that @xmath223 .
then , using the @xmath45-convexity of @xmath219 and the elementary inequality @xmath224 ( valid if @xmath225 $ ] ) yields @xmath226 one concludes by the point @xmath227 of .
thus the normal distribution satisfies with @xmath228 and @xmath229 .
moreover , note that the last inequality in the above proof holds as an equality .
now we are in position to derive an unconstraint ( extended ) robbins - monro algorithm to minimize the function @xmath39 , provided the function @xmath37 satisfies a sub - multiplicative control property , in which @xmath230 is a real parameter and @xmath231 a function from @xmath232 to @xmath233 , such that , namely @xmath234\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}<+\infty .
\end{array}\right . \ ] ] * remark .
* assumption seems almost non - parametric . however , its field of application is somewhat limited by ( [ h2 ] ) for the following reason : if there exists a positive real number @xmath235 such that @xmath236 is concave , then @xmath237 for some real constant @xmath114 ; which in turn implies that the function @xmath231 in needs to satisfy @xmath238 for some @xmath239 and some @xmath240 .
( then @xmath241 with @xmath242 if @xmath243 $ ] and @xmath244 if @xmath245 , when @xmath246 ) .
[ thmmt ] suppose @xmath9 and @xmath37 satisfy , , ( [ conddiff ] ) and for some parameters @xmath225 $ ] , @xmath247 and @xmath240 , and that the step sequence @xmath248 satisfies the usual decreasing step assumption @xmath249 then the recursive procedure defined by @xmath250 where @xmath153 is an i.i.d . sequence with the same distribution as @xmath9 and @xmath251 @xmath67 converges toward an @xmath134-valued ( square integrable ) random variable @xmath51 .
in order to apply theorem [ thmrz ] , we have to check the following fact : _ mean reversion _ : the mean function of the procedure defined by ( [ algorm ] ) reads @xmath252\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } = \frac{e^{-2\delta |\theta|^a}}{1+\widetilde f(-\theta)^{2c}}\nabla v(\theta)\ ] ] so that @xmath253 and if @xmath254 and @xmath255 , @xmath256 for every @xmath257 .
_ linear growth of @xmath258 _ : all our efforts in the design of the procedure are motivated by this assumption ( [ lingrowth ] ) which prevents explosion .
this condition is clearly fulfilled by @xmath259 since @xmath260\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}^2 & = \frac{e^{-4\delta |\theta|^a}}{(1+\widetilde f(-\theta)^{2c})^2 } { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } f^4(x-\theta)\left(\frac{p^2(x-\theta)}{p(x)p(x-2\theta)}\frac{|\nabla p(x-2\theta)|}{p(x-2\theta)}\right)^2 { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } , \\ & \le c e^{-4\delta |\theta|^a } { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } ( 1+\widetilde f(x)^{2c})^2 ( a(|x|^{a-1}+|\theta|^{a-1})+b)^2 { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}},\end{aligned}\ ] ] where we used assumption in the first line and inequality ( [ ineqtech ] ) from lemma [ lemmetec ] in the second line .
one derives that there exists a real constant @xmath114 such that @xmath261\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}^2 \le c { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } \widetilde f(x)^{4c}(1+|x|)^{2(a-1 ) } { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } ( 1+|\theta|^{2(a-1)}).\ ] ] this provides the expected condition since holds .
@xmath133 * the normal distribution .
its density is given on @xmath6 by @xmath262 so that is satisfied as well as for @xmath228 , @xmath263 .
assumption is satisfied iff @xmath264 .
then the function @xmath259 has a particularly simple form @xmath265 * _ the hyper - exponential distributions _ @xmath266\ ] ] where @xmath267 is polynomial function .
this wide family includes the normal distributions , the laplace distribution , the symmetric gamma distributions , etc . * _ the logistic distribution _ its density on the real line is given by @xmath268 is satisfied as well as for @xmath269 ( @xmath270 ) , @xmath271 .
assumption is satisfied iff @xmath272 .
a second classical approach is to consider an exponential change of measure ( or esscher transform ) .
this transformation has already been consider for that purpose in @xcite to extend the procedure with repeated projections introduced in @xcite .
we denote by @xmath273 the cumulant generating function ( or log laplace ) of @xmath9 _ i.e. _ @xmath274\or
\bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}$ ] .
we assume that @xmath275 for every @xmath145 ( which implies that @xmath273 is an infinitely differentiable convex function ) and define @xmath276 let @xmath277 denote any random variable with distribution @xmath132 .
we assume that @xmath273 satisfies @xmath278 one must be aware that what follows makes sense as a variance reduction procedure only if the distribution of @xmath26 can be simulated at the same cost as @xmath9 or at least at a reasonable cost @xmath171 @xmath279 where @xmath280 is a borel subset of a metric space and @xmath281 is an explicit borel function . by , the potential @xmath39 to be minimized is @xmath282\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}.$ ] suppose @xmath273 satisfies and @xmath37 satisfies @xmath283\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } < + \infty.\ ] ] then is fulfilled and the function @xmath39 is differentiable on @xmath6 with a gradient given by @xmath284\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } , \\ & = { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } { { \ifcase 2\or ( \or \bigl(\or \bigl(\or \biggl(\or \biggl(\or \left(\fi } \nabla \psi(\theta ) - { x^{(-\theta)}}{\ifcase 2\or ) \or \bigr)\or \bigr)\or \biggr)\or \biggr)\or \right)\fi } } f^2({x^{(-\theta ) } } ) { \ifcase 6\or ]
\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } e^{\psi(\theta)-\psi(-\theta ) } , \label{diff_esscher}\end{aligned}\ ] ] where @xmath285\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } } { { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } e^ { { { \ifcase 1\or \langle\or \bigl\langle\or \bigl\langle\or \biggl\langle\or \biggl\langle\or \left\langle\fi } \theta , x { \ifcase 1\or \rangle\or \bigr\rangle\or \bigr\rangle\or \biggr\rangle\or \biggr\rangle\or \right\rangle\fi } } } { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}}$ ] . the function @xmath273 is clearly log - convex so that @xmath286 is @xmath45-concave for every @xmath209 . on the other hand ,
by we have @xmath287 for every @xmath209 , and is fulfilled .
the formal differentiation to get is obvious and is made rigorous by applying the assumption on @xmath37 .
the second expression of the gradient uses a third change of variable @xmath288\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } e^{\psi(\theta)-\psi(-\theta)}.\cqfd\end{aligned}\ ] ] we assume that @xmath273 satisfies and @xmath37 satisfies and @xmath289\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } < + \infty.\ ] ] then the recursive procedure @xmath290 where @xmath291 is an i.i.d .
sequence with the same distribution as @xmath292 in ( [ rexesscher ] ) and @xmath293 @xmath67 converges toward an @xmath134-valued ( square integrable ) random vector @xmath51 .
we have to check the linear growth of the function @xmath294 ( condition ) .
we have @xmath295\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } & = e^{-\lambda\sqrt{d } { { \ifcase 1\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \nabla \psi(-\theta ) { \ifcase 1\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi } } } { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } f^4({x^{(-\theta ) } } ) { { \ifcase 2\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \nabla \psi(\theta ) - { x^{(-\theta)}}{\ifcase 2\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}}^2 { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } , \notag \\ & \le c e^{-\lambda\sqrt{d } { { \ifcase 1\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \nabla \psi(-\theta ) { \ifcase 1\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi } } } { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } e^{\lambda { { \ifcase 1\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } { x^{(-\theta)}}{\ifcase 1\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi } } } { { \ifcase 2\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \nabla \psi(\theta ) - { x^{(-\theta)}}{\ifcase 2\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}}^2 { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } , \notag \\ & \le c e^{-\lambda \sqrt{d } { { \ifcase 1\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \nabla \psi(-\theta ) { \ifcase 1\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi } } } { { \ifcase 6\or ( \or \bigl(\or \bigl(\or \biggl(\or \biggl(\or \left(\fi } { { \ifcase 1\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \nabla \psi(\theta ) { \ifcase 1\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}}^2 { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } e^{\lambda { { \ifcase 1\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } { x^{(-\theta)}}{\ifcase 1\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi } } } { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } + { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } { { \ifcase 1\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } { x^{(-\theta)}}{\ifcase 1\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}}^2 e^{\lambda { { \ifcase 1\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } { x^{(-\theta)}}{\ifcase 1\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi } } } { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } { \ifcase 6\or ) \or \bigr)\or \bigr)\or \biggr)\or \biggr)\or \right)\fi}}. \label{esscher - check - nec}\end{aligned}\ ] ] first , by the following inequality @xmath296 we have @xmath297 where @xmath298 if @xmath299 or @xmath300 if @xmath301 . with this notation , we have @xmath302\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } & \le \sum_{j\subset { { \ifcase 6\or \ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \bigl\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \bigl\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \biggl\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \biggl\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \left\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\fi } } } { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } e^{\lambda { { \ifcase 1\or \langle\or \bigl\langle\or \bigl\langle\or \biggl\langle\or \biggl\langle\or \left\langle\fi } e_j , { x^{(-\theta)}}{\ifcase 1\or \rangle\or \bigr\rangle\or \bigr\rangle\or \biggr\rangle\or \biggr\rangle\or \right\rangle\fi } } } { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } = \sum_{j\subset { { \ifcase 6\or \ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \bigl\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \bigl\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \biggl\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \biggl\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \left\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\fi } } } e^{\psi(\lambda e_j-\theta ) - \psi(-\theta)}.\end{aligned}\ ] ] by the concavity of @xmath303 , we have @xmath304 so that @xmath305\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } \le \sum_{j\subset { { \ifcase 6\or \ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \bigl\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \bigl\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \biggl\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \biggl\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \left\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\fi } } } e^{\lambda { { \ifcase 1\or \langle\or \bigl\langle\or \bigl\langle\or \biggl\langle\or \biggl\langle\or \left\langle\fi } \nabla \psi(-\theta ) , e_j { \ifcase 1\or \rangle\or \bigr\rangle\or \bigr\rangle\or \biggr\rangle\or \biggr\rangle\or \right\rangle\fi}}+\delta\lambda^2|e_j|^2 } \le c_{d,\lambda,\delta } e^{\lambda \sqrt{d } { { \ifcase 1\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \nabla \psi(-\theta ) { \ifcase 1\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}}}. \label{esscher - check - nec-1}\ ] ] in the same way , we have @xmath306\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } & \le \sum_{j\subset { { \ifcase 6\or \ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \bigl\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \bigl\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \biggl\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \biggl\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \left\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\fi } } } { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } { { \ifcase 1\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } x^{(\lambda e_j-\theta ) } { \ifcase 1\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}}^2 { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } e^{\psi(\lambda e_j-\theta ) - \psi(-\theta ) } , \notag \\ & \le c_{d,\lambda,\delta } e^{\lambda \sqrt{d } { { \ifcase 1\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \nabla \psi(-\theta ) { \ifcase 1\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}}}\sum_{j\subset { { \ifcase 6\or \ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \bigl\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \bigl\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \biggl\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \biggl\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \left\ { } 1,\dots , d { \ifcase 6\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\fi } } } { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } { { \ifcase 1\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } x^{(\lambda e_j-\theta ) } { \ifcase 1\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}}^2 { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}. \label{esscher - check - nec-2}\end{aligned}\ ] ] now , by differentiation of @xmath273 it is easy to check that @xmath307 which implies @xmath308\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } = \operatorname{tr\ ! } { { \ifcase 6\or ( \or \bigl(\or \bigl(\or \biggl(\or \biggl(\or \left(\fi } { \operatorname{d\!}^{\,2}\!}\psi(\lambda e_j - \theta ) { \ifcase 6\or ) \or \bigr)\or \bigr)\or \biggr)\or \biggr)\or \right)\fi } } + \operatorname{tr\ ! } { { \ifcase 6\or ( \or \bigl(\or \bigl(\or \biggl(\or \biggl(\or \left(\fi } \nabla \psi(\lambda e_j - \theta)^{\otimes 2 } { \ifcase 6\or ) \or \bigr)\or \bigr)\or \biggr)\or \biggr)\or \right)\fi}}.\ ] ] the assumption implies that @xmath309 ( for the partial order on symmetric matrices induced by nonnegative symmetric matrices ) then @xmath310 is a bounded function of @xmath145 and in turn @xmath311 has a linear growth by the fundamental formula of calculus .
consequently , for every @xmath312 , @xmath308\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } \le c { { \ifcase 6\or ( \or \bigl(\or \bigl(\or \biggl(\or \biggl(\or \left(\fi } 1 + { { \ifcase 1\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \theta { \ifcase 1\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}}^2 { \ifcase 6\or ) \or \bigr)\or \bigr)\or \biggr)\or \biggr)\or \right)\fi}}.\ ] ] plugging this into and using and we obtain @xmath313\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } \le c { { \ifcase 6\or ( \or \bigl(\or \bigl(\or \biggl(\or \biggl(\or \left(\fi } 1 + { { \ifcase 1\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \theta { \ifcase 1\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}}^2 { \ifcase 6\or ) \or \bigr)\or \bigr)\or \biggr)\or \biggr)\or \right)\fi}}$ ]
we consider a @xmath89-dimensional it process @xmath7}$ ] solution to the stochastic differential equation ( sde ) @xmath314 where @xmath91}$ ] is a @xmath92-dimensional standard brownian motion , @xmath93}$ ] is the stopped process at time @xmath94 , @xmath95\times { \cal c}([0,t],\r^d)\to \r^d$ ] and @xmath96\times { \cal c}([0,t],\r^d)\to { \cal m}(d , q)$ ] are measurable with respect to the canonical predictable @xmath315-field on @xmath88\times { \cal c}([0,t],\r^d ) $ ] . for further details we refer to @xcite , p. 124
thus , if @xmath316 and @xmath317 for every @xmath318,\r^d)$ ] , @xmath9 is a usual diffusion process with drift @xmath319 and diffusion coefficient @xmath320 .
if @xmath321 and @xmath322 for every @xmath318,\r^d)$ ] where @xmath323 , then @xmath9 is _ the continuous euler scheme with step _ @xmath324 of the above diffusion with drift @xmath319 and diffusion coefficient @xmath320 . an easy adaptation of standard proofs for regular sde s show
( see @xcite ) that strong existence and uniqueness of solutions for ( [ sdext ] ) follows from the following assumption @xmath325,\ ; \forall\ , x,\ , y\!\in { \cal c}([0,t],\r^d),\ , { { \ifcase 1\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } b(t , y)-b(t , x ) { \ifcase 1\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}}+ { { \ifcase 6\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \sigma(t , y)-\sigma(t , x ) { \ifcase 6\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi } } \le
c_{b,\sigma } { { \ifcase 6\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } x - y { \ifcase 6\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}_{\scriptscriptstyle{\infty}}}. \end{cases}\end{aligned}\ ] ] our aim is to devise an adaptive variance reduction method inspired from section [ dimfinie ] for the computation of @xmath104\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}\ ] ] where @xmath37 is an borel functional defined on @xmath87,\r^d)$ ] such that @xmath326\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}>0 \quad \text{and}\quad f(x)\!\in l^2(\p).\ ] ] in this functional setting , girsanov theorem will play the role of the invariance of lebesgue measure by translation .
the translation process that we consider in this section is of the form @xmath327 where @xmath22 is defined for every @xmath328,\r^d)$ ] and @xmath329 by @xmath330 \times { \cal c}([0,t],\r^d ) \rightarrow { \cal m}(q , p),\ ] ] a bounded borel function and @xmath329 ( represented by a borel function ) for @xmath102 . in the sequel ,
we use the following notations @xmath331 , @xmath332 where @xmath333 denotes the solution to @xmath334 .
first we need the following standard abstract lemma .
[ girs ] suppose @xmath335 holds .
+ the sde @xmath336 satisfies the weak existence and uniqueness assumptions and for every non negative borel functional @xmath337,\r^{d+1 } ) \to \r_+$ ] and @xmath338,\r^{q})$ ] we have , with the above notations , @xmath339\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } = { \e\ ! { } { { \ifcase 4\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } g { { \ifcase 3\or ( \or \bigl(\or \bigl(\or \biggl(\or \biggl(\or \left(\fi } { x^{(\theta ) } } , \int_0^.\!\ ! { { \ifcase 1\or \langle\or \bigl\langle\or \bigl\langle\or \biggl\langle\or \biggl\langle\or \left\langle\fi } f(s , { x^{(\theta),s } } ) , \operatorname{d\!}w_s { \ifcase 1\or \rangle\or \bigr\rangle\or \bigr\rangle\or \biggr\rangle\or \biggr\rangle\or \right\rangle\fi } } + \int_0^. \!\ ! { { \ifcase 1\or \langle\or \bigl\langle\or \bigl\langle\or \biggl\langle\or \biggl\langle\or \left\langle\fi } f , \theta { \ifcase 1\or \rangle\or \bigr\rangle\or \bigr\rangle\or \biggr\rangle\or \biggr\rangle\or \right\rangle\fi}}(s , { x^{(\theta),s } } ) \operatorname{d\!}s { \ifcase 3\or ) \or \bigr)\or \bigr)\or \biggr)\or \biggr)\or \right)\fi } } \\ \times e^{- \int_0^t\!\ ! { { \ifcase 1\or \langle\or \bigl\langle\or \bigl\langle\or \biggl\langle\or \biggl\langle\or \left\langle\fi } { \theta^{(\theta)}}_s , \operatorname{d\!}w_s { \ifcase 1\or \rangle\or \bigr\rangle\or \bigr\rangle\or \biggr\rangle\or \biggr\rangle\or \right\rangle\fi } } -\frac{1}{2 } { { { \ifcase 6\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } { \theta^{(\theta)}}{\ifcase 6\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}_{\scriptscriptstyle{\!l^2_{t , q}}}}}^2 } { \ifcase 4\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}},\end{gathered}\ ] ] and @xmath340\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } = { \e\ ! { } { { \ifcase 4\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } g { { \ifcase 3\or ( \or \bigl(\or \bigl(\or \biggl(\or \biggl(\or \left(\fi } x , \int_0^. \!\ !
{ { \ifcase 1\or \langle\or \bigl\langle\or \bigl\langle\or \biggl\langle\or \biggl\langle\or \left\langle\fi } f(s , x^s ) , \operatorname{d\!}w_s { \ifcase 1\or \rangle\or \bigr\rangle\or \bigr\rangle\or \biggr\rangle\or \biggr\rangle\or \right\rangle\fi } } - \int_0^. \!\ ! { { \ifcase 1\or \langle\or \bigl\langle\or \bigl\langle\or \biggl\langle\or \biggl\langle\or \left\langle\fi } f , \theta { \ifcase 1\or \rangle\or \bigr\rangle\or \bigr\rangle\or \biggr\rangle\or \biggr\rangle\or \right\rangle\fi}}(s , x^s ) \operatorname{d\!}s { \ifcase 3\or ) \or \bigr)\or \bigr)\or \biggr)\or \biggr)\or \right)\fi}}\\ \times e^{\int_0^t \!\ ! { { \ifcase 1\or \langle\or \bigl\langle\or \bigl\langle\or \biggl\langle\or \biggl\langle\or \left\langle\fi } \theta_s , \operatorname{d\!}w_s { \ifcase 1\or \rangle\or \bigr\rangle\or \bigr\rangle\or \biggr\rangle\or \biggr\rangle\or \right\rangle\fi } } -\frac{1}{2 } { { { \ifcase 6\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \theta
{ \ifcase 6\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}_{\scriptscriptstyle{\!l^2_{t , q}}}}}^2 } { \ifcase 4\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}},\end{gathered}\ ] ] this is a straightforward application of theorem 1.11 , p.372 ( and the remark that immediately follows ) in @xcite once noticed that @xmath341 , @xmath342 and @xmath343 are predictable processes with respect to the completed filtration of @xmath344 .
@xmath133 @xmath108 the dolans exponential @xmath345}$ ] is a true martingale for any @xmath329 .
@xmath108 in fact , still following the above cited remark form @xcite , the above lemma holds true if we replace @xmath22 by any progressively measurable process @xmath346 such that @xmath347\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } < + \infty$ ] .
it follows from the first identity in lemma [ girs ] that for every bounded borel function @xmath348 \times { \cal c}([0,t ] , \r^d ) \to { \cal m}(q , p)$ ] and for every @xmath349 @xmath104\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } = { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } f({x^{(\theta ) } } ) e^{- \int_0^t \ ! { { \ifcase 1\or \langle\or \bigl\langle\or \bigl\langle\or \biggl\langle\or \biggl\langle\or \left\langle\fi } { \theta^{(\theta)}}_s , \operatorname{d\!}w_s { \ifcase 1\or \rangle\or \bigr\rangle\or \bigr\rangle\or \biggr\rangle\or \biggr\rangle\or \right\rangle\fi } } -\frac{1}{2 } { { { \ifcase 6\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } { \theta^{(\theta)}}{\ifcase 6\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}_{\scriptscriptstyle{\!l^2_{t , q}}}}}^2 } { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}},\ ] ] ( set @xmath350 ) .
so , finding the estimator with the lowest variance amounts to solving the minimization problem @xmath351\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}},\ ] ] using lemma [ girs ] with @xmath352 and @xmath353 yields @xmath354\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}.\ ] ] assume @xmath355 for some @xmath235 as well as assumptions ( [ nonvide ] ) and @xmath335 .
then function @xmath39 is finite on @xmath356 and @xmath45-convex . *
assume that the bounded matrix - valued borel function @xmath100 satisfies that @xmath357 has a non - atomic kernel on the event @xmath358 @xmath171 @xmath359\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}=0\ ] ] then for every finite dimensional subspace @xmath360 , @xmath361 . if furthermore @xmath362\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } \theta(s ) \operatorname{d\!}s > 0,\ ] ] then @xmath363 . *
the function @xmath39 is differentiable at every @xmath364 and the differential @xmath365 is characterized on every @xmath366 by @xmath367\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } , \notag \\ & \ ; = { \e\ ! { } { { \ifcase 4\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } f^2({x^{(-\theta ) } } ) e^ { { { \ifcase 6\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } { \theta^{(-\theta)}}{\ifcase 6\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}_{\scriptscriptstyle{\!l^2_{t , p}}}}^2 } { { \ifcase 3\or ( \or \bigl(\or \bigl(\or \biggl(\or \biggl(\or \left(\fi } 2 { { \ifcase 1\or \langle\or \bigl\langle\or \bigl\langle\or \biggl\langle\or \biggl\langle\or \left\langle\fi } { \theta^{(-\theta ) } } , \varphi({x^{(-\theta)}}{}^ { , . } ) \psi { { \ifcase 1\or \rangle\or \bigr\rangle\or \bigr\rangle\or \biggr\rangle\or \biggr\rangle\or \right\rangle\fi}}_{\scriptscriptstyle{\ !
l^2_{t , p } } } } - \int_0^t\!\ ! { { \ifcase 1\or \langle\or \bigl\langle\or \bigl\langle\or \biggl\langle\or \biggl\langle\or \left\langle\fi } \varphi({x^{(-\theta),s } } ) \psi_s , \operatorname{d\!}w_s { \ifcase 1\or \rangle\or \bigr\rangle\or \bigr\rangle\or \biggr\rangle\or \biggr\rangle\or \right\rangle\fi } } { \ifcase 3\or ) \or \bigr)\or \bigr)\or \biggr)\or \biggr)\or \right)\fi } } { \ifcase 4\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}.\label{diff2}\end{aligned}\ ] ] * remarks .
* @xmath108 for practical implementation , the
finite dimensional " statement is the only result of interest since it ensures that @xmath368 .
@xmath108 if @xmath369 and @xmath370 , the infinite - dimensional " assumption is always satisfied . @xmath371 as concerns the function @xmath39 , we rely on equality ( [ v3 ] ) . set @xmath372 .
owing to the hlder inequality , showing that this function is finite on the whole space @xmath373 amounts to proving that @xmath374\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } \le e^ { { { \ifcase 6\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \varphi { \ifcase 6\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}_{\scriptscriptstyle{\infty}}}^2 { { { \ifcase 6\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \theta { \ifcase 6\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}_{\scriptscriptstyle{\!l^2_{t , p } } } } } r(r+1)/2 } < + \infty.\ ] ] to show that @xmath39 goes to infinity at infinity , one proceeds as follows . using the trivial equality @xmath375 and the reverse hlder inequality with conjugate exponents @xmath376 we obtain @xmath377\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}^3 { \e\ ! { } { { \ifcase 4\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } e^{\frac{1}{2 } \int_0^t\ ! {
{ \ifcase 1\or \langle\or \bigl\langle\or \bigl\langle\or \biggl\langle\or \biggl\langle\or \left\langle\fi } \theta_s,\operatorname{d\!}w_s { \ifcase 1\or \rangle\or \bigr\rangle\or \bigr\rangle\or \biggr\rangle\or \biggr\rangle\or \right\rangle\fi } } - \frac{1}{8 } { { { \ifcase 6\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \theta { \ifcase 6\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}_{\scriptscriptstyle{\!l^2_{t , q}}}}}^2 } { \ifcase 4\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}^{-2 } , \\ & \ge { \e\ ! { } { { \ifcase 4\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } f^{2/3}(x ) e^{\frac{1}{12 } { { { \ifcase 6\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \theta { \ifcase 6\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}_{\scriptscriptstyle{\!l^2_{t , q}}}}}^2 } { \ifcase 4\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}^3\end{aligned}\ ] ] by the martingale property of the dolans exponential .
let @xmath378 such that @xmath379\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } > 0 $ ] .
we have then @xmath380\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}^3 $ ] , and by the conditional jensen inequality @xmath381\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } } { \ifcase 4\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}^3\\ & = { \e\ ! { } { { \ifcase 4\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } { \boldsymbol{1 } _ { { { \ifcase 1\or \ { } f^2(x ) \ge \varepsilon { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \bigl\ { } f^2(x ) \ge \varepsilon { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \bigl\ { } f^2(x ) \ge \varepsilon { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \biggl\ { } f^2(x ) \ge \varepsilon { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \biggl\ { } f^2(x ) \ge \varepsilon { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \left\ { } f^2(x ) \ge \varepsilon { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\fi}}}}e^{\frac{1}{12 } { \p\ ! { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } f^2(x ) \ge \varepsilon { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } { \e\ ! { } { { \ifcase 2\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } { { { \ifcase 6\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \theta { \ifcase 6\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}_{\scriptscriptstyle{\!l^2_{t , q}}}}}^2 { \boldsymbol{1 } _ { { { \ifcase 1\or \ { } f^2(x ) \ge \varepsilon { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \bigl\ { } f^2(x ) \ge \varepsilon { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \bigl\ { } f^2(x ) \ge \varepsilon { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \biggl\ { } f^2(x ) \ge \varepsilon { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \biggl\ { } f^2(x ) \ge \varepsilon { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \left\ { } f^2(x ) \ge \varepsilon { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\fi } } } } { \ifcase 2\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } } { \ifcase 4\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}^3 .\end{aligned}\ ] ] now @xmath382\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } = \int_0^t \theta(s)^ * { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } \varphi_s(x^s)^ * \varphi_s(x^s ) { \boldsymbol{1 } _ { { { \ifcase 1\or \ { } f^2(x ) \ge \varepsilon { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \bigl\ { } f^2(x ) \ge \varepsilon { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \bigl\ { } f^2(x ) \ge \varepsilon { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \biggl\ { } f^2(x ) \ge \varepsilon { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \biggl\ { } f^2(x ) \ge \varepsilon { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \left\ { } f^2(x ) \ge \varepsilon { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\fi } } } } { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } \theta(s ) \operatorname{d\!}s
\ge 0.\ ] ] the assumption ( [ nonatomic ] ) implies that , for every @xmath329 , @xmath383\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } \theta(s ) \operatorname{d\!}s
\ge \int_0^t \theta(s)^*\ , { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } \varphi_s(x^s)^ * \,\varphi_s(x^s ) { \boldsymbol{1 } _ { { { \ifcase 1\or \ { } f^2(x ) \ge 0 { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \bigl\ { } f^2(x ) \ge 0 { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \bigl\ { } f^2(x ) \ge 0 { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \biggl\ { } f^2(x ) \ge 0 { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \biggl\ { } f^2(x ) \ge 0 { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\or \left\ { } f^2(x ) \ge 0 { \ifcase 1\or \\or \bigr\\or \bigr\\or \biggr\\or \biggr\\or \right\\fi}\fi } } } } { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } \theta(s ) \operatorname{d\!}s > 0,\ ] ] so that if @xmath63 runs over the compact sphere of a finite dimensional subspace @xmath3 of @xmath356 @xmath384\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } \theta(s ) \operatorname{d\!}s > 0,\ ] ] so that @xmath385\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } = + \infty,\ ] ] and one concludes by fatou s lemma using that @xmath386\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } > 0 $ ] .
the second claim easily follows from assumption ( [ general ] ) .
@xmath387 as a first step , we show that the random functional @xmath388 from @xmath356 into @xmath389 ( @xmath390 ) , is differentiable . indeed , it from the below inequality , @xmath391 where @xmath392 is clearly a bounded random functional from @xmath356 into @xmath389 , with an operator norm @xmath393 ( @xmath394 ( this follows from hlder and _ b.d.g . _
inequalities ) .
then , we derive that @xmath395 is differentiable form @xmath356 into every @xmath389 with differential @xmath396 .
this follows from standard computation based on ( [ diffl2lp ] ) , the elementary inequality @xmath397 and the fact that @xmath398 where we used both hlder and _ b.d.g . _ inequality . one concludes that @xmath399\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}$ ] is differentiable by using the @xmath400differentiability of @xmath401 with @xmath402 .
the second form of the gradient is obtained by a girsanov transform using lemma [ girs ] .
@xmath133 in view of a practical implementation of the procedure we are lead to consider some non trivial finite dimensional subspaces @xmath3 of @xmath356 .
the function @xmath39 being strictly @xmath45-convex on @xmath3 and going to infinity as @xmath403 goes to infinity , @xmath404 , the restriction of @xmath39 on @xmath3 attains a minimum @xmath405 which _ de facto _ becomes the target of the procedure .
furthermore , for every @xmath404 , @xmath406 and the quadratic function @xmath407 is a lyapunov function for the problem . like for the static framework investigated in section [ translation ] ,
our algorithm will be based on the representation ( [ diff2 ] ) for the differential @xmath408 of @xmath39 : in this representation the variance reducer @xmath63 appears inside the functional @xmath37 which makes easier a control at infinity in order to prevent from any early explosion of the procedure .
however , to this end we need to control the discrepancy between @xmath9 and @xmath409 .
this is the purpose of the following lemma .
[ xmoinsxtheta ] assume @xmath335 holds .
let @xmath100 be a bounded borel @xmath101-valued function defined on @xmath88 \times { \cal c}([0,t ] , \r^d)$ ] , let @xmath364 and let @xmath9 and @xmath277 denote a strong solutions of @xmath410 and @xmath411 driven by the same brownian motion .
then , for every @xmath412 , there exists a real constant @xmath413 such that @xmath414 } { { \ifcase 2\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } x_t - { x^{(\theta)}}_t { \ifcase 2\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi } } { \ifcase 4\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}_{\scriptscriptstyle{\!r } } } \le c_{b,\sigma } e^{c_{b , \sigma } t } { { \ifcase 4\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \int_0^t \!\ ! { { \ifcase 2\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \sigma(s , { x^{(\theta),s } } ) { \theta^{(\theta)}}_s { \ifcase 2\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi } } \operatorname{d\!}s { \ifcase 4\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}_{\scriptscriptstyle{\!r}}}. \ ] ] the proof follows the lines of the proof of the strong rate of convergence of the euler scheme ( see @xmath68 @xcite ) . @xmath133 the main result of this section is the following theorem .
suppose that assumption ( [ nonvide ] ) and @xmath335 hold .
let @xmath100 be a bounded borel @xmath101-valued function ( with @xmath102 ) defined on @xmath88 \times { \cal c}([0,t ] , \r^d)$ ] , and let @xmath37 be a functional @xmath37 satisfying @xmath415,\r^d),\qquad { { \ifcase 1\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } f(x ) { \ifcase 1\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi } } \le c_{_f}(1 + { { \ifcase 6\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } x { \ifcase 6\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}_{\scriptscriptstyle{\infty}}}^\lambda)\ ] ] for some positive exponent @xmath240 ( then @xmath416 for every @xmath417 ) . let @xmath3 be a finite dimensional subspace of @xmath356 spanned by an orthonormal basis @xmath418 .
let @xmath419 .
we define the algorithm by @xmath420 where @xmath421 satisfies ( [ stepcond ] ) , @xmath422 is a sequence of independent brownian motions for which @xmath423 is a strong solution to @xmath424 and for every standard brownian motion @xmath344 , every @xmath425-adapted @xmath117-valued process @xmath426}$ ] , @xmath427 where for @xmath419 @xmath428 then the recursive sequence @xmath429 a.s .
converges toward an @xmath134-valued ( squared integrable ) random variable @xmath51 . for a practical implementation of this algorithm
, we must have _ for all _ brownian motions @xmath430 a strong solution @xmath431 of @xmath424 .
in particular , this is the case if the driver @xmath100 is locally lipshitz ( in space ) or if @xmath9 is the continuous euler scheme of a diffusion with step @xmath324 ( using the driver @xmath432 ) .
note that if @xmath100 is continuous ( in space ) but not necessarily locally lipshitz , the euler scheme converges in law to the solution of the sde .
when the diffusion coefficient @xmath315 is bounded , it follows from lemma [ xmoinsxtheta ] that , for every @xmath433 , @xmath434 } { { \ifcase 2\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } x_t - { x^{(\theta)}}_t { \ifcase 2\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi } } { \ifcase 4\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}_{\scriptscriptstyle{\!r } } } \le c_{b,\sigma , t } { { \ifcase 6\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \varphi { \ifcase 6\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}_{\scriptscriptstyle{\infty } } } { { { \ifcase 6\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \theta { \ifcase 6\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}_{\scriptscriptstyle{\!l^2_{t , p } } } } } { { \ifcase 6\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \sigma { \ifcase 6\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}_{\scriptscriptstyle{\infty}}},\ ] ] where @xmath435\times{\cal c}([0,t ] , \r^d ) } { { \ifcase 6\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \sigma(t , x ) { \ifcase 6\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}}$ ] . first note that for every @xmath436 , the mean function @xmath66 of the algorithm reads @xmath437\or
\bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } = { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } \frac{e^{- { { \ifcase 6\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \varphi { \ifcase 6\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}_{\scriptscriptstyle{\infty } } } { { { \ifcase 6\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \theta { \ifcase 6\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}_{\scriptscriptstyle{\!l^2_{t , p}}}}}}}{1 + { { { \ifcase 6\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } { \theta^{(-\theta)}}{\ifcase 6\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}_{\scriptscriptstyle{\!l^2_{t , q}}}}}^{2\lambda+\eta } } { { \ifcase 1\or \langle\or \bigl\langle\or \bigl\langle\or \biggl\langle\or \biggl\langle\or \left\langle\fi } \operatorname{d\!}v_{|e}(\theta ) , \psi { { \ifcase 1\or \rangle\or \bigr\rangle\or \bigr\rangle\or \biggr\rangle\or \biggr\rangle\or \right\rangle\fi}}_{\scriptscriptstyle{\ ! l^2_{t , p } } } } { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}},\ ] ] so that , for every @xmath438 , @xmath439\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } > 0.\ ] ] it remains to check that for every @xmath440 , @xmath441 to apply the robbins - zygmund lemma which ensures the @xmath67 convergence of the procedure ( see section [ argmin - target ] )
. we first deal with the term @xmath442 .
let @xmath443 .
@xmath444 now @xmath445 one shows likewise that @xmath446 combining theses estimates shows that @xmath447 satisfies the linear growth assumption in @xmath448 .
if @xmath315 is unbounded it follows from assumption that , for every @xmath99\times { \cal c}([0,t],\r^d)$ ] , @xmath449 elementary computation based on and lemma [ girs ] yield @xmath450 for every @xmath417 ( assumption implies that @xmath451 for every @xmath417 ) .
following the same proof to the bounded case , we obtain easily the results with @xmath452 .
we conclude by noting that @xmath453 is an arbitrary parameter to cancel the denominator .
@xmath133 if the functional @xmath37 is bounded ( @xmath454 ) , we prove in the same way that the algorithm without correction , _
i.e. _ build with @xmath455 , a.s .
for the sake of simplicity we focus in this section on importance sampling by mean translation in a finite dimensional setting ( section [ translation ] ) although most of the comments below can also be applied at least in the path - dependent diffusions setting .
as proved by arouna ( see @xcite ) , we can consider a purely adaptive approach to reduce the variance .
it consists to perform the robbins - monro algorithm simultaneously with the monte carlo approximation .
more precisely , estimate @xmath86\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}$ ] by @xmath456 where @xmath457 is the _ same innovation _ as that used in the robbins - monro procedure @xmath458 .
this adaptive monte carlo procedure satisfies a central limit theorem with the optimal asymptotic variance @xmath459\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } { \ifcase 6\or ) \or \bigr)\or \bigr)\or \biggr)\or \biggr)\or \right)\fi } } \xrightarrow{\cal l } { \cal n}(0 , \sigma^2 _ * ) , \quad \text{whith } \quad \sigma^2 _ * = v(\theta^ * ) - { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } f(x ) { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}^2.\ ] ] this approach can be extended to the esscher transform when we use the same innovation @xmath460 ( see ) for the monte carlo procedure ( computing @xmath461 ) and the robbins - monro algorithm ( computing @xmath462 ) . likewise in the functional setting
we can combine the variance reduction procedure and the monte carlo simulations using the same brownian motion . in practice
, it is not clear that this adaptive monte carlo is better than the naive two stage procedure : performing first robbins - monro with a small number of iterations ( to get a rough estimate @xmath51 ) , then performing the monte carlo simulations with this optimized parameter .
as concerns the rate of convergence , once again this a regular stochastic algorithm behaves as described in usual stochastic approximation theory textbooks like @xcite , @xcite , @xcite .
so , as soon as the optimal variance reducer set is reduced to a single point @xmath51 , the procedure satisfies under quite standard assumptions a @xmath463 .
we will not enter into technicalities at this stage but only try to emphasize the impact of a renormalization factor @xmath464 like @xmath465 or @xmath466 induced by the function @xmath37 on the final " rate of convergence of the algorithm toward @xmath51 .
we will assume that @xmath467 and that @xmath468 for the sake of simplicity .
one can write @xmath469 the function @xmath470 corresponds to the case of a bounded function @xmath37 ( then @xmath471 ) . under simple integration assumptions
, one shows that @xmath39 is twice differentiable and that @xmath472\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}\ ] ] consequently the mean functions @xmath66 and @xmath473 related to @xmath259 and @xmath470 which read respectively @xmath474 are differentiable at @xmath51 and @xmath475 now , general results about clt say that if @xmath476 , @xmath477 with @xmath478 then @xmath479 where @xmath480 the mapping @xmath481 reaches its minimum at @xmath482 leading to the minimal asymptotic variance @xmath483\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}}{h_0'(y^*)^2}= \frac { { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } f^4(x)(\theta^*-x)^2e^{-\theta^*x } { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } } { { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } f^2(x)(x^2-\theta^*x+1 ) { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}}^2}\ ] ] by homogeneity .
so the optimal rate of convergence of the procedure is not impacted by the use of the normalizing function @xmath464 . however , coming back to condition ( [ condtcl ] ) , we see that this assumption on the coefficient @xmath484 is more stringent since @xmath485 ( in practice this factor can be rather large ) .
consequently , given the fact that @xmath486 is unknown to the user , this will induce a blind choice of @xmath484 biased to higher values . with the well - known consequence in practice
that if @xmath484 is too large the `` _ _ clt _ _ regime '' will take place later than it would with smaller values .
one solution to overcome this contradiction can be to make @xmath484 depend on @xmath487 and slowly decrease . as a conclusion , the algorithm never explodes ( and converges ) even for strongly unbounded functions @xmath37 which is a major asset compared to the version of the algorithm based on repeated projections . nevertheless , the normalizing factor which ensures the non - explosion of the procedure may impact the rate of convergence since it has an influence on the tuning of the step sequence ( which is always more or less blind " since it depends on the target @xmath51 .
in fact , we did not meet such difficulty in our numerical experiments reported below .
one classical way to overcome this problem can be to introduce the empirical mean of the algorithm implemented with a slowly decreasing step la rupert & poliak " ( see @xmath68 @xcite ) : set @xmath488 , @xmath489 and @xmath490 where @xmath491 denotes the regular robbins - monro algorithm defined by ( [ algorm ] ) starting at @xmath154 .
then @xmath492 converges toward @xmath51 and satisfies a clt with the optimal asymptotic variance ( [ asympvar ] ) .
see also a variant based on a gliding window developed in @xcite . in many applications ( see below with the spark spread options with the nig distribution ) the natural set of parameters
@xmath22 is not @xmath23 but an open connected subset of @xmath23 . nevertheless , as illustrated below , our unconstrained approach still works provided one can proceed a diffeomorphic change of parameter by setting @xmath493 where @xmath494 is a @xmath495-diffeomorphism with a bounded differential ( @xmath171 @xmath496 ) . as an illustration ,
let us consider the case where the state function @xmath497 of the procedure is designed so that @xmath498 where @xmath39 is the objective function to be minimized over @xmath22 and @xmath499 is a bounded _ positive _ borel function .
then , one replaces @xmath497 by @xmath500 and defines recursively a procedure on @xmath23 by @xmath501 in order to establish the @xmath67 convergence of @xmath502 to @xmath134 , one relies on a variant of robbins - monro algorithm , namely a stochastic gradient approach ( see @xcite for further details ) : one defines @xmath503 which turns out to be a lyapunov function for the new algorithm since @xmath504 if @xmath141 satisfies @xmath505 ) , one shows under the standard decreasing " assumption on the step sequence that @xmath506 and @xmath507 . if @xmath508 or @xmath509 , one easily derives that @xmath510 @xmath67 as @xmath511
first we consider a simple case to compare the two algorithms of section [ dimfinie ] .
the quantity to compute is @xmath104\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } = \int_{\r } f(x ) \ , p_{\operatorname{nig}}(x;\alpha,\beta,\delta,\mu ) \operatorname{d\!}x,\ ] ] where @xmath512 is the density of @xmath9 a normal inverse gaussian ( nig ) random variable of parameters @xmath513
i.e. @xmath514 , @xmath515 , @xmath516 , @xmath517 , @xmath518 where @xmath519 is a modified bessel function of the second kind and @xmath520 . we can summarize the two algorithms presented in section [ dimfinie ] , more precisely the variance reduction based on translation of the density ( see subsection [ translation ] ) and the one based on the esscher transform ( see subsection [ esscher ] ) , by the following simplified ( no computation of the variance ) pseudo - code : .... for n = 0 to m do x ~ nig(alpha , beta , mu , delta ) theta = theta - 1/(n+1000)*h1(theta , x ) for n = 0 to n do x ~ nig(alpha , beta , mu , delta ) mean = mean + f(x ) * p(x+theta)/p(x ) .... .... for n = 0 to m do x ~ nig(alpha , beta - theta , mu , delta ) theta = theta - 1/(n+1000)*h2(theta , x ) for n = 0 to n do x ~ nig(alpha , beta+theta , mu , delta ) mean = mean + f(x ) * exp(-theta*x ) mean = mean * exp(psi(theta ) ) .... * _ translation case .
_ we consider the function @xmath521 of the robbins - monro procedure of the first algorithm defined by @xmath522 where an analytic formulation of the derivative @xmath523 is easily obtained using the relation on the modified bessel function @xmath524 .
+ the assumption is satisfied with @xmath525 , and our results of subsection [ translation ] apply . * _ esscher transform . _ in the esscher approach we consider the function @xmath526 defined by @xmath527 note that @xmath273 is not well defined for every @xmath170 .
indeed , the cumulant generating function of the nig distribution is defined by @xmath528 for every @xmath529 .
moreover , we need @xmath530 to be well defined _
i.e. _ @xmath531 . to take account of these restrictions ,
we slightly modify the algorithm parametrization ( see subsection [ extension ] ) @xmath532 , and update @xmath533 in the robbins - monro procedure ( multiply the function @xmath534 by the derivative @xmath535 ) .
the payoff @xmath37 is a call option of strike @xmath536 , @xmath537 .
the parameters of the nig random variable @xmath9 are @xmath538 , @xmath539 , @xmath540 and @xmath541 .
the variance reduction obtained for different value of @xmath536 are summarized in the tabular [ tab - trans - esscher ] .
the number of iterations in the robbins - monro variance reduction procedure is @xmath542 and the number of monte carlo iterations is @xmath543 .
note that for each strike , the prices are computed using the same pseudo - random number generator initialized with the same _
seed_. .variance reduction for different strikes ( one dimensional nig example ) .
[ cols="<,^,^,^,>,^ , > " , ] 0 ( 2005 ) .
stability of stochastic approximation under verifiable conditions .
_ siam j. control optim .
_ , * 44*(1 ) , no .
1 , 283312 ( electronic ) . ( 2004 ) .
adaptative monte carlo method , a variance reduction technique .
_ monte carlo methods and appl .
_ , * 10*(1 ) , 124 .
_ numerical methods for stochastic processes _ , wiley series in probability and mathematical statistics : applied probability and statistics . a wiley - interscience publication .
john wiley & sons , inc .
, new york , 359 pp .
isbn : 0 - 471 - 54641 - 0 .
efficient variance reduction for functionals of diffusions by relative entropy , technical report , cermics - enpc ( france ) .
_ adaptive algorithms and stochastic approximation _ , * 22 * , _ applications of mathematics _ , transl . from french by s. wilson , springer - verlag , berlin .
_ stochastic approximation procedure with randomly varying truncations _ , scientia sinica series .
convergence and robustness of the robbins - monro algorithm truncated at randomly varying bounds , _ stoch .
_ , * 27*(2 ) , 217231 .
_ iterative random models _ , transl . from french , springer - verlag
_ monte carlo methods in financial engineering _ , springer .
asymptotically optimal importance sampling and stratification for pricing path - dependent options .
finance _ * 9*(2 ) , no .
2 , 117152 . ( 2000 ) .
weak approximation of killed diffusion using euler schemes , _ stoch .
_ , * 87*(2 ) , 167197 .
optimal importance sampling parameter search for lvy processes via stochastic approximation , pre - print .
stochastic approximation and recursive algorithms and applications , springer .
( 2007 ) . _
algorithmes stochastiques et options parisiennes _ ,
thse de lenpc , 151p .
an adaptive scheme for the approximation of dissipative systems , _ stoch .
proc . appl .
_ , * 117*(10 ) , 14911518 .
( 1998 ) . _ continuous martingales and brownian motion _ ,
@xmath544 edition , springer , berlin , 1998 ( @xmath545 edition , 1990 ) .
asymptotic almost sure efficiency of averaged stochastic algorithms .
siam j. control optim . * 39*(1 ) , 4972 ( electronic ) .
_ diffusions , markov processes and martingales _ , @xmath546 edition , cambridge mathematical library .
we propose below the proof of the slight extension of the regular robbins - monro algorithm when @xmath547 is not reduced to a single equilibrium point .
the key is still the convergence theorem for non negative super - martingales .
set @xmath548 , @xmath549 .
let @xmath550 .
then @xmath551 where @xmath552\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } = h(\theta_n , z_{n+1 } ) - h(\theta_n),\ ] ] is an increment of ( local ) martingale satisfying @xmath553\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } \le c(1 + { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } |\theta_n-\theta^*|^2 { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}})$ ] owing to the assumptions on @xmath259 and schwarz inequality which also implies that @xmath554\or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } \le \frac 12 { { \ifcase 6\or ( \or \bigl(\or \bigl(\or \biggl(\or \biggl(\or \left(\fi } { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } { { \ifcase 1\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \theta_n-\theta^ * { \ifcase 1\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}}^2 { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } + { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } { { \ifcase 1\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } h(\theta_n , z_{n+1 } ) { \ifcase 1\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}}^2 { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi } } } { \ifcase 6\or ) \or \bigr)\or \bigr)\or \biggr)\or \biggr)\or \right)\fi } } \le c(1 + { \e\ ! { } { { \ifcase 6\or [ \or \bigl[\or \bigl[\or \biggl[\or \biggl[\or \left[\fi } { { \ifcase 1\or \lvert\or \bigl\lvert\or \bigl\lvert\or \biggl\lvert\or \biggl\lvert\or \left\lvert\fi } \theta_n-\theta^ * { \ifcase 1\or \rvert\or \bigr\rvert\or \bigr\rvert\or \biggr\rvert\or \biggr\rvert\or \right\rvert\fi}}^2 { \ifcase 6\or ] \or \bigr]\or \bigr]\or \biggr]\or \biggr]\or \right]\fi}}},\ ] ] for an appropriate real constant @xmath42 .
then , one shows by induction on @xmath487 from ( [ ineql2 ] ) that @xmath555 is square integrable for every @xmath556 and that @xmath557 is integrable , hence a true martingale increment .
now , one derives from the assumptions ( [ stepcond ] ) and ( [ ineql2 ] ) that @xmath558 is a ( non negative ) super - martingale with @xmath559 .
this uses the mean - reverting assumption .
hence @xmath560 is @xmath561-@xmath67 converging toward an integrable r.v .
@xmath562 . consequently , using that @xmath563 , one gets @xmath564 the super - martingale @xmath565 being @xmath566-bounded , one derives likewise that @xmath567 is @xmath566-bounded since @xmath568 now , a series with nonnegative terms which is upper bounded by an ( @xmath67 ) converging sequence , @xmath67 converges in @xmath233 so that @xmath569 it follows from ( [ cvrz ] ) that , @xmath561-@xmath67 , @xmath570 which is integrable since @xmath571 is @xmath566-bounded and consequently @xmath67 finite . let @xmath572 . set @xmath573 it follows from the @xmath67 finiteness of @xmath574 that @xmath575 @xmath67 .
now we consider the compact set @xmath576 .
it is separable so there exists an everywhere dense sequence in @xmath577 , denoted for convenience @xmath578 .
the above proof shows that @xmath561-@xmath67 , for every @xmath579 , @xmath580 as @xmath581 .
then set @xmath582 which satisfies @xmath583 .
assume @xmath584 .
up to two successive extractions , there exists a subsequence @xmath585 such that @xmath586 the function @xmath66 being continuous @xmath587 which implies that @xmath588 .
hence @xmath589 .
then any limiting value @xmath590 of the sequence @xmath591 will satisfy @xmath592 which in turn implies that @xmath593 by considering a subsequence @xmath594 .
so , @xmath595 is the unique limiting value of the sequence @xmath596
@xmath171 @xmath597 as @xmath581 .
the fact that the resulting random vector @xmath598 is square integrable follows from fatou s lemma and the @xmath599-boundedness of the sequence @xmath600.@xmath133 | we propose an _ unconstrained _ stochastic approximation method of finding the optimal measure change ( in an _ a priori _ parametric family ) for monte carlo simulations .
we consider different parametric families based on the girsanov theorem and the esscher transform ( or exponential - tilting ) . in a multidimensional gaussian framework , arouna uses a projected robbins - monro procedure to select the parameter minimizing the variance ( see @xcite ) . in our approach ,
the parameter ( scalar or process ) is selected by a classical robbins - monro procedure without projection or truncation . to obtain this unconstrained algorithm
we intensively use the regularity of the density of the law without assume smoothness of the payoff .
we prove the convergence for a large class of multidimensional distributions and diffusion processes .
we illustrate the effectiveness of our algorithm via pricing a basket payoff under a multidimensional nig distribution , and pricing a barrier options in different markets . _
key words : stochastic algorithm , robbins - monro , importance sampling , esscher transform , girsanov , nig distribution , barrier options . _ |
let @xmath1 be an edge - labeled directed graph ( referred to hereafter simply as a graph ) , where @xmath2 is the vertex set , @xmath3 is the edge set , and @xmath4 is the edge labeling taking values on a finite alphabet @xmath5 @xcite .
we require that the labeling @xmath6 is deterministic : edges that start at the same vertex have distinct labels .
we further assume that @xmath7 has finite memory @xcite . the one - dimensional ( 1-d ) _ constraint _ @xmath8 that is presented by @xmath7 is defined as the set of all words that are generated by paths in @xmath7 ( i.e. , the words are obtained by reading - off the edge labels of such paths ) .
examples of 1-d constraints include runlength - limited ( rll ) constraints @xcite , symmetric runlength - limited ( srll ) constraints @xcite , and the charge constraints @xcite .
the capacity of @xmath9 is given by @xmath10 an @xmath11-track _ parallel encoder _ for @xmath8 at rate @xmath12 is defined as follows ( see figure [ fig : encoder ] ) .
-track parallel encoder.,title="fig : " ] 1 . at stage @xmath13 ,
the encoder ( which may be state - dependent ) receives as input @xmath14 ( unconstrained ) information bits . 2 .
the output of the encoder at stage @xmath15 is a word @xmath16 of length @xmath11 over @xmath5 .
3 . for @xmath17 ,
@xmath18th track _
@xmath19 of any given length @xmath20 , belongs to @xmath9 .
4 . there are integers @xmath21 such that the encoder is _ @xmath22-sliding - block decodable _ ( in short , @xmath22-sbd ) : for @xmath23 , the @xmath14 information bits which were input at stage @xmath15 are uniquely determined by ( and can be efficiently calculated from ) @xmath24 .
the decoding window size of the encoder is @xmath25 , and it is desirable to have a small window to avoid error propagation . in this work , we will be mainly focusing on the case where @xmath26 , in which case the decoding requires no look - ahead . in @xcite
, it was shown that by introducing parallelism , one can reduce the window size , compared to conventional serial encoding .
furthermore , it was shown that as @xmath11 tends to infinity , there are @xmath27-sbd parallel encoders whose rates approach @xmath28 . a key step in @xcite
is using some perturbation of the conditional probability distribution on the edges of @xmath7 , corresponding to the maxentropic stationary markov chain on @xmath7 .
however , it is not clear how this perturbation should be done : a naive method will only work for unrealistically large @xmath11 .
also , the proof in @xcite of the @xmath27-sbd property is only probabilistic and does not suggest encoders and decoders that have an acceptable running time . in this work , we aim at making the results of @xcite more tractable . at the expense of possibly increasing the memory of the encoder ( up to the memory of @xmath7 )
we are able to define a suitable perturbed distribution explicitly , and provide an efficient algorithm for computing it .
furthermore , the encoding and decoding can be carried out in time complexity @xmath29 , where the multiplying constants in the @xmath30 term are polynomially large in the parameters of @xmath7 .
denote by @xmath31 the diameter of @xmath7 ( i.e. , the longest shortest path between any two vertices in @xmath7 ) and let @xmath32 be the adjacency matrix of @xmath7 , i.e. , @xmath33 is the number of edges in @xmath7 that start at vertex @xmath34 and terminate in vertex @xmath35 .
our main result , specifying the rate of our encoder , is given in the next theorem .
[ theo : main ] let @xmath7 be a deterministic graph with memory @xmath36 .
for @xmath11 sufficiently large , one can efficiently construct an @xmath11-track @xmath37-sbd parallel encoder for @xmath38 at a rate @xmath12 such that @xmath39 where @xmath40 ( respectively , @xmath41 ) is the smallest ( respectively , largest ) _ nonzero _ entry in @xmath42 .
the structure of this paper is as follows . in section [ sec : twodimensionalconstraints ] we show how parallel encoding can be used to construct an encoder for a 2-d constraint .
as we will show , a parallel encoder is essentially defined through what we term a multiplicity matrix .
section [ sec : encoder ] defines how our parallel encoder works , assuming its multiplicity matrix is given .
then , in section [ sec : computingmultiplicitymatrix ] , we show how to efficiently calculate a good multiplicity matrix .
although 2-d constraints are our main motivation , section [ sec:1d ] shows how our method can be applied to 1-d constraints .
section [ sec : twoopt ] defines two methods by which the rate of our encoder can be slightly improved .
finally , in section [ sec : fastenumerativecoding ] we show a method of efficiently realizing a key part of our encoding procedure .
our primary motivation for studying parallel encoding is to show an encoding algorithm for a family of two - dimensional ( 2-d ) constraints .
the concept of a 1-d constraint can formally be generalized to two dimensions ( see @xcite ) .
examples of 2-d constraints are 2-d rll constraints @xcite , 2-d srll constraints @xcite , and the so - called square constraint @xcite .
let @xmath43 be a given 2-d constraint over a finite alphabet @xmath5 .
we denote by @xmath44 $ ] the set of all @xmath45 arrays in @xmath43 .
the capacity of @xmath43 @xcite is given by @xmath46\right| } \ ; .\ ] ] suppose we wish to encode information to an @xmath45 array which must satisfy the constraint @xmath43 ; namely , the array must be an element of @xmath44 $ ] . as a concrete example , consider the square constraint @xcite : its elements are all the binary arrays in which no two ` 1 ' symbols are adjacent on a row , column , or diagonal .
we first partition our array into two alternating types of vertical strips : _ data strips _ having width @xmath47 , and _ merging strips _ having width @xmath48 . in our example , let @xmath49 and @xmath50 ( see figure [ fig : example : square ] ) .
@xmath51 secondly , we select a graph @xmath1 with a labeling @xmath52 $ ] such that @xmath53 , i.e. , each path of length @xmath20 in @xmath7 generates a ( column ) word which is in @xmath54 $ ] .
we then fill up the data strips of our @xmath45 array with @xmath55 arrays corresponding to paths of length @xmath20 in @xmath7 .
thirdly , we assume that the choice of @xmath48 allows us to fill up the merging strips in a row - by - row ( causal ) manner , such that our @xmath45 array is in @xmath43 .
any 2-d constraint @xmath43 for which such @xmath47 , @xmath48 , and @xmath7 can be found , is in the family of constraints we can code for ( for example , the 2-d srll constraints belong to this family @xcite ) .
consider again the square constraint : a graph which produces _ all _ @xmath55 arrays that satisfy this constraint is given in figure [ fig : example : g ] .
also , for @xmath56 , we can take the merging strips to be all - zero .
( there are cases , such as the 2-d srll constraints , where determining the merging strips may be less trivial @xcite . )
whose paths generate all @xmath57 arrays satisfying the square constraint .
the label of an edge is given by the label of the vertex it enters.,title="fig : " ] suppose we have an @xmath37-sbd parallel encoder for @xmath38 at rate @xmath12 with @xmath58 tracks .
we may use this parallel encoder to encode information in a row - by - row fashion to our @xmath45 array : at stage @xmath15 we feed @xmath14 information bits to our parallel encoder .
let @xmath16 be the output of the parallel encoder at stage @xmath15 .
we write @xmath59 to row @xmath15 of the @xmath18th data strip , and then appropriately fill up row @xmath15 of the merging strips .
decoding of a row in our array can be carried out based only on the contents of that row and the previous @xmath36 rows . since
@xmath14 information bits are mapped to @xmath60 symbols in @xmath5 , the rate at which we encode information to the array is @xmath61 we note the following tradeoff : typically , taking larger values of @xmath47 ( while keeping @xmath48 constant ) will increase the right - hand side of the above inequality .
however , the number of vertices and edges in @xmath7 will usually grow exponentially with @xmath47 .
thus , @xmath47 is taken to be reasonably small .
note that in our scheme , a single error generally results in the loss of information stored in the respective vertical sliding - block window .
namely , a single corrupted entry in the array may cause the loss of @xmath62 rows .
thus , our method is only practical if we assume an error model in which whole rows are corrupted by errors .
this is indeed the case if each row is protected by an error - correcting code ( for example , by the use of unconstrained positions @xcite ) .
let @xmath63 be a positive integer which will shortly be specified .
the @xmath63 words @xmath19 , @xmath64 , that we will be writing to the first @xmath63 tracks are all generated by paths of length @xmath20 in @xmath7 . in
what follows , we find it convenient to regard the @xmath65 arrays @xmath66 as ( column ) words of length @xmath20 of some new 1-d constraint , which we define next .
@xmath63th kronecker power _ of @xmath1 , denoted by @xmath67 , is defined as follows .
the vertex set @xmath68 is simply the @xmath63th cartesian power of @xmath2 ; that is , @xmath69 an edge @xmath70 goes from vertex @xmath71 to vertex @xmath72 and is labeled @xmath73 whenever for all @xmath64 , @xmath74 is an edge from @xmath75 to @xmath76 labeled @xmath77 .
note that a path of length @xmath20 in @xmath78 is just a handy way to denote @xmath63 paths of length @xmath20 in @xmath7 .
accordingly , the @xmath65 arrays @xmath79 are the words of length @xmath20 in @xmath80 .
let @xmath7 be as in section [ sec : introduction ] and let @xmath32 be the adjacency matrix of @xmath7 . denote by @xmath81 the @xmath82 all - one row vector .
the description of our @xmath11-track parallel encoder for @xmath8 makes use of the following definition .
a @xmath83 nonnegative integer matrix @xmath84 is called a ( valid ) _ multiplicity matrix _ with respect to @xmath7 and @xmath11 if @xmath85 ( while any multiplicity matrix will produce a parallel encoder , some will have higher rates than others . in section [ sec : computingmultiplicitymatrix ] , we show how to compute multiplicity matrices @xmath86 that yield rates close to @xmath28 . )
recall that we have at our disposal @xmath11 tracks .
however , we will effectively be using only the first @xmath87 tracks in order to encode information .
the last @xmath88 tracks will all be equal to the first track , say .
write @xmath89 .
a vertex @xmath90 is a _ typical vertex _
( with respect to @xmath86 ) if for all @xmath34 , the vertex @xmath91 appears as an entry in @xmath92 exactly @xmath93 times . also , an edge @xmath94 is a _ typical edge _ with respect to @xmath86 if for all @xmath95 , there are exactly @xmath96 entries @xmath74 which as edges in @xmath7start at vertex @xmath34 and terminate in vertex @xmath35 .
a simple computation shows that the number of outgoing typical edges from a typical vertex equals @xmath97 ( where @xmath98 ) .
for example , in the simpler case where @xmath7 does not contain parallel edges ( @xmath99 ) , we are in effect counting in permutations with repetitions , each time for a different vertex @xmath91 .
the encoding process will be carried out as follows .
we start at some fixed typical vertex @xmath100 . out of the set of outgoing edges from @xmath101
, we consider only typical edges .
the edge we choose to traverse is determined by the information bits .
after traversing the chosen edge , we arrive at vertex @xmath102 . by , @xmath102 is also a typical vertex , and the process starts over .
this process defines an @xmath11-track parallel encoder for @xmath8 at rate @xmath103 this encoder is @xmath37-sbd , where @xmath36 is the memory of @xmath7 .
consider now how we map @xmath14 information bits into an edge choice @xmath104 at any given stage @xmath15 .
assuming again the simpler case of a graph with no parallel edges , a natural choice would be to use an instance of enumerative coding @xcite .
specifically , suppose that for @xmath105 , a procedure for encoding information by an @xmath106-bit binary vector with hamming weight @xmath107 were given .
suppose also that @xmath108 .
we could use this procedure as follows .
first , for @xmath109 and @xmath110 , the binary word given as output by the procedure will define which @xmath111 of the possible @xmath112 entries in @xmath113 will be equal to the edge in @xmath3 from the vertex @xmath114 to itself ( if no such edge exists , then @xmath115 ) .
having chosen these entries , we run the procedure with @xmath116 and @xmath117 to choose from the remaining @xmath118 entries those that will contain the edge in @xmath3 from @xmath114 to @xmath119 .
we continue this process , until all @xmath112 entries in @xmath113 containing an edge outgoing from @xmath114 have been picked .
next , we run the procedure with @xmath120 and @xmath121 , and so forth .
the more general case of a graph containing parallel edges will include a preliminary step : encoding information in the choice of the @xmath96 edges used to traverse from @xmath34 to @xmath35 ( @xmath33 options for each such edge ) .
a fast implementation of enumerative coding is presented in section [ sec : fastenumerativecoding ] .
the above - mentioned preliminary step makes use of the schnhage strassen integer - multiplication algorithm @xcite , and the resulting encoding time complexity is proportional , with a negligible penalty in terms of rate : fix @xmath34 and @xmath35 , and let @xmath122 be an integer design parameter .
assume for simplicity that @xmath123 .
the number of vectors of length @xmath122 over an alphabet of size @xmath33 is obviously @xmath124 .
so , we can encode @xmath125 bits through the choice of such a vector . repeating this process
, we can encode @xmath126 bits through the choice of @xmath127 such vectors .
the concatenation of these vectors is taken to represent our choice of edges .
note that the encoding process is linear in @xmath11 for constant @xmath122 .
also , our losses ( due to the floor function ) become negligible for modestly large @xmath122 . ] to @xmath128 .
it turns out that this is also the decoding time complexity .
further details are given in section [ sec : fastenumerativecoding ] .
the next section shows how to find a good multiplicity matrix , i.e. , a matrix @xmath86 such that @xmath129 is close to @xmath28 .
in order to enhance the exposition of this section , we accompany it by a running example ( see figure [ fig : running1 ] ) . throughout this section ,
we assume a probability distribution on the edges of @xmath7 , which is the maxentropic stationary markov chain @xmath130 on @xmath7 @xcite . without real loss of generality
, we can assume that @xmath7 is irreducible ( i.e. , strongly - connected ) , in which case @xmath130 is indeed unique .
let the matrix @xmath131 be the transition matrix induced by @xmath130 , i.e. , @xmath132 is the probability of traversing an edge from @xmath91 to @xmath133 , conditioned on currently being at vertex @xmath91 .
let @xmath134 be the @xmath82 row vector corresponding to the stationary distribution on @xmath2 induced by @xmath135 ; namely , @xmath136 and @xmath137 .
let @xmath138 and define @xmath139 taking the number of tracks in our running example ( figure [ fig : running1 ] ) to be @xmath140 gives @xmath141 .
also , our running example has @xmath142 and @xmath143 thus , @xmath144 and @xmath145 note that @xmath146 also , observe that hold when we substitute @xmath147 for @xmath86 . thus ,
if all entries of @xmath147 were integers , then we could take @xmath86 equal to @xmath147 . in a way , that would be the best choice we could have made : by using stirling s approximation , we could deduce that @xmath129 approaches @xmath28 as @xmath148 .
however , the entries of @xmath147 , as well as @xmath149 , may be non - integers .
we say that an _ integer _
matrix @xmath150 is a _ good quantization _ of @xmath151 if @xmath152{$\sum_{j \in v } { \tilde{p}}_{i , j}$ } } \;\;\leq\;\ ; { \left\lceil\sum_{j \in v } { p}_{i , j}\right\rceil } \ ; , \\ \label{eq : preflow_entries } { \left\lfloor{p}_{i , j}\right\rfloor } & \leq & \textstyle { \makebox[3.4em]{${\tilde{p}}_{i , j}$ } } \;\;\leq\;\ ; { \left\lceil{p}_{i , j}\right\rceil } \ ; , \quad \textrm{and---}\qquad \\ \label{eq : preflow_sumi } \textstyle { \left\lfloor\sum_{i \in v } { p}_{i , j}\right\rfloor } & \leq & \textstyle { \makebox[3.4em]{$\sum_{i \in v } { \tilde{p}}_{i , j}$ } } \;\;\leq\;\ ; { \left\lceil\sum_{i \in v } { p}_{i , j}\right\rceil } \ ; .\end{aligned}\ ] ] namely , a given entry in @xmath153 is either the floor or the ceiling of the corresponding entry in @xmath147 , and this also holds for the sum of entries of a given row or column in @xmath153 ; moreover , the sum of entries in both @xmath153 and @xmath147 are exactly equal ( to @xmath154 ) .
[ lemm : preflow ] there exists a matrix @xmath153 which is a good quantization of @xmath147 .
furthermore , such a matrix can be found by an efficient algorithm . .
an edge labeled @xmath155 has lower and upper bounds @xmath156 and
@xmath157 , respectively.,title="fig : " ] we recast as an integer flow problem ( see figures [ fig : preflow ] and [ fig : running3 ] ) . consider the following flow network , with upper and lower bounds on the flow through the edges @xcite .
the network has the vertex set @xmath158 with source @xmath159 and target @xmath160 .
henceforth , when we refer to the upper ( lower ) bound of an edge , we mean the upper ( lower ) bound on the flow through it .
there are four kinds of edges : 1 .
an edge @xmath161 with upper and lower bounds both equaling to @xmath162 .
2 . @xmath163 for every @xmath91 , with the upper and lower bounds @xmath164 and @xmath165 , respectively .
3 . @xmath166 for every @xmath95 , with the upper and lower bounds @xmath167 and @xmath168 , respectively .
4 . @xmath169 for every @xmath133 , with the upper and lower bounds @xmath170 and @xmath171 , respectively .
we claim that can be satisfied if a legal integer flow exists : simply take @xmath172 as the flow on the edge from @xmath173 to @xmath174 .
it is well known that if a legal _ real _ flow exists for a flow network with integer upper and lower bounds on the edges , then a legal _ integer _
flow exists as well ( * ? ? ?
* theorem 6.5 ) .
moreover , such a flow can be efficiently found @xcite .
to finish the proof , we now exhibit such a legal real flow : 1 .
the flow on the edge @xmath161 is @xmath175 .
the flow on an edge @xmath163 is @xmath176 .
the flow on an edge @xmath166 is @xmath177 .
the flow on an edge @xmath169 is @xmath178 . in running example 2 .
an edge labeled @xmath179 has lower and upper bounds @xmath180 and @xmath181 , respectively .
a legal real flow is given by @xmath156 .
a legal integer flow is given by @xmath182 .
the matrix @xmath153 resulting from the legal integer flow is given , as well as the matrix @xmath147 ( again).,title="fig : " ] @xmath183 for the remaining part of this section , we assume that @xmath153 is a good quantization of @xmath147 ( say , @xmath153 is computed by solving the integer flow problem in the last proof ) . the next lemma states that @xmath153 `` almost '' satisfies ( [ eq : sum_up ] ) .
[ lemm : disc ] let @xmath184 and @xmath185 .
then , for all @xmath91 , @xmath186 from , we get that for all @xmath91 , @xmath187 recall that is satisfied if we replace @xmath86 by @xmath147 . thus , by , we have that also holds if we replace @xmath188 by @xmath189 .
we conclude that @xmath190 .
the proof follows from the fact that entries of @xmath153 are integers , and thus so are those of @xmath191 and @xmath192
. the following lemma will be the basis for augmenting @xmath153 so that ( [ eq : sum_up ] ) is satisfied .
[ lemm : fmat ] fix two distinct vertices @xmath193 .
we can efficiently find a @xmath83 matrix @xmath194 with non - negative integer entries , such that the following three conditions hold . 1 .
[ eq : fsumentries ] @xmath195 2 . for all @xmath196 , [ eq : fsane ]
@xmath197 3 .
denote @xmath198 and @xmath199 .
then , for all @xmath91 , @xmath200 let @xmath201 be the vertices along a shortest path from @xmath202 to @xmath15 in @xmath7 . for all @xmath196 , define @xmath203 namely , @xmath204 is the number of edges from @xmath34 to @xmath35 along the path . conditions ( [ eq : fsumentries ] ) and ( [ eq : fsane ] ) easily follow from ( [ eq : f ] ) .
condition ( [ eq : fsurp ] ) follows from the fact that @xmath205 ( @xmath206 ) is equal to the number of edges along the path for which @xmath34 is the start ( end ) vertex of the edge
. the matrix @xmath153 will be the basis for computing a good multiplicity matrix @xmath86 , as we demonstrate in the proof of the next theorem .
[ theo : dfromptilde ] let @xmath207 be a good quantization of @xmath147 .
there exists a multiplicity matrix @xmath208 with respect to @xmath7 and @xmath11 , such that 1 .
[ it : dagreaterthandalpha ] @xmath209 for all @xmath95 , and 2 .
@xmath210 ( where @xmath154 is as defined in ) .
moreover , the matrix @xmath86 can be found by an efficient algorithm .
consider a vertex @xmath91 .
if @xmath211 , then we say that vertex @xmath34 has a _ surplus _ of @xmath212 . in this case , by lemma [ lemm : disc ] , we have that the surplus is equal to 1 . on the other hand , if @xmath213 then vertex @xmath34 has a _ deficiency _ of @xmath214 , which again is equal to 1 . of course , since @xmath215 , the total surplus is equal to the total deficiency , and both are denoted by @xmath216 : @xmath217 denote the vertices with surplus as @xmath218 and the vertices with deficiency as @xmath219 .
recalling the matrix @xmath220 from lemma [ lemm : fmat ] , we define @xmath221 we first show that @xmath86 is a valid multiplicity matrix .
note that @xmath222 .
thus , ( [ eq : sum_entries ] ) follows from ( [ eq : malpha ] ) , ( [ eq : preflow_sumij ] ) , and ( [ eq : fsumentries ] ) .
the definitions of surplus and deficiency vertices along with ( [ eq : fsurp ] ) give ( [ eq : sum_up ] ) .
lastly , recall that ( [ eq : sane_entries ] ) is satisfied if we replace @xmath96 by @xmath177 .
thus , by ( [ eq : preflow_entries ] ) , the same can be said for @xmath172 .
combining this with ( [ eq : fsane ] ) yields ( [ eq : sane_entries ] ) .
since the entries of @xmath223 are non - negative for every @xmath18 , we must have that @xmath209 for all @xmath95 .
this , together with ( [ eq : sum_entries ] ) and ( [ eq : preflow_sumij ] ) , implies in turn that @xmath210 . for the matrix @xmath153 in figure [ fig : running3 ]
, we have @xmath224 thus , @xmath225 .
namely , the vertex @xmath226 has a surplus while the vertex @xmath227 has a deficiency .
taking @xmath228 and @xmath229 we get @xmath230 now that theorem [ theo : dfromptilde ] is proved , we are in a position to prove our main result , theorem [ theo : main ] .
essentially , the proof involves using the stirling approximation and taking into account the various quantization errors introduced into @xmath86 .
the proof itself is given in the appendix .
the main motivation for our methods is 2-d constrained coding . however , in this section , we show that they might be interesting in certain aspects of 1-d coding as well .
given a labeled graph @xmath7 , a classic method for building an encoder for the 1-d constraint @xmath231 is the state - splitting algorithm @xcite .
the rate of an encoder built by @xcite approaches the capacity of @xmath231 . also , the word the encoder outputs has a corresponding path in @xmath7 , with the following favorable property : the probability of traversing a certain edge approaches the maxentropic probability of that edge ( assuming an unbiased source distribution ) .
however , what if we d like to build an encoder with a different probability distribution on the edges ?
this scenario may occur , for example , when there is a requirement that all the output words of a given length @xmath63 that are generated by the encoder have a prescribed hamming weight yielding a stationary markov chain with largest possible entropy , subject to a set of edges ( such as the set of edges with label ` 1 ' ) having a prescribed cumulative probability . ] .
more formally , suppose that we are given a labeled graph @xmath232 ; to make the exposition simpler , suppose that @xmath7 does not contain parallel edges .
let @xmath135 and @xmath233 be a transition matrix and a stationary probability distribution corresponding to a stationary ( but not necessarily maxentropic ) markov chain @xmath130 on @xmath7 .
we assume w.l.o.g . that each edge in @xmath7 has a positive conditional probability .
we are also given an integer @xmath11 , which we will shortly elaborate on .
we first describe our encoder in broad terms , so as that its merits will be obvious .
let @xmath234 and @xmath63 be as previously defined , and let @xmath235 be specified shortly .
we start at some fixed vertex @xmath236 . given @xmath237 information bits ,
we traverse a soon to be defined cyclic path of length @xmath63 in @xmath7 .
the concatenation of the edge labels along the path is the word we output .
of course , since the path is cyclic , the concatenation of such words is indeed in @xmath231 .
moreover , the path will have the following key property : the number of times an edge from @xmath34 to @xmath35 is traversed equals @xmath238 .
namely , if we uniformly pick one of the @xmath63 edges of the path , the probability of picking a certain edge @xmath239 is constant ( not a function of the input bits ) , and is equal to the probability of traversing @xmath239 on the markov chain @xmath130 , up to a small quantization error .
the rate @xmath240 of our encoder will satisfy ( [ eq : rbound ] ) , where we replace @xmath12 by @xmath240 and @xmath241 by the entropy of @xmath130 .
we would like to be able to exactly specify the path length @xmath63 as a design parameter .
however , we specify @xmath11 and get an @xmath63 between @xmath11 and @xmath242
. our encoding process will make use of an _ oriented tree _ , a term which we will now define .
a set of edges @xmath243 is an oriented tree of @xmath7 with root @xmath244 if @xmath245 and for each @xmath246 there exists a path from @xmath247 to @xmath244 consisting entirely of edges in @xmath248 ( see figure [ fig : orientedtree ] ) . note that if we reverse the edge directions of an oriented tree , we get a directed tree as defined in ( * ? ? ?
* theorem 2.5 ) . since reversing the directions of all edges in an irreducible graph results in an irreducible graph
, we have by ( * ? ? ?
* lemma 3.3 ) that an oriented tree @xmath248 indeed exists in @xmath7 , and can be efficiently found .
so , let us fix some oriented tree @xmath248 with root @xmath244 . by (
* theorem 2.5 ) , we have that every vertex @xmath246 which is not the root @xmath244 has an out - degree equal to 1 .
thus , for each such vertex @xmath247 we may define @xmath249 as the destination of the single edge in @xmath248 going out of @xmath247 .
.,title="fig : " ] we now elaborate on the encoding process .
the encoding consists of two steps . in the first step , we map the information bits to a collection of lists . in the second step ,
we use the lists in order to define a cyclic path .
first step : given @xmath237 information bits , we build for each vertex @xmath91 a list @xmath250 of length @xmath93 , @xmath251 the entries of each @xmath250 are vertices in @xmath2 .
moreover , the following properties are satisfied for all @xmath34 : * the number of times @xmath35 is an entry in @xmath250 is exactly @xmath238 . * if @xmath252 , then the last entry of the list equals the parent of @xmath34 .
namely , @xmath253 recalling ( [ eq : r_d ] ) , a simple calculation shows that the number of possible list collections is @xmath254 thus , we define the rate of encoding as @xmath255 also , note that as in the 2-d case , we may use enumerative coding in order to efficiently map information bits to lists .
second step : we now use the lists @xmath250 , @xmath91 , in order to construct a cyclic path starting at vertex @xmath244 .
we start the path at @xmath244 and build a length-@xmath63 path according to the following rule : when exiting vertex @xmath34 for the @xmath18th time , traverse the edge going into vertex @xmath256 . of course , our encoding method is valid ( and invertible ) iff we may always abide by the above - mentioned rule .
namely , we do nt get `` stuck '' , and manage to complete a cyclic path of length @xmath63 .
this is indeed the case : define an auxiliary graph @xmath257 with the same vertex set , @xmath2 , as @xmath7 and @xmath238 parallel edges from @xmath34 to @xmath35 ( for all @xmath196 ) .
first , recall that for sufficiently large @xmath11 , the presence of an edge from @xmath34 to @xmath35 in @xmath7 implies that @xmath258 .
thus , since @xmath7 was assumed to be irreducible , @xmath257 is irreducible as well .
also , an edge in @xmath248 from @xmath34 to @xmath35 implies the existence of an edge in @xmath257 from @xmath34 to @xmath35 . secondly , note that by ( [ eq : sum_up ] ) , the number of times we are supposed to exit a vertex is equal to the number of times we are supposed to enter it .
the rest of the proof follows from ( * ? ? ?
* , claim 2 ) , applied to the auxiliary graph @xmath257 .
namely , our encoder follows directly from van aardenne - ehrenfest and de bruijn s @xcite theorem on counting eulerian cycles in a graph .
we now return to the rate , @xmath240 , of our encoder . from ( [ eq : malpha ] ) , ( [ eq : preflow_entries ] ) , ( [ eq : preflow_sumi ] ) and theorem [ theo : dfromptilde ] , we see that for @xmath11 sufficiently large , @xmath259 is greater than some positive constant times @xmath260 .
thus , ( [ eq : rbound ] ) still holds if we replace @xmath12 by @xmath240 and @xmath241 by the entropy of @xmath130 .
recall from section [ sec : twodimensionalconstraints ] the square constraint : its elements are all the binary arrays in which no two ` 1 ' symbols are adjacent on a row , column , or diagonal . by employing the methods presented in @xcite , we may calculate an upper bound on the rate of the constraint .
this turns out to be @xmath261 .
we will show an encoding / decoding method with rate slightly larger than @xmath262 ( about @xmath263 of the upper bound ) . in order to do this
, we assume that the array has @xmath264 columns .
our encoding method has a fixed rate and has a vertical window of size 2 and vertical anticipation 0 .
we should point out now that a straightforward implementation of the methods we have previously defined gives a rate which is strictly @xmath265 than @xmath262 .
namely , this section also outlines two improvement techniques which help boost the rate .
we start out as in the example given in section [ sec : twodimensionalconstraints ] , except that the width of the data strips is now @xmath266 ( the width of the merging strips remains @xmath50 ) .
the graph @xmath7 we choose produces all width-@xmath47 arrays satisfying the square constraint , and we take the merging strips to be all - zero .
our array has @xmath264 columns , so we have @xmath267 tracks ( the last , say , column of the array will essentially be unused ; we can set all of its values to 0 ) .
define the normalized capacity as @xmath268 the graph @xmath7 has @xmath269 vertices and normalized capacity @xmath270 this number is about @xmath271 from the upper bound on the capacity of our 2-d constraint .
thus , as expected , there is an inherent loss in choosing to model the 2-d constraint as an essentially 1-d constraint .
of course , this loss can be made smaller by increasing @xmath47 ( but the graph @xmath7 will grow as well ) . from theorem [ theo : main ]
, the rate of our encoder will approach the normalized capacity of @xmath272 as the number of tracks @xmath11 grows .
so , once the graph @xmath7 is chosen , the parameter we should be comparing ourselves to is the normalized capacity .
we now apply the methods defined in section [ sec : computingmultiplicitymatrix ] and find a multiplicity matrix @xmath86 .
recall that the matrix @xmath86 defines an encoder . in our case
, this encoder has a rate of about @xmath273 .
this is @xmath274 of the normalized capacity , and is quite disappointing ( but the improvements shown in sections [ ssec : moore ] and [ ssec : breakmerge ] below are going to improve this rate ) . on the other hand ,
note that if we had limited ourselves to encode to each track independently of the others , then the best rate we could have hoped for with 0 vertical anticipation turns out to be @xmath275 ( see ( * ? ? ?
* theorem 5 ) ) .
we now define a graph @xmath276 which we call the reduction of @xmath7 .
essentially , we will encode by constructing paths in @xmath276 , and then translate these to paths in @xmath7 . in both @xmath7 and @xmath276 ,
the maxentropic distributions have the same entropy .
the main virtue of @xmath276 is that it often has less vertices and edges compared to @xmath7 .
thus , the penalty in ( [ eq : rbound ] ) resulting from using a finite number of tracks will often be smaller . for @xmath277 , we now recursively define the concept of @xmath202-equivalence ( very much like in the moore algorithm @xcite ) . * for @xmath278 , any two vertices @xmath279 are 0-equivalent . * for @xmath280 , two vertices @xmath279 are @xmath202-equivalent iff 1 ) the two vertices @xmath281 are @xmath282-equivalent , and 2 ) for each @xmath282-equivalence class @xmath283 , the number of edges from @xmath284 to vertices in @xmath283 is equal to the number of edges from @xmath285 to vertices in @xmath283 .
denote by @xmath286 the partition induced by @xmath202-equivalence .
for the graph @xmath7 given in figure [ fig : example : g ] , @xmath287 note that , by definition , @xmath288 is a refinement of @xmath286 .
thus , let @xmath289 be the smallest @xmath202 for which @xmath286 = @xmath288 .
the set @xmath290 can be efficiently found ( essentially , by the moore algorithm @xcite ) .
define a ( non - labeled ) graph @xmath291 as follows .
the vertex set of @xmath276 is @xmath292 for each @xmath293 , let @xmath294 be a fixed element of @xmath283 ( if @xmath283 contains more than one vertex , then pick one arbitrarily ) .
also , for each @xmath295 , let @xmath296 be the class @xmath293 such that @xmath297 .
let @xmath298 ( @xmath299 ) and @xmath300 ( @xmath301 ) denote the start and end vertex of an edge @xmath239 in @xmath7 ( @xmath302 ) , respectively .
the edge set @xmath303 is defined as @xmath304 where @xmath305 namely , the number of edges from @xmath306 to @xmath307 in @xmath302 is equal to the number of edges in @xmath7 from some fixed @xmath308 to elements of @xmath307 , and , by the definition of @xmath289 , this number does not depend on the choice of @xmath284 .
the graph @xmath302 is termed the _ reduction _ of @xmath7 .
the reduction of @xmath7 from figure [ fig : example : g ] is given in figure [ fig : example : greduced ] .
note that since @xmath7 was assumed to be irreducible , we must have that @xmath302 is irreducible as well . from figure
[ fig : example : g].,title="fig : " ] the entropies of the maxentropic markov chains on @xmath7 and @xmath302 are equal .
let @xmath309 be the adjacency matrix of @xmath302 , and recall that @xmath310 is the adjacency matrix of @xmath7 .
let @xmath311 and @xmath312 be the perron eigenvalue and right perron eigenvector of @xmath313 , respectively @xcite . next ,
define the vector @xmath314 as @xmath315 it is easily verifiable that @xmath316 is a right eigenvector of @xmath317 , with eigenvalue @xmath318 .
now , since @xmath319 is a perron eigenvector of an irreducible matrix , each entry of it is positive .
thus , each entry of @xmath316 is positive as well .
since @xmath317 is irreducible , we must have that @xmath316 is a perron eigenvector of @xmath317 .
so , the perron eigenvalue of @xmath317 is also @xmath318 .
the next lemma essentially states that we can think of paths in @xmath302 as if they were paths in @xmath7 .
[ lemm : pathconversion ] let @xmath320 .
fix some @xmath321 , and @xmath322 .
there exists a one - to - one correspondence between the following sets .
first set : paths of length @xmath20 in @xmath302 with start vertex @xmath323 and end vertex @xmath324 .
second set : paths of length @xmath20 in @xmath7 with start vertex @xmath244 and end vertex in @xmath324 .
moreover , for @xmath325 , the first @xmath15 edges in a path belonging to the second set are a function of only the first @xmath15 edges in the respective path in the first set .
we prove this by induction on @xmath20 .
for @xmath326 , we have @xmath327 to see this , note that we can assume w.l.o.g . that @xmath328 , and then recall ( [ eq : etag ] ) . for @xmath329 , combine the claim for @xmath330 with that for @xmath331 . notice that @xmath332 .
we now show why @xmath302 is useful .
let @xmath333 be the multiplicity matrix found by the methods previously outlined , where we replace @xmath7 by @xmath302 .
let @xmath334 .
we may efficiently encode ( and decode ) information to @xmath335 in a row - by - row manner at rate @xmath336 .
we conceptually break our encoding scheme into two steps . in the first step ,
we `` encode '' ( map ) the information into @xmath337 paths in @xmath302 , each path having length @xmath20 .
we do this as previously outlined ( through typical vertices and edges in @xmath302 ) .
note that this step is done at a rate of @xmath336 .
in the second step , we map each such path in @xmath302 to a corresponding path in @xmath7 . by lemma [ lemm : pathconversion ] , we can indeed do this ( take @xmath323 as the first vertex in the path , @xmath324 as the last vertex , and @xmath328 ) . by lemma [ lemm : pathconversion ]
we see that this two - step encoding scheme can easily be modified into one that is row - by - row . applying the reduction to our running example ( square constraint with @xmath266 and @xmath50 ) ,
reduces the number of vertices from @xmath338 in @xmath7 to @xmath339 in @xmath302 .
the computed @xmath333 increases the rate to about @xmath340 , which is @xmath341 of the normalized capacity .
let @xmath342 be the @xmath337th kronecker power of the moore - style reduction @xmath302 .
recall that the rate of our encoder is @xmath343 where @xmath260 is the number of typical edges in @xmath342 going out of a typical vertex .
the second improvement involves expanding the definition of a typical edge , thus increasing @xmath260 .
this is best explained through an example .
suppose that @xmath302 has figure [ fig : breakmerge ] as a subgraph ; namely , we show all edges going out of vertices @xmath344 and @xmath227 .
also , let the numbers next to the edges be equal to the corresponding entries in @xmath333 .
the main thing to notice at this point is that if the edges to @xmath345 and @xmath346 are deleted ( `` break '' ) , then @xmath344 and @xmath227 have exactly the same number of edges from them to vertex @xmath35 , for all @xmath347 ( after the deletion of edges , vertices @xmath344 and @xmath227 can be `` merged '' ) .
let @xmath92 be a typical vertex .
a short calculation shows that the number of entries in @xmath92 that are equal to @xmath344 ( @xmath227 ) is @xmath348 ( @xmath349 ) .
recall that the standard encoding process consists of choosing a typical edge @xmath113 going out of the typical vertex @xmath92 and into another typical vertex @xmath350 .
we now briefly review this process . consider the @xmath351 entries in @xmath92 that are equal to @xmath344 .
the encoding process with respect to them will be as follows ( see figure [ fig : breakmerge1 ] ) : * out of these @xmath351 entries , choose @xmath352 for which the corresponding entry in @xmath350 will be @xmath353 . since there is exactly one edge from @xmath344 the @xmath353 in @xmath302 , the corresponding entries in @xmath113 must be equal to that edge . * next , from the remaining @xmath354 entries , choose @xmath355 for which the corresponding entries in @xmath350 will be @xmath226 .
there are two parallel edges from @xmath344 to @xmath226 , so choose which one to use in the corresponding entries in @xmath113 .
* we are left with @xmath356 entries , the corresponding entries in @xmath350 will be @xmath107 . also , we have one option as to the corresponding entries in @xmath113 .
a similar process is applied to the entries in @xmath92 that are equal to @xmath227 .
thus , the total number of options with respect to these entries is @xmath357 , @xmath350 , where we got from @xmath92 to @xmath350 by the standard encoding process.,title="fig : " ] next , consider a different encoding process ( see figure [ fig : breakmerge2 ] ) . * out of the @xmath351 entries in @xmath92 that are equal to @xmath344 , choose @xmath352 for which the corresponding entry in @xmath350 will be @xmath353 . as before ,
the corresponding entries in @xmath113 have only one option . * out of the @xmath358 entries in @xmath92 that are equal to @xmath227 ,
choose @xmath359 for the corresponding entry in @xmath350 will be @xmath346 . again , one option for entries in @xmath113 . * now , of the remaining @xmath360 entries in @xmath92 that are equal to @xmath344 or @xmath227 , choose @xmath361 for which the corresponding entry in @xmath350 will be @xmath226 .
we have two options for the entries in @xmath113 .
* we are left with @xmath362 entries in @xmath92 that are equal to @xmath344 or @xmath227 .
these will have @xmath107 as the corresponding entry in @xmath350 , and one option in @xmath113 .
, @xmath350 , where we got from @xmath92 to @xmath350 by the improved encoding process .
the shaded part corresponds to vertices that were merged.,title="fig : " ] thus , the total number of options is now @xmath363 the important thing to notice is that in both cases , we arrive at a typical vertex @xmath350 . to recap , we first `` broke '' the entries in @xmath92 that are equal to @xmath344 into two groups : those which will have @xmath353 as the corresponding entry in @xmath350 and those which will have @xmath226 or @xmath107 as the corresponding entry . similarly , we broke entries in @xmath92 that are equal to @xmath227 into two groups .
next , we noticed that of these four groups , two could be `` merged '' , since they were essentially the same .
namely , removing some edges from the corresponding vertices in @xmath302 resulted in vertices which were mergeable .
of course , these operations can be repeated .
the hidden assumption is that the sequence of breaking and merging is fixed , and known to both the encoder and decoder . the optimal sequence of breaking and merging is not known to us .
we used a heuristic .
namely , choose two vertices such that the sets of edges emanating from both have a large overlap
. then , break and merge accordingly .
this was done until no breaking or merging was possible .
we got a rate of about @xmath262 , which is @xmath364 of the normalized capacity .
recall from section [ sec : encoder ] that in the course of our encoding algorithm , we make use of a procedure which encodes information into fixed - length binary words of constant weight .
a way to do this would be to use enumerative coding @xcite .
immink @xcite showed a method to significantly improve the running time of an instance of enumerative coding , with a typically negligible penalty in terms of rate .
we now briefly show how to tailor immink s method to our needs .
denote by @xmath106 and @xmath107 the length and hamming weight , respectively , of the binary word we encode into .
some of our variables will be _ floating - point _
numbers with a mantissa of @xmath365 bits and an exponent of @xmath345 bits : each floating - point number is of the form @xmath366 where @xmath156 and @xmath157 are integers such that @xmath367 note that @xmath368 bits are needed to store such a number .
also , note that every positive real @xmath369 such that @xmath370 has a floating point approximation @xmath371 with relative precision @xmath372 we assume the presence of two look - up tables .
the first will contain the floating - point approximations of @xmath373 .
the second will contain the floating - point approximations of @xmath374 , where @xmath375 in order to exclude uninteresting cases , assume that @xmath376 and is such that @xmath377 .
also , take @xmath345
large enough so that @xmath378 is less than the maximum number we can represent by floating point . thus , we can assume that @xmath379 and @xmath380 .
notice that in our case , we can bound both @xmath106 and @xmath107 from above by the number of tracks @xmath11 .
thus , we will actually build beforehand two look - up tables of size @xmath381 bits .
let @xmath371 denote the floating - point approximation of @xmath369 , and let @xmath382 and @xmath383 denote floating - point multiplication and division , respectively .
for @xmath384 we define @xmath385 note that since we have stored the relevant numbers in our look - up table , the time needed to calculate the above function is only @xmath386 .
the encoding procedure is given in figure [ fig : enumerativecoder ] .
we note the following points : * the variables @xmath106 , @xmath387 , @xmath107 and @xmath388 are integers ( as opposed to floating - point numbers ) . * in the subtraction of @xmath389 from @xmath387 in line 5 , the floating - point number @xmath389 is `` promoted '' to an integer ( the result is an integer ) .
* name : * @xmath390 + * input : * integers @xmath391 such that @xmath105 and @xmath392 .
* output : * a binary word of length @xmath106 and weight @xmath107 .
if ( @xmath107 = = @xmath393 ) // stopping condition : / * 1 * / + return @xmath394 ; / * 2 * / + for ( @xmath395 ; @xmath396 ; @xmath388++ ) \ { / * 3 * / + if ( @xmath397 ) / * 4 * / + @xmath398 ; / * 5 * / + else / * 6 * / + return @xmath399 ; / * 7 * / + } / * 8 * / + we must now show that the procedure is valid , namely , that given a valid input , we produce a valid output . for our procedure , this reduce to showing two things : 1 ) if the stopping condition is not met , a recursive call will be made .
2 ) the recursive call is given valid parameters as well .
namely , in the recursive call , @xmath387 is non - negative .
also , for the encoding to be invertible , we must further require that 3 ) @xmath400 for @xmath401 .
condition 2 is clearly met , because of the check in line 4 .
denote @xmath402 ( and so , @xmath403 ) .
condition 3 follows from the next lemma .
[ lemm : boundroundoff ] fix @xmath105 .
then , @xmath404 the proof is essentially repeated invocations of ( [ eq : fp ] ) on the various stages of computation .
we leave the details to the reader .
finally , condition 1 follows easily from the next lemma .
fix @xmath105 .
then , @xmath405 the claim will follow if we show that @xmath406 this is immediate from lemma [ lemm : boundroundoff ] and the binomial identity @xmath407 note that the penalty in terms of rate one suffers because of using our procedure ( instead of plain enumerative coding ) is negligible .
namely , @xmath408 can be made arbitrarily close to @xmath409 .
since we take @xmath380 and @xmath379 , we can show by amortized analysis that the running time of the procedure is @xmath410 .
specifically , see ( * ? ? ?
* section 17.3 ) , and take the potential of the binary vector corresponding to @xmath387 as the number of entries in it that are equal to ` @xmath393 ' .
the decoding procedure is a straightforward `` reversal '' of the encoding procedure , and its running time is also @xmath410 .
let @xmath411 be as in , where we replace @xmath96 by @xmath172 and @xmath412 by @xmath188 . by the combinatorial interpretation of , and the fact that @xmath209 for all @xmath95
, it easily follows that @xmath413 .
thus , @xmath414 denote by @xmath415 the base of natural logarithms . by stirling s formula
we have @xmath416 and from we get that @xmath417 by ( [ eq : preflow_sumij ] ) and , @xmath418 since @xmath419 , we have @xmath420 moreover , by and , the rhs of the last equation equals @xmath421 we conclude that @xmath422 lastly , recall that @xmath423 and @xmath424 .
thus , @xmath425 where @xmath426 is the entropy of the stationary markov chain @xmath130 with transition matrix @xmath135 . recall that @xmath130 was selected to be maxentropic : @xmath427 .
this fact , along with and a short calculation , finishes the proof .
the first author would like to thank roee engelberg for very helpful discussions .
b. h. marcus , r. m. roth , and p. h. siegel , `` constrained systems and coding for recording channels , '' in _ handbook of coding theory _ , v. pless and w. huffman , eds.1em plus 0.5em minus 0.4emamsterdam : elsevier , 1998 , pp . | a constant - rate encoder decoder pair is presented for a fairly large family of two - dimensional ( 2-d ) constraints . encoding and decoding is done in a row - by - row manner , and is sliding - block decodable .
essentially , the 2-d constraint is turned into a set of independent and relatively simple one - dimensional ( 1-d ) constraints ; this is done by dividing the array into fixed - width vertical strips . each row in the strip
is seen as a symbol , and a graph presentation of the respective 1-d constraint is constructed .
the maxentropic stationary markov chain on this graph is next considered : a perturbed version of the corresponding probability distribution on the edges of the graph is used in order to build an encoder which operates _ in parallel _ on the strips .
this perturbation is found by means of a network flow , with upper and lower bounds on the flow through the edges .
a key part of the encoder is an enumerative coder for constant - weight binary words .
a fast realization of this coder is shown , using floating - point arithmetic .
@xmath0 the work of tuvi etzion was supported in part by the united states
israel binational science foundation ( bsf ) , jerusalem , israel , under grant no .
2006097 .
the work of ron m. roth was supported in part by the united states
israel binational science foundation ( bsf ) , jerusalem , israel , under grant no .
2002197 . |
supersymmetric ( susy ) models of particle physics are very attractive in that they stabilize the gauge hierarchy problem , and provide an avenue for the incorporation of gravity via local susy , or supergravity@xcite .
they also receive some indirect experimental support via the unification of gauge couplings under minimal supersymmetric standard model ( mssm ) renormalization group evolution ( rge)@xcite , and they provide several different candidates ( neutralinos , gravitinos , axions / axinos , @xmath4 ) which can serve as cold dark matter ( cdm ) in the universe .
if evidence for susy is found at lhc , then a paramount question will be : what is the mechanism of susy breaking , and how is it communicated to the visible sector ? some of the possibilities proposed in the literature include : gravity - mediation ( sugra ) with a gravitino mass @xmath5 tev@xcite , gauge - mediation ( gmsb ) with @xmath6 tev@xcite , and anomaly mediation ( amsb ) with @xmath7 tev@xcite .
anomaly - mediated supersymmetry breaking ( amsb ) models have received much attention in the literature due to their attractive properties@xcite : 1 . the soft supersymmetry ( susy ) breaking terms
are completely calculable in terms of just one free parameter ( the gravitino mass , @xmath8 ) , 2 .
the soft terms are real and flavor invariant , thus solving the susy flavor and @xmath9 problems and 3 .
the soft terms are actually renormalization group invariant@xcite , and can be calculated at any convenient scale choice . in order to realize the amsb set - up , it was proposed that the hidden sector be `` sequestered '' on a separate brane from the observable sector in an extra - dimensional universe , so that tree - level supergravity breaking terms do not dominate the soft term contributions .
such a set - up can be realized in brane - worlds , where susy breaking takes place on one brane , with the visible sector residing on a separate brane .
a further attractive feature of amsb models arises due to the scale of their gravitino mass .
sugra - type models with @xmath5 tev suffer from the cosmological gravitino problem .
there are two parts to the gravitino problem@xcite .
1 . if the re - heat temperature after inflation @xmath10 gev , then the high rate of thermal gravitino production leads to an overabundance of neutralino dark matter@xcite .
2 . even for lower values of @xmath11 gev , thermal production of @xmath12 followed by late decays to @xmath13 pairs injects high energy particles into the cosmic soup during or after bbn , thus disrupting one of the pillars of big - bang theory@xcite .
if @xmath14 5 tev , then the lifetime @xmath15 drops below @xmath16 sec , and gravitino decay occurs before or at the onset of bbn . in amsb models where @xmath7 tev
, the gravitino is much too short - lived to be afflicted by the bbn bounds . in spite of their attractive features ,
amsb models suffer from the well - known problem that slepton mass - squared parameters are found to be negative , giving rise to tachyonic states .
the original `` solution '' to this problem was to posit that scalars acquire as well a universal mass @xmath17 , which when added to the amsb ssb terms , renders them positive@xcite .
the derived form of soft susy breaking terms , supplemented by a universal scalar mass @xmath17 and implemented at the gut scale , constitutes what is usually called the minimal amsb , or mamsb model . in mamsb and the additional models described below
, it is assumed that electroweak symmetry is broken radiatively due to the large top quark mass , so that the magnitude of the @xmath18 parameter is determined to gain the correct value of @xmath19 , and the bilinear soft term @xmath20 is traded for the ratio of higgs field vevs , @xmath21 .
an alternative set - up for amsb has been advocated in ref .
@xcite , known as hypercharged anomaly - mediation ( hcamsb ) .
it is a string - motivated scenario which uses a similar setup as the one envisioned for amsb . in hcamsb ,
the mssm resides on a d - brane , and the hypercharge gaugino mass is generated in a geometrically separated hidden sector .
an additional contribution to the @xmath22 gaugino mass @xmath23 is generated , and its magnitude is parametrized by an additional parameter @xmath24 . the large value of @xmath23 feeds into slepton mass evolution through the mssm rge , and acts to lift the weak - scale slepton soft masses beyond tachyonic values .
thus , the hcamsb model naturally solves the tachyonic slepton mass problem which is endemic to pure amsb scenarios .
a third scenario has recently been proposed in ref .
@xcite , under the name gaugino amsb , or inoamsb .
the inoamsb model is suggested by recent work on the phenomenology of flux compactified type iib string theory@xcite , which reduces to @xmath25 supergravity below the compactification scale .
the essential features of this scenario are that the gaugino masses are of the anomaly - mediated susy breaking ( amsb ) form , while scalar and trilinear soft susy breaking terms are highly suppressed : they are taken as @xmath26 at energy scale @xmath27 , at first approximation .
the normally large value of @xmath23 as generated in amsb models feeds into the scalar soft term evolution , lifting slepton soft masses to generate an allowable sparticle mass spectrum , while at the same time avoiding tachyonic sleptons or charged lsps ( lightest susy particles ) .
charged lsps are common in models with negligible soft scalar masses , such as no - scale@xcite or gaugino mediation models@xcite .
since scalar and trilinear soft terms are highly suppressed , the susy induced flavor and @xmath9-violating processes are also suppressed in inoamsb .
all three of these models mamsb , hcamsb and inoamsb share the common feature that the lightest mssm particle is a neutral wino , while the lightest chargino is wino - like with a mass @xmath28 .
the @xmath29-@xmath30 mass gap is of order @xmath31 mev@xcite , so that dominantly @xmath32 , with the decay - produced pion(s ) being very soft .
the small mass gap makes the @xmath29 rather long lived ( @xmath33 sec ) , and it may yield observable highly ionizing tracks ( hits ) of order @xmath34 in length at lhc detectors@xcite .
an important consequence of wino - like neutralinos is that the thermal abundance of neutralino cold dark matter falls generally an order of magnitude or so below the measured abundance : _
cdmh^2=0.11230.0035 68% cl according to the wmap7 data analysis@xcite .
this latter fact has led many to consider amsb - like models as perhaps less interesting than sugra - type models , wherein the bino - like or mixed bino - higgsino neutralino can more easily yield the measured relic abundance . in this paper ,
we address the question of the dark matter abundance in amsb models . while the calculated thermal abundance of wino - like neutralinos is found to be below measured values ( for @xmath35 gev ) , we find that there exists a variety of attractive methods to augment the wino abundance , thus bringing the calculated abundance into accord with experiment . these include : 1 .
decay of scalar ( _ e.g. _ moduli ) fields into sparticles , ultimately terminating in @xmath30 production@xcite . in this case
, the lsp is expected to be a relic wino - like neutralino , which would constitute _ all _ of the cdm .
thermal production@xcite of gravitinos @xmath12 and also possibly gravitino production via moduli@xcite or inflaton@xcite decay , followed by @xmath36 ( where @xmath37 assorted sm debris ) . here
also , the lsp would be a relic wino - like neutralino , which would constitute all of the cdm .
3 . thermal production of heavy axinos@xcite followed by @xmath38@xcite . here
, the lsp is again a relic wino - like neutralino , but the cdm would consist of a wino - like wimp plus axion mixture .
a scenario where @xmath39 , so the @xmath40 is instead the stable lsp@xcite . in this case ,
a combination of thermally produced axinos plus vacuum mis - alignment produced axions would constitute the cdm@xcite . in sec .
[ sec : models ] , we present some details of the three amsb models which we investigate . in sec .
[ sec : dm ] , we present four methods of reconciling the amsb cdm relic abundance with the measured value . given cases 1 and 2 , and possibly 3 , where we expect all ( or some fraction ) of the measured abundance to consist of relic winos , in sec .
[ sec : winodet ] we present rates for direct and indirect detection of wino - like neutralinos . unlike sugra models , the wino cdm direct detection rate is bounded from below .
we find the current experiments like xenon-100 should be able to explore the parameter space of amsb - like models with a wino - like neutralino up to @xmath41 gev .
next generation detectors such as ton - scale noble liquids or supercdms should be able to push to @xmath42 gev .
this would correspond to a reach in @xmath43 gev , _
i.e. _ well beyond the projected reach of lhc .
we also find excellent prospects for indirect detection of wino - like cdm via detection of wino annihilation into @xmath44s , @xmath45s , @xmath46s or @xmath47s in the galactic halo .
in fact , kane _ et al .
_ have already proposed wino cdm as an explanation for the recent anomalies seen by pamela , atic , fermi and others@xcite .
neutrino telescopes such as icecube will also have a reach for wino - like neutralinos , especially for large @xmath21 . in case 4
, we would expect the axino @xmath40 to be the lsp , and so here no direct or indirect wimp detection signals are expected .
however , it may be the case that large amounts of _ axions _
@xmath48 are produced in the early universe , in which case direct detection of axions may be possible at admx@xcite .
we present parameter expectations in sec .
[ ssec : ata ] for the scenario of mixed axion / axino cdm to occur . in sec .
[ sec : conclude ] , we present a summary and our conclusions .
the amsb contribution to the gaugino mass is given by , m_i=_g_i
g_im_= m_3/2 , [ eq : amsb1 ] where @xmath49 is the corresponding beta function , defined by @xmath50 .
the constants @xmath51 for @xmath52 .
the amsb contribution to soft susy breaking scalar masses is given by , m_^2=-14\{d_g+ d_f}m_^2 [ eq : amsb2 ] where @xmath53 is the @xmath54-function for the corresponding superpotential yukawa coupling , and @xmath55 , with @xmath56 the wave function renormalization constant .
complete expressions for the mssm can be found _
e.g. _ in ref s @xcite .
since these give rise to tachyonic slepton masses , each term is supplemented by @xmath57 , where @xmath17 is some additional universal contribution to scalar masses .
finally , the anomaly - mediated contribution to the trilinear susy breaking scalar coupling is given by , a_f=_f f m _ , [ eq : amsb3 ] where @xmath58 labels the appropriate yukawa coupling ( _ e.g. _ @xmath59 is the top - quark yukawa coupling ) .
thus , the parameter space of the `` minimal '' amsb model ( mamsb ) is given by@xcite m_0 , m_3/2 , , sign ( ) ( mamsb ) .
in the mamsb model , we assume as usual that electroweak symmetry is broken radiatively by the large top - quark yukawa coupling .
then the ssb @xmath20 term and the superpotential @xmath18 term are given as usual by the scalar potential minimization conditions which emerge from requiring an appropriate breakdown of electroweak symmetry .
the above expressions for the soft susy breaking terms are usually imposed as gut - scale boundary conditions , and weak scale values are calculated via renormalization group evolution . in hcamsb ,
susy breaking is localized at the bottom of a strongly warped hidden region , geometrically separated from the visible region where the mssm resides .
the warping suppresses contributions due to tree - level gravity mediation@xcite and the anomaly mediation can become the dominant source of susy breaking in the visible sector .
possible exceptions to this sequestering mechanism are gaugino masses of @xmath60 gauge symmetries @xcite . thus , in the mssm , the mass of the bino the gaugino of @xmath22 can be the only soft susy breaking parameter not determined by anomaly mediation@xcite . depending on its size
, the bino mass @xmath23 can lead to a small perturbation to the spectrum of anomaly mediation , or it can be the largest soft susy breaking parameter in the visible sector : as a result of rg evolution , its effect on other soft susy breaking parameters can dominate the contribution from anomaly mediation .
we parametrize the hcamsb ssb contribution @xmath61 using a dimensionless quantity @xmath24 such that @xmath62 ; then , @xmath24 governs the size of the hypercharge contribution to soft terms relative to the amsb contribution@xcite . the soft susy breaking terms are then exactly the same as in mamsb , except there is no @xmath63 contribution to scalar masses , and the @xmath22 gaugino mass is given by m_1 = ( + ) m_3/2 , so that @xmath64 takes us back to pure amsb soft terms , with their concommitant tachyonic sleptons .
then the parameter space of hcamsb models is given by , m_3/2 , , sign ( ) ( hcamsb ) , where the dimensionless @xmath24 typically ranges between @xmath65 for allowable spectra@xcite .
while the lightest neutralino is mainly wino - like in mamsb and hcamsb models , it is important phenomenologically to realize that for certain regions of parameter space , the @xmath30 picks up a substantial higgsino component , thus becoming a mixed wino / higgsino particle .
this occurs in mamsb at large @xmath17 values , and in hcamsb at large @xmath24 values . the situation is shown in fig .
[ fig : z1frac ] for @xmath66 tev , @xmath67 and @xmath68 , where we show @xmath69 ( the wino component of @xmath30 in the notation of ref .
@xcite ) and @xmath70 ( the higgsino component of @xmath30 ) as a function of _ a _ ) . @xmath17 in mamsb and _ b _ ) .
@xmath24 for hcamsb .
we see indeed that in both these cases , @xmath71 is increasing as the relevant parameter increases .
phenomenologically viable versions of string theory require the stabilization of all moduli fields as well as weak to intermediate scale supersymmetry breaking .
models satisfying these criteria were first developed in the context of type iib string theory using flux compactifications and non - perturbative effects on calabi - yau orientifolds ( cyos)@xcite . the low energy limit of type - iib string theory after compactification on a cyo is expected to be @xmath25 supergravity ( sugra ) .
two classes of the above models which yield an interesting supersymmetry breaking scenario have been studied : * those with only a single khler modulus ( skm models ) .
these are essentially of the kklt type @xcite but with uplift coming from one - loop quantum effects . *
large volume scenario ( lvs)@xcite models which require at least two moduli . in both of these types of models ,
the moduli fields are stabilized using a combination of fluxes and non - perturbative effects .
additionally , supersymmetry is broken by the moduli fields acquiring non - zero f - terms and interacting gravitationally with the mssm .
for both models , the gauginos acquire mass predominately through the weyl anomaly while the classical contribution to the scalar masses and trilinear coupling constants are naturally suppressed . here
, we take the limit where scalar and trilinear soft breaking parameters are exactly zero at the gut scale : @xmath72 , while gaugino masses are of the amsb form .
the parameter space of inoamsb models is then given by m_3/2 , , sign ( ) ( inoamsb ) .
as shown in ref .
@xcite , the inoamsb model solves the problem of tachyonic scalars in amsb , since now the gut scale scalar masses vanish .
it also solves the problem of charged lsps which is endemic to no - scale sugra or gaugino - mediated susy breaking ( inomsb ) models , which have @xmath72 but with universal gaugino masses equal to @xmath73 . for inoamsb models with scalar masses @xmath74 at the gut scale the large gut - scale @xmath22 gaugino mass @xmath23 pulls all scalar masses to large values , leaving no tachyons and a wino - like neutralino as the lightest mssm particle .
the above mamsb and hcamsb models have been included into the isasugra subprogram of the event generator isajet@xcite .
in addition , sparticle mass spectra for the inoamsb model can easily be generated using usual msugra input parameters with @xmath72 , but with non - universal gaugino masses as specified by amsb models .
after input of mamsb , hcamsb or inoamsb parameters , isasugra then implements an iterative procedure of solving the mssm rges for the 26 coupled renormalization group equations , taking the weak scale measured gauge couplings and third generation yukawa couplings as inputs , as well as the above - listed gut scale ssb terms .
isasugra implements full 2-loop rg running in the @xmath75 scheme , and minimizes the rg - improved 1-loop effective potential at an optimized scale choice @xmath76@xcite to determine the magnitude of @xmath18 and @xmath77 .
all physical sparticle masses are computed with complete 1-loop corrections , and 1-loop weak scale threshold corrections are implemented for the @xmath78 , @xmath79 and @xmath80 yukawa couplings@xcite .
the off - set of the weak scale boundary conditions due to threshold corrections ( which depend on the entire superparticle mass spectrum ) , necessitates an iterative up - down rg running solution .
the resulting superparticle mass spectrum is typically in close accord with other sparticle spectrum generators@xcite .
once the weak scale sparticle mass spectrum is known , then sparticle annihilation cross sections may be computed . to evaluate the thermally produced neutralino relic density
, we adopt the isared program@xcite , which is based on calchep@xcite to compute the several thousands of neutralino annihilation and co - annihilation feynman diagrams .
relativistic thermal averaging of the cross section times velocity is performed@xcite . as an example , in fig .
[ fig : wino_rd ] , we show the thermally produced neutralino relic density @xmath81 versus @xmath8 for all three models : mamsb , hcamsb and inoamsb .
we take @xmath67 and @xmath68 . for mamsb
, we also take @xmath82 , and for hcamsb , we take @xmath83 . on the upper axis , we also indicate the corresponding values of @xmath84 .
we see that the relic abundance is typically well below wmap7-measured levels , until @xmath85 tev , corresponding to @xmath86 tev , and @xmath87 tev : well beyond any conceivable lhc reach .
the well - known tiny relic abundance arises due to the large @xmath88 annihilation and also @xmath89 and @xmath90 co - annihilation processes .
shortly after the introduction of amsb models , moroi and randall proposed a solution to the amsb dark matter problem based on augmented neutralino production via the decays of moduli fields in the early universe@xcite .
the idea here is that string theory is replete with additional moduli fields : neutral scalar fields with gravitational couplings to matter . in generic supergravity theories ,
the moduli fields are expected to have masses comparable to @xmath8 .
when the hubble expansion rate becomes comparable to the moduli mass @xmath91 , then an effective potential will turn on , and the moduli field(s ) will oscillate about their minima , producing massive excitations , which will then decay to all allowed modes : _
e.g. _ gauge boson pairs , higgs boson pairs , gravitino pairs , @xmath4 .
the neutralino production rate via moduli decay has been estimated in ref .
it is noted in ref .
@xcite that the abundance given by _
_ 1h^2~0.1 ( ) ( ) ^1/4 ( ) ( ) ^3/2 with @xmath92 @xmath93/sec yields nearly the measured dark matter abundance for wino - like neutralino annihilation cross sections and @xmath94 tev .
, where @xmath95 is the ( large ) volume of the compactified manifold : @xmath96 in planck units . in this case , the mechanism would not so easily apply . ]
these authors dub this the `` non - thermal wimp miracle '' .
a necessary condition for augmented neutralino production via scalar field decay is that the re - heat temperature of radiation @xmath97 induced by moduli decays is bounded by @xmath98 mev ( in order to sustain big bang nucleosynthesis ( bbn ) as we know it ) , and @xmath99 , where @xmath100 is the freeze - out temperature for thermal neutralino production @xmath101 .
if @xmath97 exceeds @xmath100 , then the decay - produced neutralinos will thermalize , and the abundance will be given by the thermal calculation as usual .
this `` low re - heat '' neutralino production mechanism has been investigated extensively by gondolo and gelmini@xcite .
the low re - heat neutralino abundance calculation depends on the input value of @xmath97 and the ratio @xmath102 , where @xmath79 is the average number of neutralinos produced in moduli decay , and @xmath91 is the scalar field mass .
they note that theories with an underabundance of thermally produced neutralino cdm with @xmath103 can always be brought into accord with the measured dm abundance for at least one and sometimes two values of @xmath97 . field decay . since this case does nt attain in amsb models ( unless @xmath104 gev ) , we will neglect it here . ]
while the low @xmath105 mev scenario with dm generation via scalar field decay is compelling , we note here that it is also consistent with some baryogenesis mechisms : _
e.g. _ affleck - dine baryogenesis wherein a large baryon asymmetry is generated early on , only to be diluted to observable levels via moduli decay@xcite , or a scenario wherein the baryon asymmetry is actually generated by the moduli decay@xcite .
an alternative possibility for augmenting the production of wino - like neutralinos in amsb models is via gravitino production and decay in the early universe .
while gravitinos would not be in thermal equilibrium during or after re - heat , they still can be produced thermally via radiation off ordinary sparticle scattering reactions in the early universe .
the relic density of thermally produced gravitinos as calculated in ref s
@xcite is given by _ ^tph^2=_i=1
^ 3_ig_i^2(1 + ) ( ) ( ) ( ) , where @xmath106 and @xmath107 are the gauge couplings and gaugino masses evaluated at scale @xmath108 , and @xmath109 and @xmath110 .
each gravitino ultimately cascade decays down to the wino - like @xmath30 state , so the neutralino relic density is given by _ _
1h^2 = _ _ 1^tph^2 + _ ^tph^2 .
[ eq : gravo ] a plot of the value of @xmath97 and @xmath8 which is required to yield @xmath111 from eqn [ eq : gravo ] is shown in fig . [
fig : tr_m32 ] for mamsb ( @xmath112 ) , hcamsb ( @xmath113 ) and inoamsb using @xmath67 and @xmath68 .
the region above the @xmath111 curves would yield too much dark matter , while the region below the curves yields too little .
we should consider the curves shown in fig .
[ fig : tr_m32 ] as only indicative of the simplest scenario for wino production via gravitino decay .
three other effects can substantially change the above picture from what is presented in eq .
[ eq : gravo ] . * on the one hand , if moduli fields @xmath114 exist with mass @xmath115 , then gravitinos can also be produced via moduli production and decay@xcite .
the exact abundance of these moduli - produced gravitinos is very model dependent , and depends on the moduli and gravitino mass and branching fractions . *
a second case arises if we consider gravitino production via inflaton decay at the end of inflation@xcite .
this production mechanism depends on unknown properties of the inflaton : _
e.g. _ its mass and branching fractions , and the re - heat temperature generated by inflaton decay .
these latter quantities are very model dependent . *
additional entropy production generated via the inflaton , moduli and gravitino decays may also dilute the above relic abundance in eq .
[ eq : gravo ] .
we will bear in mind that these possibilities permit much lower or much higher values of @xmath97 and @xmath8 than those shown by the @xmath116 contour of fig .
[ fig : tr_m32 ] .
a third mechanism for increasing the wino - like relic abundance is presented in ref .
@xcite , in the context of the pqmssm .
if we adopt the peccei - quinn ( pq ) solution to the strong @xmath9 problem within the context of supersymmetric models , then it is appropriate to work with the pq - augmented mssm , which contains in addition to the usual mssm states , the axion @xmath48 , the @xmath117-parity even saxion field @xmath118 , and the spin-@xmath119 @xmath117-parity odd axino @xmath40 .
the axino can serve as the lightest susy particle if it is lighter than the lightest @xmath117-odd mssm particle .
the @xmath48 and @xmath40 have couplings to matter which are suppressed by the value of the pq breaking scale @xmath120 , usually considered to be in the range @xmath121 gev@xcite . in ref .
@xcite , it is assumed that @xmath122 , where @xmath30 is the lsp . in the amsb scenarios considered here
, we will assume @xmath123 gev , so as to avoid overproduction of dark matter via gravitinos . with these low values of @xmath97 ,
we are also below the axino decoupling temperature @xmath124 , so the axinos are never considered at thermal equilibrium@xcite .
however , axinos can still be produced thermally via radiation off usual mssm scattering processes at high temperatures .
the calculation of the thermally produced axino abundance , from the hard thermal loop approximation , yields@xcite _
^tp = h^25.5 g_s^6 ( ) ( ) ^2 ( ) ( ) [ eq : oh2_tp ] where @xmath125 is the strong coupling evaluated at @xmath108 and @xmath126 is the model dependent color anomaly of the pq symmetry , of order 1 .
since these axinos are assumed quite heavy , they will decay to @xmath127 or @xmath128 modes , which further decay until the stable lsp state , assumed here to be the neutral wino , is reached .
if the temperature of radiation due to axino decay ( @xmath129 ) exceeds the neutralino freeze - out temperature @xmath100 , then the thermal wino abundance is unaffected by axino decay . if @xmath130 , then the axino decay will _ add _ to the neutralino abundance
. however , this situation breaks up into two possibilities : _ a _ ) .
a case wherein the axinos can dominate the energy density of the universe , wherein extra entropy production from heavy axino decay may dilute the thermal abundance of the wino - like lsps , and _
b _ ) . a case where they do nt . in addition , if the yield of winos from axino decay is high enough , then additional annihilation of winos after axino decay may occur ; this case is handled by explicit solution of the boltzmann equation for the wino number density . along with a component of wino - like neutralino cdm
, there will of course be some component of vacuum mis - alignment produced axion cdm : thus , in this scenario , we expect a wimp / axion mixture of cdm . in this case , we again consider the pqmssm , as in subsec .
[ ssec : haxino ] .
but now , we consider a light axino with @xmath39 , so that @xmath40 is the stable lsp@xcite . here , the thermally produced wino - like neutralinos will decay via @xmath131 , so we will obtain a very slight dark matter abundance from neutralino decay : @xmath132 , since each thermally produced neutralino gives rise to one non - thermally produced ( ntp ) axino .
we will also produce axinos thermally via eqn [ eq : oh2_tp ] .
finally , we will also produce axion cdm via the vacuum mis - alignment mechanism@xcite : @xmath133 ( we will take here the initial mis - alignment angle @xmath134 ) .
the entire cdm abundance is then the sum _
in this case , the tp axinos constitute cdm as long as @xmath135 mev .
the ntp axinos constitute warm dm for @xmath136 gev@xcite , but since their abundance is tiny , this fact is largely irrelevant .
the entire cdm abundance then depends on the parameters @xmath120 , @xmath137 and @xmath97 ; it also depends extremely weakly on @xmath81 , since this is usually small in amsb models . as an example , we plot in fig .
[ fig : oh2_ata ] the three components of mixed axion / axino dm abundance from hcamsb benchmark point 1 in ref .
@xcite : @xmath138 , @xmath66 tev , @xmath67 and @xmath68 .
the neutralino thermal dm abundance would be @xmath139 if the @xmath30 was stable .
we require instead @xmath140 , and plot the three components of @xmath141 versus @xmath142 , for three values of @xmath143 , @xmath144 and @xmath145 gev .
the value of @xmath137 is determined by the constraint @xmath146 .
we see that at low values of @xmath142 , the ntp axino abundance is indeed tiny .
also the axion abundance is tiny since the assumed initial axion field strength is low .
the tp axino abundance dominates . as @xmath142 increases , the axion abundance increases , taking an ever greater share of the measured dm abundance .
the tp axino abundance drops with increasing @xmath142 , since the effective axino coupling constant is decreasing . around @xmath147 gev , the axion abundance becomes dominant .
it is in this range that admx@xcite would stand a good chance of measuring an axion signal using their microwave cavity experiment . in fig .
[ fig : trvsmax ] , we again require @xmath146 for hcamsb benchmark point 1 , but this time plot the value of @xmath97 which is needed versus @xmath137 , for various values of @xmath142 . the plots
terminate at high @xmath97 in order to avoid reaching the axion decoupling temperature @xmath148 .
dashed curves indicate regions where over 50% of the dm is warm , instead of cold .
solid curves yield the bulk of dm as being cold .
we see that for very light axino masses , and large values of @xmath120 , the value of @xmath97 easily reaches beyond @xmath149 gev , while maintaining the bulk of dark matter as cold .
such high values of @xmath97 are good enough to sustain baryogenesis via non - thermal leptogenesis@xcite , although thermal leptogenesis requires @xmath10 gev@xcite .
since @xmath120 is quite large , we would expect that the dominant portion of dm is composed of relic axions , rather than axinos ; as such , detection of the relic axions may be possible at admx@xcite . while fig s [ fig : oh2_ata ] and [ fig : trvsmax ] were created for the hcamsb model , quite similar results
are obtained for the mamsb or inoamsb models .
for amsb dark matter cases 1 and 2 above , it is expected that the thermal wino abundance will be supplemented by either moduli or gravitino decay in the early universe , thus increasing the wino abundance into accord with measured values . in these cases , it may be possible to detect relic wino - like wimps with either direct or indirect detection experiments@xcite .
also , in case 3 above , it is expected that the dm abundance is comprised of an axion / wino mixture .
if the wino component of this mixture is substantial , then again direct or indirect wimp detection may be possible , while if axions are dominant , then a wimp signal is less likely , but direct detection of relic axions is more likely : in a nearly equal mixture of wimps and axions , possibly detection of both could occur ! in case 4 above , we would expect no wimp signals to occur in either direct or indirect detection experiments .
direct detection of wimps depends on the wimp - nucleon scattering cross section , but also on assumptions about the local wimp density ( usually assumed to be @xmath150 ) , and the velocity distribution of the relic wimps ( usually assumed to follow a maxwellian distribution @xmath151 where @xmath152 km / sec , the sun s velocity about the galactic center ) . in our case , where wimps are mainly produced non - thermally via moduli , gravitino or axino decay , the original velocity distribution due to decays will be red - shifted away and the current distribution will arise mainly from gravitational infall , as is the case with thermal wimp production .
the direct detection reach plots are usually presented in terms of the wimp - nucleon scattering cross section .
then , the experimental reach depends on factors like the mass and spin of the nuclear target , and the assumed local wimp density and velocity profiles . direct detection of wimps
is usually broken down into two components : detection via spin - independent ( si ) interactions , and detection via spin - dependent interactions ( sd ) . for si interactions
, it may be best to use heavy target nuclei , since the si nucleon - wimp interactions sum coherently over the nuclear mass . for the si wimp - nucleon cross section ,
we use the isatools subroutine isares@xcite . our results for si direct detection of wino - wimps
is shown in fig .
[ fig : sidd ] , in the @xmath153 plane . here , we scan over @xmath8 for all models , and @xmath17 ( for mamsb ) and @xmath24 ( for hcamsb ) .
we show results for @xmath67 and 40 , while taking @xmath68 .
the inoamsb results occur as lines , since there is no @xmath17 or @xmath24 dependence .
we keep only solutions that obey the lep2 limit on a wino - like chargino : @xmath154 gev@xcite .
several crucial features emerge from the plot .
first , we note that for a given value of @xmath155 , the value of @xmath156 is bounded from below , unlike the case of the msugra model .
that means that wino - wimp dark matter can be either detected or excluded for a given @xmath155 value .
second , we note that the cross section values generally fall in the range that is detectable at present or future dd experiments .
the purple contour , for instance , exhibits the cdms reach based on 2004 - 2009 data , and already excludes some points , especially those at large @xmath21 .
we also show the reach of xenon-100 , lux , xenon-100 upgrade , and xenon 1 ton@xcite .
these experiments should be able to either discover or exclude amsb models with @xmath155 values below @xmath157 and 500 gev respectively .
these wimp masses correspond to values of @xmath158 and 3850 gev , respectively !
the latter reach far exceeds the 100 fb@xmath159 of integrated luminosity reach of lhc for @xmath84 .
reach of lhc for hcamsb is found to be @xmath160 tev . in ref .
@xcite , the 100 fb@xmath159 reach of lhc for inoamsb was found to be @xmath161 tev . ] for inoamsb models , where the minimal value of @xmath162 exceeds that of mamsb or hcamsb for a given @xmath155 value , the xenon 1 ton reach is to @xmath163 gev , corresponding to a reach in @xmath84 of 6200 gev ! in fig .
[ fig : sddd ] , we show the sd direct detection cross section @xmath164 versus @xmath155 for mamsb , hcamsb and inoamsb models with @xmath67 and 40 .
we also show a recent limit on this cross section from the coupp experiment , which is above the theory expectation by two orders of magnitude .
we also show two limits from icecube in 2009 , which do approach the theory region , but only for rather large values of @xmath155 .
the icecube sd reach is quite significant , because the rate for wimp annihilation in the core of the sun mainly depends on the sun s ability to sweep up neutralinos as it passes along its orbit .
the target here is the solar hydrogen , where the sd cross section usually dominates the si one , since the atomic mass is minimal ( an enhancement by number of nucleons per nucleus is usually necessary to make the si cross section competetive with the sd one ) . since icecube is mainly sensitive to very high energy muons with @xmath165 gev , it can access mainly higher values of @xmath155 .
the icecube deepcore reach is also shown .
the deepcore project will allow icecube to access much lower energy muons , and thus make it more useful for generic wimp searches .
while deepcore will access a portion of parameter space , it will not reach the lower limit on sd cross sections as predicted by amsb models .
next , we present rates for indirect detection ( i d ) of wino - like dm via neutrino telescopes , and via detection of gamma rays and anti - matter from wimp annihilation in the galactic halo .
the i d detection rates depend ( quadratically@xcite ) on the assumed galactic dm density ( halo ) profile .
we will show results using two profiles : isothermal and navarro - frenk - white ( nfw)@xcite ( see _ e.g. _
@xcite for plots of several recent halo profiles ) .
most halo models are in near accord at the earth s position at @xmath166 kpc from the galactic center .
however , predictions for the dm density near the galactic center differ wildly , which translates to large uncertainties for dm annihilation rates near the galactic core . the corresponding uncertainty will be smaller for anti - protons , and smaller still for positrons ; since these particles gradually lose energy while propagating through the galaxy , they can reach us only from limited distances over which the halo density is relatively well - known .
possible clumping of dm yields an additional source of uncertainty in i d detection rates . in fig .
[ fig : idd_m0_tb10]_a_. ) , we show for comparison the si direct detection scattering cross section versus @xmath17 in the mamsb model for @xmath66 tev and @xmath67 . for these parameters , the wino - like neutralino has mass @xmath167 gev .
we also indicate an approximate reach of xenon-10 and xenon-100 .
while the si direct detection cross section is just below xenon-100 reach for low @xmath17 , as @xmath17 increases , the value of @xmath18 drops , much as it does in msugra as we approach the focus point region . for large @xmath17 ,
the @xmath30 becomes mixed wino - higgsino , and its direct detection cross section increases into the range which is accessible to xenon-100 .
, for @xmath66 tev , @xmath67 and @xmath68 . , scaledwidth=75.0% ] in fig .
[ fig : idd_m0_tb10]_b_. ) , we show the flux of muons from @xmath168 conversions at earth coming from neutralino annihilation to sm particles within the solar core . here , we use the isajet / darksusy interface for our calculations@xcite , and require @xmath165 gev .
the predicted rate depends , in this case , mainly on the sun s ability to sweep up and capture neutralinos , which depends mainly on the _ spin - dependent _ neutralino - nucleon scattering cross section ( since in this case , the neutralinos mainly scatter from solar hydrogen , and there is no mass number enhancement ) , which is mostly sensitive to @xmath169 exchange .
the rates are again low for low @xmath17 with wino - like neutralinos .
they nearly reach the icecube detectability level at large @xmath17 where the neutralinos , while remaining mainly wino - like , have picked up an increasing higgsino component , so that the neutralino couplings to @xmath56 become large . in fig .
[ fig : idd_m0_tb10]_c_. ) , we show the expected flux of gamma rays with @xmath170 gev , as required for the fermi gamma - ray space telescope ( fgst ) , arising from dm annihilations in the galactic core . in this case
, we see a signal rate which is flat with respect to @xmath17 . here
, the rate depends mainly on the @xmath88 annihilation cross section , which occurs via chargino exchange ; since the @xmath30s remain mainly wino - like , and the chargino mass hardly varies , the annihilation rate hardly varies with @xmath17 .
the predictions for two halo profiles differ by over an order of magnitude , reflecting the large uncertainty in our knowledge of the dm density at the center of our galaxy .
both projections are above the approximate reach of the fgst . in fig .
[ fig : idd_m0_tb10]_d_.)-_f_. ) , we show the expected flux of positrons @xmath45 , antiprotons @xmath46 and antideuterons @xmath171 from neutralino halo annihilations .
each of these frames show detectable rates by pamela@xcite ( for @xmath45s and @xmath46s ) and by gaps@xcite ( for anti - deuterons ) .
these elevated idd rates ( compared to msugra@xcite for similar @xmath21 values ) for anti - matter detection reflect the elevated rate for the @xmath172 annihilation into @xmath173 cross section .
the halo model uncertainty for anti - matter detection is much smaller than in the @xmath44-ray case , since for charged particle detection , it is necessary that the anti - matter is generated relatively close to earth , where the dm density profile is much better known . in fig .
[ fig : idd_m0_tb40 ] , we show rates for direct and indirect detection of wino - like wimps in mamsb versus @xmath17 with @xmath66 tev , @xmath174 and @xmath68 . the si direct detection rate shown in fig .
[ fig : idd_m0_tb40]_a_. ) shows a notable enhancement at low @xmath17 , and the usual enhancement at large @xmath17 due to the increasing higgsino component of @xmath30 .
the low @xmath17 enhancement arises because the mass of the heavy higgs scalar @xmath175 has dropped with increasing @xmath21 , and is now quite light : @xmath176 gev for @xmath177 gev ( compared to @xmath178 gev for the same @xmath17 with @xmath67 . the @xmath179 loop diagram via @xmath175 exchange is enhanced , resulting in a huge direct detection cross section .
this range of @xmath17 is already excluded by dd wimp searches ! , for @xmath66 tev , @xmath174 and @xmath68 .
, scaledwidth=75.0% ] in fig .
[ fig : idd_m0_tb40]_b_. ) , we show the muon flux from mamsb models versus @xmath17 for @xmath66 tev and @xmath174 . in this case
, we again see a huge enhancement at low @xmath17 . while normally the sd @xmath180 cross section dominates the solar accretion rate for wimps , in this case , the 3 order - of - magnitude increase in si cross section shown in frame _
a_. ) contributes and greatly increases the solar capture rate , and hence the muon flux from the sun .
of course , this region would already be excluded by present dd limits .
we also see a curious `` anti - resonance '' effect in @xmath181 around @xmath182 gev . in this case ,
@xmath183 , and neutralino annihilation is enhanced by the @xmath184 resonance . normally , for amsb models , @xmath185 ( @xmath186 or @xmath56 ) is the dominant annihilation mechanism .
but on the higgs resonance , @xmath187 instead dominates .
the energy distribution of neutrinos from @xmath79 decay is far softer than that from @xmath188 or @xmath56 decay , leading to @xmath189 conversions to lower energy muons .
since we require @xmath165 gev for icecube , fewer muons are detected , and hence the anti - resonance effect . at large @xmath17 and @xmath174 ,
the muon flux is again enhanced by the increased wimp scattering rate via its increasing higgsino component . in fig .
[ fig : idd_m0_tb40]_c - f_. ) , we see the flux of gamma rays and anti - matter versus @xmath17 at large @xmath21 . here , the rate versus @xmath17 is again flat , reflecting the usually constant @xmath190 annihilation rate into vector bosons .
the exception occurs at @xmath191 gev , where annihilation through the @xmath184-resonance enhances the halo annihilation rate@xcite . at large @xmath17 and @xmath174 ,
the @xmath45 and @xmath46 detection rates drop .
this is due to the changing final state from @xmath190 annihilation : at low @xmath17 it is mainly to vector bosons , leading to a hard @xmath45 and @xmath46 distribution . at large @xmath17 , annihilations to @xmath192 increase and
become prominent , but the energy distribution of @xmath45 and @xmath46 softens , and since we require @xmath193 gev , the detection rate drops . in frame _
f_. ) , showing the @xmath47 rate , the rate actually increases at large @xmath17 , since here we already require quite low energy @xmath47s for detection , and the distribution only reflects the increased annihilation rate . in this subsection , we present wino - like wimp dd and i d rates in the hcamsb model for @xmath66 tev , versus varying @xmath24 . as shown in ref .
@xcite , a low value of @xmath194 corresponds to pure anomaly - mediation , while large @xmath24 gives an increasing mass @xmath23 to the hypercharge gaugino at the gut scale .
the large value of @xmath23 pulls sparticle masses to larger values via rg evolution , with the pull increasing in accord with the matter state s hypercharge quantum number : thus at large @xmath24 we expect relatively heavy @xmath195 states , but comparatively light @xmath196 and @xmath197 states .
in fact , the @xmath60 rge effect coupled with the large @xmath78-quark yukawa coupling@xcite leads to relatively light , and dominantly left- , top squark states @xmath198 at large @xmath24 .
for very high values of @xmath199 , the value of @xmath200 diminishes until radiative ewsb no longer occurs ( much as in the focus point region of msugra ) . in fig .
[ fig : iddhc_tb10]_a_. ) , we show the si neutralino direct detection rate versus @xmath24 for @xmath67 .
the cross section @xmath162 is of order @xmath201 pb at low @xmath24 , consistent with pure wino - like neutralinos . as @xmath24 increases , the increasing sparticle masses feed into @xmath202 , diminishing the term @xmath203 and leading to a lessened downward push by the top yukawa coupling .
since @xmath204 at the weak scale , the @xmath200 term is also diminished , leading to an increasing higgsino component of @xmath30 .
the increased higgsino component yields an enhanced @xmath162 via higgs exchange at large @xmath24 .
, for @xmath66 tev , @xmath67 and @xmath68 .
, scaledwidth=75.0% ] in fig .
[ fig : iddhc_tb10]_b_. ) , we plot the muon flux in the hcamsb model vs. @xmath24 .
the muon flux is quite small at low @xmath24 , but at high @xmath24 , the increasing higgsino component of @xmath30 leads to an increased @xmath164 via @xmath56-exchange . in fig
s [ fig : iddhc_tb10]_c - f_. ) , we plot the @xmath44-ray , @xmath45 , @xmath46 and @xmath47 fluxes versus @xmath24 . in these cases ,
the rates are large due to the large @xmath205 annihilation cross section and is relatively flat with @xmath24 . at the largest @xmath24 values , annihilation to @xmath192 states
is enhanced , leading to diminished rates for @xmath45 and @xmath46 ( due to softened energy distributions ) but to a slightly increased rate for @xmath47s . in fig .
[ fig : iddhc_tb40 ] , we show the same rates vs. @xmath24 in the hcamsb model , except now for @xmath174 .
the si direct detection rate in fig .
[ fig : iddhc_tb40]_a_. ) is enhanced relative to the @xmath67 case due to the much lighter higgs mass @xmath206 and the increased @xmath79-quark yukawa coupling . the rate diminishes as @xmath24 increases due to increasing squark and higgs masses , until very high @xmath24 is reached , and the rate is enhanced by the growing higgsino component of @xmath30 .
the parameter space terminates above @xmath207 due to lack of rewsb . in fig .
[ fig : iddhc_tb40]_b_. ) , we see that the muon flux due to solar core annihilations is also enhanced . in this case , the increase is again due to the large enhancement in si scattering cross section , which feeds into the solar accretion rate . as in fig .
[ fig : idd_m0_tb40]_b_. ) , we find an anti - resonance dip at the @xmath24 value where @xmath208 , and wimp annihilations occur instead mainly into @xmath192 rather than @xmath209 states .
fig s [ fig : iddhc_tb40]_c - f_. ) show the halo annihilation rates for hcamsb at @xmath174 .
these rates are generally flat with changing @xmath24 , and do not suffer an increase compared with low @xmath21 results , since @xmath205 still dominates the annihilation rate .
the exception occurs at @xmath210 , where @xmath208 , and halo annihilation is enhanced by the pseudoscalar higgs resonance .
, for @xmath66 tev , @xmath174 and @xmath68 .
, scaledwidth=75.0% ] in fig .
[ fig : idd_m32_tb10 ] , we show direct and indirect wino dm detection rates versus @xmath8 for @xmath67 and @xmath68 . for mamsb , we take @xmath211 tev , while for hcamsb , we take @xmath212 .
the associated mass spectra versus @xmath8 can be found in ref .
@xcite for mamsb and hcamsb , and in ref .
@xcite for the inoamsb model .
spectra for all three models are shown in table 1 of ref .
@xcite for the case of @xmath66 tev . in fig .
[ fig : idd_m32_tb10]_a_. ) , the si direct detection cross section is shown for all three models . in this case , we see for a given value of @xmath8 , the inoamsb model gives the highest cross section , while mamsb gives the lowest .
the larger inoamsb cross section is due in part because inoamsb models have a smaller @xmath18 value for a given value of @xmath8 , and so si scattering via higgs exchange ( which involves a product of higgsino and gaugino components ) is enhanced . , for @xmath67 and @xmath68 . for mamsb , we take @xmath211 tev , while for hcamsb , we take @xmath212 . in these plots ,
we adopt the nfw dm halo profile .
, scaledwidth=75.0% ] in fig .
[ fig : idd_m32_tb10]_b_. ) , we show the relative rates for indirect wino detection due to wimp annihilation into @xmath213 states in the solar core , with subsequent muon detection from @xmath214 conversions in antarctic ice , as might be seen by icecube .
we require @xmath165 gev . the muon flux is mainly related to the spin - dependent direct detection rate , which enters the sun s ability to capture wimps . here again , inoamsb yields the highest rates , and mamsb the lowest .
this follows the relative values of @xmath18 in the three models : low @xmath18 in inoamsb leads to a larger higgsino component of @xmath30 , and an increased sd scattering rate via @xmath169 exchange .
a rough reach of the icecube detector is shown , and indicates that the low @xmath8 portion of parameter space of inoamsb and hcamsb may be accessible to @xmath214 searches . in fig .
[ fig : idd_m32_tb10]_c - f_. ) , we show the i d rates for detection of @xmath44 , @xmath45s , @xmath46s and @xmath47s , for the energy ranges indicated on the plots .
all these plots adopt the nfw halo profile . in all these cases ,
all three models yield almost exactly the same detection rates for a given value of @xmath8 .
this is due to the dominance of @xmath185 halo annihilations , which mainly depend on the gaugino component of @xmath30 , which is nearly all wino - like .
the rough reach of fermi - lat , pamela and gaps is shown for reference .
the high rates for wino halo annihilations should yield observable signals .
as mentioned previously , kane _ et al .
_ promote wino - like wimps as a source of the pamela anomaly@xcite . in this case , a large @xmath46 signal should be seen as well , although the pamela @xmath46 rate seems to agree with sm background projections . in fig .
[ fig : idd_m32_tb40 ] , we show direct and indirect wino detection rates versus @xmath8 as in fig . [
fig : idd_m32_tb10 ] , except now for a large value of @xmath174 . in frame _
a_. ) , we see that the si direct detection rates are all elevated with respect to the @xmath67 case .
the well - known large @xmath21 enhancement@xcite occurs due to enhanced higgs exchange contributions , where now the value of @xmath206 is lower and the @xmath79-quark yukawa coupling is larger . for low @xmath8
, the inoamsb model has the smallest value of @xmath206 and the lowest value of @xmath18 for a given @xmath8 value , and thus the highest value of @xmath162 . as @xmath8 increases , the value of @xmath206 increases for inoamsb and hcamsb , while it actually _ decreases _ for mamsb .
thus , for @xmath215 tev , the mamsb model yields the highest value of @xmath162 . , for @xmath174 and @xmath68 . for mamsb , we take @xmath211 tev , while for hcamsb , we take @xmath212 . in these plots , we adopt the nfw dm halo profile .
, scaledwidth=75.0% ] in fig .
[ fig : idd_m32_tb40]_b_. ) , we show the muon flux for icecube due to wino annihilation in the solar core . here again , the rates are elevated compared to the @xmath67 case .
the inoamsb model yields the highest flux at low @xmath8 , since it has the lowest @xmath18 value , and the highest higgsino component , which enters into the @xmath169 exchange diagram for @xmath216 scattering . at @xmath217 tev in the mamsb model , we obtain @xmath208 , and the solar core annihilations mainly proceed to @xmath192 states instead of @xmath209 , which diminishes the muon energy distribution and hence the detection rate for @xmath18s with @xmath165 gev . at higher @xmath8 values , the resonance is passed , and annihilation once again proceeds dominantly into @xmath209 . for high @xmath8 , the mamsb model yields the highest muon flux , due to its elevated value of @xmath162 . in fig .
[ fig : idd_m32_tb40]_c - f_. ) , we find relatively little change in halo annihilation rates due to an increase in @xmath21 , since the annihilations mainly proceed via @xmath205 , which depends mainly on gauge couplings .
the exception occurs in the mamsb model , where we do get the resonance enhancement of halo annihilations when @xmath208 .
we also obtain some @xmath21 enhancement of the @xmath47 detection rate for inoamsb and mamsb at large @xmath8 because in these cases the @xmath187 annihilation rate , which does receive @xmath21 enhancement , contributes to the detection of rather low energy @xmath47s .
in this paper , we have investigated aspects of cold dark matter in three models of anomaly mediation : mamsb , hcamsb and inoamsb . typically , each gives rise to a wino - like lightest neutralino , unless very high values of @xmath17 ( for mamsb ) or @xmath24 ( for hcamsb ) are used , in which case the @xmath30 becomes a mixed wino - higgsino state . in this class of models with a wino - like @xmath30 ,
the thermal abundance of neutralino cdm is well below measured values , unless @xmath104 gev .
we discuss four ways to reconcile the predicted abundance of cdm with experiment : 1 .
enhanced neutralino production via scalar field ( _ e.g. _ moduli ) decay , 2 .
enhanced neutralino production via gravitino decay , where gravitinos may arise thermally , or by moduli or inflaton decay , 3 .
enhanced neutralino production via heavy axino decay , and 4 .
neutralino decay to axinos , where the bulk of cdm comes from a mixture of vacuum mis - alignment produced axions and thermally produced axinos .
cases 1 and 2 should lead to a situation where all of cdm is comprised of wino - like wimps ; they will be very hard , perhaps impossible , to tell apart .
case 3 would contain a mixture of axion and wino - like wimp cdm .
it is a scenario where it is possible that both a wimp and an axion could be detected .
case 4 predicts no direct or indirect detection of wimps , but a possible detection of relic axions .
it is important to note that more than one of these mechanisms may occur at once : for instance , we may gain additional neutralino production in the early universe from moduli , gravitino and axino decay all together . in sec .
[ sec : winodet ] , we presented rates for direct and indirect detection of relic wino - like wimps .
the si direct detection cross sections are bounded from below .
ultimately , ton - scale noble liquid or supercdms experiments should probe out to @xmath42 gev , which would exceed the 100 fb@xmath159 reach of lhc ; a non - observation of signal would put enormous stress on amsb - like models as new physics .
we also evaluated sd direct detection : current experiments have little reach for amsb - like models , although icecube deepcore and possibly coupp upgrades may probe more deeply .
we also presented indirect wimp detection rates for all three amsb models .
the icecube experiment has some reach for wimps from amsb models , especially at high @xmath21 or when the @xmath30 picks up a higgsino component .
we noted an interesting inverse resonance effect in the muon flux detection rate , caused by transition from solar core annihilations to @xmath209 states , to annihilations to mainly @xmath192 states .
the detection of @xmath44s , @xmath45s , @xmath46s and @xmath47s are all elevated in amsb - like models compared to msugra , due to the high rate for @xmath185 annihilation in the galactic halo .
the results do depend on the assumed halo profile , especially for @xmath44-ray detection in the direction of the galactic core . generally ,
if a signal is seen in the @xmath45 channel , then one ought to be seen in the @xmath46 channel , and ultimately in the @xmath44 , @xmath47 ( if / when gaps flies ) or direct detection channel .
in addition , a sparticle production signal should ultimately be seen at lhc , at least for @xmath218 gev , once 100 fb@xmath159 of integrated luminosity is accrued . as a final remark ,
we note here that the dark matter detection signals all provide complementary information to that which will be provided by the cern lhc . at lhc , each model mamsb , hcamsb and inoamsb will provide a rich assortment of gluino and squark cascade decay signals which will include multi - jet plus multi - lepton plus missing @xmath219 events . in all cases ,
the wino - like lightest neutralino state will be signaled by the well - known presence of highly ionizing tracks ( hits ) from quasi - stable charginos with track length of order _
cm_s , before they decay to soft pions plus a @xmath30 .
it is noted in ref s @xcite and @xcite that the three models should be distinguishable at lhc by the differing opposite - sign / same flavor dilepton invariant mass distributions . in the case of mamsb , with @xmath220 ,
we expect a single mass edge from @xmath221 decay . in hcamsb ,
the sleptons are rather heavy , and instead @xmath222 occurs at a large rate , leading to a bump in @xmath223 , upon a continuum distribution . in inoamsb , with @xmath224 , but with @xmath225 and @xmath226 split in mass ( due to different @xmath22 quantum numbers ) , a characteristic _
double mass edge _ is expected in the @xmath227 invariant mass distribution .
we thank shanta de alwis for comments on the manuscript .
this work was supported in part by the u.s .
department of energy .
99 h. baer and x. tata , _ weak scale supersymmetry : from superfields to scattering events _ , ( cambridge university press , 2006 ) s. dimopoulos , s. raby and f. wilczek , _ phys .
* d 24 * ( 1981 ) 1681 ; s. dimopoulos and h. georgi , _ nucl .
* b 193 * ( 1981 ) 150 ; l. ibanez and g. ross , _ phys .
* b 105 * ( 1981 ) 439 ; n. sakai , _
z. phys . * c*_*11 * ( 1981 ) 153 ; m. einhorn and d. r. t. jones , _ nucl . phys . _ * b 196 * ( 1982 ) 475 ; w. marciano and g. senjanovic , _ phys .
* d 25 * ( 1982 ) 3092 .
u. amaldi , w. de boer and h. furstenau , _ phys .
* b 260 * ( 1991 ) 447 ; j. ellis , s. kelley and d. v. nanopoulos , _ phys .
* b 260 * ( 1991 ) 131 ; p. langacker and luo , _ phys .
_ * d 44 * ( 1991 ) 817 .
d. freedman , s. ferrara and p. van nieuwenhuizen , _ phys .
* d 13 * ( 1976 ) 3214 ; s. deser and b. zumino , _ phys .
lett . _ * b 62 * ( 1976 ) 335 ; s. ferrara , d. freedman , p. van nieuwenhuizen , p. breitenlohner , g. gliozzi and j. scherk , _ phys .
* d 15 * ( 1977 ) 1013 ; e. cremmer , s. ferrara , l. girardello and a. van proeyen , _ nucl .
* b 212 * ( 1983 ) 413 m. dine , w. fischler and m. srednicki , _ nucl .
* b 189 * ( 1981 ) 575 ; s. dimopoulos and s. raby , _ nucl .
* b 192 * ( 1982 ) 353 ; m. dine and w. fischler , _ phys .
* b 110 * ( 1982 ) 227 and _ nucl .
* b 204 * ( 1982 ) 346 ; c. nappi and b. ovrut , _ phys .
* b 113 * ( 1982 ) 175 ; l. alvarez - gaume , m. claudson and m. wise , _ nucl .
* b 207 * ( 1982 ) 96 ; s. dimopoulos and s. raby , _ nucl . phys . _
* b 219 * ( 1983 ) 479 ; m. dine , a. nelson , y. nir and y. shirman , _ phys .
_ * d 53 * ( 1996 ) 2658 ; for a review , see g. giudice and r. rattazzi , _ phys .
_ * 322 * ( 1999 ) 419 .
l. randall and r. sundrum , _ nucl .
_ * b 557 * ( 1999 ) 79 .
g. giudice , m. luty , h. murayama and r. rattazzi , _ j. high energy phys . _
* 12 * ( 1998 ) 027 . for recent discussion , see m. dine and n. seiberg , _ j. high energy phys . _ * 0703 * ( 2007 ) 040 ; s. de alwis , _ phys . rev . _
* d 77 * ( 2008 ) 105020 .
i. jack and d. r. t. jones , _ phys .
* b 465 * ( 1999 ) 148 .
s. weinberg , _ phys .
_ * 48 * ( 1982 ) 1303 ; r. h. cyburt , j. ellis , b. d. fields and k. a. olive , _ phys .
_ * d 67 * ( 2003 ) 103521 ; k. jedamzik , _ phys . rev . _ * d 70 * ( 2004 ) 063524 ; m. kawasaki , k. kohri and t. moroi , _ phys .
_ * b 625 * ( 2005 ) 7 .
m. kawasaki , k. kohri and t. moroi , _ phys .
* d 71 * ( 2005 ) 083502 .
r. dermisek , h. verlinde and l - t .
wang , _ phys . rev .
* 100 * ( 2008 ) 131804 .
h. baer , s. de alwis , k. givens , s. rajagopalan and h. summy , _ j. high energy phys . _ * 1005 * ( 2010 ) 069 .
de alwis , _ j. high energy phys . _ * 1003 * ( 2010 ) 078 . j. r. ellis , c. kounnas and d. v. nanopoulos , nucl . phys .
b * 247 * ( 1984 ) 373 ; j. r. ellis , a. b. lahanas , d. v. nanopoulos and k. tamvakis , phys .
b * 134 * ( 1984 ) 429 ; g. a. diamandis , j. r. ellis , a. b. lahanas and d. v. nanopoulos , phys .
b * 173 * ( 1986 ) 303 ; for a review , see a. lahanas and d. v. nanopoulos , _ phys .
rept . _ * 145 * ( 1987 ) 1 .
d. e. kaplan , g. d. kribs and m. schmaltz , _ phys .
* d 62 * ( 2000 ) 035010 ; z. chacko , m. a. luty , a. e. nelson and e. ponton , _ j. high energy phys . _ * 01 * ( 2000 ) 003 . h. c. cheng , b. dobrescu and k. matchev , _ nucl .
* b 543 * ( 1999 ) 47 .
e. komatsu _ et al .
_ ( wmap collaboration ) , arxiv:1001.4538 ( 2010 )
. t. moroi and l. randall , _ nucl .
* b 570 * ( 2000 ) 455 .
m. bolz , a. brandenburg and w. buckmuller , _ nucl . phys . _
* b 606 * ( 2001 ) 518 . j. pradler and f. steffen , _ phys .
* d 75 * ( 2007 ) 023509 .
k. kohri , m. yamaguchi and j. yokoyama , _ phys .
* d 72 * ( 2005 ) 083510 ; t. asaka , s. nakamura and m. yamaguchi , _ phys .
rev . _ * d 74 * ( 2006 ) 023520 .
m. endo , f. takahashi and t. t. yanagida , _ phys .
* d 76 * ( 2007 ) 083509 .
l. covi , j. e. kim and l. roszkowski , _ phys .
* 82 * ( 1999 ) 4180 ; l. covi , h. b. kim , j. e. kim and l. roszkowski , _ j. high energy phys . _ * 0105 * ( 2001 ) 033 .
a. brandenburg and f. steffen , jcap*0408 * ( 2004 ) 008 .
choi , j. e. kim , h. m. lee and o. seto , _ phys .
rev . _ * d 77 * ( 2008 ) 123501 .
for a recent review , see p. sikivie , hep - ph/0509198 ; m. turner , _ phys .
_ * 197 * ( 1990 ) 67 ; j. e. kim and g. carosi , arxiv:0807.3125 ( 2008 ) .
h. baer and h. summy , _ phys .
* b 666 * ( 2008 ) 5 .
h. baer , a. box and h. summy , _ j. high energy phys .
_ * 0908 * ( 2009 ) 080 .
k. rajagopal , m. turner and f. wilczek , _ nucl . phys . _ * b 358 * ( 1991 ) 447 ; for recent reviews , see l. covi and j. e. kim , arxiv:0902.0769 and f. steffen , _ eur . phys .
j. _ * c 59 * ( 2009 ) 557 .
p. grajek , g. kane , d. phalen , a. pierce and s. watson , _ phys .
* d 79 * ( 2009 ) 043506 ; g. kane , r. lu and s. watson , _ phys .
* b 681 * ( 2009 ) 151 .
d. feldman , g. kane , r. lu and b. nelson , _ phys .
* b 687 * ( 2010 ) 363 .
t. gherghetta , g. giudice and j. wells , _ nucl .
_ * b 559 * ( 1999 ) 27 ; j. l. feng , t. moroi , l. randall , m. strassler and s. su , _ phys . rev .
* 83 * ( 1999 ) 1731 ; j. l. feng and t. moroi , _ phys . rev . _
* d 61 * ( 2000 ) 095004 ; f. paige and j. wells , hep - ph/0001249 ; a. datta and k. huitu , _ phys .
_ * d 67 * ( 2003 ) 115006 ; s. asai , t. moroi , k. nishihara and t. t. yanagida , _ phys . lett . _ * b 653 * ( 2007 ) 81 ; s. asai , t. moroi and t. t. yanagida , _ phys
. lett .
_ * b 664 * ( 2008 ) 185 ; h. baer , j. k. mizukoshi and x. tata , _ phys . lett . _
* b 488 * ( 2000 ) 367 ; a. j. barr , c. lester , m. parker , b. allanach and p. richardson , _ j. high energy phys . _
* 0303 * ( 2003 ) 045 .
s. kachru , l. mcallister and r. sundrum , _ j. high energy phys .
_ * 0710 * ( 2007 ) 013 . h. verlinde , l - t .
wang , m. wijnholt and i. yavin , _ j. high energy phys . _ * 0802 * ( 2008 ) 082 ; m. buican , d. malyshev , d. r. morrison , h. verlinde and m. wijnholt , _ j. high energy phys . _ * 0701 * ( 2007 ) 107 .
h. baer , r. dermisek , s. rajagopalan and h. summy , _ j. high energy phys . _
* 0910 * ( 2009 ) 078 .
m. r. douglas and s. kachru , rev .
phys . * 79 * , 733 ( 2007 ) [ arxiv : hep - th/0610102 ] .
s. kachru , r. kallosh , a. linde and s. trivedi , _ phys .
_ * d 68 * ( 2003 ) 046005 .
v. balasubramanian , p. berglund , j. p. conlon and f. quevedo , _ j. high energy phys .
_ * 03 * ( 2005 ) 007
. f. paige , s. protopopescu , h. baer and x. tata , hep - ph/0312045 .
h. e. haber , r. hempfling and a. hoang , _
z. phys .
* c*_*75 * ( 1997 ) 539 .
d. pierce , j. bagger , k. matchev and r. zhang , _ nucl .
* b 491 * ( 1997 ) 3 .
g. belanger , s. kraml and a. pukhov , _ phys .
* d 72 * ( 2005 ) 015003 .
isared , by h. baer , c. balazs and a. belyaev , _ j. high energy phys .
_ * 0203 * ( 2002 ) 042 .
comphep v.33.23 , by a. pukhov _
_ , hep - ph/9908288 ( 1999 ) .
p. gondolo and g. gelmini , nucl
b360 * , 145 ( 1991 ) ; j. edsj and p. gondolo , phys .
rev . d*56 * , 1879 ( 1997 ) .
b. acharya , g. kane , s. watson and p. kumar , _ phys .
* d 80 * ( 2009 ) 083529 .
g. gelmini and p. gondolo , _ phys .
* d 74 * ( 2006 ) 023510 ; g. gelmini , p. gondolo , a. soldatenko and c. yaguna , _ phys .
* d 74 * ( 2006 ) 083514 ; g. gelmini , p. gondolo , a. soldatenko and c. yaguna , _ phys .
* d 76 * ( 2007 ) 015010 .
m. kawasaki and k. hakayama , _ phys .
rev . _ * d 76 * ( 2007 ) 043502 .
r. kitano , h. murayama and m. ratz , _ phys .
* b 669 * ( 2008 ) 145 .
m. bolz , a. brandenburg and w. buchmuller , _ nucl .
* b 606 * ( 2001 ) 518 l. f. abbott and p. sikivie , _ phys .
* b 120 * ( 1983 ) 133 ; j. preskill , m. wise and f. wilczek , _ phys .
* b 120 * ( 1983 ) 127 ; m. dine and w. fischler , _ phys .
* b 120 * ( 1983 ) 137 ; m. turner , _ phys .
* d 33 * ( 1986 ) 889 ; l. visinelli and p. gondolo , _ phys .
rev . _ * d 80 * ( 2009 ) 035024 . k. jedamzik , m. lemoine and g. moultaka , jcap*0607 * ( 2006 ) 010
. l. duffy _
_ , _ phys .
* 95 * ( 2005 ) 091304 and _ phys .
* d 74 * ( 2006 ) 012006 ; for a review , see s. asztalos , l. rosenberg , k. van bibber , p. sikivie and k. zioutas , _ ann .
nucl . part .
sci._*56 * ( 2006 ) 293 .
g. lazarides and q. shafi , _ phys .
_ * b 258 * ( 1991 ) 305 ; k. kumekawa , t. moroi and t. yanagida , _ prog . theor . phys . _
* 92 * ( 1994 ) 437 ; t. asaka , k. hamaguchi , m. kawasaki and t. yanagida , _ phys .
* b 464 * ( 1999 ) 12 .
w. buchmller , p. di bari and m. plumacher , _ nucl .
phys . _ * b 643 * ( 2002 ) 367 and erratum - ibid,*b793 * ( 2008 ) 362 ; _ annal .
* 315 * ( 2005 ) 305 and _ new jou .
phys._*6 * ( 2004 ) 105 .
p. grajek , g. kane , d. phalen , a. pierce and s. watson , arxiv:0807.1508 ; p. ullio , _ j. high energy phys .
_ * 0106 * ( 2001 ) 053 .
h. baer , c. balazs , a. belyaev and j. ofarrill , jcap * 0309 * , ( 2003 ) 007 .
lepsusywg , note lepsusywg/02 - 04.1 . see http://dmtools.berkeley.edu/limitplots/ , maintained by r. gaitskell and j. filippini . for reviews , see _ e.g. _
g. jungman , m. kamionkowski and k. griest,_phys .
* 267 * ( 1996 ) 195 ; a. lahanas , n. mavromatos and d. nanopoulos , _ int .
j. mod .
* d 12 * ( 2003 ) 1529 ; m. drees , hep - ph/0410113 ; k. olive , `` tasi lectures on astroparticle physics '' , astro - ph/0503065 ; g. bertone , d. hooper and j. silk , _ phys . rept .
_ * 405 * ( 2005 ) 279 . for a recent review of axion
/ axino dark matter , see f. steffen , arxiv:0811.3347 ( 2008 ) .
j. navarro , c. s. frenk and s. d. m. white , _ astrophys .
j. _ * 462 * ( 1996 ) 563 .
h. baer , e. k. park and x. tata , _ new jou
. phys._*11 * ( 105024 ) 2009 .
p. gondolo , j. edsjo , p. ullio , l. bergstrom , m. schelke and e. a. baltz , jcap * 0407 * ( 2004 ) 008 ; for a review , see f. boudjema , j. edsjo and p. gondolo , arxiv:1003.4748 .
m. pearce ( pamela collaboration ) , _ nucl .
phys . _ * 113 * _ ( proc . suppl . ) _
( 2002 ) 314 .
k. mori , c. j. hailey , e. a. baltz , w. w. craig , m. kamionkowski , w. t. serber and p. ullio , _ astrophys .
j. _ * 566 * ( 2002 ) 604 ; c. j. hailey _ et al . _ , jcap * 0601 * ( 2006 ) 007 ; h. baer and s. profumo , jcap * 0512 * ( 2005 ) 008 . h. baer and j. ofarrill , jcap * 0404 * ( 2004 ) 005 ; h. baer , a. belyaev , t. krupovnickas and j. ofarrill , jcap * 0408 * ( 2004 ) 005 .
m. drees and m. nojiri , _ phys .
rev . _ * d 47 * ( 1993 ) 376 ; h. baer and m. brhlik , _ phys .
* d 57 * ( 1998 ) 567 | supersymmetric models based on anomaly - mediated susy breaking ( amsb ) generally give rise to a neutral wino as a wimp cold dark matter ( cdm ) candidate , whose thermal abundance is well below measured values . here , we investigate four scenarios to reconcile amsb dark matter with the measured abundance : 1 .
non - thermal wino production due to decays of scalar fields ( _ e.g. _ moduli ) , 2 .
non - thermal wino production due to decays of gravitinos , 3 .
non - thermal wino production due to heavy axino decays , and 4 .
the case of an axino lsp , where the bulk of cdm is made up of axions and thermally produced axinos . in cases 1 and 2 , we expect wino cdm to constitute the entire measured dm abundance , and we investigate wino - like wimp direct and indirect detection rates .
wino direct detection rates can be large , and more importantly , are bounded from below , so that ton - scale noble liquid detectors should access all of parameter space for @xmath0 gev .
indirect wino detection rates via neutrino telescopes and space - based cosmic ray detectors can also be large . in case 3
, the dm would consist of an axion plus wino admixture , whose exact proportions are very model dependent . in this case
, it is possible that both an axion and a wino - like wimp could be detected experimentally . in case 4 .
, we calculate the re - heat temperature of the universe after inflation . in this case , no direct or indirect wimp signals should be seen , although direct detection of relic axions may be possible . for each dm scenario ,
we show results for the minimal amsb model , as well as for the hypercharged and gaugino amsb models .
* neutralino , axion and axino cold dark matter + in minimal , hypercharged and gaugino amsb * + howard baer@xmath1 , radovan derm ' iek@xmath2 , shibi rajagopalan@xmath3 , heaya summy@xmath3 + _ 1 .
dept . of physics and astronomy , university of oklahoma , norman , ok 73019 , usa + 2 .
dept . of physics , indiana university , bloomington in 47405 , usa + _ |
the l3 detector @xcite started in 1989 to collect data at lep in @xmath1 collisions at a centre of mass energy @xmath4 .
it has contributed to the success of the lepi programme , when more than twenty millions of z bosons were produced testing the standard model of electroweak interactions @xcite ( sm ) with an impressive precision @xcite . in
1995 the campaign for the gradual increase of the lep beam energy started , bringing the experiments in the so called lepii era . in the present paper
i will give a snapshot of several l3 results from the high integrated luminosity runs of 1997 and 1998 at an average @xmath5 of @xmath6 and @xmath3 , respectively .
these two energies will be indicated as @xmath7 and @xmath8 hereafter and correspond to 55pb@xmath0 and 176pb@xmath0 of integrated luminosity , respectively .
some of the results described here are published , others are preliminary . results from the lower energy runs at @xmath9 with 10pb@xmath0 of integrated luminosity and @xmath10 with 21pb@xmath0 , are sometimes included in the discussed analyses .
these results , which are only a subsample of the lepii physics program carried out by the l3 experiment , are classified in the following according to the type and number of particles primarily produced in the @xmath11 interactions : two fermions , two w , just one w , two or more photons , one z and a photon and two z. tests of theories of gravity with extra spatial dimensions are finally presented .
fermion pair production is a fundamental process to be studied at lepii both as a verification of sm predictions and as a necessary check of the understanding of the detector performance .
a kinematically favoured configuration in fermion pair production at energies above the z pole is the emission of an hard initial state photon which lowers the effective centre of mass energy , @xmath12 , to the z resonance .
this process is known as `` radiative return to the z '' and yields high energy photons either in the detector or almost collinear with the beams and hence undetected .
it is customary to express the results of the cross sections and forward backward asymmetries of the fermions by separating the full data sample ( @xmath13 ) and the purely high energy one , from which the radiative return to the z events are rejected ( @xmath14 ) .
table 1 reports the l3 results for the two considered energies @xcite for quark , muon , tau and electron pairs . a good agreement with the sm predictions is observed .
figure [ fig:1 ] presents the evolution with @xmath5 of the fermion pair cross section as predicted by the sm and measured by the l3 experiment . from the analysis of events with a single photon visible in the detector
@xcite it is possible to derive the radiative neutrino pair production cross section . at @xmath8
the measured cross section is : @xmath15 where the first error is statistical and the second systematic .
the extrapolation of this value to the total cross section reads : @xmath16 + channel & @xmath13 & @xmath14 & @xmath13 & @xmath14 + @xmath17 & @xmath18&@xmath19 & & + @xmath20 & @xmath21&@xmath22 & @xmath23&@xmath24 + @xmath25 & @xmath26&@xmath27 & @xmath28&@xmath29 + @xmath30 & @xmath31&@xmath32 & @xmath33&@xmath34 + + @xmath17 & @xmath35&@xmath36 & &
+ @xmath20 & @xmath37&@xmath38 & @xmath39&@xmath40 + @xmath25 & @xmath41&@xmath42 & @xmath43&@xmath44 + @xmath30 & @xmath45&@xmath46 & @xmath47&@xmath48 + figure [ fig:1 ] shows the evolution of these two cross sections as a function of @xmath5 as well as a comparison of @xmath49 with the sm expectations in presence of 2 , 3 or 4 neutrino species . from the photon energy spectrum it is possible to determine the number @xmath50 of light neutrino species as @xmath51 , with respectively statistic and systematic errors
the average of this result with l3 results at the z pole @xcite yields : @xmath52 the study of fermion pairs constitutes also an interesting probe of possible new physics beyond the sm .
the examples of supersymmetry and contact interactions will be highlighted in the following .
the minimal supersymmetric standard model @xcite ( mssm ) introduces a new quantum number , the r parity @xcite .
this quantity distinguishes ordinary particles and supersymmetric ones , requiring an even number of the latter in each interaction vertex and hence constraining the lightest supersymmetric particle to be stable . as a consequence of the negative results of the search for supersymmetric particles at the present colliders ,
it is interesting to investigate possible signatures of supersymmetric models with broken r
parity @xcite , where triple vertices with a supersymmetric particle and two sm ones are allowed .
some of these signatures are the production of a pair of electrons in @xmath1 collisions mediated by the @xmath53 or @xmath54channel muon or tau sneutrino exchange , the production of a muon pair via the @xmath53channel exchange of a tau sneutrino or that of a tau pair via a muon sneutrino .
these sneutrinos are the scalar partner of the sm neutrinos .
the presence of these processes would lead to a resonant structure in the lepton pair production cross section around the mass of the exchanged sneutrino . from the investigation @xcite of the cross sections and asymmetries of final states electrons , muons and tau ,
no evidence for such signatures is found and limits at 95% confidence level ( cl ) are derived on the coupling constant of the sneutrino as a function of its mass , as reported in figure [ fig : snu ] , for the electron and muon signatures .
contact interactions can be thought of as a general formalism to describe new physics from a scale much higher than the energy of an investigated process .
an example is the contact interaction structure used by fermi to describe the beta decay @xcite fifty years before colliders reached the necessary energy to produce the w boson , whose mass is now known to be the scale of the process .
analogously a new interaction of coupling constant @xmath55 and scale @xmath56 yet far above direct experimental reach can be probed in fermion pair production in @xmath1 interactions .
it is sketched in figure [ fig : ci ] and is parametrised via an effective lagrangian @xcite : @xmath57 where @xmath58 and @xmath59 are the left and right handed initial state electron and final state fermion fields and the coefficients @xmath60 allow to choose which helicities contribute to the fermion pair production within the different models , as listed in table 2 .
the contact interaction will manifest itself in the differential cross sections of a given phase space parameter as a function of @xmath61 in the interference terms of sm and new physics and as a function of @xmath62 for the pure new physics part .
the search for contact interactions @xcite proceeds by performing a @xmath63 fit to the charged fermion pair cross sections and asymmetries measurements presented above , with @xmath56 as a free parameter with the convention @xmath64 .
the results of those fits are compatible with the sm for all the possible choices of the helicities and 95% cl limits as high as @xmath65 are set on the scale @xmath56 of some models .
all the limits are presented in figure [ fig : ci ] , where @xmath66 and @xmath67 denote respectively the limits in the case of the upper and lower signs of the @xmath68 parameters of table 2 .
lower energy data @xcite are also included .
.helicities contributions in different models of contact interactions . [ cols="^,^,^,^,^,^,^,^,^,^",options="header " , ] from the analysis @xcite of the presented distributions as well as hadronic w pair events and @xmath69 pairs , together with the discriminating variables of the zz event selections and the hadronic cross section , no statistically significant hints for lsg are found in the l3 data at @xmath7 and @xmath8 and the 95% cl limits presented in table 4 are set on @xmath70 . assuming that no higher order operators give sizeable contributions to the lsg mediated boson and fermion pair production and that the meaning of the cutoff parameter is the same for all the investigated processes , it is possible to combine the boson and fermion limits .
they are as high as @xmath71 for @xmath72 and @xmath73 for @xmath74 at 95% cl and are competitive with those achieved from a combined analysis of the lepii data @xcite
. the l3 collaboration also investigated the direct production of a graviton , lost in the extra dimensions , associated with a photon @xcite or a z @xcite .
the phase space favours the first process .
these searches did not not yield any evidence for the expected lsg signatures @xcite allowing to set 95% cl limits on the lsg scale in excess of @xmath75 .
this walk through part of the recent l3 results shows a quite wide activity , even on subjects that were not expected to be covered at the start of the lepii experimental program @xcite . on this point
all the experiments match the efforts of the accelerator crew that is impressing the community with record breaking performance in term of energy , luminosity and operation efficiency .
several important results have been omitted from this review due to the lack of space and the framework of the discussion .
the most important worth mentioning are the 95% cl mass limits on the sm higgs boson at @xmath76 @xcite and that on the lightest mssm neutralino at @xmath77 @xcite , both achieved with the analysis of the full data sample collected by l3 up to @xmath8 .
moreover the l3 experiment has contributed with a large amount of results to the field of two photon physics , as reviewed in reference @xcite .
let me close this review with the hope that something more than precise cross section measurements and interesting limits is awaiting the lep community in the secrets of the last gev still to be squeezed from the machine in the high energy runs of 1999 up to a centre of mass energy of @xmath78 , and even beyond , in the final run of the year 2000 .
i am grateful to my colleagues of the l3 experiment , too numerous to be named here , with whom i shared this challenging physics adventure since i arrived at cern as an undergraduate student , through the end of the century .
it is only with the lively daily interaction with many of them on several of the subjects i described in this work , that i learned something about experimental high energy physics .
i wish to thank the organisers of this conference in particular for having helped me to find a swimming pool on the leninskii prospekt and to fix some complex part of my travel arrangements , two tasks at more than 95% cl beyond the reach of my russian phrase book .
99 l3 collab . , b. adeva , nucl .
. meth . * a 289 * ( 1990 ) 35 ; l3 collab . , o. adriani , physics reports * 236 * ( 1993 ) 1 ; m. chemarin , nucl . inst . meth . * a 349 * ( 1994 ) 345 ; m. acciarri , nucl . inst
. meth . * a 351 * ( 1994 ) 300 ; i. c. brock , nucl . instr . and meth . * a 381 * ( 1996 ) 236 ; a. adam , nucl . inst
. meth . * a 383 * ( 1996 ) 342 ; g. basti , nucl .
. meth . * a 374 * ( 1996 ) 293 ; | a data sample corresponding to an integrated luminosity of 232pb@xmath0 was collected in 1997 and 1998 by the l3 experiment at lep in @xmath1 collisions at centre of mass energies between @xmath2 and @xmath3 .
pair production of fermions and bosons is studied and compared with the standard model expectations .
events with a single detected photon or w boson are also considered .
the measurement of several standard model cross sections is discussed .
the presence and the magnitude of triple couplings of charged and neutral electroweak gauge bosons is investigated .
these processes are used to probe new physics beyond the standard model , including the existence of extra spatial dimensions .
-2.0 cm -3.0 cm * recent results from the l3 experiment * + salvatore mele + ep division , cern , ch1211 , genve 23 , switzerland + _ on behalf of the l3 collaboration _ |
according to a finding by bondi and gold ( 1950 ) the magnetic multipole moments which result from electric currents in a perfectly conducting fluid occupying a simply connected body can not grow boundlessly .
they remain for any fluid motion within bounds determined by the initial magnetic flux through the surface of the body . in a simple spherical mean field dynamo model of @xmath0type proposed by krause and steenbeck ( 1967 )
, however , the magnetic field grows endlessly both inside and outside the fluid body even in the limit of perfect conductivity .
this conflict has been resolved in a more sophisticated model by rdler ( 1982 ) that properly considers the specific structure of the mean electromotive force resulting from the constraints on the fluid motion at the boundary of the body .
it was shown analytically that this structure indeed ensures the boundedness of the magnetic multipole moments and thus excludes an infinite growth of the magnetic field outside the fluid body in the high conductivity limit . however , the question remained open whether or not a dynamo may work inside the fluid body , which then would have to be invisible , or hidden , in the sense that its magnetic field is completely confined inside this body .
only a few arguments were given which support the conjecture that such a dynamo may exist . in between the issue of the adjustment of a dynamo in the high conductivity limit to the constraints posed by the bondi and gold theorem
was also addressed in papers by hollerbach , galloway and proctor ( 1995 and 1998 ) .
they considered a dynamo in a spherical fluid shell with a chaotic flow surrounded by insulating space inside and outside and demonstrated that it works also in the limit of high conductivity , in which then the magnetic field in the insulating spaces vanishes . in the present paper
we return to the mean field model proposed by rdler ( 1982 ) .
after a few short explanations concerning the case of high conductivity and the mean field concept in dynamo theory ( sections 2 and 3 ) and the simple model by krause and steenbeck ( 1967 ) ( section 4 ) we deliver a systematic treatment of our model and present numerical results giving evidence for the existence of a hidden fast dynamo in the above sense ( section 5 ) .
finally we discuss a few conclusions ( section 6 ) .
we consider here dynamo models consisting of an electrically conducting fluid occupying a simply connected region surrounded by free space .
let us assume that the magnetic flux density , @xmath1 , is governed by the induction equation @xmath2 inside the fluid body , continues as a potential field @xmath3 in outer space and vanishes at infinity . as usual
, @xmath4 means the velocity of the fluid and @xmath5 its magnetic diffusivity . for steady @xmath4
we may expect solutions @xmath6{0mm}{8.2pt } } } } } \exp ( pt ) \big)$ ] , where @xmath7{0mm}{8.2pt}}}}$ ] is a steady complex field and @xmath8 a complex quantity independent of space and time coordinates .
the real part @xmath9 of @xmath8 is the growth rate of @xmath1 .
we speak of a dynamo if there is at least one solution @xmath1 with a non negative @xmath9 . for many applications to cosmic objects
the limit of high electrical conductivity , that is @xmath10 , deserves special interest . to define this limit more precisely ,
let us measure all length in units of @xmath11 being a typical scale of @xmath4 or @xmath1 , and the time in units of @xmath12 , with @xmath13 being a typical magnitude of @xmath4
. then ( [ i1 ] ) applies with @xmath5 replaced by @xmath14 , where @xmath15 is the magnetic reynolds number defined by @xmath16 and the high conductivity limit corresponds to @xmath17 .
we mention two important aspects of dynamo action in the high conductivity limit .
to explain the first one we compare dynamos which differ only in the value of @xmath15 . if then the largest growth rate @xmath18 remains positive and takes a finite positive value as @xmath17 we speak of a fast dynamo , otherwise , that is , if @xmath18 tends to zero or to a negative value , of a slow dynamo .
the second aspect is that in the case of perfect conductivity , @xmath19 , the magnetic field has to satisfy the requirements posed by the theorem by bondi and gold ( 1950 ) . for the sake of simplicity
we suppose the fluid body to be a sphere and formulate this theorem as proposed by rdler ( 1982 ) .
we use spherical coordinates @xmath20 and represent the potential @xmath21 introduced with ( [ i2 ] ) in the form @xmath22 with spherical harmonics @xmath23 defined by @xmath24 where the @xmath25 are associated legendre polynomials .
the @xmath26 are complex constants satisfying @xmath27 which define multipole moments , e.g. , those with @xmath28 the dipole moment , with @xmath29 the quadrupole moment etc .
depending on the motions at the boundary of the fluid body the @xmath26 may vary in time . according to the theorem by bondi and gold , however , they are bounded in the sense that @xmath30 with @xmath31 given by the initial distribution of the magnetic flux at the boundary .
this in particular excludes any exponential growth of the @xmath26 .
a third aspect which concerns the conditions to be satisfied by @xmath1 at the boundary will be discussed below .
in cases in which the magnetic field and the fluid motion are of turbulent nature , or show by other reasons complex structures in space or time , the mean field approach to dynamo models has proved to be useful ; see , e.g. , krause and rdler ( 1980 ) . in this approach both the magnetic flux density @xmath1 and the fluid velocity @xmath4 are considered as a sum of a mean field , @xmath32{0mm}{8.2pt}}}}}$ ] or @xmath33 , defined by a proper averaging procedure , and a fluctuating part , @xmath34 or @xmath35 . provided that the reynolds averaging rules apply
, we may conclude from equations ( [ i1 ] ) that @xmath36{0mm}{8.2pt}}}}}+ { { \mbox{\boldmath $ \nabla$}}}\x ( { \ol{{{\mbox{\boldmath $ u$}}}}}\x { { { \,\hspace{0.01em}\ol{\!\hspace{-0.01em}{{\mbox{\boldmath $ b$}}}}{\rule[0pt]{0mm}{8.2pt}}}}}+ { \mbox{\boldmath $ \cal{e}$ } } ) - \partial { { { \,\hspace{0.01em}\ol{\!\hspace{-0.01em}{{\mbox{\boldmath $ b$}}}}{\rule[0pt]{0mm}{8.2pt}}}}}/ \partial t = { { \mbox{\boldmath $ 0 $ } } } \ , , \q { { \mbox{\boldmath $ \nabla$}}}\cdot { { { \,\hspace{0.01em}\ol{\!\hspace{-0.01em}{{\mbox{\boldmath $ b$}}}}{\rule[0pt]{0mm}{8.2pt}}}}}= 0 \ , , \label{i10}\ ] ] inside the fluid body , where @xmath37 is an electromotive force due to fluctuations , @xmath38 for a given motion @xmath37 is a linear functional of @xmath32{0mm}{8.2pt}}}}}$ ] . under the usually accepted assumption of sufficiently weak variation of @xmath32{0mm}{8.2pt}}}}}$ ] in space and time
it can be represented as @xmath39{0mm}{8.2pt}}}}_j - b_{ijk } \partial { { \,\hspace{0.01em}\ol{\!\hspace{-0.01em}b}{\rule[0pt]{0mm}{8.2pt}}}}_j / \partial x_k \ , . \label{i12}\ ] ] here we rely on cartesian coordinates and use the summation convention .
the tensors @xmath40 and @xmath41 are , apart from @xmath5 , determined by @xmath33 and @xmath42 .
if , for instance , @xmath43 and @xmath42 represents a homogeneous isotropic turbulence we have @xmath44{0mm}{8.2pt}}}}}- \beta \ , { { \mbox{\boldmath $ \nabla$}}}\x { { { \,\hspace{0.01em}\ol{\!\hspace{-0.01em}{{\mbox{\boldmath $ b$}}}}{\rule[0pt]{0mm}{8.2pt}}}}}\label{i13}\ ] ] with coefficients @xmath45 and @xmath46 determined by @xmath35 or , what is the same here , by @xmath4 and being independent of space coordinates .
the first term on the right hand side describes the @xmath45effect , that is , the occurrence of a mean electromotive force parallel or antiparallel to @xmath1 but vanishes if the turbulence is reflectionally symmetric .
the second term gives rise to introduce a mean field conductivity different from the usual one .
several results are available concerning the connection of @xmath45 and @xmath46 with the properties of the @xmath4field .
we mention in particular those obtained in the second order correlation approximation ; see krause and rdler ( 1980 ) . in the high conductivity limit
, in this context defined by @xmath47 with @xmath48 and @xmath49 being correlation length and time of the @xmath4field , @xmath45 and @xmath46 then take the values @xmath50 and @xmath51 given by @xmath52 in general both @xmath45 and @xmath46 do not vanish in this limit .
one of the simplest mean field dynamo models , which can be treated analytically , has been proposed by krause and steenbeck ( 1967 ) .
the fluid body is supposed to be a sphere of radius r , again surrounded by free space .
any mean motion of the fluid is ignored , @xmath53 , and thinking of homogeneous isotropic turbulence the electromotive force @xmath37 is taken in the form ( [ i13 ] ) with constant @xmath45 and @xmath46 .
specifying now equations ( [ i10 ] ) to this case we write simply @xmath1 instead of @xmath32{0mm}{8.2pt}}}}}$ ] and use @xmath54 and @xmath55 as units of length and time . looking then for solutions @xmath1 varying like @xmath56 with @xmath57 we may reduce these equations to @xmath58 where @xmath59 is a dimensionless measure of the @xmath45effect , @xmath60 in this model equation ( [ i15 ] )
was completed by the requirements that @xmath1 is continuous across the boundary and has the structure given by ( [ i2 ] ) and ( [ i4 ] ) in outer space .
the problem posed in this way has been investigated for axisymmetric @xmath1fields by krause and steenbeck ( 1967 ) for the steady case and by voigtmann ( 1968 ) for the time
dependent case . a detailed treatment of the general case including non axisymmetric time
dependent fields is given in krause and rdler ( 1980 ) .
we only mention here a few particular results .
there are independent solutions @xmath1 which possess the form of single multipole fields , that is dipole fields , quadrupole fields etc . in outer space . in all cases
@xmath8 is real , it takes a negative value for @xmath61 and increases monotonously with @xmath62 , runs through zero for some marginal value of @xmath62 and behaves like @xmath63 as @xmath64 . comparing all these solutions we find the smallest marginal value for a solution of dipole type . denoting this value by @xmath65 we have @xmath66 consider now the high conductivity limit .
then the condition of dynamo action resulting from ( [ i16 ] ) and ( [ i17 ] ) takes the form @xmath67 .
it seems well possible to satisfy this condition , that is , the dynamo may well work in this limit .
the dimensional growth rate is then given by @xmath68 .
since in the case of a dynamo with growing magnetic fields @xmath8 must be positive , this growth rate is positive too , that is , we have a fast dynamo .
the structure of the magnetic field depends , if its multipole character is specified , only on the value of the parameter @xmath59 irrespective of that of @xmath5 .
the field is non
zero both inside and outside the fluid body , and this can not change as @xmath10 . if a dynamo works and the field grows exponentially in time then the @xmath26 introduced with ( [ i4 ] ) do so , too .
this , however , is in conflict with the bondi gold theorem .
one of the shortcomings of the model considered so far , which might be the reason for this conflict , is the assumption of an electromotive force @xmath37 in the form ( [ i13 ] ) with constant @xmath45 and @xmath46 , which can be justified for homogeneous isotropic turbulence only . near the boundary of the fluid body
the turbulence must necessarily deviate from homogeneity and isotropy , and @xmath37 has to take a more complex form . as already mentioned , in a paper by rdler ( 1982 ) a modified model was studied with a spherically symmetric distribution of the turbulence .
then the radial direction necessarily occurs as preferred direction in the turbulence and therefore we have a more complex form of @xmath37 . it was shown that with this modification the conflict is resolved . more precisely , with these assumptions any growth of the magnetic field at the boundary of the fluid body or in outer space can be ruled out .
turning now to this model we assume again that the mean magnetic flux density @xmath32{0mm}{8.2pt}}}}}$ ] is governed by equations ( [ i10 ] ) in a spherical fluid body @xmath69 and continues as a potential field as given by ( [ i2 ] ) , or ( [ i4 ] ) , in the outer space @xmath70 .
we further exclude any mean motion , @xmath43 , and specify the electromotive force @xmath37 in accordance with a spherically symmetric turbulence . for the sake of simplicity we write in the following again @xmath1 instead of @xmath32{0mm}{8.2pt}}}}}$ ] , and @xmath4 instead of @xmath35 , and we rely again on spherical coordinates @xmath71 , @xmath72 .
spherical symmetry of the turbulence is understood here in the sense that all averaged quantities determined by the @xmath4field are invariant under arbitrary rotations of this field about arbitrary axes through the center @xmath73 of the fluid body . from this definition
we may conclude by standard reasoning ( see , e.g. , krause and rdler ( 1980 ) ) that the electromotive force @xmath37 must have the form @xmath74 with scalar coefficients @xmath75 determined by @xmath4 and depending on position only through @xmath76 but not through @xmath77 or @xmath72 . here
@xmath78 is the radial unit vector and @xmath79 the symmetric part of the gradient tensor of @xmath80 so that , if we refer to cartesian coordinates , @xmath81 . some more details concerning the derivation of ( [ m1 ] ) are given in rdler ( 1982 ) .
if the turbulence is reflectionally symmetric about planes containing the center of the body @xmath82 and @xmath83 are equal to zero . in the case of homogeneous isotropic turbulence only the coefficients @xmath84 and @xmath85 can be non zero and the others have to vanish , so that we return to ( [ i13 ] ) .
later we will also assume that the turbulence is steady , that is , that all averaged quantities depending on @xmath4 are invariant under shifts along the time axis . in this case
the coefficients @xmath86 are independent of time .
even if we accept again the second order correlation approximation the calculation of the coefficients @xmath86 for points near the boundary of the fluid body is , at least for finite conductivity , rather complex .
we note here only results for the high conductivity limit , defined as above by @xmath87 , which were already obtained in a slightly different form by rdler ( 1982 ) . denoting the mentioned coefficients in this limit by @xmath88 we have @xmath89 and @xmath90 with @xmath91 and @xmath92 and analogous definitions of @xmath93 and @xmath94 . according to our assumptions the @xmath95 , like the averaged quantities under the integrals ,
depend on @xmath96 via @xmath76 only .
it seems natural to assume that @xmath97 and @xmath98 are non negative everywhere . at the boundary of the fluid body we have @xmath99 and therefore @xmath100
let us now represent the magnetic flux density @xmath80 as a sum of a poloidal and toroidal part , @xmath101 where @xmath102 , and expand the defining scalars @xmath103 and @xmath104 in series of spherical harmonics @xmath105 .
it can easily be followed up that the equations and conditions governing @xmath1 imply no coupling between contributions to @xmath1 differing in @xmath106 or @xmath107 so that we may restrict ourselves to the simple solutions defined by @xmath108 due to the factor @xmath54 in ( [ r1 ] ) the dimensions of @xmath103 and @xmath104 coincide .
preparing the definition of dimensionless quantities we introduce the constants @xmath109 and @xmath110 with the dimension of a velocity and a magnetic diffusivity .
we will assume homogeneous isotropic turbulence in some central region of the fluid body and identify these constants with @xmath50 and @xmath51 given by ( [ i14 ] ) for this region .
in the following we measure all lengths in units of @xmath54 , the time in units of @xmath111 , further the @xmath112 in units of @xmath109 , the @xmath113 in units of @xmath114 , the @xmath115 in units of @xmath110 , and @xmath116 in units of @xmath117 . using standard methods
we may then reduce equations ( [ i10 ] ) for @xmath1 to @xmath118 with @xmath119 @xmath120 and @xmath121 @xmath122 equations ( [ i2 ] ) for the outer space are equivalent with @xmath123 note that @xmath124 coincides with @xmath125 for the chosen @xmath106 and @xmath107 .
as a consequence of the divergence relation for @xmath1 its normal component and thus @xmath103 has to be continuous across the boundary @xmath126 . for finite conductivity of the fluid , that is @xmath127 , we may exclude surface currents so that the tangential components have to be continuous too , and so @xmath128 and @xmath104
. together with ( [ r7 ] ) we may conclude that @xmath129 the original problem of the determination of @xmath1 occurs then as the problem of solving the equations ( [ r3 ] ) , completed by ( [ r6a ] ) and ( [ r6b ] ) , with the boundary conditions ( [ r8 ] ) . in the case of perfect conductivity ,
@xmath130 , again the continuity of @xmath103 has to be required .
however , surface currents can no longer be excluded , which correspond to discontinuities of @xmath128 and @xmath104 .
more precisely , such currents are given by @xmath131 { { \mbox{\boldmath $ \hat{r}$}}}\x { { \mbox{\boldmath $ \nabla$}}}y_n^m + [ t ] { { \mbox{\boldmath $ \nabla$}}}y_n^m $ ] ) , where @xmath132 means the magnetic permeability of the fluid and @xmath133 $ ] stands for @xmath134 .
in this case the equations ( [ r3 ] ) have to be completed by conditions which we will formulate later .
remaining with the case of infinite conductivity we first demonstrate , repeating the ideas described by rdler ( 1982 ) , that our model is compatible with the bondi gold theorem . for this purpose
we consider the first equation ( [ r3 ] ) , completed by ( [ r6a ] ) , for @xmath135 .
putting there @xmath130 , inserting @xmath136 according to ( [ m2 ] ) and using ( [ m4 ] ) we find @xmath137 and hence @xmath138 remembering that @xmath139 and that @xmath140 is non negative we see that ( [ i6 ] ) is indeed satisfied .
we now turn our attention to the question whether in the limit of infinite conductivity magnetic fields can grow inside the fluid body .
we rely on equations ( [ r3 ] ) to ( [ r7 ] ) , put @xmath130 and insert @xmath141 according to ( [ m2 ] ) . with the idea to have a model with a minimum of free parameters we choose simply @xmath142 with @xmath143 this corresponds to the assumption of homogeneous isotropic turbulence inside the spherical region @xmath144 and of inhomogeneous anisotropic turbulence in the surrounding shell @xmath145 .
note that @xmath146 at @xmath126 and therefore ( [ m4 ] ) is satisfied .
finally we assume that there is initially no magnetic field at the boundary of the fluid body , that is @xmath147 at @xmath126 .
due to ( [ r9 ] ) we have then @xmath148 at @xmath126 at any time .
starting from the second equation ( [ r3 ] ) together with ( [ r6b ] ) and using ( [ m2 ] ) , ( [ r12 ] ) and ( [ r13 ] ) we arrive at a relation for @xmath149 analogous to ( [ r9 ] ) from which we may conclude that also @xmath150 at @xmath126 at any time .
the behaviour of @xmath1 is then governed by ( [ r3 ] ) , completed by ( [ r6a ] ) , ( [ r6b ] ) , ( [ m2 ] ) , ( [ r12 ] ) and ( [ r13 ] ) , and the role of boundary conditions is taken by @xmath147 at @xmath126 .
( we note that the remarks concerning the boundary condition for @xmath103 in rdler ( 1982 ) , section 3.6 , are incorrect . )
the problem posed in this way has been investigated numerically .
there are indeed solutions @xmath1 , or @xmath103 and @xmath104 , which for sufficiently large @xmath151 grow exponentially in time .
the growth rate @xmath8 introduced as above again proves to be real . the fastest growing solution for a given @xmath151 is always one of dipole type , that is @xmath28 . we restrict our attention in this paper to such solutions and , furthermore , to the case @xmath152 .
the dependence of the growth rate @xmath8 on @xmath151 is shown in figure [ f1 ] .
clearly dynamo action is possible if @xmath153 with @xmath154 let us define an effective radius @xmath155 of our model by equating of @xmath156 to the value of @xmath65 for the simple model discussed above given by ( [ i17 ] ) , that is @xmath157 . comparing this with the corresponding relation for the modified model under consideration , that is @xmath158
, we find @xmath159 .
this means that the shell with the fractional radius between @xmath160 and @xmath161 showing deviations from a homogeneous isotropic turbulence is less effective for dynamo action than the central region of the fluid body without such deviations .
figures [ f2 ] and [ f3 ] show the radial distribution of the energy densities of the poloidal and the toroidal parts of the magnetic field and field pictures for two examples with @xmath162
. it would be interesting to study a sequence of dynamo models with positive @xmath163 approaching zero . in this context
we must have in mind that @xmath163 occurs not only with the dissipation terms @xmath164 and @xmath165 in ( [ r3 ] ) .
in addition the coefficients @xmath166 entering @xmath167 and @xmath168 via ( [ r6a ] ) and ( [ r6b ] ) depend on the magnetic diffusivity , that is , on a parameter proportional to @xmath163 .
there are , however no results available which describe this dependence in the neighbourhood of a jump of the magnetic diffusivity as it occurs at the boundary .
several numerical calculations have been carried out on the basis of equations ( [ r3 ] ) including the dissipation terms but using expressions for @xmath166 as given by ( [ m2 ] ) and ( [ m3 ] ) , that is , for infinite conductivity .
of course , in these calculations the boundary conditions ( [ r8 ] ) have been used .
indeed the solutions obtained in this way approach the corresponding ones with @xmath130 in a very satisfying manner as @xmath169 .
the only difference occurs in the distribution of the electric currents near the boundary , what is understandable since for @xmath130 a part of them occurs as surface currents .
it is interesting to observe how the magnetic energy concentrates itself more and more inside the conducting body as @xmath169 .
table [ t1 ] shows for one example the dependence of the ratio of the energies @xmath170 and @xmath171 in the outer and in all space on @xmath163 .
c@c + @xmath172 & @xmath173 + + @xmath174 & @xmath175 + @xmath176 & @xmath177 + @xmath178 & @xmath179 +
as already shown in an earlier paper by rdler ( 1982 ) , the mean field model reconsidered here , in contrast to a more simplified one , meets the requirements posed by the bondi gold theorem .
it was , however , only conjectured but not really demonstrated that it admits dynamo action inside the fluid body in the high conductivity limit .
the present paper clearly demonstrates the possibility of a fast dynamo inside the fluid body , whose magnetic field is then , as required by the bondi gold theorem , completely confined in the fluid body and in that sense invisible from outside .
the model is also in another respect , which is not necessarily connected with the high conductivity limit , of interest for the fundamentals of mean field dynamo theory . in most of the mean field dynamo models elaborated in view of cosmic objects only a few contributions to the mean electromotive force have been taken into account , and others were cancelled in the vague hope that they are of minor importance .
we demonstrate here the possibility of dynamo action in an idealised model which is consistent in the sense that it includes all contributions to the electromotive force which occur under this idealisation .
it remains to be discussed whether the concentration of the dynamo generated magnetic field in the interior of a highly conducting body , which was demonstrated above , may indeed occur in astrophysical bodies . | the paper deals with a simple spherical mean field dynamo model of @xmath0type in the case of high electrical conductivity .
a spherically symmetric distribution of turbulent motions is assumed inside a spherical fluid body surrounded by free space , and some complete form of the corresponding mean electromotive force is taken into account . for a turbulence lacking reflectional symmetry finite growth rates of magnetic fields prove to be possible even in the limit of perfect conductivity , that is , the model corresponds to a fast dynamo . in accordance with the theorem by bondi and gold the dynamo generated field
is completely confined in the fluid body . |
we are interested in this article in the large time behavior of solutions of first - order hamilton - jacobi equations , set in a bounded domain with nonlinear neumann boundary conditions , including the case of dynamical boundary conditions .
the main originality of this paper is twofold : on one hand , we obtain results for these nonlinear neumann type problems in their full generality , with minimal assumptions ( at least we think so ) and , on the other hand , we provide two types of proofs following the two classical approaches for these asymptotic problems : the first one by the pde methods which has the advantages of allowing to treat cases when the hamiltonians are non - convex , the second one by an optimal control / dynamical system approach which gives a little bit more precise description of the involved phenomena .
for cauchy - neumann problems with linear neumann boundary conditions , the asymptotic behavior has been established very recently and independently by the second author in @xcite by using the dynamical approach and the first and third authors in @xcite by using the pde approach . in order to be more specific ,
we introduce the following initial - boundary value problems u_t+h(x , du)= 0 & in @xmath0 , + b(x , du)=0 & on @xmath1 , + u(x,0)=u_0(x ) & on @xmath2 and u_t+h(x , du)= 0 & in @xmath0 , + u_t+b(x , du)=0 & on @xmath1 , + u(x,0)=u_0(x ) & on @xmath2 , where @xmath3 is a bounded domain of @xmath4 with a @xmath5-boundary and @xmath6 is a real - valued unknown function on @xmath7 .
we , respectively , denote by @xmath8 and @xmath9 its time derivative and gradient with respect to the space variable .
the functions @xmath10 are given real - valued continuous function on @xmath11 ; more precise assumptions on @xmath12 and @xmath13 will be given at the beginning of section [ preliminary ] . throughout this article , we are going to treat these problems by using the theory of viscosity solutions and thus the term viscosity " will be omitted henceforth .
we also point out that the boundary conditions have to be understood in the viscosity sense : we refer the reader to the `` user s guide to viscosity solutions '' @xcite for a precise definition which is not recalled here .
the existence and uniqueness of solutions of ( cn ) or ( dbc ) are already well known .
we refer to the articles @xcite and the references therein .
the standard asymptotic behavior , as @xmath14 , for solutions of hamilton - jacobi equations is the following : the solution @xmath15 is expected to look like @xmath16 where the constant @xmath17 and the function @xmath18 are solutions of an _ additive eigenvalue _ or _ ergodic problem_. in our case , we have two different ergodic problems for ( cn ) and ( dbc ) : indeed , looking for a solution of the form @xmath19 for ( cn ) , where @xmath20 is constant and @xmath21 a function defined on @xmath2 , leads to the equation ( e1 ) h(x , dw(x))= a & in @xmath3 , + b(x , dw(x))=0 & on @xmath22 while , for ( dbc ) , the function @xmath21 has to satisfy ( e2 ) h(x , dw(x))= a & in @xmath3 , + b(x , dw(x))= a & on @xmath22 .
we point out that one seeks , here , for a pair @xmath23 where @xmath24 and @xmath25 such that @xmath21 is a solution of ( e1 ) or ( e2 ) . if @xmath23 is such a pair , we call @xmath21 an _ additive eigenfunction _ or _ ergodic function _ and @xmath20 an _ additive eigenvalue _ or _ ergodic constant_. a typical result , which was first proved for hamilton - jacobi equations set in @xmath4 in the periodic case by p .-
lions , g. papanicolaou and s. r. s. varadhan @xcite , is that there exists a unique constant @xmath26 for which this problem has a _ bounded _
solution , while the associated solution @xmath21 may not be unique , even up to an additive constant .
this non - uniqueness feature is a key difficulty in the study of the asymptotic behavior .
the main results of this article are the following : under suitable ( and rather general ) assumptions on @xmath12 and @xmath13 + ( i ) there exists a unique constant @xmath17 such that ( e1 ) ( resp . , ( e2 ) ) has a solution in @xmath27 .
+ ( ii ) if @xmath6 is a solution of ( cn ) ( resp . , ( dbc ) ) , then there exists a solution @xmath28 of ( e1 ) ( resp . , ( e2 ) ) , such that @xmath29 the rest of this paper consists in making these claims more precise by providing the correct assumptions on @xmath12 and @xmath13 , by recalling the main existence and uniqueness results on ( cn ) and ( dbc ) , by solving ( e1 ) and ( e2 ) , and proving ( i ) and finally by showing the asymptotic result ( ii ) . in an attempt to make the paper concise ,
we have decided to present the full proof of ( ii ) for ( cn ) only by the optimal control / dynamical system approach while we prove the ( dbc ) result only by the pde approach . to our point of view , these proofs are the most relevant one , the two other proofs following along the same lines and being even simpler . in the last decade , the large time behavior of solutions of hamilton - jacobi equation in compact manifold @xmath30 ( or in @xmath31 , mainly in the periodic case ) has received much attention and general convergence results for solutions have been established .
g. namah and j .- m .
roquejoffre in @xcite are the first to prove under the following additional assumption @xmath32 where @xmath30 is a smooth compact @xmath33-dimensional manifold without boundary .
then a. fathi in @xcite proved the same type of convergence result by dynamical systems type arguments introducing the `` weak kam theory '' .
contrarily to @xcite , the results of @xcite use strict convexity ( and smoothness ) assumptions on @xmath34 , i.e. , @xmath35 for all @xmath36 and @xmath37 ( and also far more regularity ) but do not require . afterwards
j .- m . roquejoffre @xcite and a. davini and a. siconolfi in @xcite refined the approach of a. fathi and they studied the asymptotic problem for hamilton - jacobi equations on @xmath30 or @xmath33-dimensional torus . the second author ,
y. fujita , n. ichihara and p. loreti have investigated the asymptotic problem specially in the whole domain @xmath31 without the periodic assumptions in various situations by using the dynamical approach which is inspired by the weak kam theory .
see @xcite . the first author and p. e. souganidis obtained in @xcite more general results , for possibly non - convex hamiltonians , by using an approach based on partial differential equations methods and viscosity solutions , which was not using in a crucial way the explicit formulas of representation of the solutions
later , by using partially the ideas of @xcite but also of @xcite , results on the asymptotic problem for unbounded solutions were provided in @xcite .
there also exists results on the asymptotic behavior of solutions of convex hamilton - jacobi equation with boundary conditions .
the third author @xcite studied the case of the state constraint boundary condition and then the dirichlet boundary conditions @xcite .
roquejoffre in @xcite was also dealing with solutions of the cauchy - dirichlet problem which satisfy the dirichlet boundary condition pointwise ( in the classical sense ) : this is a key difference with the results of @xcite where the solutions were satisfying the dirichlet boundary condition in a generalized ( viscosity solutions ) sense .
these results were slightly extended in @xcite by using an extension of pde approach of @xcite .
we also refer to the articles @xcite for the large time behavior of solutions to time - dependent hamilton - jacobi equations .
recently e. yokoyama , y. giga and p. rybka in @xcite and the third author with y. giga and q. liu in @xcite has gotten the large time behavior of solutions of hamilton - jacobi equations with noncoercive hamiltonian which is motivated by a model describing growing faceted crystals .
we refer to the article @xcite for the large - time asymptotics of solutions of nonlinear neumann - type problems for viscous hamilton - jacobi equations .
this paper is organized as follows : in section [ preliminary ] we state the precise assumptions on @xmath12 and @xmath13 , as well as some preliminary results on ( cn ) , ( dbc ) , ( e1 ) and ( e2 ) .
section [ pde ] is devoted to the proof of convergence results by the pde approach . in section [ dsa ]
we devote ourselves to the proof of convergence results by the optimal control / dynamical system approach .
then we need to give the variational formulas for solutions of ( cn ) and ( dbc ) and results which are related to the weak kam theory , which are new and interesting themselves .
in appendix we give the technical lemma which is used in section [ pde ] and the proofs of basic results which are presented in section [ preliminary ] . before closing the introduction
, we give a few comments about our notation .
we write @xmath38 for @xmath39 , @xmath40 and @xmath41 . for @xmath42 for @xmath43
we denote by @xmath44 , @xmath45 , @xmath46 , @xmath47 the space of real - valued continuous , lower semicontinuous , upper semicontinuous , lipschitz continuous and on @xmath48 with values in @xmath13 , respectively . for @xmath49
we denote by @xmath50 and @xmath51 the set of all measurable functions whose absolute value raised to the @xmath52-th power has finite integral and which are bounded almost everywhere on @xmath48 with values in @xmath13 , respectively .
we write @xmath53 for the sets of @xmath54-th continuous differentiable functions for @xmath55 . for
given @xmath56 and @xmath57 , we use the symbol @xmath58,b)$ ] to denote the set of absolutely continuous functions on @xmath59 $ ] with values in @xmath13 .
we call a function @xmath60 a modulus if it is continuous and nondecreasing on @xmath61 and vanishes at the origin .
in this section , we introduce the key assumptions on @xmath62 and we present basic pde results on ( cn ) and ( dbc ) ( existence , comparison , ... , etc . ) which will be used throughout this article . the proofs
are given in the appendix .
we use the following assumptions .
* @xmath3 is a bounded domain of @xmath4 with a @xmath5-boundary . in the sequel ,
we denote by @xmath63 a @xmath64-defining function for @xmath3 , i.e. a @xmath64-function which is negative in @xmath3 , positive in the complementary of @xmath2 and which satisfies @xmath65 on @xmath22 .
such a function exists because of the regularity of @xmath66 if @xmath67 , we have @xmath68 where @xmath69 is the unit outward normal vector to @xmath22 at @xmath70 in order to simplify the presentation and notations , we will use below the notation @xmath71 for @xmath72 , even if @xmath73 is not on @xmath22 .
of course , if @xmath74 , @xmath75 is still an outward normal vector to @xmath22 at @xmath73 , by assumption @xmath75 does not vanish on @xmath22 but it is not anymore a unit vector .
* the function @xmath12 is continuous and coercive , i.e. , @xmath76 , there exists a constant @xmath77 such that @xmath78 for all @xmath79 and @xmath80 .
* there exists @xmath81 such that @xmath82 for all @xmath83 , @xmath84 and @xmath85 with @xmath86 .
* there exists a constant @xmath87 such that @xmath88 for any @xmath83 and @xmath89 .
* the function @xmath90 is convex for any @xmath83 .
we briefly comment these assumptions .
assumption ( a1 ) is classical when considering the large time behavior of solutions of hamilton - jacobi equations since it is crucial to solve ergodic problems .
assumption ( a2 ) is a non - restrictive technical assumption while ( a3)-(a4 ) are ( almost ) the definition of a nonlinear neumann boundary condition .
finally the convexity assumption ( a5 ) on @xmath13 will be necessary to obtain the convergence result .
we point out that the requirements on the dependence of @xmath12 and @xmath13 in @xmath73 are rather weak : this is a consequence of the fact that , because of ( a1 ) , we will deal ( essentially ) with lipschitz continuous solutions ( up to a regularization of the subsolution by sup - convolution in time .
therefore the assumptions are weaker than in the classical results ( cf .
@xcite ) .
a typical example for @xmath13 is the boundary condition arising in the optimal control of processes with reflection which has control parameters : @xmath91 where @xmath92 is a compact metric space , @xmath93 are given continuous functions and @xmath94 is a continuous vector field which is oblique to @xmath22 , i.e. , @xmath95 for any @xmath83 and @xmath96 .
our first result is a comparison result .
[ thm : comparison ] let @xmath97 and @xmath98 be a subsolution and a supersolution of ( cn ) ( resp .
, ( dbc ) ) , respectively .
if @xmath99 on @xmath2 , then @xmath100 on @xmath7 . then , applying carefully perron s method ( cf .
@xcite ) , we have the existence of lipschitz continuous solutions .
[ thm : existence ] for any @xmath101 , there exists a unique solution @xmath102 of ( cn ) or ( dbc ) .
moreover , if @xmath103 , then @xmath6 is lipschitz continuous on @xmath7 and therefore @xmath104 and @xmath105 are uniformly bounded . finally , if @xmath6 and @xmath18 are the solutions which are respectively associated to @xmath106 and @xmath107 , then @xmath108 finally we consider the additive eigenvalue / ergodic problems .
[ thm : additive]there exists a solution @xmath109 of ( e1 ) ( resp . , ( e2 ) ) . moreover , the additive eigenvalue is unique and is represented by @xmath110 the following proposition shows that , taking into account the ergodic effect , we obtain bounded solutions of ( cn ) or ( dbc ) . this result is a straightforward consequence of theorems [ thm : additive ] and [ thm : comparison ] . [
prop : bound ] let @xmath17 be the additive eigenvalue for ( e1 ) ( resp . , ( e2 ) ) .
let @xmath6 be the solution of ( cn ) ( resp .
, ( dbc ) )
. then @xmath111 is bounded on @xmath7 . from now on , replacing @xmath15 by @xmath112 , we can normalize the additive eigenvalue @xmath17 to be @xmath113 . as a consequence @xmath12
is also replaced by @xmath114 and @xmath13 by @xmath115 in the ( dbc)-case . in order to obtain the convergence result
, we use the following assumptions . * either of the following assumption ( a6)@xmath116 or ( a6)@xmath117 holds .
* there exists @xmath118 such that , for any @xmath119 $ ] , there exists @xmath120 such that if @xmath121 and @xmath122 for some @xmath79 and @xmath123 , then for any @xmath124 $ ] , @xmath125 * there exists @xmath118 such that , for any @xmath119 $ ] , there exists @xmath126 such that if @xmath127 and @xmath122 for some @xmath79 and @xmath128 , then for any @xmath129 , @xmath130 in the optimal control / dynamical system approach , the following assumptions are used * the function @xmath12 is convex , i.e. , for each @xmath131 the function @xmath132 is convex on @xmath31 and either of the following assumption ( a7)@xmath116 or ( a7)@xmath117 holds .
* there exists a modulus @xmath133 satisfying @xmath134 for @xmath40 such that for any @xmath135 , if @xmath136 , @xmath137 and @xmath138 , then @xmath139 we point out that we use the notation @xmath140 for the convex subdifferential of the function @xmath132 where @xmath73 is fixed .
our main result is the following theorem .
[ thm : large - time ] assume ( a0)-(a6 ) or ( a0)-(a5 ) and ( a7 ) .
for any @xmath101 , if @xmath6 is the solution of ( cn ) ( resp . , ( dbc ) ) associated to @xmath106 , then there exists a solution @xmath141 of ( e1 ) ( resp . , ( e2 ) ) , such that @xmath142 \(i ) under the convexity assumption on @xmath12 of ( a7 ) , the assumptions ( a1)(a6 ) are equivalent to ( a1)(a5 ) and ( a7 ) , and therefore ( a1)(a5 ) and ( a7 ) imply ( a1)(a6 ) .
indeed , under ( a1 ) and the convexity assumption on @xmath12 , conditions ( a7)@xmath143 are equivalent to ( a6)@xmath143 , respectively , this equivalence in the plus case has been proved in ( * ? ? ?
* appendix c ) .
the proof in the munis case is similar to that in the plus case , which we leave to the interested reader .
( ii ) we notice that if @xmath12 is smooth with respect to the @xmath52-variable , then ( a6 ) is equivalent to a _
one - sided directionally strict convexity _ in a neighborhood of the level set @xmath144 for all @xmath145 , i.e. , * _ there exists @xmath118 such that , for any @xmath119 $ ] , there exists @xmath126 such that if @xmath121 and @xmath122 ( or if @xmath127 and @xmath122 ) for some @xmath146 and @xmath128 , then for any @xmath124 $ ] , @xmath147 _
\(iii ) let us take the hamiltonian @xmath148 for instance .
if we consider the homogeneous neumann condition , then we can easily see that the additive eigenvalue is @xmath149 .
this hamiltonian is not convex but satisfies ( a6 ) ( and ( a6 ) ) .
as we mentioned it in the introduction , we provide the proof of theorem [ thm : large - time ] only in the case of the nonlinear dynamical - type boundary value problem ( dbc ) . in the case of the nonlinear neumann - type boundary condition ,
the proof is simpler and we will only give a remark at the end of this section . in order to avoid technical difficulties , we assume that @xmath106 is lipschitz continuous ( and therefore the solution @xmath6 of ( dbc ) is lipschitz continuous on @xmath7 )
. we can easily remove it by using . as in @xcite
the _ asymptotic monotonicity _ of solutions of ( dbc ) is a key property to get convergence .
[ thm : asymp - mono ] + ( i ) * ( asymptotically increasing property ) * assume that ( a6)@xmath116 holds . for any @xmath119 $ ] , there exists @xmath150 such that @xmath151 and @xmath152 for all @xmath79 , @xmath153 with @xmath154 .
+ ( ii ) * ( asymptotically decreasing property ) * assume that ( a6)@xmath117 holds .
for any @xmath119 $ ] , there exists @xmath150 such that @xmath151 and @xmath155 for all @xmath79 , @xmath153 with @xmath154 .
the * proof of theorem [ thm : large - time ] * follows as in @xcite : we reproduce these arguments for the convenience of the reader .
since @xmath156 is compact in @xmath157 , there exists a sequence @xmath158 which converges uniformly on @xmath2 as @xmath159 .
theorem [ thm : comparison ] implies that we have @xmath160 for any @xmath161 .
therefore , @xmath162 is a cauchy sequence in @xmath163 and it converges to a function denoted by @xmath164 . fix any @xmath79 and @xmath153 with @xmath154 . by theorem [ thm : asymp - mono ]
we have @xmath165 or @xmath166 for any @xmath167 and @xmath168 . sending @xmath159 and then @xmath169 , we get , for any @xmath154 @xmath170 or @xmath171 therefore , we see that the functions @xmath172 are uniformly bounded and equi - continuous , and they are also monotone in @xmath173 .
this implies that @xmath174 uniformly on @xmath2 as @xmath175 for some @xmath176 .
moreover , by a standard stability property of viscosity solutions , @xmath21 is a solution of ( dbc ) . since @xmath177 uniformly in @xmath7 as @xmath159 , we have @xmath178 where @xmath179 as @xmath159 , uniformly in @xmath73 and @xmath173 . taking the half - relaxed semi - limits as @xmath14 , we get @xmath180 sending @xmath159 yields @xmath181 for all @xmath79 . and the proof of theorem [ thm : large - time ] is complete .
now we turn to the * proof of theorem [ thm : asymp - mono]*. noticing that the additive eigenvalue is @xmath113 again , by proposition [ prop : bound ] the solution @xmath6 of ( dbc ) is bounded on @xmath7 .
we consider any solution @xmath18 of ( e2 ) .
we notice that @xmath182 is still a solution of ( e2 ) for any constant @xmath183 . therefore subtracting a positive constant to @xmath18 if necessary
, we may assume that @xmath184 and we fix such a constant @xmath185 .
we define the functions @xmath186 by @xmath187 for @xmath119 $ ] . by the uniform continuity of @xmath6 and @xmath18
, we have @xmath188 .
it is easily seen that @xmath189 and @xmath190 for all @xmath191 and @xmath119 $ ] .
[ lem : main - lem ] ( i ) assume that ( a6)@xmath116 holds .
the function @xmath192 is a supersolution of @xmath193 for any @xmath194 $ ] and some @xmath183 , where @xmath195 and @xmath196 ( ii ) assume that ( a6)@xmath117 holds .
the function @xmath197 is a subsolution of @xmath198 for any @xmath194 $ ] and some @xmath183 . before proving lemma [
lem : main - lem ] we notice that lemma [ lem : main - lem ] implies @xmath199 as @xmath200 .
indeed noting that @xmath201 is nondecreasing and the function @xmath202 is increasing for @xmath40 , we see that the comparison principle holds for both of neumann problems @xmath203 and @xmath204 moreover we have @xmath205 , @xmath149 are solutions of and @xmath206 , @xmath149 are a subsolution and a solution of , respectively .
therefore from these observations we see @xmath207 and @xmath208 , which imply the conclusion .
we only prove ( i ) , since we can prove ( ii ) similarly .
fix @xmath119 $ ] and let @xmath192 be the function given by .
we recall that @xmath209 for any @xmath210 , @xmath211 .
let @xmath212 and @xmath213 be a strict local minimum of @xmath214 , i.e. , @xmath215 for all @xmath216\setminus\{(\xi,{\sigma})\}$ ] and some small @xmath40 . since we can get the conclusion by the same argument as in @xcite in the case where @xmath217 , we only consider the case where @xmath218 in this proof .
moreover since there is nothing to check in the case where @xmath219 or @xmath220 , we assume that @xmath221 we choose @xmath222 such that @xmath223 we write @xmath224 for @xmath225 henceforth .
next , for @xmath37 small enough , we consider the function @xmath226 where @xmath227 is the function which is defined just after ( a0 ) .
we notice that , for @xmath228 , @xmath229 is a strict minimum point of this function .
this implies that , for @xmath37 small enough , this function achieves its minimum over @xmath230\}$ ] at some point @xmath231 which converges to @xmath229 when @xmath232 .
then there are two cases : either ( i ) @xmath233 or ( ii ) @xmath234 .
we only consider case ( ii ) here too since , again , the conclusion follows by the same argument as in @xcite in case ( i ) . in case ( ii ) , since @xmath235 , the @xmath236-term vanishes and we have @xmath237 by the strict minimum point property . for
any @xmath238 let @xmath239 and @xmath240 be , respectively , the functions given in lemma [ lem : func - c ] with @xmath241 and @xmath242 , and let @xmath243 and @xmath244 be , respectively , the functions given in lemma [ lem : coercivity ] with @xmath245 and @xmath246 for @xmath247 .
we set @xmath248\}$ ] .
we define the function @xmath249 by @xmath250 in view of lemma [ lem : coercivity ] , if @xmath251 , where @xmath48 is the constant in @xmath252 , then @xmath253 achieves its minimum over @xmath254 at some point @xmath255 which depends on @xmath256 . by taking a subsequence
if necessary we may assume that @xmath257 set @xmath258 and then we have @xmath259 therefore we may assume that @xmath260 for small @xmath247 .
* claim : * there exists a constant @xmath261 such that @xmath262 for all @xmath263 .
we only consider the estimate of @xmath264 , since we can obtain the estimate of @xmath265 similarly .
the inequality @xmath266 implies @xmath267 for some @xmath268 . combining this and the inequality in lemma [ lem : coercivity ] ( i ) we get the conclusion of claim for @xmath269 . in the sequel ,
we denote by @xmath270 a quantity which tends to @xmath113 as @xmath271 .
derivating @xmath253 with respect to each variable @xmath272 at @xmath273 formally , we have @xmath274 we remark that we should interpret @xmath275 , and @xmath276 as the viscosity solution sense here .
we first consider the case where @xmath277 . in view of claim , ( a3)(a5 ) and
lemma [ lem : coercivity ] ( ii ) we have @xmath278 where @xmath279 is a modulus .
therefore @xmath280 for @xmath281 which are small enough compared to @xmath37 .
similarly if @xmath282 then we have @xmath283 we next consider the case where @xmath284 .
note that we have @xmath285 for any @xmath83 , @xmath286 , where @xmath287 and @xmath254 is the function defined by . by lemma [ lem : coercivity ] and the homogenity with degree @xmath149 of @xmath288 with respect to the @xmath52-variable , we have @xmath289 again , for @xmath281 which are small enough compared to @xmath37 , we have @xmath290 therefore , by the definition of viscosity solutions we have @xmath291 in view of the above claim by taking a subsequence if necessary we may assume that @xmath292 as @xmath271 for some @xmath293 .
sending @xmath294 , @xmath295 and then @xmath296 in , we obtain @xmath297 we use these three inequality in the following way : first , using ( a2 ) , the second one leads to @xmath298 for some constant @xmath183 .
it remains to estimate @xmath299 .
set @xmath300 . by ( a6)@xmath116 we obtain @xmath301 for some @xmath126 .
we therefore have @xmath302 using this estimate in our first inequality yields the desired result , namely @xmath303 we remark that the solution of ( cn ) has the asymptotic monotonicity property . in order to prove this
, we mainly use the following lemma in place of lemma [ lem : main - lem ] . set @xmath304 for @xmath119 $ ] .
+ ( i ) assume that ( a6)@xmath116 holds .
the function @xmath192 is a supersolution of @xmath305 for any @xmath119 $ ] .
+ ( ii ) assume that ( a6)@xmath117 holds .
the function @xmath197 is a subsolution of @xmath306 for any @xmath119 $ ] .
as we mentioned in the introduction , we mainly concentrate on problem ( cn ) in this section .
we begin this section with an introduction to the skorokhod problem .
let @xmath307 and set @xmath308 by the convex duality , we have @xmath309 note that @xmath310 we set @xmath311 and observe that @xmath312 , @xmath313 for @xmath314 , @xmath315 is a convex subset of @xmath316 and @xmath317 is a closed convex subset of @xmath31 .
observe as well that if @xmath318 belongs to @xmath315 for some @xmath307 , then @xmath319 and hence @xmath314 .
for example , if @xmath320 for some functions @xmath321 , then @xmath322 \infty&\text { otherwise , } \end{cases}\ ] ] and @xmath323 let @xmath324 and @xmath325 , and let @xmath326,{\mathbb{r}}^n)$ ] , @xmath327 , { \mathbb{r}}^n)$ ] and @xmath328,{\mathbb{r}})$ ] .
we introduce a set of conditions : @xmath329 , \\ & l(t)\geq 0 \ \
\ \text { for a.e . } t\in[0,\,t ] , \\ & l(t)=0 \ \
\text { if } \eta(t)\in{\omega}\ \ \ \text { for a.e . } t\in[0,\,t ] , \end{aligned } \right.\ ] ] and @xmath330,\,{\mathbb{r}})$ such that } \lab{i:2 - 5}\ ] ] @xmath331.\ ] ] observe here that the inclusion @xmath332 is equivalent to the condition that @xmath333 if @xmath334 , and @xmath335 and @xmath336 if @xmath337 .
condition is therefore equivalent to the condition that @xmath338,\lab{i:2 - 6 } \\ & \text{and } \
\dot\eta(t)=v(t ) \
\text { if } \ l(t)=0 \
\text { for a.e . } t\in[0,\,t ] .
\notag\end{aligned}\ ] ] here we have used the fact that @xmath339 is lower semi - continuous ( hence borel ) function bounded from below by the constant @xmath340 .
the expression @xmath341 in is actually defined only for those @xmath342 $ ] such that @xmath334 , but we understand that @xmath343 0 & \text { otherwise . } \end{cases}\ ] ] similarly we henceforth use the convention that zero times an undefined quantity equals zero . with use of this convention
, we define the function @xmath344 on @xmath345 $ ] by @xmath346 we remark that under assumption we have for a.e .
@xmath347 $ ] , @xmath348 in the case where @xmath320 for some @xmath349 , it is easily seen that condition is equivalent to @xmath350.\ ] ] ( compare this together with with ( 1.4 ) in @xcite . ) in this case we have @xmath351 for a.e .
@xmath347 $ ] .
now , given a point @xmath352 , a constant @xmath325 and a function @xmath327,{\mathbb{r}}^n)$ ] , the _ skorokhod problem _ is to find a pair of functions @xmath353,\,{\mathbb{r}}^n)$ ] and @xmath354,\,{\mathbb{r}})$ ] for which and are satisfied .
[ i : skor]let @xmath352 , @xmath325 and @xmath355,\,{\mathbb{r}}^n)$ ] .
there exists a solution @xmath356 of the skorokhod problem .
moreover , there exists a constant @xmath357 , independent of @xmath73 , @xmath358 and @xmath18 , such that , for any solution @xmath356 of the skorokhod problem , the inequalities @xmath359 and @xmath360 hold for a.e .
@xmath347 $ ] . for fixed @xmath352 and @xmath325
, @xmath361 denotes the set of the all triples @xmath362 of functions
@xmath353,\,{\mathbb{r}}^n)$ ] , @xmath355,{\mathbb{r}}^n)$ ] and @xmath354,\,r)$ ] such that conditions , are satisfied , and @xmath363 denotes the set of the triples @xmath362 of functions @xmath364 , @xmath18 and @xmath365 on @xmath366 such that , for all @xmath325 , the restriction of @xmath362 to the interval @xmath345 $ ] belongs to @xmath361 .
note here that if @xmath367 for some @xmath352 and @xmath325 and if we extend the domain of @xmath362 to @xmath61 by setting @xmath368 then the extended @xmath362 belongs to @xmath363 .
moreover we set @xmath369 let @xmath370 .
for @xmath371 we set @xmath372 where @xmath373 and @xmath374 .
we call the function @xmath375 the hamiltonian of @xmath12 . this function has the properties : @xmath376 is lower semicontinuous on @xmath377 , convex in @xmath378 and coercive , i.e. , @xmath379 for all @xmath352 .
the function @xmath375 may take the value @xmath380 , but @xmath381 for some constant @xmath40 .
[ i : visco - en ] the function @xmath382 is continuous on @xmath383 and a solution of @xmath384 moreover , we have @xmath385 we set @xmath386 for @xmath352 .
the above theorem ensures that @xmath387 and @xmath382 is a solution of ( cn ) . in what follows
we give an outline of proofs of theorems [ i : skor ] and [ i : visco - en ] .
indeed , most of the arguments parallel to those for similar assertions in @xcite for ( cn ) with linear neumann boundary condition .
[ i : selec ] let @xmath388 and @xmath247 .
there exists @xmath389 such that for all @xmath307 , @xmath390 and @xmath391
. let @xmath392 and set @xmath393 note that @xmath394 on @xmath395 , that , as @xmath396 , @xmath397 uniformly on @xmath398 for every @xmath399 and that for @xmath307 , the function @xmath400 is in @xmath401 . also , by ( a4 ) we see that @xmath402 for some @xmath399 depending only on @xmath403 and @xmath404 .
hence , @xmath405 . moreover , it is easy to see that @xmath406 . indeed ,
if @xmath407 as @xmath408 and @xmath409 , then @xmath410 noting that @xmath411 , we may choose a subsequence @xmath412 , converging to a point @xmath364 , of @xmath413 .
from the above inequality with @xmath414 , we get in the limit @xmath415 this shows that @xmath416 , which implies that @xmath417 and @xmath418 .
if we set @xmath419 and @xmath420 , then we have for all @xmath421 , @xmath422 thus , we find that @xmath423 for all @xmath307 . moreover , for each fixed @xmath247 , if @xmath392 is small enough , then we have @xmath424 finally , we note that @xmath389 and conclude the proof .
[ i : bound - by - v ] let @xmath325 .
there is a constant @xmath357 , depending only on @xmath425 and @xmath426 , such that for any @xmath427 , @xmath428.\ ] ] recall that @xmath429 is a lipschitz bound of the functions @xmath90 @xmath307 . an immediate consequence of the above lemma is that for @xmath430 , if @xmath431 is an interval of @xmath61 and @xmath432 , with @xmath433 , then @xmath434 .
let @xmath435 and set @xmath436 on @xmath366 .
we choose a function @xmath437,{\mathbb{r}})$ ] so that @xmath438 for a.e .
@xmath439 $ ] .
if @xmath440 for a.e . @xmath439 $ ] , then we have @xmath441 for a.e .
@xmath439 $ ] , which yields @xmath442 for a.e .
@xmath439 $ ] , and we are done .
henceforth we assume that the set @xmath443\mid l(s)>0\}$ ] has positive measure .
we choose a subset @xmath444 of @xmath445 having full measure so that @xmath446 , that @xmath447 and @xmath438 for all @xmath448 and that @xmath364 is differentiable everywhere in @xmath444 .
we set @xmath449 for @xmath448 , and note that @xmath450 for all @xmath448 . using the defining function @xmath227 ( cf .
( a0 ) ) and noting that @xmath451 for all @xmath452 $ ] , we find that for any @xmath448 , @xmath453 that is , @xmath454 for all @xmath448 . fix any @xmath448 . since @xmath450 , we have @xmath455 and @xmath456 .
accordingly , we get @xmath457 and hence , @xmath458 .
finally , we note that latexmath:[$ completes the proof .
fix @xmath352 , @xmath325 and @xmath355,{\mathbb{r}}^n)$ ] . due to lemma
[ i : selec ] , there exists a @xmath460 such that @xmath390 for all @xmath461 . according to ( * ? ? ?
* theorem 4.1 ) , there exists a pair @xmath462,{\mathbb{r}}^n){\times}l^1([0,\,t],{\mathbb{r}})$ ] such that @xmath463 , @xmath464 for all @xmath439 $ ] and for a.e .
@xmath439 $ ] , @xmath465 we set @xmath466 for @xmath439 $ ] , and observe that we have for a.e .
@xmath439 $ ] , @xmath467 completing the existence part of the proof .
the remaining part of the proof is exactly what lemma [ i : bound - by - v ] guarantees .
set @xmath468 .
we first prove that @xmath382 is a subsolution of , .
let @xmath469 and @xmath470 .
assume that @xmath471 attains a strict maximum at @xmath472 .
we need to show that if @xmath473 , then @xmath474 , and if @xmath475 , then either @xmath476 we are here concerned only with the case where @xmath475 . the other case can be treated similarly .
to prove , we argue by contradiction .
thus we suppose that ( [ i:2 - 11 ] ) were false .
we may choose an @xmath477 so that @xmath478 and @xmath479 .
by lemma [ i : selec ] , we may choose @xmath389 so that @xmath390 for all @xmath307 and @xmath480 .
note that @xmath481 .
set @xmath482 $ ] . by replacing @xmath247
if needed , we may assume that @xmath483 and for all @xmath484 , @xmath485 where @xmath486 and @xmath487 are assumed to be defined and continuous on @xmath488 .
we may assume that @xmath489 .
set @xmath490 , and note that @xmath491 and @xmath492 for @xmath493 .
we choose a point @xmath494)\cap q$ ] so that @xmath495 .
now , we consider the skorokhod problem with the function @xmath496 in place of @xmath497 . for the moment
we denote by @xmath498 the set of all @xmath499,{\mathbb{r}}^n){\times}l^1([0,\,t],{\mathbb{r}}^n){\times}l^1([0,\,t],{\mathbb{r}})$ ] satisfying and , with the function @xmath496 in place of @xmath497 .
we apply ( * ? ? ?
* lemma 5.5 ) , to find a triple @xmath500 such that for a.e .
@xmath501 , @xmath502 note here that , since @xmath503 , we have @xmath504 and @xmath505 for a.e
. @xmath506 $ ] .
we set @xmath507 and note that @xmath508 for all @xmath509 and @xmath510 . using the dynamic programming principle , we obtain @xmath511 hence , setting @xmath512 , we get @xmath513 using and , we obtain @xmath514 which is a contradiction .
thus , @xmath382 is a subsolution of , .
next , we turn to the proof of the supersolution property of @xmath382 .
let @xmath470 and @xmath469 .
assume that @xmath515 attains a strict minimum at @xmath472 .
we need to show that if @xmath473 , then @xmath516 , and if @xmath475 , then either @xmath517 as before , we only consider the case where @xmath475 . to prove , we suppose by contradiction that @xmath518 and @xmath519 .
there is a constant @xmath247 such that @xmath520 where @xmath521 $ ] .
here we may assume that @xmath522 and @xmath523 .
set @xmath524 .
we may choose a point @xmath525 so that @xmath526
. we select a triple @xmath527 so that @xmath528 where @xmath374 .
we set @xmath529 .
it is clear that @xmath530 and @xmath531 for all @xmath532 $ ] .
accordingly , we have @xmath533 and hence , @xmath534 note by the fenchel - young inequality and that for a.e .
@xmath532 $ ] , @xmath535 where @xmath536 .
consequently , in view of we get @xmath537 which is a contradiction .
the function @xmath382 is thus a supersolution of , .
it remains to show the continuity of @xmath382 on @xmath538 . in view of theorem [ thm : comparison ] , we need only to prove that @xmath539 indeed , once this is done , we see by theorem [ thm : comparison ] that @xmath540 on @xmath538 , which guarantees that @xmath541 . to show , fix any @xmath247 .
we may select a function @xmath542 such that @xmath543 for all @xmath307 and @xmath544 indeed , we can first approximate @xmath106 by a sequence of @xmath64 functions and then modify the normal derivative ( without modifying too much the function itself ) by adding a function of the form @xmath545 where @xmath546 is a @xmath64 , increasing function such that @xmath547 , @xmath548 and @xmath549 for all @xmath550 . then we may choose a constant @xmath551 so that the function @xmath552 is a ( classical ) subsolution of , .
then , for any @xmath553 and @xmath554 , we have @xmath555 setting @xmath556 and @xmath557 for @xmath558 $ ] and using the fenchel - young inequality , we observe that for a.e .
@xmath558 $ ] , @xmath559 combining these observations , we obtain @xmath560 which ensures that @xmath561 for all @xmath562 , and moreover , @xmath563 for all @xmath352 . next , fix any @xmath553 and set @xmath564 , @xmath565 and @xmath440 for @xmath211 . observe that @xmath554 and that @xmath566 and @xmath567 this shows that @xmath568 for all @xmath352 .
thus we find that is valid , which completes the proof .
next we present the variational formula for the solution of ( dbc ) .
the basic idea of obtaining this formula is similar to that for ( cn ) , and thus we just outline it or skip the details .
we define the function @xmath569 on @xmath570 by @xmath571 where the infimum is taken all over @xmath554 , @xmath572 , and @xmath573 $ ] is given by @xmath574
. then we extend the domain of definition of @xmath569 to @xmath575 by setting @xmath576 for @xmath352 .
in the definition of @xmath569 we apparently use the set @xmath577 ( the skorokhod problem for @xmath3 and @xmath13 ) , but the underlining idea is to consider the skorokhod problem for the domain @xmath578 and the function @xmath579 in place of @xmath3 and @xmath497 , respectively . indeed ,
setting @xmath580 , @xmath581 and @xmath582 , we observe that the vector @xmath583 is the unit outer normal at @xmath584 , the conditions ( a1)(a7 ) are satisfied with @xmath585 and @xmath586 in place of @xmath13 and @xmath3 and the lagrangian @xmath587 of @xmath588 is given by @xmath589 where @xmath590 is the indicator function of the set @xmath591 , i.e , @xmath592 if @xmath593 and @xmath594 if @xmath595 .
if we set for @xmath584 , @xmath596 then it is easily seen that @xmath597 .
the skorokhod problem for @xmath598 and @xmath585 is to find for given @xmath599 , @xmath600 and @xmath601,{\mathbb{r}}^{n+1})$ ] a pair of functions @xmath602,{\mathbb{r}}^{n+1})$ ] and @xmath354,{\mathbb{r}})$ ] such that @xmath603 , @xmath604 for @xmath439 $ ] , @xmath605 for a.e .
@xmath439 $ ] , @xmath440 if @xmath606 for a.e .
@xmath439 $ ] , and @xmath607 for a.e . @xmath439 $ ] and for some @xmath437,{\mathbb{r}})$ ] .
it is easily checked that for given @xmath599 , @xmath600 and @xmath601,{\mathbb{r}}^{n+1})$ ] , the pair of functions @xmath608,{\mathbb{r}}^{n+1})$ ] and @xmath354,{\mathbb{r}})$ ] is a solution of the skorokhod problem for @xmath598 and @xmath585 if and only if @xmath609 and @xmath610 for all @xmath439 $ ] .
if we take into account of the form , then we need to consider the skorokhod problem only with @xmath611 .
that is , in our minimization at @xmath553 , we have only to consider the infimum all over @xmath554 and @xmath612 such that @xmath613 for @xmath211 .
note that this function @xmath612 is decreasing on @xmath366 and that @xmath614 if and only if @xmath615 , which justifies the choice of @xmath616 in .
we have the following theorems concerning ( dbc ) .
[ i : visco - es ] the function @xmath569 is a solution of _ ( dbc ) _ and continuous on @xmath575 .
moreover , if @xmath617 , then @xmath618 . in the above theorem , the subsolution ( resp .
, supersolution ) property of ( dbc ) assumes as well the inequality @xmath619 ( resp .
, @xmath620 ) on @xmath621 .
we do not give here the proof of the above theorem , since one can easily adapt the proof of theorems [ i : visco - en ] , using theorems [ thm : comparison ] and [ thm : existence ] , with minor modifications .
a typical modification is the following : in the proof of the viscosity property of @xmath569 , we have to replace the cures @xmath622 , with @xmath211 , which are used in the proof of theorem [ i : visco - en ] , by the curves @xmath623 , with @xmath211 , where @xmath624 . a further remark on the modifications of the proof is the use of the following lemma in place of ( * ?
* lemma 5.5 ) .
[ i : selec-2 ] let @xmath371 , @xmath352 , @xmath625,{\mathbb{r}}^n)$ ] and @xmath247 .
then there is a triple @xmath554 such that for a.e .
@xmath626 , @xmath627 where @xmath628 and @xmath629 for @xmath630 .
the above lemma can be proved in a parallel fashion as in the proof of ( * ? ? ?
* lemma 5.5 ) , and we leave it to the reader to prove the lemma . in this section
we establish the existence of extremal curves ( or optimal controls ) @xmath631 for the variational formula .
we set @xmath468 .
[ i : exist - extremal - en ] let @xmath632 and let @xmath633 be the unique solution of _
( cn)_. let @xmath553
. then there exists a triple @xmath634 such that @xmath635 where @xmath572 .
moreover , @xmath636,{\mathbb{r}}^n)$ ] and @xmath637,{\mathbb{r}}^{n+2})$ ] .
fix @xmath638 . in view of formula
, we may choose a sequence @xmath639 such that for @xmath55 , @xmath640 where @xmath641 .
we show that the sequence @xmath642 is uniformly integrable on @xmath643 $ ] .
once this is done , due to lemma [ i : bound - by - v ] , the sequences @xmath644 and @xmath645 are also uniformly integrable on @xmath643 $ ] .
if we choose a constant @xmath646 so that @xmath647 , then @xmath648 for all @xmath649 and hence , @xmath650 for a.e .
@xmath558 $ ] .
due to lemma [ i : bound - by - v ] , there is a constant @xmath651 , independent of @xmath54 , such that @xmath652 for a.e .
@xmath558 $ ] , which implies that @xmath653 for a.e .
@xmath558 $ ] . since @xmath12 is coercive , for each @xmath654 there
exists a constant @xmath655 such that @xmath656 for all @xmath657 . combining these two estimates ,
we get for all @xmath654 , @xmath658 we fix any @xmath659 and measurable @xmath660 $ ] , and , using the above estimate with @xmath661 and @xmath662 , observe that @xmath663 and hence , @xmath664}|u|+1+c(c_1)t+c(2c_1+a)|e|,\end{aligned}\ ] ] where @xmath665 denotes the lebesgue measure of @xmath445 . from this , we easily deduce that @xmath642 is uniformly integrable on @xmath643 $ ] .
thus , the sequences @xmath644 , @xmath642 and @xmath645 are uniformly integrable on @xmath643 $ ] .
next , we show that @xmath666 is uniformly integrable on @xmath643 $ ] . to this end , we fix two finite sequences @xmath667 and @xmath668 so that @xmath669 set @xmath670 and @xmath671 . in view of the dynamic programming principle
, we have for @xmath672 , @xmath673 subtracting these from yields @xmath674 hence , if @xmath675 is a lipschitz bound of @xmath6 , then we get @xmath676 now , using with @xmath661 , we find that @xmath677 from which we infer that @xmath666 is uniformly integrable on @xmath643 $ ] .
we apply the dunford - pettis theorem to the sequence @xmath678 , to find an increasing sequence @xmath679 and functions @xmath680,{\mathbb{r}}^n)$ ] , @xmath681,{\mathbb{r}})$ ] such that , as @xmath682 , @xmath683 weakly in @xmath684,{\mathbb{r}}^{2n+2})$ ] . setting @xmath685 for @xmath558 $ ] , we have @xmath686 uniformly on @xmath643 $ ] as @xmath682 .
then , as in the last half of the proof of ( * ? ? ?
* lemma 7.1 ) , we infer that @xmath687 it is now obvious that @xmath688 now , we show that @xmath689 .
it is clear that @xmath464 for all @xmath558 $ ] and @xmath605 for a.e .
@xmath558 $ ] .
it is thus enough to show that @xmath690 for a.e .
@xmath558 $ ] .
setting @xmath691 and @xmath692 on @xmath643 $ ] , we have @xmath693,\ ] ] let @xmath694,{\mathbb{r}})$ ] satisfy @xmath695 for all @xmath558 $ ] .
we have @xmath696 sending @xmath697 along the subsequence @xmath698 , we find that @xmath699 this implies that @xmath700 for a.e .
@xmath558 $ ] , and conclude that @xmath701 .
next , we set @xmath702 for @xmath558 $ ] . since @xmath690 for a.e
@xmath558 $ ] , we see that @xmath703 for a.e . @xmath558 $ ] and @xmath704,{\mathbb{r}})$ ] . using and , we get @xmath705 therefore , we have @xmath706 for a.e .
@xmath558 $ ] and @xmath707 finally , we check the regularity of the triple @xmath689 and the function @xmath708 .
fix any interval @xmath709\subset[0,\,t]$ ] , and observe as in that @xmath710 here we may choose a constant @xmath711 so that @xmath712 for a.e .
@xmath558 $ ] . combining the above and , with @xmath713 in place of @xmath714 and @xmath715 , and setting @xmath716
, we get @xmath717 from which we conclude that @xmath718,{\mathbb{r}}^{n+1})$ ] as well as @xmath719,{\mathbb{r}}^{n+1})$ ] .
an immediate consequence of the previous theorem is the following .
[ i : exist - extremal - sn ] let @xmath720 be a solution of _ ( e1 ) _ , with @xmath721 .
let @xmath352 .
then there is a triple @xmath722 such that for any @xmath371 , @xmath723 where @xmath374 .
moreover , @xmath724 and @xmath725 .
note that the function @xmath726 is a solution of ( cn ) .
using theorem [ i : exist - extremal - en ] , we define inductively the sequence @xmath727 as follows .
we first choose a @xmath728 so that @xmath729 next , we assume that @xmath730 , with @xmath731 , is given , and choose a @xmath732 so that @xmath733 once the sequence @xmath734 is given , we define the @xmath722 and the function @xmath708 on @xmath366 by setting for @xmath735 and @xmath736 , @xmath737 it is clear that @xmath554 , @xmath572 and is satisfied . thanks to theorem [ i : exist - extremal - en ] , we have @xmath738,{\mathbb{r}}^n)$ ] and @xmath739,{\mathbb{r}}^{n+2})$ ] for @xmath740 .
moreover , in view of the proof of theorem [ i : exist - extremal - en ] , we see easily that @xmath741)}<\infty$ ] , from which we conclude that @xmath742 . throughout this section
we fix a subsolution @xmath743 of ( e1 ) with @xmath721 , @xmath325 and a lipschitz curve @xmath364 in @xmath488 , i.e. , @xmath744,{\mathbb{r}}^n)$ ] and @xmath745)\subset\bar{\omega}$ ] . henceforth in this section
we assume that there is a bounded , open neighborhood @xmath746 of @xmath621 for which @xmath12 , @xmath13 and @xmath33 are defined and continuous on @xmath747 , @xmath748 and @xmath749 , respectively .
moreover , we assume by replacing @xmath425 and @xmath404 in ( a3 ) , ( a4 ) respectively by other positive numbers if needed that ( a1 ) , with @xmath746 in place of @xmath3 , and ( a3)(a5 ) , with @xmath746 in place of @xmath621 , are satisfied .
( of course , these are not real additional assumptions . )
[ i : exist - p - n ] there exists a function @xmath750,{\mathbb{r}}^n)$ ] such that for a.e .
@xmath347 $ ] , @xmath751 , @xmath752 , and @xmath753 if @xmath754 . to prove the above theorem
, we use the following lemmas .
[ i : basic - p ] let @xmath755 , @xmath756 and @xmath757,{\mathbb{r}}^n)$ ] .
assume that @xmath758 uniformly on @xmath488 as @xmath271 and , for a.e .
@xmath347 $ ] , @xmath759 then there exists a function @xmath750,{\mathbb{r}}^n)$ ] such that for a.e . @xmath347 $ ] , @xmath760 , @xmath761 , and @xmath762 if @xmath763 .
observe first that for all @xmath347 $ ] , @xmath764 next , we observe by the coercivity of @xmath12 that @xmath765 is bounded in @xmath766,{\mathbb{r}}^n)$ ] , and then , in view of the banach - sack theorem , we may choose a sequence @xmath767 and a function @xmath750,{\mathbb{r}}^n)$ ] so that @xmath768 is in the closed convex hull of @xmath769 , @xmath770 strongly in @xmath771,{\mathbb{r}}^n)$ ] as @xmath408 and @xmath772 for a.e .
@xmath347 $ ] as @xmath408 . by and
the convexity of @xmath12 and @xmath13 , we see that , for a.e . @xmath347 $ ] , @xmath773 , and @xmath774 if @xmath763 .
moreover , we have , for all @xmath347 $ ] , @xmath775 now , by sending @xmath408 , we get for a.e
. @xmath347 $ ] , @xmath761 , and @xmath762 if @xmath763 , and , for all @xmath347 $ ] , @xmath776 .
the proof is complete .
[ i : localized - main]let @xmath777 and @xmath247 .
then there are an open neighborhood @xmath382 of @xmath778 in @xmath488 , a sequence @xmath779 of open neighborhoods of @xmath780 in @xmath746 and a sequence @xmath781 of @xmath64 functions on @xmath782 , such that for each @xmath783 the function @xmath784 satisfies @xmath785 and , as @xmath682 , @xmath786 uniformly on @xmath382 .
we now prove theorem [ i : exist - p - n ] by assuming lemma [ i : localized - main ] , the proof of which will be given after the proof of theorem [ i : exist - p - n ] .
in view of lemma [ i : basic - p ] , it is enough to show that for each @xmath247 there exists a function @xmath787,{\mathbb{r}}^n)$ ] such that for a.e .
@xmath347 $ ] , @xmath788 , @xmath789 , and @xmath790 if @xmath763 . to show this
, we fix any @xmath247 .
it is sufficient to prove that for each @xmath791 $ ] , there exist a neighborhood @xmath792 of @xmath612 , relative to @xmath345 $ ] , and a function @xmath793 such that for a.e .
@xmath794 , @xmath795 , @xmath796 , and @xmath797 if @xmath763 .
fix any @xmath798 $ ] .
consider first the case where @xmath799 .
there is a @xmath800 such that @xmath801 , where @xmath802\cap[0,\,t]$ ] .
we may choose an open neighborhood @xmath382 of @xmath778 such that @xmath803 . by the mollification technique , for any @xmath804
, we may choose a function @xmath805 such that @xmath806 and @xmath807 for all @xmath808 .
then , setting @xmath809 for @xmath794 and @xmath804 , we have @xmath810 and @xmath811 for a.e . @xmath794 and all @xmath804 .
hence , by lemma [ i : basic - p ] , we find that there is a function @xmath793 such that for a.e .
@xmath794 , @xmath812 and @xmath813 .
next consider the case where @xmath814 .
thanks to lemma [ i : localized - main ] , there are an open neighborhood @xmath382 of @xmath815 in @xmath488 , a sequence @xmath816 of open neighborhoods of @xmath780 in @xmath746 and a sequence @xmath817 of @xmath64 functions on @xmath818 such that for any @xmath819 , @xmath820 for all @xmath821 , @xmath822 for all @xmath823 and @xmath824 for all @xmath808 .
we now choose a constant @xmath800 so that if @xmath802\cap[0,\,t]$ ] , then @xmath825 . set @xmath826 for @xmath794 . then , for a.e .
@xmath347 $ ] , we have @xmath827 , @xmath828 , and @xmath829 if @xmath763 . lemma [ i : basic - p ] now ensures that there exists a function @xmath830 such that for a.e .
@xmath794 , @xmath831 , @xmath832 , and @xmath833 if @xmath763 .
the proof is now complete . for the proof of lemma [ i : localized - main ] , we need the following lemma .
[ i : subdiff ] let @xmath834 be a subsolution of _ ( e1 ) _ , with @xmath721 .
let @xmath777 and @xmath835 .
assume that @xmath836 for all @xmath371 .
then @xmath837 in particular , we have @xmath838 .
we choose a fucntion @xmath839 so that @xmath840 and the function @xmath841 attains a strict maximum at @xmath778 .
let @xmath842 be a function such that @xmath843 and @xmath844 for all @xmath307 .
for @xmath247 , let @xmath845 be a maximum point of the fucntion @xmath846 on @xmath488 .
it is obvious that @xmath847 as @xmath848 and @xmath849 .
suppose that @xmath850 .
then we have @xmath851 , which is impossible by the choice of @xmath52 .
that is , we have @xmath852 .
observe that for any @xmath307 , @xmath853 , which guarantees that @xmath854 .
thus we have @xmath855 , which implies that @xmath838 .
we fix any @xmath856 and @xmath777 .
since @xmath3 is a @xmath64 domain , we may assume after a change of variables if necessary that @xmath857 and for some constant @xmath40 , @xmath858 of course , we have @xmath859 for all @xmath860 .
now , we choose a constant @xmath675 so that for @xmath861 , if @xmath862 , then @xmath863 . we next choose a constant @xmath399 so that @xmath864 for all @xmath865 . replacing @xmath866 by @xmath867
if @xmath868 , we may assume that @xmath869 .
we now show that @xmath6 is a subsolution of @xmath870 to do this , we fix any @xmath871 and @xmath872 .
we need to consider only the case when @xmath307 .
we may assume that @xmath873 .
otherwise , we have nothing to prove .
we set @xmath874 .
note that @xmath875 and , therefore , @xmath876 . since the function @xmath877 is non - decreasing on @xmath878
, we see that @xmath879 for all @xmath173 large enough . also , it is obvious that @xmath880 for all @xmath173 large enough .
therefore , we see that @xmath881 if @xmath173 is large enough and conclude that @xmath882 .
since @xmath883 is a closed subset of @xmath31 , we see that @xmath884 . from the definition of @xmath612
, we observe that @xmath881 for @xmath885 .
we now invoke lemma [ i : subdiff ] , to find that @xmath886 .
we recall the standard observation that if @xmath887 , then @xmath888 for all @xmath889 .
hence , we must have either @xmath890 or @xmath891 for any @xmath892 . set @xmath893\mid h(x,\,p+t\tn(x))>0\}$ ] , and observe that @xmath894 and @xmath895 .
there is a sequence @xmath896 $ ] converging to @xmath616 such that @xmath897 , which implies that @xmath898 .
hence we have @xmath899 .
thus we have @xmath530 , @xmath895 and @xmath899 . by the choice of @xmath254
, we have @xmath900 , and hence @xmath901 . noting that @xmath902 and @xmath530 , we see by the monotonicity of @xmath903 that @xmath904 .
thus we find that @xmath6 is a subsolution of . following the arguments of lemma [
lem : coercivity ] , we can show that there exist a function @xmath905 and a constant @xmath800 such that for all @xmath378 , @xmath906 and @xmath907 < { \varepsilon } & \text { if } \
\xi_n\leq { \delta}. \end{cases}\ ] ] we may also assume that @xmath908 and all the derivatives of @xmath546 are bounded on @xmath31 .
we introduce the sup - convolution of @xmath6 as follows : @xmath909 where @xmath804 .
we write @xmath910 for @xmath911 for convenience , and note that @xmath912 if @xmath913 and @xmath914 if @xmath915 .
set @xmath916 , @xmath917 and @xmath918 .
note that @xmath382 is an open neighborhood of @xmath857 relative to @xmath488 , @xmath919 is a open neighborhood of @xmath780 and @xmath920 .
we choose an @xmath921 so that @xmath922 for all @xmath923 , and assume henceforth that @xmath923 .
we now prove that if @xmath924 is small enough , then @xmath925 satisfies in the viscosity sense @xmath926 to this end , we fix any @xmath927 and @xmath928
. choose @xmath929 so that @xmath930 .
it is a standard observation that if @xmath931 , then @xmath932 .
next , let @xmath933 denote the projection of @xmath934 on the half space @xmath935 .
that is , @xmath936 if @xmath937 and @xmath938 otherwise .
we note that @xmath939 and @xmath940 .
hence , @xmath941 and furthermore , @xmath942 , where @xmath943 .
accordingly , we get @xmath944 .
we may assume that @xmath945 , so that @xmath931 .
if @xmath946 , then @xmath947 and we have @xmath948 .
moreover , writing @xmath949 for the modulus of @xmath12 on @xmath950 , we get @xmath951
. we may assume by reselecting @xmath952 by a smaller positive number that @xmath953 .
thus we have @xmath954 . next , assume that @xmath955 .
we have @xmath956 , and hence , @xmath957 . since @xmath958 , we find that @xmath959 , and @xmath960 where @xmath357 is a lipschitz bound of @xmath961
. we may assume by replacing @xmath952 by a smaller positive number if needed that @xmath962 . then we have @xmath963 , and therefore , @xmath948 . as before ,
we get @xmath964 .
thus we conclude that if @xmath923 , then @xmath965 is satisfied in @xmath966 in the viscosity sense .
next , we assume that @xmath967 . since @xmath968
, we have @xmath969 and @xmath970 .
thus , @xmath925 satisfies @xmath971 in @xmath919 in the viscosity sense . since @xmath972 in @xmath966 in the viscosity sense ,
the functions @xmath925 on @xmath966 , with @xmath923 , have a common lipschitz bound .
therefore , by replacing @xmath866 a smaller positive number if necessary , we may assume that for any @xmath923 , @xmath973 in @xmath919 in the viscosity sense .
finally , we fix @xmath819 and choose an @xmath974 so that @xmath975 for all @xmath808 . by mollifying @xmath976 , we may find a function @xmath977 such that @xmath978 for all @xmath979 , @xmath980 for all @xmath981 and @xmath982 for all @xmath983 .
the collection of the open subset @xmath984 of @xmath488 , the sequence @xmath985 of neighborhoods of @xmath986 and the sequence @xmath781 of functions gives us what we needed .
[ i : inf - en ] let @xmath987 be a nonempty collection of subsolutions of , . set @xmath988 for @xmath989 .
assume that @xmath6 is a real - valued function on @xmath990 .
then @xmath6 is a subsolution of , .
set @xmath468 .
fix @xmath469 and @xmath470 , and assume that @xmath991 attains a strict maximum at @xmath472 .
we may assume that @xmath992 has the form : @xmath993 for some functions @xmath994 and @xmath995 .
fix any @xmath247 . by lemma [ i : selec ]
, there exists @xmath996 such that @xmath997 and @xmath998 for all @xmath307 . by this first condition
, we see that the functions @xmath999 , with @xmath1000 , are subsolutions of @xmath1001 by ( * ? ? ?
* theorem 2.8 ) , we find that @xmath6 is a subsolution of , which implies that either @xmath1002 , or @xmath475 and @xmath1003 .
but , this last inequality guarantees that @xmath1004 .
since @xmath247 is arbitrary , we see that either @xmath1005 , or @xmath475 and @xmath1006 . hence , @xmath6 is a subsolution of , .
the above proof reduces the nonlinear boundary condition to the case of a family of linear neumann conditions . one can prove the above lemma without such a linearization procedure by treating directly the nonlinear condition and using lemma [ i : localized - main ] as in the proof of ( * ? ? ?
* theorem 2.8 ) .
it is worthwhile to noticing that another convexity lemma is valid .
that is , if @xmath6 and @xmath18 are subsolutions of , , then so is the function @xmath1007 , with @xmath1008 .
note that the propositions , corresponding to this convexity lemma and lemma [ i : inf - en ] , are valid for ( dbc ) . in this section ,
we give the second proof of theorem [ thm : large - time ] for ( cn ) .
we write @xmath468 throughout this section .
we define the function @xmath1009 on @xmath488 by @xmath1010 where @xmath1011 is the function on @xmath488 given by @xmath1012 it is a standard observation that @xmath106 and @xmath1009 are lipschitz continuous functions on @xmath488 and are , respectively , a subsolution and a supersolution of ( e1 ) , with @xmath721 . moreover , using lemma [ i : inf - en ] , we see that @xmath1009 is a solution of ( e1 ) , with @xmath721 .
[ i : formula1]we have @xmath1013 for all @xmath352 . for the proof of this lemma
, we refer to the proof of ( * ? ? ?
* proposition 4.4 ) , which can be easily adapted to the present situation , and we skip it here .
we show that @xmath1014 therefore , in order to prove , it is enough to show the pointwise convergence in .
moreover , by lemma [ i : formula1 ] , we need only to show that @xmath1015 now , we fix any @xmath352 .
since @xmath1009 is a solution of ( e1 ) , with @xmath721 , by theorem [ i : exist - extremal - sn ] , there is an extremal triple @xmath554 such that , if we set @xmath1016 , then we have @xmath1017 which is equivalent to the condition that @xmath1018 for a.e .
moreover , we have @xmath1019 and @xmath1020 . by theorem [ i : exist - p - n ] , there exists a function @xmath1021 such that for a.e .
@xmath1022 , @xmath1023 , @xmath761 and @xmath762 if @xmath763 .
observe that @xmath1024 for a.e . @xmath1022 .
next , combining the above relations together with the fenchel - young inequality , @xmath1025 , we observe that if we set @xmath436 , then for a.e .
@xmath889 , @xmath1026 thus , all the inequalities above are indeed equalities . in particular , we find that @xmath1027 for a.e .
@xmath889 , which shows that @xmath1028 and @xmath1029 for a.e .
@xmath889 .
we here consider only the case when ( a7)@xmath1030 is valid .
it is left to the reader to check the other case when ( a7)@xmath1031 holds .
the argument outlined below is parallel to the last half of the proof of ( * ? ? ?
* theorem 1.3 ) . since ( a7)@xmath1030 is assumed , there exist a constant @xmath1032 and a function @xmath1033 satisfying @xmath1034 such that for any @xmath1035 and @xmath1036 , if @xmath1037 and @xmath1038 for some @xmath138 , then @xmath1039 this ensures that for a.e . @xmath889 and all @xmath1035 , @xmath1040 we fix @xmath247 , and note ( see for instance the proof of ( * ? ? ? * theorem 1.3 ) ) that there is a positive constant @xmath1041 and , for each @xmath1042 , a constant @xmath1043 such that @xmath1044 .
we choose @xmath1045 so that @xmath1046 .
fix any @xmath1047 , and set @xmath1048 , @xmath1049 , @xmath1050 and @xmath1051 . note that @xmath1052 and @xmath396 as @xmath175 .
we set @xmath1053 and @xmath1054 for @xmath889 . using and noting that @xmath1055 and @xmath1056 , we get @xmath1057 hence , noting that @xmath1058 and @xmath1059 and using the dynamic programming principle , we find that @xmath1060 which implies that @xmath1061 .
the proof is now complete .
finally we briefly sketch some arguments in order to explain how to adapt the dynamical approach to theorem [ thm : large - time ] for ( cn ) to that of theorem [ thm : large - time ] for ( dbc ) . as usual , we assume that @xmath1062 and let @xmath1009 be the same function as in .
we infer that @xmath1063 for all @xmath352 . to prove the inequality @xmath1061 for @xmath352 , we fix @xmath352 and @xmath247 , and select a triple @xmath554 , a constant @xmath1064 and a family @xmath1065 $ ] as in the dynamical approach .
we fix @xmath371 and introduce the function @xmath612 on @xmath366 given by @xmath613 .
define the constant @xmath1066 $ ] by @xmath1067 .
since @xmath1068 , we have @xmath1069 as @xmath175 .
we set @xmath1070 , @xmath1049 , @xmath1071 and @xmath1072 .
we note that @xmath1073 and that , as @xmath175 , @xmath1074 and @xmath1075 .
we assume henceforth that @xmath173 is large enough so that @xmath1076 .
we define @xmath1077 , @xmath1078 , @xmath1079 and @xmath1080 as in the dynamical approach for ( cn ) .
we define the function @xmath1081 on @xmath366 by @xmath1082 for @xmath211 .
it is easily seen that @xmath1083 .
then we compute similarly that @xmath1084 where @xmath1033 is a function satisfying @xmath1034 , and @xmath1085 from which we conclude that @xmath1061 for all @xmath352 .
once the additive eigenvalues of ( e1 ) and ( e2 ) are normalized so that @xmath1086 , the correspondence between the initial data @xmath106 and the asymptotic solution @xmath1009 is the same for both ( cn ) and ( dbc ) .
we assume that @xmath1086 , and present in this section another formula for the function @xmath1009 given by .
we introduce the aubry ( or , aubry - mather ) set @xmath92 for ( e1 ) , with @xmath721 .
we first define the function @xmath1087 by @xmath1088 and then the aubry ( or , aubry - mather ) set @xmath92 for ( e1 ) , with @xmath721 , as the subset of @xmath488 consisting of those points @xmath1089 where the function @xmath1090 is a solution of ( e1 ) , with @xmath721 . the function @xmath1009 given by is represented as @xmath1091 the function @xmath1011 has the formula similar to the above : @xmath1092 .
accordingly , we have @xmath1093 . we do not give the proof of these formulas , and instead we refer the reader to ( * ? ? ?
* proposition 4.4 ) and ( * ? ? ?
* proposition 6.3 ) where these formulas are established for ( e1 ) with the linear neumann condition .
in order to prove theorem [ thm : comparison ] , we have to build test - functions . for the convenience of the reader
, we briefly recall how to construct these functions and refer to @xcite for more general cases as well as more details . by ( a2 ) and ( a3 ) there exists a function @xmath1098 such that @xmath1099 for some @xmath1094 . noting that @xmath1100 is increasing , we see that @xmath1101 is decreasing for any @xmath84 . therefore we see that a regularized function @xmath1102 by a mollification kernel satisfies the desired properties .
[ lem : func - c ] for any @xmath1103 , there exists @xmath1104 such that for any @xmath1105 , @xmath238 , and @xmath218 , there exists a function @xmath1106 such that @xmath1107 for any @xmath84 , where @xmath1108 is a modulus .
[ lem : coercivity ] for fixed constants @xmath1110 , we denote by @xmath1111 , @xmath1108 , and @xmath1112 the functions and the constant given in lemma [ lem : func - c ] for @xmath238 and @xmath218 .
we introduce the function @xmath995 defined , for @xmath1113 , by @xmath1114 for @xmath1115 and @xmath659 . if @xmath1116 , then we first consider case ( a )
. using ( a0 ) and a taylor expansion at the point @xmath1129 , it is easy to see that @xmath1130 using the continuity of @xmath1131 , and therefore of @xmath1132 , we see that @xmath1133 for some modulus @xmath279 .
therefore , taking into account the restriction @xmath1119 and changing perhaps the modulus @xmath279 , we have @xmath1134 and this yields @xmath1135 as @xmath294 , @xmath1136 , and @xmath1137 , which implies that @xmath1138 in case ( b ) by ( a3 ) , ( a4 ) and lemma [ lem : func - c ] , we get ( changing perhaps the modulus @xmath279 ) @xmath1139 since @xmath1140 . gathering the two cases , we have the result
. similarly we obtain @xmath1141 if @xmath1142 .
let @xmath1145 denote the function @xmath1146}\{u(x , s)-(1/{\gamma})(t - s)^2\}\ ; , \ ] ] for any @xmath1147 .
this sup - convolution procedure is standard in the theory of viscosity solutions ( although , here , it acts only on the time - variable ) and it is known that , for @xmath1148 small enough , @xmath1145 is a subsolution of ( cn ) in @xmath1149 , where @xmath1150 ( see @xcite for instance ) .
moreover , it is easy to check that @xmath1151 in @xmath1149 and therefore by the coercivity of @xmath12 and the @xmath5-regularity of @xmath22 we have , for all @xmath1152 , @xmath1153 $ ] @xmath1154 for some @xmath1155 . finally , as @xmath1156 , @xmath1157 .
therefore it is enough to consider @xmath1158 for @xmath1147 small enough , and we follow the classical proof by introducing @xmath1159 for @xmath1160 .
this maximum is achieved at @xmath1161 $ ] , namely @xmath1162 .
clearly @xmath612 depends on @xmath364 but we can assume that it remains bounded away from @xmath113 , otherwise we easily get a contradiction .
we only consider the case where @xmath218 .
we first consider problem ( cn ) .
we introduce the function @xmath243 defined by : @xmath1163 where @xmath995 is the function given by lemma [ lem : coercivity ] with @xmath1164 and @xmath1165 .
it is worth pointing out that , compared to the proof of lemma [ lem : main - lem ] , the change @xmath1166 , consists in exchanging the role of @xmath73 and @xmath1089 , which is natural since the variable @xmath73 is used here for the subsolution while it was corresponding to a supersolution in the proof of lemma [ lem : main - lem ] .
we define the function @xmath1167\to{\mathbb{r}}$ ] by @xmath1168 let @xmath1169 achieve its maximum at @xmath1170 $ ] . by standard arguments
, we have @xmath1171 by taking a subsequence if necessary . in view of the lipschitz continuity of @xmath1172
, we have @xmath1173 . taking ( formally ) the derivative of @xmath1169 with respect to each variable @xmath1174 at @xmath1175
, we have @xmath1176 we remark that we should interpret @xmath1177 and @xmath1178 in the viscosity solution sense here .
we also point out that the viscosity inequalities we are going to write down below , hold up to time @xmath358 , in the spirit of @xcite , lemma 2.8 , p. 41 .
therefore , by the definition of viscosity solutions of ( cn ) , using the arguments of user s guide to viscosity solutions @xcite , there exists @xmath1180 such that @xmath1181 with @xmath1182 . by we may assume that @xmath1183 as @xmath294 for some @xmath84 by taking a subsequence if necessary .
sending @xmath294 and then @xmath296 in the above inequalities , we have a contradiction since @xmath1184 while the @xmath12-terms converge to the same limit .
therefore @xmath612 can not be assumed to remain bounded away from @xmath113 and the conclusion follows .
we next consider problem ( dbc ) .
let @xmath1185 be defined as above and let @xmath1186 be the function given by lemma [ lem : func - c2 ] .
we define the function @xmath1187 by @xmath1188 for @xmath1189 .
we define the function @xmath1167\to{\mathbb{r}}$ ] by @xmath1190 let @xmath1191 achieve its maximum at @xmath1192^{2}$ ] and set @xmath1193 derivating ( formally ) @xmath1191 with respect to each variable @xmath1194 at @xmath1195 , we have @xmath1196 we remark that we should interpret @xmath1197 and @xmath1198 in the viscosity solution sense here .
we consider the case where @xmath277 .
note that @xmath1199 and then we have @xmath1200 for @xmath281 which are small enough compared to @xmath37 . in the case where @xmath284 we similarly obtain @xmath1201 for @xmath281 which are small enough compared to @xmath37 .
the existence part being standard by using the perron s method ( see @xcite ) , we mainly concentrate on the regularity of solutions when @xmath1202 .
we may choose a sequence @xmath1203 so that @xmath1204 and @xmath1205 for some @xmath357 which is uniform for all @xmath55 .
we fix @xmath55 .
we claim that @xmath1206 and @xmath1207 are , respectively , a sub and supersolution of ( cn ) or ( dbc ) with @xmath1208 for a suitable large @xmath1094 .
we can easily see that @xmath1209 are a sub and supersolution of ( cn ) or ( dbc ) in @xmath3 if @xmath1210 .
we recall ( see @xcite for instance ) that if @xmath67 , then @xmath1211 where @xmath1212 denotes the super - differential of @xmath1213 at @xmath1214 .
we need to show that @xmath1215 for all @xmath1216 . by ( a2 ) it is clear enough that there exists @xmath1217 such that , if @xmath1218 , then @xmath1219
. then choosing @xmath1220 , the above inequality holds .
a similar argument shows that @xmath1221 is a supersolution of ( cn ) for @xmath1222 large enough .
it is worth pointing out that such @xmath1222 is independent of @xmath54
. we can easily check that @xmath1209 are a sub and supersolution of ( dbc ) on @xmath1 too . by perron s method ( see @xcite ) and theorem [ thm : comparison ] , we obtain continuous solutions of ( cn ) or ( dbc ) with @xmath1208 that we denote by @xmath1223 . as a consequence of perron s method
, we have @xmath1224 to conclude , we use a standard argument : comparing the solutions @xmath1225 and @xmath1226 for some @xmath1227 and using the above property on the @xmath1223 , we have @xmath1228 as a consequence we have @xmath1229 and , by using the equation together with ( a1 ) , we obtain that @xmath1230 is also bounded . finally sending @xmath697 by taking a subsequence
if necessary we obtain the lipschitz continuous solution of ( cn ) or ( dbc ) .
we finally remark that , if @xmath1231 , we can obtain the existence of the uniformly continuous solution on @xmath7 by using the above result for @xmath1232 and ( [ eq : contsol ] ) which is a direct consequence of theorem [ thm : comparison ] .
we first prove ( i ) . for any @xmath1233 we consider @xmath1234 following similar arguments as in the proof of theorem [ thm : existence ] , it is easy to prove that , for @xmath357 large enough @xmath1235 and @xmath1236 are , respectively , a subsolution and a supersolution of or .
we remark that , because of ( a1 ) and the regularity of the boundary of @xmath3 , the subsolutions @xmath21 of such that @xmath1237 on @xmath2 satisfy @xmath1238 in @xmath3 for some @xmath1239 and therefore they are equi - lipschitz continuous on @xmath2 . with these informations ,
perron s method provides us with a solution @xmath1240 of .
moreover , by construction , we have @xmath1241 for a fixed @xmath1242 . because of and
the regularity of the boundary @xmath22 , @xmath1243 is a sequence of equi - lipschitz continuous and uniformly bounded functions on @xmath2 . by ascoli - arzela s theorem
, there exist subsequences @xmath1244 and @xmath1245 such that @xmath1246 as @xmath408 for some @xmath1247 and @xmath1248 . by a standard stability result of viscosity solutions
we see that @xmath1249 is a solution of ( e1 )
. * acknowledgements . *
this work was partially done while the third author visited mathematics department , university of california , berkeley .
he is grateful to professor lawrence c. evans for useful comments and his kindness .
g. barles and j .-
roquejoffre , _ ergodic type problems and large time behaviour of unbounded solutions of hamilton - jacobi equations _ , comm .
partial differential equations * 31 * ( 2006 ) , no . 7 - 9 , 12091225 .
y. giga , q. liu and h. mitake , _ large - time behavior of one - dimensional dirichlet problems of hamilton - jacobi equations with non - coercive hamiltonians _
, j. differential equations 252 ( 2012 ) , 12631282 . .
h. ishii , _
asymptotic solutions for large time of hamilton - jacobi equations in euclidean n space _ , ann .
h. poincar anal .
non linaire , * 25 * ( 2008 ) , no 2 , 231266 . h. ishii , weak kam aspects of convex hamilton - jacobi equations with neumann type boundary conditions , j. math .
pures appl .
( 9 ) * 95 * ( 2011 ) , no .
1 , 99135 .
e. yokoyama , e , y. giga , and p. rybka , _ a microscopic time scale approximation to the behavior of the local slope on the faceted surface under a nonuniformity in supersaturation _
d * 237 * ( 2008 ) , no .
22 , 28452855 . | in this article , we study the large time behavior of solutions of first - order hamilton - jacobi equations , set in a bounded domain with nonlinear neumann boundary conditions , including the case of dynamical boundary conditions .
we establish general convergence results for viscosity solutions of these cauchy - neumann problems by using two fairly different methods : the first one relies only on partial differential equations methods , which provides results even when the hamiltonians are not convex , and the second one is an optimal control / dynamical system approach , named the `` weak kam approach '' which requires the convexity of hamiltonians and gives formulas for asymptotic solutions based on aubry - mather sets . |
the present paper deals with finding large free sets for infinite set mappings of finite order . for a set @xmath19 and a cardinal @xmath2 , we denote by @xmath20^{\lambda}$ ] ( resp . ,
@xmath20^{<{\lambda}}$ ] ) the set of all subsets of @xmath19 with @xmath2 elements ( resp . , less than @xmath2 elements ) . for cardinals
@xmath1 and @xmath2 , a subset @xmath21 of the powerset @xmath22 of @xmath1 , and a map @xmath23^{<{\lambda}}$ ] , we say that a subset @xmath24 of @xmath1 is _ free with respect to @xmath25 _ if @xmath26 for each @xmath27 . for cardinals
@xmath2 , @xmath7 , and for a positive integer @xmath14 , the statement @xmath28 holds if every map @xmath6^r\to[{\kappa}]^{<{\lambda}}$ ] ( we say a _ set mapping of order @xmath14 _ ) has a @xmath7-element free set . likewise , @xmath29 means that every map @xmath6^{<{\omega}}\to[{\kappa}]^{<{\lambda}}$ ] has a @xmath7-element free set .
denote by @xmath10 the @xmath11-th successor of an infinite cardinal @xmath2 .
free set theorem ( see @xcite or ( * ? ? ?
* theorem 46.1 ) ) states that @xmath30 iff @xmath9 , for all infinite cardinals @xmath1 and @xmath2 and every non - negative integer @xmath11 . whether larger free subsets can be found leads to unexpected discoveries .
an argument that originates in lzr @xcite shows that the relation @xmath31 holds for each infinite cardinal @xmath2 ( cf .
* corollary 44.2 ) ) .
therefore , the relation @xmath32 holds _ a fortiori _
, for each positive integer @xmath33 .
further , hajnal - mt @xcite and hajnal @xcite , respectively , proved that the relation @xmath34 holds for every infinite cardinal @xmath2 , every @xmath35 , and every integer @xmath36 ( cf .
* theorem 46.2 ) ) . in the presence of the generalized continuum hypothesis @xmath37 , we can say more .
indeed , it follows from ( * ? ? ?
* theorem 45.5 ) that the relation @xmath38 holds for every integer @xmath39 and every infinite cardinal @xmath2 . on the other hand , without assuming @xmath37 ,
the situation gains a touch of strangeness .
set @xmath40 , @xmath41 , and for each positive integer @xmath11 , @xmath42 is the least positive integer such that @xmath43 ( the latter notation meaning that for each @xmath44 ^ 5\to{\{0,1\}}$ ] , either there exists a @xmath45-element subset @xmath19 of @xmath42 such that @xmath46 ^ 5={\{0\}}$ ] or there exists a @xmath47-element subset @xmath19 of @xmath42 such that @xmath46 ^ 5={\{1\}}$ ] ) .
the existence of the sequence @xmath48 is ensured by ramsey s theorem .
it is established in komjth and shelah @xcite that for each positive integer @xmath11 , there exists a generic extension of the universe in which @xmath49 .
in particular , there exists a generic extension of the universe in which @xmath50 . by using algebraic tools of completely different nature and purpose introduced in @xcite called _ compatible norm - coverings _ , the first author established in ( * ? ? ?
* thorme 3.3.13 ) the relation @xmath51 for each positive integer @xmath14 and each infinite cardinal @xmath2 , thus improving by one the cardinality of the free set given by kuratowski s free set theorem , and thus solving in the affirmative the question , raised in @xcite , whether @xmath52 . in the present paper
, we develop a self - contained approach to such questions , and we extend the methods in order to find further large free sets results . in particular , we establish the relation @xmath53 for each infinite cardinal @xmath2 and for all positive integers @xmath11 and @xmath14 with @xmath15 ; here @xmath54 denotes the largest integer below any real number @xmath55 .
our main idea is to extend the @xmath28 notation by introducing a partially ordered set ( poset ) parameter , thus defining the notation @xmath3 , for cardinals @xmath1 , @xmath2 and a poset @xmath0 , see definition [ d : infcombp ] . in particular ,
the statements @xmath56^{<{\omega}},\subseteq)$ ] and @xmath29 are equivalent , for all cardinals @xmath1 , @xmath2 , and @xmath7 ( cf .
proposition [ p : twoarrequiv ] ) .
then we define the _ kuratowski index _ of a finite poset @xmath0 as the least non - negative integer @xmath11 such that @xmath57 holds for each infinite cardinal @xmath1 ( cf . definition [ d : kurind ] ) , with a minor adjustment for antichains .
this definition is tailored in order to ensure that the finite boolean lattice @xmath58 of all subsets of @xmath11 has kuratowski index @xmath11 , for each positive integer @xmath11 .
smaller posets have smaller kuratowski index ( proposition [ p : kurindincr ] ) while the kuratowski index function is subadditive on finite direct products ( cf .
proposition [ p : addkur ] ) . as a corollary ,
the kuratowski index is bounded above by the order - dimension ( cf .
proposition [ p : uppbd ] ) .
we apply these results to the _ truncated cubes _ @xmath59 ( cf .
notation [ not : b(k , m ) ] ) .
in particular , we present the first author s proof of the relation , which uses the fact that the order - dimension of @xmath60 ( and thus also its kuratowski index ) is equal to @xmath61 . by using a further estimate of the order - dimension established in a 1971 paper by spencer @xcite
, we deduce the relation ( cf .
proposition [ p : largefreesets ] ) .
for example , @xmath16 .
another estimate of the order - dimension of @xmath59 , originating from a 1986 paper by fredi and kahn @xcite , not so good asymptotically but giving larger free sets for smaller alephs ( cf .
proposition [ p : fuka ] ) , yields , for example , the relation @xmath62 finally , dushnik s original exact estimate yields , for example , the relation @xmath63
we shall use basic set - theoretical notation and terminology about ordinals and cardinals .
we denote by @xmath64 , or @xmath65 ( resp . ,
@xmath66 ) the image ( resp . , inverse image ) of a set @xmath19 under @xmath67 .
cardinals are initial ordinals .
we denote by @xmath68 the cofinality of an ordinal @xmath69 .
we denote by @xmath70 the first limit ordinal , mostly denoted by @xmath71 in case it is viewed as a cardinal .
we identify every non - negative integer @xmath11 with the finite set @xmath72 ( so @xmath73 ) .
we denote by @xmath74 the successor cardinal of a cardinal @xmath1 , and we define @xmath75 , for a non - negative integer @xmath11 , by @xmath76 and @xmath77 .
we denote by @xmath78 the powerset of a set @xmath19 , and we set @xmath79^{\kappa}&:={{\{y\in{\mathfrak{p}}(x)\mid\operatorname{card}y={\kappa}\}}}\,,\\ [ x]^{<{\kappa}}&:={{\{y\in{\mathfrak{p}}(x)\mid\operatorname{card}y<{\kappa}\}}}\,,\\ [ x]^{{{\leqslant}}{\kappa}}&:={{\{y\in{\mathfrak{p}}(x)\mid\operatorname{card}y\leq{\kappa}\}}}\ , , \end{aligned}\ ] ] for every cardinal @xmath1
. all our posets will be nonempty . for posets @xmath0 and @xmath80 ,
a map @xmath81 is _ isotone _ if @xmath82 implies that @xmath83 , for all @xmath84 .
we denote by @xmath85 the least element of @xmath0 if it exists .
an element @xmath86 in a poset @xmath0 is _ join - irreducible _ if @xmath87 implies that @xmath88 , for every ( possibly empty ) finite subset @xmath19 of @xmath0 ; we denote by @xmath89 the set of all join - irreducible elements of @xmath0 , endowed with the induced partial ordering .
we set @xmath90 for all subsets @xmath80 and @xmath19 of @xmath0 ; in case @xmath91 is a singleton , then we shall write @xmath92 ( resp . , @xmath93 ) instead of @xmath94 ( resp . , @xmath95 ) .
we set @xmath96 , for each @xmath97 .
a subset @xmath80 of @xmath0 is a _ lower subset of @xmath0 _ if @xmath98 .
we say that @xmath0 is a _ tree _ if @xmath0 has a smallest element and @xmath99 is a chain for each @xmath97 .
the statement referred to in the section title is the following .
[ d : infcombp ] for cardinals @xmath1 , @xmath2 and a poset @xmath0 , let @xmath3 hold if for every mapping @xmath100^{<{\lambda}}$ ] , there exists a one - to - one map @xmath101 such that @xmath102 in most cases , we shall use the symbol @xmath3 only in case @xmath0 is _ lower finite _ , in which case it is sufficient to assume that the mapping @xmath25 is defined only on @xmath103^{<{\omega}}$ ] ( we can always set @xmath104 for infinite @xmath19 ) .
once this is set , it is of course sufficient to assume that @xmath25 is _ isotone _ ( replace @xmath105 by @xmath106 ) . [ l : subposetleadsto ] let @xmath0 and @xmath80 be posets , with @xmath80 lower finite , and let @xmath1 , @xmath2 be cardinals .
if @xmath107 and @xmath0 embeds into @xmath80 , then @xmath3 .
we may assume that @xmath0 is a sub - poset of @xmath80 .
let @xmath6^{<{\omega}}\to[{\kappa}]^{<{\lambda}}$ ] be isotone . by assumption , there exists a one - to - one map @xmath108 such that @xmath109 the restriction @xmath67 of @xmath110 to @xmath0 is one - to - one . as @xmath25 is isotone and @xmath111 for each @xmath112 ,
is obviously satisfied .
the following lemma states that the @xmath3 relation can be `` verified on the join - irreducible elements '' .
[ l : equivkurat ] let @xmath0 be a lower finite poset and let @xmath2 and @xmath1 be infinite cardinals .
then @xmath3 iff for every isotone mapping @xmath113^{<{\omega}}\to[{\kappa}]^{<{\lambda}}$ ] , there exists a one - to - one map @xmath114 such that @xmath115 it is trivial that @xmath3 entails the given condition . for the converse
, we shall fix a bijection @xmath116^{<{\omega}}\to{\kappa}$ ] , and set @xmath117 that is , @xmath118 is the union of all finite subsets @xmath119 of @xmath1 such that @xmath120 . in particular , @xmath118 belongs to @xmath103^{<{\lambda}}$ ] whenever @xmath19 belongs to @xmath103^{<{\lambda}}$ ] .
now let @xmath6^{<{\omega}}\to[{\kappa}]^{<{\lambda}}$ ] be an isotone map .
we can define a map @xmath113^{<{\omega}}\to[{\kappa}]^{<{\lambda}}$ ] by the rule @xmath121^{<{\omega}}\,.\ ] ] by assumption , there exists a one - to - one map @xmath122 which satisfies the condition .
as @xmath0 is lower finite , we can define a map @xmath123 by the rule @xmath124 as @xmath0 is lower finite , every element of @xmath0 is the join of all elements of @xmath89 below it . hence , as both @xmath110 and @xmath125 are one - to - one , @xmath67 is one - to - one as well .
now we prove that the condition holds .
let @xmath82 in @xmath0 and let @xmath126 such that @xmath127 , we must prove that @xmath128 . as @xmath129 and @xmath25 is isotone
, we get @xmath130 thus @xmath131 .
this means that for each @xmath132 , @xmath133 belongs to @xmath134 .
as @xmath133 also belongs to @xmath135 ( because @xmath136 ) , it follows , using , that @xmath137 , and so @xmath138 . as this holds for each @xmath132
, we obtain that @xmath128 , as was to be proved .
we shall now recall a few known facts about the relation @xmath139 ( cf .
section [ s : intro ] ) . in case @xmath140 and @xmath141 ,
the existence of @xmath1 such that @xmath29 is a large cardinal axiom , that entails the existence of @xmath142 in case @xmath143 ( cf .
@xcite , and also @xcite for further related consistency strength results ) .
the relation @xmath29 follows from the infinite partition property @xmath144 ( _ existence of the @xmath145 erds cardinal _ ) where @xmath146 , see ( * ? ? ?
* theorem 45.2 ) and the discussion preceding it .
this notation is related to ours _ via _ the following easy result .
[ p : twoarrequiv ] the statements @xmath56^{<{\omega}},\subseteq)$ ] and @xmath29 are equivalent , for every cardinal @xmath7 and all infinite cardinals @xmath1 , @xmath2 .
we use the characterization of the @xmath147 relation given by lemma [ l : equivkurat ] . first observe that @xmath148^{<{\omega}})={{\{{\{{\xi}\}}\mid{\xi}<{\rho}\}}}$ ] .
assume that the statement @xmath149^{<{\omega}}$ ] holds , and let @xmath6^{<{\omega}}\to[{\kappa}]^{<{\lambda}}$ ] . by using the easy direction of lemma [ l : equivkurat ] ,
we obtain a one - to - one map @xmath150 such that , putting @xmath151 , @xmath152^{<{\omega}}\,.\ ] ] this means that @xmath24 is free with respect to @xmath25 .
conversely , assume that @xmath29 holds , and let @xmath100^{<{\lambda}}$ ] . by assumption
, there exists @xmath153^{\rho}$ ] which is free with respect to @xmath25 .
then holds , for any one - to - one map @xmath154 .
hence @xmath56^{<{\omega}},\subseteq)$ ] holds .
the following observation illustrates the difficulties of checking definition [ d : infcombp ] on arbitrary posets .
[ p : nofreenonlf ] there exists no cardinal @xmath1 satisfying @xmath155 . observe that @xmath156 .
let @xmath157 hold if the symmetric difference @xmath158 is finite , for all subsets @xmath19 and @xmath119 of @xmath1 .
let @xmath159 be a set that meets the @xmath160-equivalence class @xmath20_{\equiv_{\mathrm{fin}}}$ ] of @xmath19 in exactly one point , for each @xmath161 , and denote by @xmath105 the unique element of @xmath20_{\equiv_{\mathrm{fin}}}\cap\delta$ ] if @xmath19 is at most countable , @xmath162 otherwise .
if the relation @xmath155 holds , then there is a one - to - one map @xmath163 such that @xmath164 for all @xmath165 in @xmath166 . as @xmath167
, there exists @xmath168 such that @xmath169 .
put @xmath170 .
it follows from the relation @xmath171 that @xmath172 , thus @xmath173 while @xmath174 , a contradiction .
[ d : kurind ] a non - negative integer @xmath11 is _ * a * kuratowski index _ of a finite poset @xmath0 if either @xmath0 is an antichain , or @xmath0 is not an antichain , @xmath175 , and the relation @xmath57 holds for each infinite cardinal @xmath1 .
furthermore , we shall call _ * the * kuratowski index of @xmath0 _ the least such @xmath11 if it exists ( we shall see shortly that it does ) , and we shall denote it by @xmath176 . the reason of our adjustment of definition [ d : kurind ] for antichains lies essentially in the fact that the relation @xmath3 trivially holds for every antichain @xmath0 and all cardinals @xmath1 , @xmath2 such that @xmath177 .
[ p : existskur(p ) ] the number of join - irreducible elements in @xmath0 is a kuratowski index of @xmath0 , for each finite poset @xmath0 .
hence , @xmath176 is defined and @xmath178 .
the statement is trivial in case @xmath0 is an antichain , so suppose that @xmath0 is not an antichain .
set @xmath179 , let @xmath2 be an infinite cardinal , and set @xmath180 .
it follows from kuratowski s free set theorem @xcite that the relation @xmath181 holds ( cf . proposition [ p : twoarrequiv ] ) .
as the assignment @xmath182 defines an embedding from @xmath0 into @xmath183 , it follows from lemma [ l : subposetleadsto ] that the relation @xmath3 holds .
[ p : kurattree ] let @xmath184 be a well - founded tree and let @xmath1 be an infinite cardinal such that @xmath185 and @xmath186 for each @xmath187
. then the relation @xmath188 holds .
as @xmath184 is well - founded , there exists a unique map @xmath7 from @xmath184 to the ordinals ( the rank map ) such that @xmath189 for each @xmath190 .
the range @xmath191 of the map @xmath7 is an ordinal and @xmath192 .
fix a partition @xmath193 of @xmath1 such that @xmath194 for each @xmath195 .
fix also a one - to - one map @xmath196 .
we define a strict ordering @xmath197 on @xmath184 ( lexicographical product ) by @xmath198 observe that @xmath197 is a strict well - ordering of @xmath184 and that it extends the original strict ordering @xmath199 .
let @xmath100^{<{\kappa}}$ ] .
we define , by @xmath197-induction , a map @xmath200 , as follows .
let @xmath190 and suppose that @xmath201 has been defined and belongs to @xmath202 , for each @xmath187 such that @xmath203 .
it follows from the assumptions on @xmath184 that the set @xmath204 ( where @xmath205 is evaluated with respect to the original ordering of @xmath184 ) has cardinality less than @xmath1 .
hence , @xmath206 is nonempty .
we pick @xmath207 .
we claim that @xmath67 is one - to - one .
indeed let @xmath208 distinct , we prove that @xmath209 .
we may assume that @xmath203 .
if @xmath210 , then @xmath211 , so @xmath212 and the conclusion follows .
if @xmath213 , then the conclusion follows from @xmath214 , @xmath215 , and @xmath216 . furthermore , for all @xmath217 in @xmath184 , @xmath218 does not belong to @xmath219 , thus , _ a fortiori _ , not to @xmath220 .
therefore , @xmath221 for all @xmath222 in @xmath184 . from lemma [ l : subposetleadsto ]
we deduce immediately the following .
[ p : kurindincr ] if a poset @xmath0 embeds into a finite poset @xmath80 , then @xmath223 .
the proof of the following result is inspired by the proof of kuratowski s free set theorem , see ( * ? ? ?
* theorem 46.1 ) .
it is closely related to ( * ? ? ?
* lemme 3.3.7 ) .
[ l : prodkurat ] let @xmath0 and @xmath80 be lower finite posets with zero and let @xmath69 , @xmath224 , @xmath225 be infinite cardinals such that @xmath226 and @xmath227 . if @xmath228 and @xmath229 , then @xmath230 .
we use the equivalent form of the @xmath147 relation provided by lemma [ l : equivkurat ] .
let @xmath231^{<{\alpha}}$ ] . as @xmath232 , there exists a partition @xmath233 of the set @xmath225 such that @xmath234 and @xmath235 . as @xmath236 , we can define a map @xmath237^{<{\beta}^+}$ ] by the rule @xmath238^{<{\omega}}\right)}\,,\quad\text{for each } y\in{\mathfrak{p}}(v)\,.\ ] ] as @xmath229 , there exists a one - to - one map @xmath239 such that @xmath240 as @xmath227 , we can define a map @xmath241^{<{\alpha}}$ ] by the rule @xmath242 as @xmath228 , there exists a one - to - one map @xmath243 such that @xmath244 as @xmath245 and @xmath225 is the disjoint union of @xmath246 and @xmath247 , we can define a one - to - one map @xmath248 by the rule @xmath249 let @xmath250 in @xmath251 , we verify that the following statement holds : @xmath252 so let @xmath253 such that @xmath254 , that is , @xmath255 .
we must prove that @xmath256 .
we separate cases . * _ case _ 1 . * @xmath257 .
as @xmath258 while @xmath259 , it follows from that @xmath137 , that is , @xmath138 .
* _ case _ 2 .
* @xmath260 . as @xmath261 while @xmath262 , it follows from that @xmath263 , that is , @xmath264 .
this completes the proof of .
[ p : addkur ] the inequality @xmath265 holds , for any finite posets @xmath0 and @xmath80 with zero .
we may assume that @xmath0 and @xmath80 are both nonzero .
set @xmath266 and @xmath267 .
let @xmath1 be an infinite cardinal and set @xmath268 .
as @xmath269 and @xmath270 , it follows from the product lemma ( lemma [ l : prodkurat ] ) that @xmath271 .
now observe that @xmath272 .
denote by @xmath273 the _ order - dimension _ of a poset @xmath0 , that is , the smallest number @xmath1 of chains such that @xmath0 embeds into a product of @xmath1 chains .
the order - dimension of a finite poset is finite .
it follows immediately from proposition [ p : kurattree ] that the kuratowski index of a nontrivial finite tree is @xmath274 .
hence , by applying proposition [ p : addkur ] , we obtain immediately the following upper bound for @xmath176 .
[ p : uppbd ] le @xmath11 be a positive integer and let @xmath0 be a finite poset . if @xmath0 embeds , as a poset , into a product of @xmath11 trees , then @xmath275 .
in particular , @xmath276 .
as @xmath277 ( for the assignment @xmath182 defines a meet - embedding from @xmath0 into @xmath278 ) , this bound is sharper than the previously observed bound @xmath279 ( cf
. proposition [ p : existskur(p ) ] ) .
in fact , it dates back to baker @xcite that @xmath280 , where @xmath281 denotes the _ width _ function ( the width of a poset is the supremum of the cardinalities of all antichains of that poset ) .
hence , @xmath282 while proposition [ p : uppbd ] provides an upper bound for the kuratowski index of a finite poset , our next result , proposition [ p : kurgewbr ] , will provide a lower bound .
the classical definition of breadth , see baker @xcite or ditor ( * ? ? ?
* section 4 ) , runs as follows .
let @xmath11 be a positive integer .
a join - semilattice @xmath0 has _ breadth at most @xmath11 _ if for every nonempty finite subset @xmath19 of @xmath0 , there exists a nonempty @xmath283 with at most @xmath11 elements such that @xmath284 .
this is a particular case of the following definition of breadth , valid for every poset , and , in addition , _ self - dual _ : we say that a poset @xmath0 has breadth at most @xmath11 if for all @xmath285 , @xmath286 in @xmath0 , if @xmath287 for all @xmath288 in @xmath289 , then there exists @xmath290 such that @xmath291 .
we denote by @xmath292 the least non - negative integer @xmath11 , if it exists , such that @xmath0 has breadth at most @xmath11 , and we call it _ * the * breadth of @xmath0_. [ p : kurgewbr ] let @xmath0 be a finite poset with zero such that @xmath99 is a join - semilattice for each @xmath97 . then the inequality @xmath293 holds .
we may assume that @xmath294 . hence @xmath295 is a positive integer .
pick @xmath97 such that @xmath296 .
there are @xmath297 such that @xmath298 for each @xmath299 , where the join @xmath300 is evaluated in @xmath99 .
the assignment @xmath301 defines a join - embedding from @xmath58 into @xmath0 , thus , by proposition [ p : kurindincr ] , @xmath302 ( the latter equality is part of the content of kuratowski s free set theorem , cf .
proposition [ p : twoarrequiv ] ) . as an immediate application of propositions [ p : uppbd ] and [ p : kurgewbr ]
, we obtain the following .
[ c : kur(prdntr ) ] let @xmath11 be a positive integer and let @xmath0 be a product of @xmath11 nontrivial finite trees
. then @xmath303 .
in this section , we shall give estimates of kuratowski indexes of _ truncated cubes _ , with applications to finding large free sets for set mappings of order greater than one .
[ not : b(k , m ) ] for integers @xmath14 , @xmath304 , @xmath305 , and @xmath33 such that @xmath306 and @xmath307 for each @xmath308 , we define the _ truncated @xmath33-dimensional cubes _ @xmath309 endowed with containment .
diagrams indexed by @xmath310 , for @xmath311 , are widely used in gillibert @xcite .
as @xmath312 for each positive integer @xmath33 , we shall assume that @xmath313 whenever we consider the poset @xmath59 .
the following result is related to ( * ? ? ?
* corollaire 3.3.3 ) .
[ p : freeb(k , m ) ] let @xmath14 and @xmath33 be integers with @xmath314 and let @xmath1 and @xmath2 be infinite cardinals .
then the following statements are equivalent : 1 .
@xmath315 ; 2 . for every @xmath6^{{{\leqslant}}r}\to[{\kappa}]^{<{\lambda}}$ ]
there exists @xmath153^m$ ] such that @xmath26 for each @xmath316^{{{\leqslant}}r}$ ] ; 3 .
@xmath317 .
an easy argument , similar to the one used in the proof of proposition [ p : twoarrequiv ] , yields the equivalence of ( i ) and ( ii ) .
furthermore , as every map from @xmath103^r$ ] to @xmath103^{<{\lambda}}$ ] trivially extends to a map from @xmath103^{{{\leqslant}}r}$ ] to @xmath103^{<{\lambda}}$ ] , ( ii ) implies ( iii ) .
finally assume that ( iii ) holds , and let @xmath6^{{{\leqslant}}r}\to[{\kappa}]^{<{\lambda}}$ ] .
we must find an @xmath33-element free set , with respect to @xmath25 , of @xmath1 .
we may assume that @xmath25 is isotone . by applying ( ii ) to the restriction @xmath318 of @xmath25 to @xmath103^r$ ] , we obtain an @xmath33-element subset @xmath24 of @xmath1 which is free with respect to @xmath318 .
let @xmath316^{{{\leqslant}}r}$ ] . for each @xmath319 , from @xmath320 and @xmath321 it follows that there exists @xmath322^r$ ] such that @xmath323 . by applying this to all elements @xmath319
, we obtain that we can write @xmath324 , for a positive integer @xmath325 and subsets @xmath326 , , @xmath327 of @xmath24 .
as @xmath328 for each @xmath329 and as @xmath25 is isotone , we obtain that @xmath26 .
therefore , @xmath24 is free with respect to @xmath25 , and so ( ii ) holds .
as @xmath59 is a finite lattice with breadth @xmath330 , it follows from proposition [ p : kurgewbr ] that @xmath331 , for all integers @xmath14 and @xmath36 such that @xmath314 .
furthermore , as @xmath59 has exactly @xmath33 join - irreducible elements ( namely the singletons ) , it follows from proposition [ p : existskur(p ) ] that @xmath332 . on the other hand , according to the results of lzr , hajnal - mt , and hajnal cited in section [ s : intro ] , it follows from proposition [ p : freeb(k , m ) ] that the relation@xmath333 holds for all integers @xmath334 and @xmath36 .
in particular , @xmath335 , and thus , as the converse inequality holds , @xmath336 , whenever @xmath334 and @xmath337 . arguing as above , in the presence of @xmath37 we obtain from the relation @xmath38 ( cf .
* theorem 45.5 ) ) that @xmath336now for all integers @xmath14 and @xmath33 such that @xmath314 . without assuming @xmath37 , komjth and shelah s result ( cf .
section [ s : intro ] ) yields that @xmath338 may be larger than or equal to @xmath339 .
in particular , @xmath340 in any set - theoretical universe satisfying @xmath37 , while @xmath341 in some generic extension . and therefore , the _ kuratowski index function is not absolute _ ( in the set - theoretical sense ) .
we sum up in the following proposition some of the results above .
[ p : morekurb(r , m ) ] 1 .
@xmath342 for all integers @xmath14 and @xmath33 such that @xmath314 .
@xmath336 for all integers @xmath14 and @xmath33 such that @xmath314 and @xmath334 .
3 . assume that @xmath37 holds .
then @xmath336 for all integers @xmath14 and @xmath33 such that @xmath314 .
4 . there exists a model of @xmath343 where @xmath341 . while the integer @xmath344 of proposition [ p : morekurb(r , m ) ]
is quite large , smaller values yield new kuratowski indexes that were not available by using earlier methods .
the following result is due to the first author ( * ? ? ?
* lemme 3.3.12 ) .
the proof of the order - dimension exact estimate @xmath345 is contained in @xcite , however ( * ? ? ?
* lemme 3.3.12 ) provides an easy direct proof in that case .
of course it is obvious that the breadth of @xmath60 is @xmath61 , and so the conclusion of lemma [ l : orddimbrm ] follows from propositions [ p : uppbd ] and [ p : kurgewbr ] .
[ l : orddimbrm ] @xmath346 , for every positive integer @xmath11
. the following corollary is observed in ( * ? ? ?
* thorme 3.3.13 ) .
[ c : pierren+2 ] the relation @xmath347 holds for each infinite cardinal @xmath2 and each positive integer @xmath11 . in particular , this answers in the affirmative the question , raised on @xcite , whether @xmath52 .
these methods show that in order to prove the existence of large free sets for set mappings of type a positive integer @xmath14 ( as defined in ( * ? ? ?
* section 46 ) ) , it is sufficient to establish upper bounds for the order - dimension of finite lattices of the form @xmath59 .
this problem gets somewhat simplified by using the following easy result .
[ l : dim1tor ] the equality @xmath348 holds , for all integers @xmath33 and @xmath14 such that @xmath314 .
as @xmath59 contains @xmath349 , the inequality @xmath350 is trivial .
conversely , set @xmath351 and let @xmath352 be a product of chains ( any lattice would do ) with an order - embedding @xmath353 .
we set @xmath354 clearly @xmath355 is isotone .
let @xmath356 such that @xmath357 , we must prove that @xmath358 .
we may assume that @xmath359 .
in particular , @xmath360 , and thus , by part of the proof of proposition [ p : freeb(k , m ) ] , we can write @xmath361 , for a positive integer @xmath325 and @xmath362^r$ ] . for each @xmath363 and each @xmath364 , @xmath365 thus , as @xmath125 is an order - embedding , @xmath366 . as this holds for all possible choices of @xmath367 and @xmath368
, we obtain that @xmath358 .
therefore , @xmath355 is an order - embedding , and so @xmath59 order - embeds into the same product of chains as @xmath349 does .
getting estimates of the order - dimension of @xmath349 has given rise to a great deal of work , starting with dushnik @xcite .
further refinements can be found , for example , in kierstead @xcite .
we shall illustrate how large free sets can be obtained from small order - dimensions in proposition [ p : largefreesets ] .
we shall use dushnik s work @xcite . for integers
@xmath33 , @xmath14 with @xmath306 , dushnik denotes by @xmath369 the minimal number @xmath370 such that there exists a set @xmath371 of @xmath370 linear orderings of @xmath33 such that for each @xmath372^r$ ] and each @xmath373 , there exists @xmath374 such that @xmath375 for each @xmath376 .
then dushnik establishes in ( * ? ? ?
* theorem iii ) that @xmath377 in order to get estimates of @xmath378 , we shall use hajnal s work quoted in spencer s paper @xcite .
for integers @xmath11 and @xmath14 such that @xmath379 , spencer denotes by @xmath380 the maximal cardinality of an `` @xmath14-scrambling '' family of subsets of @xmath11 , and he establishes on ( * ? ? ? * lemma , p. 351 ) the inequality @xmath381 where @xmath54 denotes the largest integer below any real number @xmath55 .
furthermore , spencer establishes on @xcite the inequality @xmath382 now let @xmath11 and @xmath14 be integers such that @xmath383 , and set @xmath384 it follows from that @xmath385 , thus , by the isotonicity of the function @xmath370 , it follows from that @xmath386 , and thus , by , @xmath387 .
therefore , by lemma [ l : dim1tor ] , @xmath388 as well , and therefore , by proposition [ p : uppbd ] , @xmath389 .
now an immediate application of proposition [ p : freeb(k , m ) ] yields the relation @xmath390 , for each infinite cardinal @xmath1 .
this completes the proof of the following result .
[ p : largefreesets ] let @xmath11 and @xmath14 be positive integers with @xmath15 .
then the relation @xmath391 holds for each infinite cardinal @xmath2 .
denote by @xmath392 and @xmath393 the base two logarithm , resp .
the natural logarithm of a positive real number @xmath55 . set @xmath394 , for all integers @xmath11 , @xmath14 such that @xmath15 .
elementary calculations yield the asymptotic estimate @xmath395 we illustrate on table [ ta : smalle ] the behavior of the function @xmath396 on @xmath397 and ( relatively ) small values of @xmath11 .
the given values of @xmath11 are minimal for the corresponding values of @xmath398 : for example , @xmath399 while @xmath400 . the value
@xmath401 is the first one for which proposition [ p : largefreesets ] gives a nontrivial large free set result .
.height 12pt depth 6pt width 0ptsmall values of @xmath398 [ cols="<,>,>,>,>,>,>,>,>,>",options="header " , ] in particular , we deduce from proposition [ p : largefreesets ] the following relations : @xmath402 and so on .
smaller values of @xmath11 also give nontrivial large free sets , however , for those values the bounds that we shall discuss now are better .
we recall the following estimate of the order - dimension of @xmath349 .
this estimate is first stated explicitly in brightwell _ et al . _
* proposition 1.5 ) , where it is attributed to fredi and kahn @xcite .
however , the estimate is not stated explicitly in the latter paper , although the proof is an easy modification of the proof of that paper s ( * ? ? ?
* proposition 2.3 ) .
[ p : fuka ] let @xmath14 and @xmath33 be integers such that @xmath314 and let @xmath403 be a positive integer satisfying the inequality @xmath404 then @xmath405 . if @xmath405 , then , by lemma [ l : dim1tor ] , @xmath406 , thus , by proposition [ p : uppbd ] , @xmath407 , and thus , by proposition [ p : freeb(k , m ) ] , @xmath408 .
in particular , we obtain the relation @xmath409 which is better than the relation @xmath410 given by table [ ta : smalle ] .
however , for @xmath411 and @xmath412 the size of the free set given by proposition [ p : largefreesets ] becomes larger again .
we can even obtain further large free sets results by using dushnik s original exact estimate of the order - dimension of a truncated cube under certain conditions @xcite .
[ p : dush50 ] let @xmath33 , @xmath368 , and @xmath304 be positive integers such that @xmath413 , @xmath414 , and @xmath415 then @xmath416 . in the context of proposition [ p : dush50 ] , it follows from lemma [ l : dim1tor ] that@xmath417 , thus , by proposition [ p : uppbd ] , @xmath418 , and thus , by proposition [ p : freeb(k , m ) ] , the relation @xmath419 holds .
interesting values are obtained for @xmath420 and @xmath421 , giving respectively the relations @xmath422 for @xmath423 and @xmath424 , we obtain , respectively , @xmath425 of course , the `` origin '' @xmath71 can be changed , in any of these relations , to any infinite cardinal , so , for example , the relation @xmath426 holds for any infinite cardinal @xmath2 .
our work @xcite makes a heavy use of kuratowski indexes of finite lattices , in particular in order to evaluate `` critical points '' between quasivarieties of algebraic structures . now evaluating the kuratowski index of a finite poset may be a hard problem , even in the case of easily describable finite lattices .
for example , consider the finite lattices @xmath0 and @xmath80 represented in figure [ fig : lattices ] .
both @xmath0 and @xmath80 have breadth two and order - dimension three .
furthermore , @xmath0 embeds , as a poset , into @xmath80 , thus , by proposition [ p : kurindincr ] , @xmath223 .
therefore , by propositions [ p : uppbd ] and [ p : kurgewbr ] , we obtain the inequalities @xmath427 we do not know which one of those inequalities can be strengthened to an equality .
the statement @xmath428 is equivalent to _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ for every infinite cardinal @xmath2 and every @xmath429^{<{\omega}}\to[{\lambda}^+]^{<{\lambda}}$ ] , there are distinct @xmath430 such that @xmath431 , @xmath432 , and @xmath433 for all @xmath288 in @xmath434 , _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ for every infinite cardinal @xmath2 and every @xmath429^{<{\omega}}\to[{\lambda}^+]^{<{\lambda}}$ ] , there are distinct @xmath430 such that @xmath431 and @xmath436 for all @xmath288 in @xmath434 . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ such results could be relevant in further improving `` large free sets '' results such as corollary [ c : pierren+2 ] and proposition [ p : largefreesets ] . in particular , the gap between @xmath398 ( which gives a positive large free set result in proposition [ p : largefreesets ] ) and @xmath45 ( which gives komjth and shelah s negative large free set result ) looks huge .
can the bound @xmath398 be improved in proposition [ p : largefreesets ] ? on the other hand , we do not even know whether the relation @xmath437 holds .
p. erds , a. hajnal , a. mt , and r. rado , `` combinatorial set theory : partition relations for cardinals '' .
studies in logic and the foundations of mathematics * 106*. north - holland publishing co. , amsterdam , 1984 .
347 p. isbn : 0 - 444 - 86157 - 2 a. hajnal and a. mt , _ set mappings , partitions , and chromatic numbers_. logic colloquium 73 ( bristol , 1973 ) , p. 347379 . studies in logic and the foundations of mathematics , vol . * 80 * , north - holland , amsterdam , 1975 . | we introduce an extension , indexed by a partially ordered set @xmath0 and cardinal numbers @xmath1 , @xmath2 , denoted by @xmath3 , of the classical relation @xmath4 in infinite combinatorics . by definition
, @xmath5 holds if every map @xmath6^n\to[{\kappa}]^{<{\lambda}}$ ] has a @xmath7-element free set . for example , kuratowski s free set theorem states that @xmath8 holds iff @xmath9 , where @xmath10 denotes the @xmath11-th cardinal successor of an infinite cardinal @xmath2 . by using the @xmath3 framework , we present a self - contained proof of the first author s result that @xmath12 , for each infinite cardinal @xmath2 and each positive integer @xmath11 , which solves a problem stated in the 1985 monograph of erds , hajnal , mt , and rado .
furthermore , by using an order - dimension estimate established in 1971 by hajnal and spencer , we prove the relation @xmath13 , for every infinite cardinal @xmath2 and all positive integers @xmath11 and @xmath14 with @xmath15 .
for example , @xmath16 .
other order - dimension estimates yield relations such as @xmath17 ( using an estimate by fredi and kahn ) and @xmath18 ( using an exact estimate by dushnik ) . |
in this work we discuss the low frequency response of two - dimensional conductors subject to a perpendicular quantizing magnetic field@xcite and subject to time - dependent oscillating voltages applied to the contacts of the sample .
we use an approach@xcite@xmath0@xcite closely related to the scattering approach to dc conduction@xcite in a mesoscopic phase - coherent multiterminal sample .
the approach treats current and voltage contacts on equal footing and views the entire sample and the nearby conductors as one electric entity .
+ it is well - known from mesoscopic conduction theory that all dc resistances are rational functions of transmission probabilities for carriers to reach a contact if they are injected at another contact.@xcite moreover , only the states at the fermi surface which connect different contacts count . in a conductor subject to a quantizing magnetic field
the only states at a hall plateau which provide such a connection are the _ edge states _ which are the quantum mechanical generalization of the semi - classical ` skipping orbits ' of the charge carriers along the sample boundary . as a consequence
the dc conductance is quantized and determined by the topology of the edge - state arrangement.@xcite for a lucid discussion of the current patterns and hall potentials in the presence of transport , we refer the reader to a recent work by komiyama and hirai.@xcite + most of the early work on edge states relies on a picture valid for noninteracting electrons which must be modified in very pure samples with smooth boundaries.@xcite interaction leads to a decomposition of the electron gas into compressible and incompressible regions , where the former ones are associated with the edge states .
interaction must also be considered to describe edge states in the fractional quantum - hall regime.@xcite a detailed description of the contacts for chiral edge - state models has recently been provided by kane and fisher.@xcite while in these works the main concern is with short - range interactions , we emphasize in our work the long - range part of the coulomb interaction . for this case ,
edge states are important since they contribute to screening , in contrast to the bulk states.@xcite due to the coulomb interaction , the ac response is not only determined by the topology of the edge - state arrangement but also by its geometry . + in the following , we concentrate on a two - dimensional electron gas with spin - split landau levels .
coulomb interaction is taken into account by consideration of the capacitances of all edge channels and all gates .
we assume that additional contributions due to parasitic capacitances associated with contacts can be neglected , that the contacts are quantized , and that inter - edge scattering is absent .
the gates together with the edge channels which are separated by regions of incompressible electron gas can be seen as an arrangement of mesoscopic conductors embedded in a dielectric medium .
we describe an @xmath1-terminal sample at a hall plateau by an arrangement of such conductors @xmath2 connected to contacts @xmath3 ( see , e.g. , fig .
[ fig1 ] ) .
we are interested in the dynamical conductance @xmath4 which determines the fourier amplitudes of the current , @xmath5 , at a contact @xmath6 in response to an oscillating voltage @xmath7 at contact @xmath8 @xmath9 the voltage variation @xmath10 is related to the variation of the electro - chemical potential @xmath11 in reservoir @xmath12 by @xmath13 , where @xmath14 is the electron charge .
+ since all the present gates and conductors are included in the description , and due to the long - range nature of the coulomb interaction the theory for the conductance has to satisfy two basic requirements , namely _ current conservation _ and _ gauge invariance_. current conservation corresponds to zero total current , @xmath15 .
hence the total charge is conserved .
gauge invariance means that a global voltage shift , @xmath16 , does not affect the currents . in particular
, we fix the voltage scale such that @xmath17 corresponds to the equilibrium state where all electrochemical potentials @xmath18 are equal .
current conservation and gauge invariance demand the admittance matrix @xmath19 to satisfy @xmath20 .
for example , the admittance of a two - terminal device is determined by a single scalar quantity , @xmath21 . due to microreversibility the linear response coefficients
must additionally satisfy the onsager - casimir reciprocity relations @xmath22 where @xmath23 is the magnetic field .
we are interested in the low - frequency limit and write @xmath24 while the theory for the dc - conductances @xmath25 of quantized hall systems is well established,@xcite the _ emittance matrix @xmath26 has been investigated only recently.@xcite below we recall these results and , in addition , provide new results for the second order term @xmath27 .
to get some feeling for the emittance and the second order term , consider a two - terminal device with a macroscopic capacitor @xmath28 in series with a resistor @xmath29 .
for this purely capacitive structure with zero dc - conductance one finds that the emittance is the capacitance , @xmath30 , and that @xmath31 contains the @xmath32 time constant associated with charge relaxation .
these simple results must be modified for _ mesoscopic _ conductors and conductors which connect _ different _ reservoirs.@xcite firstly , it is not the geometrical capacitance but rather the _ electro - chemical capacitance _ which relates charges on mesoscopic conductors to voltage variations in the reservoirs .
secondly , conductors which connect different reservoirs allow a transmission of charge which leads to inductance - like contributions to the emittance .
_ an important property of a single edge channel is its uni - directional transparency and the absence of backscattering.@xcite the scattering matrix @xmath33 relates the out - going current amplitude at contact @xmath34 to the incoming current amplitude at contact @xmath12 . the part of the scattering matrix associated with edge channel @xmath35 can be written in the form @xmath36 with a scattering amplitude @xmath37 .
the energy dependence of the phase @xmath38 determines the density of states of the edge channel , @xmath39 at the fermi energy .
we mention that for noninteracting particles the density of states is related to the electric field at the edge.@xcite @xmath40 and @xmath41 are the injection and emission probability , respectively , of channel @xmath35 at contact @xmath34 : @xmath42 since an edge channel is always connected to one contact at each of its ends , one has @xmath43 .
furthermore , @xmath44 ( @xmath45 ) is the number of edge channels which enter ( leave ) the sample at contact @xmath34 .
clearly , the transmission probabilities are functions of the magnetic field @xmath23 and obey the micro - reversibility conditions @xmath46 .
this follows directly from the fact that an inversion of the magnetic field inverses the arrows attached to the edge states . in terms of the @xmath40 and
@xmath41 the dc conductance becomes@xcite @xmath47 hence the topology of the edge - channel arrangement determines the dc conductance .
note that current conservation , gauge invariance , and the onsager - casimir relations follow immediately from the properties of @xmath40 and @xmath41 .
while the dc - conductance is completely determined by the overall transmission , the ac - conductance contains information on the nonequilibrium state within the sample .
application of an oscillatory voltage at a sample contact leads to a periodic charging of the sample , and the induced electric fields have to be taken into account self - consistently . in order to treat the dynamic charge and potential distribution in the nonequilibrium state we spatially
discretize the two - dimensional electron gas .
more concrete , we assign to each edge channel @xmath35 a single electrochemical potential of the carriers , a single electrical potential , and a total charge .
we neglect spatio - temporal effects inside edge channels such as edge plasmons with finite wave numbers.@xcite however , our discussion takes into account the zero wave - number modes of the edge states .
the electrochemical potential shift @xmath48 in edge channel @xmath35 can be written in terms of the voltages @xmath49 in the reservoirs , @xmath50 . to simplify notation
we write quantities associated with edge channels as @xmath51-dimensional vectors , e.g. @xmath52 .
the nonequilibrium distribution of charges and electric potentials @xmath53 and @xmath54 , respectively , are related by @xmath55 , where @xmath56 is the geometrical capacitance matrix for the edge channels and gates . as usual , @xmath56 can be obtained from the poisson equation .
a relation between the charges and the voltages of the reservoirs , @xmath57 , defines the electrochemical capacitance matrix @xmath58 .
note that while @xmath56 is real and frequency independent , @xmath59 is a function of frequency . here
, we define @xmath56 by assuming that each edge channel is connected to its own reservoir , hence charge conservation and gauge invariance imply for both @xmath56 and @xmath60 the sum rules @xmath61 .
both matrices are symmetric and even functions of the magnetic field .
further we need the dynamic densities of states of the edge channels . in the present notation
they can be summarized in a diagonal matrix @xmath62 with elements @xmath63.@xcite here , @xmath64 is the dwell time of carriers in edge channel @xmath35 .
the specific form of the @xmath65 can be understood by noticing that the charging of edge channel @xmath35 is a relaxation process with a time @xmath66 .
the nonequilibrium charge is @xmath67 from which one obtains the dynamic electrochemical capacitance @xmath68 where @xmath69^{-1}{\bf c}$ ] is the static part of the electrochemical capacitance matrix .
the dependence of the effective capacitance @xmath70 on the densities of states is reasonable : small densities of states make it more difficult to charge the conductor . for large densities of states , on the other hand ,
one recovers the geometrical capacitance .
the low frequency admittance can be calculated with the help of a recent theory which generalizes the scattering approach to ac conduction.@xcite the ingredients which we need are the current response to an external voltage shift @xmath71 and the current response to an internal voltage shift @xmath72 for carriers at the fermi energy @xmath73 and for carriers with energies @xmath74 .
we do not enter into details here but present only the result .
we find @xmath75 using the micro - reversibility properties of the @xmath76 and the symmetry of the capacitance matrix it is again easy to prove current conservation , gauge invariance , and the onsager - casimir reciprocity relations .
the results ( [ eq6 ] ) and ( [ eq7 ] ) yield the admittance to order @xmath77 .
the emittance @xmath78 @xmath79 can be understood as the sum of all ` two - body ' coulomb interactions between edge states and/or gates.@xcite in a pure capacitive arrangement where @xmath80 the emittance matrix is a capacitance matrix , i.e. diagonal elements are positive and off - diagonal elements are negative .
as we will see below , this can change in the presence of edge channels which connect different reservoirs .
the result for the second - order term , @xmath81 @xmath82 @xmath83 , can thus similarly be interpreted as a ` three - body ' interaction with an intermediate step via edge channel ( or gate ) @xmath35 .
this simple result , of course , relies on the discrete model and is modified as soon as one includes internal spatio - temporal edge excitations .
= 50 mm = 50 mm
let us discuss the corbino disc and the quantum hall bar with @xmath84 channels along an edge as shown in fig .
[ fig1 ] for a magnetic field @xmath23 such that @xmath85 .
channels at the upper and the lower edges are labeled by odd and even numbers , respectively .
equation ( [ eq6 ] ) permits now to treat the case where different edge states are at different potentials . for the sake of clarity , however , we assume that the capacitances between edge channels at the same edge are very large such that they are at equal potentials .
we take thus each set together to form a single @xmath84-channel conductor associated with only even or only odd labels .
the geometric capacitance between the two sets of channels is denoted by @xmath28 ( @xmath86 ) .
while for the disc the reservoirs are capacitively coupled , for the bar all edge channels transmit charge between different reservoirs .
hence , the dc - conductance of the disc and the bar are @xmath87 and @xmath88 , respectively .
the dynamic capacitance between the two sets of edge channels is @xmath89 with a static electrochemical capacitance given by @xmath90 .
the charge relaxation resistance is given by @xmath91 due to the purely capacitive coupling the low frequency admittance of the corbino disc is @xmath92 as one expects , this result is equivalent to the low frequency admittance of a mesoscopic capacitor with @xmath84 channels on each side.@xcite on the other hand , one finds for the bar @xmath93 while the emittance of the disc behaves like a capacitance , the emittance of the bar is inductive - like .
inductive behavior is a typical property of conductors where transmission between different reservoirs dominates.@xcite the crossover between the two limits of a purely transmitting and a purely capacitive arrangement can be investigated by considering tunneling between edge channels . in a recent work ,
@xcite we have shown that the emittance of a quantum point contact without magnetic field exhibits a change of the sign as a function of the transparency .
+ note that if all @xmath94 are of the same order of magnitude , the charge relaxation resistance @xmath29 scales like @xmath95 .
nevertheless , if @xmath96 scales with @xmath84 ( as is the case on a hall plateau and for large @xmath28 ) one has @xmath97 . at a transition between two hall plateaus
, however , the density of states of the innermost edge channel diverges and one expects the charge relaxation resistance to be of the order of a single resistance quantum .
figure [ fig1]c . shows a two - terminal bar with two additional symmetric gates each located close to an edge state .
macroscopic gates have huge densities of states which means that @xmath98 , where @xmath99 is the geometric capacitance between a gate and the nearest edge channel .
note that the off - diagonal elements of the capacitance matrix are negative .
we assume that the other off - diagonal capacitance elements are much smaller , i.e. @xmath100 and @xmath101 .
now , let us ask for the current through contact @xmath102 in response to a voltage oscillation at gate @xmath103 .
we find to leading order of @xmath104 for the associated admittance @xmath105 with @xmath106 .
clearly , since edge channel @xmath107 emits into reservoir @xmath102 , and gate @xmath103 interacts only via gate @xmath108 with this edge channel , the response is proportional to @xmath109 and of second order in frequency . on the other hand , by inverting the magnetic field , edge channel @xmath102 emits into reservoir @xmath102 . now , the relevant capacitance turns out to be @xmath110 ; to leading order of @xmath104 we find @xmath111 . thus while @xmath112 exhibits a capacitive response given by @xmath113 , the capacitive response in @xmath114
is absent .
such a dramatic magnetic field asymmetry has been verified in an experiment by chen et al .
@xcite in the absence of gate @xmath108 , i.e. for @xmath115 .
+ the result ( [ eq7 ] ) implies that the conductance @xmath19 between two contacts which are capacitively coupled to the sample is a symmetric function of the magnetic field .
this is in general true up to @xmath116.@xcite to higher order in frequency , however , the magnetic - field symmetry of the current response can change in the presence of voltage probes . as an example
, we assume @xmath115 and that contact @xmath107 in fig .
acts as a voltage probe . using @xmath117
, we eliminate @xmath118 from the eqs .
( [ eq1 ] ) and obtain the reduced admittance matrix @xmath119 associated with the contacts @xmath102 , @xmath103 , and @xmath108 .
we find @xmath120 and @xmath121 .
hence , the impedance of a purely capacitive arrangement is in general not symmetric but only governed by the onsager - casimir symmetry relations , @xmath122 .
this conclusion has been drawn by sommerfeld et al.@xcite who found an asymmetric capacitance of a three - terminal corbino disc .
our theory predicts such an asymmetric behavior in a simple way . +
another interesting question concerns the hall resistance @xmath123 of the sample shown in fig . [ fig1]c .
if @xmath104 vanishes one expects exact quantization because the gates do nothing but measure the edge potentials .
an expansion to first order in @xmath104 yields the low frequency resistance @xmath124 in the case where the two gates are decoupled , one concludes indeed that the resistance is quantized , @xmath125 .
however , due to the coulomb coupling between the gates this result is modified even for the dc case .
the integer quantum hall effect@xcite corresponds to the quantization of the hall resistance and the vanishing of the longitudinal resistance of the ideal four - probe quantum hall bar of fig .
[ fig1]d . at zero frequency .
two of the contacts serve as voltage probes , whereas the two remaining contacts are used as source and sink for the current . with the help of our theory
the dc - results can be extended to the low - frequency case . if the contacts @xmath107 and @xmath108 are the voltage probes , the longitudinal resistance is defined by @xmath126 .
the hall resistance is defined by @xmath127 , provided the contacts @xmath103 and @xmath108 are voltage probes .
let us assume that the elements @xmath128 of the electrochemical capacitance matrix are known .
for the specific geometry of fig .
we assume @xmath129 and calculate the low frequency admittance . from eqs .
( [ eq1 ] ) , ( [ eq6 ] ) , and ( [ eq7 ] ) we find then the longitudinal resistance @xmath130 the leading term of the longitudinal resistance is determined by the coulomb coupling between the current circuit and the voltage circuit which are represented by edge channels @xmath102 and @xmath107 , respectively .
this is even true for finite @xmath131.@xcite moreover , for the specific symmetry @xmath132 the dissipation vanishes also in @xmath133 . for the hall resistance one finds @xmath134 we mention , that if we had chosen the symmetry@xcite @xmath135 , the imaginary part of the hall resistance would have vanished in first order of @xmath136 .
we presented a theory of ac - conductance for quantized hall samples which takes the coulomb interaction into account self - consistently .
it is based on a discretized model where each edge channel is described by a homogeneous mesoscopic conductor .
the whole sample is viewed as an electric entity and the results satisfy the important requirements of gauge invariance and current conservation .
obviously , the theory provides a manifold of experimental tests .
the internal dynamics enters by a single relaxation process associated with the uniform charging or decharging of edge channels with a certain relaxation time .
a next step should be to include internal dynamics due to spatio - temporal excitations@xcite along the edge channel within a microscopic theory .
this work was supported by the swiss national science foundation under grant nr .
99 k. von klitzing , g. dorda , and m. pepper , phys . rev
. lett . * 45 * , 494 ( 1980 ) .
t. christen and m. bttiker , phys .
b * 53 * 2064 , ( 1996 ) .
m. bttiker , j. phys . : condens .
matter * 5 * , 9361 ( 1993 ) .
m. bttiker , h. thomas , and a. prtre , phys .
a * 180 * , 364 ( 1993 ) ; z. phys .
b * 94 * , 133 ( 1994 ) . for a review see : m. bttiker , in ` semiconductors and semimetals ' , vol
. 35 , edited by m. reed , academic press , p.191 , ( 1992 ) .
s. komiyama and h. hirai , phys .
b * 54 * , july ( 1996 ) .
d. b. chklovskii , b. i. shklovskii , and l. i. glazman , phys .
b * 46 * , 4026 ( 1992 ) ; n. r. cooper and j. t. chalker , phys .
b * 48 * , 4530 ( 1993 ) ; k. lier and r. r. gerhardts , phys . rev .
b * 50 * , 7757 ( 1994 ) . c. w. j. beenakker , phys . rev .
lett . * 64 * , 216 ( 1990 ) ; a. m. chang , solid state commun . *
74 * , 871 ( 1990 ) ; a. h. mcdonald , phys .
lett . * 64 * , 222 ( 1990 ) ; x. g. wen , phys . rev .
b * 43 * , 11025 ( 1991 ) . c. l. kane and m. p. a. fisher , phys . rev .
b * 52 * , 17393 ( 1995 ) .
e. yahel , d. orgad , a. palevski , and h. shtrikman , phys .
lett . * 76 * , 2149 ( 1996 ) ; a. h. mcdonald , t. m. rice , and w. f. brinkman , phys .
b * 28 * , 3648 ( 1983 ) ; s. takaoka et al . ,
lett . * 72 * , 3080 ( 1994 ) . v. i. talyanskii et al .
, j. phys . : condens .
matter * 7 * , l435 ( 1995 ) .
p. k. h. sommerfeld , r. w. van der heijden , and f. m. peeters , phys .
b * 53 * , 13250 ( 1996 ) . a. prtre , h. thomas , and m. bttiker , phys . rev .
b ( unpublished ) .
t. christen and m. bttiker , phys .
lett . * 76 * , july ( 1996 ) .
w. chen , t. p. smith iii , m. bttiker , and m. shayegan , phys .
lett . * 73 * , 146 ( 1994 ) . | in the framework of the edge - channel picture and the scattering approach to conduction , we discuss the low frequency admittance of quantized hall samples up to second order in frequency .
the first - order term gives the leading order phase - shift between current and voltage and is associated with the displacement current .
it is determined by the emittance which is a capacitance in a capacitive arrangement of edge channels but which is inductive - like if edge channels predominate which transmit charge between _ different _ reservoirs .
the second - order term is associated with the charge relaxation .
we apply our results to a corbino disc and to two- and four - terminal quantum hall bars , and we discuss the symmetry properties of the current response . in particular , we calculate the longitudinal resistance and the hall resistance as a function of frequency . |
starting point of our analysis are toroidal silica microcavites@xcite that intrinsically combine ultra - high-@xmath0 optical whispering gallery modes with around 20 mechanical modes in the 0 - 100 mhz range@xcite which are observable in an interferometric readout using the structure s optical modes ( see methods ) .
recent work has shown that the intrinsic coupling of optical and mechanical modes of toroidal microcavities via radiation - pressure can give rise to the effect of dynamical backaction@xcite , which allows realization of narrow bandwidth photonic oscillators@xcite or photonic rf down - converters@xcite as well as radiation pressure cooling of a mechanical oscillator@xcite . the radial breathing mode ( cf .
1a ) in particular exhibits strong optomechanical coupling , low effective mass ( @xmath2 ) and high frequency enabling the first demonstration of resolved sideband cooling of a micromechanical oscillator@xcite . to understand their mechanical @xmath0-factors which have remained completely unexplored so far , we first present a study of mechanical dissipation in microtoroids and subsequently show how the results from this study can be harnessed to devise structures with unprecedentedly low dissipation in the frequency range above @xmath3 . + in order to assess the contribution of clamping losses , the diameter of the silicon pillar holding the silica resonator ( cf . fig .
1a ) is varied . to this end
consecutive @xmath4 etching cycles , undercutting the silica structure but leaving the silica itself unaffected , are applied .
thus , the relative undercut @xmath5 ( @xmath6 : radius of the cavity , @xmath7 : length of the free standing membrane , cf .
inset of fig .
1a ) can be controlled .
the dependence of the measured @xmath0-factors of the radial breathing mode ( rbm ) on the relative undercut is depicted in fig .
intuitively , it may be expected that higher @xmath0-factors are attained for larger undercut due to a reduced clamping area .
interestingly , however , the @xmath0-factors of six different microresonators of similar size show a strongly non - monotonous dependence on the relative undercut which contradicts the simple expectation of smaller losses for smaller clamping areas .
moreover , the behaviour is remarkably reproducible for the different samples . a plot of the measured frequencies along with an fem simulation ( see methods ) of the toroids radially symmetric mechanical modes depicted in fig .
1a shows excellent agreement of measured and simulated frequencies .
furthermore , the fem simulation indicates an avoided crossing between different mechanical modes as first predicted by mindlin in 1951@xcite .
a measurement with a different sample allows _ directly _ observing this avoided crossing by employing highly sensitive quantum limited optical displacement sensing ( see methods ) which allows to also monitor the flexural modes exhibiting weaker optomechanical coupling .
the frequencies and the corresponding @xmath0-factors of both the radial breathing mode and a flexural mode obtained in this second measurement are shown in fig .
the frequencies of both modes decrease for larger relative undercut , where the flexural mode shows a steeper slope .
thus , it can be observed that the eigenfrequencies of both modes approach each other but , as predicted by fem simulations , they do not cross .
moreover , both mode patterns hybridize and eventually swap their dispersion lines during the avoided crossing ( inset of fig . [
it is exactly in the region in which both modes couple , where the @xmath0-factor of the rbm is found to be strongly reduced .
this distinctive behaviour can be well explained with a generic model of two coupled harmonic oscillators @xmath8 , representing the bare radial ( r ) and flexural ( f ) mode , with angular frequencies @xmath9 and damping rates @xmath10 .
an offset of the silica torus from the equatorial plane of the silica disk@xcite as well as the and measured @xmath0-factors .
* mechanical @xmath0-factors ( dots with error bars , left axis ) and corresponding simulated @xmath11 values ( dashed lines , right axis ) for the radial breathing and a flexural mode of an on - chip toroid show a linear relation in the parameter range depicted where @xmath12 .
inset : mechanical @xmath0-factors depending on the background gas pressure .
below 1 mbar the gas damping contribution becomes negligible .
the black and red lines correspond to fits in the molecular ( @xmath13 ) and viscous ( @xmath14 ) gas damping regime . ]
single - sided clamping of the disk to the silicon pillar gives rise to an appreciable coupling rate @xmath15 between both modes .
the coupled eigenvalues of the system then read @xmath16 and thus the coupled eigenfrequencies @xmath17 and @xmath0-factors @xmath18 can be extracted . assuming a linear dependence of the bare angular frequencies @xmath19 and @xmath0-factors @xmath20 on the undercut @xmath21
, these can be used to asymptotically fit the data ( cf . fig .
1b ) . using the values
obtained from this first fit , only @xmath15 remains as free parameter in equation ( [ coupledequation ] ) .
a least square fit to both the measured frequencies and @xmath0-factors yields a coupling rate of @xmath22 , much larger than the damping rates @xmath23 .
the corresponding coupled frequencies @xmath24 and @xmath0-factors @xmath25 match the data remarkably well ( cf . fig .
1b ) confirming that indeed both modes behave as two coupled harmonic oscillators giving rise to an avoided crossing between the rbm and the low-@xmath0 flexural mode .
moreover , the data presented here , to the authors best knowledge , manifest the first direct observation of normal mode coupling of two different mechanical modes within a micromechanical resonator
. + for the @xmath0-factor of the rbm the avoided crossings are detrimental .
the hybridization with the low-@xmath0 flexural modes strongly reduces the mechanical quality of the rbm in the coupling region due to enhanced displacement amplitudes in the clamping area .
drawing on the quantitative power of fem simulations , it is indeed possible to show that the displacement amplitudes at the pillar strongly influence the mechanical @xmath0-factor . in order to study and elucidate the role of clamping losses the parameter @xmath11 , defined as : @xmath26 is introduced , where @xmath27 is the total energy stored in the oscillator , @xmath28 denotes the density of silica , and @xmath29 the speed of sound in silica , @xmath30 is the mechanical oscillation frequency and @xmath31 denotes the out of plane oscillation amplitude at position @xmath32 , with the integration extending over the clamping area @xmath33 .
@xmath11 [ t ! ] is the mechanical quality factor that is expected if the clamping area @xmath33 is modelled as a membrane , radiating acoustic energy with a power of @xmath34 .
this acoustic loss then leads to a @xmath0-factor of @xmath35 as given by equation ( [ eq : d ] ) .
note that a similar expression can be derived following a phonon tunneling approach@xcite .
2 the @xmath0-factors measured for two different modes of one microresonator are shown together with the simulated values of @xmath11 .
strikingly , @xmath11 is directly proportional to the measured @xmath0-factors , where @xmath12 , revealing good agreement of measured mechanical loss and fem simulations .
this unambiguously shows that the mechanical quality factors are clamping loss limited .
the prefactor of @xmath36 may be attributed to impedance matching conditions between the silica disk and the silicon pillar which are not taken into account in this model . note that these , and all following measurements , were performed in a vacuum chamber ( @xmath37 ) to fully eliminate the possible influence of gas damping ( cf .
2 ) . besides explaining the origin of the mechanical dissipation of different modes and proving that plainly reducing the clamping area to a minimum is _ not _ sufficient for optimized @xmath0-factors the parameter @xmath11 moreover allows _ a - priori _ modelling of the oscillator s mechanical quality .
this constitutes a powerful , direct tool to design high-@xmath0 mechanical oscillators with negligible clamping losses , which we have used to devise the novel structures described below .
+ one such geometry is shown in fig .
3a consisting of a toroid resonator with four spokes connecting it to the silicon support .
incorporation of the silica spokes into a toroidal microresonator is possible by following the fabrication steps as shown in fig .
importantly , the design allows independent control of the spokes width , length and position for any major and minor diameter of the optical toroid microcavities yielding further degrees of freedom to tune their mechanical frequencies .
moreover , using the aforementioned fem model of the clamping loss limited @xmath0-factors , i.e. maximizing @xmath11 , the geometry can be chosen such that the discrete eigenfrequencies of the spokes decouple the toroid s radial motion from the center of the silica disk , strongly mitigating clamping losses .
[ f:4]a shows the dependence of the measured @xmath0-factors on the simulated values of @xmath11 for several different samples of vastly different geometries with and without spokes .
indeed , maximizing @xmath11 using the spokes support allows obtaining reproducibly increased mechanical @xmath0-factors .
the quality factors which can be achieved at room temperature are boosted by approximately a factor of ten , e.g. a value of 50,000 is attained at @xmath38 as shown in fig .
the corresponding clamping loss limit as described by our model parameter @xmath11 can be pushed beyond the @xmath39-level using an optimized design . yet , as shown in fig .
[ f:4]a a saturation of the @xmath0-factors becomes evident and the data allow a fit according to a saturation model @xmath40 .
further loss mechanisms which will be addressed below become dominant for samples with strongly reduced clamping losses and for room temperature operation set an upper limit of @xmath41 .
the slope of the linear part @xmath42 obtained from this fit is slightly larger than the one obtained from fig .
[ f:2 ] ( dashed line in fig .
[ f:4]a ) .
this can be attributed to different impedance matching conditions as the sample of fig .
[ f:2 ] was fabricated using a @xmath43 silica preform whereas the new samples in fig .
[ f:4 ] consist of a @xmath44 silica layer .
+ it is emphasized that the optical properties of the toroid microresonators are not affected by the optimized mechanical design .
the @xmath45 laser assisted reflow process@xcite creating the surface tension induced surface finish of the silica toroid can still be applied and the of a @xmath46 sample as a function of temperature is shown .
the damping decreases from @xmath47 at room temperature to a record value of @xmath48 at @xmath49 . a fit ( red line ) including losses due to tls ( @xmath50 ) , anharmonicity ( @xmath51 ) and a constant background ( @xmath52 ) yields excellent agreement with the data and allows extracting an upper bound of @xmath53 for the remaining clamping losses .
further suppression of the intrinsic losses could enable @xmath0-factors approaching the extracted clamping loss limit of @xmath54 .
inset : damping @xmath55 of a different sample at @xmath56 illustrating the reliability of the fit ( red line ) of the three aforementioned contributions over a wide temperature range from @xmath57 to @xmath58.,title="fig:"][t ! ] subsequent @xmath4 underetch does not affect the optical quality . as a result , the thermal bistability and optical mode - splitting@xcite , salient features of ultra high-@xmath0 silica microtoroids@xcite , can be observed in the novel spoke - supported microresonators . to illustrate this point , fig .
[ f:4]c shows the transmission spectrum of a spokes resonator with an unloaded optical finesse of @xmath59 displaying mode splitting@xcite . at the same time
, the mechanical design is not adversely affected by the laser reflow process , even for very small spokes ( cf .
thus , optical and mechanical properties of microtoroidal optomechanical resonators can be decoupled and optimized _
independently_. the mechanical frequencies and @xmath0-factors are controlled by choosing appropriate spokes dimensions without affecting the choice of minor and major diameter of the toroid defining its optical frequencies and finesse@xcite . + in order to elucidate the apparent room temperature limit to the mechanical quality of @xmath60 and to quantify contributions of different origin , we studied the temperature dependence of the @xmath0-factors . in fig .
[ f:5 ] the damping of the rbm is depicted as a function of temperature .
the damping of @xmath47 measured for this sample at @xmath61 and room temperature continuously decreases when the temperature is increased . in particular , at @xmath49 a record value of @xmath62 is attained , on par with the best published values in this frequency range@xcite . besides demonstrating a further reduction of dissipation , the observed temperature dependence allows a detailed understanding of the limitations to the mechanical quality . as will be shown below , these are due to dissipation caused by two - level systems ( tls ) common to amorphous materials@xcite and anharmonicity@xcite .
the inset in fig .
[ f:5 ] shows the damping @xmath63 obtained with a @xmath64 sample measured from 10 to @xmath58@xcite .
a two parameter fit of the tls losses together with the anharmonic contribution ( no adjustable parameters ) and a constant background ( clamping losses ) yields excellent agreement with the data over the full temperature range as well as with the parameter range reported in experiments conducted with bulk silica@xcite ( see methods ) . using the same fit for the high temperature data ( @xmath65 to @xmath49 ) displayed in fig .
[ f:5 ] also yields excellent agreement .
moreover , the extracted temperature independent background sets an upper bound for remaining clamping losses of @xmath53 experimentally confirming that a strong reduction of clamping losses is achieved . for this specific sample
the @xmath11 model predicts six times lower clamping losses .
thus , it is likely that asymmetries in the structure cause additional clamping losses not taken into account in the above fem model leading to the observed background of @xmath66 .
it is noted that thermoelastic damping@xcite for which an upper bound of @xmath67 is found for the resonators considered here , can safely be neglected in this analysis .
the intrinsic losses of silica ( i.e. those caused by tls and anharmonicity ) are known to exhibit a local minimum at a temperature of around @xmath68@xcite and thus a further increase in temperature is expected to yield even higher @xmath0-factors than those reported here .
+ the quantification of remaining clamping loss limit yielding values above @xmath69 is also very important for the prospects of future low temperature experiments , as the intrinsic losses in amorphous materials can be strongly reduced by cryogenic operation@xcite . after reaching a maximum at @xmath70
the intrinsic losses decrease again at lower temperatures ( cf .
data in fig .
[ f:4 ] ) .
it is well known@xcite that the @xmath0-factors reach a plateau of @xmath71 before rapidly increasing below a drop - off temperature @xmath72 which shows a strong frequency dependence . for khz frequencies ,
the drop - off temperature resides in the @xmath73-range@xcite and therefore silica is commonly not considered as a suitable material for low temperature oscillators .
it is noted , however , that for high oscillation frequencies the drop - off temperature strongly increases and for e.g. @xmath74 ( @xmath75 ) in @xcite the intrinsic loss limited @xmath0-factors were measured to reach 50,000 at @xmath76 while quality factors exceeding 100,000 may be extrapolated for a temperature of @xmath77@xcite .
thus , the spoke supported resonators are very promising optomechanical devices for cryogenic operation with the prospect of even improving the demonstrated outstanding room temperature @xmath0-factors at low temperature , possibly approaching the extracted clamping loss limit beyond @xmath78 .
in summary , we presented novel micron - scale on - chip optomechanical resonators that for the first time allow independent control of optical and mechanical degrees of freedom within one and the same device .
avoided crossings of mechanical modes ( as first predicted by mindlin@xcite ) were observed which enabled a detailed understanding of mechanical dissipation , allowing the design of optomechanical systems combining mechanical @xmath0-factors of 80,000 , on par with the best published results in the frequency range above 30 mhz , with ultra - high optical finesse .
this property yields improved performance for applications such as realizing narrow band filters and narrow line - width radiation pressure driven``photonic clocks''@xcite .
yet , it is also an essential ingredient in studies which seek to demonstrate quantum signatures of mesoscopic objects@xcite . to illustrate this point , the ratio of displacement spectral density provided by radiation pressure quantum back - action and the thermal noise spectral density is given@xcite : @xmath79 = \frac{4 \xi^2}{1 + 4 \omega_\mathrm{m}^2/\kappa^2}\frac{\hbar\ , q\ , \mathcal{f}^2}{\lambda ( c / n)\ , \pi\ , k_b t\ , m \omega_\mathrm{m}}\,p$ ] ( @xmath80 for a linear cavity and @xmath81 for a ring - cavity , @xmath82 : power impinging on the cavity ) .
for cryogenic operation at temperature @xmath83 , a @xmath0-factor of 80,000 ( which seems feasible given our results and the measurements at @xmath84 in @xcite ) and the realistic parameters @xmath85 ( wavelength ) , @xmath86 ( effective mass ) , @xmath87 ( mechanical frequency ) , @xmath88 ( finesse ) and a cavity radius of @xmath89 ( the cavity bandwidth @xmath90 for a toroidal cavity is then given by @xmath91 , where @xmath29 denotes the speed of light in vacuo and @xmath92 the refractive index of silica ) this expression reaches unity already at a launched power @xmath82 as low as @xmath93 .
thus , radiation pressure quantum back - action which so far has remained experimentally inaccessible may get within reach .
moreover , at a bath temperature of @xmath83 , and for a mechanical @xmath0-factor of 80,000 already one tenth of the radiation - pressure cooling rates of @xmath94 demonstrated in silica microtoroids@xcite would enable laser cooling of the rbm in the resolved sideband regime to an occupation number of @xmath95 , below unity , and would thus allow to venture into the regime of cavity _ quantum _ optomechanics .
+ pertaining to the wider implications of our work , the studied aspects of geometry dependent mechanical dissipation via mode coupling ( which is also important for nanomechanical beam resonators@xcite ) and the demonstrated quantification of clamping losses can be of broad interest and could directly be applicable to other whispering gallery mode resonators made of amorphous@xcite and crystalline material@xcite as well as to a wide range of other micro- and nanomechanical oscillators , such as photonic crystal based systems@xcite which are also expected to exhibit optomechanical coupling .
to monitor the mechanical modes thermal noise , laser light is coupled to the high-@xmath0 optical whispering gallery mode of the silica resonators using a tapered optical fibre .
an external cavity diode laser is locked detuned by approximately half the optical line - width @xmath96 with respect to a cavity resonance @xmath97 .
thus , the small , thermal fluctuations @xmath98 of the cavity radius cause changes in the optical path - length which imprint themselves linearly into an amplitude modulation @xmath99 of the power transmitted by the cavity , i.e. @xmath100 .
spectral analysis of the transmitted power thus allows extracting the mechanical frequencies @xmath101 and dissipation rates @xmath102 from which the corresponding quality factor @xmath103 is inferred . to minimize radiation pressure induced back - action , i.e. to rule out any amplification@xcite or cooling@xcite which would lead to a modification of the natural line - width@xcite ,
the laser power used in these measurements is kept at a very low level , well below the threshold for the parametric instability@xcite . as an additional check the measurements
are always performed both on the red and blue detuned wing of the optical resonance .
the @xmath0-factors from both measurements usually agree within less than @xmath104 , indicating negligible radiation pressure backaction . + in order to achieve a higher displacement sensitivity allowing to observe further mechanical modes of the microresonators , and in particular the avoided crossing in fig .
[ f:2]b ) , an adapted version of the hnsch - couillaud polarization spectroscopy@xcite using a nd : yag laser locked to cavity resonance is employed@xcite .
the polarization of the field injected into the taper is chosen such that only a small portion couples to a polarization non - degenerate whispering gallery mode of the cavity while the larger part bypasses the cavity serving as a local oscillator . analyzing the phase - shift of the field exiting the cavity via a homodyne detection a shot - noise limited sensitivity @xmath105}=\frac{\lambda}{8 \pi \mathcal{f}}\sqrt{\frac{\hbar \omega}{p}}\sqrt{1 + 4\omega_\mathrm{m}^2/\kappa^2}$ ] of order @xmath106
is achieved at the frequencies of interest from 10 - 100 mhz@xcite , where @xmath82 denotes the power impinging on the cavity .
thus , not only the radial breathing mode but more than twenty different mechanical modes of the toroid between 0-@xmath107 can be monitored .
all observed modes can be identified with the according mode pattern obtained by fem simulation with a relative frequency deviation of less than @xmath104@xcite . a commercial software package ( comsol multiphysics )
is used for 3d finite element modelling of the cavity structures . using fixed boundary conditions at the bottom of the conical silicon pillar ,
the eigenvalue solver provided by the package calculates the eigenfrequencies @xmath30 and the corresponding stress and strain fields of the meshed structure by diagonalizing its coupled linear equations of motion .
summing over all elements the total mechanical energy @xmath27 of the mode can be calculated .
typically 100,000 elements are used where in the clamping region the characteristic size of an element is of order @xmath108 . in order to calculate the
parameter @xmath11 , the squared out of plane displacement amplitude @xmath109 is summed over the plane connecting the silica resonator and the silicon substrate .
+ for fitting the data in fig .
5 we employed the known model for losses caused by two level systems @xmath110 and anharmonic damping @xmath111 . the expression for @xmath110 originating from microscopic two level systems described by a distribution of double - well potentials reads@xcite : @xmath112 where @xmath113 is the error function , @xmath114 denotes the temperature and @xmath115 the oscillation frequency .
@xmath116 is a scale for the distribution of the potential barrier @xmath117 separating both wells , and @xmath118 scales the asymmetry in the depth of the wells ( all in temperature units ) .
@xmath119 is a dimensionless exponent , @xmath120 a relaxation time and @xmath121 a dimensionless parameter .
measurements spanning from @xmath122 to above @xmath123 as well as from a few up to a few hundred kelvin allow an excellent fit using the parameters @xmath124 , @xmath125 , @xmath126 , @xmath127 and @xmath128@xcite . for the fit presented in this work
we adopt the above mean values for @xmath119 , @xmath116 and @xmath129 while using @xmath120 and @xmath121 as fit parameters .
the fitted values are @xmath130 ( low temperature fit at @xmath131 ) , @xmath132 ( high temperature fit at @xmath61 ) and @xmath133 , falling within the parameter range as specified in@xcite .
the anharmonic contribution @xmath51 is given by@xcite : @xmath134 here , @xmath135 denotes the debye and @xmath136 the sound velocity . for the fit in fig .
5 we adopt @xmath137@xcite , the grneisen parameter @xmath138@xcite , the mean lifetime of thermal phonons @xmath139 determined in@xcite and the known values for the specific heat per unit volume @xmath140 , the sound velocity @xmath141 and the density @xmath28 of silica ( e.g. @xcite ) without any fit parameter . + tjk acknowledges support via an independent max planck junior research group grant , a marie curie excellence grant ( jrg - uhq ) , the dfg funded nanosystems initiative munich ( nim ) and a marie curie reintegration grant ( rg - uhq ) .
the authors gratefully acknowledge j. kotthaus , a. rogach and f. brsgens for access to clean - room facilities for micro - fabrication , a. marx for assistance in scanning electron microscopy and i. wilson - rae for stimulating discussions . | cavity - enhanced radiation - pressure coupling of optical and mechanical degrees of freedom gives rise to a range of optomechanical phenomena , in particular providing a route to the quantum regime of mesoscopic mechanical oscillators .
a prime challenge in cavity optomechanics has however been to realize systems which simultaneously maximize optical finesse and mechanical quality .
here we demonstrate for the first time independent control over both mechanical and optical degree of freedom within _ one and the same _ on - chip resonator .
the first direct observation of mechanical normal mode coupling in a micromechanical system allows for a quantitative understanding of mechanical dissipation .
subsequent optimization of the resonator geometry enables intrinsic material loss limited mechanical @xmath0-factors , rivalling the best values reported in the high mhz frequency range , while simultaneously preserving the resonators ultra - high optical finesse . besides manifesting a complete understanding of mechanical dissipation in microresonator based optomechanical systems , our results provide an ideal setting for cavity optomechanics . over the past years
it has become experimentally possible to study the coupling of optical and mechanical modes via cavity enhanced radiation pressure , which gives rise to a diverse set of long anticipated optomechanical phenomena@xcite such as radiation pressure driven oscillations@xcite and , as demonstrated in 2006 , dynamic backaction cooling@xcite .
moreover , this coupling can be exploited to perform highly sensitive measurements of displacement@xcite which may enable the observation of radiation pressure quantum backaction@xcite or related phenomena .
major goals in the emerging field of cavity - optomechanics@xcite , such as ground - state cooling@xcite necessitate high optical finesse and high mechanical quality factors at mechanical oscillation frequencies exceeding the optical cavity s linewidth .
while recently impressive progress has been made in creating experimental settings in which radiation pressure effects can be studied@xcite , a prime challenge still concerns attaining simultaneously high optical finesse and high mechanical @xmath0-factors .
nearly all approaches so far have combined traditional cm - sized optical elements ( mirrors ) with micro- or nanoscale mechanical oscillators which simultaneously act as mirrors@xcite ( or as dispersive element@xcite ) .
other approaches have used the intrinsic mechanical modes of the optical elements@xcite . yet , in general , the required high reflectivity of the micro - element limits its dimension to wavelength size and thus sets an upper limit to the mechanical frequencies that can be achieved . also , it is exceedingly difficult to attain high mechanical quality factors while maintaining high optical reflectivity as independent control of optical and mechanical degrees of freedom is generally not possible .
thus , although remarkably high mechanical @xmath0-factors at low frequencies have been obtained@xcite , this and previous approaches@xcite have so far not succeeded in combining mechanical @xmath0-factors comparable to those achieved in the field of nems and mems@xcite with the best values of optical finesse@xcite .
but exactly the combination of both is important for applications such as low loss , narrow - band `` photonic clocks''@xcite and indispensable for fundamental studies aiming at approaching and detecting quantized motion in mesoscopic optomechanical systems@xcite .
+ here we show for the first time independent control over both optical and mechanical degree of freedom in _ one and the same _ microscale optomechanical resonator .
the demonstration of mechanical normal mode coupling@xcite within a micromechanical device and the concomitant geometry dependence of clamping losses allows a _ quantitative _ understanding of mechanical dissipation .
we demonstrate monolithic spoke - supported silica resonators comprising a toroidal boundary which allows ultra - high optical finesse ( @xmath1)@xcite rivalling the best values obtained in fabry - perot cavities@xcite .
independent control of their mechanical properties leads to strongly reduced clamping losses allowing for unprecedented mechanical quality factors ( e.g. 80,000 at 38 mhz ) rivalling the best published values of strained silicon nitride nanoresonators@xcite as well as radial contour - mode disk resonators@xcite at similarly high frequencies which allow entering the resolved sideband regime@xcite using state - of - the - art optical cavities .
it is shown that the measured @xmath0-factors are only limited by temperature dependent intrinsic dissipation@xcite , which can be reduced by low temperature operation . moreover ,
the observed geometry dependent loss mediated by normal mode coupling may be of relevance across a wide range of micro- and nanomechanical oscillators , for which a detailed understanding of dissipation is lacking . |
a few years ago wilczek and collaborators proposed a new method to derive the hawking flux from spherical black holes@xcite .
they have proved , using a dimensional reduction procedure , that near the event horizon effective field theories become two - dimensional and exhibit gauge and gravitational anomalies .
they showed , in the unruh vacuum , that the hawking flux emerges in order to cancel these anomalies .
recent applications and a modification of this method can be found in refs . . a more technically simplified version of the method for asymptotically flat spacetimes , using only the covariant form of the anomalies ,
was suggested in ref . .
the purpose of this brief article , which is based in ref . , is to apply this method of covariant anomalies in asymptotically non - flat spacetimes of non - spherical topology .
consider the four - dimensional black hole solutions of einstein s equations with negative cosmological constant @xmath0 .
these asymptotically locally anti - de sitter ( ads ) black holes are known as topological black holes ( tbhs)@xcite and their metric is @xmath1 where @xmath2 is the ads radius and @xmath3 is a constant proportional to the black hole mass .
their topology is @xmath4 , where @xmath5 is a two - dimensional manifold of constant negative curvature .
the angular part of , with @xmath6 and @xmath7 , is the line element of the manifold @xmath5 , which is locally isomorphic to the hyperbolic manifold @xmath8 and of the form @xmath9 , where @xmath10 is a discrete subgroup of isometries of @xmath11 .
the manifold @xmath5 is a compact two - dimensional surface of constant negative curvature and of genus @xmath12@xcite .
below , i will concentrate only on the genus @xmath13 case and deal with the event horizon @xmath14 , which corresponds to the outer horizon for @xmath15 and to the unique horizon for @xmath16 .
now consider a complex scalar field @xmath17 in the background of the tbh of genus two .
its action is @xmath18 where the first term is the free part of the action and @xmath19 is the part which includes a mass term , potential terms and self - interaction terms .
i perform a partial wave decomposition of @xmath20 in terms of the functions @xmath21 , which are defined as @xmath22 where @xmath23 are the associated legendre functions , @xmath24 and @xmath25 takes discrete real values @xmath26 . these functions
form a complete orthonormalized set of functions on the manifold @xmath5 of genus two@xcite . upon transforming to the `` tortoise ''
coordinate @xmath27 defined by @xmath28 and taking the near - horizon limit @xmath29 , one finds that the effective radial potentials for partial wave modes of the field , the mass term and the interaction terms contain the suppression factor @xmath30 and vanish exponentially fast@xcite . keeping only dominant terms ,
i find that physics near the event horizon can be effectively described by an infinite collection of two - dimensional free massless scalar fields , each propagating in a two - dimensional spacetime , which is given by the @xmath31 part of the tbh metric of @xcite . in this reduced two - dimensional spacetime , following the ideas of wilczek et al.@xcite , i neglect the ingoing modes in the region near the event horizon , which behave as left - moving modes , since they can not classically affect physics outside the horizon .
then , the effective two - dimensional theory exhibits a gravitational anomaly in this region , in the form of the non - conservation of the energy - momentum tensor@xcite .
the time component of the covariant gravitational anomaly is@xcite @xmath32 @xmath33 is the covariant energy - momentum tensor and the prime denotes differentiation with respect to @xmath34 . solving this equation with the proposed in refs .
boundary condition @xmath35 , i find the non - conserved energy - momentum tensor @xmath36 the hawking flux is measured at infinity , where there is no gravitational anomaly and in refs .
it was given by the conserved energy - momentum tensor .
one sees that the gravitational anomaly @xmath37 vanishes at asymptotic infinity .
however , @xmath38 is non - zero there , due to the fact that the spacetime asymptotically is ads .
these remarks and , which can be written as @xmath39 , imply that the conserved energy - momentum tensor , or the energy flux @xmath40 measured at infinity , is @xmath41 the energy flux @xmath40 has a form equivalent to blackbody radiation of temperature @xmath42@xcite , which is exactly the hawking temperature of the tbh of @xcite .
hence , @xmath40 can be identified with the hawking flux from this tbh .
however , i should note that the actual hawking radiation observed at infinity is calculated through the grey - body factors , due to the curvature of the spacetime away from the horizon@xcite .
another interesting non - spherical black hole solution is a tbh conformally coupled to a scalar field @xmath43 .
this black hole solution is known as the mtz black hole@xcite @xmath44 where @xmath45 is newton s constant and @xmath46 is a constant proportional to the black hole mass .
the mtz black hole has the same topology as the tbhs previously analyzed .
therefore , i will concentrate again only on the case of @xmath5 being a genus two compact surface of constant negative curvature and deal with the event horizon @xmath47 , which is the unique horizon for @xmath48 and the outer horizon for @xmath49 . considering a complex scalar field @xmath20 in the background of the mtz black hole of genus two , which does not interact with the scalar field @xmath43 , the dimensional reduction procedure proceeds as in
. all the arguments of hold also in this case@xcite and the conserved energy - momentum tensor is @xmath50 this energy flux has a form equivalent to blackbody radiation of temperature @xmath51@xcite , which is the hawking temperature of the mtz black hole .
the conformally coupled scalar field @xmath43 , which is time - independent , does not contribute to the hawking flux .
this was proved explicitly in ref . by showing that its action near the horizon vanishes , due to the presence of the suppression factor @xmath30 .
conclusively , i derived the hawking temperature and flux , without backscattering effects , from non - spherical black holes in asymptotically non - flat spacetimes , via the method of covariant anomalies , and showed that an infinite set of two - dimensional scalar fields near the horizon acts as the thermal source of this flux .
robinson and f. wilczek , _ phys .
lett . _ * 95 * , 011303 ( 2005 ) .
s. iso , h. umetsu and f. wilczek , _ phys .
lett . _ * 96 * , 151302 ( 2006 ) .
s. iso , h. umetsu and f. wilczek , _ phys .
d _ * 74 * , 044017 ( 2006 ) .
l. bonora , m. cvitan , s. pallua and i. smolic , _ phys .
d _ * 80 * , 084034 ( 2009 ) ; j. j. peng and s. q. wu , arxiv:0906.5121 [ hep - th ] ; s. w. wei , r. li , y. x. liu and j. r. ren , _ eur .
phys . j. c _ * 65 * , 281 ( 2010 ) ; r. becar , p. gonzalez , g. pulgar and j. saavedra , arxiv:0808.1735 [ gr - qc ] .
t. morita , _ phys .
b _ * 677 * , 88 ( 2009 ) . r. banerjee , _ int
* 17 * , 2539 ( 2009 ) .
e. papantonopoulos and p. skamagoulis , _ phys .
d _ * 79 * , 084022 ( 2009 ) .
l. vanzo , _ phys .
d _ * 56 * , 6475 ( 1997 ) .
e. n. argyres , c. g. papadopoulos , e. papantonopoulos and k. tamvakis , _ j. phys .
a _ * 22 * , 3577 ( 1989 ) . c. martinez , r. troncoso and j. zanelli , _ phys
d _ * 70 * , 084035 ( 2004 ) . | i derive the hawking flux from black holes of constant negative curvature and from a black hole of constant negative curvature conformally coupled to a scalar field , using the covariant gravitational anomalies method . |
the optical depth of spiral disks has been the topic of many and varied studies since the claim by @xcite that they are transparent .
the subject became controversial when @xcite and @xcite argued that disks were virtually opaque . at the cardiff meeting @xcite many possible methods to attack the problem were proposed .
the dust disks of spirals may obscure objects in the high redshift universe @xcite or conceal mass in their disks @xcite .
an excellent review of the current state of knowledge on the opacity of spiral disks is given by @xcite .
early approaches to this subject were presented in @xcite and more recent developments are : + ( 1 ) disks are more opaque in the blue @xcite , ( 2 ) they are practically transparent in the near infrared @xcite , making these bands the best mass - to - luminosity estimator @xcite , ( 3 ) disks are practically transparent in the outer parts but show significant absorption in the inner regions @xcite , ( 4 ) the extinction correlates with galaxy luminosity @xcite , and ( 5 ) spiral arms are more opaque than the disk @xcite the majority of these studies are based on either inclination effects on the light distribution of a large sample of disks or a dust and light model to fit the observed profiles .
while there is some agreement on the view that spiral disks are substantially optically thick in their central regions and become optically thin in their outer parts , the exact radial extinction profile remains uncertain .
most measurements to date use the disk light itself to measure the extinction , and consequently require an assumption on the relative distribution of dust and light in a spiral disk .
the extinction in a disk can be derived from far infrared and sub - mm emission arising from the cold dust in disks .
however , these methods make the assumption that the emission characterises the dust in the disk .
but the far infrared and sub - mm emission is likely to be dominated by the warmest component of the dust , which tends to be the smaller grains on the outside of molecular clouds facing an energy source @xcite . in this case
the far - infrared and sub - mm emission will underestimate the average opacity .
@xcite mention this in their estimate of gas - to - dust ratio in spiral disks based on iras observations . to obtain a better characterisation of the absorption in a spiral disk without knowledge of the distribution of stars and dust in the disk ,
a known background source is needed .
@xcite proposed to use an occulted galaxy for this purpose , assuming it has a symmetric light distribution @xcite . @xcite and the companion paper of this work @xcite use the number of distant galaxies in the field as their background source , calibrating this number with simulations following their `` synthetic field method '' ( sfm ) . in @xcite
, we describe the details of this method .
both the `` occulting galaxy method '' and the sfm have the benefit of not using the disk s own light to measure the extinction .
the drawbacks of the keel and white method are the assumption of symmetry of both galaxies and the few suitable pairs available .
the sfm is more universally applicable but limited by the poor statistics .
it does however not need to assume any distribution of either the disk s light or the absorbers in it .
the sfm does require high - resolution images from hst and remains limited to the arms and disk of spiral galaxies due to crowding .
in this paper we report radial extinction profiles for spiral disks of different hubble types based on 32 hubble space telescope , wide field planetary camera 2 ( hst / wfpc2 ) fields in 29 galaxies of hubble types sab and later . in section 2
we describe the sample and its selection in detail .
the sfm is briefly outlined in section 3 ; a more complete description of the method and the recent improvements we have made to it is given in the companion paper @xcite . in section 4
we discuss the radial profile of average opacity for individual galaxies and , in section 5 , the composite average radial extinction profile for our entire sample .
the effects of inclination , spiral arm prominence , and hubble type on the profiles are discussed as well .
section 6 discusses the average ( v - i ) color of the field galaxies we found and section 7 , the tentative relation between average surface brightness and opacity .
we discuss some of the implications of our results , and end with the conclusions on disk opacity from the numbers and colors of field galaxies seen through them .
our sample of hst / wfpc2 fields was selected from the mast archive at stsci , based on criteria for both the target galaxy and the hst data .
the total exposure times and original proposal i d and reference are listed in table 1 , the basic data on the galaxies in table 2 .
the total solid angle of this sample is 146 arcmin@xmath2 .
the foreground galaxy should be a spiral , ideally face - on , spanning enough sky to cover a significant number of field galaxies .
this solid angle constraint limits the maximum distance for application of the sfm to approximately 30 mpc .
hst starts to resolve the disk population of spirals at close distances , making the field too crowded for field galaxy identification ( see @xcite ) .
this imposes a minimum distance of a few mpc .
the galaxies are type sab and later , plotted in figure 1 , according to the rc3 @xcite . no limit on inclination
was imposed as long as spiral arms could be discerned .
the majority of this sample is from the distance scale key project . the project s observing strategy @xcite was geared towards maximizing the number of cepheid variables detected .
this resulted in the selection of fields in the optical disk of face - on later - type spirals with a prominent arm visible .
these selection criteria are to be considered when interpreting the opacity measurements of these fields .
the sample of spiral disks has multi - epoch imaging in 2 photometric band available , as the original science driver for many of these data sets was to sample the cepheid light curves .
the reacquisition resulted in a slight shift in the pointing at each epoch .
the unintentional dither allowed us to drizzle the combined images to a pixel scale of 0.05 .
the data reduction is described in detail by @xcite .
@xcite predicted that improved resolution would mainly benefit the statistics of distant galaxy counts in the local group .
however , as the sample spans a range of distances and the statistics are likely to be poor , the maximum possible sampling was selected .
the pc chip was not used for analysis because it often lies on the most crowded region , its noise characteristics are different from the wf chips , and there are fewer reference fields available for the sfm .
this sample of hst pointings was selected as it is reasonably uniform , which allows similar processing using hubble deep field ( hdf ) background fields .
the hst / wfpc2 pointing should neither be on the center of the galaxy where crowding is too much of a problem , nor be outside the disk of the target galaxy where the expected opacity is likely too low to be measured with the sfm .
we selected pointings with the f814w ( i ) and f555w ( v ) filters as the identification of objects is based on several parameters characterising structure and on the v - i color .
a minimum exposure time of about 2000 seconds in both filters was adopted .
the choice of filters and exposure time was based on our earlier experiences with spiral and irregular galaxies @xcite . and to maximize the number of suitable fields .
@xcite describe in detail the data - reduction and the automated sfm , so we will give only a short summary here . to calibrate the numbers of field galaxies found in the science fields , simulated or `` synthetic '' fields are made .
these are the original science fields with an extincted hubble deep field added .
the numbers of simulated field galaxies found suffer from the same confusion and crowding as the number from the science field and therefore depend solely on the dimming applied .
a series of simulations give us the relation between field galaxy numbers and applied dimming .
the following relation is fitted to the simulated numbers : @xmath3 @xmath4 is the dimming in magnitudes , @xmath5 the number of field galaxies in a simulation , @xmath6 and @xmath7 are the normalization and slope of the fit ( i.e. the number of field galaxies expected with no extinction and how this number diminishes with increasing extinction ) .
the slope c characterises the effects of crowding and confusion on the relation between opacity and galaxy numbers .
it is usually slightly over unity but the unique character of each field led us to characterise c using the simulations in each field or combination of fields .
the intersection between this relation and the _ real _ number of field galaxies gives us the average extinction ( @xmath8 ) typical for the solid angle where these field galaxies were found .
the field galaxies are identified in the science fields , using automated selection , based on structural parameters and ( v - i ) color , together with a visual verification . in the synthetic fields ,
the visual check of objects was substituted by anti - correlating the automatically selected objects with the automatic selection in the science field , removing both the real distant galaxies as well as contaminants and leaving only added objects . the uncertainty in the number of field galaxies from the science field is a combination of the poisson uncertainty and the uncertainty due to field galaxy clustering .
the uncertainty in the simulated numbers is poisson only as they come from a known background , the hubble deep fields .
the uncertainty in average opacity is derived from the uncertainties in field galaxy numbers based on the poisson uncertainty as expressed by @xcite , and the clustering uncertainty based on the 2-p correlation function found by @xcite and equation 1 .
the crowding and confusion bias is calibrated with the simulations . for a detailed error discussion of the sfm see @xcite , 4.3 . a difference in dust extinction from our own galaxy between the reference fields ( hdf - n / s ) and the pointing at the foreground galaxy will introduce a bias in the extinction measurement of the specific disk .
@xcite used galaxy counts to measure the extinction towards grb 970228 . and found excellent agreement with other measurements of galactic extinction @xcite .
@xcite produced an all - sky map of galactic extinction based on cobe and iras maps and we use their values for galactic extinction ( table 2 ) .
most galaxies in the sample do not show a significant difference in galactic extinction compared to the average of the galactic extinction towards the hdf - n / s ( @xmath8 = 0.039 mag ) .
however , the numbers of galaxies from each science field were nevertheless corrected for the difference in galactic extinction using equation 1 .
@xcite presented results based on the sfm for individual wide field chips , characterised as `` arm '' or `` disk '' regions based on predominance in the chip .
@xcite segmented the wfpc2 mosaics based on morphological component ( arm , inter - arm , disk ) or projected radius from the center .
however , the statistics from individual wfpc2 fields barely allow any meaningful opacity measurements for solid angles smaller than a single wf field . in table 3
we present opacity values for the projected radial annuli in each foreground galaxy in our sample .
the radii are expressed in @xmath9 , half the @xmath10 from rc3 catalog @xcite .
the error - bars are computed from the uncertainties in the numbers of real and simulated field galaxies from counting and clustering @xcite . for the galaxies for which we have two wfpc2 pointings ( ngc5194 , ngc3621 and
ngc4414 ) , the radial extinction from the combined counts is also shown .
estimates of extinction in a galaxy disk based on a single wfpc2 chip suffer from poor statistics @xcite , which make a radial dependence hard to establish ( see table 3 and @xcite ) .
this led us to apply the sfm to a large sample of foreground galaxies in order to estimate the general extinction properties of galaxy disks from the combined numbers of field galaxies seen through these disks .
we caution against averaging the values in table 3 to derive average profiles .
figure 2 shows the radial opacity plot from all our fields combined .
the numbers of field galaxies from both the science fields and the simulations from all fields were combined based on their projected radial distance from the respective galaxy s centers , expressed in @xmath9 .
the top panel in figure 2 shows the total combined solid angle for each radial bin for which the average opacity was determined .
the middle panel shows the number of field galaxies from the science field and the average number found in the simulations without any dimming ( a=0 ) .
the bottom panel shows the opacity for each bin derived from the intersection of equation 1 fitted to the simulations and the real number of galaxies . the solid angle or the number of simulated galaxies without any dimming is a good indicator for the reliability of our opacity estimate , as reflected by the error bars in the bottom panel of figure 2 .
the estimates are limited at small radii by the high surface brightness and crowding of the foreground galaxy center , effectively masking some of the solid angle available at those radii . at
higher radii the uncertainty comes from a lack of solid angle covered ( see figure 2 , top panel ) .
this is a selection effect of our sample , as most of the wfpc2 fields were pointed at the optical disk of the galaxy . in order to determine the effects on average radial opacity of disk inclination , prominence of spiral arms or hubble type , radial opacity plots of subsets of our sample
were constructed .
the solid angle used is then some fraction of those in the top panel in figure 2 , which consequently increases the uncertainty in the opacity measurement . to counter this , instead of the radial bins of 0.1 @xmath9 ( figure 2 ) ,
radial binning of 0.2 @xmath9 was applied .
the inclination of the foreground disks affects the measured opacity , but the amount of this effect depends on the dust geometry . in the case of a uniform thick screen , the path length attenuating the field galaxies is increased .
however in the case of a screen of dark clouds in the disk , the effect is on the apparent filling factor of clouds . because the correction depends on which dust geometry is assumed , we present the radial results ( figure 2 ) without any correction and explore corrections using several different models , illustrated in figure 3 .
the homogeneous screen results in a multiplicative factor cos(i ) to be applied to the opacity value ( a ) or in a correction to the number of field galaxies found in the science field , as follows : @xmath11 the other models consider a screen with fully opaque clouds or patches . depending on the thickness of these clouds ,
the apparent filling factor depends differently on inclination . in these models ,
the clouds have an average oblateness @xmath12 ( @xmath13 ) with major axis a and minor axis b of the clouds . following the geometry of figure 3 ,
the relation between the number of field galaxies , seen through the foreground disk face - on ( @xmath14 ) , at an inclination i ( @xmath15 ) , and the average number of field galaxies in the field behind the foreground galaxy ( @xmath6 ) can then be expressed as : @xmath16 \ [ 1 \ - \ cos(i)\ ] \
n_0 + \left\ { \ \epsilon \ [ 1-cos(i ) ] \ + \ cos(i ) \right\ } \ n_i\ ] ] where the extreme cases for @xmath12 are spherical clouds ( @xmath17 ) and flat patches ( @xmath18 ) .
the oblateness @xmath12 parameterises the ratio between the scale height of the dust and its extent in the plane of the disk . in images of edge - on disks , the visible dust lanes are confined to a thinner disk than the stars .
it is therefore likely that @xmath12 is not 0 . for purposes of illustration we use a value of 0.5 in figure 8 .
figure 4 shows the radial profile from figure 2 corrected for both extreme cases and uniform screens in all galaxies .
the contributions from each galaxy to the composite radial profile were corrected according to equation 2 and equation 3 before addition .
figure 5 shows the total uncorrected radial profile for four subdivisions of our sample based on inclination . from figure 5
it seems clear that other effects are much more important than the inclination of the foreground disk . for this reason we ignoring the effects of inclination on our measurements . as the effects of inclination are debatable on the basis of preferred dust geometry , we present further results without inclination corrections except where noted . another effect on the average radial opacity profile in figure 2
is due to the presence of spiral arms . as our sample is predominantly from the cepheid distance project ,
the wfpc2 images all feature spiral arms .
if these are more opaque than the disk proper @xcite , then the radial profile presented in figure 2 could be biased towards higher opacities .
separate radial plots for the arm , the inter - arm part of the disk and outside any spiral arm part of the disk are shown in figure 7 based on the counts from the typical regions in the entire sample , with the exception of the lsb galaxies ugc 2302 and ugc 6614 .
they were left out , as not much spiral structure can be discerned ; they appear completely transparent ( see figure 8) . in order to differentiate between the effects of spiral arms and disks in the opacity plots
, the images were segmented into crowded , arm , disk ( inter - arm ) and disk ( outside arm ) .
these regions were flagged in the mosaiced wfpc2 fields using the gipsy ( groningen image processing system ) function _ blot _ , in the same way as ngc1365 in @xcite .
the choice of typical regions was made in order to compare the arm and inter - arm results of @xcite and @xcite .
a typical mask is presented in figure 6 and masks for all galaxies in the sample are presented in @xcite .
it should be noted that this segmentation into typical regions is subjective , and based on those sections of the foreground galaxies covered by the wfpc2 observations we use .
radial extinction profiles of the typical regions are presented in figure 7 .
the arm regions show much more opacity and a much more pronounced radial dependence of that opacity .
there is a radial dependence as well for the inter - arm parts of the disk , although not as steep .
the outside parts of the disk of the spiral galaxy , however , show little or no relation between opacity and radius .
the opaque components of a spiral disk appear to be the spiral arms while the disk itself is more transparent but much more extended .
figure 7 also shows the total solid angle of the radial annuli over which the opacities are determined . from these
we can conclude that the regions deemed ` arm ' are not dominating the whole of the fields .
the radial effect of the arms only becomes visible in the total radial opacity plot ( figure 2 and top left panel figure 7 ) , at the lower radii where the arms and the inter - arm region dominate . in figure 8
we present the profiles corrected for inclination assuming an @xmath12 of 0.5 .
the general trends remain but now for somewhat lower opacity values .
@xcite and @xcite presented their extinction values from occulting galaxy pairs as a function of the radius , scaled with @xmath9 .
@xcite compared ground - based photometry , and @xcite used spectroscopic measurements of the occulted galaxy light .
these extinction points are plotted in figure 9 for arm and inter - arm regions .
the extinction curves from figure 7 and the inclination corrected one from figure 8 are plotted as well .
both the arm and the inter - arm extinction values as a function of radius agree well with the values obtained from the occulting galaxy technique .
it is remarkable how well the results compare , considering they were obtained from completely different samples of spiral galaxies and using different techniques .
the values from the occulting galaxy technique are slightly lower than ours .
there are several possible reasons for this .
the spectroscopic results from @xcite ( triangles in figure 9 ) possibly favour the more transparent regions in a disk ( domingue 2004 , private communication ) . but more importantly , the sample of occulting foreground galaxies consists of a different makeup of spiral galaxy subtypes than that of this paper .
it should be noted that there are no galaxies common to both our sample and that of the occulting galaxy technique .
single galaxy in both our sample and the occulting galaxy method s .
@xcite noted that their later types ( sbc ) seem more opaque , as do we ( figure 11 ) .
figure 10 compares the results for the most prevalent hubble subtypes in the occulting galaxy method ( sb and sbc ) with our results for those .
the arm values seem to match up but there is a difference in the inter - arm results .
it is unclear to us whether this points to a structural effect in either technique .
a likely explanation is that the definition of `` inter - arm '' applies to slightly different regions in the spiral disks for @xcite , and his paper .
this paper s definition of typical regions is based on the high - resolution mosaic , whereas the @xcite is on their ground - based imaging .
there is therefore the possibility that we include sections in the interarm regions which the occulting techniques would not resolve as `` inter - arm '' , raising our values of opacity for those regions with respect to the occulting galaxy technique .
the sfm provides an independent verification of the occulting galaxy technique using a fundamentally different approach .
in addition , a component was added to the distinction between arm and inter - arm parts of the disk , the `` outside '' , meaning not directly enclosed by spiral arms . the fact that this component is not fully transparent raises the possibility of a dust disk extending beyond the spiral arms ( figure 8) . with the effect of spiral arms on the opacity profile ,
the hubble type is likely to be of influence on the profile .
figure 10 shows the extinction profiles averaged per hubble type .
hubble types sab through sd are presented , as well as the average opacity profile of the two lsb galaxies in our sample .
hubble types sab through scd in our sample show disk extinction up to the @xmath9 .
the sab result is tentative , owing to the poor statistics from only 2 wfpc2 fields .
the sd and lsb galaxies however appear effectively transparent .
lsb galaxies were in fact assumed to be transparent by @xcite when discussing the morphology of field galaxies seen though them .
however both profiles , sd and lsb , are based on only two wfpc2 fields , accounting for the higher uncertainties . comparing early type galaxies with later types
, it appears that the later type galaxies ( sbc - sc ) show more extinction and at larger radii .
the sb galaxies show a bump which appears to be associated with higher extinction from the more tightly wound spiral arms .
figures 12 and 13 show the radial profiles for arm and disk -both inter - arm and outside regions- for our sample divided into early and late spirals .
a finer separation in hubble type would have resulted in even higher uncertainties , due to lack of solid angle and hence statistics .
the solid angle of each radial annulus is also plotted .
purely arm regions do not dominate the opacity profiles .
however , the interarm regions are similar in behaviour ( figures 7 and 8)
. the inner parts of the profile are therefore more arm - like , while the remainder is disk dominated .
the radial dependence of arm and inter - arm regions is more pronounced in the case of earlier galaxies than for later types .
the opacity in the spiral arms is substantially higher then in the disk , for both early and late types .
the total profiles show bumps at 0.9 and 1.1 @xmath9 for the early and late types , respectively . in the case of the early types
this seems due to the spiral arm contribution at that radius .
the bump for the late types is not as significant , but might be related to the general position of spiral arms as well .
the average color of the field galaxies found in the science fields can , in principle , tell us something about the actual dust geometry responsible for the drop in numbers . if there is a correlation between the average reddening of the field galaxies and the average opacity of the foreground spiral , then the dust extinction responsible for the drop in number of field galaxies is , at least in part , in the form of a diffuse screen , reddening the visible galaxies . if , however , the average color of the field galaxies does not change with opacity , then the drop in their number is likely due to fully opaque clouds with transparent sections between them to allow for the detection of the unreddened surviving field galaxies . +
our method influences the average color of the detected field galaxies in several ways .
the v band ( f555w ) for the synthetic field background was constructed from the original hdf images in the f606w and f450w filters .
however , @xcite estimated that this introduces a negligible error .
crowding introduces blended objects in the synthetic fields .
the synthetic counts are corrected for this effect but it will influence the average color .
the simulated dust extinction we used was grey , so no preferred reddening was introduced in the synthetic fields .
however , our automated selection procedure for field galaxy candidates selects against very blue objects , introducing a preference for red galaxies .
a similar selection effect may take place in the visual check of the science fields , as blue objects are treated as suspected foreground objects .
overall , these selection biases will cause the average color of the field galaxies in the science and synthetic fields to be redder than the average for an unobstructed field of distant galaxies .
the effects of blended objects on the average colors of science and synthetic fields are not identical , as not all blends have been removed from the synthetic counts .
nonetheless , we can compare the trends of both these average colors with radius .
the average ( v - i ) color of the field galaxies found in the science fields , the simulation without extinction ( a = 0 ) , and the opacity derived from the galaxy numbers are plotted as a function of radius in figure 14 .
the average ( v - i ) color of the field galaxies in the science fields does not change with radius and hence average opacity .
the average color of the synthetic field objects appears to become bluer with radius .
however , beyond 1.4 @xmath9 the number of objects is very small and the averages should be treated with caution . comparing the average color of distant galaxies from science and synthetic fields for the inner part of the disk , the average of the science field objects is redder than the synthetic field average .
this difference in average color is likely the result of blends with blue foreground objects being inadvertently included in the the selection of synthetic field objects .
the number of objects from the synthetic fields was corrected for this effect ( see also @xcite but the average color was not . the science field objects do not suffer from this as they were checked visually for blended objects .
the field galaxies seen in the science fields are likely visible in parts of the disk which are nearly transparent or , alternatively , the dust screen in the disk behaves according to the `` grey '' extinction law .
this is remarkable , considering the bias towards redder objects throughout our method .
figure 15 shows the average color - extinction measurements for the field galaxies in the science fields .
the galactic extinction law is shown for comparison .
each point has been determined in a radial annulus , for all the fields combined ( figure 2 ) and for the typical regions ( figure 7 , but with the finer radial sampling of figure 2 . ) . without distinction between regions , no trend with opacity
can be discerned for the average color .
the average color in the arm regions does not seem to rise much with opacity .
the arm extinction is decidedly grey , something also found by @xcite . for the disk regions ,
no distinct trend can be seen , and they do not seem to follow the galactic law very well . @xcite found the disk region of ngc4536 to be more galactic in its reddening law .
there is an average reddening with respect to the average ( v - i ) color of the hdf objects identified as galaxies by our algorithm without a foreground disk .
this reddening is likely to be the result of the effects of the visual check and contamination of the color measurement by stray disk light .
these color - extinction relations for both arms and disk seem to be greyer than the galactic extinction slope , favouring the possibility that at least part of the extinction is in opaque clouds .
uncertainties are such , however , that even for our increased statistics no good relation between opacity and color can be found .
the counts of distant galaxies were added per radial annulus , expressed in @xmath9 . additionally , the flux and solid angle from each field can be added for each radial annulus . the averaged surface brightness ( @xmath19 ) and the average opacity ( a ) from each radial bin in figure 2 ,
are plotted in figure 16 .
these values for the radial bins per typical region in figure 7 are plotted as well , but with finer sampling . there is a hint of a relationship between luminosity and extinction .
this is consistent with the relation found by @xcite and @xcite between overall disk opacity and the galaxy s total luminosity .
the solid angles over which the surface brightness and opacity were averaged were selected by radius and not luminosity .
areas with different surface brightnesses are therefore combined at each radius , smoothing out any relation between opacity and surface brightness .
hence , most of the points in figure 15 are around the same opacity and the range of surface brightness values is small .
nevertheless , the values of high extinction and brightness ( @xmath20 ) do show some relationship between surface brightness and opacity , with arm regions displaying more extinction at the same surface brightness levels than the inter - arm disk .
the disk values appear to show no correlation , but these values are per definition not for high surface brightness levels . and yet , together with the higher values of opacity found for the spiral arms , the points are not inconsistent with a relation between the opacity and luminosity .
opacity measurements in partitions of the wfpc2 images based on average surface brightness instead of radius should reveal any relation more clearly .
this comparison between extinction and emission will be presented , in more detail , in a later paper .
from the number of field galaxies found through the disks of spiral galaxies , a quantative picture of extinction as a function of radius can be found .
the sfm is too limited by statistics to obtain a good result for individual fields .
however a meaningful measurement can be derived from a combination of several galaxies .
the effect of spiral arms on the radial extinction is quite distinct and dependent on hubble type . from these radial plots
it becomes clear that the dust in the disk is not one smooth layer sandwiched between the disk s stars but two separate components : radially dependent spiral arms and a more transparent but also extended disk .
in addition , there is a highly opaque central bulge component . whether the average opacity is a result of dust clouds or a smooth screen of grey dust , is difficult to determine from the numbers , colors and luminosities of the field galaxies found through the disks of our sample .
however , the relative independence on inclination of the average opacity measured from numbers of spiral galaxies and the grey relation between opacity and average field galaxy color , both point towards a patchy distribution of the absorbing dust in the disk .
assuming that the optical depth of the disk can be expressed as @xmath21 , where f is the area filling factor , the average opacity of @xmath22 would require a filling factor of 40% of the disk , a figure that rises to 85 % in the spiral arms .
while the visible dust lanes can account for at least a part of these clouds , other dust clouds are likely embedded in the disks and arms .
this patchy coverage explains the occasional distant galaxy seen through a spiral disk @xcite , which has sometimes been used as anecdotal evidence for transparent disks .
the disk is in fact relatively transparent where background galaxies are seen .
however , by calibrating the number of the distant galaxies found , a very different picture emerges .
while this result is consistent with earlier findings , these values are likely upper limits , not lower ones .
any inclination correction will lower the face - on value for extinction .
however , the patchy nature of dust extinction is likely responsible for a high variety of extinction values in the disk .
integrated measurements such as light profiles , however , will be affected according to the presented opacity values . a possible relation between extinction and brightness
would also cast doubt on the fixed mass - to - light ratios generally assumed when modelling the kinematics of a spiral disk .
the effects of dust extinction on the number of field galaxies found in our fields lead us to the following conclusions : + ( 1 ) the sfm gives an unbiased but uncertain measure of opacity for single wfpc2 fields ( table 3 ) .
+ ( 2 ) on average , the disk of a spiral galaxy has an opacity in @xmath23 of @xmath241 magnitude ( figure 2 ) .
+ ( 3 ) the extinction measured from the number of field galaxies seems to be independent of the inclination of the foreground disk over the range in @xmath25 covered by our sample ( @xmath26 ) ( figure 5 ) .
this is consistent with fully opaque flattened clouds covering a fraction of the area as the cause of the observed average opacity .
+ ( 4 ) the absorption profile for typical regions in the disk is strongly influenced by the spiral arms .
spiral arm regions are the most opaque and display a radial dependence .
the disk regions enclosed by a spiral arm are also more opaque than other disk regions and display a similar radial dependence ( figure 7 ) .
+ ( 5 ) the radial extinction curves derived from numbers of background galaxies and those reported by the occulting galaxy technique agree reasonably well ( figure 9 and 10 ) , although a systematic effect in either ( or both ) technique seems to be present in the interarm results ( figure 10 ) .
+ ( 6 ) sc galaxies show much more opacity in their central regions than other types ( figure 11 ) .
+ ( 7 ) all hubble types earlier than scd in our sample show substantial disk extinction in @xmath23 up to @xmath9 ( figure 11 ) .
+ ( 8) the numbers from sd and lsb galaxies are consistent with a transparent disk , but these measurements are limited by statistics .
( figure 11 ) .
+ ( 9 ) both early and late type spirals exhibit the extinction profiles of the two distinct components : a radially dependent one ( arm and inter - arm ) , and a more extended disk ( outside arms ) ( figures 12 and 13 ) .
+ ( 10 ) the average color of the distant galaxies identified in the science fields does not change with either radius or the opacity derived from their numbers ( figure 14 ) .
+ ( 11 ) the grey nature of the absorption derived from field galaxy counts and colors holds for all typical regions ( figure 15 ) .
the absence of inclination effects and the grey nature of the absorption is consistent with strongly absorbing dense clouds masking off the distant galaxies .
their covering factor in the disk would be around 40% , regardless of cloud sizes ( which our technique can not provide ) .
+ ( 12 ) the average surface brightness in radial annuli and the corresponding average opacity derived from distant galaxy counts appear correlated .
they are consistent with a rise of opacity with surface brightness .
( figure 16 ) .
although the range of surface brightness is small and the uncertainty in opacity rises with higher surface brightness , this is in good agreement with earlier results for bright galaxies .
this would constitute a relation between dust mass and light .
+ the `` synthetic field method '' has proven itself to be a useful , model - independent , technique to measure the total opacity of spiral disks .
it can be applied to any spiral disk at intermediate distance for which there is high resolution imaging available .
we will present further results for opacity as a function of surface brightness in a future paper .
the authors would like to thank the anonymous referee for his or her comments and harry ferguson for insightful discussions on the `` synthetic field method '' .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jpl , caltech , under contract with nasa .
this work is primarily based on observations made with the nasa / esa hubble space science telescope ( stsci ) and obtained from the data archive at the space telescope institute .
support for this work was provided by nasa through grant number hst - ar-08360 from stsci .
stsci is operated by the association of universities for research in astronomy , inc .
, under the nasa contract nas 5 - 26555 .
we are also grateful for the financial support of the stsci director s discretionary fund ( grant numbers 82206 and 82304 ) , and of the kapteyn institute of groningen university , l. , bresolin , f. , kennicutt , r. c. , saha , a. , stetson , p. b. , freedman , w. l. , mould , j. r. , madore , b. f. , sakai , s. , ford , h. c. , gibson , b. k. , graham , j. a. , han , m. , hoessel , j. g. , huchra , j. , hughes , s. m. , illingworth , g. d. , phelps , r. , prosser , c. f. , and silbermann , n. a. : 1998 , , 655 , l. , freedman , w. l. , hill , r. j. , saha , a. , madore , b. f. , kennicutt , r. c. , stetson , p. b. , ford , h. c. , graham , j. a. , hoessel , j. g. , han , m. , huchra , j. , hughes , s. m. , illingworth , g. d. , kelson , d. , mould , j. r. , phelps , r. , silbermann , n. a. , sakai , s. , turner , a. , harding , p. , and bresolin , f. : 1996 , , 568 , w. l. , hughes , s. m. , madore , b. f. , mould , j. r. , lee , m. g. , stetson , p. , kennicutt , r. c. , turner , a. , ferrarese , l. , ford , h. , graham , j. a. , hill , r. , hoessel , j. g. , huchra , j. , and illingworth , g. d. : 1994 , , 628 , w. l. , madore , b. f. , gibson , b. k. , ferrarese , l. , kelson , d. d. , sakai , s. , mould , j. r. , kennicutt , r. c. , ford , h. c. , graham , j. a. , huchra , j. p. , hughes , s. m. g. , illingworth , g. d. , macri , l. m. , and stetson , p. b. : 2001 , , 47 , b. k. , hughes , s. m. g. , stetson , p. b. , freedman , w. l. , kennicutt , r. c. , mould , j. r. , bresolin , f. , ferrarese , l. , ford , h. c. , graham , j. a. , han , m. , harding , p. , hoessel , j. g. , huchra , j. p. , illingworth , g. d. , kelson , d. d. , macri , l. m. , madore , b. f. , phelps , r. l. , prosser , c. f. , saha , a. , sakai , s. , sebo , k. m. , silbermann , n. a. , and turner , a. m. : 1999 , , 48 , j. a. , ferrarese , l. , freedman , w. l. , kennicutt , r. c. , mould , j. r. , saha , a. , stetson , p. b. , madore , b. f. , bresolin , f. , ford , h. c. , gibson , b. k. , han , m. , hoessel , j. g. , huchra , j. , hughes , s. m. , illingworth , g. d. , kelson , d. d. , macri , l. , phelps , r. , sakai , s. , silbermann , n. a. , and turner , a. : 1999 , , 626 , j. a. , phelps , r. l. , freedman , w. l. , saha , a. , ferrarese , l. , stetson , p. b. , madore , b. f. , silbermann , n. a. , sakai , s. , kennicutt , r. c. , harding , p. , bresolin , f. , turner , a. , mould , j. r. , rawson , d. m. , ford , h. c. , hoessel , j. g. , han , m. , huchra , j. p. , macri , l. m. , hughes , s. m. , illingworth , g. d. , and kelson , d. d. : 1997 , , 535 , b. w. , allen , r. j. , and van der kruit , p. c. : 2002 , in _ asp conf .
ser . 273 : the dynamics , structure & history of galaxies : a workshop in honour of professor ken freeman ed .
da costa & h. jerjen .
san francisco , astronomical society of the pacific _ , pp 337+ , s. m. g. , han , m. , hoessel , j. , freedman , w. l. , kennicutt , r. c. , mould , j. r. , saha , a. , stetson , p. b. , madore , b. f. , silbermann , n. a. , harding , p. , ferrarese , l. , ford , h. , gibson , b. k. , graham , j. a. , hill , r. , huchra , j. , illingworth , g. d. , phelps , r. , and sakai , s. : 1998 , , 32 , d. d. , illingworth , g. d. , saha , a. , graham , j. a. , stetson , p. b. , freedman , w. l. , kennicutt , r. c. , mould , j. r. , ferrarese , l. , huchra , j. p. , madore , b. f. , prosser , c. f. , bresolin , f. , ford , h. c. , gibson , b. k. , hoessel , j. g. , hughes , s. m. g. , macri , l. m. , sakai , s. , and silbermann , n. a. : 1999 , , 614 , d. c. , filippenko , a. v. , gates , e. l. , li , w. , eastman , r. g. , barth , a. j. , bus , s. j. , chornock , r. , coil , a. l. , frink , s. , grady , c. a. , harris , a. w. , malkan , m. a. , matheson , t. , quirrenbach , a. , and treffers , r. r. : 2002a , , 35 , l. m. , huchra , j. p. , stetson , p. b. , silbermann , n. a. , freedman , w. l. , kennicutt , r. c. , mould , j. r. , madore , b. f. , bresolin , f. , ferrarese , l. , ford , h. c. , graham , j. a. , gibson , b. k. , han , m. , harding , p. , hill , r. j. , hoessel , j. g. , hughes , s. m. g. , kelson , d. d. , illingworth , g. d. , phelps , r. l. , prosser , c. f. , rawson , d. m. , saha , a. , sakai , s. , and turner , a. : 1999a , , 155 , l. m. , huchra , j. p. , stetson , p. b. , silbermann , n. a. , freedman , w. l. , kennicutt , r. c. , mould , j. r. , madore , b. f. , bresolin , f. , ferrarese , l. , ford , h. c. , graham , j. a. , gibson , b. k. , han , m. , harding , p. , hill , r. j. , hoessel , j. g. , hughes , s. m. g. , kelson , d. d. , illingworth , g. d. , phelps , r. l. , prosser , c. f. , rawson , d. m. , saha , a. , sakai , s. , and turner , a. : 1999b , , 155 , j. r. , hughes , s. m. g. , stetson , p. b. , gibson , b. k. , huchra , j. p. , freedman , w. l. , kennicutt , r. c. , bresolin , f. , ferrarese , l. , ford , h. c. , graham , j. a. , han , m. , hoessel , j. g. , illingworth , g. d. , kelson , d. d. , macri , l. m. , madore , b. f. , phelps , r. l. , prosser , c. f. , rawson , d. , saha , a. , sakai , s. , sebo , k. m. , silbermann , n. a. , and turner , a. m. : 2000 , , 655 , d. m. , macri , l. m. , mould , j. r. , huchra , j. p. , freedman , w. l. , kennicutt , r. c. , ferrarese , l. , ford , h. c. , graham , j. a. , harding , p. , han , m. , hill , r. j. , hoessel , j. g. , hughes , s. m. g. , illingworth , g. d. , madore , b. f. , phelps , r. l. , saha , a. , sakai , s. , silbermann , n. a. , and stetson , p. b. : 1997 , , 517 , s. , ferrarese , l. , kennicutt , r. c. , graham , j. a. , silbermann , n. a. , mould , j. r. , freedman , w. l. , bresolin , f. , ford , h. c. , gibson , b. k. , han , m. , harding , p. , hoessel , j. g. , huchra , j. p. , hughes , s. m. , illingworth , g. d. , kelson , d. , macri , l. , madore , b. f. , phelps , r. l. , saha , a. , sebo , k. m. , stetson , p. b. , and turner , a. : 1999 , , 540 , n. a. , harding , p. , ferrarese , l. , stetson , p. b. , madore , b. f. , kennicutt , r. c. , freedman , w. l. , mould , j. r. , bresolin , f. , ford , h. , gibson , b. k. , graham , j. a. , han , m. , hoessel , j. g. , hill , r. j. , huchra , j. , hughes , s. m. g. , illingworth , g. d. , kelson , d. , macri , l. , phelps , r. , rawson , d. , sakai , s. , and turner , a. : 1999 , , 1 , n. a. , harding , p. , madore , b. f. , kennicutt , r. c. , saha , a. , stetson , p. b. , freedman , w. l. , mould , j. r. , graham , j. a. , hill , r. j. , turner , a. , bresolin , f. , ferrarese , l. , ford , h. , hoessel , j. g. , han , m. , huchra , j. , hughes , s. m. g. , illingworth , g. d. , phelps , r. , and sakai , s. : 1996 , , 1 , a. , ferrarese , l. , saha , a. , bresolin , f. , kennicutt , r. c. , stetson , p. b. , mould , j. r. , freedman , w. l. , gibson , b. k. , graham , j. a. , ford , h. , han , m. , harding , p. , hoessel , j. g. , huchra , j. p. , hughes , s. m. g. , illingworth , g. d. , kelson , d. d. , macri , l. , madore , b. f. , phelps , r. , rawson , d. , sakai , s. , and silbermann , n. a. : 1998 , , 207 l l l l l l ngc 925 & 26400.0 & 9000.0 & & 5397 & @xcite + ngc 1365 & 66560.0 & 16060.0 & & 5972 & @xcite + ngc 1425 & 58800.0 & 29700.0 & & 5972/6431 & @xcite + ngc 1637 & 26400.0 & 13200.0 & & 9155 & @xcite + ngc 2541 & 28760.0 & 12760.0 & & 5972 & @xcite + ngc 2841 & 26400.0 & 11000.0 & & 8322 & @xcite + ngc 3031 ( m81 ) & 2000.0 & 2000.0 & & 9073 & @xcite + ngc 3198 & 27760.0 & 12560.0 & & 5972 & @xcite + ngc 3319 & 26400.0 & 10400.0 & & 6431 & @xcite + ngc 3351 ( m95 ) & 31900.0 & 9830.0 & & 5397 & @xcite + ngc 3621 - 1 & 5200.0 & 7800.0 & & 8584 & @xcite + ngc 3621 - 2 & 20759.0 & 7380.0 & & 5397 & @xcite + ngc 3627 ( m66 ) & 58800.0 & 25000.0 & & 6549 & @xcite + ngc 4321 ( m100 ) & 32750.0 & 17150.0 & & 5397 & @xcite + ngc 4414 - 1 & 1600.0 & 1600.0 & & 8400 & hubble heritage + ngc 4414 - 2 & 32430.0 & 10230.0 & & 5397 & @xcite + ngc 4496a & 68000.0 & 16000.0 & & 5427 & @xcite + ngc 4527 & 60000.0 & 25000.0 & & 7504 & @xcite + ngc 4535 & 48800.0 & 31200.0 & & 5397/6431 & @xcite + ngc 4536 & 68000.0 & 20000.0 & & 5427 & @xcite + ngc 4548 ( m98 ) & 48500.0 & 30900.0 & & 6431 & @xcite + ngc 4559 & 2000.0 & 2000.0 & & 9073 & @xcite + ngc 4571 & 10400.0 & 26400.0 & & 6833 & @xcite + ngc 4603 & 58800.0 & 14800.0 & & 6439 & @xcite + ngc 4639 & 58800.0 & 13000.0 & & 5981 & @xcite + ngc 4725 & 32430.0 & 10230.0 & & 5397 & @xcite + ngc 6946 & 2000.0 & 2000.0 & & 9073 & @xcite + ngc 7331 & 40660.0 & 9860.0 & & 5397 & @xcite + ugc 2302 & 15000.0 & 15600.0 & & 8255 & @xcite + ugc 6614 & 10100.0 & 10100.0 & & 8213 & @xcite
+ ngc 5194 - 1 ( m51 ) & 2000.0 & 2000.0 & & 9073 & @xcite + ngc 5194 - 2 ( m51 ) & 2000.0 & 2000.0 & & 9073 & @xcite +
l l l l l l l l l l ngc 925 & sabd & 36.820469 & 33.578880 & -70 & 62.48 & 11.22 & 9.16 & 0.147 & + ngc 1365 & sbb & 53.401909 & -36.140659 & 49 & 34.41 & 10.47 & 17.95 & 0.039 & + ngc 1425 & sbb & 55.548061 & -29.893511 & -51 & 57.32 & 5.75 & 21.88 & 0.025 & + ngc 1637 & sab(rs)c & 70.367622 & -2.858040 & 27 & 30.68 & 3.98 & 8.2 & 0.078 & ( 1 ) + ngc 2541 & sacd & 123.666969 & 49.061440 & -7 & 56.63 & 6.31 & 11.22 & 0.097 & + ngc 2841 & sab & 140.511063 & 50.976479 & -30 & 60.33 & 8.13 & 14.1 & 0.030 & ( 2 ) + ngc 3031 ( m81 ) & sa(s)ab & 148.88826 & 69.06526 & -38.5 & 58.87 & & 3.63 & 0.155 & + ngc 3198 & sbc & 154.979126 & 45.549690 & 37 & 61.77 & 8.51 & 13.80 & 0.024 & + ngc 3319 & sb(rs)cd & 159.789719 & 41.686871 & 39 & 69.51 & 6.17 & 13.30 & 0.028 & + ngc 3351 ( m95 ) & sbb & 160.990555 & 11.703610 & -2 & 14.53 & 7.41 & 10.00 & 0.054 & + ngc 3621 & sac & 169.567917 & -32.812599 & -16 & 58.13 & 12.3 & 6.64 & 0.156 & + ngc 3627 ( m66 ) & sab(s)b & 170.062607 & 12.991290 & 1 & 57.38 & 9.12 & 10.05 & 0.063 & + ngc 4321 ( m100 ) & sabbc & 185.728745 & 15.822380 & -57 & 39.65 & 7.41 & 15.21 & 0.051 & + ngc 4414 & sac & 186.612869 & 31.223545 & -20 & 48.70 & 3.63 & 17.70 & 0.038 & + ngc 4496a & sbm & 187.913361 & 3.939467 & 65 & 58.67 & 3.98 & 14.86 & 0.048 & + ngc 4527 & sab(s)bc & 188.535400 & 2.653810 & 67 & 70.73 & 6.17 & 14.1 & 0.043 & ( 3 ) +
ngc 4535 & sabc & 188.584625 & 8.197760 & 27 & 28.36 & 7.08 & 15.78 & 0.038 & + ngc 4536 & sab(rs)bc & 188.613037 & 2.187880 & -70 & 62.61 & 7.08 & 14.93 & 0.035 & + ngc 4548 ( m98 ) & sbb & 188.860123 & 14.496320 & -60 & 22.48 & 5.37 & 16.22 & 0.074 & + ngc 4559 & sab(rs)cd & 188.990372 & 27.959761 & -37 & 61.77 & 10.72 & 10.88 & 0.034 & + ngc 4571 & sa(r)d & 189.234879 & 14.217357 & 75 & 23.07 & 3.63 & 14.9 & 0.091 & ( 4 ) + ngc 4603 & sa(rs)bc & 190.229980 & -40.976402 & 30 & 50.21 & 3.39 & 33.3 & 0.325 & ( 5 ) + ngc 4639 & sabbc & 190.718140 & 13.257536 & -25 & 43.95 & 2.75 & 21.98 & 0.050 & + ngc 4725 & sabab & 192.610886 & 25.500759 & 47 & 45.25 & 10.72 & 12.36 & 0.023 & + ngc 5194 ( m51 ) & sa(s)bc & 202.46957 & 47.19526 & 45.0 & 25.58 & 11.22 & 8.4 & 0.067 & + ngc 6946 & sab(rs)cd & 308.718048 & 60.153679 & 70 & 39.65 & 11.48 & 11.48 & 0.663 & + ngc 7331 & sab & 339.267090 & 34.415920 & -12 & 57.38 & 10.47 & 14.72 & 0.176 & + ugc 2302 & sb(rs)m & 42.285831 & 2.127265 & 60.0 & 1.0 & 4.79 & 14.7 & 0.156 & ( 6 ) + ugc 6614 & sa(r)a & 174.811844 & 17.143578 & -65 & 40.54 & 1.66 & 84.68 & 0.055 & ( 7 ) + ngc 925 & @xmath27 & @xmath28 & @xmath29 & @xmath30 & & & & & & & + ngc 1365 & & @xmath31 & @xmath32 & & & & & & & & + ngc 1425 & & & @xmath33 & @xmath34 & @xmath35 & @xmath36 & @xmath37 & & & & + ngc 1637 & & @xmath38 & @xmath39 & @xmath40 & @xmath41 & & & & & & + ngc 2541 & @xmath42 & @xmath43 & @xmath44 & @xmath45 & & & & & & & + ngc 2841 & @xmath46 & @xmath47 & @xmath48 & @xmath49 & @xmath50 & & & & & & + ngc 3031 & @xmath51 & @xmath52 & & & & & & & & & + ngc 3198 & & @xmath53 & @xmath54 & @xmath55 & & & & & & & + ngc 3319 & @xmath56 & @xmath57 & @xmath58 & @xmath59 & & & & & & & + ngc 3351 & & @xmath60 & @xmath61 & & @xmath62 & @xmath63 & @xmath64 & @xmath65 & & @xmath66 & + ngc 3621 & @xmath67 & @xmath68 & @xmath69 & & & & & & & & + ngc 3621 - 1 & & @xmath70 & & & & & & & & & + ngc 3621 - 2 & @xmath71 & @xmath72 & @xmath73 & & & & & & & & + ngc 3627 & & @xmath74 & @xmath75 & & & & & & & & + ngc 4321 & & @xmath76 & @xmath77 & @xmath78 & & & & & & & + ngc 4414 & & @xmath79 & @xmath80 & @xmath81 & @xmath82 & @xmath83 & & & & & + ngc 4414 - 1 & & @xmath84 & @xmath85 & @xmath86 & @xmath87 & @xmath88 & & & & & + ngc 4414 - 2 & & @xmath89 & @xmath90 & @xmath91 & @xmath92 & @xmath93 & & & & & + ngc 4496a & & @xmath94 & @xmath95 & @xmath96 & & & & & & & + ngc 4527 & @xmath97 & @xmath98 & @xmath99 & @xmath100 & & & & & & & + ngc 4535 & @xmath101 & @xmath102 & @xmath103 & @xmath104 & @xmath105 & & & & & & + ngc 4536 & & @xmath106 & @xmath107 & @xmath69 & & & & & & & + ngc 4548 ( m98 ) & & @xmath108 & @xmath109 & @xmath110 & @xmath111 & @xmath112 & @xmath113 & @xmath114 & @xmath115 & @xmath116 & + ngc 4559 & @xmath117 & @xmath118 & @xmath119 & @xmath120 & & & & & & & + ngc 4571 & @xmath121 & @xmath122 & @xmath123 & @xmath124 & & & & & & & + ngc 4603 & @xmath125 & @xmath126 & @xmath127 & & & & & & & & + ngc 4639 & @xmath128 & @xmath129 & @xmath130 & @xmath131 & & & & & & & + ngc 4725 & & @xmath132 & @xmath133 & @xmath134 & & & & & & & + ngc 5194 & & @xmath135 & @xmath136 & @xmath137 & & & & & & & + ngc 5194 - 1 & & @xmath138 & @xmath139 & @xmath140 & & & & & & & + ngc 5194 - 2 & & @xmath141 & @xmath142 & @xmath143 & & & & & & & + ngc 6946 & & @xmath144 & @xmath145 & @xmath146 & @xmath147 & & & & & & + ngc 7331 & & @xmath148 & @xmath149 & @xmath150 & & & & & & & + ugc 2302 & @xmath151 & @xmath91 & @xmath152 & @xmath153 & & & & & & & + ugc 6614 & & & @xmath154 & @xmath155 & @xmath156 & @xmath41 & @xmath157 & @xmath158 & @xmath159 & @xmath160 & @xmath161 + | dust extinction can be determined from the number of distant field galaxies seen through a spiral disk . to calibrate this number for the crowding and confusion introduced by the foreground image
, @xcite and @xcite developed the `` synthetic field method '' ( sfm ) , which analyses synthetic fields constructed by adding various deep exposures of unobstructed background fields to the candidate foreground galaxy field .
the advantage of the sfm is that it gives the average opacity for area of galaxy disk without assumptions about either the distribution of absorbers or of the disk starlight .
however it is limited by low statistics of the surviving field galaxies , hence the need to combine a larger sample of fields .
this paper presents the first results for a sample of 32 deep hst / wfpc2 archival fields of 29 spirals .
the radial profiles of average dust extinction in spiral galaxies based on calibrated counts of distant field galaxies is presented here , both for individual galaxies as well as for composites from our sample .
the effects of inclination , spiral arms and hubble type on the radial extinction profile are discussed .
the dust opacity of the disk apparently arises from two distinct components ; an optically thicker ( @xmath0 ) but radially dependent component associated with the spiral arms and a relatively constant optically thinner disk ( @xmath1 ) .
these results are completely in agreement with earlier work on occulted galaxies .
the early type spiral disks in our sample show less extinction than the later types .
lsb galaxies and possibly sd s appear effectively transparent .
the average color of the field galaxies seen through foreground disks does not appear to change with radius or opacity .
this grey behaviour is most likely due to the patchy nature of opaque clouds .
the average extinction of a radial annulus and its average surface brightness seem to correlate for the brighter regions .
this leads to the conclusion that the brighter parts of the spiral disk , such as spiral arms , are also the ones with the most extinction associated with them . |
the influence of strong external magnetic fields on the structure of the qcd phase diagram is a very important field of research due to its consequences on several physical phenomena : the measurements in heavy ion collisions at very high energies , the behavior of the first stages of the universe and the understanding of compact astrophysical objects like magnetars . the inclusion of a magnetic field in the lagrangian density of the nambu jona - lasinio ( njl ) model and of the polyakov nambu
jona - lasinio ( pnjl ) model gives rise to the magnetic catalysis ( mc ) effect , i.e. , the enhancement of the quark condensate due to the magnetic field @xcite , but fails to account for the inverse magnetic catalysis ( imc ) found in lqcd calculations @xcite where the suppression of the quark condensate takes place due to the strong screening effect of the gluon interactions . in order to overcome this discrepancy ,
it was proposed , by using the su(2 ) njl model @xcite and the su(3 ) njl / pnjl models @xcite , that the model coupling , @xmath0 , can be seen as proportional to the running coupling , @xmath1 , and consequently , a decreasing function of the magnetic field strength allowing to include the impact of @xmath2 in both models .
indeed , the strong screening effect of the gluon interactions in the region of low momenta weakens the interaction which is reflected into a decrease of the scalar coupling with the intensity of the magnetic field @xcite . since there is no lqcd data available for @xmath2 , by using the njl model we can fit @xmath3 in order to reproduce the pseudocritical chiral transition temperatures , @xmath4 , obtained in lqcd calculations @xcite . the resulting fit function that reproduces the @xmath4 is @xmath5 with @xmath6 , @xmath7 , @xmath8 , and @xmath9 and where @xmath10 .
we also have used @xmath11 mev . in the njl model , the renormalized pseudocritical chiral transition temperatures , @xmath12 ,
are plotted in left panel of fig .
[ fig : temp_crit ] as a function of @xmath13 : with the magnetic field dependent coupling @xmath3 ( green line ) , given by eq .
( [ eq : fit ] ) ; with lqcd results ( red dots ) ; and the usual constant coupling @xmath14 ( black dashed dot line ) , that shows magnetic catalyzes with increasing @xmath12 for all range of magnetic fields .
mev ) as a function of @xmath13 in the njl model with a magnetic field dependent coupling @xmath3 and a constant coupling @xmath15 , and the lattice results @xcite .
( right panel ) the chiral ( @xmath16 ) and deconfinement ( @xmath17 ) transitions temperatures as a function of @xmath13 in the pnjl , using @xmath3 given by eq .
( [ eq : fit ] ) .
, title="fig : " ] mev ) as a function of @xmath13 in the njl model with a magnetic field dependent coupling @xmath3 and a constant coupling @xmath15 , and the lattice results @xcite .
( right panel ) the chiral ( @xmath16 ) and deconfinement ( @xmath17 ) transitions temperatures as a function of @xmath13 in the pnjl , using @xmath3 given by eq .
( [ eq : fit ] ) .
, title="fig : " ] now , using @xmath3 given in eq .
( [ eq : fit ] ) , we calculate the chiral and deconfinement transitions temperatures as a function of @xmath13 in the pnjl model .
the results are shown in the right panel of fig .
[ fig : temp_crit ] : due to the existing coupling between the polyakov loop field and quarks within the pnjl model , the @xmath3 does not only affect the chiral transition but also the deconfinement transition .
consequently , both temperatures transitions decrease with increasing magnetic filed strength .
the nature of the phase transition and the existence of the critical end point ( cep ) are open issues for theoretical studies about the qcd phase diagram @xcite . from the experimental point of view
the existence / location of the cep is also a very timely topic .
this renders important to know the conditions that can change the position of the cep in the phase diagram , namely the presence of strong magnetic fields . in the following ,
we will study two scenarios for the effect of a static external magnetic field on the location of the cep when symmetric matter ( @xmath18 ) is considered : + case i @xmath19 where we take the usual @xmath14 and no imc effects are included ; + case ii @xmath19 where we will use @xmath3 given by eq .
( [ eq : fit ] ) which will allow us to consider the imc effects on the qcd phase diagram . the results for case i are plotted in the left panel of fig .
[ cep ] and reproduce qualitatively the results previously obtained within the njl model in @xcite : as the intensity of the magnetic field increases , the transition temperature increases and the baryonic chemical potential decreases until the critical value @xmath20 gev@xmath21 . for stronger magnetic fields both @xmath22 and @xmath23 increase . in the right panel of fig .
[ cep ] the cep is given in a @xmath22 versus baryonic density plot .
it is seen that when @xmath13 increases from 0 to 1 gev@xmath21 the baryonic density at the cep increases from 2@xmath24 to @xmath25 @xcite . with respect to case ii
the results for the cep are presented in fig .
[ cep_imc ] , red points .
we clearly observe a different behavior when compared with case i ( black points ) : at @xmath26 both cep s coincide but , already for small values of @xmath27 , the cep is moved to lower temperatures and chemical potentials .
nevertheless , until @xmath28 gev@xmath21 the pattern is similar for both cases .
however , for stronger magnetic fields the position of the cep in case ii oscillates between @xmath29 and @xmath30 mev while the chemical potential takes increasingly smaller values : a completely different behavior when compared with case i , where both values of @xmath22 and @xmath23 for the cep increase .
the reason of this behavior lies in fact that the restoration of chiral symmetry is stressed by the decreasing of the coupling @xmath3 .
the increasing of the magnetic filed is not sufficient to counteract this effect as can be seen if fig .
[ cep_mass ] , where we plot the quarks masses ( @xmath31-black line ; @xmath32-red line ; @xmath33-blue line ) as function of @xmath23 for the respective temperature where the cep occurs ( @xmath34 ) at @xmath35 and @xmath36 gev@xmath21 . at @xmath35 gev@xmath21 ( left panel )
@xmath0 is barely affected by the magnetic field , the values of the quark masses are very close to each other for both cases , and the cep occurs at smaller temperatures and at close , but smaller , chemical potentials . when @xmath36 gev@xmath21 , the quark masses in case i are increased with respect to the @xmath26 case ( due to mc effect ) , being the restoration of chiral symmetry more difficult to achieve . however , when @xmath3 , case ii , the masses of the quarks are already smaller than the @xmath26 case ( due to imc effect ) leading to an faster restoration of chiral symmetry at small temperatures and chemical potentials .
eventually , with the increase of @xmath27 , the cep would disappear in the temperature axis and the transition to the chiral restored phase is always of first order .
this work was partially supported by project no .
pest - oe / fis / ui0405 /2014 developed under the initiative qren financed by the ue / feder through the program compete @xmath19 `` programa operacional factores de competitividade '' , and by grant no .
sfrh / bd/51717/2011 from fct .
r. l. s. farias , k. p. gomes , g. i. krein and m. b. pinto , phys .
c * 90 * ( 2014 ) 025203 , [ arxiv:1404.3931 [ hep - ph ] ] .
m. ferreira , p. costa , o. loureno , t. frederico and c. providncia , phys .
d * 89 * ( 2014 ) 116011 , [ arxiv:1404.5577 [ hep - ph ] ] . | the effect of an external magnetic field in qcd phase diagram , namely , in the the location of the critical end point ( cep ) is investigated . using the 2 + 1 flavor nambu-jona - lasinio model with polyakov loop
, it is shown that when an external magnetic field is applied its effect on the cep depends on the strength of the coupling .
if the coupling depends on the magnetic field , allowing for inverse magnetic catalysis , the cep moves to lower chemical potentials eventually disappearing , and the chiral restoration phase transition is always of first order . |
let @xmath6 be the elliptic modular function .
for a positive integer @xmath7 , the classical modular polynomial @xmath8 is the minimal polynomial of the function @xmath9 over the field @xmath10 . as a polynomial in two variables
, we have @xmath11 .
the polynomial @xmath12 is symmetric , with integer coefficients , and has degree @xmath13 in both variables ( * ? ? ?
* ch . 5 ) .
for a nonzero polynomial @xmath14 with complex coefficients , let @xmath15 , where @xmath16 ranges over the coefficients of @xmath14 .
cohen proves in @xcite that the bound @xmath17 holds , with @xmath18 . for @xmath19 prime ,
this yields @xmath20 the purpose of this paper is to prove an explicit upper bound on the @xmath21 term . for every prime @xmath0 we have @xmath22 .
such an explicit bound is needed to obtain rigorous results from algorithms that compute @xmath1 , including @xcite . to prove theorem 1
, we retrace the proof of cohen , specializing to the case that @xmath19 is prime and seeking only an upper bound .
these restrictions simplify our presentation , but the core of the argument remains the same .
our main contribution is to make the bound fully explicit .
we also add two refinements to the proof , lemma 5 and lemma 8 , that sharpen the constants we obtain in theorem 1 .
these improvements have a significant impact in the range of @xmath0 practical for computation .
we recall that the modular polynomial may be written as @xmath23 see ( * ? ? ?
5.2 ) . to simplify our notation , for any @xmath24 ,
let @xmath25 denote the univariate polynomial @xmath26 , and let us define @xmath27 .
[ lemma : hlybound ] let @xmath0 be prime and let @xmath28 . for @xmath29
we have @xmath30 each coefficient @xmath16 of a monic polynomial with roots @xmath31 satisfies @xmath32 .
applying this to ( [ modular_equation ] ) yields the lemma . to bound @xmath2 , it suffices to bound @xmath33 for real @xmath34 in an interval @xmath35 $ ] .
[ lemma : interpolation ] assume @xmath36 for @xmath37 $ ] .
then @xmath38 this follows immediately from lemma [ lemma : a4 ] of the appendix .
as explained in @xcite , the real value @xmath39 increases monotonically with the real parameter @xmath40 .
we have @xmath41 , and for each @xmath34 in the interval @xmath42 $ ] there is a unique @xmath43 for which @xmath44 .
we find that @xmath45 , hence for @xmath46 $ ] we have @xmath47 .
our main task is to bound the sum @xmath48 as a function of @xmath0 , for any fixed @xmath49 .
to do so , we use the following lemma . [
lemma : jbound ] let @xmath28 and let @xmath50 .
we have the bound @xmath51 where @xmath52 for all @xmath53 . from ( * ? ?
* lemma 1 ) we have @xmath54 the lemma follows immediately upon taking logarithms . to profitably apply lemma [ lemma : jbound ] to ( [ sl1 ] )
, we first shift the argument of @xmath55 using the invariance of @xmath56 under @xmath57 .
fixing the integer @xmath58 $ ] , let @xmath59 be the unit interval @xmath60 $ ] .
we may rewrite ( [ sl1 ] ) as @xmath61 where @xmath62 now ranges over the @xmath0 integers in the interval @xmath63 . using the farey series of order @xmath64 , we may partition @xmath59 into subintervals as @xmath65 where @xmath66 contains @xmath67 .
the endpoints @xmath68 and @xmath69 both satisfy @xmath70 see ( * ? ? ?
* ch . 3 ) and ( * ? ? ?
* lemma 3 ) .
there are @xmath71 subintervals in ( [ ipartition ] ) , and each contains at least one @xmath72 .
the bound @xmath73 is well known ( * ? ? ?
3.7 ) , and for @xmath74 we have the explicit bound @xmath75 proven in lemma [ lemma : a2 ] of the appendix ( see @xcite for an asymptotically tighter bound ) .
we now assume @xmath76 , so that @xmath74 .
the bound we eventually prove will also hold when @xmath77 . for each @xmath78 in @xmath79 , let us pick integers @xmath80 and @xmath81 with @xmath82 .
we may then apply the unimodular transformation @xmath83 to @xmath84 in each term of ( [ sl2 ] ) .
using @xmath85 , we obtain @xmath86 now let @xmath87 and @xmath88 , so that @xmath89 note that each @xmath72 is contained in a particular @xmath78 that determines @xmath90 and @xmath91 .
we may now bound @xmath92 by @xmath93 the notation in the right - hand side of ( [ sl3 ] ) is as follows .
for each @xmath78 we pick a unimodular matrix @xmath94 as above , and with @xmath95 we put @xmath96 and @xmath97 , where @xmath98 is the function defined in lemma [ lemma : jbound ] . as the following lemma shows ,
the nave bound @xmath99 that we get from lemma [ lemma : jbound ] can be significantly improved , on average .
[ lemma : epsilonbound ] with the notation as above , let @xmath100 . for all
@xmath76 and @xmath101 we have @xmath102 among the points of the form @xmath103 in the interval @xmath78 , we call the smallest and largest _ exterior _ points .
the others are called _ interior _ points . for each interior point @xmath104
we have @xmath105 yielding @xmath106 .
we now divide the exterior points into two categories .
we call an exterior point _
good _ if it is at least a distance @xmath107 from both endpoints of the interval in which it lies .
an exterior point is _ bad _ otherwise . by a calculation similar to that
above , we find that @xmath108 holds for each good exterior point .
there are @xmath109 half intervals of the form @xmath110 or @xmath111 , and each contains at most one exterior point .
adjacent half intervals that do not lie in the same @xmath78 can not both contain a bad exterior point .
hence , there are at most @xmath112 bad exterior points . for the purposes of computing an upper bound
, we assume that there are exactly this many , and that every interval has two exterior points , implying that there are exactly @xmath113 good exterior points .
in the worst case , the bad exterior points lie in the intervals @xmath78 with the largest values of @xmath91 .
if we order the intervals according to @xmath114 and divide them into quartiles , this roughly means that each interval in the top two quartiles contains two bad exterior points , while each interval in the bottom two quartiles contains two good exterior points . by corollary [ cor : a3 ] of the appendix , up to an absolute error of 2 , at least of the intervals
have @xmath115 , at least have @xmath116 , and at least @xmath117 have @xmath118 .
let @xmath119 , with @xmath98 as in lemma [ lemma : jbound ] .
we use @xmath120 and @xmath121 to estimate @xmath122 for the bad exterior points in the upper two quartiles , and we use @xmath123 and @xmath124 for the good exterior points in the lower two quartiles . for the interior points ,
we note that for @xmath125 , the interval @xmath78 has width less than @xmath126 and can contain at most one interior point .
similarly , if we have @xmath127 , then there can be at most two interior points .
we may therefore use @xmath128 to estimate @xmath122 for the interior points in the top quartile ( at most one per interval ) , @xmath129 for the interior points in the second quartile ( at most two per interval ) , and @xmath130 for the remaining interior points . applying lemma [ lemma : a2 ]
, we use @xmath131 to bound @xmath132 . putting everything together
, we obtain the function @xmath133 as an upper bound on @xmath134 .
for any fixed @xmath76 we find that for @xmath101 the function @xmath135 is maximized when @xmath136 , and in fact the bound @xmath137 holds for all @xmath76 and @xmath101 .
computing @xmath138 then yields the lemma .
our remaining task is to bound the terms @xmath139 appearing in ( [ sl3 ] ) . setting @xmath140
, we have @xmath141 which we view as a function of @xmath142 .
we now apply the identity @xmath143 as in ( * ? ? ?
* lemma 6 ) , to obtain @xmath144 and we use the following lemma to bound the right - hand side .
[ lemma : tsumbound ] let @xmath0 be prime , let @xmath145 , and for @xmath146 , define @xmath147 .
assume @xmath58 \ge 2 $ ] .
then we have @xmath148 noting absolute convergence , we reorder the sums , obtaining @xmath149 with @xmath150 the inner sum becomes @xmath151 for @xmath152 this is a ramanujan sum , and we may apply the identity @xmath153 due to hlder @xcite ( or see ( * ? ? ? * thm .
we then have @xmath154 for all @xmath152 , and for @xmath155 we note that @xmath156 . for @xmath157 we therefore have @xmath158 where @xmath159 . since @xmath160 , we may bound ( [ triplesum ] ) by @xmath161 to treat the infinite sum in ( [ twosums ] ) we evaluate the first two terms and bound the tail with an integral , using @xmath162 .
this yields @xmath163 where we have used @xmath164 valid for @xmath40 .
we now apply lemma [ lemma : a1 ] of the appendix , whose error term is an increasing function of @xmath165 , to the first sum in ( [ twosums ] ) . using @xmath58\le \sqrt{l}$ ]
we obtain @xmath166 applying ( [ leftsum ] ) and ( [ rightsum ] ) to ( [ twosums ] ) yields the lemma . extending the sum in ( [ ghksum ] ) to all integers allowed us to conveniently bound the triple sum in ( [ sl3 ] ) via lemma [ lemma : tsumbound ] .
this yields the correct leading term of @xmath167 in theorem 1 , however the @xmath21 term is overestimated significantly .
lemma [ lemma : ghktail ] proves a lower bound on the tail of the middle sum in ( [ ghksum ] ) which is then subtracted from the bound in lemma [ lemma : tsumbound ] to sharpen the @xmath21 term .
[ lemma : ghktail ] let the notation be as above .
for @xmath168 we have @xmath169 consider an interval @xmath78 for which @xmath170 . by ( [ endpoints ] ) ,
the width of @xmath78 is less than @xmath171 , hence it contains at most two points of the form @xmath103 .
it follows that at most two of the terms @xmath172 in the middle sum of ( [ ghksum ] ) correspond to terms present in the left sum .
we assume there are exactly two , and these must be of the form @xmath139 and @xmath173 .
the sum is maximized when @xmath174 , hence the overestimate introduced by the inequality in ( [ ghksum ] ) is at least @xmath175 where we have used the identity @xmath176 . for each @xmath177 $ ] , there are @xmath178 intervals @xmath78 to which ( [ tanhbound ] ) applies .
the total overestimate related to these intervals is at least @xmath179 to bound the sum in ( [ overhk ] ) , we apply lemma [ lemma : a1 ] twice and take the difference : @xmath180 the term @xmath181 arises from the error term @xmath182 in lemma [ lemma : a1 ] .
we make the worst - case assumption that the upper error is maximally negative and the lower error is maximally positive , yielding @xmath183 combining ( [ overhk ] ) and ( [ overhksum ] ) we obtain the lower bound @xmath184 we now apply the same argument to intervals @xmath78 with @xmath185 . for such intervals
there are at most three terms in the middle sum of ( [ ghksum ] ) that correspond to terms in the left sum , and the sum of these terms is maximized when the middle term has @xmath186 .
using the identity @xmath187 , we obtain @xmath188 to compute @xmath189 , we now take the difference of the maximal positive errors arising from lemma [ lemma : a1 ] , consistent with our worst - case assumption in ( [ eps1 ] ) above , obtaining @xmath190 continuing in this fashion , we consider @xmath78 with @xmath191 for an integer @xmath192 .
let us define @xmath193 for @xmath194 and @xmath195 we then obtain the lower bound @xmath196 assuming @xmath197 , we let @xmath198 . for any fixed @xmath197 , as @xmath43 varies over @xmath199 we find that @xmath200 is minimized when @xmath136 ( we note that this is not true of the first few @xmath201 ) .
we then compute @xmath202 as a lower bound on the total overestimate in the middle sum of ( [ ghksum ] ) . subtracting the right - hand side of ( [ deltabound ] ) from the bound in lemma 6 yields the lemma . for every prime @xmath0 we have @xmath22 .
the modular polynomials @xmath1 are well known for @xmath203 , and the theorem holds in these cases , so assume @xmath204 . let @xmath205 $ ] , and let @xmath101 satisfy @xmath206 , as discussed following lemma [ lemma : interpolation ]
. from lemma [ lemma : hlybound ] we have @xmath207 where @xmath92 is the sum in ( [ sl3 ] ) . applying lemma [ lemma : jbound ] ,
we obtain @xmath208 for @xmath92 we have two different bounds , depending on whether @xmath0 is less than or greater than 3600 , and this yields two bounds for @xmath2 . for @xmath209 , we bound @xmath92 by adding the bound in lemma [ lemma : epsilonbound ] to the bound in lemma [ lemma : tsumbound ] . plugging this and ( [ s1 ] ) into ( [ hly ] ) yields @xmath210 for
@xmath101 this bound is maximized when @xmath136 .
lemma [ lemma : interpolation ] then yields @xmath211 for @xmath212 we instead bound @xmath92 by adding the bound in lemma [ lemma : epsilonbound ] to the bound in lemma [ lemma : ghktail ] . in this case
we obtain @xmath213 a direct computation , using the bound @xmath214 and the algorithm described in @xcite , finds that theorem 1 holds for @xmath215 ( see table 1 ) . for all @xmath212 ,
the bound of the theorem is greater than @xmath216 , hence the theorem 1 holds for @xmath212 .
[ cor1 ] for every prime @xmath0 we have @xmath217 .
for @xmath212 the corollary follows from the bound @xmath216 in ( [ b2 ] ) . for @xmath215
it is verified by the values of @xmath2 in table 1 .
table 1 is listed on the following three pages in a three column format . for each prime @xmath218 ,
the table lists three values .
the first is the integer @xmath219 this represents the number of bits required to store the absolute value of the largest coefficient of @xmath1 , and we have @xmath220 .
the second is @xmath221 which reflects the constant in the @xmath21 term of @xmath2 .
the third value is @xmath222 the ratio of the bound in corollary [ cor1 ] to the height of @xmath1 . within the range of table 1
, we find that for @xmath223 the bound of corollary [ cor1 ] exceeds @xmath2 by less than fifteen percent , as indicated by the values for @xmath224 .
however , as shown by the values for @xmath225 , our bounds are clearly not optimal .
based on the data in the table , we make the following conjectures . for every prime @xmath226 we have @xmath227 . for @xmath0 prime we have @xmath228
[ cols=">,>,>,>,>,>,>,>,>,>,>,>",options="header " , ]
here we prove certain number - theoretic bounds used in our estimates above .
in most cases stronger asymptotic results are known , but we seek fully explicit bounds .
our results are derived from bounds for the summatory functions @xmath229 where @xmath230 is the mbius function .
the function @xmath231 counts squarefree positive integers , and it is well known that @xmath232 , see @xcite for example .
we let @xmath233 , and note the following result from @xcite .
the proofs below use the riemann - stieltjes integral , as defined , for example , in ( * ? ? ?
* ch . 7 ) .
the existence of all the integrals we use , including improper integrals , is easily verified and assumed without further comment . for a real number @xmath242
, we use @xmath243 $ ] to denote the largest integer @xmath244 and define @xmath245 $ ] . to make our error terms explicit , for any positive real value
@xmath246 we use @xmath247 to denote an unspecified real number with absolute value at most @xmath246 . throughout this appendix ,
all sums range over positive integers .
both bounds are verified by machine for @xmath255 , so we assume @xmath238 .
the first result is standard , but some care is required to express the bound in terms of @xmath242 rather than the integer @xmath243 $ ] . from (
* ? ? ? * eq . 9.89 )
we have @xmath256 } { \medspace=\medspace}\log [ x ] + \gamma + \frac{1}{2[x ] } - \frac{1}{12[x]^2 } + \frac{1}{120[x]^4 } - \frac{\theta}{256[x]^6},\ ] ] where @xmath257 is a positive real number less than 1 .
we wish to bound the quantity @xmath258 } { \medspace=\medspace}\log\left(1+\frac{\{x\}}{[x]}\right ) - \frac{1}{2[x ] } + \frac{1}{12[x]^2 } - \frac{1}{120[x]^4 } + \frac{\theta}{256[x]^6}.\\\ ] ] expanding @xmath259}\right)$ ] , the right - hand side of ( [ heq1 ] ) can be expressed as @xmath260 } - \frac{\{x\}^2}{2[x]^2 } + \frac{\{x\}^3}{3[x]^3 } - \frac{\{x\}^4}{4[x]^4 } - \frac{1}{2[x ] } + \frac{1}{12[x]^2 } - \frac{1}{120[x]^4 } + \epsilon\left(\frac{1}{5[x]^5}\right).\ ] ] we now apply @xmath261}=\frac{1}{x}+\frac{\{x\}}{x[x]}$ ] repeatedly to ( [ heq2 ] ) , which eventually yields @xmath262 } { \medspace=\medspace}\frac{2\{x\}-1}{2x } & { \medspace+\medspace}\frac{6\{x\}^2 - 6\{x\}+1}{12x^2 } { \medspace+\medspace}\frac{2\{x\}^3 - 3\{x\}^2+\{x\}}{6x^3}\\ & { \medspace+\medspace}\frac{30\{x\}^4 - 60\{x\}^3 + 30\{x\}^2 - 1}{120x^4 } { \medspace+\medspace}\epsilon\left(\frac{5}{[x]^5}\right).\end{aligned}\ ] ] recalling that @xmath238 , we note that if @xmath263 then the sum of the terms on the rhs is less than @xmath264 .
when @xmath265 then the first two terms are both positive , with sum less than @xmath266 , while the sum of the remaining three terms is negative , with absolute value less than @xmath266 .
this proves the first result . to prove the second result
, we integrate by parts and apply corollary [ mubounds ] : @xmath267 subtracting @xmath268 from both sides and taking the limit as @xmath269 yields @xmath270 corollary [ mubounds ] implies @xmath271 , and the total error is @xmath272 .
for the first bound , we recall that @xmath275 .
therefore @xmath276 to bound the tail , we integrate by parts and apply corollary [ mubounds ] : @xmath277 the second bound is proved similarly . when bounding the tail one uses @xmath278 to complete the proof .
the lemma is machine verified for @xmath255 , so we assume @xmath282 .
we recall that @xmath283 , and therefore @xmath284 by lemma [ lemma : harmonic ] we have @xmath285 , which yields @xmath286 applying lemma [ lemma : musums ] to the first two sums and lemma [ lemma : harmonic ] to the third sum yields @xmath287 for @xmath282 we can replace the error term by @xmath288 .
the lemma is machine verified for @xmath291 , so we assume @xmath238 . as above , we apply @xmath283 to obtain @xmath292 ^ 2+\left[\frac{x}{d}\right]\right).\ ] ] we note that @xmath293=1 $ ] , by ( * ? ? ?
* thm . 3.12 ) , and we have @xmath294 ^ 2 & { \medspace=\medspace}\sum_{d\le x}\mu(d)\left(\frac{x}{d}-\left\{\frac{x}{d}\right\}\right)^2\\\notag & { \medspace=\medspace}\sum_{d\le x}\mu(d)\left(\frac{x}{d}\right)^2{\medspace-\medspace}2\sum_{d\le x}\mu(d)\left\{\frac{x}{d}\right\}\left(\frac{x}{d}\right){\medspace+\medspace}\sum_{d\le x}\mu(d)\left\{\frac{x}{d}\right\}^2\\\notag & { \medspace=\medspace}x^2\sum_{d\le x}\frac{\mu(d)}{d^2}{\medspace-\medspace}2x\sum_{d\le x}\left\{\frac{x}{d}\right\}\frac{\mu(d)}{d } { \medspace+\medspace}\epsilon(x).\end{aligned}\ ] ] this first sum is addressed by lemma [ lemma : musums ] . for the second we use @xmath295 by (
* lemma 3.2 ) , the first sum is @xmath296 .
applying lemma [ lemma : harmonic ] we obtain @xmath297 since @xmath298 , the error term is @xmath299 . applying this to ( [ a2eq2 ] ) yields @xmath300 ^ 2{\medspace=\medspace}\frac{6x^2}{\pi^2 } { \medspace+\medspace}\epsilon\left(\frac{x}{300}\right ) + \epsilon\left(\frac{6x\log x}{\pi^2}+\frac{21x}{10}\right ) { \medspace+\medspace}\epsilon(x),\ ] ] and from ( [ a2eq1 ] )
we then have @xmath301 since @xmath302 , we may replace the error term with @xmath303 .
let @xmath319 for @xmath320 . if @xmath321 is any @xmath322-subset of @xmath323 then we have @xmath324 .
therefore @xmath325 ) , and we need only consider @xmath326 .
now suppose @xmath327 , with @xmath328 .
we then have @xmath329 which implies @xmath330 .
thus we may assume @xmath331 .
the lemma clearly holds for @xmath332 , so we assume @xmath333 and set @xmath334 $ ] .
we then have @xmath335 and it remains only to show that @xmath336 . using the explicit bounds @xmath337 for stirling s formula @xcite , one obtains the inequality @xmath338 valid for @xmath339 . for fixed @xmath313 and @xmath326 the right - hand side of ( [ binombound ] )
is an increasing function of @xmath322 .
we thus obtain @xmath340 using @xmath341 ( since @xmath333 ) .
taking logarithms , we have @xmath342 which completes the proof .
[ lemma : a4 ] let @xmath343 $ ] be a nonzero polynomial of degree at most @xmath313 in each variable .
suppose @xmath344 for all real @xmath34 in the interval @xmath35 $ ] , for some real @xmath345 and @xmath346 .
then we have @xmath347 let @xmath348 .
it suffices to show @xmath349 satisfies the bound in the lemma for each nonzero @xmath350 . for any @xmath24 ,
the coefficient of @xmath351 in @xmath352 is @xmath353 . as in (
* lemma 9 ) , let us pick @xmath354 interpolation points @xmath355 $ ] : @xmath356 we may interpolate @xmath357 from the pairs @xmath358 , where @xmath359 is the coefficient of @xmath351 in @xmath360 . applying the lagrange interpolation formula yields @xmath361 by lemma [ lemma : heightfromroots ] , the coefficients of @xmath362 have absolute values bounded by @xmath363 , where @xmath364 and @xmath365 we also have @xmath366 and by the hypothesis of the lemma , @xmath367 , where @xmath368 .
thus the absolute values of the coefficients of @xmath357 are bounded by @xmath369 assuming @xmath357 is nonzero , we take logarithms and obtain @xmath370 as desired . | for a prime @xmath0 , let @xmath1 be the classical modular polynomial , and let @xmath2 denote its logarithmic height . by specializing a theorem of cohen , we prove that @xmath3 . as a corollary
, we find that @xmath4 also holds .
a table of @xmath2 values is provided for @xmath5 . |
order @xmath0 qcd corrections to structure functions containing heavy quarks and to single heavy quark inclusive distributions in deep - inelastic electroproduction ( @xmath1 , @xmath2 where @xmath3 stands for any final hadronic state allowed by quantum - number conservation and @xmath4 is a proton of momentum @xmath5 ) were recently published in @xcite and @xcite , respectively . by combining the next - to - leading order ( nlo ) heavy quark structure functions with the corresponding light - quark structure functions @xcite ,
the heavy quark content of the nucleon has been studied as a function of @xmath6 and @xmath7 @xcite .
event rates for charm production integrated over bins in @xmath8 and @xmath9 relevant to hera data have been calculated in @xcite .
to further the study of deep - inelastic electroproduction of heavy quarks we have recalculated the virtual - photon - parton cross sections of @xcite in an exclusive fashion @xcite .
this enables us to study the single and double differential distributions and correlations among all outgoing particles in the virtual - photon induced reaction @xmath10 with @xmath11 or @xmath12 jet and to easily incorporate experimental cuts . by examining distributions and correlations that are trivial at lowest order ( for example
, the azimuthal angle between the heavy quark and the heavy antiquark ) one directly tests the hard scattering cross section that is predicted by perturbative qcd and factorization .
therefore our results should give a clean test of nlo perturbative qcd .
nevertheless we remind the reader that this is a fixed order perturbative calculation and suffers from the same problems as all nlo calculations .
there are regions in phase space where it will break down .
for example , in the above mentioned azimuthal angle distribution , if one looks at the prediction too near the back - to - back configuration there will be an extra enhancement of the cross section due to multiple soft gluon emission which is not included in our nlo result . in this short report
we discuss some details of the calculation and present interesting distributions .
the transverse and longitudinal photon components are treated separately and the latest cteq3 parton densities @xcite , consistent with the newly released hera data @xcite , are used in the kinematic regime appropriate for production of charm quarks at hera .
we make our predictions at fixed values of @xmath9 @xmath13 and @xmath8 @xmath14 . we stress that here we only consider _ extrinsic _ heavy quark production , involving bethe - heitler and compton production from massless partons . for a discussion of _ intrinsic _ production , where the heavy quark is considered to be part of the proton s wavefunction ,
see brodsky _ et al_. @xcite .
a variable flavor scheme which joins the extrinsic heavy flavor production picture at @xmath15 with a light mass parton density picture at @xmath16 has been discussed by aivazis _
et al_. @xcite . by comparing the fixed flavor scheme calculation of @xcite with the variable flavor number scheme of @xcite
it is concluded in @xcite that the former yields the most stable and reliable results for @xmath17 in the threshold region ( @xmath18 where @xmath19 is the mass of the heavy quark ) .
the contribution by w.l .
van neerven to these proceedings reviews the progress towards implementing the variable flavour number scheme in nlo @xcite . in our exclusive computation
we use the subtraction method which is based on the replacement of divergent ( collinear or soft ) terms in the squared matrix elements by generalized plus distributions .
this allows us to isolate the soft and collinear poles within the framework of dimensional regularization without calculating all the phase space integrals in a spacetime dimension @xmath20 as usually required in a traditional single particle inclusive computation . in this method
the expressions for the squared matrix elements in the collinear limit appear in a factorized form , where poles in @xmath21 multiply splitting functions and lower order squared matrix elements .
the cancellation of collinear singularities is then performed using the factorization theorem @xcite .
the expressions for the squared matrix elements in the soft limit appear in a factorized form where poles in @xmath21 multiply lower order squared matrix elements .
the cancellation of soft singularities takes place upon adding the contributions from the renormalized virtual diagrams .
since the final result is in four - dimensional space time , we can compute all relevant phase space integrations using standard monte carlo integration techniques @xcite and produce histograms for exclusive , semi - inclusive , or inclusive quantities related to any of the outgoing particles .
we can also incorporate any reasonable set of experimental cuts .
our computer code has no small phase space slice parameters and the parameters defining the generalized plus distributions may be tuned to give fast numerical convergence , which is an advantage of using this subtraction method .
analysis of the partonic reaction @xmath23 does not involve soft or virtual contributions so we can use it to most simply explain the method .
one particular graph out of the order @xmath24 set is shown in fig.1 .
we project this set of graphs on a particular photon polarization state @xmath25 and square the amplitude .
the answer is written as @xmath26 where the coefficient absorbs overall factors .
( 7,7 ) now we write the photon - parton cross section as @xmath27 where the 3 indicates a three - body final state . to proceed we need some notation for the variables
@xmath28 , @xmath29 , @xmath30 , @xmath31 , @xmath32 and @xmath33 . then if we introduce @xmath34 and @xmath35 we can write @xmath36 .
this representation shows that @xmath37 as @xmath38 .
this @xmath39 divergence yields the collinear divergence in @xmath40 , which is due to the presence of the feynman graph in fig.1 . to handle this divergence
we multiply and divide by @xmath39 so that @xmath41 where @xmath42 is now finite as @xmath43 and @xmath44 is divergent .
we next replace the factor @xmath45 in the ( @xmath46-dimensional ) three - body phase space , where @xmath47 .
this has the effect of splitting the @xmath46-dimensional cross section into two divergent parts .
however it lets us take the limit @xmath48 and get back to 4-dimensions , with @xmath49 regularizing the collinear singularity . replacing the divergent factor @xmath50 in @xmath51 one obtains the following decomposition : ( for more details see @xcite ) @xmath52 with @xmath53 and @xmath54
the function @xmath55 satisfies @xmath56 with @xmath57 for reaction ( 1 ) we find that @xmath58 \nonumber \\ & \times & ( 1+\epsilon/2)^{-1 } b_{i , qed } ( xk_1).\end{aligned}\ ] ] where @xmath59 it the born amplitude with @xmath60 replaced by @xmath61 .
the factor in the square brackets is the altarelli - parisi splitting function in @xmath46 dimensions so we can define new ( collinear free ) cross sections @xmath62 with @xmath63 \big\ } \
, , \nonumber \\ & & \\
d\sigma^{(f)}_{i , q } & = & - \big ( \frac{1}{16 \pi^2 } \big)^2 c_{i , q } \beta_5 f^q_i(x , y,\theta_1,\theta_2 ) \nonumber \\ & \times & \big ( \frac{1}{1+y } \big)_{\omega } dx dy \sin \theta_1 d\theta_1d\theta_2 . \end{aligned}\ ] ] the finite functions @xmath64 are available in @xcite .
as the quark channel only contains collinear singularities , we also use it to illustrate how the generalized plus distributions are implemented numerically and how the @xmath49 dependence disappears in the sum ( [ quarkfinal ] ) . to this end consider @xmath65 where we have suppressed all indices and arguments of @xmath66 other than @xmath67 . in the bottom line
we see that the infinity encountered at the lower integration limit @xmath68 is cancelled in the sum of the two integrals . in practice
one can only reasonably take the lower limit to be @xmath69 where @xmath70 in double precision fortran before round off errors enter . none the less ,
the final results are stable with respect to the variation of @xmath71 in the range @xmath72 to @xmath73 .
the upper limit of the second integral gives a contribution @xmath74 which cancels against the @xmath75 term in @xmath76 .
the first integral in the bottom line is commonly called the `` event '' and has a positive definite weight .
the second integral plus the factorized collinear contribution is commonly called the `` counter - event '' and may have either positive or negative weight . the implementation of this procedure for the gluon channel is similar but more complicated due to the presence of both soft and collinear divergences .
recalling that the probability density is related to the momentum density via @xmath77 we write the hadronic cross section as @xmath78 we now specialize to the case where @xmath79 is a proton , as in the case of hera . using the relations @xmath80 where @xmath81 with @xmath82 , and relations for the scaling functions , we find @xmath83 \big\ } \ , , \nonumber \\ & & \end{aligned}\ ] ] where @xmath84 .
we have set @xmath85 and shown the @xmath86 dependence of the running coupling @xmath0 explicitly .
the lower boundary on the integration is given by @xmath87 .
this formula yields the standard heavy quark structure functions @xmath17 and @xmath88 for electron proton scattering , and we will present results as differentials of these functions . from the formalism described in the previous section we are left with events of positive weight and counter - events of either positive or negative weight .
our program outputs the final state four vectors of the event and counter - event together with the corresponding weight .
we histogram these into bins to produce differential distributions .
( 7,7 ) we start by considering the production of charm quarks .
we use @xmath89 and simply choose the factorization ( renormalization ) scale as @xmath90 .
note that there are many possible choices of scale as we have all components of the final four vectors . aside from the @xmath91 dependence
, this choice reduces to @xmath92 for electroproduction of massless quarks and @xmath93 for the photoproduction of charm quarks .
we introduce a @xmath91 dependence by adding in the average of the magnitude of the transverse momenta of the heavy quark and heavy antiquark .
as mentioned earlier we use the cteq3 m parton densities @xcite in the @xmath94 scheme and the two loop @xmath0 with @xmath95 .
the first distribution we present basically measures the transverse momentum of the additional jet which recoils against the heavy quark pair .
the @xmath91 distribution of the charm - anticharm pair is shown in fig . 2 where we plot @xmath96 as a function of @xmath91 .
( 7,7 ) the histograms are presented at fixed @xmath9= 12 @xmath97 for @xmath8 values of @xmath98 , @xmath99 , @xmath100 and @xmath101 respectively .
one sees that the @xmath91 distribution peaks at small @xmath91 and has a negative contribution in the lowest bin .
this is a region where the dominant contribution is from counter - events so the weights can be negative .
the results of this calculation require missing contributions from even higher order perturbation theory before this bin will have a positive weight .
the depth of the negative bins is a function of @xmath8 , @xmath9 , and the choice of scale .
note that at larger @xmath91 the structure function is dominated by the contribution from the square of the bremsstrahlung graphs so the weights are positive .
figure 3 shows the corresponding results for fixed @xmath102 plotted for the @xmath9 values of @xmath103 @xmath104 , @xmath105 @xmath104 , @xmath106 @xmath104 and @xmath107 @xmath104 respectively .
the distributions peak near small @xmath91 and are either small or negative in the first bin .
the histograms with @xmath108 and @xmath109 ( the dotted line ) are the same in figs . 2 and 3 .
we have also used the same scales on the axes so one can easily see that there is a greater change if we fix @xmath8 and vary @xmath9 than if we fix @xmath9 and vary @xmath8 .
we will continue to use the same scale for all the pairs of later plots to simplify the comparison between them .
we now turn to the distributions in the azimuthal angle between the outgoing charm quark and charm antiquark which we denote as @xmath110 .
this is the angle between the @xmath111 vectors of the heavy quark - antiquark in the c. m. frame of the virtual - photon - hadron system .
since we integrate over the azimuthal angle between the plane containing the incoming and outgoing leptons and the plane defined by the incoming parton and outgoing heavy quark ( to limit our discussion to @xmath112 and @xmath113 ) we can only plot relative azimuthal correlations . in the born approximation this distribution is a delta function at @xmath114 , as their four momenta must balance . due to the radiation of the additional light mass parton ,
the distribution has a tail extending below @xmath114 and has a valley at @xmath114 .
the distributions become negative in the highest bins .
this negative region is a general feature of all exclusive calculations .
figure 4 shows results for @xmath115 at the same values of fixed @xmath9 and variable @xmath8 as chosen previously in fig .
2 , while fig . 5 shows the results for fixed @xmath8 and variable @xmath9 as chosen previously in fig .
3 . note again that the dotted histograms are the same in figs . 4 and 5 , and there is more variation for fixed @xmath8 and changing @xmath9 than for fixed @xmath9 and changing @xmath8 . in this report
we have outlined the nlo calculation of the virtual - photon - parton ( wilson ) coefficient functions in the fully differential production of heavy quarks plus one jet .
more correlations and checks are given in @xcite .
the computer program we have written for the exclusive calculation has the advantage that the four vectors of the heavy quark , heavy - antiquark and/or one additional light parton jet are produced for each event and can be subjected to experimental cuts .
we have not done this in any of the plots shown here but the computer program is available and can be easily modified to incorporate acceptances of the detectors at hera .
the program has been used to reanalyse the emc data on charm production in @xcite .
in this report we have presented some plots which depend on information from the four vector of the additional jet .
we showed the distributions in the transverse - momentum ( @xmath91 ) of the heavy quark antiquark pair and in the azimuthal angle ( @xmath110 ) between the @xmath111 vectors of the heavy quark and heavy antiquark .
all quantities were predicted in the c. m. frame of the photon - proton system after integration over the azimuthal angle between the plane containing the incoming and outgoing lepton and the plane containing the incoming proton and outgoing heavy quark .
the results were presented as distributions in @xmath17 at specific points in @xmath8 , @xmath9 and @xmath116 .
none of these distributions can be reproduced by any @xmath117-factor multiplication as the corresponding born distributions are proportional to delta - functions . in all cases
the histograms of these distributions have negative bins .
these are regions where the nlo calculation is not sufficient and a nnlo order calculation ( or some form of resummation ) should be made .
a general statement about the magnitude of the nlo contribution compared to the lo one is difficult to make as the size and sign of the corrections may vary strongly between different regions of phase space .
however , we see that all plots have larger @xmath9 variation at fixed @xmath8 as comparied to varying @xmath8 at fixed @xmath9 . by varying the renormalization ( factorization ) scale
we observed that the distributions presented here changed in normalization but not in shape . in our study
we also calculated averages of various quantities and found the typical variation between central and extreme scale choices of 5 percent for charm production .
s. j. brodsky , p. hoyer , a. h. mueller and w - k .
tang , nucl .
b369 * ( 1992 ) 519 ; g. ingelman , l. jnsson and m. nyborg , phys . rev . * d47 * ( 1993 ) 4872 . for a complete list of references
see r. vogt and s. j. brodsky nucl
* b438 * ( 1995 ) 261 . | we discuss some results for heavy quark correlations in next - to - leading order in deep inelastic electroproduction . |
the homeomorphism type of a compact , closed , simply connected symplectic 4-manifold is uniquely determined by its euler characteristic @xmath3 , signature @xmath4 , and the intersection form @xmath5 .
if @xmath5 is indefinite , as it happens in most of the cases , it is determined by @xmath3,@xmath4 and its _
parity_. it is natural to ask for which combinations ( @xmath3 , @xmath4 , @xmath6 ) there exists such symplectic manifold , and how many distinct representatives there are .
the first question is called the _ geography problem _ , the second is referred to as the _ botany_.
both have been previously studied in the context of complex surfaces .
symplectic 4-manifolds can be equipped with an almost complex structure , and this provides the possibility to use some notions coming from the complex geometry . in particular , we can define chern classes and chern numbers .
chern classes , in general , depend on the choice of an almost complex structure , but @xmath7 is the same as euler characteristic @xmath3 , and @xmath8 , which in the complex case denotes the self - intersection of a canonical divisor , can be extended uniquely by hirzebruch signature formula @xmath9 .
similarly , we can define the holomorphic euler characteristic @xmath10 by setting @xmath11 .
the existence of the almost complex structure guarantees this number to be an integer .
the convention in 4-manifold geography is to use pairs ( @xmath10 , @xmath8 ) instead of ( @xmath3 , @xmath4 ) , and we will follow it in this article .
important fact about chern numbers is the bogomolov
miyaoka yau inequality , asserting that @xmath12 holds for complex surfaces of general type .
it was conjectured by by fintushel and stern that this relation holds also for symplectic 4-manifolds .
most recent attempts ( @xcite , @xcite , @xcite ) in exploring the geography problem for simply connected symplectic 4-manifolds of positive signature , or equivalently those with @xmath13 , provided families of examples close to the bogomolov miyaoka yau line ( b m y line ) .
these constructions followed similar pattern started with a complex surface with nontrivial fundamental group lying on , or close to the b
m y line , and performed an operation which preserved symplectic structure and reduced the topological complexity of the original surface .
the operation in use was either symplectic connected normal sum used several times , fibered knot surgery or logarithmic transformation in a cusp neighborhood .
the botany question for complex surfaces has a partial answer the number of non - diffeomorphic complex surfaces of general type in a given homeomorphism type is always finite , although there is no uniform bound on this number .
the other large family , elliptic surfaces , is also well understood ( see @xcite for further references ) .
however , for simply connected elliptic surfaces in a given homeomorphism type there are infinitely many different smooth structures .
these two results reflects the nature of restrictions given by the complex ( geometric ) structure . in symplectic case ,
the picture is less clear due to the availability of many gluing constructions , which are topological in nature .
fintushel and stern in @xcite provided another infinite family of symplectic 4-manifolds with different seiberg
witten ( @xmath14 ) invariants , all of them homeomorphic to the standard @xmath15 surface @xmath16 .
the method , fibered knot surgery , is different from the logarithmic transformation , and can be used with similar success for any symplectic 4-manifold containing cusp fiber and satisfying some additional properties ( e.g. existence of a @xmath17-embedded torus fiber ) . in particular
, it can be used also for symplectic manifolds coming from ( some ) complex surfaces of general type .
this indicates that the botany problem for symplectic 4-manifolds comprises infinitely many diffeomorphism types for a much larger class of homeomorphism types labeled by triples ( @xmath10 , @xmath8 , @xmath6 ) . in the following
we will construct a family @xmath18 of simply connected symplectic 4-manifolds close to the b m
moreover , for each @xmath18 there exist infinitely many mutually non - diffeomorphic symplectic 4-manifolds , all homeomorphic to @xmath18 .
[ thm1 ] there are simply connected symplectic 4-manifolds @xmath18 with chern numbers @xmath19 , @xmath20 , holomorphic euler characteristic @xmath21 , and signature @xmath22 .
* acknowledgement : * the author would like to thank zoltn szab for many helpful discussions and suggestions .
in @xcite , hirzebruch gave an example of a sequence @xmath23 of minimal complex surfaces of general type with the following chern numbers @xmath24 @xmath25 we briefly recall his construction , and also discuss some topological properties of smooth models of these algebraic surfaces .
let @xmath26 and @xmath27 be the elliptic curve @xmath28 let @xmath29 be a complex surface , and denote its points by @xmath30 .
there is a well defined complex multiplication by algebraic integers @xmath31 $ ] on @xmath27 .
the elliptic curve in @xmath32 given by @xmath33 is then isomorphic to @xmath27 for every algebraic integer @xmath34 . consider the following curves on @xmath29 @xmath35 all four curves pass through the origin and do not intersect each other anywhere else .
it is , in fact , the consequence of the special choice of @xmath36 as a generator of the elliptic curve @xmath27 .
then @xmath37 is also a unit in @xmath38 $ ] and the curves @xmath39 , @xmath40 do not intersect outside of the origin . other cases are trivial .
let @xmath41 be the lattice in @xmath29 consisting of @xmath42 points @xmath43 for each point of @xmath41 there are four curves passing through it , parallel to the curves @xmath44 , @xmath45 , @xmath39 and @xmath40 . denote the union of @xmath46 curves parallel to @xmath47 as @xmath48 for @xmath49 .
altogether we have @xmath50 elliptic curves forming four parallel families . except for the points in @xmath41
there are no other intersection points . after blowing up @xmath42 points of @xmath41
we get a smooth 4-manifold @xmath51 with euler characteristic @xmath42 . there are @xmath42 exceptional curves @xmath52 ( @xmath53 ) resulting from blow - ups .
denote by @xmath54 the proper transforms of @xmath48 after the blow - up ; these curves are smooth and do not intersect each other .
since each of the homology classes @xmath55 $ ] for @xmath56 is divisible by @xmath46 , we can construct a complex algebraic surface @xmath23 which is a branched @xmath57-fold cover of @xmath58 branched over @xmath54 for each @xmath56 , where each @xmath54 will have the branching index @xmath59 .
denote this covering map as @xmath60 .
then a calculation of euler characteristic gives @xmath61 for the chern number @xmath8 we have @xmath62 for the actual computation , and for the further details we refer the reader to @xcite .
the map @xmath60 can be extended via the natural projections to the factors of the torus @xmath29 .
then we have the following diagram : @xmath63 the map @xmath64 respects the natural fibration of @xmath29 , singular fibers occur over the @xmath46 points in the set @xmath65 , and are formed by the curve @xmath66 and exceptional curves @xmath52 . the situation for the map @xmath67 is analogous . composed maps @xmath68 , @xmath69 give two different ( singular ) fibrations of space @xmath23 over a 2-dimensional torus .
in fact , these maps are holomorphic outside the blown - up points and corresponding exceptional curves , which leads us to the observation that a generic fiber of @xmath68 is a smooth holomorphic curve . for a regular point @xmath70 in @xmath44 , i.e. point not belonging to @xmath71 , the inverse image
@xmath72 is an elliptic curve parallel to @xmath45 .
this fiber is intersected @xmath73 times by the curves @xmath74 and @xmath75 .
note , that these points are branching points of the map @xmath76 restricted to the map of two dimensional fibers in @xmath23 and @xmath58 over the point @xmath70 . by the following computation of euler characteristic of a regular fiber @xmath77
we get @xmath78 hence the genus of a regular fiber is @xmath79 . for the singular value @xmath80 of the composed map @xmath68 and the respective singular fiber @xmath81 the situation is a bit more complicated .
we know that @xmath80 belongs to @xmath71 , and since the construction is completely symmetric with respect to the points in @xmath71 , it is enough to describe the singular fiber over one point
. we can choose the point @xmath82 . then the inverse image @xmath83 consists of @xmath46 exceptional spheres @xmath52 corresponding to the blown up points over @xmath80 , connected with and intersected by the proper transform @xmath84 of the torus @xmath45 .
moreover , each of the exceptional spheres is intersected at exactly 3 different points by @xmath85 , @xmath86 and @xmath75 , respectively .
this gives us the complete information about the map @xmath87 .
the singular fiber @xmath88 consists of @xmath46 tori copies of @xmath84 , connected with @xmath46 copies of branched covers @xmath89 of exceptional spheres @xmath52 ( @xmath90 ) .
the euler characteristic of @xmath89 is @xmath91 and the euler characteristic of the whole fiber @xmath88 is @xmath92 to summarize the above mentioned facts we state the lemma .
[ tor1 ] there exists a complex surface @xmath23 of general type with chern numbers @xmath93 , @xmath94 , holomorphic euler characteristic @xmath95 , signature @xmath96 , and is equipped with a map @xmath97 .
the genus of a regular fiber @xmath98 is @xmath99 .
in this section we describe the topology of @xmath23 via the study of its fundamental group .
since our ultimate goal is to construct simply connected 4-manifold obtained from @xmath23 by a certain gluing , it would be convenient to find an embedded surface with small genus which generates as much of the fundamental group of @xmath23 as possible .
we use the fact that the maps @xmath68 , @xmath69 map the space @xmath23 to tori @xmath44 and @xmath45 . denote by @xmath100 a regular fiber over some regular point @xmath101 , and
@xmath102 a regular fiber over some regular point @xmath103 . recall that the resulting surfaces @xmath104 , @xmath105 are smooth complex curves . in the previous section we computed the genera of these curves , they are the same , and equal @xmath106 . the number of intersection points between @xmath104 and @xmath105 can be deduced from the following observation .
surface @xmath105 projects down by the map @xmath76 to an embedded torus in @xmath58 .
the map @xmath76 restricted to @xmath105 is then an @xmath57-fold branched cover . when we project resulting torus further down by the map @xmath64 into @xmath44 we get , actually , one - to - one map .
therefore the restricted map @xmath107 is an @xmath57-fold branched cover as well . as a consequence
we get that the number of intersection points between @xmath104 and @xmath105 is exactly @xmath57 .
note that all intersections are transversal and positive .
the following lemma shows that the union @xmath108 contains all the topological information we need .
[ fund1 ] the natural inclusion of fundamental groups @xmath109 is surjective .
we will show that any loop @xmath110 can be deformed into a loop lying inside @xmath108 .
we can assume that the base point @xmath111 of the fundamental group @xmath112 is one of intersection points of @xmath104 and @xmath105 .
the map @xmath68 is singular over points in the set @xmath71 .
since the singular locus @xmath113 in @xmath23 has dimension 2 , by using standard transversality argument we can assume that the representative of @xmath114 avoids singularities .
consider the projection @xmath115 .
this gives a loop inside @xmath44 avoiding singular points in @xmath71 .
the inverse image of @xmath116 , the space @xmath117 is therefore a surface bundle over @xmath116 , and the fiber of this bundle over the point @xmath111 is exactly @xmath104 . using the fact that @xmath105 is a branched cover of @xmath44
, we can deduce that the intersection of the bundle and the surface @xmath105 is an @xmath57-fold cover of the loop @xmath116 , in particular @xmath105 intersects each fiber in @xmath57 points .
since the loop @xmath114 lies inside the bundle whose fiber is connected , we can homotopically deform @xmath114 into @xmath118 which will lie inside @xmath105 except for the part lying above the point @xmath111 and in the fiber @xmath104 .
this proves the lemma .
[ tor3 ] there exists an embedded symplectic surface @xmath119 in @xmath23 such that the inclusion map @xmath120 is surjective .
the surface @xmath119 has genus @xmath121 and self - intersection @xmath122 .
we obtain @xmath119 from @xmath108 by standard resolution of singular intersection points in @xmath123 .
it is easy to see that @xmath124 .
since @xmath104 , @xmath105 are regular fibers , each of them have zero self - intersection number .
the resolution of each positive intersection point increases the self - intersection by 2 , hence @xmath125 ^ 2 = ( [ f_1 ] + [ f_2])^2 = 2[f_1][f_2 ] = 2 n^3 $ ] .
finally , the resolution of a singular point is only a local operation , but it might , and in fact does , change the fundamental group of respective surfaces .
nevertheless , the map @xmath126 is surjective , its kernel being generated by loops in @xmath119 which can be contracted to singular points in @xmath108 .
this proves the lemma .
surfaces @xmath104 and @xmath105 from the last section were non - degenerate holomorphic curves . after the resolution of their intersection points we got a smooth embedded surface @xmath119 , which is embedded symplectically .
we will perform connected symplectic sum operation ( as defined by gompf in @xcite ) to the ( symplectic ) manifold @xmath23 along embedded symplectic surface @xmath119 to obtain simply connected symplectic 4-manifold @xmath18 .
the ingredient we still need to provide is the second summand a simply connected symplectic manifold @xmath127 , preferably with small chern numbers @xmath8 and @xmath7 , and with embedded symplectic surface of same genus and opposite self - intersection as @xmath119 .
we begin with a @xmath15 surface @xmath16 , equipped with the standard elliptic fibration @xmath128 . in order to obtain a non - degenerate holomorphic curve ( or symplectic surface ) of relatively large negative self - intersection , we need to select a number of points on a curve , and blow them up . resulting proper transform of the original curve
will have the required negative self - intersection , decreasing by 1 after each blow - up .
the adjunction formula for complex and symplectic manifolds implies that this is basically the only effective procedure how to obtain such surfaces .
picking and blowing - up @xmath129 points on a selected section of @xmath16 gives a smoothly embedded rational complex curve @xmath130 with self - intersection @xmath131 inside complex surface @xmath132 .
note that the elliptic fibration remained untouched outside blown - up points .
now we can perform fibered knot surgery ( see @xcite ) , in a small tubular neighborhood of a regular elliptic fiber @xmath27 in a neighborhood of one of the cusp fibers in @xmath16 .
let @xmath133 be a torus knot @xmath134 with genus @xmath135 , and @xmath136 be a result of fibered knot surgery on @xmath137 [ tor4 ] the 4-manifold @xmath127 is symplectic and simply connected , and it contains symplectically embedded surface @xmath138 of genus @xmath139 and self - intersection @xmath140 ^ 2 = -2n^3 = - [ f]^2 $ ] .
chern numbers of @xmath127 are @xmath141 , @xmath142 , holomorphic euler characteristic @xmath143 , and the signature @xmath144 .
moreover , @xmath145 .
the fact that @xmath145 follows from the triviality of @xmath146 , since this space fibers over @xmath147 with a ( singular ) simply connected fiber @xmath147 coming from one of cusps .
the rest of the statement follows directly from the construction of @xmath127 and straightforward computations .
* of theorem [ thm1 ] * we have done almost all the work in lemmas [ tor1 ] , [ fund1 ] , [ tor3 ] and [ tor4 ] .
it remains to show that the manifold @xmath18 is simply connected .
this can be obtained by combining facts that @xmath149 is surjective , that @xmath150 , and by using seifert - van kampen theorem for the union @xmath151 .
computations of characteristics are routine .
[ infa ] for each 4-manifold @xmath18 there exist infinitely many symplectic smooth manifolds of the same homeomorphism type which are not diffeomorphic to it . there exist infinitely many smooth manifolds of the same homeomorphism type not admitting any symplectic structure , as well . the @xmath15 surface @xmath16 contains the nucleus @xmath152 , and at least one symplectically @xmath17-embedded torus in the complement @xmath153 , because this complement is a _ milnor fiber _ @xmath154 ( see @xcite , @xcite ) .
in our construction we used only the inside of the nucleus we used a regular fiber in the cusp neighborhood for the fibered knot surgery @xmath136 , and a section for the symplectic connected sum @xmath148 .
it means that the manifold @xmath18 could be produced in such a way that the @xmath17-embedded torus in the complement @xmath155 stayed intact during the surgeries .
this gives us a possibility to perform another fibered knot surgery in the neighborhood of the remaining cusp . by reasoning similar to that in @xcite
we see that the manifolds after the knot surgery will have different @xmath14-invariants than those of @xmath18 , provided the alexander polynomial of the knot is non - trivial .
the infinite number of fibered knots with different alexander polynomials provide the infinite number of symplectic manifolds homeomorphic but not diffeomorphic to @xmath18 . it can be shown ( see @xcite , cor .
1.3 ) , that taking the fibered knot surgery along any knot , which is not fibered , and does not have a _ monic _
alexander polynomial will produce a manifold not admitting a symplectic structure .
again , there is an infinite number of such knots with different non - monic alexander polynomials .
similar results were obtained previously by j. park , @xcite , and stipsicz , @xcite . both constructed infinite families of simply connected 4-manifolds with the ratio @xmath156 very close to 9 .
however , their examples did not get above the line @xmath157 , with @xmath158 , and @xmath159 , respectively . in @xcite , stipsicz improved his result and constructed a family of simply connected symplectic 4-manifolds with @xmath160 asymptotically approaching the line @xmath161 . in comparison with our result
, this family converges more rapidly , meaning that for comparable values of @xmath10 manifolds constructed by stipsicz have higher value of @xmath8 , and are closer to the b
m y line .
reason for this is , basically , the starting manifold .
our construction starts with @xmath23 below the line , while the other construction uses complex surfaces lying on the b m y line . in @xcite ,
j. park used stipsicz s method and populated a large part of the region corresponding to the positive signature , however the euler characteristics of these 4-manifolds have to be quite large . some desirable results going further in this direction
would be the construction of infinite family of 4-manifolds in constant distance from the b
m y line ( compare with the results of chen , @xcite , for complex surfaces of general type ) , or the construction of an example above it , which would contradict the conjecture by fintushel and stern .
at the present time , the available tools and techniques are insufficient for such constructions . | we construct a new family @xmath0 of simply connected symplectic 4-manifolds with the property @xmath1 ( as @xmath2 ) . |
we study here different compositions of point processes , say @xmath0 @xmath1 where @xmath2 and @xmath3 are mutually independent and can possibly be birth , death or homogeneous poisson processes .
these compositions arise in the analysis of different population models , when , in some specific experimental circumstances , the time @xmath4 must be replaced by a stochastic process @xmath5 @xmath6 in our analysis we concentrate our attention on the case where the time increments are unit - valued and thus are properly represented by point processes , such as linear birth and poisson processes . in these cases the composed process can be regarded as a randomly sampled population model .
the simplest case of the composition of two independent homogeneous poisson processes ( with different rates ) has been already studied in @xcite .
more recently , the iterated poisson process has been considered in @xcite .
our investigation concentrates here on the distribution of the first - passage times , i.e.@xmath7which , in this context , substantially differ from the upcrossing and downcrossing times , i.e.@xmath8indeed , the composed processes @xmath9 jump over ( resp . under ) any level @xmath10 with positive probability .
the hitting probabilities @xmath11 display different behaviors .
for example , for the iterated birth process @xmath12 ( with @xmath13 independent linear birth processes ) , the probabilities @xmath11 attain their maximal value at @xmath14 . in the linear death process at poisson distributed time the hitting probabilities display an oscillating behavior , which can be perceived when the initial size of the population is sufficiently large .
furthermore , we observe that in all the cases considered here the hitting probabilities @xmath11 depend only on the parameter of the outer process
. moreover , the processes @xmath9 display sample paths with upward or downward jumps of arbitrary size . other models with the same feature , such as the space - fractional poisson processes or , in general , time - changed poisson processes with berntein subordinators , have been analyzed in @xcite .
multiple jumps are also displayed by the generalized fractional birth processes studied in @xcite .
as in this last work , this property is reflected in the form of the equation governing the state probabilities @xmath15 @xmath16 where the time derivative @xmath17 is shown to depend , also here , on all @xmath18 , for @xmath19 in particular , our analysis concerns the following specific cases : * @xmath20 @xmath21(section 3 ) * @xmath22 @xmath21(section 4 ) * @xmath23 @xmath21(section 5 ) * @xmath24 @xmath21(section 6 ) , where @xmath25 and @xmath26 are independent linear birth ( yule - furry ) processes , @xmath27 is the homogeneous poisson process , @xmath28 is the linear death process and @xmath29 is the sublinear death process .
the latter is characterized by the fact that the probability of a further death in @xmath30 depends on the number of deaths recorded up to time @xmath4 ( unlike the linear case where this probability depends on the number of surviving individuals ) .
the sublinear birth process was introduced in don , while the sublinear death process has been investigated in @xcite , in a fractional context .
the processes considered here can be useful to model the population evolution sampled at poisson times
. the iterated birth process can be more appropriate for the cases where the time separating occurrences is rapidly decreasing .
this structure can be applied , for example , in some experimental studies of diseases or epidemic diffusions .
time - changed birth processes of different forms have been studied in din and @xcite , with various applications to finance .
time - changed poisson and birth processes arise also in the study of fractional point processes ( see , for example , @xcite , @xcite , @xcite and cah2 ) .
* list of the main symbols * @xmath31 where @xmath32 is the number of the initial components of the population and @xmath33
let us first consider the composition of a linear birth ( yule - furry ) process @xmath34 , with one initial progenitor and birth rate @xmath35 with an independent process of the same kind , with parameter @xmath36 i.e.@xmath37as a consequence of the definition ( [ def ] ) , the initial number of progenitors of the iterated linear birth process is random , since @xmath38 a.s .
the probability mass function of this process can be written , for any @xmath39 as @xmath40we observe that@xmath41we prove now that the probabilities @xmath42 @xmath43 @xmath44 decrease as @xmath10 increases , since we have that@xmath45 \\ & = & e^{-\lambda t-\alpha } \sum_{l=1}^{k-1}\binom{k-2}{l-1}\frac{(-e^{-\alpha } ) ^{l}}{1-e^{-\alpha ( l+1)}(1-e^{-\lambda t } ) } \\ & = & -e^{-\lambda t-2\alpha } \sum_{m=0}^{\infty } ( 1-e^{-\lambda t})^{m}e^{-2\alpha m}\sum_{l=0}^{k-2}\binom{k-2}{l}(-1)^{l}e^{-\alpha l(m+1 ) } \\ & = & -e^{-\lambda t-2\alpha } \sum_{m=0}^{\infty } ( 1-e^{-\lambda t})^{m}e^{-2\alpha m}(1-e^{-\alpha ( m+1)})^{k-2}<0,\end{aligned}\]]for any @xmath46 the sample paths of the iterated birth process display upward jumps of size larger or equal to one . for this reason
the analysis of the first - passage time through an arbitrary level @xmath47 , i.e.@xmath48is of a certain importance .
moreover , we shall prove that @xmath49 is strictly less than one , as it happens for the iterated poisson process ( see @xcite ) and thus any level @xmath10 can be avoided with positive probability , because of `` multiple jumps '' .
we present the explicit distribution of @xmath50 in the next theorem .
the distribution of @xmath50 reads@xmath51 , \label{bb}\]]for any @xmath52 @xmath53 by a conditioning argument we get @xmath54which coincides with ( [ bb ] ) .
we note that , in the special case @xmath55 , we get@xmath56 ^{2}}% dt=\lambda e^{\lambda t}(1-e^{-\alpha } ) [ q_{1}^{\mathcal{z}}(t)]^{2}dt\ ] ] by integrating ( [ bb ] ) , we get that@xmath57 , \label{cc}\]]which can be alternatively written as@xmath58 in order to analyze some special cases , we supply also the following finite sum form of the hitting probabilities given in ( [ cc]):@xmath59 \notag \\ & = & e^{-\alpha k}\sum_{r=0}^{k-2}\binom{k-1}{r}(-1)^{r}\frac{e^{-\alpha ( r+1)}% } { 1-e^{-\alpha ( r+1)}}\left ( e^{\alpha ( k - r-1)}-1\right ) .
\label{ic}\end{aligned}\]]it is now easy to show that the following relationships hold , for @xmath60:@xmath61since we have that@xmath62 \\
\pr \{t_{4}^{\mathcal{z}}<\infty \ } & = & e^{-2\alpha } \left [ 1-e^{-\alpha } % \frac{(1-e^{-\alpha } ) ( 1+e^{-3\alpha } ) } { 1+e^{-\alpha } + e^{-2\alpha } } \right ] .\end{aligned}\]]the following figures ( produced by the software r ) describe the behavior of the probabilities ( [ ic ] ) , for different values of @xmath63 we remark that the scales are not the same in different figures . ]
the figures above show that the probabilities @xmath64 , for @xmath65 , are decreasing functions which , for @xmath66 vanish more rapidly than for @xmath67 the rate of decreasing is slow , for all @xmath68 since , from ( [ pre ] ) , we get that@xmath69}{1-e^{-\alpha } + e^{-\alpha ( j+1)}}-\frac{% e^{-3\alpha } ( 1-e^{-\alpha } ) } { 1-e^{-\alpha } + e^{-2\alpha } } \\ & = & ( 1-e^{-\alpha } ) \left\ { \sum_{j=2}^{\infty } \frac{(1-e^{-\alpha j})[1+e^{-\alpha ( j+1)}]}{1-e^{-\alpha } + e^{-\alpha ( j+1)}}-\frac{% e^{-3\alpha } } { 1-e^{-\alpha } + e^{-2\alpha } } \right\ } = \infty , \end{aligned}\]]since@xmath70}{% 1-e^{-\alpha } + e^{-\alpha ( j+1)}}=\frac{1}{1-e^{-\alpha } } .\ ] ]
we now pass to the analysis of the process defined as @xmath71obtained as a composition of a linear birth process @xmath72 with an independent homogeneous poisson process @xmath73 with rate @xmath74 the process @xmath75 can be regarded as a linear birth process randomly sampled at poisson times .
we first note that , for @xmath76@xmath77\right\ } .
\notag\end{aligned}\]]in particular , for @xmath78 , we get@xmath79and thus the waiting time for the appearance of the first offspring is exponentially distributed with parameter @xmath80 it is easy to show that @xmath81the variance of @xmath75 can be obtained as follows @xmath82 + \mathbb{e}var\left [ \left . b_{\alpha } ( n_{\lambda } ( t))\right\vert n_{\lambda } ( t)\right ] \label{he } \\ & = & var\left [ n_{0}e^{\alpha n_{\lambda } ( t)}\right ] + \mathbb{e}\left [ n_{0}e^{2\alpha n_{\lambda } ( t)}-n_{0}e^{\alpha n_{\lambda } ( t)}\right ] \notag \\ & = & n_{0}(n_{0}+1)e^{\lambda t(e^{2\alpha } -1)}-n_{0}^{2}e^{\lambda t(e^{\alpha } -1)}-n_{0}e^{2\lambda t(e^{\alpha } -1)}. \notag\end{aligned}\]]for the sake of simplicity , from now on , we assume @xmath83 and denote for brevity @xmath84 as @xmath85 .
the factorial moments of @xmath75 can be evaluated as follows : @xmath86 ... [ % \mathcal{x}(t)-r+1]\right\vert \mathcal{x}(t)=1\right\ } \label{ci } \\ & = & \sum_{k = r}^{\infty } k(k-1) ...
(k - r+1)q_{k}^{\mathcal{x}}(t ) \notag \\ & = & [ \text{by ( \ref{si } ) ] } \notag \\ & = & r!e^{-\lambda t}\sum_{k = r}^{\infty } \binom{k}{r}\sum_{j=0}^{\infty } e^{-\alpha j}\left ( 1-e^{-\alpha j}\right ) ^{k-1}\frac{(\lambda t)^{j}}{j ! } \notag \\ & = & r!e^{-\lambda t}\sum_{j=0}^{\infty } e^{-\alpha j}\frac{(\lambda t)^{j}}{j!% } \sum_{l=0}^{\infty } \binom{r+l}{l}\left ( 1-e^{-\alpha j}\right ) ^{r+l-1 } \notag \\ & = & r!e^{-\lambda t}\sum_{j=0}^{\infty } e^{-\alpha j}\frac{(\lambda t)^{j}}{j!% } \frac{\left ( 1-e^{-\alpha j}\right ) ^{r-1}}{[1-(1-e^{-\alpha j})]^{r+1 } } \notag\end{aligned}\ ] ] @xmath87by using result ( [ ci ] ) we can check formulas ( [ bu ] ) and ( [ he ] ) , by means of independent calculations .
our main interest here is to study the distribution of the first - passage time for an arbitrary level @xmath10 , i.e.@xmath88for @xmath46 in the next theorem we derive the following explicit distribution of @xmath89 for @xmath21 and @xmath47 we have that @xmath90\ } \label{two}\ ] ] we start by considering that@xmath91which coincides with ( [ two ] ) . in the particular case
@xmath92 formula ( [ two ] ) reduces to@xmath93where @xmath94 is an exponential r.v . with parameter @xmath95 and
this clearly shows that for the first - passage time through the level @xmath96 the following result holds@xmath97 for any @xmath47 , by integrating formula ( [ two ] ) , we get@xmath98\sum_{m=0}^{\infty } e^{-\alpha ( l+1)m } \notag \\ & = & e^{-\alpha k}\sum_{m=0}^{\infty } e^{-\alpha m}\sum_{l=0}^{k-2}(-1)^{l}% \binom{k-1}{l}[e^{\alpha ( k-1)-\alpha l(m+1)}-e^{-\alpha lm } ] \notag \\ & = & \sum_{m=0}^{\infty } e^{-\alpha ( m+1)}\left [ ( 1-e^{-\alpha ( m+1)})^{k-1}-(-1)^{k-1}e^{-\alpha ( m+1)(k-1)}\right ] + \notag \\ & & -e^{-\alpha k}\sum_{m=0}^{\infty } e^{-\alpha m}\left [ ( 1-e^{-\alpha m})^{k-1}-(-1)^{k-1}e^{-\alpha m(k-1)}\right ] \notag \\ & = & \sum_{m=0}^{\infty } \left [ e^{-\alpha ( m+1)}(1-e^{-\alpha ( m+1)})^{k-1}-e^{-\alpha k}e^{-\alpha m}(1-e^{-\alpha m})^{k-1}\right ] \notag \\ & = & ( 1-e^{-\alpha k})\sum_{m=1}^{\infty } e^{-\alpha m}(1-e^{-\alpha m})^{k-1 } \notag \\ & = & ( 1-e^{-\alpha k})\sum_{m=1}^{\infty } \pr \{b_{\alpha } ( m)=k\}. \notag\end{aligned}\]]as in the case of the iterated birth process , the hitting probabilities ( for ) are not affected by the parameter @xmath99 of the inner process ( @xmath100 , in this case)@xmath101 moreover , we remark that the relationship given in the last line is equal to the analogous one presented in ( [ pre ] ) ( except for the additional term in ( [ pre ] ) , which is due to the different starting point of the birth process with respect to the poisson one):@xmath102for any @xmath46 thus , for the iterated linear birth process , the probability of reaching any level @xmath10 in a finite time is strictly smaller than the corresponding probability for @xmath103 .
this difference decreases monotonically with @xmath104 for @xmath55 the first line of formula ( [ for ] ) reduces to ( [ rel])@xmath101 in the special case @xmath14 we have that@xmath105 \\ & = & e^{-\alpha } \frac{1+e^{-2\alpha } } { 1+e^{-\alpha } } \\ & = & \pr \{\left .
t_{2}^{\mathcal{x}}<\infty \right\vert \mathcal{x}(0)=1\}% \frac{1+e^{-2\alpha } } { 1+e^{-\alpha } } < \pr \{\left .
t_{2}^{\mathcal{x}% } < \infty \right\vert \mathcal{x}(0)=1\}<1.\end{aligned}\]]we show now that , for any @xmath47 the distribution of the first - passage time for the level @xmath10 is monotonically decreasing : we present the following heuristic argument @xmath106 \\ & = & \frac{(1-e^{-\alpha } ) } { \alpha } \left\ { \left ( 1+ ... +e^{-\alpha ( k-2)}\right ) \left [ \frac{1}{k}-\frac{1}{k-1}\right ] + \frac{e^{-\alpha ( k-1)}}{k}\right\ } \\ & \leq & \frac{(1-e^{-\alpha } ) } { \alpha } \left [ -\frac{e^{-\alpha ( k-2)}}{% k(k-1)}(k-1)+\frac{e^{-\alpha ( k-1)}}{k}\right ] < 0,\end{aligned}\]]for
@xmath107 the plots of fig.2 confirm the decreasing structure of the probabilities @xmath108 ] from formula ( [ for ] ) we derive the following equality@xmath109it is clear from ( [ io ] ) that , for large values of @xmath10 , we get@xmath110and thus the hitting probabilities slowly change with @xmath10 as fig . 2 confirms . by applying formula ( [ io ] )
we can prove also the following relationship@xmath111 + \sum_{m=1}^{\infty } e^{-3\alpha m}(1-e^{-\alpha m})^{k-3}. \notag\end{aligned}\]]formula ( [ in ] ) shows that , for large values of @xmath10 , @xmath112
we here consider linear and sublinear death processes at poisson times .
the linear death process , with initial population of @xmath113 individuals and death rate @xmath114 , has a binomial distribution , i.e.@xmath115the probabilities ( [ i ] ) satisfy the difference - differential equations@xmath116with @xmath117 moreover , we denote the sublinear death process , with death rate @xmath114 , as @xmath118 @xmath21 and its distribution is given by@xmath119the probabilities @xmath120 satisfy the following difference - differential equations @xmath121 with @xmath122 in order to catch the probabilistic mechanism underlying ( [ mo ] ) we should consider that @xmath123we note that in the linear death process the probability of cancellation of an individual in @xmath30 is proportional to the number of existing components at time @xmath124 on the other hand , in the sublinear case , this probability is proportional to the number of deaths recorded up to time @xmath124 this property makes the sublinear process more suitable for describing epidemics and the diffusion of rumors . the expected value of @xmath118 @xmath21 can be evaluated as follows@xmath125 -\frac{1}{\mu } % ( 1-e^{-\mu t})\frac{d}{dt}\left [ \sum_{k=0}^{n_{0}}(1-e^{-\mu t})^{k}\right ] \notag \\ & = & n_{0}\left [ 1-\left ( 1-e^{-\mu t}\right ) ^{n_{0}+1}\right ] -(1-e^{-\mu t})% \left [ e^{\mu t}-e^{\mu t}\left ( 1-e^{-\mu t}\right ) ^{n_{0}+1}-(n_{0}+1)\left ( 1-e^{-\mu t}\right ) ^{n_{0}}\right ] \notag \\ & = & n_{0}+\left ( 1-e^{-\mu t}\right ) ^{n_{0}+1}+e^{\mu t}\left ( 1-e^{-\mu t}\right ) ^{n_{0}+2}+1-e^{\mu t } \notag \\ & = & n_{0}+1-e^{\mu
t}[1-(1-e^{-\mu t})^{n_{0}+1}],\qquad n_{0}\geq 1 .
\notag\end{aligned}\]]it is easy to check that @xmath126 as expected . for the linear death process
, the hitting time of the @xmath127-th event , i.e.@xmath128has the following distribution@xmath129thus we get@xmath130for any @xmath104 on the other hand , in the sublinear case , the distribution of @xmath131reads@xmath132moreover we get@xmath133for any @xmath104 let@xmath134where @xmath135 is a linear death process with parameter @xmath136 and @xmath100 is an independent homogeneous poisson process with parameter @xmath137 then , for @xmath138 the probability mass function reads @xmath139since@xmath140the waiting time of the first death of @xmath141 is exponentially distributed with parameter @xmath142 and has the same form of the corresponding probability of the subordinated linear birth process , which was given in formula ( [ st ] ) .
the probability generating function of @xmath141 can be written as follows:@xmath143^{n_{0 } } \label{due } \\ & = & \sum_{m=0}^{n_{0}}\binom{n_{0}}{m}(-1)^{m}(1-u)^{m}\exp \{-\lambda t(1-e^{-\mu m}\}. \notag\end{aligned}\]]the mean value and variance are respectively given by@xmath144and @xmath145 + \mathbb{e}var\left [ \left . d_{\mu } ( n_{\lambda } ( t))\right\vert n_{\lambda } ( t)\right ] \\ & = & var\left [ n_{0}e^{-\mu n_{\lambda } ( t)}\right ] + \mathbb{e}\left [ n_{0}e^{-\mu n_{\lambda } ( t)}(1-e^{-\mu n_{\lambda } ( t)})\right ] \notag \\ & = & n_{0}(n_{0}-1)e^{\lambda t(e^{-2\mu } -1)}-n_{0}^{2}e^{2\lambda t(e^{-\mu } -1)}+n_{0}e^{\lambda t(e^{-\mu } -1)}. \notag\end{aligned}\ ] ] we are interested now in the differential equation satisfied by the distribution of the process @xmath146 as a preliminary result we prove the following lemma .
the probability generating function given in ( [ due ] ) satisfies the following initial - value problem:@xmath147 g^{\mathcal{y}}(u , t ) \\
g^{\mathcal{y}}(u,0)=u^{n_{0}}% \end{array}% \right . ,
\quad t , u\geq 0 , \label{tre}\]]where @xmath148 is the shift operator . by considering the equation governing the state probabilities of the poisson process
, we can write@xmath149where @xmath150 is the probability generating function of @xmath28 . since , for any @xmath151@xmath152 g^{d}(u , t ) \\
& = & -\mu ^{2}(1-u)\frac{\partial } { \partial u}g^{d}(u , t)+\mu ^{2}(1-u)^{2}% \frac{\partial ^{2}}{\partial u^{2}}g^{d}(u , t)\end{aligned}\]]@xmath153and , analogously@xmath154we can write@xmath155where @xmath156 are the stirling numbers of the second kind ( see @xcite ) . thus we get ( by considering that @xmath157)@xmath158 g^{\mathcal{y}% } ( u , t),\end{aligned}\]]where we have applied the dobinski formula for the bell polynomial @xmath159 , i.e.@xmath160equation ( [ tre ] ) follows by applying the well - known exponential generating function of the bell s polynomial ( see e.g. @xcite ) , i.e.@xmath161 the probability mass function of @xmath162 satisfies the following equation@xmath163with the initial condition @xmath164 in order to derive the differential equation satisfied by ( [ uno ] ) we rewrite the first line in ( [ tre ] ) as follows:@xmath165 \\ & = & \lambda \sum_{k=0}^{n_{0}}q_{k}^{\mathcal{y}}(t)\sum_{j=0}^{k}(1-e^{-\mu } ) ^{j}\binom{k}{j}\sum_{l^{\prime } = k - j}^{k}\binom{j}{k - l^{\prime } } % ( -1)^{j - k+l^{\prime } } u^{l^{\prime } } -\lambda g^{\mathcal{y}}(u , t ) \\ & = & \lambda \sum_{k=0}^{n_{0}}q_{k}^{\mathcal{y}}(t)\sum_{l=0}^{k}u^{l}\binom{% k}{l}\sum_{j = k - l}^{k}(1-e^{-\mu } ) ^{j}\binom{l}{k - j}(-1)^{j - k+l}-\lambda g^{% \mathcal{y}}(u , t ) \\ & = & \lambda \sum_{l=0}^{n_{0}}u^{l}\sum_{k = l}^{n_{0}}\binom{k}{l}q_{k}^{% \mathcal{y}}(t)\sum_{j = k - l}^{k}(1-e^{-\mu } ) ^{j}\binom{l}{l - k+j}% ( -1)^{j - k+l}-\lambda g^{\mathcal{y}}(u , t ) \\ & = & [ j = k - l+j^{\prime } ] \end{aligned}\ ] ] @xmath166 \\ & = & \lambda \sum_{l=0}^{n_{0}}u^{l}\sum_{r=0}^{n_{0}-l}\binom{r+l}{l}q_{r+l}^{% \mathcal{y}}(t)(1-e^{-\mu } ) ^{r}e^{-\mu l}-\lambda g^{\mathcal{y}}(u , t).\end{aligned}\ ] ] by comparing equation ( [ bo ] ) with ( [ acc ] ) , we can see that , in the subordinated case the time - derivative of @xmath167 depends on the probabilities @xmath168 , for any @xmath169 , while , in the standard case , the derivative of @xmath170 depends only on @xmath171 this is due to the fact that the process @xmath172 performs downward jumps of arbitrary size , whose law can be derived from ( [ uno ] ) and can be written as follows:@xmath173the first line in ( [ iss ] ) means that , during the infinitesimal interval @xmath174 either the poisson process ( the `` time '' ) does not change or it changes and all the individuals survive during a time span of unit length . in our view the main result of this section is the density of the first passage - time through the level * * * * @xmath10 , i.e.@xmath175which is presented in the following theorem .
the probability density of @xmath176 reads , for any @xmath177@xmath178 ^{n_{0}-k}-\left [ 1-e^{-\mu j}\right ] ^{n_{0}-k}\right\ } .
\notag\end{aligned}\ ] ] we start by considering that , for @xmath179@xmath180 @xmath181 ^{n_{0}-k}-\left [ 1-e^{-\mu j}\right ] ^{n_{0}-k}\right\ } + \notag \\ & & + \lambda dte^{-\lambda t}\binom{n_{0}}{k}e^{-\mu k}(1-e^{-\mu } ) ^{n_{0}-k } , \notag\end{aligned}\ ] ] which coincides with ( [ fr ] ) .
by integrating ( [ fr ] ) we get that , for @xmath179@xmath182 ^{n_{0}-k}-\left [ 1-e^{-\mu j}\right ] ^{n_{0}-k}\right\ } .
\label{ic3}\]]we note that the probability ( [ ic3 ] ) has the same structure of formula ( 36 ) in @xcite , which is related to the iterated poisson process , despite the fact that the outer process @xmath28 has decreasing paths .
an alternative form of ( [ fr ] ) , as a finite sum , can be obtained as follows @xmath183,\]]which , by integration , gives@xmath184for any @xmath185we note that , for @xmath186 the extinction probability is given by@xmath187for @xmath188 , we have instead that@xmath189 ] unexpectedly enough fig.3 ( which is obtained here for @xmath190 ) shows that the probabilities @xmath191 do not display a monotonic behavior for sufficiently large values of @xmath113 .
let @xmath192where @xmath193 is a sublinear death process with parameter @xmath194 and @xmath195 is an independent poisson process with parameter @xmath137 then the probability mass function reads @xmath196for @xmath197 , while , for @xmath198 , is given by@xmath199clearly @xmath200 is a decreasing function of the initial number of individuals , as can be directly checked , by considering ( [ mo ] ) .
the expected value of @xmath201 @xmath44 can be evaluated , by considering ( [ e ] ) , as follows@xmath202\right\ } \frac{% ( \lambda t)^{j}}{j ! } \\ & = & n_{0}+1-e^{-\lambda t(1-e^{\mu } ) } + \sum_{j=0}^{\infty } ( 1-e^{-\mu j})^{n_{0}+1}\frac{(\lambda te^{\mu } ) ^{j}}{j ! } \\ & \leq & n_{0}-\left ( 1-e^{-\lambda t(1-e^{-\mu } ) } \right ) .\end{aligned}\]]in the case of the subordinated sublinear death process , we have no markovianity and thus we can not evaluate explicitly the distribution of the hitting times @xmath203 @xmath204 for this reason we consider here the instant of the first downcrossing of the level @xmath10 , i.e.@xmath205for @xmath206 , we can write @xmath207 \\ & = & \sum_{l = k}^{n_{0}}\sum_{r=0}^{n_{0}-l}\binom{n_{0}-l}{r}(-1)^{r}\exp \left\ { -\lambda t(1-e^{-\mu ( 1+r)})\right\ } \\ & = & \sum_{r=0}^{n_{0}-k}(-1)^{r}\exp \left\ { -\lambda t(1-e^{-\mu ( 1+r)})\right\ } \sum_{l = k}^{n_{0}-r}\binom{n_{0}-l}{r}.\end{aligned}\]]the inner sum can be treated as follows@xmath208 \\ & = & \binom{n_{0}-k+1}{r+1},\end{aligned}\]]so that we get@xmath209which vanishes , for @xmath210 , for any @xmath206 . for @xmath186 by applying ( [ no2 ] ) we have , instead , the following extinction probability:@xmath211which , for @xmath210 , is equal to @xmath212
we thank dr .
bruno toaldo for providing the figures presented in this paper . | in this paper we study the iterated birth process of which we examine the first - passage time distributions and the hitting probabilities . furthermore , linear birth processes , linear and sublinear death processes at poisson times are investigated .
in particular , we study the hitting times in all cases and examine their long - range behavior .
the time - changed population models considered here display upward ( birth process ) and downward jumps ( death processes ) of arbitrary size and , for this reason , can be adopted as adequate models in ecology , epidemics and finance situations , under stress conditions . * keywords and phrases * : yule - furry process , linear and sublinear death processes , hitting times , extinction probabilities , first - passage times , stirling numbers , bell polynomials . _
ams mathematical subject classification ( 2010 ) : 60g55 , 60j80 . _ |
a density of a dilute gas is governed by the boltzmann equation @xmath0where @xmath1 is a distribution function for the gas particles at a time @xmath2 a position @xmath3and a velocity @xmath4 throughout this paper , the collision operator takes the form@xmath5where @xmath6\omega , $ ] @xmath7\omega $ ] and @xmath8 with @xmath9 ( hard potential ) and @xmath10 for all @xmath11 .
if the gas is contained in a bounded region or flows past a solid bodies , the boltzmann equation must be accompanied by boundary conditions , which describe the interaction of the gas molecules with the solid walls .
let the domain @xmath12 be a smooth bounded domain .
we consider three basic types of boundary conditions(@xcite,@xcite ) for @xmath1 at @xmath13 with @xmath14 , where @xmath15 is an outward unit normal vector at @xmath16 : + 1 . * in - flow injection boundary condition * : incoming particles are prescribed ; @xmath17 2 .
* diffuse reflection boundary condition * : incoming particles are a probability average of the outgoing particles ; @xmath18 with a normalized maxwellian @xmath19 , a normalized constant @xmath20 such that @xmath21 3 . *
bounce - back reflection boundary condition * : incoming particles bounce back at the reverse the velocity ; @xmath22 the purpose of this paper is to investigate possible formation and propagation of discontinuity for the nonlinear boltzmann equation under those boundary conditions . in order to state our results ,
we need following definitions . throughout this paper
, we assume the domain @xmath23 is open and bounded and connected . for simplicity ,
we assume that the boundary @xmath24 is smooth , i.e. for each point @xmath25 , there exists @xmath26 and a smooth function @xmath27 such that - upon relabeling and reorienting the coordinates axes if necessary - we have @xmath28 the outward normal vector at @xmath29 is given by @xmath30 given @xmath31 , let @xmath32=[x(s;t , x , v),v(s;t , x , v)]=[x-(t - s)v , v]$ ] be a trajectory ( or a characteristics ) for the boltzmann equation ( boltzmann ) : @xmath33with the initial condition : @xmath34=[x , v].$ ] for @xmath35 , we define the * backward exit time * , @xmath36 to be the last moment at which the back - time straight line @xmath37 $ ] remains in the interior of @xmath38 @xmath39 we also define the * backward exit position * in @xmath24 @xmath40 and we always have @xmath41 we denote the phase boundary in the phase space @xmath42 as @xmath43 and split it into outgoing boundary @xmath44 the incoming boundary @xmath45 and the grazing boundary @xmath46 @xmath47 @xmath48 we need to study the grazing boundary @xmath49 more carefully .
[ grazingboundary ] we define the * concave(singular ) grazing boundary * which is a subset of the grazing boundary @xmath49 : @xmath50 and the * outward inflection grazing boundary * in the grazing boundary @xmath49 : @xmath51 and the * inward inflection grazing boundary * in the grazing boundary @xmath49 : @xmath52 and the * convex grazing boundary * in the grazing boundary @xmath49 : @xmath53 it turns out that the concave ( singular ) grazing boundary @xmath54 is the only part at which discontinuity can be created and propagates into the interior of the phase space @xmath55 .
define the * discontinuity set * in @xmath56 as @xmath57\big\ } \cup\big\ { \ ( t , x , v ) \in ( 0,\infty)\times\{\omega\times\mathbb{r}^3 \cup \gamma_+\ } : \
t\geq t_{\mathbf{b}}(x , v ) \text { and } \
( x_{\mathbf{b}}(x , v),v)\in\gamma_0^{\mathbf{s } } \ \big\ } , \label{d}\end{aligned}\ ] ] and the * continuity set * in @xmath58 as @xmath59 \big\}\nonumber\\ & & \ \ \ \ \ \ \cup \ \big\ { \ ( t , x , v ) \in ( 0,\infty)\times\{\omega\times\mathbb{r}^3 \cup \gamma_+\ } : \
t_{\mathbf{b}}(x , v ) \ \ \text{or } \ \ ( x_{\mathbf{b}}(x , v),v ) \in \gamma_- \cup \gamma_0^{i- } \ \big\}.\label{c}\end{aligned}\ ] ] for bounce - back reflection boundary condition case ( [ fbounceback ] ) we need slightly different definitions : the * bounce - back discontinuity set * and the * bounce - back continuity set * are @xmath60 \big\}\\ & & \ \ \ \ \ \cup
\ \big\ { ( t , x , v ) \in [ 0,\infty)\times\{\omega\times\mathbb{r}^3 \cup\gamma_+ \ } : \
t < t_{\mathbf{b}}(x , v ) \ \text{or } \ \ \left [ \ ( x_{\mathbf{b}}(x , v),v)\in\gamma_- \cup \gamma_0^{i- } \ \ \text{and } \ \
t<2t_{\mathbf{b}}(x , v)+t_{\mathbf{b}}(x ,- v ) \
\right ] \ \ \ \ \ \ \ \ \\ & & \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \
\text{or } \
\ \left [ \ ( x_{\mathbf{b}}(x ,- v),-v)\in \gamma_- \cup \gamma_0^{i- } \ \ \text{and } \ \ ( x_{\mathbf{b}}(x , v),v)\in\gamma_- \cup \gamma_0^{i- } \ \right ] \big\ } , \label{dbb}\end{aligned}\ ] ] respectively . the discontinuity set @xmath61 consists of two sets : the first set of ( [ d ] ) is the grazing boundary part @xmath49 of @xmath61 .
this set mainly consists of the phase boundary where the backward exit time @xmath62 is not continuous ( lemma [ tbcon ] ) .
the second set of ( [ d ] ) is mainly the interior phase space part of @xmath61 , i.e. @xmath63 , which is a subset of a union of all forward trajectories in the phase space emanating from @xmath64 .
notice that @xmath61 does not include the forward trajectories emanating from @xmath65 because those forward trajectories are not in the interior phase space @xmath66 .
we also exclude the case @xmath67 from @xmath61 . in fact , considering the pure transport equation , @xmath67 implies the transport solution at @xmath31 equals the initial data at @xmath68 and if the initial data is continuous , we expect the transport solution is continuous around @xmath31
. notice that we exclude the initial plane @xmath69 from @xmath61 because we assume that the boltzmann solution is continuous at @xmath70 .
the continuity set @xmath71 consists of points either emanating from the initial plane or from @xmath72 , but not @xmath64 .
+ furthermore we define a set including the grazing boundary @xmath49 and all forward trajectories emanating from the whole grazing boundary @xmath49 .
[ grazingset ] the * grazing set * is defined as @xmath73 and the * grazing section * is @xmath74 obviously the grazing set @xmath75 includes the discontinuity set @xmath61 . in order to study the continuity property of the boltzmann solution we define : for a function @xmath76 defined on @xmath77 we define the * discontinuity jump in space and velocity * @xmath78_{x , v } \ = \ \lim_{\delta\downarrow 0}\sup_{\begin{scriptsize}(x^{\prime},v^{\prime } ) , ( x^{\prime\prime},v^{\prime\prime } ) \in \{\bar{\omega}\times\mathbb{r}^3 \backslash \mathfrak{g}\ } \cap \{b((x , v);\delta ) \backslash ( x , v)\}\end{scriptsize } } |\phi(t , x^{\prime},v^{\prime})-\phi(t , x^{\prime\prime},v^{\prime\prime})|,\ ] ] and the * discontinuity jump in time and space and velocity * @xmath79_{t , x , v } \ = \ \lim_{\delta\downarrow 0}\sup_{\begin{scriptsize}\begin{array}{ccc } t^{\prime},t^{\prime\prime}\in b(t;\delta)\\(x^{\prime},v^{\prime } ) , ( x^{\prime\prime},v^{\prime\prime})\in \{\bar{\omega}\times\mathbb{r}^3 \backslash \mathfrak{g}\ } \cap \{b((x , v);\delta ) \backslash ( x , v)\ } \end{array}\end{scriptsize } } |\phi(t^{\prime},x^{\prime},v^{\prime})-\phi(t^{\prime\prime},x^{\prime\prime},v^{\prime\prime})|,\ ] ] where @xmath75 is defined in definition [ grazingset ] .
we say a function @xmath80 is discontinuous in space and velocity ( in time and space and velocity ) at @xmath31 if and only if @xmath81_{x , v } \neq 0 \ ( [ \phi]_{t , x , v } \neq 0)$ ] and continuous in space and velocity ( in time and space and velocity ) at @xmath31 if and only if @xmath81_{x , v } = 0 \ ( [ \phi]_{t , x , v } = 0)$ ] .
notice that the function @xmath80 is only defined away from the grazing set @xmath75 .
once the discontinuity jump of given function @xmath80 is zero at @xmath31 then the function @xmath80 can be extended to @xmath82 near @xmath31 . because of those definitions we can consider a function which has a removable discontinuity as a continuous function . and a non - zero discontinuity jump @xmath83_{t , x , v}\neq 0 $ ] means @xmath80 has a `` real '' discontinuity which is not removable .
now we are ready to state the main theorems of this paper . in
order to state theorems in the unified way we use a weight function @xmath84 such that @xmath85 .
[ formationofsingularity ] let @xmath12 be an open subset of @xmath86 with a smooth boundary @xmath24 .
assume @xmath12 is non - convex , @xmath87 choose any @xmath88 with @xmath89 .
for any small @xmath90 , 1 .
there exist @xmath91 and an initial datum @xmath92 which is continuous on @xmath93 , and an in - flow boundary datum @xmath94 which is continuous on @xmath95 , satisfying @xmath96 and @xmath97 such that if @xmath98 on @xmath82 is boltzmann solution of ( [ boltzmann ] ) with the in - flow boundary condition ( [ finflow ] ) then @xmath98 is discontinuous in space and velocity at @xmath99 , i.e. @xmath100_{x_0,v_0}\neq 0.$ ] 2 .
there exist @xmath91 and an initial datum @xmath92 which is continuous on @xmath93 , satisfying @xmath101 and @xmath102 such that if @xmath98 on @xmath82 is boltzmann solution of ( [ boltzmann ] ) with the diffuse boundary condition ( [ fdiffuse ] ) with the compatibility condition ( [ cdiffuse ] ) then @xmath98 is discontinuous in space and velocity at @xmath99 , i.e. @xmath100_{x_0,v_0}\neq 0.$ ] 3
. there exist @xmath103 and an initial datum @xmath92 which is continuous on @xmath93 , satisfying ( [ smallness ] ) and @xmath104 such that if @xmath98 on @xmath82 is boltzmann solution of ( [ boltzmann ] ) with the bounce - back boundary condition ( [ fbounceback ] ) then @xmath98 is discontinuous in space and velocity at @xmath99 , i.e. @xmath100_{x_0,v_0}\neq 0.$ ] the smallness of given data ( [ gsmallness ] ) , ( [ smallness ] ) ensures the global existence of boltmzann solution for all boundary conditions @xcite .
notice that we can observe the formation of discontinuity for any point of the concave(singular ) grazing boundary @xmath64 of any generic non - convex domain @xmath12 .
if we assume that @xmath92 is continuous on @xmath93 and @xmath94 is continuous on @xmath95 , and that the compatibility conditions ( [ cinflow ] ) , ( [ cdiffuse ] ) and ( [ cbounceback ] ) are valid up to @xmath105 , then theorem 1 implies the continuity breaks down at the concave(singular ) grazing boundary @xmath64 after a short time @xmath106 for in - flow ( [ finflow ] ) , diffuse ( [ fdiffuse ] ) boundary condition and @xmath107 for bounce - back ( [ fbounceback ] ) boundary condition . for this generic cases ,
we said the boltzmann solution @xmath98 has a * local - in - time formation of discontinuity * at @xmath99 .
+ + once we have the formation of discontinuity at @xmath108 , we further establish that the discontinuity propagates along the forward characteristics .
[ propagation ] let @xmath12 be an open bounded subset of @xmath86 with a smooth boundary @xmath24 .
let @xmath1 be the boltzmann solution of ( [ boltzmann ] ) with the initial datum @xmath109 which is continuous on @xmath93 , and with one of the following boundary conditions : 1 .
for in - flow boundary condition ( [ finflow ] ) , let ( [ cinflow ] ) and ( [ gsmallness ] ) be valid and @xmath94 be continuous on @xmath95 .
2 . for diffuse boundary condition ( [ fdiffuse ] ) , assume ( [ smallness ] ) and ( [ cdiffuse ] ) .
3 . for bounce - back boundary condition ( [ fbounceback ] ) , assume ( [ smallness ] ) and ( [ cbounceback ] ) .
then for all @xmath110 we have @xmath111_{t , x_0 + ( t - t_0)v_0,v_0}\leq e^{-c_1(1+|v_0|)^{\gamma}(t - t_0)}[f(t_0)]_{x_0,v_0},\label{1sidedinequality}\ ] ] where @xmath112 only depends on @xmath113 . on the other hand ,
assume @xmath100_{x_0,v_0}\neq 0 $ ] , and @xmath114 for in - flow and diffuse boundary conditions and @xmath115 for bounce - back boundary condition , and a strict concavity of @xmath116 at @xmath117 along @xmath118 , i.e. @xmath119 then for all @xmath110 , the boltzmann solution @xmath98 is discontinuous in time and space and velocity at @xmath120 , i.e. @xmath121_{t , x_0 + ( t - t_0)v_0,v_0}\neq 0 $ ] and @xmath122_{x_0,v_0}\leq [ f]_{t , x_0 + ( t - t_0)v_0,v_0},\label{decaydiscontinuity}\ ] ] where @xmath123 , and @xmath124 which is positive for sufficiently small @xmath113 .
the strict concavity condition ( [ convexity ] ) rules out some technical issue of the backward exit time @xmath125 .
our theorem characterize the propagation of discontinuity before the forward trajectory reaches the boundary . in the case
that the forward trajectory reaches the boundary , i.e. @xmath126 , the situation is much more complicated .
denote @xmath127 .
if the trajectory hits on the boundary non - tangentially , i.e. @xmath128 , for in - flow and diffuse boundary cases , the discontinuity disappears because of the continuity of the in - flow datum and the average property of diffuse boundary operator . for bounce - back case
the discontinuity is reflected and continues to propagate along the trajectory .
if the trajectory hits on the boundary tangentially , i.e. @xmath129 , there are three possibilities .
firstly if @xmath130 then the situation is same as the case @xmath128 above . secondly , if the trajectory is contained in the boundary for awhile , i.e. there exists @xmath90 so that @xmath131 for @xmath132 then it is hard to predict the propagation of discontinuity in general . assuming certain condition on @xmath12 for example definition [ linesegment ]
, we can rule such a unlikely case .
the last case is that @xmath133 .
assume we have a sequence of @xmath134 and @xmath135 so that @xmath136 , and a directional strict concavity ( [ convexity ] ) is valid for each @xmath137 .
we can show the propagation of discontinuity also between the first and the second intersections , i.e. @xmath121_{t , x_0 ( t - t_0)v_0,v_0 } \neq 0 $ ] for @xmath138 in general . for @xmath139 , if we have very simple geometry , for example the first picture of figure 2 , we can show the propagation of discontinuity , i.e. @xmath121_{t , x_0 ( t - t_0)v_0,v_0 } \neq 0 $ ] for @xmath140 even for @xmath141 .
but in general , for example the second picture of figure 2 , we can not show @xmath121_{t , x_0 ( t - t_0)v_0,v_0 } \neq 0 $ ] for @xmath140 for @xmath142 .
[ fig2 ] [ fig3 ] + the next result states that theorem [ formationofsingularity ] and theorem [ propagation ] capture all possible singularities ( discontinuities ) , despite nonlinearity in the boltzmann equation . in other words ,
the singularity of the boltzmann solution is propagating as the linear boltzmann equation and no new singularities created from the nonlinearity of boltzmann equation .
[ continuityawayfromd ] let @xmath12 be an open bounded subset of @xmath86 with a smooth boundary @xmath24 .
let @xmath1 be a boltzmann solution of ( [ boltzmann ] ) with the initial datum @xmath109 which is continuous on @xmath143 and with one of 1 . in - flow
boundary condition ( [ finflow ] ) .
assume ( [ gsmallness ] ) is valid and the compatibility condition @xmath144 and @xmath94 is continuous on @xmath145 .
2 . diffuse boundary condition ( [ fdiffuse ] ) .
assume ( [ smallness ] ) is valid and the compatibility condition @xmath146 3 .
bounce - back boundary condition ( [ fbounceback ] ) .
assume ( [ smallness ] ) is valid and the compatibility condition @xmath147 then @xmath1 is a continuous function on @xmath71 for 1,2 and a continuous function on @xmath148 for 3 .
if the domain @xmath12 does not include a line segment ( definition [ linesegment ] ) then the continuity set @xmath71 and @xmath148 are the complementary of @xmath61 and @xmath149 respectively . therefore @xmath1 is continuous on @xmath150 for 1,2 and continuous on @xmath151 for 3 .
[ linesegment ] assume @xmath152 be open and the boundary @xmath24 be smooth .
we say the boundary @xmath24 * does not include a line segment * if and only if for each @xmath153 and for all @xmath154 there is no @xmath90 such that @xmath155 is a linear function for @xmath156 where @xmath157 from ( [ boundaryfunction ] ) .
there are many references for the mathematical study of different aspects of the boundary value problem of the boltzmann equation , for example @xcite@xcite@xcite and the references therein . in @xcite ,
an unified @xmath158 theory in the near maxwellian regime is developed to establish the existence , uniqueness and exponential decay toward a maxwellian , for all four basic types of the boundary conditions and rather general domains .
the qualitative study of the particle - boundary interaction in a bounded domain and its effects on the global dynamics is a fundamental problem in the boltzmann theory .
one of challenging questions is the regularity theory of kinetic equations in bounded domain .
this problem is hard because even for simplest kinetic equations with the differential operator @xmath159 , the phase boundary @xmath160 is always characteristic but not uniformly characteristic at the grazing set @xmath161 . in a convex domain a continuity of the boltzmann solution away from @xmath49
is established in @xcite for all four basic boundary conditions . in a convex domains ,
backward trajectories starting at interior points of the phase space can not reach points of the grazing boundary @xmath49 , due to velocity lemma(@xcite@xcite ) , where possible singularities may exist . in general ,
on the other hand , in a non - convex domain , backward trajectories starting at interior points of the phase space can reach the grazing boundary .
therefore we expect singularities will be created at some part of grazing boundary @xmath49 and propagate inside of the phase space .
this question has been attracting a lot of attentions from early 90s , see references in pp.9192 in sone s book @xcite . for boltzmann equation ,
most of works are numerical studies @xcite@xcite@xcite and few mathematical studies .
once we enlarge our survey to propagation of singularities which already exist on initial data or boundary data , there are some mathematical works @xcite@xcite@xcite@xcite@xcite as well as numerical works @xcite@xcite . in @xcite , for linear bgk model , a propagation of discontinuity , which exists already in the boundary data , is studied mathematically and also numerically . in @xcite , for the full boltzmann equation in the near vacuum regime , a propagation of sobolev @xmath162 singularity , which exists already in the initial data , is studied and same effect has been recently shown in the near maxwellian regime @xcite@xcite . in vlasov theory , we refer to @xcite@xcite@xcite for the boundary value problem .
singular solutions were studied in @xcite extensively . in @xcite , the non - convexity condition of boundary
is replaced by the inward electric field which has a similar effect with non - convexity of the boundary . in convex domains , hlder estimates of vlasov solution with specular reflection boundary
is solved recently @xcite@xcite , but sovolev - type estimate is still widely open .
+ + our results give a rather complete characterization of formation and propagation of singularity for the nonlinear boltzmann equation near maxwellian in general domain under in - flow , diffuse , bounce - back boundary conditions .
there is no restriction of the time interval .
more precisely we show that for any non - convex point @xmath16 of the boundary and velocity tangent to @xmath24 at @xmath16 , there exists an initial datum ( and in - flow datum , for in - flow boundary condition case ) such that the boltzmann solution has a jump discontinuity at @xmath163 .
once the discontinuity occurs at the grazing boundary , this discontinuity propagates inside along the forward trajectory until it hits the boundary again . and except those points , the grazing boundary and forward trajectories emanating from the grazing boundary
, we can show that the boltzmann solution is continuous.(continuity away from @xmath61 ) * 1 . the equality induced by non - convex domain * + we consider near maxwellian regime and linearized boltzmann equation ( [ linearboltzmannequation ] ) .
the formation of discontinuity is a consequence of following estimate .
assume @xmath164 as below picture so that for sufficiently small @xmath165 the backward trajectory @xmath166 is in an interior of the phase space .
for simplicity we impose the trivial in - flow boundary condition @xmath167 which corresponds @xmath168 ( [ inflow ] ) .
consider points @xmath169 in @xmath170 and @xmath171 missing the non - convex part near @xmath163 and both sequences converge @xmath163 as @xmath172 .
[ fig4 ] + now suppose the solution @xmath173 of the linearized boltzmann equation be continuous around @xmath163 .
then the boltzmann solution @xmath173 at @xmath169 @xmath174 and at @xmath171 , @xmath175 converges each other as @xmath176
. then we have the following equality @xmath177 thanks to @xcite , the pointwise estimate of @xmath173 , with some standard estimates of @xmath178 , the right hand side of above equality has magnitude @xmath179 .
if you choose @xmath180 then the above equality ( [ inequalityinducedbynonconvexbody ] ) can not be true for sufficiently small @xmath181 unless the trivial case @xmath182 .
therefore the boltzmann solution @xmath173 can not be continuous at @xmath163 . for diffuse ( [ fdiffuse ] ) ,
bounce - back ( [ fbounceback ] ) boundary conditions we also obtain the equality induced by non - convex domain similar as ( [ inequalityinducedbynonconvexbody ] ) .
+ + this argument bases on the idea that free transport effect is dominant to collision effect if time @xmath165 and the perturbation @xmath183 is small .
continuity of the gain term @xmath184 * + the smoothing effect of the gain term @xmath184 is one of the fundamental features of the boltzmann theory .
there are lots of results about the smoothing effect in sobolev regularity , for example @xmath185 with some assumption on various collision kernels @xcite@xcite@xcite .
to study the propagation of singularity and regularity , in the case of angular cutoff kernel ( [ gradcutoff ] ) , it is standard to use duhamel formulas and combine the velocity average lemma and the regularity of @xmath184 @xcite . for detail , see villani s note @xcite especially pp .
7779 . + +
in order to study the propagation of discontinuity and continuity we need a totally different smooth effect of @xmath184 . for the discontinuity induced by the non - convex domain ,
we need following : recall the grazing set @xmath75 in definition [ grazingset ] .
a test function @xmath76 is continuous on @xmath186\times(\omega\times\mathbb{r}^3 ) \backslash \mathfrak{g}$ ] and bounded on @xmath186\times\omega\times\mathbb{r}^3 $ ] .
then @xmath187\times \omega \times \mathbb{r}^3 ) .\label{continuityq+}\ ] ] recall that the grazing set @xmath188 .
the grazing section @xmath189 is a union of straight lines in velocity space @xmath86 and two dimensional lebesque measure of @xmath190 is zero ( hongjie dong s lemma , lemma 17 of @xcite ) .
moreover , using continuous behavior of @xmath191 in @xmath16 , one can invent a very effective covering of @xmath191 ( guo s covering , lemma 18 of @xcite ) . because of those geometric and size restriction on @xmath75 , even the gain term @xmath184 is an integration operator in @xmath192 alone
, we can prove the smoothing effect of @xmath184 in @xmath193\times \omega \times \mathbb{r}^3 ) $ ] for @xmath194 and @xmath192 , see theorem [ continuity ] .
notice that those smoothing effect on @xmath195 has been believed to be true for long time without a mathematical proof in numerical communities @xcite , p1587 of @xcite , p502 of @xcite .
+ + the main idea to prove the smoothing effect in @xmath195 is to use the carleman s representation for @xmath196 which has been a very effective tool @xcite@xcite@xcite . @xmath197 with the hyperplane @xmath198 .
we will show the smallness of @xmath199 for @xmath200 .
assume we have sufficient decay of @xmath80 for large @xmath192 .
replace the integrable kernel @xmath201 by smooth compactly supported function and cut off the singular part of @xmath202 to control the above difference as @xmath203 where @xmath204 is chosen to be @xmath205 for convenience .
one can easily control the integration at the first line . because for the first term , integrating over @xmath206 , we can cut off a small neighborhood of @xmath191 from @xmath207 .
away from that neighborhood , using the continuity of @xmath80 away from @xmath191 we can control the integrand pointwisely . in order to control the second line integration
we have to control the difference in big braces . to do that we choose a special change of variables for @xmath208 , ( [ changeofvariables2 ] ) .
under this change of variables the second line is bounded by @xmath209 the second integration term above is a function of @xmath210 and @xmath192 .
unfortunately one can not expect a pointwise control ( smallness ) of the second integration for all @xmath192 : even the grazing section @xmath191 , where @xmath211 might have discontinuity , is small , i.e. 2-dimensional lebesque measure of @xmath212 is zero , the measure on the plane @xmath213 could be large(even infinite ) .
however , in section 3.3 , we can show that that bad situation happens for very rare @xmath206 in @xmath214 and use the integration over @xmath206 to control the above integration . + * 3 .
new proof of continuity of boltzmann solution with diffuse boundary condition * + in section 5.2 we prove a continuity away from @xmath61 of boltzmann solution with diffuse boundary condition using simple iteration scheme ( [ hm ] ) with iteration diffuse boundary condition ( [ iterationdiffuse ] ) .
this iteration scheme has several advantages .
first it preserves a continuity away from @xmath61 as @xmath215 increasing , that is , if @xmath216 is continuous away from @xmath61 then @xmath217 is also continuous away from @xmath61 .
second , the sequence @xmath218 has uniform @xmath219 bound and moreover it is cauchy in @xmath219 for in - flow boundary condition @xmath220 .
therefore @xmath221 , a solution of the linear boltzmann equation is continuous local in time . combining with uniform - in - time boundedness of boltzmann solution ( @xcite )
, we achieve the continuity for all time . in order to apply this idea to diffuse boundary condition ,
we use guo s idea @xcite : a norm of the diffuse boundary operator is less than 1 effectively , if we trace back several bounces .
this approach gives simpler proof for the continuity of boltzmann equation with diffuse boundary condition with convex domain ( see lemma @xmath222 of @xcite ) . in section 2 , we state some preliminary facts which are useful tools for this paper . in section 3 , we state and
prove the continuity of @xmath223 ( theorem 4 ) . in section @xmath224
, we deal with in - flow boundary , diffuse boundary and bounce - back boundary , respectively . for each section , first we prove the formation of discontinuity ( theorem 1 ) .
then we show the continuity away from @xmath61 ( theorem 3 ) . using this continuity , combining with continuity of @xmath184
, we show the propagation of discontinuity ( theorem 2 ) .
in this section we study continuity properties of the backward exit time @xmath62 and , a measure theoretic property and geometric covering of the grazing set @xmath75 , and estimates of boltzmann operators and the carleman s representation .
+ + we use lemma 1 of @xcite , basic properties of the backward exist time @xmath62 : [ huang ] @xcite let @xmath12 be an open bounded subset of @xmath86 with a smooth boundary @xmath24 .
let @xmath31 be connected with @xmath225 backward in time through a trajectory of ( [ ode ] ) .
the backward exit time @xmath62 is lower semicontinuous . 2 if @xmath226then @xmath227
are smooth functions of @xmath228 so that@xmath229 for a convex domain , if a point @xmath163 is in the interior of the phase space , i.e. @xmath230 then the condition ( [ negative ] ) is always satisfied and hence @xmath62 is smooth due to lemma [ huang ] .
however for a non - convex domain , there is a point @xmath163 in @xmath55 but @xmath231 , i.e. @xmath232 .
we further investigate a continuity property of @xmath125 for that case .
indeed , discontinuity behavior of @xmath62 for @xmath233 is a main ingredient of the formation of discontinuity .
[ tbcon ] let @xmath152 be an open set with a smooth boundary @xmath24 .
assume @xmath234 with @xmath235 and @xmath236 .
consider @xmath237 , i.e. @xmath238 .
@xmath239 recall @xmath240 and @xmath64 in definition [ grazingboundary ] . throughout this proof , without loss of generality
we assume that @xmath24 is a graph of @xmath241 locally and @xmath242 and @xmath243 .
moreover assume @xmath244 and @xmath245 so that @xmath246 .
first , let @xmath247 . by the definition of @xmath240 , we have @xmath248 and @xmath249 for @xmath250 . using the continuity of @xmath241
, choose sufficiently small @xmath251 such that @xmath252 and @xmath253 for @xmath254 .
fix @xmath255 and @xmath256 .
we define @xmath257 for @xmath258 , @xmath259 for @xmath260 . for @xmath261 , @xmath262 for @xmath260 . using the continuity of @xmath241 and @xmath263 ,
there exists @xmath264 so that @xmath265 , i.e. @xmath266 . if @xmath267 and @xmath268 then @xmath269 and @xmath270 so that @xmath271 . next , let @xmath233 . by the definition of the concave grazing boundary @xmath64
, we have @xmath248 and @xmath249 for @xmath250 .
choose a sequence @xmath272 .
there exists @xmath273 such that @xmath274 for sufficiently large @xmath275 .
this implies that @xmath276 but @xmath277 as @xmath172 .
+ + in the next two lemmas , we consider the grazing set @xmath75 ( definition [ grazingset ] ) including the discontinuity set @xmath61 .
the next lemma , lemma 17 of @xcite due to hongjie dong , is important to control a size of @xmath75 .
we denote @xmath278 as a standard 2-dimensional lebesque measure and @xmath279 as a standard 3-dimensional lebesque measure .
recall that the grazing section @xmath191 in definition [ grazingset ] .
[ hongjie ] @xcite if @xmath24 is @xmath280 then the grazing section @xmath191 restricted to @xmath281 has zero 2-dimensional lebesque measure , i.e. @xmath282 for all @xmath283 . with condition @xmath284
, we can construct the guo s covering which is little bit stronger than the original one in lemma 18 in @xcite .
[ bouncebacklower ] @xcite assume @xmath285 is valid for all @xmath283 .
let @xmath286 .
then for any @xmath287 and @xmath288 there exist @xmath289 and @xmath290 balls @xmath291 , as well as open sets @xmath292 of @xmath293 which are radial symmetric , i.e. @xmath294 with @xmath295 and @xmath296 for all @xmath297 such that for any @xmath298 there exists @xmath299 so that @xmath300and for @xmath301 @xmath302 or equivalently @xmath303 combining lemma [ hongjie ] and lemma [ bouncebacklower ] , we have following lemma , which is useful to prove theorem [ continuity ] .
namely , a function which is continuous away from the grazing set @xmath75 is uniformly continuous except arbitrary small open set containing @xmath75 .
[ uniformlycontinuous ] assume @xmath80 is continuous on @xmath186\times ( \omega\times \{v\in\mathbb{r}^3 : \frac{1}{m}\leq |v|\leq n \})\backslash \mathfrak{g}$ ] .
for fixed @xmath304 and @xmath273 and @xmath288 , there exist @xmath305 and an open set @xmath306 which is radial symmetric , i.e. @xmath307 with @xmath308 and @xmath309 such that @xmath310 for @xmath311 and @xmath312 let @xmath313 .
due to guo s covering @xcite , lemma [ bouncebacklower ] , we can choose @xmath314 including @xmath16 and @xmath315 , as well as @xmath316 so that @xmath317 with @xmath318 .
notice that @xmath319 . we can choose an open set @xmath320 so that @xmath321 and @xmath322 .
since both of @xmath323 and @xmath324 are compact subsets of @xmath325 , we have a positive distance between two sets , i.e. @xmath326 assume @xmath327 . fix @xmath283 and @xmath328 .
then @xmath329 implies that @xmath330 . for such @xmath331 and
@xmath332 consider the function @xmath80 as it s restriction on a compact set @xmath186\times\bar{b}(x;\delta)\times b_n \backslash o_{x_i}$ ] .
therefore @xmath333\times\bar{b}(x;\delta)\times b_n \backslash o_{x_i}}$ ] is uniformly continuous .
hence @xmath334 can be controlled small uniformly if @xmath90 is sufficiently small .
+ + we will use the carleman s representation @xcite@xcite in the proof of theorem [ continuity ] crucially .
let @xmath335 be defined by ( [ qgl ] ) and let @xmath336 and @xmath337 , @xmath338 make @xmath339 almost everywhere .
then the * carleman s representation * is @xmath340 where @xmath213 is a hyperplane containing @xmath338 and perpendicular to @xmath341 , i.e. @xmath342 + in the proof of theorem [ continuity ] we need to control the integration over @xmath213 in ( [ carleman ] ) frequently : [ estimwhb ] for a rapidly decreasing function @xmath343 , we have @xmath344 where @xmath345 only depends on @xmath80 .
for fixed @xmath206 and @xmath192 , let us denote @xmath346 , with @xmath347 , be the orthonormal basis of @xmath86 such that any @xmath348 can be written as @xmath349 . since @xmath350 from ( [ hyperplane ] ) , there is @xmath351 such that @xmath352 where @xmath353
. then we can write @xmath354 and @xmath355 .
moreover @xmath356 .
we can write the left hand side of ( [ difference1 ] ) as @xmath357 in terms of the standard perturbation @xmath173 such that @xmath358 the boltzmann equation can be rewritten as @xmath359where the standard linear boltzmann operator , see @xcite , is given by @xmath360with the collision frequency @xmath361 for @xmath362 and @xmath363we recall two estimates of operators @xmath364 and @xmath365 from @xcite .
the grad estimate for hard potentials : @xmath366recall @xmath367 in ( [ weight])@xmath368 let @xmath369
. then there exists @xmath370 and @xmath371 such that for @xmath372 @xmath373 for the nonlinear collision operator @xmath374 also we recall a standard estimate @xmath375 for @xmath376 .
in this section we mainly prove the following theorem : + [ continuity ] assume @xmath1 is continuous on @xmath186\times(\omega\times\mathbb{r}^3 ) \backslash \mathfrak{g}$ ] and @xmath377\times\bar{\omega}\times\mathbf{r}^3 ) } < + \infty,\ ] ] where @xmath378 with @xmath379 and @xmath380 .
then @xmath381 is continuous in @xmath382\times \omega \times \mathbb{r}^3 $ ] and @xmath383\times \bar{\omega } \times\mathbb{r}^3 } |\nu^{-1 } \bar{w}^{-1 } q_{+}(f , f)(t , x , v)|<\infty.\label{bdq}\ ] ] theorem [ continuity ] , a smooth effect in @xmath195 , is the crucial ingredient to prove theorem [ propagation ] and theorem [ continuityawayfromd ] .
this smooth effect of the gain term ensures that there is no singularity created by the nonlinearity of botlzmann equation .
+ + it is easy to show the boundedness ( [ bdq ] ) from @xmath384\times(\bar{\omega}\times\mathbf{r}^3))}^2\end{aligned}\ ] ] where we used ( [ standardbound ] ) and @xmath385 .
+ + next we will show the continuity part of theorem [ continuity ] .
the goal of following three subsections is to show @xmath386\times \omega\times\mathbb{r}^3 , \ \text{there is \ } \delta>0 \text { \ such that}\nonumber\\ & & |q_{+}(\bar{w}h,\bar{w}h)(\bar{t},\bar{x},\bar{v})- q_{+}(\bar{w}h,\bar{w}h)(t , x , v)| < \varepsilon \ \ \text{for \ } |(\bar{t},\bar{x},\bar{v})-(t , x , v)| < \delta .
\label{continuitystatement}\end{aligned}\ ] ] in this section , we use the carleman s representation to split @xmath387 in a natural way ( [ diff ] ) , and introduce two change of variables ( [ changeofvariables1 ] ) and ( [ changeofvariables2 ] ) .
+ + it is convenient to define @xmath388 where @xmath389\times ( \bar{\omega}\times\mathbb{r}^3 ) ) } = ||\bar{w}^{-1 } f ||_{l^{\infty}([0,t]\times(\bar{\omega}\times\mathbf{r}^3))}.$ ] choose @xmath390 . using the carleman s representation ( [ carleman ] ) we have @xmath391 @xmath392 in order to control the first term of ( [ diff ] ) , we need to compare arguments of @xmath393 and @xmath394 . for that purpose ,
we introduce the following change of variables : for fixed @xmath192 and @xmath332 in @xmath86 we define @xmath395 then two planes @xmath396 have same normal direction . the distance between to planes
is @xmath397 .
assume ( [ changeofvariables1 ] ) .
clearly @xmath398 where @xmath399 is @xmath400 identity matrix
. the normal direction of @xmath401 is @xmath402 which is also the normal direction of @xmath403 . to measure a distance between two planes @xmath403 and @xmath401
, we consider the line passing @xmath192 and directing @xmath404 , which is @xmath405 .
the solution of @xmath406 is a solution of @xmath407 .
easily we have @xmath408 . since @xmath409 is unit - speed line we know that @xmath410 is the distance between @xmath396 .
+ + an important property of ( [ changeofvariables1 ] ) is that two planes @xmath396 have the same normal direction . in order to control the second term of ( [ diff ] ) , we need to compare arguments of @xmath411 and @xmath412 especially @xmath348 and @xmath413 .
for that purpose , we introduce the following change of variables : for fixed @xmath414 and @xmath332 in @xmath86 , we define a unit jacobian change of variables @xmath415 in this change of variables @xmath416 if and only if @xmath417 . assume ( [ changeofvariables1 ] ) and ( [ changeofvariables2 ] )
. clearly @xmath418 .
we can check following equality : @xmath419 by definition , @xmath348 is equivalent to @xmath420
. then , from the above equality , we conclude @xmath421 which is equivalent to @xmath416 .
+ under the first change of variables ( [ changeofvariables1 ] ) , we can rewrite the first term of @xmath422 as @xmath423 under the second change of variables ( [ changeofvariables2 ] ) , we can rewrite the second term of @xmath422 as @xmath424 we will estimate ( [ estimategainterm ] ) and ( [ estimategainterm2 ] ) separately in following two sections .
we divide into several cases : @xmath425 @xmath426 . from lemma [ estimwhb ] , for @xmath427 we can estimate @xmath428 hence we have @xmath429 @xmath430 @xmath431 and @xmath432 , or @xmath431 and @xmath433 .
also assume @xmath434 .
@xmath435 where we have used @xmath436 and @xmath437 and lemma [ estimwhb ] .
@xmath438 @xmath431 and @xmath439 .
@xmath440 since @xmath441 is integrable we can choose a smooth function @xmath442 with compact support such that @xmath443 therefore we can bound ( [ a1 ] ) by two parts @xmath444 from ( [ estimate2 ] ) , it is easy to control the first term @xmath445 now we are going to estimate the second term ( [ estimate112 ] ) . applying lemma [ uniformlycontinuous ] to @xmath446 , we can choose @xmath447 and an open set @xmath448 with @xmath449 such that @xmath450 for @xmath451 and @xmath452 .
therefore we can split the second part ( [ estimate112 ] ) as integration over @xmath453 and @xmath454 and control as @xmath455 in summary , combinig @xmath456 and @xmath457 , we have established @xmath458 choosing sufficiently large @xmath459 and @xmath288 then @xmath460 the estimate of ( [ estimategainterm2 ] ) is much more delicate .
the reason is that we can not expect @xmath461 in ( [ estimategainterm2 ] ) is small for all @xmath462 .
we know that @xmath463 may not be continuous on @xmath464 .
even @xmath191 is radial symmetric and has a small measure by lemma [ hongjie ] , a bad situation , the intersection of @xmath191 and @xmath213 could have large ( even infinite ) 2-dimensional lebesque measure , can happen .
however we can show that such bad situations only happen for very rare @xmath206 s in @xmath86 . using the integration over @xmath462 , we are able to control ( [ estimategainterm2 ] ) small .
+ + recall @xmath465 and @xmath466 in ( [ estimategainterm2 ] ) .
we divide into several cases : * case 1 :* @xmath426 .
follow exactly same proof of * case 1 * of the previous subsection , we conclude @xmath467 * case 2 :* @xmath431 and @xmath432 .
we go back to original formula , the second term of ( [ diff ] ) , and use lemma [ estimwhb ] to estimate @xmath468 * case 3 :* @xmath431 , @xmath469 , and @xmath470 or @xmath471 . in the case of @xmath470
, we have @xmath472 in the case of @xmath471 we have @xmath473 * case 4 :* @xmath474 and @xmath475 . in order to remove the unboundedness of @xmath476 in ( [ estimategainterm2 ] ) , we choose a positive smooth function @xmath477 with compact support such that @xmath478 splitting @xmath479 into two parts @xmath480 where we used ( [ z1 ] ) for the first line .
from now we will focus to estimate ( [ z1 ] ) .
* case 5 :* @xmath481 and @xmath482 or @xmath483 .
this region included the part where the collision kernel @xmath484 has a singular behavior .
@xmath485 * case 6 :* @xmath486 and @xmath487 and @xmath488 and @xmath489 .
we estimate @xmath490 we need this step because of the singular behavior of @xmath491 where @xmath492 with @xmath493 .
the function @xmath494 is not continuous at @xmath495 and continuous away from @xmath496 , i.e. the restriction of @xmath497 on a compact set , @xmath498 is uniformly continuous . from @xmath499 and @xmath500
we have lower bound of @xmath501 similarly from @xmath488 and @xmath502 we have a lower bound of @xmath503 therefore for any @xmath504 , we can choose @xmath90 so that @xmath505 for @xmath487 and @xmath488 and @xmath489 .
+ now we split ( [ case5 ] ) by two parts @xmath506 using ( [ e1case5 ] ) , the continuity of @xmath484 away from @xmath496 , the first line above is bounded by @xmath507 + in the remainder of this section we will focus on ( [ ab2 ] ) : + * estimate of ( [ ab2 ] ) * + @xmath508 where we used @xmath509 . recall our choice of @xmath510 and @xmath208 from ( [ changeofvariables1 ] ) and ( [ changeofvariables2 ] ) to have @xmath511 we will use the followin strategy : separate @xmath512 into two parts @xmath513 the first part is the integration over @xmath453 , a neighborhood of @xmath191 that contains possible discontinuity of @xmath514 .
moreover we expect the measure of the neighborhood @xmath453 is small so we can control the first term . for the second term , we will use the continuity of the integrand @xmath515 .
however if @xmath516 then @xmath191 could be a large measure set in @xmath517 .
for example if @xmath518 then @xmath191 is @xmath519plane and @xmath520 is also @xmath519plane . therefore we have to divide two cases @xmath521 and @xmath516 and study separately . + * case of * @xmath522 + in the case of @xmath521 , assume @xmath523 for sufficiently small @xmath524 .
we will divide the velocity space @xmath525 into @xmath526 the important property of @xmath527 is that if @xmath528 then @xmath213 does not contain zero .
we can split the integration part of ( [ abc ] ) , @xmath529 into @xmath530 notice that @xmath531 has a small measure : @xmath532 therefore we have @xmath533 now we are going to estimate ( [ b2 ] ) .
here we use a property of @xmath534 : for @xmath535 we have @xmath536 where we also have used @xmath469 and @xmath431 .
from lemma [ uniformlycontinuous ] we use @xmath453 , an open radial symmetric subset of @xmath537 with a small measure and @xmath515 is uniformly continuous on @xmath454 , to split ( [ b2 ] ) into @xmath538 for the last line , we use lemma [ uniformlycontinuous ] to know estimate @xmath539 , for @xmath540 and @xmath541 .
therefore @xmath542 in order to show that @xmath543 is small , we introduce following projection : [ projection ] assume @xmath544 .
let @xmath545 .
we define a projection @xmath546 for @xmath547 , define the restricted projection @xmath548 then for @xmath549 the jacobian of @xmath550 is bounded : @xmath551 where @xmath552 is defined by @xmath553 . without loss of generality , we may assume @xmath554 .
using spherical coordinate , @xmath555 and a jacobian matrix of @xmath550 , @xmath556 therefore a jacobian of @xmath550 is @xmath557 notice that @xmath558 because @xmath559 and @xmath560 we have @xmath561 assume we choose @xmath562 . by definition
we know that @xmath563 and the 2-dimension lebesque measure of @xmath564 is bounded by @xmath565 therefore we have an upper bound of ( [ u1 ] ) : @xmath566 where @xmath567 . in case of @xmath521 , from ( [ sum28 ] ) , ( [ sum30 ] ) and ( [ sum29 ] )
, we have @xmath568 + * case of * @xmath569 + in this case , we do not have a upper bound of the jacobian of @xmath550 .
instead we will use the structure of @xmath191 of lemma [ bouncebacklower ] crucially . in case of @xmath516 , we split @xmath570 @xmath571 for @xmath206 , we use a spherical polar coordinates @xmath572 so that @xmath573 by definition , @xmath574 is a plane containing the origin and normal to @xmath206 . we know
that @xmath574 is generated by two unit vectors @xmath575 we will use a polar coordinate @xmath576 for @xmath577 , i.e. @xmath578 direct computation gives @xmath579 @xmath580 therefore we have following identity @xmath581 recall the standard 3-dimensional polar coordinates and 2-dimensional polar coordinates : @xmath582 and use above identities to control @xmath583 by @xmath584 we focus on the inner integration @xmath585 and divide it into @xmath586\cup[\frac{\pi}{2}+\varrho,\frac{3\pi}{2}-\varrho]\cup[\frac{3\pi}{2}+\varrho,2\pi]}\label{x1}.\end{aligned}\ ] ] easily @xmath587 . for ( [ x1 ] ) , we use @xmath588 and @xmath589 on @xmath590\cup[\frac{\pi}{2}+\varrho,\frac{3\pi}{2}-\varrho]\cup[\frac{3\pi}{2}+\varrho,2\pi]$ ] and @xmath591 $ ] to have @xmath592 where we used ( [ identity ] ) .
to sum we have @xmath593 + on the other hand for ( [ c2 ] ) we can use lemma [ uniformlycontinuous ] to have @xmath594 from ( [ ssum1 ] ) and ( [ ssum2 ] ) we have @xmath595 to summarize , from ( [ sum21 ] ) , ( [ sum22 ] ) , ( [ sum23 ] ) , ( [ sum24 ] ) , ( [ sum25 ] ) , ( [ sum26 ] ) , ( [ sum27 ] ) , ( [ sum31 ] ) and ( [ sum32 ] ) , we have established @xmath596 we choose @xmath597 sufficiently large and @xmath524 sufficiently small so that we can control @xmath598 . combining with the result of previous subsection ( [ finalest1 ] ) ,
we conclude ( [ continuitystatement ] ) and and prove theorem [ continuity ] . the following is a consequence of theorem [ continuity ] .
[ continuity1 ] assume @xmath601 is continuous on @xmath186\times ( \bar{\omega}\times\mathbb{r}^3)\backslash\mathfrak{g}$ ] and @xmath602\times ( \bar{\omega}\times\mathbb{r}^3)).\ ] ] then @xmath603 and @xmath604 are continuous in @xmath186\times\omega\times\mathbb{r}^3 $ ] and @xmath383\times\bar{\omega}\times\mathbb{r}^3}|\nu^{-1}(v)w(v)k(f)|<\infty , \ \
\sup_{[0,t]\times\bar{\omega}\times\mathbb{r}^3}|\nu^{-1}(v)w(v)\gamma_{+}(f , f)| < \infty.\label{boundedkgamma}\ ] ] the above boundedness is a direct consequence of ( [ wk ] ) and ( [ gamma1 ] ) .
thanks to theorem [ continuity ] , we already established the continuity of @xmath605 . therefore we only need to show the continuity of @xmath606 choose @xmath607 so that @xmath452
. we will estimate @xmath608 where we used a change of variables @xmath609 for the underlined term . using the taylor s expansion we control @xmath610 where @xmath611 for some @xmath612 and @xmath613 .
therefore we control @xmath614 where we have used the the angular cutoff assumption ( [ gradcutoff ] ) .
now we estimate ( [ difference12 ] ) with following steps : * case 1 :* @xmath615 .
since @xmath616 , we estimate @xmath617 * case 2 :* @xmath618 .
a function @xmath173 is continuous on @xmath186\times(\bar{\omega}\times b(0;n))\backslash\mathfrak{g}$ ] . by the lemma [ uniformlycontinuous ]
, we can choose @xmath619 with @xmath620 with @xmath621 for @xmath622 with @xmath623 .
therefore @xmath624 is bounded by @xmath625 from ( [ 1 - 1 ] ) , ( [ 1 - 2 ] ) and ( [ 1 - 3 ] ) , we summarize @xmath626 which is less than @xmath627 for sufficiently large @xmath628 and sufficiently small @xmath629
. + + + in following sections , we will prove theorem 1 , theorem 2 and theorem 3 , for each boundary conditions . in order to write theorems in the unified way @xcite for all boundary condition cases , we use the weight function @xmath630 in ( [ weight ] ) and define @xmath631 in terms of @xmath514 , the boltzmann equation ( [ boltzmann ] ) is equivalent to @xmath632 where @xmath633 with boundary conditions : @xmath634
in this section , we consider the linearized boltzmann equation ( [ hbol ] ) with the in - flow boundary condition ( [ inflow ] ) .
first we will show the formation of discontinuity using a pointwise estimate of the boltzmann solution @xcite .
then we use the continuity of collision operators , theorem [ continuity ] , to show a continuity of solution on the continuity set @xmath71 and the propagation of discontinuity on the discontinuity set @xmath61 .
we prove part 1 of theorem [ formationofsingularity ] . without loss of generality
we may assume @xmath635 and @xmath636 and @xmath88 .
locally the boundary is a graph , i.e. @xmath637 . the condition @xmath88 implies @xmath638 and @xmath639 which means @xmath640 for @xmath641 .
( see figure 3 ) + for simplicity we assume a zero boundary datum , i.e. @xmath642 . from theorem 1 of @xcite , we have a global solution of the linearized boltzmann equation ( [ hbol ] ) with zero in - flow boundary condition , satisfying @xmath643 for some @xmath644 . in the proof
we do not use the decay estimate but just boundedness @xmath645 recall the constants @xmath646 and @xmath647 from ( [ wk ] ) and ( [ gamma1 ] ) . choose @xmath648
sufficiently small so that @xmath649 where @xmath650 for any @xmath651 with @xmath652 .
this choice is possible because the right hand side of ( [ 1/2 ] ) is a continuous function of @xmath653 and it has a value @xmath654 when @xmath655 .
furthermore assume a condition for our initial datum @xmath656 : there is sufficiently small @xmath657 such that @xmath658 and @xmath659 we claim the boltzmann solution @xmath514 with such an initial datum @xmath656 and zero in - flow boundary condition is not continuous at @xmath660 .
we will use a contradiction argument : suppose @xmath661_{x_0,v_0 } = 0.\label{contradiction}\ ] ] choose sequences of points @xmath662 and @xmath663 .
because of our choice , for sufficiently large @xmath664 , the characteristics @xmath665 $ ] is near to @xmath666 , i.e. @xmath667 hence the boltzmann solution at @xmath668 is @xmath669 combining @xmath670 with ( [ contradiction ] ) , we conclude @xmath671 on the other hand , using ( [ boundedness ] ) we can estimate @xmath672 which is contradiction to ( [ hhh ] ) .
we aim to prove part 1 of theorem [ continuityawayfromd ] in this section .
first we recall lemma 12 of @xcite , the representation for solution operator @xmath673 for the homogeneous transport equation with in - flow boundary condition : [ ginflowdecay ] @xcite let @xmath674 and @xmath675 let @xmath676 be the solution to the transport equation @xmath677for @xmath678 @xmath679 next we prove a generalized version of lemma 13 in @xcite .
[ transportequation ] let @xmath12 be an open subset of @xmath86 with a smooth boundary @xmath24 and an initial datum @xmath680 be continuous in @xmath681 and a boundary datum @xmath682 be continuous in @xmath186\times \{\gamma_- \cup \gamma_0^{i-}\}$ ] .
also assume @xmath683 and @xmath76 are continuous in the interior of @xmath186\times\omega\times\mathbb{r}^3 $ ] and satisfy @xmath684\times\omega\times\mathbb{r}^3}\big|q(t , x , v)\big| < \infty$ ] and @xmath684\times\omega}\big|\phi(\cdot,\cdot , v)\big| < \infty$ ] for all @xmath338 .
let @xmath685 be the solution of @xmath686 assume the compatibility condition on @xmath72 , @xmath687 then the boltzmann solution @xmath685 is continuous on the continuity set @xmath71 .
furthermore , if the boundary @xmath24 does not include a line segment ( definition [ linesegment ] ) then @xmath685 is continuous on a complementary set of the discontinuity set , i.e. @xmath688\times\bar{\omega}\times\mathbb{r}^3\}\backslash\mathfrak{d}$ ] .
continuity on @xmath689\}$ ] is obvious from the assumption .
fix @xmath690 .
notice that @xmath691}(s)=0,\label{eqntalongxv}\ ] ] along the characteristics @xmath692 until the characteristics hits on the boundary . choose @xmath390 and use a change of variables @xmath693 with @xmath694 $ ] to have @xmath695}(s)=0,\label{eqntalongxv1}\end{aligned}\ ] ] where @xmath696 and @xmath697 . by the definition @xmath71 , we can separate two cases : @xmath67 , @xmath698 .
+ from the assumption @xmath699 , we know that ( [ eqntalongxv ] ) holds for @xmath700 .
now we choose @xmath701 near @xmath31 so that @xmath702 , and @xmath696 is in the interior of @xmath12 for all @xmath694 $ ] . taking the integration over @xmath703 $ ] of @xmath704 to have @xmath705}(s ) q(s , x(s),v(s))e^{-\int_s^t \phi(\tau , x(\tau),v(\tau))d\tau}\\ & & \
\ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ \ -\mathbf{1}_{[t-\bar{t}+\max\{0,\bar{t}-t_{\mathbf{b}}(\bar{x},\bar{v})\},t]}(s ) q(s+(\bar{t}-t),\bar{x}(s),\bar{v}(s ) ) e^{-\int_s^t \phi(\tau+(\bar{t}-t),\bar{x}(\tau),\bar{v}(\tau))d\tau } \big\}ds.\end{aligned}\ ] ] since @xmath656 and @xmath80 is continuous , it is easy to see that the first line above goes to zero when @xmath706 . for the remainder
we separate cases : @xmath707 and @xmath708 .
if @xmath707 the remainder is bounded by @xmath709 and @xmath80 , and the second term is small as @xmath706 .
the case @xmath708 is similar .
+ we only have to consider cases of @xmath710 and @xmath711 . by definition @xmath712 . from lemma [ tbcon ] , we know that @xmath62 is a continuous function when @xmath713 . in the case of @xmath710 , for @xmath607
, we have @xmath714 . taking the integration over @xmath703 $ ] of @xmath704 to have @xmath715 using the continuity of @xmath125 and @xmath716 and @xmath80 , it is easy to show that @xmath717 as @xmath706 . in the case of
@xmath711 we can choose @xmath607 so that @xmath718 .
taking the integration over @xmath703 $ ] of @xmath704 to have @xmath719 where the first three lines can be small using the compatibility condition and continuity of @xmath656 in @xmath681 and a continuity of @xmath682 on @xmath186\times\{\gamma_- \cup \gamma_0^{i-}\}$ ] and continuity of @xmath80 . for the fourth line above
, we use the continuity of @xmath716 and @xmath80 .
+ if the boundary @xmath24 does not include a line segment ( definition [ linesegment ] ) we have @xmath720\times\bar{\omega}\times\mathbb{r}^3\}\backslash\mathfrak{d}$ ] . + * proof of part 1 of theorem [ continuityawayfromd ] * we will use the following iteration scheme @xmath721 with @xmath722 and @xmath723 with @xmath724 . for simplicity
we define @xmath725 * step 1 * : we claim @xmath726 for all @xmath727 and for any @xmath728 where @xmath729\times\bar{\omega}\times\mathbb{r}^3\ } , \label{ct}\ ] ] where the continuity set @xmath71 is defined in ( [ c ] ) .
we will use mathematical induction to show ( [ mcontii ] ) .
we choose @xmath730 then ( [ mcontii ] ) is satisfied for @xmath731 .
assume ( [ mcontii ] ) for all @xmath732 .
rewrite @xmath733 then the equation of @xmath217 is @xmath734 from theorem [ continuity ] and corollary [ continuity1 ] we know that @xmath735 and @xmath736 is continuous in @xmath186\times\omega\times\mathbb{r}^3 $ ] .
apply lemma [ transportequation ] where @xmath76 corresponds to @xmath737 and @xmath683 corresponds to the right hand side of ( [ hmequation ] ) .
then we check ( [ mcontii ] ) for @xmath738 .
* step 2 * : we claim that there exist @xmath739 and @xmath90 such that if @xmath740 and @xmath741 then there exists @xmath742 so that @xmath743 for all @xmath744 .
moreover @xmath745 is cauchy in @xmath746\times\bar{\omega}\times\mathbb{r}^3)$ ] .
+ first we will show a boundedness ( [ boundedh - m ] ) for all @xmath744 .
we use mathematical induction on @xmath215 .
assume @xmath747 where @xmath728 will be determined later . integrating ( [ hm ] ) along the trajectory , we have @xmath748 and @xmath749 where we choose @xmath750 and then @xmath751 and then @xmath752 .
+ newt we will show the sequence @xmath218 is cauchy in @xmath746\times\bar{\omega}\times\mathbb{r}^3)$ ] .
the equation of @xmath753 is @xmath754 where @xmath755 from ( [ wk ] ) and ( [ gamma1 ] ) , we have a bound of @xmath756 , @xmath757 integrating ( [ hm1 ] ) along the trajectory , we have @xmath758 if we choose @xmath759 and @xmath760 then @xmath761 then we have @xmath762 which means that the sequence @xmath218 is cauchy in @xmath746\times\bar{\omega}\times\mathbb{r}^3)$ ] . *
step 3 * : from previous steps we obtain that @xmath514 with @xmath763 is continuous function on @xmath764 .
now we claim that @xmath514 is continuous in @xmath71 .
notice that @xmath765 only depends on @xmath766 and @xmath767 .
using unform bound of @xmath768 ( theorem 1 of @xcite ) we can obtain the continuity for @xmath514 for all time by repeating @xmath186,[t,2t], ... $ ] .
if the boundary @xmath24 does not include a line segment ( definition [ linesegment ] ) then every step is valid with @xmath769 instead of @xmath71 and @xmath186\times\{\bar{\omega}\times\mathbb{r}^3\}\backslash\mathfrak{d}$ ] instead of @xmath764 .
* proof of 1 of theorem [ propagation ] * * proof of ( [ 1sidedinequality ] ) * + in order to show the upper bound of discontinuity jump ( [ 1sidedinequality ] ) , we will show @xmath770_{x_0 + ( t - t_0)v_0,v_0 } \leq [ h]_{t_0,x_0,v_0 } e^{-(\frac{1}{c_{\nu}}+ \frac{c^{\prime}}{c_w}||h_0||_{\infty})(1+|v_0|)^{\gamma}(t - t_0)},\ ] ] when @xmath88 and @xmath771 .
choose two points @xmath772 and compare the representation @xmath773 it is easy to see that if @xmath774 then as @xmath775 we have @xmath776 and if @xmath777 then as @xmath775 we have @xmath778 therefore the first four lines converge to @xmath779_{t_0,x_0,v_0 } \times e^{-\nu(v_0)(t - t_0)-\int_{t_0}^t \nu(\sqrt{\mu}\frac{h}{w})(\tau , x_0 -(t_0-\tau)v_0,v_0)d\tau}$ ] .
for the last two lines , using the continuity of @xmath780 we conclude that it converges to zero .
therefore we have @xmath781_{x_0 + ( t - t_0)v_0,v_0 } & \leq & [ h]_{t_0,x_0,v_0 } e^{-\nu(v_0)(t - t_0)-\int_{t_0}^t \nu(\sqrt{\mu}\frac{h}{w})(\tau , x_0 -(t_0 -\tau)v_0 , v_0 ) d\tau}\\ & \leq & [ h]_{t_0,x_0,v_0 } \times e^{-(\frac{1}{c_{\nu}}-c_w c^{\prime}||h_0||_{\infty})(1+|v_0|)^{\gamma } ( t - t_0)},\end{aligned}\ ] ] where we used @xmath782 remark that * proof of ( [ 1sidedinequality ] ) * is valid for in - flow , diffuse and bounce - back cases . * * proof of ( [ decaydiscontinuity ] ) * * + assume @xmath783_{x_0,v_0}\neq 0 $ ] and @xmath784 with @xmath88 .
further assume that the boundary @xmath24 is strictly concave at @xmath117 along @xmath118 direction ( [ convexity ] ) .
+ * step 1 * claim : we can choose sequences @xmath785 such that @xmath786_{x_0,v_0}\neq 0 $ ] . + from @xmath783_{x_0,v_0}\neq 0 $ ] we may assume @xmath787_{x_0,v_0}\neq 0,\ ] ] for all @xmath788 . and for each @xmath664 we can choose desired sequences .
+ + * step 2 * claim : for given @xmath273 , up to subsequence we may assume that @xmath789 we remark that a continuity @xmath790 on @xmath791 , i.e. @xmath792_{t_0,x_0,v_0 } = w(v_0 ) [ \ g |_{[0,\infty)\times\gamma_-}]_{t_0,x_0,v_0}=0 \ \ \ \text{for all } \
( t_0,x_0,v_0)\in[0,\infty)\times\{\gamma_- \cup \gamma_0^{\mathbf{s}}\ } \label{contg}\end{aligned}\ ] ] is crucially used in this step . in order to show the final goal ( [ twosequences ] ) of this step , we need to prove following statement .
@xmath793 we will prove ( [ canreachthere ] ) later and show ( [ twosequences ] ) using ( [ canreachthere ] ) .
it suffices to show that there are only finite @xmath664 such that @xmath794 suppose there are infinitely many @xmath795 satisfying ( [ sequence1 ] ) .
if @xmath273 is sufficiently small then ( [ sequence1 ] ) implies that @xmath796 and @xmath797 .
the boltzmann solution @xmath514 at @xmath798 is @xmath799 and a similar representation for @xmath800 .
compare representations of @xmath801 and @xmath800 to conclude @xmath802_{t_0 -t_{\mathbf{b}}(x_0,v_0),x_{\mathbf{b}}(x_0,v_0),v_0 } \times e^{-\nu(v_0)(t_0 -t_{\mathbf{b}}(x_0,v_0))-\int_{t_0 -t_{\mathbf{b}}(x_0,v_0)}^{t_0 } \nu(\sqrt{\mu}\frac{h}{w})(\tau , x_0 -(t_0-\tau)v_0,v_0)d\tau } , \label{compareh}\end{aligned}\ ] ] where we used the continuity of @xmath803 and @xmath804 . further using the in - flow boundary condition @xmath805
, we have @xmath806_{t_0 , x_0,v_0 } w(v_0 ) e^{-\nu(v_0)(t_0 -t_{\mathbf{b}}(x_0,v_0))-\int_{t_0 -t_{\mathbf{b}}(x_0,v_0)}^{t_0 } \nu(\sqrt{\mu}\frac{h}{w})(\tau , x_0 -(t_0-\tau)v_0,v_0)d\tau } = 0,\label{continuityuptogamma0}\end{aligned}\ ] ] where we used the continuity of @xmath682 on @xmath807 , ( [ contg ] ) at the last line .
this is contradict because we choose the sequences @xmath808 satisfying @xmath809_{x_0,v_0}\neq 0 $ ] in * step 1*. + now suppose there are infinitely many @xmath810 satisfying ( [ sequence2 ] ) . because of ( [ canreachthere ] ) we have @xmath811 and @xmath812 .
the boltzmann solution @xmath514 at @xmath813 is @xmath814 and same representation for @xmath815 . using the continuity of @xmath656
we see that @xmath816 which is also contradiction .
+ now we prove ( [ canreachthere ] ) .
we can choose @xmath273 sufficiently small so that @xmath817 .
from @xmath818 we know that a line segment between @xmath117 and @xmath819 has only one intersection point @xmath117 with @xmath24 , i.e.@xmath820 .
furthermore we can choose @xmath524 so large that @xmath821}b(x_0-sv_0;\varrho ) \cap
\partial\omega \subset b(x_0;\varepsilon)$ ] .
choose @xmath822 sufficiently large so that @xmath823}b(x_0-sv_0;\varrho ) $ ] for all @xmath824 .
if @xmath825 then @xmath826 and this implies @xmath827 .
+ + * step 3 * claim : choose @xmath165 so that @xmath828 and denote @xmath829 .
then there exists @xmath822 so that @xmath830 for all @xmath831 .
+ using ( [ canreachthere ] ) we only have to prove @xmath832 . from ( [ twosequences ] ) we know that @xmath833 .
we assume that @xmath834 and @xmath835 and @xmath836 .
let s define @xmath837 since @xmath838 we have @xmath839 and @xmath840 .
because of the strict concavity along @xmath118 direction at @xmath117 ( [ convexity ] ) , for sufficiently large @xmath275 so that @xmath841 we have @xmath842 where the hessian of @xmath241 is evaluated at @xmath843 . since @xmath844 we have @xmath845 for @xmath846 therefore @xmath847 and @xmath848 .
this is contradiction because @xmath849 the consequence of this step is that for @xmath850 we have a representation of @xmath514 at @xmath31 @xmath851 + * step 4 * claim : for given @xmath273 there exists @xmath90 so that if @xmath852 and @xmath853 and @xmath854 and @xmath855 then @xmath856 we have @xmath857 @xmath858 and similarly @xmath859 @xmath860 let s compare the arguments of two representations : @xmath861 and @xmath862 we can choose desired @xmath273 to conclude ( [ unifconti ] ) .
+ + * step 5 * claim : choose @xmath165 so that @xmath863 and denote @xmath829 .
let @xmath864_{x_0,v_0 } $ ] and @xmath865 be chosen in * step 4*. then we can choose @xmath866 so that @xmath867 and @xmath868 and @xmath869 .
+ if there are infinitely many @xmath870 so that @xmath871 and @xmath872 then up to subsequence we can define @xmath873 .
therefore we may assume @xmath874 for all @xmath664 .
we assume that @xmath875 and @xmath835 and @xmath836 .
now we illustrate how to choose such a @xmath870 .
denote @xmath876 and @xmath877 .
first we will choose @xmath878 and @xmath879 so that @xmath880 and @xmath881 .
the condition ( [ grazing ] ) implies that @xmath882 in order to use the implicit function theorem we define @xmath883 and compute , using ( [ convexity ] ) @xmath884 for @xmath267 , @xmath268 and the hessian is evaluated at @xmath885 .
hence @xmath886 is a smooth function near @xmath267 and @xmath887 . in order to study the behavior of @xmath888 we use the taylor s expansion : from @xmath889 we have @xmath890 where the hessian @xmath891 is evaluated at @xmath892 and the hessian @xmath893 is evaluated at @xmath894 with @xmath895 .
for @xmath267 and @xmath896 we know that the right hand side of the above equation converges to @xmath897 hence we have a control of @xmath888 , i.e @xmath898 from ( [ tangent ] ) , @xmath899 equals @xmath900 where @xmath901 for some @xmath902 . using the smoothness of @xmath241 we can bound ( [ dd1 ] ) as @xmath903 to sum for fixed @xmath16 and direction
@xmath904 we can choose @xmath905 such that @xmath906 and @xmath905 is controlled by ( [ ubounded ] ) .
finally we choose @xmath907 and find the corresponding @xmath905 so that @xmath908 .
define @xmath909 .
then we have desired @xmath870 for sufficiently large @xmath664
. + + * step 6 * to sum for @xmath910 we have @xmath872 and @xmath871 and @xmath911_{x_0,v_0}$ ] .
hence the representation of the boltzmann solution @xmath514 at @xmath910 is given by @xmath912 using ( [ htxv ] ) we have @xmath913_{x_0,v_0}e^{-\nu(v_0)(t - t_0)-\int_{t_0}^t \nu(\sqrt{\mu}\frac{h}{w})(\tau , x_0 + ( \tau - t_0)v_0,v_0)d\tau},\end{aligned}\ ] ] which implies that @xmath770_{x , v } \geq \frac{1}{4}[h(t_0)]_{x_0,v_0 } \times e^{-(c_{\mu } + c^{\prime}c_w ||h_0||_{\infty})(1+|v|)^{\gamma}(t - t_0 ) } \neq 0.\ ] ] * remark * _ through step 1 to step 6 , we only used the in - flow boundary datum @xmath682 explicitely in step 2 .
all the other steps are valid for diffuse and bounce - back boundary condition cases . in step 2 , we only used ( [ contg ] ) the continuity of @xmath914 on @xmath791 . therefore , if we can show the continuity of @xmath98 on @xmath791 then we can prove ( [ decaydiscontinuity ] ) . for diffuse and bounce - back boundary
we will prove such a continuity to conclude ( [ decaydiscontinuity ] ) . _
in this section , we consider the linearized boltzmann equation ( [ hbol ] ) with the diffuse boundary condition ( [ diffuse ] ) . in spite of the averaging effect of diffuse reflection operator , we can observe the formation and propagation of discontinuity .
continuity away from @xmath61 is also established .
we prove part 2 of theorem [ formationofsingularity ] .
the idea of proof is similar to in - flow case but we also use @xmath915 not only @xmath916 as a parameter . without loss of generality we may assume @xmath635 and @xmath917 and @xmath88 . locally the boundary is a graph , i.e. @xmath637 and @xmath640 for @xmath918 .
( see figure 3 ) assume that @xmath919 is sufficiently small so that the global solution @xmath514 of ( [ hbol ] ) with diffuse boundary ( [ diffuse ] ) has a uniform bound ( [ boundedness ] ) , from theorem 4 of @xcite .
choose @xmath920 sufficiently small and @xmath921 sufficiently large so that @xmath922 where @xmath923 and @xmath646 and @xmath647 from ( [ wk ] ) and ( [ gamma1 ] ) .
more precisely , first choose @xmath921 large enough to have @xmath924 then choose @xmath925 as @xmath926 assume the condition for initial datum @xmath656 : there is sufficiently small @xmath927 such that @xmath928 and @xmath929 we claim that the boltzmann solution @xmath514 with such initial datum @xmath656 is not continuous at @xmath930 .
we will use a contradiction argument : assume the boltzmann solution @xmath514 is continuous at @xmath99 , i.e ( [ contradiction ] ) is valid . choose sequences of points @xmath931 and @xmath932 .
because of our choice , for sufficiently large @xmath664 , we have @xmath933 hence the boltzmann solution at @xmath668 is @xmath934 using the diffuse boundary condition ( [ diffuse ] ) , the boltzmann solution at @xmath935 is @xmath936 using a pointwise boundedness ( [ boundedness ] ) of @xmath514 , and @xmath937 , we can estimate @xmath938 which is contradiction to ( [ contradiction ] ) . instead of using the argument of @xcite to show continuity in the case of diffuse reflection boundary condition we will use the sequence ( [ hm ] ) with the boundary condition ( [ iterationdiffuse ] ) and lemma [ transportequation ] . this argument
also gives a new proof of the continuity of boltzmann solution in a strictly convex domain with simpler way than @xcite .
* proof of 2 of theorem [ continuityawayfromd ] * we will use the sequence ( [ hm ] ) with @xmath722 with following boundary condition @xmath939 with @xmath940 .
* step 1 * : we claim that @xmath941 is continuous function on @xmath186\times\gamma$ ] even if @xmath942\times\bar{\omega}\times\mathbb{r}^3)$ ] is only continuous on @xmath186\times\bar{\omega}\times\mathbb{r}^3 \backslash \mathfrak{g}$ ] .
we will show as @xmath706 , @xmath943 using the fact @xmath944 as @xmath945 and the exponentially decay weight function of @xmath946 it suffices to show that @xmath947 for sufficiently large @xmath948 . using lemma [ uniformlycontinuous ] we can choose open set @xmath949 so that @xmath950 is small and @xmath216 is uniformly continuous on @xmath951 .
therefore we can make @xmath952 small using the smallness of @xmath453 and make @xmath953 small using the uniformly continuity of @xmath216 on @xmath951 .
hence ( [ 136 ] ) is valid . *
step 2 * : we claim @xmath954 for all @xmath727 where @xmath764 is defined in ( [ ct ] ) . by induction
choose @xmath730 and ( [ mconti ] ) is satisfied for @xmath731 .
assume ( [ mconti ] ) for all @xmath732 .
let @xmath955 .
then the equation of @xmath217 is @xmath956 from theorem [ continuity ] and corollary [ continuity1 ] we know that @xmath957 and @xmath736 are both continuous in @xmath186\times\omega\times\mathbb{r}^3 $ ] . because of step 1 we know that @xmath958 is also continuous function on @xmath186\times\gamma$ ] .
so we can apply lemma [ transportequation ] to conclude ( [ mconti ] ) is valid for @xmath738 . * step 3 * : we claim @xmath216 is a cauchy sequence in @xmath764 for some small @xmath728 .
first we will compute some constants explicitly . from ( [ normalizedconstant ] )
the normalized constant @xmath959 is @xmath960^{-1}$ ] .
choose @xmath961 and then we can compute the right hand side of above term : @xmath962_{0}^{\infty } = 2\pi.\end{aligned}\ ] ] therefore we have @xmath963 .
next we will show @xmath964 where @xmath965 .
we follow the computation of lemma 25 in @xcite . for @xmath966 , in the case of @xmath967 we can see that @xmath968 has a maximum value at @xmath969 which is @xmath970 and @xmath971 where @xmath972 and combining with ( [ 1111 ] ) we conclude ( [ coefficient ] ) .
first we will show a boundedness ( [ boundedh - m ] ) .
let @xmath216 be a solution of ( [ hm ] ) with @xmath973 and the boundary condition ( [ iterationdiffuse ] ) .
then there exist @xmath974 such that if @xmath975 then @xmath976 we will use mathematical induction .
choose @xmath977 and assume @xmath978 and @xmath979 for @xmath732 , where @xmath980 will be determined later .
from lemma 24 of @xcite the representation of @xmath217 which is a solution of ( [ hm ] ) with the boundary condition ( [ iterationdiffuse ] ) is given by @xmath981 } + \underbrace{\int_0^t e^{-\nu(v)(t - s)}q^m ( s , x-(t - s)v , v ) ds}_{\mathbf{i } } \big\}\label{hm11}\\ & + & \mathbf{1}_{0<t_1}(t , x , v)\big\ { \underbrace{\int_{t_1}^{t } e^{-\nu(v)(t - s)}q^m ( s , x-(t - s)v , v ) ds}_{\mathbf{ii } } + \frac{e^{-\nu(v)(t - t_1)}}{\tilde{w}(v ) } \int_{\prod_{j=1}^{k}\mathcal{v}_j}h \big\},\label{hm12}\end{aligned}\ ] ] where @xmath982 was defined ( [ qmm ] ) and @xmath983 } d\sigma_l(0 ) + \sum_{l=1}^{k } \underbrace{\int_0^{t_l } \mathbf{1}_{t_{l+1}\leq 0 < t_l } q^{m - l}(s , x_l
-(t_l -s)v_l , v_l ) d\sigma_l(s ) ds}_{\mathbf{iii}}\label{hm13}\\ + \sum_{l=1}^{k } \underbrace{\int_{t_{l+1}}^{t_l } \mathbf{1}_{0<t_{l+1 } } q^{m - l}(s , x_l-(t_l - s)v_l , v_l ) d\sigma_l(s ) ds}_{\mathbf{iv } } + \underbrace{\mathbf{1}_{0 < t_{k+1 } } h^{m - k+1}(t_{k+1},x_{k+1},v_{k } ) d\sigma_{k}(t_{k+1})}_{[\mathrm{many \ bounces}]}.\label{hm14}\end{aligned}\ ] ] here @xmath984 is evaluated at @xmath985 of @xmath986 + first we can estimate @xmath987 $ ] in ( [ hm11 ] ) and ( [ hm13 ] ) , @xmath988 where we used ( [ coefficient ] ) . + next we estimate @xmath989 $ ] term in ( [ hm14 ] ) which is crucial estimate in this proof .
we use lemma 23 in @xcite to bound a contribution of @xmath989 $ ] term in ( [ hm14 ] ) in the last term of ( [ hm12 ] ) by @xmath990 where we used ( [ coefficient ] ) at the last step .
the remainders * i * , * ii * , * iii * and * iv * are contributions of @xmath991 .
we introduce a notation @xmath992 where the above inequality holds for @xmath993 and @xmath994 and @xmath995 where we used the induction hypothesis ( [ inductionhypothesis ] ) for ( [ inequalityh ] ) and ( [ inequalityh2 ] ) .
easily we have @xmath996 to summarize , we can estimate all terms of representation of @xmath997 in ( [ hm11 ] ) to obtain @xmath998\\ & & \ \ \ \ \ \ \ \ \ \ \ + 1 + \tilde{c}_{\beta } \rho^{2\beta-4}\big\ } + c_{\gamma}c||h_0||_{\infty } \sup_{0\leq s\leq t _ * } ||h^{m+1}(s)||_{\infty}.\end{aligned}\ ] ] choose @xmath999 .
choose @xmath1000 sufficiently large so that @xmath1001 and then choose @xmath1002 sufficiently small so that @xmath1003 and then choose @xmath739 sufficiently large @xmath1004 and choose @xmath1005 . finally assume @xmath1006
. then we have @xmath1007 \big\}\\ & \leq & \frac{20}{19}||h_0||_{\infty } \left\ { \frac{c}{10 } + c\big [ \frac{1}{30 } + \frac{1}{30 } + \frac{1}{20 } + \frac{1}{15 } \big ] \right\}\leq c||h_0||_{\infty}.\end{aligned}\ ] ] + next we will show that @xmath216 is a cauchy sequence in @xmath219 .
let @xmath216 be a solution of ( [ hm ] ) with @xmath1008 and the boundary condition ( [ iterationdiffuse ] ) .
then there exist @xmath974 so that if @xmath975 then @xmath216 is cauchy in @xmath1009\times\bar{\omega}\times\mathbb{r}^3)$ ] . the equation of @xmath753 is @xmath1010 where @xmath1011 is defined at ( [ tildeqm ] ) . from lemma 24 of @xcite we have the representation @xmath1012 where @xmath1013}}.\end{aligned}\ ] ] first using lemma 24 of @xcite , we estimate @xmath1014}$ ] term for sufficiently large @xmath1015 by @xmath1016 easily we have @xmath1017 where @xmath1018 with @xmath1019 . to summarize
, we can estimate all terms of representation of @xmath1020 in ( [ hm+1 ] ) for any @xmath1021 to obtain @xmath1022 chosen later .
choose @xmath1000 sufficiently large so that @xmath1023 and then choose @xmath1024 so small that @xmath1025 and @xmath1026 .
then we have @xmath1027 using ( [ termbyterm ] ) for @xmath1028 so that @xmath1029 and @xmath1030it is easy to show @xmath1031 we apply the above inequality term by term in ( [ termbyterm ] ) to have @xmath1032 now we estimate @xmath1033 - 1}\ { \sup_{0\leq s\leq t}||h^{2k}-h^{2k-1}||_{\infty } + ... + \sup_{0\leq s\leq t}||h^{1}-h^{0}||_{\infty } \}\end{aligned}\ ] ] @xmath1034 - 1 } \sum_{l=0}^{m - n-1 } \{(1+\tilde{\tau})^k -1\}^{\left[\frac{m - l-1}{k}\right]-\left[\frac{n}{k}\right]}\ { \sup_{0\leq s\leq t}||h^{2k}-h^{2k-1}||_{\infty } + ... + \sup_{0\leq s\leq t}||h^{1}-h^{0}||_{\infty}\}\\ & \leq & \tilde{\tau } \{(1+\tilde{\tau})^k -1\}^{\left[\frac{n}{k}\right]-1 } \frac{1}{2-(1+\tilde{\tau})^k } \ { \sup_{0\leq s\leq t}||h^{2k}-h^{2k-1}||_{\infty } + ... + \sup_{0\leq s\leq t}||h^{1}-h^{0}||_{\infty}\},\end{aligned}\ ] ] where we choose @xmath1035 - 1 $ ] so that @xmath1036 . if @xmath1037 is chosen sufficiently small so that @xmath1038 then @xmath1039 - 1 } \rightarrow 0 $ ] as @xmath176 which implies that @xmath1040 as @xmath1041
. thus @xmath216 is cauchy in @xmath219 .
* step 4 * : we claim that @xmath514 is continuous in @xmath71 . notice that @xmath765 only depends on @xmath766 and @xmath767 ( theorem 1 of @xcite ) . using a unform bound of @xmath768
, we can obtain the continuity of @xmath514 for all time by repeating @xmath186,[t,2t], ... $ ] .
if the boundary @xmath24 does not include a line segment ( [ linesegment ] ) then every step is valid with @xmath769 instead of @xmath71 and @xmath186\times\{\bar{\omega}\times\mathbb{r}^3\}\backslash\mathfrak{d}$ ] instead of @xmath764 .
* proof of 2 of theorem [ propagation ] * + * proof of ( [ 1sidedinequality ] ) * : the proof is exactly same as in - flow case in section 4.3 .
+ * proof of ( [ decaydiscontinuity ] ) * the proof is exactly same as the proof of in - flow case in section 4.3 except * step 2*. as we mentioned in remark of * step 2 * , we need to show a continuity of a boundary datum on @xmath1042 . in diffuse
reflection boundary condition case , we need @xmath1043_{t , y , v } = \lim_{\delta\downarrow 0 } \sup_{\begin{scriptsize}\begin{array}{ccc } t^{\prime},t^{\prime\prime}\in b(t;\delta)\\(y^{\prime},v^{\prime } ) , ( y^{\prime\prime},v^{\prime\prime})\in \gamma_- \cap b((y , v);\delta ) \backslash ( y , v ) \end{array}\end{scriptsize } } & = & \lim_{\delta\downarrow 0 } \sup_{\begin{scriptsize}\begin{array}{ccc } t^{\prime},t^{\prime\prime}\in b(t;\delta)\\(y^{\prime},v^{\prime } ) , ( y^{\prime\prime},v^{\prime\prime})\in \gamma_- \cap b((y , v);\delta ) \backslash ( y , v ) \end{array}\end{scriptsize } } \left|\frac{1}{\tilde{w}(v^{\prime})}\int_{\mathcal{v}(y^{\prime } ) } h(t^{\prime},y^{\prime},\mathfrak{v } ) \tilde{w}(\mathfrak{v } ) d\sigma(\mathfrak{v})- \frac{1}{\tilde{w}(v^{\prime\prime})}\int_{\mathcal{v}(y^{\prime\prime } ) } h(t^{\prime\prime},y^{\prime\prime},\mathfrak{v } ) \tilde{w}(\mathfrak{v } ) d\sigma(\mathfrak{v})\right|\end{aligned}\ ] ] for @xmath1044 .
this is already proven in section 5.2 continuity away from @xmath61 .
in this section , we consider the linear boltzmann equation ( [ hbol ] ) with the bounce - back boundary condition ( [ bounceback ] ) .
we prove part 3 of theorem [ formationofsingularity ] . without loss of generality
we may assume @xmath635 and @xmath636 and @xmath88 .
locally the boundary is a graph , i.e. @xmath637 . the condition @xmath88 implies @xmath638 and @xmath639 which means @xmath640 for @xmath918 .
( see figure 3 ) recall the constants @xmath646 and @xmath647 from ( [ wk ] ) and ( [ gamma1 ] ) .
choose @xmath1045 sufficiently small so that @xmath1046 assume a condition for the initial datum @xmath656 : there is sufficiently small @xmath657 such that + @xmath1047 and @xmath1048 we will use a contradiction argument : assume the boltzmann solution @xmath514 is continuous at @xmath99 , i.e. ( [ contradiction ] ) is valid .
choose sequences of points @xmath662 and @xmath1049 .
because of our choice , for sufficiently large @xmath664 , we have @xmath1050 hence the boltzmann solution at @xmath668 and @xmath1051 is @xmath1052 using a pointwise boundedness ( [ boundedness ] ) of @xmath514 with ( [ wk ] ) and ( [ gamma1 ] ) , we have @xmath1053 therefore using ( [ 1/2b ] ) , @xmath1054 which is contradiction to ( [ contradiction ] ) .
@xcite [ bouncebackformula]let @xmath1068 and @xmath76 with @xmath684\times\omega}|\phi(\cdot,\cdot , v)| < \infty$ ]
. there exists a unique solution @xmath1069 of @xmath1070with the bounce - back reflection @xmath1071 for @xmath1072 for almost any @xmath1073 @xmath1074where @xmath1075 and @xmath1076 in ( [ bouncebackcycle ] ) .
[ bbtransportequation ] let @xmath12 be an open subset of @xmath86 with a smooth boundary @xmath24 and an initial datum @xmath1077 be continuous in @xmath1078 .
also assume @xmath683 and @xmath76 be continuous in the interior of @xmath186\times\omega\times\mathbb{r}^3 $ ] and @xmath684\times\omega\times\mathbb{r}^3 } |q(t , x , v)| < \infty$ ] and @xmath684\times\omega}|\phi(\cdot,\cdot , v)| < \infty$ ] for all @xmath338 .
let @xmath685 be the solution of @xmath1079 assume the compatibility condition on @xmath72 @xmath1080 then the boltzmann solution @xmath685 is continuous on @xmath148 .
further , if the boundary @xmath24 does not include a line segment ( [ linesegment ] ) then @xmath685 is continuous on a complementary set of the discontinuity set , i.e. @xmath186\times \{\bar{\omega}\times\mathbb{r}^3\}\backslash \mathfrak{d}_{bb}$ ] .
the proof is similar to the proof of lemma 16 of @xcite .
take any point @xmath1081\times\bar{\omega}\times\mathbb{r}^3 $ ] and recall its back - time cycle and ( [ bouncebackformular ] ) .
assume @xmath1082 . using ( [ bouncebackformular ] )
, @xmath685 takes the form @xmath1083 take any point @xmath1084 . by the definition of @xmath148 we assume that @xmath230 or @xmath1085 and we can separate three cases : @xmath1086 with @xmath1087 and @xmath1088 with @xmath1089 . + simply we have @xmath1090 and use the continuity of @xmath683 and @xmath76 to conclude the continuity of @xmath685 .
+ a representation of @xmath685 takes the form @xmath1091 thanks to lemma [ huang ] and lemma [ tbcon ] , the condition @xmath712 implies continuity of @xmath1092 .
therefore we can show the continuity of @xmath685 .
+ we have ( [ hmmm ] ) for @xmath685 . thanks to ( [ bouncebackx ] ) and ( [ tk1 ] ) and lemma [ huang ] and lemma [ tbcon ]
, the conditions @xmath1088 and @xmath1089 imply continuity of @xmath1093 .
therefore we can show the continuity of @xmath685 .
+ * proof of part 1 of theorem 3 * + following the in - flow and diffuse cases , we use the iteration scheme ( [ hm ] ) which is equivalent to ( [ hmequation ] ) with bounce - back boundary condition @xmath1094 and an initial condition @xmath722 .
* step 1 * : we claim that @xmath1095 is a continuous function in @xmath1096 for all @xmath727 and for any @xmath728 where @xmath1097\times\bar{\omega}\times\mathbb{r}^3\}$ ] .
choose @xmath1098 and use mathematical induction .
assume @xmath1095 is continuous @xmath1096 for @xmath732 .
apply lemma [ bbtransportequation ] to concluse that @xmath217 is continuous in @xmath1096 . * step 2 * : we claim that there exist @xmath739 and @xmath90 such that if @xmath741 then there exists @xmath1099 so that @xmath1100 and @xmath745 is cauchy in @xmath746\times\bar{\omega}\times\mathbb{r}^3)$ ] .
+ fisrt we will show the boundedness using mathematical induction . assume @xmath1101 where @xmath728 will be chosen later .
applying lemma [ bouncebackformula ] , @xmath80 and @xmath716 correspond with @xmath1102 and the right hand side of ( [ hm ] ) respectively to have a representation of @xmath997 @xmath1103 where @xmath1104=\left[x_{\mathbf{cl}}(s;t , x , v),v_{\mathbf{cl}}(s;t , x , v)\right]$ ] is in ( [ bouncebackcycle ] ) .
the above term is bounded by @xmath1105 where the constants are coming from basic estimates , ( [ wk ] ) and ( [ gamma1 ] ) .
choose @xmath750 and @xmath1106 and @xmath1107 .
then we have @xmath1108 .
+ next we will show @xmath745 is cauchy in @xmath746\times\bar{\omega}\times\mathbb{r}^3)$ ] .
recall @xmath1109 from ( [ tildeqm ] ) .
the equation of @xmath753 is ( [ hm1 ] ) with a zero initial condition @xmath1110 and a bounce - back boundary condition @xmath1111 . applying lemma [ bouncebackformula ] to ( [ hm1 ] ) we have @xmath1112 where @xmath1104=\left[x_{\mathbf{cl}}(s;t , x , v),v_{\mathbf{cl}}(s;t , x , v)\right]$ ] is in ( [ bouncebackcycle ] ) .
then we have exactly same estimates of in - flow case to conclude @xmath218 is cauchy .
* step 3 * : same argument as in - flow case but substitute @xmath1113 for @xmath1114 respectively .
* proof of 2 of theorem [ propagation ] * + * proof of ( [ 1sidedinequality ] ) * : the proof is exactly same as in - flow case in section 4.3 .
+ * proof of ( [ decaydiscontinuity ] ) * recall that we have @xmath783_{x_0,v_0}\neq 0 $ ] for @xmath88 and @xmath115 .
the proof is exactly same as the proof of in - flow case in section 4.3 except * step 2*. we need to show a continuity of a boundary datum on @xmath1042 . in bounce - back reflection
boundary condition case , we need to show @xmath1115_{t_0,x_0,v_0 } = \lim_{\delta\downarrow 0 } \sup_{\begin{scriptsize}\begin{array}{ccc } t^{\prime},t^{\prime\prime}\in b(t;\delta)\\(y^{\prime},v^{\prime } ) , ( y^{\prime\prime},v^{\prime\prime})\in \gamma_- \cap b((x_0,v_0);\delta ) \backslash ( x_0,v_0 ) \end{array}\end{scriptsize } } because @xmath1116 is in the incoming boundary @xmath170 , using the bounce - back boundary condition , we have @xmath1117 .
further due to the condition @xmath1118 we have @xmath1119 and @xmath1120 and similar representation for @xmath1121 . using the continuity of @xmath1122 and @xmath1123 we have @xmath1124_{t_0,x_0,v_0 } & = & \lim_{\delta\downarrow 0 } \sup_{\begin{scriptsize}\begin{array}{ccc } t^{\prime},t^{\prime\prime}\in b(t;\delta)\\(y^{\prime},v^{\prime } ) , ( y^{\prime\prime},v^{\prime\prime})\in \gamma_- \cap b((x_0,v_0);\delta ) \backslash ( x_0,v_0 ) \end{array}\end{scriptsize } } & & \times e^{-\nu(v_0)t_0 -\int_0^{t_0}\nu(\sqrt{\mu}\frac{h}{w})(\tau , x_0+(t_0-\tau)v_0,v_0)d\tau},\end{aligned}\ ] ] where we used the continuity of the initial datum @xmath656 in the last equality .
+ + * acknowledgements . *
the author is indebted to his advisor yan guo for helpful discussions .
the research is supported in part by frg no .
524230 .
99 aoki , k. : private communications .
arlotti , l. ; banasiak , j. ; lods , b. : on general transport equations with abstract boundary conditions .
the case of divergence free force field , _ preprint _ 2009 .
aoki , k. ; bardos , c. ; dogbe , c. ; golse , f. : a note on the propagation of boundary induced discontinuities in kinetic theory , _ math . models methods appl .
_ 11 ( 2001 ) , no . 9 , 15811595 .
aoki , k. ; takata , s. ; aikawa , h. ; golse , f. : a rarefied gas flow caused by a discontinuous wall temperature , _ phys . fluids _ 13 ( 2001 ) , no .
9 , 26452661 boudin , l. ; desvillettes , l. : on the singularities of the global small solutions of the full boltzmann equation , _ monatshefte math .
_ 131 ( 2000 ) , 91108 .
bernis , l. ; desvillettes , l. : propagation of singularities for classical solutions of the vlasov - poisson - boltzmann equation , _ discrete contin .
dyn . syst .
_ 24 ( 2009 ) , no .
1 , 1333 .
cercignani , c. : propagation phenomena in classical and relativistic rarefied gases , _ transport theory statist .
_ 29 ( 2000 ) , no . 3 - 5 , 607614 .
duan , r. ; li , m .-
r . ; yang , t. : propagation of singularities in the solutions to the boltzmann equation near equilibrium , _ math . models methods appl .
_ 18 ( 2008 ) , no . 7 , 10931114 .
greenberg , w. ; van der mee , c. ; protopopescu , v. : boundary value problems in abstract kinetic theory . operator theory : advances and applications , 23 .
birkhauser verlag , basel , 1987 .
guiraud , j .-
: an @xmath1125 theorem for a gas of rigid spheres in a bounded domain , theories cinetiques classiques et relativistes , pp . 2958 .
centre nat .
recherche sci . , paris , 1975 guo , y. : singular solutions of the vlasov - maxwell system on a half line , _ arch .
rational mech .
anal . _ 131 ( 1995 ) ,
3 , 241 - 304 .
guo , y. : classical solutions to the boltzmann equation for molecules with an angular cutoff , _ arch .
anal . _ 169 ( 2003 ) ,
4 , 305 - 353 .
guo , y. : decay and continuity of boltzmann equation in bounded domains , to appear in _ arch . rat .
_ grad , h. : asymptotic theory of the boltzmann equation .
rarefied gas dynamics.in : proceedings of the 3rd international symposium , pp.26 - 59 , paris , 1962 hwang , h - j : regularity for the vlasov - poisson system in a convex domain , _
siam j. math .
_ , 36 ( 2004 ) , no.1 , 121171 hwang , h - j ; velazquez j. : global existence for the vlasov - poisson system in bounded domains , to appear in _ arch .
_ kim , c. : boltzmann equation in a bounded domain : specular reflection with 2-d general geometry , _ in preparation_. lions , p .-
compactness in boltzmann s equation via fourier integral operators and applications .
i , ii , iii .
_ j. math .
kyoto univ .
_ 34 ( 1994 ) , no .
2 , 391427 , 429461 , 539584 .
mischler , s. : on the initial boundary value problem for the vlasov - poisson - boltzmann system .
_ 210 ( 2000 ) , no .
2 , 447466 .
sone , y. : molecular gas dynamics .
theory , techniques , and applications .
modeling and simulation in science , engineering and technology .
birkhauser boston , inc .
, boston , ma , 2007 .
sone , y. ; takata , s. : discontinuity of the velocity distribution function in a rarefied gas around a convex body and the s layer at the bottom of the knudsen layer , _ transport theor .
_ 21 ( 1992 ) , 501530 .
takata , s. ; sone , y. ; aoki , k. : numerical analysis of a uniform flow of a rarefied gas past a sphere on the basis of the boltzmann equation for hard - sphere molecules , _ phys .
fluids _ * a 5 * ( 1993 ) , 716737 .
ukai , s. ; solutions of the boltzmann equation .
patterns and waves , 3796 , stud .
appl . , 18 , north - holland , amsterdam , 1986 villani , c. ; a review of mathematical topics in collisional kinetic theory . handbook of mathematical fluid dynamics , vol .
i , 71305 , north - holland , amsterdam , 2002 .
voigt , j. : functional analytic treatment of the initial boundary value problem for collisionless gases , habilitationsschrift , munchen , 1981 ( http://www.math.tu-dresden.de/ voigt / vopubl / habilschr / habil80.pdf ) wennberg , b. : regularity in the boltzmann equation and the radon transform , _ commun .
in p.d.e . _ 19(1994 ) , 2057 - 2074 .
wennberg , b. : the geometry of binary collisions and generalized radon transforms , _ arch .
rational mech .
_ 139 ( 1997 ) , no .
3 , 291302 . | the formation and propagation of singularities for boltzmann equation in bounded domains has been an important question in numerical studies as well as in theoretical studies .
consider the nonlinear boltzmann solution near maxwellians under in - flow , diffuse , or bounce - back boundary conditions . we demonstrate that discontinuity
is created at the non - convex part of the grazing boundary , then propagates only along the forward characteristics inside the domain before it hits on the boundary again . |
resolved images of circumstellar disks around young stars provide the most direct tool for investigating the formation of planets . at millimeter wavelengths , the thermal dust emission is generally optically thin and measures the radial distribution of circumstellar dust @xcite .
however , since circumstellar disks in nearby star forming regions typically have radii between 100 and 500 au , sub - arcsecond angular resolution is required to spatially resolve the dust emission , even in nearby star - forming clouds .
millimeter - wave interferometers are essential for such studies .
since sub - arcsecond observations at millimeter wavelengths require both high sensitivity and high dynamical range , only a small number of bright disks have been observed at resolutions of 0.4 - 1 to date @xcite . the combined array for research in millimeter - wave astronomy ( carma ) and the new extended configuration of the sub - millimeter array are rapidly enabling more extensive high resolution surveys of circumstellar disks , particularly in the taurus and ophiuchus star forming regions ( * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* hereafter paper i ) .
the highest angular resolution achieved so far by millimeter - wave interferometers is 0.3 - 0.4 , corresponding to spatial scales of 40 - 50 au at the distance of taurus and ophiuchus . in most cases
, the dust density appears to increase smoothly inward down to the orbital radius resolved by the observations , typically @xmath325 au .
however , central cavities in the dust distribution are revealed in a number of disks @xcite .
it remains a matter of debate whether these cavities are caused by dynamical interactions , inside out disk dispersal mechanisms , dust opacity variations , or viscous evolution ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* paper i ) .
nevertheless , these observations still lack the angular resolution required to resolve the innermost part of the disk where the density of the circumstellar material is highest and the formation of planets is more probable . here we describe carma observations of the thermal dust emission towards the young stars dg tau and ry tau at an angular resolution of 0.15 at 1.3 mm and 0.3 at 2.8 mm . at the distance of taurus
( 140 pc ) , 0.15 corresponds to spatial scales of 20 au , such that emission on orbital scales comparable to saturn can be resolved .
this is more than a factor of two improvement over previous observations of circumstellar disks at these wavelengths .
dg tau and ry tau are classical t tauri stars of spectral type m0 and k1 respectively @xcite .
stellar ages inferred from stellar evolutionary models are less than 1 myr ( see paper i for more details and references ) .
the relative youth of both systems is confirmed by the presence of large amounts of gas and dust extending to 0.1 pc and by associated stellar jets and outflows ( see , e.g , * ? ? ?
* ; * ? ? ?
* ) . from near - infrared to millimeter wavelengths ,
both objects exhibit strong emission in excess of that from the stellar photospheres .
this is attributed to rotating disks with radii of few hundred au that first absorb and then re - emit radiation from the central stars .
@xcite . our earlier carma observations of 1.3 mm thermal dust emission from these disks , at a resolution of 0.7 , suggested disk masses between 5 and 150% of the stellar mass for both sources ( see paper i ) . these high disk masses and the youth of ry tau and dg tau make these prime targets to investigate the earliest stages of planet formation .
our new observations of ry tau and dg tau have a factor of 5 better angular resolution and a factor of 3 better sensitivity than the previous data .
this paper investigates three main questions related to the formation of planets in young circumstellar disks .
( 1 ) what is the surface density distribution in the observed disks down to an orbital radius of 10 au ? ( 2 )
are there any signatures of planet formation contained in the dust distribution ? finally , ( 3 ) , do the dust properties vary with orbital radius ?
a qualitative answer to the first two questions is proposed in section [ sec : morp ] where we present the observations and discuss the morphology of the dust emission .
a quantitative analysis is described in section [ sec : mod ] , where we compare the observations with theoretical models of disk emission .
implications of these results for disk structure , for the possible presence of planets , and for the radial variation of the dust opacity are considered in section [ sec : res ] .
the conclusions are presented in section [ sec : conc ] .
we observed thermal dust emission from the ry tau and dg tau circumstellar disks using carma in the a , b , and c configurations .
the date of observation , array configurations used , baseline range , sizes and orientations of the synthesized beams , integrated fluxes , seeing and noise levels are summarized in table [ tab : obs ] .
the c - configuration observations were presented in paper i. llllllll + dg tau ' '' '' & & a+b+c & & [email protected] , 103 & 367@xmath514 & 0.03 & 0.96 + ... ' '' '' & 2007 oct 08 & c & 21 - 279 & [email protected] , 112 & & + ... & 2007 dec 14 & b & 81 - 937 & [email protected] , 116 & & + ... & 2009 jan 31 & a & 130 - 1884 & [email protected] , 102 & & + ry tau ' '' '' & & a+b+c & & [email protected] , 81 & 227@xmath57 & 0 & 0.90 + ... ' '' '' & 2007 oct 22 & c & 16 - 280 & [email protected] , 102 & & + ... & 2008 dec 30 & b & 82 - 935 & [email protected] , 109 & & + ... & 2009 jan 19 & a & 139 - 1884 & [email protected] , 82 & & + ' '' '' + dg tau ' '' '' & & a+b & & [email protected] , 131 & 58@xmath56 & 0.07 & 0.45 + ... ' '' '' & 2008 jan 14 & b & 80 - 798 & [email protected] , 125 & & + ... & 2009 feb 05 & a & 136 - 1678 & [email protected] , 163 & & + ry tau ' '' '' & & a+b & & [email protected] , 82 & 36@xmath53 & 0.07 & 0.28 + ... ' '' '' & 2008 feb 01 & b & 82 - 945 & [email protected] , 95 & & + ... & 2009 feb 10 & a & 123 - 1884 & [email protected] , 82 & & + the observations were obtained at lo frequencies of 228.1 ghz ( @xmath6 mm ) and 106.2 ghz ( @xmath7 mm ) .
the carma correlator at the time of the observations contained three bands , each of which was configured to 468 mhz bandwidth to provide maximum continuum sensitivity .
the band pass shape was calibrated by observing 3c273 ; flux calibration was set by observing uranus and 3c84 .
the radio galaxy 3c111 was observed every 9 minutes to correct for atmospheric and instrumental effects .
variations of the atmospheric conditions on time scales shorter than 9 minutes are not corrected , in effect resulting in seeing .
we quantified the atmospheric seeing by measuring the size of the phase calibrator image ; if the seeing is negligible the phase calibrator appears as a point source .
otherwise , the seeing produces a gaussian smoothing that can be quantified through the full width half maximum ( fwhm ) of the resulting image .
we find that at 1.3 mm the effect of seeing is negligible for ry tau but produces a fwhm of 0.03 for dg tau .
atmospheric conditions were slightly worse during the 2.8 mm observations , resulting in seeing of 0.07@xmath8 for both objects .
these seeing estimates do not account for variations in the atmospheric conditions on angular scales of 10 , corresponding to the separation between the source and the calibrator .
values for the atmospheric seeing are summarized in table [ tab : obs ] and are adopted in the model fitting described in section [ sec : mod ] .
the raw data were reduced using the miriad software package .
the maps of the continuum emission shown in figure [ fig : cont ] were derived using gildas software .
corresponding complex visibilities are shown in figure [ fig : uvamp ] . at 1.3 mm , natural weighting of the a
, b and c configuration observations produced a fwhm synthesized beam size of @xmath3 0.15 .
the noise levels are 0.96 mjy / beam and 0.90 mjy / beam respectively for dg tau and ry tau .
dust emission at 2.8 mm was observed in the a and b configurations at angular resolution of @xmath3 0.35 and noise levels of 0.45 mjy / beam and 0.28 mjy / beam for dg tau and ry tau respectively .
in figure [ fig : cont ] , the dust emission in both disks is clearly resolved and characterized by a smooth and centrally symmetric radial profile .
dg tau intensity contours are almost circular suggesting a disk inclination smaller than 30 . for ry tau , the intensity contours are elongated in the north - east direction suggesting a disk position angle of about 24 measured east from north and a disk inclination of at least 65 .
for both sources the disk orientations agree with those found in paper i. the 1.3 mm dust continuum emission from the ry tau disk shows two spatially resolved peaks separated by about 0.2(28 au ) , and oriented along the apparent major axis of the disk .
details of the central [email protected] region are displayed in the inset in the upper left panel of figure [ fig : cont ] , and the radial profile of the surface density along the disk major axis is shown in figure [ fig : radialprofile ] . the intensity at both peaks is 29 mjy / beam , which is 2 mjy / beam ( i.e. 2.2@xmath0 ) higher than the intensity at the center of the disk .
we also estimated the expected central surface brightness by fitting a gaussian to the surface brightness distribution at angular distances larger than 0.15 .
the fitted gaussian is shown as the solid curve in figure [ fig : radialprofile ] .
a gaussian function was chosen since it provides a reasonable parametric representation of the dust emission .
interpolating this gaussian fit to the center of the disk suggests an expected central surface brightness of 31 mjy / beam , which is 4@xmath0 higher than the measured value .
the significance level of the two intensity peaks , the fact that they appear in the map before cleaning , their orientation along the disk major axis , and the symmetry with respect to the central star , suggest that they are real and , therefore , that the dust emission decreases inside an orbital radius of about 14 au .
this is analogous to the situation in " transitional disks , where the inner gaps observed in the dust emission are attributed to dusty depleted inner regions ( see , e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
at a first sight , this interpretation is incompatible with ry tau s large near- and mid - infrared excesses , which suggest the presence of warm dust within 10 au of the star @xcite . if , however , the inner disk is only partially depleted and dust emission remains optically thick in the infrared , the observed double intensity peak and the spectral energy distribution can be reconciled .
a number of physical mechanisms could reduce the dust density in the inner region of circumstellar disks .
for example , planets less massive than jupiter may carve partially depleted gaps in the surface density distribution by tidal interaction with the surrounding material @xcite .
this possibility is discussed in more detail in section [ sec : surf ] . in paper
i , we also proposed that a surface density profile that gradually decreases towards the star may originate naturally from the viscous evolution of a disk if viscosity decreases with radius .
finally , it is also possible that the decrease in dust emission may originate from a lowering of opacity due to the growth of dust grains to centimeter sizes @xcite .
unambiguously disentangling these models requires even higher angular resolution observations than are yet available .
we must note that radial velocities studies @xcite and hipparcos observations of the variability of the photocenter @xcite suggest that ry tau is a binary . the hipparcos data implies a minimum projected separation of 23.6 mas and a position angle of 304@xmath534 , almost perpendicular to the position angle of the disk inferred from our observations .
assuming that the binary and the disk have the same inclination , the spatial separation between the binary components is 6 - 9 au , and could explain the double peak in the dust continuum emission .
indeed , the presence of a stellar mass companion orbiting at a radius of 6 - 9 au would push the inner radius of the circumstellar disk to a distance of 9 - 13 au by tidal interactions @xcite .
however , the binary nature of ry tau has been rendered questionable by near - infrared interferometric observations that suggest an inner disk radius at 0.1 au from the central star and exclude the presence of a stellar mass companion between 0.35 au and 4 au down to a stellar flux ratio of 0.05 @xcite .
a stellar companion was also undetected in recent spectroscopic and aperture masking observations ( duy cuong nguyen and adam kraus private communication ) . as discussed above ,
the spectral energy distribution is also inconsistent with the existence of a large inner gap completely depleted of gas and dust as could be expected for a stellar companion @xcite .
these results suggest that ry tau is indeed a single star , and the variability observed by hipparcos and the radial velocity variations may be attributed to brightness changes in the circumstellar environment ( see the discussion in * ? ? ?
* and references therein ) .
a notable characteristic of our images of the dust emission is the high degree of central symmetry and , with the exception of the innermost region , the almost complete absence of features in the surface brightness distribution .
if the emission is optically thin ( we will examine this assumption is section [ sec : surf ] ) , this translates to a smooth radial profile for the dust .
the degree of symmetry of the emission can be quantified by analyzing the imaginary part of the correlated flux , plotted in figure [ fig : uvamp ] as a function of the angular frequency @xmath9 .
point symmetric emission will have a zero imaginary part at all spatial frequencies . for ry tau , the deviations from zero are comparable to the noise in the observations ( see the left panels of figure [ fig : uvamp ] ) radial profile of the 1.3 mm surface brightness in ry tau disk measured along the major axis of the disk at a position angle of 24 east from north .
the intensity error bars ( red points ) correspond to the noise level of 0.9 mjy / beam .
the solid curve shows a gaussian fit , while the dashed line indicates the zero intensity level .
the bracket in the upper left indicates the angular resolution . ]
the surface brightness distribution for the dg tau disk has a central peak coincident with the stellar position and smoothly decreases outward to reach the noise level at an angular distance of about 0.5 . at both 1.3 and 2.8 mm
the emission appears fairly symmetric , and indeed the imaginary part of the correlated flux is zero for @xmath10 k@xmath11 ( see upper right panel of figure [ fig : uvamp ] ) .
the imaginary part increases up to 50 mjy at shorter spatial frequencies , suggesting that the emission may be asymmetric on angular scales larger than @xmath12 . comparing the real and imaginary parts of the correlated flux at the shortest spatial frequencies , we find that the asymmetric part of the emission contributes about 14% to the total flux .
as already noted , the high angular resolution observations of ry tau and dg tau allow us to examine disk properties that bear on planet formation .
in particular the radial density profile of the circumstellar material is critical to understanding where planet formation may occur , or where it has occurred . with observations at more than one wavelength , we may also consider radial variations of the grain properties .
as we describe below , measurements of the dust radial profile and the variation of the dust properties with radius are best undertaken by comparing the observations with theoretical disk models in the fourier domain .
there , the effects of different angular resolutions , sensitivity and atmospheric seeing can be more easily taken into account .
moreover , comparison with theoretical models is the only way to quantify the contribution from optically thick emission , leading to an improved estimate of the dust mass .
lllll + dg tau & 0.075 & 3 & 0.082 & 0.056 + ry tau & 0.035 & 3 & 0.131 & 0.075 + + dg tau & 5.0 & 3.5 & 0.012 & 0.0078 + ry tau & 5.0 & 3.7 & 0.026 & 0.015 to investigate the dust radial distribution around dg tau and ry tau , we consider two different models for the disk surface density .
the first model consists of the classical power law parameterization @xmath13 where @xmath14 is the disk surface density at a radius of 40 au . @xmath15 and @xmath16 are the inner and outer disk radii respectively .
the second model is the similarity solution for the evolution of a viscous keplerian disk @xcite .
as discussed in paper i , this has the form @xmath17 \right\ } } \end{gathered}\ ] ] where @xmath18 is the surface density at radius @xmath19 , sometimes called the _
transition radius_. for @xmath20 , the surface density has a power law profile characterized by the slope @xmath21 , while at larger radii the surface density falls exponentially .
these two different parameterizations are used to calculate the dust emission by solving the structure of an hydrostatic disk heated by the stellar radiation @xcite .
the temperature on the disk mid plane is self - consistently calculated by adopting the _ two layer _ approximation of @xcite .
the disk temperature , which depends mainly on the disk optical depth at optical and infrared wavelengths , is roughly proportional to @xmath22 for both surface density profiles ( see paper i and reference therein for a detailed discussion on the disk temperature radial profile ) . fundamental to any disk model is the prescription adopted for the dust opacity . although the dust contributes only about 1% to the total disk mass , it dominates the disk opacity in the wavelength range where most of the stellar and disk radiation is emitted .
we assume that the dust size distribution follows a power law @xmath23 where @xmath24 is the radius of a dust grain .
the assumptions on the slope @xmath25 , on the minimum and maximum grain sizes , on the dust chemical composition and on the grain structure define the frequency dependence of the dust opacity @xmath26 , and , ultimately , the disk emission .
the dust opacity is calculated for compact spheres composed of astronomical silicates and organic carbonates @xcite .
we assume a mass ratio of 1 between silicates and organics , which leads to grain density of 2.5 g/@xmath27 .
the dust opacity averaged over the grain size distribution is calculated by fixing the minimum grain size to 0.005 @xmath28 m .
the maximum grain size @xmath29 and the slope @xmath25 are set to reproduce the observed slope of the spectral energy distribution as discussed below . at millimeter wavelengths
the dust opacity can be approximated by a power law @xmath30 @xcite . if the dust emission is optically thin and the rayleigh - jeans approximation is satisfied , the slope @xmath1 of the dust opacity is related to the spectral index @xmath31 of the observed disk emission @xmath32 ( @xmath33 ) by the relation @xmath34 .
this relation is only approximate if the dust emission is optically thick at some radii . in paper
i , we derived values for @xmath1 of 0.5 and 0.7 for dg tau and ry tau respectively from an analysis of the sed , taking into account the optically thick contribution to the total dust emission . for the assumed dust composition and structure
, these values of @xmath1 can be reproduced with different choices of the maximum grain size @xmath29 and the grain size slope @xmath25 ( see appendix [ app : a ] ) . to investigate how the assumptions on the grain size distribution affect the model fitting , we adopt two different dust models that correspond to the extreme cases of low ( @xmath35 ) and high ( @xmath36 ) opacity .
the corresponding dust opacities at both 1.3 mm and 2.8 mm are given in table [ tab : dust ] .
finally , we assume that the dust opacity is constant throughout the disk .
this is indeed one of the main assumption we want to test by modeling the observed dust emission at 1.3 mm and 2.8 mm and will be discussed in detail in section [ sec : betavar ] .
models and observations are compared in fourier space to avoid the non linear effects introduced by the cleaning process .
the best fit models are found by @xmath37 minimization with five free parameters : the disk inclination @xmath38 , the disk position angle pa , @xmath16 , @xmath14 , and @xmath39 for the power law surface density ( equation [ eq : pow ] ) , and @xmath38 , pa , @xmath19 , @xmath18 and @xmath21 for the similarity solution ( equation [ eq : sim ] ) .
the disk inner radius @xmath15 is fixed at 0.1 au . for both surface density models
we find best fit solutions for both the high ( @xmath36 ) and low ( @xmath35 ) dust opacity models .
the 1.3 mm and 2.8 mm data are fitted independently . to minimize @xmath37 and evaluate the constraints on the model parameters , we use a bayesian approach that adopts uniform prior probability distributions . in practice
we sample the @xmath37 probability distribution by varying the free parameters using the markov chain monte carlo method described in paper i. llllllll dg tau ' '' '' & 1.3 & 24@xmath59 & 119@xmath523 & [email protected] & [email protected] & [email protected] & 1.0608 + ... & 2.8 & 31@xmath512 & 144@xmath519 & [email protected] & [email protected] & [email protected] & 1.0629 + rytau ' '' '' & 1.3 & 66@xmath52 & 24@xmath53 & [email protected] & [email protected] & [email protected] & 1.0896 + ... & 2.8 & 71@xmath56 & 20@xmath54 & [email protected] & [email protected] & [email protected] & 1.1894 + object & @xmath11 ( mm ) & @xmath38 ( ) & pa ( ) & @xmath16 ( au ) & p & @xmath14 ( g/@xmath40 ) & @xmath41 + dg tau ' '' '' & 1.3 & 27@xmath58 & 120@xmath526 & [email protected] & [email protected] & [email protected] & 1.0611 + ... & 2.8 & 32@xmath511 & 144@xmath518 & [email protected] & [email protected] & [email protected] & 1.0629 + rytau ' '' '' & 1.3 & 66@xmath52 & 24@xmath53 & [email protected] & [email protected] & [email protected] & 1.0897 + ... & 2.8 & 71@xmath56 & 20@xmath54 & [email protected] & [email protected] & [email protected] & 1.1894 + once a best fit solution is found , we confirm that this indeed corresponds to an absolute minimum of @xmath37 , as opposed to a local minimum , by running multiple monte carlo simulations with random initializations and verifying that they all converge to the same solution .
each parameter is allowed to vary in a large range : 0 - 80 for the inclination , @xmath590 for the position angle , @xmath42 au for @xmath19 and @xmath16 , @xmath43 for @xmath39 and @xmath21 , and @xmath44 g/@xmath40 for @xmath14 and @xmath18 .
the best fit disk models found for high and low dust opacities are listed in table [ tab : res_clubs ] and table [ tab : res_spades ] respectively .
each table lists the parameters for the similarity solution disk model in the upper part , and for the power law disk model in the lower part .
the probability distributions for each free parameter are shown in figure [ fig : rytau_sim ] and figure [ fig : rytau_pow ] for ry tau in the case of the similarity solution and power law respectively .
the same quantities for dg tau are shown in figure [ fig : dgtau_sim ] and figure [ fig : dgtau_pow ] . in these figures , the black and red histograms indicate the probability distributions derived by fitting the 1.3 mm and 2.8 mm observations , respectively ; solid and dashed curves represent the @xmath36 and @xmath35 dust opacity models . for each parameter
we derive the uncertainty range that corresponds to a likelihood of 99.7% ( 3@xmath0 ) by fitting a normal distribution to the probabilities .
finally , figure [ fig : uvamp_deproj_real ] shows comparisons between the observed real part of the correlated flux ( filled squares with error bars ) , the best fit models for the similarity solution ( solid curve ) , and a power law surface density ( dashed curve ) .
llllllll dg tau ' '' '' & 1.3 & 24@xmath511 & 119@xmath524 & [email protected] & [email protected] & [email protected] & 1.0608 + ... & 2.8 & 31@xmath512 & 144@xmath520 & [email protected] & [email protected] & [email protected] & 1.0629 + rytau ' '' '' & 1.3 & 66@xmath52 & 24@xmath53 & [email protected] & [email protected] & [email protected] & 1.0896 + ... & 2.8 & 71@xmath56 & 20@xmath54 & [email protected] & [email protected] & [email protected] & 1.1893 + object & @xmath11 ( mm ) & @xmath38 ( ) & pa ( ) & @xmath16 ( au ) & p & @xmath14 ( g/@xmath40 ) & @xmath41 + dg tau ' '' '' & 1.3 & 27@xmath59 & 120@xmath524 & [email protected] & [email protected] & [email protected] & 1.0611 + ... & 2.8 & 32@xmath511 & 144@xmath519 & [email protected] & [email protected] & [email protected] & 1.0629 + rytau ' '' '' & 1.3 & 66@xmath52 & 24@xmath53 & [email protected] & [email protected] & [email protected] & 1.0897 + ... & 2.8 & 71@xmath55 & 20@xmath54 & [email protected] & [email protected] & [email protected] & 1.1893 + the best fit solutions for the @xmath36 and @xmath35 dust opacity models are shown in figure [ fig : rytau_pow]-[fig : dgtau_sim ] with solid and dashed curves respectively . in all cases ,
@xmath36 and @xmath35 models lead to very similar values for the disk position angle , the disk inclination and the radial profiles of the surface density defined by @xmath39 and @xmath16 in the case of the power law models , and @xmath21 and @xmath19 for the similarity solution models .
as discussed in paper i , these parameters are essentially independent of the dust opacity .
this is mainly because the disk mid - plane temperature @xmath45 varies by only a few percent between the different dust models , as long as the disk is optically thick to the stellar radiation . since @xmath46 , the radial profile of the surface density varies by only small fraction when
different dust models are assumed .
by contrast , the surface density normalization ( @xmath18 and @xmath14 ) varies with the dust opacity so that the product @xmath47 remains almost constant if the emission is optically thin .
consequently , a lower dust opacity requires a higher dust mass in order to emit the same amount of radiation at millimeter wavelengths .
the ratio @xmath48 is then approximately equal to the ratio between the dust opacities listed in table [ tab : dust ] . from the analysis of the surface density of the best fit model we find that the ry tau emission is always optically thin at both 1.3 and 2.8 mm .
however , dg tau emission is optically thick within 20 au at 1.3 mm for both the similarity solution and the power law models .
the 1.3 mm flux emitted within this region is about 25% of the total flux . at 2.8 mm
the emission is always optically thin in the case of the similarity solution while it is optically thick within 6 au in the power law case . in this case
the optically thick contribution is 5% of the total flux .
different dust opacities lead to different values for the total mass of dust in the disks . for dg tau
we obtain total dust masses of about 33 and 233 earth masses ( @xmath49 ) in the case of the high and low opacity dust models respectively .
disk masses of @xmath310 and 50 @xmath49 are found for ry tau .
massive disks can also be obtained by extending the grain size distribution larger than 5 cm .
for example , in paper i we derived total dust masses of about 1331 and 216 @xmath49 for dg tau and ry tau respectively by assuming a maximum grain size of 10 cm and a slightly different grain composition .
additional uncertainties in the disk mass come from the dust chemical composition . as discussed in appendix [ app : a ] , the presence of ice or vacuum in the grains leads to smaller dust opacities at millimeter wavelengths and consequently produces higher disk masses .
we therefore estimate that the circumstellar disks around dg tau and ry tau contain a minimum mass of dust of 30 and 10 @xmath49 respectively , while the upper limit is not constrained due to the uncertainties on the grain size distribution . assuming the standard dust / gas ratio of 0.01 , these values correspond to minimum disk masses of 0.009 and 0.003 @xmath50 for the dg tau and ry tau respectively . surface density for dg tau ( top panel ) and ry tau ( lower ) disk .
the red and blue curves show the best fit solutions for the power law and the similarity solution models respectively in the low opacity case @xmath35 .
the dashed curves show the 3@xmath0 uncertainty range for the surface density .
the surface densities for the high opacity case will have similar shapes , but will be about a factor of 6 lower .
, title="fig : " ] surface density for dg tau ( top panel ) and ry tau ( lower ) disk .
the red and blue curves show the best fit solutions for the power law and the similarity solution models respectively in the low opacity case @xmath35 .
the dashed curves show the 3@xmath0 uncertainty range for the surface density .
the surface densities for the high opacity case will have similar shapes , but will be about a factor of 6 lower .
, title="fig : " ] figure [ fig : uvamp_deproj_real ] shows the comparison between models and observations in terms of the real part of the correlated flux as a function of the baseline length . to correct for the disk inclination we deprojected the baseline assuming the inclinations and position angles listed in table [ tab : res_clubs ] . in this figure , the results for @xmath36 and @xmath35 dust models lead to indistinguishable curves .
similarity solution and power law models are represented with solid and dashed curves respectively , and the observations are shown by black dots with error bars .
it is clear that both the similarity solution and power law disk models provide satisfactory fits to the observations .
the similarity solution model provides smaller values of @xmath37 ( see table [ tab : res_clubs ] and [ tab : res_spades ] ) and , in the case of dg tau , a better fit to the observations between 400 and 800 k@xmath11 . in this range of spatial frequencies ,
the power law solution is characterized by a wiggle due to the sharp truncation of the dust emission at 72 au . on the other hand
, the exponential tapering of the similarity solution leads to a smooth visibility profile that matches extremely well the observations .
the same behavior is present in the lower panel which compares the model and the observations at 2.8 mm .
however , in this case the observations at @xmath51k@xmath11 are too to distinguish between the two models .
although not conclusive , this result make the similarity solution model a more appealing explanation for the dust emission in circumstellar disks , confirming the conclusions of @xcite .
figure [ fig : sigma ] shows the surface density derived from the 1.3 mm observations for both the power law and the similarity solution model in the case of high dust opacity .
the two models lead to similar values of @xmath52 in the region where most of the 1.3 mm flux is emitted , namely between @xmath315 and 50 au . in this region ,
the surface density in ry tau disk is almost constant with the radius , while it decreases roughly as @xmath53 in the case of dg tau .
inside 15 au and outside 50 au , the observations lack both the angular resolution and the sensitivity required to directly constrain the surface density . as a consequence ,
the values of @xmath52 strongly depend on the assumed model and can differ by one order of magnitude at the disk inner radius . in this section
we discuss the implications of the inferred surface density on the presence of planets .
the analysis is limited to surface density profiles obtained by fitting the observations at 1.3 mm , which have the highest angular resolution .
for the similarity solution model , the surface density has a radial profile characterized by @xmath54 and @xmath55 au .
the transition radius @xmath19 agrees well with our earlier observations ( 21@xmath53 ) but @xmath21 is significantly larger than the value of [email protected] from paper i. the discrepancy is probably due to the fact that the earlier observations were taken in poorer weather conditions and the model fitting did not account for the atmospheric seeing .
figure 10 shows the residuals after subtracting the best fit model to the new observations .
note that the power law model gives very similar residuals .
the residuals are as high as 3 - 6@xmath0 and are found at angular scales larger than 0.7 where the emission is slightly asymmetric ( see section [ sec : morp ] ) . in this outermost disk region ,
the surface density may deviate significantly from the symmetric radial profile assumed in the model .
we calculate that variations of @xmath56 g/@xmath40 with respect to the best fit surface density profile over a spatial region comparable with the beam size may produce the observed residuals .
larger variations of the surface density on smaller angular scales are also possible .
the black curves show the residuals for the 1.3 mm dg tau observations after subtracting the best - fit model for the similarity solution .
contours start at 3@xmath0 and are spaced by 1@xmath0 .
the thin red curves and the color scale show the observed dust emission , with contours spaced by 3@xmath0 . ]
the residuals do not show global deviations from the smooth surface density profile , apparently excluding the possibility of gaps in the dust distribution that might be produced by a planet .
of course , low mass planets may not produce any discernible gap and may still exist in the dg tau disk .
the formation of a gap is possible only if the efficiency in removing the material close to the planet orbital radius via tidal torques is larger than the mass accretion rate due to the disk viscosity ( see , e.g , * ? ? ?
if we assume the @xmath31 prescription for the disk viscosity @xcite , and call @xmath57 the pressure scale height of the disk , a planet orbiting at radius @xmath58 can open a radial gap in the disk surface density only if @xmath59 @xcite .
moreover , for a disk in hydrostatic equilibrium with the gravitational field of the central star , the pressure scale @xmath57 is proportional to @xmath60 @xcite .
since the temperature is @xmath61 ( see paper i for more details ) , the formation of a gap requires @xmath62 where @xmath63 is expressed in jupiter masses , @xmath58 is au , and the numerical constant is calculated for a disk temperature of 194 k at 1 au as determined from our disk model .
typical values of @xmath31 are in the range @xmath64 , and imply that a planet can open a gap at 1 au only if its mass is larger than about 0.1 @xmath65 . to open a gap at 30 au
, the mass must be larger than about 0.5 @xmath65 . to investigate the effects that a planet more massive than 0.1 @xmath65 might have on the observations of the dust continuum emission
, we simulated the presence of a planet in the dg tau disk by opening a gap in the surface density distribution corresponding to the best fit models discussed above . for simplicity , we assumed that the planet describes a circular orbit and that the gap can be represented by a circular ring . to be compatible with numerical simulations of planet - disk interaction , the half - width of the ring @xmath66
is assumed to be equal to twice the hill radius @xmath67{m_p/(3m_{\star})}$ ] ( e.g. * ? ? ?
* ; * ? ? ?
* ) . in the region between @xmath68
the surface density is depleted by a fraction @xmath69 that depends on the mass of the planet and on the disk viscosity . for @xmath70
, we can assume @xmath71 for planet masses m@xmath72 , @xmath73 for @xmath74 @xmath65 , @xmath75 for @xmath76 and @xmath77 for @xmath78 @xcite .
therefore , only planets more massive than 1 @xmath65 will produce completely cleaned gaps .
we simulated gaps corresponding to planets in the mass range 0.3 - 5 @xmath65 and with orbital radii between 1 and 90 au . for each model we calculated the residuals as the difference between the observations of dg tau at 1.3 mm and the model image .
if the gap is too small compared to our angular resolution , or too faint compared with our sensitivity , the residuals will be similar to the case without gaps shown in figure [ fig : res_dgtau ] . in this case
we say that the gap is not detected .
on the other hand , a large and deep gap will produce a bright ring in the residual . to quantify how reliable the detection of a gap is ,
we define a signal - to - noise ratio of the gap , _ gap snr _ , in the following way .
first , we deproject the residual for the inclination and position angle of dg tau disk .
then we take the radial average of the residuals at the distance corresponding to the orbital radius of the planet adopting a radial bin width equal to the fwhm of the synthesized beam ( i.e. , 0.17 ) .
we define the _ gap snr _ as the mean residual in the radial bin divided by the uncertainty in the mean . in this way ,
detected gaps correspond to _
gap snr_@xmath79 . the signal - to - noise ratio of the detection of a gap generated by a planet as a function of the orbital radius @xmath58 and the planet mass .
the different curves correspond to masses between 0.5 and 10 jupiter masses as labeled .
the thick red curves indicate the signal - to - noise ratio measured from the 1.3 mm images after subtracting the best fit model for the similarity solution.,title="fig : " ] the signal - to - noise ratio of the detection of a gap generated by a planet as a function of the orbital radius @xmath58 and the planet mass .
the different curves correspond to masses between 0.5 and 10 jupiter masses as labeled .
the thick red curves indicate the signal - to - noise ratio measured from the 1.3 mm images after subtracting the best fit model for the similarity solution.,title="fig : " ] for _ gap snr_@xmath80 , planets should produce a detectable gap .
the upper and lower panel refers to the case of dg tau and ry tau respectively .
the results are summarized in the upper panel of figure [ fig : planet_deviation ] .
planets with masses and radii that lead to _
gap snr_@xmath80 produce detectable gaps .
no gaps are detected in our observations of the dg tau disk at more than 3@xmath0 ( see the red curve ) .
this enable us to constrain the masses and orbital radii of any planets that may be present .
in particular , we can exclude that planets more massive than jupiter exist between 5 and 40 au , or that planets with masses slightly smaller than jupiter exists between 10 and 25 au .
the observations lack both the angular resolution and the sensitivity required to detect gaps produced by planets with a mass smaller than about 0.5 m@xmath81 .
an important caveat is that a planet may exist but may not have had enough time to completely open a gap in the disk .
the gap formation time scale @xmath82 results from the trade - off between the efficiency of the tidal torque exercised by the planet in removing angular momentum , and the accretion of new material coming from larger radii in the gap due to the disk viscosity .
a lower limit of the gap formation time scale is obtained in the zero viscosity limit . in this case ,
an analytic formulation is provided by @xcite in the form @xmath83 where @xmath84 is the orbital period , @xmath85 and @xmath86 as defined above .
assuming keplerian rotation , we can rewrite the time scale for the gap formation as @xmath87 the upper panel of figure [ fig : tau_delta ] shows the calculated values of @xmath88 for the stellar mass of dg tau ( 0.3 @xmath50 ) . in the case of a planet with a mass between 0.3 and 0.5 m@xmath81 orbiting at a radius larger than 40 au , the minimum time scale for the gap formation is comparable with the age of the system ( 0.1 myr ) . for more massive planets , or for closer radii ,
the minimum gaps time scale is a small fraction of the age of the system .
we conclude that , for dg tau , the observations lack the sensitivity and angular resolution required to investigate the presence of planets less massive than about 0.5 m@xmath81 at any orbital radius .
our analysis indicates that no planets more massive than jupiter are present between 5 and 50 au , unless they are younger than @xmath89 years .
minimum time scale ( @xmath82)for the formation of a gap as a function of the orbital radius and mass of the planet .
the different curves correspond to planet masses between 0.3 and 5 m@xmath81 as labeled in the figure .
the upper and lower panels show the time scale for a central star of 0.3 @xmath50 and 2 @xmath50 respectively . ]
the similarity solution for the disk surface density is characterized by @xmath90 and @xmath91 au .
as shown in figure [ fig : sigma ] , the surface density increases roughly as @xmath92 from the inner radius at 0.1 au up to about 26 au and then decreases exponentially outward .
this supports the suggestion in section [ sec : morp ] that the ry tau inner disk might be partially dust depleted with respect to power law disk models .
we note that this surface density profile may provide an explanation for both the double peak intensity at 1.3 mm and the disk excess at infrared wavelengths .
indeed , within 10 au the model disk remains optically thick at optical and infrared wavelengths , exhibiting the infrared excess typical of classical disks . at larger radii ,
the surface density in ry tau disk decreases smoothly and the residuals calculated by subtracting the best fit models to the 1.3 mm dust emission map do not show any structure at more than 3@xmath0 .
this excludes strong deviations from an unperturbed viscous disk profile .
the lower panel of figure [ fig : planet_deviation ] shows the signal - to - noise ratio of the detection of a gap generated by planets of 1 , 5 and 10 jupiter masses as a function of the orbital radius . due to the higher disk inclination and stellar mass ,
a planet orbiting around ry tau would produce a less visible gap .
in particular , our observations seems to exclude the presence of planets more massive than 5 jupiter masses between 10 and 60 au . given the higher stellar mass of ry tau ,
the minimum time scale for the formation of gaps is one order of magnitude larger than the case of dg tau ( see lower panel of figure [ fig : tau_delta ] ) .
this implies that planets less massive than jupiter orbiting at more than about 30 au may not have had enough time to form a gap in the disk .
a comparison of the best fit solutions obtained for the wavelengths of 1.3 mm and 2.8 mm enable us to investigate the dependence of dust opacity on the orbital radius .
if the dust opacity is constant throughout the disk as assumed in section [ sec : mod ] , the model fitting necessarily leads to the same surface density profile for observations at two different wavelengths .
otherwise , different @xmath52 would suggest a radial variation in the relative dust opacities at the observed wavelengths . to understand this point
, we assume that the dust emission is optically thin . in this case the observations constrain the product @xmath93 , where @xmath94 is the surface density obtained by fitting the observations at the wavelength @xmath11 . in the more general case in which the dust opacity depends on the orbital radius we can write @xmath95 the left side of this equation contains the opacity discussed in section [ sec : mod ] and the surface density derived from the model fitting .
the right side contains the unknown `` true '' surface density @xmath96 in the case in which the `` true '' dust opacity @xmath97 varies with the radius .
the ratio of equation [ eq : sigmak ] for two different wavelengths @xmath98 and @xmath99 leads to @xmath100 here we assumed that at each radius the dust opacity can be expressed by a power law @xmath101 . finally , taking the logarithm of this latter equation we can write @xmath102 where @xmath103 and @xmath104 has the form @xmath105.\ ] ] if @xmath106 , the dust opacity slope is constant throughout the disk and assumes the value discussed in section [ sec : mod ] .
otherwise , we can use the latter equation to investigate the radial variation of @xmath1 .
the best fit solutions for @xmath52 obtained at 1.3 mm and 2.8 mm are shown in figure [ fig : rytau_pow]-[fig : dgtau_sim ] with black and red curves respectively .
the best fit parameters are summarized in table [ tab : res_clubs ] and [ tab : res_spades ] .
the quoted uncertainties correspond to a likelihood of 99.7% ( i.e. 3@xmath0 ) and are calculated by fitting a normal distribution to the measured probability distributions . for ry tau , the disk model obtained by fitting the two wavelengths
separately are in agreement within 3@xmath0 . for dg tau the solutions disagree by more than 3@xmath0 only in the case of the similarity solution and high dust opacity . figure [ fig : dbeta ] shows the radial variation of @xmath1 as defined in equation [ eq : betar ] for both dg tau and ry tau .
the region marked with color indicates values of @xmath1 within 3@xmath0 from the radial profile corresponding to the best fit solution for the surface density in the case of the similarity solution model .
values of @xmath1 outside this region are rejected by our observations . the results for ry tau and dg tau are clearly consistent with a large variety of radial profiles of @xmath1 .
for both sources , @xmath1 is better constrained between radii of 20 and 70 au . however , even in this interval , the observations constrain possible variation of @xmath1 to within only @xmath107 .
nevertheless , across most of the disk the circumstellar dust differs from that observed in the interstellar medium ( ism ) .
dust in the ism is characterized by sub - micron dust grains and by a millimeter opacity slope of @xmath108 .
in contrast , in both sources , @xmath1 is smaller than 1.7 up to at least 80 au , suggesting that the circumstellar dust has been processed and , in particular , has increased its size up to a maximum value that varies between 20 @xmath28 m and a few centimeters .
although in both sources @xmath1 may be constant throughout the disk ( see dashed lines ) , our results suggest that @xmath1 decreases with the radius in dg tau disk . as discussed in section [ sec : mod ] , @xmath1 depends on a number of poorly constrained quantities , such as composition , structure , and size of the dust grains .
for example , varying @xmath1 from @xmath31 to @xmath30.2 , similar to what is suggested for dg tau between 10 and 60 au , may be due to the maximum grain size increasing from @xmath109 @xmath28 m to 1 cm for @xmath110 , or , alternatively , to a decrease of @xmath25 from 4 to 3 if the maximum grain size is between 1 and 10 cm . in short
, the interpretation of @xmath1 only in terms of the grain size distribution can be very misleading .
it seems most plausible that both dust composition and the relative contributions of smaller and larger grains change through the disk , contributing to the variation of the dust opacity .
it is clear that better constraints on the radial profile of @xmath1 are required before pushing the investigation further . in this regard
, we note that the current constraints on the radial variation of @xmath1 are limited by two factors .
first , although the angular resolution of the observations described here is significantly better than hitherto possible , the dust surface density is well constrained only between 15 and 50 au , where most of the observed flux is emitted ( see the discussion in section [ sec : surf ] ) . at smaller and larger
radii the surface density is uncertain by almost one order of magnitude .
second , our analysis is hampered by the small separation in wavelength between the observations since @xmath104 is proportional to @xmath111 .
the uncertainties shown in figure [ fig : dbeta ] can be reduced by a factor of 2 by extending the observations at 7 mm .
these observations will become possible with the expanded correlator on the evla .
we have presented carma observations of the dust thermal emission at the wavelengths of 1.3 mm and 2.8 mm from the circumstellar disks around the pre - main sequence stars ry tau and dg tau .
the observations are characterized by unprecedented angular resolution of @xmath30.15 and 0.30 at 1.3 mm and 2.8 mm respectively , corresponding to spatial scales of 20 and 40 au at the distance of taurus .
based on these images , we have addressed three fundamental questions related to the formation of planets in the disk around pre - main sequence stars .
what is the radial density distribution of circumstellar dust ? does the dust emission show any indication of the presence of ( proto)-planets ? do the dust properties vary with orbital radius ? by analyzing the morphology of the surface brightness of the dust emission and comparing the observations with theoretical disk models , we make the following conclusions : * both the classical power law disk surface density ( hayashi 1981 ) and the similarity solution for the viscous evolution of a keplerian disk ( hartmann 1998 ) fit the observations well . the surface density is well constrained between 15 and 50 au . in this region ,
the two models lead to values of @xmath112 that agree within 30% for a fixed dust opacity . at smaller and larger radii ,
the surface density depends on the assumed model and varies by almost an order of magnitude .
we have verified that the assumptions on the dust opacity have a small effect on the model fitting and , therefore , on the radial profile of the dust density .
however , the total disk mass may vary by almost two order of magnitude for different dust compositions and grain size distributions . *
the dust emission in dg tau is mostly radially symmetric .
it is characterized by a single , central peak and smoothly decreases up to an angular distance of about 0.5 .
theoretical disk models reproduce the observation very well , with randomly distributed residuals between 3 and 6@xmath0 .
no systematic deviation from the similarity solution for the surface density of a viscous disk are observed . by simulating the presence of planets in the disk via the gap in the surface density produced by tidal torques ,
we find that the observations exclude the presence of planets more massive than jupiter orbiting between 5 and 40 au from the central star , unless the planets are very young ( @xmath113 yr ) and have not had the time to open a gap in the disk .
the observations lack both the angular resolution and sensitivity to investigate the presence of planets less massive than about 0.5 jupiter masses .
+ for ry tau , the dust emission is characterized by two peaks separated by about 28 au that suggest a decrease in the surface density , or dust opacity , within 14 au of the central star .
we found that the similarity solution for the disk surface density is characterized by a negative value of @xmath21 , and provides a reasonable explanation for the double peak intensity observed at 1.3 mm .
depletion of millimeter dust grains @xcite , decreasing values of the disk viscosity , or the presence of planetesimals , may produce the observed dust morphology . at larger radii
, the dust emission shows a very smooth profile with no asymmetries or gaps .
the lack of gaps in the disk suggests that any planets between 10 and 50 au are less massive than about 5 jupiter masses , or , as for dg tau , are very young .
* the best - fit models to the 1.3 mm and 2.8 mm data were compared to investigate the radial dependence of the slope opacity @xmath1 , assuming that the dust opacity at millimeter wavelengths is expressed by a power law @xmath114 .
we can exclude cases in which @xmath1 varies by more than 0.7 within 70 au .
nevertheless , between 20 and 70 au , the disks around dg tau and ry tau are characterized by values of @xmath1 that are smaller than that found in the ism .
this implies that the dust has been reprocessed and has grown in size up to a radius of at least 20 @xmath28 m .
the investigation of the radial variation of @xmath1 is still limited by the angular resolution and by the small separation in wavelength between the observations . in the future ,
alma and the evla will play crucial roles in the investigation of the radial dependence of the dust properties by increasing the angular resolution and the interval in wavelength .
we thank antonella natta and steven beckwith for reading the manuscript and providing helpful suggestions .
we thank the ovro / carma staff and the carma observers for their assistance in obtaining the data .
support for the combined array for research in millimeter astronomy construction was derived from the gordon and betty moore foundation , the kenneth t. and eileen l. norris foundation , the associates of the california institute of technology , the states of california , illinois , and maryland , and the national science foundation .
ongoing carma development and operations are supported by the national science foundation under a cooperative agreement ( grant ast 08 - 38260 ) , and by the carma partner universities .
this work was performed in part under contract with the jet propulsion laboratory ( jpl ) funded by nasa through the michelson fellowship program .
jpl is managed for nasa by the california institute of technology .
the slope of the dust opacity @xmath1 measured at millimeter wavelengths has been widely adopted to constrain the size and opacity of the circumstellar dust ( see , e.g. , * ? ? ?
* ) . in this appendix
we investigate how these latter two quantities depend on the assumption on the grain size distribution and composition . the color scale in figure [ fig : k_beta ]
shows the dust opacity at 1.3 mm calculated for three different dust compositions . in model
a we assume that the grains are compact spheres composed of astronomical silicates and organic carbonates @xcite .
this is the dust model adopted in the paper . in model
b the grain is composed of silicates , carbonates and water ice .
finally , in model c the ice is replaced by a vacuum , resulting in a porous grain made of silicates and carbonates .
we assume a mass ratio of 1 between silicates and organics , and of 0.7 between silicates and the water ice or the vacuum @xcite .
the resulting bulk densities are 2.5 g/@xmath27 for model a , 1.9 g/@xmath27 for model b and 1.5 g/@xmath27 for model c. we assume a grain size distribution @xmath23 and fix the minimum grain size to 0.005 @xmath28 m .
we then calculate the dust opacity by varying the maximum grain size @xmath29 between @xmath115 and 10 cm and the slope @xmath25 between 2 and 5 .
the solid thick curves show the values of @xmath116 required to obtain the values of @xmath1 measured for dg tau and ry tau , 0.5 and 0.7 respectively . from the figure , it is clear that the value of @xmath1 sets a lower bound to the maximum grain radius in the grain size distribution ( see , e.g. , * ? ? ?
this lower bound is a function of the grain composition and increases by almost an order of magnitude between model a and c. for example , in order to have @xmath117 , the maximum grain size must be at least @xmath118 cm for compact grains in model a , or about 0.2 cm for the porous grains in model c. note , however , that we can obtain the same value of @xmath1 with @xmath119 cm and @xmath25 between 3.2 and 3.7 .
deriving the maximum grain size from the measure of @xmath1 is evidently strongly degenerate .
this introduces a large uncertainty on the dust opacity and ultimately on the total mass of circumstellar dust derived from millimeter observations .
even if we limit the analysis to the generally adopted value @xmath120 ( see , e.g. , * ? ? ?
* ) , the dust opacity for @xmath121 varies from about 8.4 @xmath40/g of dust in the case of the model a and @xmath122 cm , to 1.2 @xmath40/g of dust for model c and @xmath119 cm , leading to disk masses that differ by almost one order of magnitude .
the color gradient depicts the dust opacity as a function of the maximum grain size @xmath29 and the slope of the grain size distribution @xmath25 .
results are shown for three different dust models ( a , b and c ) as described in the text .
the minimum grain size is fixed at @xmath123 cm .
thin solid curves show the dust opacity contours and are spaced by 0.5 dex .
the thick solid curves show the possible pairs of @xmath116 that lead to values for the opacity slope @xmath1 equal to 0.5 ( dg tau ) and 0.7 ( ry tau ) . the letters @xmath36 and @xmath35 indicate the high and low dust opacity models adopted in fitting the observed dust emission .
, width=283 ] alexander , r. d. , clarke , c. j. & pringle , j. e. , 2006 , mnras , 369 , 216 alexander , r. d. & armitage , p. j. , 2007 , , 375 , 500 akeson , r. l. , boden , a. f. , monnier , j. d. , millan - gabet , r. , beichman , c. , beletic , j. , calvet , n. , hartmann , l. , hillenbrand , l. , koresko , c. , sargent , a. & tannirkulam , a. , 2005 , , 1173 , 1181 andrews , s. m. , wilner , d. j. , hughes , a. m. , qi , c. & dullemond , c. p. , 2009 , , 700 , 1502 beckwith , s.v.w . , sargent , a.i . ,
chini , r.s . & guesten , r. , 1990 , , 99 , 924 beckwith , s.v.w . & sargent , a.i . , 1991 , , 381 , 250 bertout , c. , robichon , n. & arenou , f , 1999 , , 352 , 574 brauer , f. , dullemond , c. p. & henning , th .
, 1998 , , 480 , 859 bryden g. , xingming c. , lin d. n. c. , nelson r. p. & papaloizou j. c. b. , 1999 , , 514 , 344 brown , j. m. , blake , g. a. , qi , c. , dullemond , c. p. & wilner , d. j. , 2008 , , 675 , 109 brown , j. m. , blake , g. a. , qi , c. , dullemond , c. p. & wilner , d. j. , 2009 , , 704 , 496 calvet , n. , dalessio , p. , watson , d. m. , franco - hernndez , r. , furlan , e. , green , j. , sutter , p. m. , forrest , w. j. , hartmann , l. , uchida , k. i. , keller , l. d. , sargent , b. , najita , j. , herter , t. l. , barry , d. j. & hall , p. , 2005
, , 630 , 185 chiang , e. i. & goldreich , p. , 1997
, , 490,368 chiang e. i. & murray - clay r. a. , 2007 , nature physics , volume 3 , issue 9 , pp .
604 - 608 dalessio , p. , calvet , n. & hartmann , l. , 2001 , , 553 , 321 dullemond , c. p. , dominik , c. , & natta , a. 2001 , , 560 , 957 dullemond , c. p. & dominik , c. , 2005 , , 434 , 971 guilloteau , s. , dutrey , a. & simon , m. , 1999 , , 348 , 570 hartmann , l. , calvet , n. , gullbring , e. & dalessio , p. , 1998
, , 495 , 385 hayashi , t. , 1981 , progress of theoretical physics supplement , no .
35 - 53 herbig , g. h. , bell & k. robbin , 1988 , hird catalog of emission - line stars of the orion population .
gerbig and k.r . bell .
lick observatory bulletin 1111 , santa cruz : lick observatory , jun 1988 , 90 p. hughes , a. m. , andrews , s. m. , espaillat , c. , wilner , d. j. , calvet , n. , dalessio , p. , qi , c. , williams , j. p. & hogerheijde , m. r. , 2009 , , 698 , 131 hughes , a. m. , wilner , d. j. , qi , c. & hogerheijde , m. r. , 2008 , , 678 , 1119 kenyon , s. & hartmann , l. , 1995 , , 101 , 117 koerner d.w .
& sargent a.i . , 1995 , , 109 , 2138 isella , a. , testi , l. , natta , a. , neri , r. , wilner , d. & qi , c. , 2007 , , 469 , 213 isella , a. , carpenter j.m . &
sargent a.i . , 2009 , , 701 , 260 lin , d.n.c .
& papaloizou j.c.b . , 1993 , protostars and planets iii , p. 749 - 835
lynden - bell , d. , & pringle , j.e .
, 1974 , , 168 , 603 mcgroarty f. & ray t. p. , 2004
, , 420 , 975 muzerolle , j. , calvet , n. & hartmann , l. , 1998 , , 492 , 743 natta , a. , testi , l. & muzerolle , j. , 2004 , , 424 , 603 natta , a. , testi , l. , calvet , n. , henning , th . , waters , r. & wilner , d. , protostars and planets v , b. reipurth , d. jewitt , and k. keil ( eds . ) , university of arizona press , tucson , 951 pp . , 2007 . , p.767 - 781 pitu , v. , guilloteau , s. & dutrey , a. , 2005 , , 443 , 945 pitu , v. , dutrey , a. , guilloteau , s. , chapillon e. & pety , j. , 2006 , 460 , 43 pitu , v. , dutrey , a. & guilloteau , s. , 2007 , , 467 , 163 pollack , j. b. , hollenbach , d. , beckwith , s. , simonelli , d. p. , roush , t. & fong , w. , 1994 , , 421 , 615 pott j .- u
. , perrin m. d. , furlan e. , ghez a. m. , herbst t. m. & metchev s. , 2009 , apj in press , arxiv:0912.0846 ricci , l. , testi , l. , natta , a. , neri , r. , cabrit , s. & herczeg , g. j. 2010 , in press , arxiv:0912.3356 robitaille , t. p. , whitney , b. a. , indebetouw , r. & wood , k. , 2007 , , 169 , 328 schegerer a. a. , wolf s. , ratzka th . &
leinert ch . , 2008 , , 478 , 779 shakura , n. i. & syunyaev , r. a. , 1973 , , 24 , 337 simon , m. , dutrey , a. & guilloteau , s. , 2000 , , 545 , 1034 strom , k. m. , strom , s. e. , edwards , s. , cabrit , s. & skrutskie , m. f. , 1989 , , 97 , 1451 st - onge , g. & bastien , p. , 2008
, , 674 , 1032 testi , l. , bacciotti , f. , sargent , a.i . ,
ray t.p . & eislffel j. , 2002 , , 394 , 31 testi , l. , natta , a. , shepherd , d.s . & wilner d. , 2003 , , 403 , 323 weingartner , j. c. & draine , b. t. , 2001 , , 563 , 842 wilner , d. j. , ho , p. t. p. , kastner , j. h. & rodrguez , l. f. , 2000 , , 534 , 101 wolf s. , moro - martin a. & dangelo g. , 2007 , p&ss , 55 , 569 zubko , v.g . , mennella , v. , colangeli , l. & bussoletti , e. , 1996 , , 282 , 1321 | we present carma observations of the thermal dust emission from the circumstellar disks around the young stars ry tau and dg tau at wavelengths of 1.3 mm and 2.8 mm .
the angular resolution of the maps is as high as 0.15 , or 20 au at the distance of the taurus cloud , which is a factor of 2 higher than has been achieved to date at these wavelengths .
the unprecedented detail of the resulting disk images enables us to address three important questions related to the formation of planets .
( 1 ) what is the radial distribution of the circumstellar dust ? ( 2 ) does the dust emission show any indication of gaps that might signify the presence of ( proto-)planets ? ( 3 ) do the dust properties depend on the orbital radius ? we find that modeling the disk surface density in terms of either a classical power law or the similarity solution for viscous disk evolution , reproduces the observations well .
both models constrain the surface density between 15 and 50 au to within 30% for a given dust opacity . outside this range ,
the densities inferred from the two models differ by almost an order of magnitude .
the 1.3 mm image from ry tau shows two peaks separated by 0.2 with a decline in the dust emission toward the stellar position , which is significant at about 2 - 4@xmath0 . for both ry tau and dg tau , the dust emission at radii larger than 15 au displays no significant deviation from an unperturbed viscous disk model . in particular
, no radial gaps in the dust distribution are detected . under reasonable assumptions ,
we exclude the presence of planets more massive than 5 jupiter masses orbiting either star at distances between about 10 and 60 au , unless such a planet is so young that there has been insufficient time to open a gap in the disk surface density .
the radial variation of the dust opacity slope , @xmath1 , was investigated by comparing the 1.3 mm and 2.8 mm observations .
we find mean values of @xmath1 of 0.5 and 0.7 for dg tau and ry tau respectively .
variations in @xmath1 are smaller than @xmath2 between 20 and 70 au .
these results confirm that the circumstellar dust throughout these disks differs significantly from dust in the interstellar medium . |
a major surprise arising from direct detection of gas - giant planets around young stars is that the spectral properties of these objects differ from those of field l and t dwarfs .
these young planets have redder near - ir colors , fainter near - ir absolute magnitudes , and peculiar spectra compared to their field analogs . over the last several years , this development has fostered closer scrutiny of the long - standing paradigm that a simple physical sequence connects the lowest mass stars to brown dwarfs to gas - giant planets .
we now know that most field brown dwarfs are not good analogs to young exoplanets .
in contrast to field t dwarfs of similar temperature , the young planets around hr 8799 and 2mass j1207@xmath039 have no methane absorption and very red colors .
these properties are thought to arise from extreme atmospheric conditions tied to their young ( @xmath61030 myr ) ages and low gravities , e.g. , enhanced vertical mixing , non - equilibrium chemistry , and unusual clouds ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this is corroborated by recent studies of the youngest ( @xmath710100 myr ) field brown dwarfs , which find that low - gravity l dwarfs also show very red colors and spectral peculiarities ( e.g. * ? ? ?
* ; * ? ? ?
complicating this interpretation , however , is the existence of very red l dwarfs that do not show spectral signatures of youth ( e.g. * ? ? ?
* ; * ? ? ?
thus , similarities between the colors and spectra of field brown dwarfs and young planets can have ambiguous interpretations .
there are two prominent shortfalls in our observational knowledge .
( 1 ) there are only a handful of very red young l dwarfs currently known ( see compilation in @xcite ) , and only one of them has a parallax measurement @xcite .
( 2 ) most l dwarfs do not have as red near - ir colors as young exoplanets , and _
none _ are as faint in their near - ir absolute magnitudes .
therefore , the utility of young field objects as exoplanet analogs may be limited , since the existing samples of these two types of objects are small and do not really overlap .
we have found an extraordinary young l dwarf that will help shed light on these topics .
it is the reddest field object found to date and the first to have absolute magnitudes comparable to directly imaged young dusty exoplanets .
we have been undertaking a search for t dwarfs using the ps .
( ps1 ) @xmath8 survey @xcite .
we select objects using ps . and 2mass based on their colors and proper motions and then obtain near - infrared photometry for further screening .
we observed pso [email protected] ( hereinafter pso j318.5@xmath022 ) using wfcam on the uk infrared telescope ( ukirt ) on the 2010 september 15 ut .
conditions were photometric with 0.91.0 seeing .
we found that the @xmath9 color for pso j318.5@xmath022 was @xmath10 mag ( table 1 ) , significantly redder than any previously known field dwarf .
we obtained @xmath11100 near - ir ( 0.82.5 ) spectra on 2011 july 21 ut from nasa s infrared telescope facility .
( by coincidence , we observed pso j318.5@xmath022 immediately before obtaining the spectrum of the similarly red l dwarf wise j0047 + 68 published in @xcite . ) conditions were photometric with 1 seeing .
we used the near - ir spectrograph spex @xcite in prism mode with the 0.8 slit .
the total on - source integration time was 16 min .
all spectra were reduced using version 3.4 of the spextool software @xcite .
we also used the final spectrum to synthesize near - ir colors for pso j318.5@xmath022 ( table 1 ) . to assess the gravity of pso j318.5@xmath022 , we obtained @xmath12 near - ir spectra using the gnirs spectrograph @xcite on the gemini - north 8.1-m telescope .
cross - dispersed spectra were obtained using the 0.3 slit and the 32 lines mm grating on the nights of 2013 june 26 , june 30 , and july 01 ut .
the total integration time was 5400 s. we reduced the data using a version of spextool modified for gnirs cross - dispersed data .
we combined the telluric - corrected spectra from the three nights using a robust weighted mean to produce the final 0.952.5 spectrum .
we conducted astrometric monitoring of pso j318.5@xmath022 with the facility near - ir camera wircam at the canada - france - hawaii telescope , obtaining 9 epochs over 2.0 years starting on 2011 july 26 ut .
our methods are described in @xcite . using 116 reference stars in the field of pso j318.5@xmath022 ,
the resulting median astrometric precision per epoch was 4.0mas , and the best - fit proper motion and parallax solution had @xmath13 with 13 degrees of freedom .
we applied a relative - to - absolute parallax correction of @xmath14mas derived from the besanon model of the galaxy .
table 1 gives our astrometry results .
we did not find any objects co - moving with pso j318.5@xmath022 in our 10@xmath154@xmath1610@xmath154 field of view within a range between 0.6mag fainter and 3.3mag brighter at @xmath3 band .
the colors of pso j318.5@xmath022 are extreme , with mag , mag , and @xmath17 mag , all being the reddest among field l dwarfs ( figure [ fig : colorcolor ] and also see @xcite ) .
such colors are thought to arise from an unusually dusty atmosphere that results from a low surface gravity ( young age ) .
the position of pso j318.5@xmath022 on the near - ir color - magnitude diagram is similarly extreme , being significantly fainter in @xmath3-band absolute magnitude than field l dwarfs ( figure [ fig : colorcolor ] ) .
it coincides with the colors and magnitudes of the directly imaged planets around hr 8799 and 2mass j1207@xmath039b .
we determine the near - ir spectral type of pso j318.5@xmath022 using the @xcite system , which provides gravity - insensitive types consistent with optical spectral types . for late - l dwarfs , visual classification in the @xmath3 and @xmath18 bands and index - based classifications with the h@xmath19od index are applicable .
for the gnirs spectrum of pso j318.5@xmath022 , we visually assign a @xmath3-band type of l9@xmath11 and a @xmath18-band type of l6@xmath11 .
the h@xmath19od index corresponds to [email protected] .
the weighted mean of these three determinations leads to a final type of l7@xmath11 .
spectral typing of our low - resolution spex spectrum gives the same classification ( table 1 ) .
pso j318.5@xmath022 shows a triangular @xmath2-band continuum , which is considered a hallmark of youth ( e.g. * ? ? ?
however , @xcite caution that very red l - dwarfs having no signatures of youth ( low gravity ) can display a triangular @xmath2-band shape ( e.g. , 2mass j2148 + 40 in figure [ fig : spectra ] ) . at moderate resolution , there are other indicators of youth for late - l dwarfs .
our gnirs spectrum displays a weak 1.20 @xmath20 m feh band as well as weak ( 1.14 @xmath20 m ) and ( 1.17 and 1.25 @xmath20 m ) lines , which indicate a low gravity . using the gravity - sensitive indices of @xcite , we classify pso j318.5@xmath022 as vl - g , which allers & liu suggest correspond to ages of @xmath71030 myr based on the ( small ) sample of young late - m / early - l dwarfs with good age constraints .
altogether , pso j318.5@xmath022 visually appears most similar to the red l dwarfs wise j0047 + 68 @xcite and 2mass j2244 + 20 @xcite , in accord with the similar spectral types and gravity classifications for these three objects .
allers & liu ( 2013 ) note that objects of the same age and spectral type ( temperature ) can display different spectral signatures of youth , based on 2 young l dwarfs in the ab dor moving group .
our new discovery affirms this idea .
the spectra of 2mass j1207@xmath039b and pso j318.5@xmath022 are quite different ( figure [ fig : spectra ] ) , despite their similar colors and absolute magnitudes and the fact they may be coeval ( section [ sec : membership ] ) . pso j318.5@xmath022 shows a negative continuum slope from 2.122.28 @xmath20 m , whereas 2mass j1207@xmath039b has a positive slope .
the @xmath2-band continuum of pso j318.5@xmath022 displays a `` shoulder '' at 1.58 @xmath20 m , whereas 2mass j1207@xmath039b has a very peaked continuum .
overall , the spectrum of 2mass j1207@xmath039b appears most similar to that of 2mass j0355 + 11 ( both objects have near - ir types of l3 vl - g ) , despite their large differences in ages , colors , and absolute magnitudes .
altogether , these comparisons hint that determining relative ages and temperatures from nir spectra may be unexpectedly complex .
to assess the physical parameters , we fit the low - resolution near - ir spectra of pso j318.5@xmath022 and 2mass j0355 + 11 with the ames / dusty model atmospheres @xcite and the bt - settl model atmospheres @xcite with two different assumed solar abundances ( [ agss ] and @xcite [ cifist ] ) . since both objects have parallaxes , the scaling factors @xmath21 from the fits provide an estimate of the objects radii . using minimization , all three sets of models
indicates that pso j318.5@xmath022 ( @xmath22 ) is 100200 k cooler than 2mass j0355 + 11 ( @xmath23 ) .
bt - settl / agss models give the best fits , with the resulting temperatures being comparable to previous fitting of field l dwarfs but with both young objects having lower surface gravities ( = 4.04.5 dex ) than field objects ( , though the number of fitted field objects is small ) .
most strikingly , the fitted radii are implausibly small ( @xmath60.8 ) , indicating discord between the data and models even though the quality of fit seems adequate by eye .
a similar radius mismatch occurs in model atmosphere fitting of young dusty planets ( e.g. , @xcite ; cf . , @xcite ) .
there is no radial velocity ( rv ) for pso j318.5@xmath022 , but we can place modest constraints on its space motion .
published rvs find that low - gravity late - m and l dwarfs reside in the range of [ @xmath020 , + 25 ] , with the smaller rv range compared to field objects being expected given the young ages .
figure [ fig : uvwxyz ] plots the spatial ( @xmath24 ) and kinematic ( @xmath25 ) location of pso j318.5@xmath022 compared to the young moving groups ( ymgs ) from @xcite . and @xmath26 are positive toward the galactic center , @xmath27 and @xmath28 are positive toward the direction of galactic rotation , and @xmath29 and @xmath30 are positive toward the north galactic pole . ]
its @xmath24 position is coincident with the pic , tuc - hor , and ab dor moving groups .
however , the velocity information makes the tuc - hor and ab dor groups implausible , given the large offset from pso j318.5@xmath022 .
overall , the match with the pic members is excellent , i.e. , the velocity offset between the group and pso j318.5@xmath022 would be only 3 if pso j318.5@xmath022 has an rv of @xmath06 .
we also computed the kinematic distance ( @xmath31 ) and angle @xmath32 from @xcite .
@xmath31 is an object s distance assuming that its tangential velocity ( as derived from its observed proper motion ) is the same as the mean tangential velocity of a ymg . including the uncertainties in the picgroup s @xmath25 , pso j318.5@xmath022 has @xmath33 pc , in good agreement with its parallactic distance .
@xmath32 gives the angle between the object s proper motion and that of the ymg at the object s sky position .
pso j318.5@xmath022 has @xmath34 , whereas the pic members from @xcite mostly have @xmath35 degs . in other words , for pso j318.5@xmath022 the amplitude and direction of its tangential velocity given its sky location agrees well with the space motion of known pic members .
finally , we computed the ymg membership probability using the online calculator of @xcite . given the position and parallax of pso j318.5@xmath022 , their bayesian method reports a membership probability of 99.9% in the pic ymg and 0.01% in the field .
( removing the parallax as an input , their method gives 99.6% and a predicted distance of [email protected] pc , in good agreement with our parallactic distance . )
the absolute value of this method s probabilities is imperfect , since their input model for the solar neighborhood does not account for the relative numbers of objects in different ymgs and the field .
however , the high probability for pso j318.5@xmath022 is comparable to the values computed for most of the known pic members ( and exceeds the rest of them ) .
the malo et al .
method also does not consider the fact that pso j318.5@xmath022 is spectroscopically young , which would boost the probability that it is a ymg member rather than a field object ( since most field objects have old ages ) .
pso j318.5@xmath022 , 2mass j0355 + 11 , and 2mass j1207@xmath039b are all moving group members and thus play an important role as `` age benchmarks '' ( e.g. * ? ? ?
* ; * ? ? ?
* ) for very red l dwarfs . with a measured and an age estimate
, we use the @xcite evolutionary models with @xmath36 to compute the remaining physical properties .
( results from the other commonly available evolutionary models are similar , comparable to the uncertainties . ) to derive , we combined our 0.92.4 low - resolution spectra flux - calibrated using the @xmath37 photometry , the _ wise _
@xmath38 and @xmath39 photometry , and the best - fitting atmospheric model ( section [ sec : spectra ] ) for the flux at bluer and redder wavelengths .
we integrated the fluxes and used the parallaxes to determine , with the uncertainties in the input data ( spectra , photometry , and parallax ) handled in a monte carlo fashion .
we find @xmath40 dex and @xmath41 dex for pso j318.5@xmath022and 2mass j0355 + 11 , respectively , very similar to field mid / late - l dwarfs @xcite .
young , very red l dwarfs are sometimes ( and confusingly ) described as `` underluminous '' compared to field objects , but we find that the luminosities of young and old field objects are comparable.39b can be described as `` underluminous , '' as its luminosity ( @xmath42 dex ; * ? ? ?
* ) is somewhat low relative to field l dwarfs . ]
rather , these objects have fainter absolute magnitudes in the bluer near - ir bands ( e.g. , @xmath3 band ) but comparable s
. they are therefore better described as being `` very red . '' for pso
j318.5@xmath022 , using an age of 12@xmath43 myr ( distributed uniformly ) from @xcite , we find a mass of @xmath44 , temperature of 1160@xmath4 k , and of 3.86@xmath45 dex ( figure [ fig : evolmodels ] ) .
adopting a larger age range of 10100 myr ( distributed uniformly ) , the mass becomes 12@xmath13 , with and increasing only slightly ( table 1 ) .
such cool temperatures correspond to a field @xmath6t5.5 dwarf @xcite and yet pso j318.5@xmath022 shows no sign of in its spectrum , similar to the situation for the planets around hr 8799 and 2mass j1207@xmath039 . .
the evolutionary - model radii are almost twice as large as those from our model atmosphere fits , as has previously been noted in analysis of directly imaged planets ( section [ sec : spectra ] ) .
our same calculations for 2mass j0355 + 11 give a mass of @xmath46 , temperature of 1420@xmath47 k , and of @xmath48 dex , assuming an age of 125@xmath125 myr @xcite based on its membership in the ab dor moving group @xcite .
this object is more massive and hotter than pso j318.5@xmath022 , as expected given its brighter absolute magnitudes and older age .
@xcite describe 2mass j0355 + 11 as having a strong resemblance to a giant exoplanet .
while it does have similar @xmath2 and @xmath18-band spectra to 2mass j1207@xmath039b ( but not in @xmath3 band nor the overall sed shape ; figure [ fig : spectra ] ) , 2mass j0355 + 11 is @xmath62 mag brighter in @xmath3 band ( figure [ fig : colorcolor ] ) and @xmath65@xmath16 more massive than 2mass j1207@xmath039b .
likewise , the spectrum of hr 8799b has been shown to resemble some field dwarfs ( e.g. , 2mass j2148 + 40 and 2mass j2244 + 20 by @xcite and 2mass j2139 + 02 by @xcite ) , yet hr 8799b is much fainter and lower mass .
this strengthens our finding that _ near - ir spectral morphology can not reliably determine the underlying physical properties of very red young objects_. the temperatures for these two objects are substantially lower then those measured in the same fashion ( using , age , and evolutionary models ) for field l dwarfs of comparable spectral types .
@xcite find of 17001950 k for spectral types l3l5 ( corresponding to the near - ir and optical types for 2mass j0355 + 11 ) and 1500 k for l7 ( the near - ir type for pso j318.5@xmath022 ) .
thus , the temperatures of these young , very red l dwarfs are @xmath6400 k cooler than comparable field objects .
@xcite show that young l - type companions tend to have cooler temperatures than ( old ) field objects of the same type .
our results show that this offset also occurs for young free - floating objects .
pso j318.5@xmath022 shares a strong physical similarity to the young dusty planets hr 8799bcd and 2mass j1207@xmath039b , as seen in its colors , absolute magnitudes , spectrum , luminosity , and mass .
most notably , it is the first field l dwarf with near - ir absolute magnitudes as faint as the hr 8799 and 2mass j1207@xmath039 planets , _ demonstrating that the very red , faint region of the near - ir color - magnitude diagram is not exclusive to young exoplanets . _ its probable membership in the pic
moving group makes it a new substellar benchmark at young ages and planetary masses .
we find very red , low - gravity l dwarfs have @xmath6400 k cooler temperatures relative to field objects of comparable spectral type , yet have similar luminosities .
comparing very red l dwarf spectra to each other and to directly imaged planets highlights the challenges of diagnosing physical properties from near - ir spectra .
pso j318.5@xmath022 is among the lowest - mass free - floating objects identified in the solar neighborhood .
@xcite have determined parallaxes and luminosities of field y dwarfs and thereby estimate masses of 720 , assuming ages of 15 gyr .
the ( likely y dwarf ) companion wd 0806@xmath0661b has a precise age from its white dwarf primary , leading to a mass of 610 @xcite . @xcite
have identified a candidate young late - t dwarf member of the ab dor moving group with an estimated mass of 47 , though a parallax and rv are needed to better determine its properties and membership ( and thus age and mass ) .
pso j318.5@xmath022 was not discovered by previous large - area searches for l dwarfs using 2mass .
pso j318.5@xmath022 has only tenuous detections in the 2mass @xmath3 and @xmath2 bands , and the resulting colors lie outside the selection criteria of @xcite .
additionally , the object is too far south to be detected by sdss , and it is not in the areas surveyed so far by ukidss or vista .
however , both the ps1 and vista datasets are growing relative to what has been searched here , both in depth and area , and upcoming surveys ( e.g. , lsst ) will have even greater reach .
thus , wide - field surveys mapping the whole sky offer a promising avenue for understanding exoplanets directly imaged in the tiny areas around the brightest stars .
the pan - starrs1 surveys have been made possible by the institute for astronomy , the university of hawaii , the pan - starrs project office , the institutions of the pan - starrs1 science consortium ( http://www.ps1sc.org ) , nsf , and nasa .
we thank brendan bowler for assistance with the figures and michael cushing for providing a pre - release update of spextool .
our research has employed the _ wise _ and 2mass data products ; nasa s astrophysical data system ; and the spex prism spectral libraries maintained by adam burgasser .
this research was supported by nsf grants ast09 - 09222 ( awarded to mcl ) and ast-0709460 ( awarded to eam ) as well as afrl cooperative agreement fa9451 - 06 - 2 - 0338 finally , the authors wish to recognize 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 .
, f. , homeier , d. , & freytag , b. 2011 , in astronomical society of the pacific conference series , vol .
448 , 16th cambridge workshop on cool stars , stellar systems , and the sun , ed . c. johns - krull , m. k. browning , & a. a. west , 91 elias , j. h. , joyce , r. r. , liang , m. , muller , g. p. , hileman , e. a. , & george , j. r. 2006 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol
. 6269 , society of photo - optical instrumentation engineers ( spie ) conference series rayner , j. t. , toomey , d. w. , onaka , p. m. , denault , a. j. , stahlberger , w. e. , watanabe , d. y. , & wang , s. 1998 , in proc .
spie : infrared astronomical instrumentation , ed .
a. m. fowler , vol .
3354 , 468479 -0.6 in 22 ( red circle ) compared to known ymgs from @xcite . for the @xmath25 plots , we adopt an rv of ( section [ sec : membership ] ) , with the larger red circles being more positive velocities and the red line being the full rv range .
uncertainties arising from errors in the distance and proper motion ( but not the rv ) are show at the 3 red circles .
the members of the nearest ymgs to pso j318.5@xmath022 are shown in color and the others in grey , with the ellipses representing the rms of the known members .
consideration of both the @xmath24 and @xmath25 data indicate the pic group is the only possibility .
the middle ( black - outlined ) circle shows the @xmath25 position of pso j318.5@xmath022 for an rv of @xmath06 .
[ fig : uvwxyz],title="fig:",width=499 ] -0.5 in -3 in -2 in 22 and known young substellar objects from @xcite and the compilation of @xcite .
the @xcite cloudless evolutionary models are plotted in color , with 0.0020.010 tracks in green ( spaced by 0.002 ) , 0.011 - 0.014 tracks in yellow ( where d - burning may complicate the interpretation and spaced by 0.001 ) , and @xmath490.016 in grey ( spaced by 0.002 up to 0.030 and then by 0.010 ) .
the thin light grey lines show the cloudy evolutionary models with @xmath36 .
[ fig : evolmodels],title="fig:",width=768 ] -1.3 in 2mass & 2mass j21140802@xmath02251358 + ps . &
pso [email protected] + _ wise _ & wise [email protected] + 2mass ra , dec ( ep 1999.34 ) & 318.53344 , @xmath022.85996 + ps .
ra , dec ( ep 2010.0 ) & 318.53380 , @xmath022.86032 + _ wise _ ra , dec ( ep 2010.34 ) & 318.53388 , @xmath022.86037 + proper motion @xmath50 ( mas / yr ) & 137.3 @xmath1 1.3 , @xmath0138.7 @xmath1 1.4 + proper motion amplitude @xmath20 ( /yr ) & 195.0 @xmath1 1.3 + proper motion pa ( ) & 135.3 @xmath1 0.4 + parallax @xmath51 ( mas ) & 40.7 @xmath1 2.4 + distance @xmath52 ( pc ) & 24.6 @xmath1 1.4 + @xmath53 ( km / s ) & 22.7 @xmath1 1.3 + ps1 @xmath54 ( ab mag ) & @xmath5523.6 ( 3@xmath56 ) + ps1 @xmath57 ( ab mag ) & @xmath5523.1 ( 3@xmath56 ) + ps1 @xmath58 ( ab mag ) & @xmath5522.9 ( 3@xmath56 ) + ps1 @xmath59 ( ab mag ) & 20.80 @xmath1 0.09 + ps1 @xmath60 ( ab mag ) & 19.51 @xmath1 0.07 + 2mass @xmath3 ( mag ) & 16.71 @xmath1 0.20 + 2mass @xmath2 ( mag ) & 15.72 @xmath1 0.17 + 2mass @xmath61 ( mag ) & 14.74 @xmath1 0.12 + mko @xmath28 ( mag ) & 18.81 @xmath1 0.10
+ mko @xmath3 ( mag ) & 17.15 @xmath1 0.04 + mko @xmath2 ( mag ) & 15.68 @xmath1 0.02 + mko @xmath18 ( mag ) & 14.41 @xmath1 0.02 + wise @xmath38 ( mag ) & 13.22 @xmath1 0.03 + wise @xmath39 ( mag ) & 12.46 @xmath1 0.03 + wise @xmath62 ( mag ) & 11.8@xmath1 0.4 + wise @xmath63 ( mag ) & @xmath558.6 ( 2@xmath56 ) + 2mass @xmath64 ( mag ) & 1.678 @xmath1 0.007 ( 0.06 ) + 2mass @xmath65 ( mag ) & 1.159 @xmath1 0.004 ( 0.03 ) + 2mass @xmath66 ( mag ) & 2.837 @xmath1 0.006 ( 0.07 ) + mko @xmath67 ( mag ) & 1.37@xmath1 0.02(0.05 ) + mko @xmath64 ( mag ) & 1.495 @xmath1 0.007 ( 0.06 ) + mko @xmath68 ( mag ) & 1.279 @xmath1 0.004 ( 0.03 ) + mko @xmath69 ( mag ) & 2.775 @xmath1 0.007 ( 0.07 ) + @xmath70 ( mag ) & 0.111
@xmath1 0.002 + @xmath71 ( mag ) & @xmath00.072 @xmath1 0.001 + @xmath72 ( mag ) & 0.049 @xmath1 0.001 + log(/ ) ( dex ) & @xmath04.42 @xmath1 0.06 + @xmath3-band type ( spex , gnirs ) & l8 @xmath1 1 , l9 @xmath1 1 + @xmath2-band type ( spex , gnirs ) & l6 @xmath1 1 , l6 @xmath1 1 + h@xmath19od type ( spex , gnirs ) & l6.0 @xmath1 0.9 , l6.0 @xmath1 0.8 + gravity score ( spex , gnirs ) & xxx2 , 2x21 + final near - ir spectral type & l7 @xmath1 1 + near - ir gravity class & vl - g + mass ( ) & 6.5 ( @xmath01.0 , + 1.3 ) + @xmath73 ( k ) & 1160 ( @xmath040 , + 30 ) + @xmath74 ( cgs ) & 3.86 ( @xmath00.08 , + 0.10 ) + radius ( ) & 1.53 ( @xmath00.03 , + 0.02 ) + mass ( ) & 12 @xmath1 3 + @xmath73 ( k ) & 1210 ( @xmath050 , + 40 ) + @xmath75 ( cgs ) & 4.21 ( @xmath00.16 , + 0.11 ) + radius ( ) & 1.40 ( @xmath00.04 , + 0.06 ) + | we have discovered using ps . an extremely red late - l dwarf , which has and , making it the reddest known field dwarf and second only to 2mass j1207@xmath039b among substellar companions .
near - ir spectroscopy shows a spectral type of l7@xmath11 and reveals a triangular @xmath2-band continuum and weak alkali ( and ) lines , hallmarks of low surface gravity .
near - ir astrometry from the hawaii infrared parallax program gives a distance of [email protected] pc and indicates a much fainter @xmath3-band absolute magnitude than field l dwarfs .
the position and kinematics of pso j318.5@xmath022 point to membership in the pic moving group .
evolutionary models give a temperature of 1160@xmath4 k and a mass of 6.5@xmath5 , making pso j318.5@xmath022 one of the lowest mass free - floating objects in the solar neighborhood .
this object adds to the growing list of low - gravity field l dwarfs and is the first to be strongly deficient in methane relative to its estimated temperature .
comparing their spectra suggests that young l dwarfs with similar ages and temperatures can have different spectral signatures of youth .
for the two objects with well constrained ages ( pso j318.5@xmath022 and 2mass j0355 + 11 ) , we find their temperatures are @xmath6400 k cooler than field objects of similar spectral type but their luminosities are comparable , i.e. , these young l dwarfs are very red and unusually cool but not `` underluminous . ''
altogether , pso j318.5@xmath022 is the first free - floating object with the colors , magnitudes , spectrum , luminosity , and mass that overlap the young dusty planets around hr 8799 and 2mass j1207@xmath039 . |
the experimental achievement of quantum degeneracy in a dilute trapped gas of cold fermionic atoms @xcite has stimulated theoretical interest in the properties of this fundamental system .
attention has focused on the critical temperature @xcite and the detection @xcite of cooper pairing as well as on the properties of mixtures of various fermionic and bosonic species @xcite .
another important problem concerns the interactions between ultracold fermions @xcite . owing to the exclusion principle , spin - polarized fermi atoms
do not interact via s - wave collisions , whereas they dominate in the low - energy regime for bosons and have pronounced effects on the statics and dynamical properties of cold boson gases @xcite .
hence , in the absence of low - energy collisions other types of forces come into play .
a good candidate is a dipole - dipole interaction between atoms or molecules , not analyzed so far in the context of cold trapped fermions .
some atoms possess permanent magnetic dipole moments of considerable magnitude ( chromium , for instance , has @xmath0 ) .
it was also proposed to induce electric dipoles in atoms @xcite .
huge permanent electric dipole moments occur naturally in diatomic polar molecules @xcite .
the behavior of atomic bosonic dipoles in traps has been investigated in @xcite , which addressed the question of instabilities in the system caused by an attractive component of dipolar interactions .
the conclusion drawn was that a large enough positive scattering length ( providing repulsive interactions ) can stabilize a system of bosonic dipoles . for strong ( e.g. molecular ) dipoles , when supposedly the scattering length can be neglected , it is the trap geometry that plays a crucial role @xcite the system is stable provided the trap assures the domination of repulsive interactions in the gas .
a full quantum mechanical description of the system of many interacting fermions is of course very complex .
but the lessons of semiclassical atomic physics can be applied , in particular the thomas - fermi approach @xcite and its refinements ( see , e.g. , @xcite ) .
its success in describing static properties of atoms is well known , and we note that , in recent years , these methods were also used successfully for studying dynamical processes of atoms and molecules in superstrong light pulses @xcite .
our paper is organized as follows : in section ii , the thomas - fermi model is revisited with an eye on the dipole - dipole forces . the dirac correction to the interaction energy term is discussed and scaling properties derived with the help of the virial theorem . in section iii the results of approximate ( nearly analytic ) and numerical minimizations of the thomas - fermi - dirac energy functional are presented .
unexpectedly , we find that the system may withstand larger dipolar forces both for very flat and for highly elongated traps .
let s begin by recalling some basic things , mainly collected from refs .
@xcite , with due attention to the changed situation : here fully spin - polarized fermions there fermions with no net spin .
the spatial one - particle density is denoted by @xmath1 , it is normalized to the total particle number @xmath2 , @xmath3 where @xmath4 denotes the volume element .
the spatial one - particle density matrix @xmath5 and the one - particle wigner function @xmath6 are related by @xmath7 the spatial and momental one - particle densities are obtained by integrating @xmath6 over the other variable , @xmath8 they are needed for the calculation of the kinetic energy , @xmath9 and the external potential energy ( of the harmonic trap ) , @xmath10 n({\!\vec{\,r}})\,,\ ] ] where @xmath11 is the mass of the atom species considered , @xmath12 is the ( transverse ) trap frequency , and @xmath13 is the aspect ratio of the cylindrical trap .
the trap is spherically symmetric for @xmath14 ; for @xmath15 , the equipotential surfaces are prolate ( `` cigar shaped '' ) ellipsoids ; for @xmath16 , they are oblate ( `` lentil shaped '' ) ellipsoids . for the dipole - dipole interaction energy , @xmath17 , we need ( the diagonal part of ) the two - particle density matrix @xmath18 , @xmath19 where @xmath20\,.\ ] ] we note that the contact term , proportional to @xmath21 , is required by the condition that the magnetic field made by the point dipole be divergence - free @xcite . an alternative way of presenting @xmath22 is @xmath23\cdot{\!\vec{\,\mu}}\,,\ ] ] which is a particularly convenient starting point for evaluating the fourier transform @xmath24\cdot{\!\vec{\,\mu}}\,.\ ] ] the vanishing divergence just mentioned is here immediately recognized , inasmuch as @xmath25=0 $ ] .
the semiclassical approximation now employed in the spirit of what the tfd trio ( thomas @xcite , fermi @xcite , and dirac @xcite ) did , although in a technically different manner is two - fold : we approximate @xmath26 by products of @xmath27 factors ( dirac ) , @xmath28 and @xmath27 by a brutally simple wigner function ( thomas and fermi ) , @xmath29^{\frac{2}{3}}-{\!\vec{\,p}}^2\bigr)\,,\ ] ] where @xmath30 is heaviside s unit step function .
this gives the density functional of the kinetic energy as @xmath31=\int({{\mathrm{d}}}{\!\vec{\,r}})\,\frac{\hbar^2}{m}\frac{1}{20\pi^2 } \bigl[6\pi^2n({\!\vec{\,r}})\bigr]^{\frac{5}{3}}\,,\ ] ] and the density functional for the potential energy in the trap is @xmath32 of ( [ eq : etrap1 ] ) .
the dipole - dipole interaction energy consists of two parts , corresponding to the two summands in ( [ eq : n2tfd ] ) , @xmath33={e_{\mathrm{dd}}}^{\mathrm{(dir)}}[n]+{e_{\mathrm{dd}}}^{\mathrm{(ex)}}[n]\ ] ] with the direct term @xmath34={\frac{1}{2}}\int({{\mathrm{d}}}{\!\vec{\,r}}')({{\mathrm{d}}}{\!\vec{\,r } } '' ) n({\!\vec{\,r}}'){v_{\mathrm{dd}}}({\!\vec{\,r}}'-{\!\vec{\,r}}'')n({\!\vec{\,r}}'')\ ] ] and the exchange term @xmath35&=&-{\frac{1}{2}}\int({{\mathrm{d}}}{\!\vec{\,r}}')({{\mathrm{d}}}{\!\vec{\,r}}'')\ , n^{(1)}({\!\vec{\,r}}';{\!\vec{\,r}}'')n^{(1)}({\!\vec{\,r}}'';{\!\vec{\,r } } ' ) \nonumber\\ & & \hphantom{-{\frac{1}{2}}\int({{\mathrm{d}}}{\!\vec{\,r}}')({{\mathrm{d}}}{\!\vec{\,r}}'')}\quad\times { v_{\mathrm{dd}}}({\!\vec{\,r}}'-{\!\vec{\,r } } '' ) \nonumber\\ & = & -{\frac{1}{2}}\int({{\mathrm{d}}}{\!\vec{\,r}})\int({{\mathrm{d}}}{\!\vec{\,s}})\,{v_{\mathrm{dd}}}({\!\vec{\,s}})\nonumber\\ & & \times n^{(1)}({\!\vec{\,r}}+{{\textstyle\frac{1}{2}}}{\!\vec{\,s}};{\!\vec{\,r}}-{{\textstyle\frac{1}{2}}}{\!\vec{\,s } } ) n^{(1)}({\!\vec{\,r}}-{{\textstyle\frac{1}{2}}}{\!\vec{\,s}};{\!\vec{\,r}}+{{\textstyle\frac{1}{2}}}{\!\vec{\,s}})\ , .
\nonumber\\ \label{eq : edd4}\end{aligned}\ ] ] now we note that @xmath36 depends only on the length @xmath37 of vector @xmath38 , not on its direction @xmath39 , because in the tf approximation ( [ eq : wigtf ] ) the product @xmath40 involves only @xmath41 and @xmath42 . as a consequence ,
it is permissible to replace , in ( [ eq : edd4 ] ) , @xmath43 by its average over the solid angle associated with @xmath38 , @xmath44\,.\ ] ] we thus arrive at @xmath45&=&{\frac{1}{2}}\int({{\mathrm{d}}}{\!\vec{\,r}})\left[n({\!\vec{\,r}})\right]^2 \,\frac{\mu_0}{4\pi}\,\frac{8\pi}{3}\,{\!\vec{\,\mu}}^2 \nonumber\\ & = & { \frac{1}{2}}\int({{\mathrm{d}}}{\!\vec{\,r}})({{\mathrm{d}}}{\!\vec{\,r}}')n({\!\vec{\,r } } ) \frac{\mu_0}{4\pi}\left [ \frac{8\pi}{3}{\!\vec{\,\mu}}^2\delta({\!\vec{\,r}}-{\!\vec{\,r}}')\right]n({\!\vec{\,r}}')\ , , \nonumber\\ \label{eq : edd6}\end{aligned}\ ] ] and accordingly @xmath46={\frac{1}{2}}\int({{\mathrm{d}}}{\!\vec{\,r}})({{\mathrm{d}}}{\!\vec{\,r}}')\ , n({\!\vec{\,r}}){\overline{v}_{\mathrm{dd}}}({\!\vec{\,r}}-{\!\vec{\,r}}')n({\!\vec{\,r}}')\ ] ] with @xmath47\,,\ ] ] which is @xmath48 of ( [ eq : vdd1 ] ) with the contact term removed . in ( [ eq : vdd2 ] ) and ( [ eq : vdd3 ] ) this corresponds to multiplying the unit dyadic @xmath49 by @xmath50 @xcite . the rotational symmetry of the trap potential in ( [ eq : etrap1 ] ) distinguishes the @xmath51 axis , and we take for granted that this is also the direction of spin polarization , @xmath52 then @xmath53 and the whole system is invariant under rotations around the @xmath51 axis .
the total tfd energy functional is then given by the sum of the kinetic energy ( [ eq : ekintfd ] ) , the potential energy in the trap ( [ eq : etrap1 ] ) , and the dipole - dipole interaction energy ( [ eq : edd7 ] ) , @xmath54 & = & \!\int\!({{\mathrm{d}}}{\!\vec{\,r } } ) \biggl[\frac{\hbar^2}{m}\frac{1}{20\pi^2 } \bigl[6\pi^2n({\!\vec{\,r}})\bigr]^{\frac{5}{3 } } + { { \textstyle\frac{1}{2}}}m\omega^2r^2n({\!\vec{\,r}})\biggr ] \nonumber\\ & & + { \frac{1}{2}}\int({{\mathrm{d}}}{\!\vec{\,r}})({{\mathrm{d}}}{\!\vec{\,r}}')n({\!\vec{\,r}}){\overline{v}_{\mathrm{dd}}}({\!\vec{\,r}}-{\!\vec{\,r } } ' ) n({\!\vec{\,r}}')\ , .
\label{eq : etottfd}\end{aligned}\ ] ] the density that minimizes @xmath55 under the constraint ( [ eq : n ] ) obeys the nonlinear integral equation @xmath56^{\frac{2}{3 } } + { { \textstyle\frac{1}{2}}}m\omega^2\bigl[x^2+y^2+(\beta z)^2\bigr ] \nonumber\\&&\qquad\qquad+\int({{\mathrm{d}}}{\!\vec{\,r}}'){\overline{v}_{\mathrm{dd}}}({\!\vec{\,r}}-{\!\vec{\,r } } ' ) n({\!\vec{\,r } } ' ) = { { \textstyle\frac{1}{2}}}m\omega^2r^2\ , , \label{eq : nonlininteq}\end{aligned}\ ] ] where @xmath57 is a convenient way of writing the lagrange multiplier for the constraint .
scaling transformations of the form @xmath58 are consistent with the constraint ( [ eq : n ] ) .
they affect the various pieces of @xmath55 in accordance with @xmath59 so that ( @xmath60 ) @xmath61 in the infinitesimal vicinity of @xmath62 , all first - order changes of @xmath63 originate in the explicit change of @xmath2 , @xmath64 , and therefore @xmath65 must hold irrespective of the value of parameter @xmath66 . in view of the linear @xmath66 dependence ,
we get two independent statements , @xmath67 they enable us to express @xmath68 , @xmath32 , and @xmath17 in terms of @xmath63 and @xmath69 , @xmath70 in conjunction with ( @xmath71 ) @xmath72 they imply that the ground state energy @xmath73 is of the form ( we leave the @xmath13 dependence implicit ) @xmath74 with @xmath75 clearly , the dimensionless number @xmath76 measures the relative strength of the dipole - dipole interaction .
the universal function @xmath77 is to be found numerically ( for each @xmath13 value of interest ) . for @xmath78
, we can solve ( [ eq : nonlininteq ] ) immediately , @xmath79^{\frac{3}{2}}\,,\ ] ] where , by convention , @xmath80^{\frac{3}{2}}=0 $ ] for negative arguments , and @xmath81 as a consequence of ( [ eq : n ] ) , and so we find @xmath82 for the spherically symmetric case @xmath14 , @xmath17 vanishes in first - order perturbation theory , so that @xmath83 with @xmath84 .
these scaling laws invite the use of correspondingly chosen dimensionless variables , such as @xmath85 for the position and @xmath86 for the density .
the constraint ( [ eq : n ] ) then appears as @xmath87 and the tfd energy acquires the form @xmath88&= & \frac{3}{10}(6\pi^2)^{\frac{2}{3 } } \int({{\mathrm{d}}}{\!\vec{\,x}})\ , [ g({\!\vec{\,x}})]^{\frac{5}{3 } } \nonumber \\ & & + \frac{1}{2 } \int ( { { \mathrm{d}}}{\!\vec{\,x}})\ , \bigl(x_1 ^ 2+x_2 ^ 2+\beta^2x_3 ^ 2\bigr)g({\!\vec{\,x } } ) \nonumber\\ & & + { \frac{1}{2}}n^{\frac{1}{6}}\varepsilon \int ( { { \mathrm{d}}}{\!\vec{\,x}})({{\mathrm{d}}}{\!\vec{\,x}}')\,g({\!\vec{\,x } } ) \frac{1 - 3\cos^{2}\theta}{|{\!\vec{\,x}}-{\!\vec{\,x}}'|^3}g({\!\vec{\,x}}')\ , , \nonumber\\ \label{eq : scaledetfd}\end{aligned}\ ] ] where @xmath89 is the angle between the polarization direction ( the @xmath90 direction ) and the relative position vector @xmath91 , @xmath92 the minimum of the scaled tfd energy is @xmath93 of ( [ eq : scale8a ] ) ( with its implicit @xmath13 dependence ) ; it is obtained for the @xmath94 that obeys the dimensionless analog of ( [ eq : nonlininteq ] ) , @xmath95^{\frac{2}{3 } } + { { \textstyle\frac{1}{2}}}\bigl(x_1 ^ 2+x_2 ^ 2+\beta^2x_3 ^ 2\bigr ) \nonumber\\&&\qquad\qquad+n^{\frac{1}{6}}\varepsilon\int({{\mathrm{d}}}{\!\vec{\,x } } ' ) \frac{1 - 3\cos^{2}\theta}{|{\!\vec{\,x}}-{\!\vec{\,x}}'|^3}g({\!\vec{\,x } } ' ) = { { \textstyle\frac{1}{2}}}x^2\ , , \label{eq : nonlininteq'}\end{aligned}\ ] ] where the value of @xmath96 is determined by the constraint ( [ eq : n ] ) .
owing to its intrinsic semiclassical approximations , one expects a few - percent deviation of the tfd energy from the true ground - state energy .
it is , therefore , not really necessary to solve the nonlinear integral equation ( [ eq : nonlininteq ] ) . a reasonable variational estimate , in conjunction with a few full - blown numerical solutions for comparison , will do .
the gaussian ansatz @xmath97 is convenient .
in addition to its aspect ratio @xmath98 , the shape parameter , it contains the scale parameter @xmath99 , so that the virial theorems of sec .
[ sec : virial ] will be obeyed for the optimal choice of @xmath99 . for this scaled density ,
the scaled kinetic energy is given by @xmath100 and the scaled value of @xmath32 is @xmath101 they exhibit the anticipated dependence on the scale parameter @xmath99 , and so does the dipole - dipole interaction energy , @xmath102 this integral can , of course , be evaluated in terms of elementary functions , but as it stands we see immediately that the integrand , and thus the integral , is positive for @xmath103 , so that @xmath104{r@{\enskip\mbox{for}\enskip}l@{\quad}l } { e_{\mathrm{dd}}}>0 & \gamma>1 & \text{(oblate gaussian)},\\[1ex ] { e_{\mathrm{dd}}}=0 & \gamma=1 & \text{(spherical gaussian)},\\[1ex ] { e_{\mathrm{dd}}}<0 & \gamma<1 & \text{(prolate gaussian ) } , \end{array}\ ] ] consistent with the expectation that an oblate density has a larger magnetic interaction energy than a prolate density . more explicitly , then , we have @xmath105&&\times\left\ { \begin{array}{c@{\enskip\mbox{for}\enskip}l@ { } } \displaystyle\ ! \frac{1-\vartheta\cot\vartheta}{\sin^2\vartheta}-{\frac{1}{3}}&\displaystyle \gamma=\frac{1}{\cos\vartheta}>1\,,\\[3ex ] \displaystyle\ ! \frac{\vartheta\coth\vartheta-1}{\sinh^2\vartheta}-{\frac{1}{3}}&\displaystyle \gamma=\frac{1}{\cosh\vartheta}<1\ ,
. \end{array}\right .
\nonumber\\ \label{eq : gauss - dd'}\end{aligned}\ ] ] the gaussian density ( [ eq : gauss ] ) can not mimic the @xmath106 solution ( [ eq : n0 ] ) very well .
but nevertheless , the resulting estimate of the @xmath106 energy , @xmath107 is only @xmath108 in excess of the correct value ( [ eq : e0 ] ) .
this accuracy is sufficient for our purposes ; and , in any case , an error of a few percent is a small price for the enormous simplification that the gaussian ansatz brings about .
dipolar interactions are partially attractive and partially repulsive , depending on the configuration of the dipoles .
one should keep in mind the simple situation of two dipoles in the plane perpendicular to their polarization , which repel each other , as opposed to the situation of two attracting dipolar particles placed along the direction of their polarization . extending this picture to a cloud of trapped dipoles
one would expect that attraction dominates in prolate traps , and repulsion in oblate traps ( provided the dipoles are polarized along the trap axis as is the case here ) . in the case of predominant attraction one may surmise that instabilities occur . by varying the two system parameters ( @xmath109 and @xmath13 ) we have investigated the issue of stability , see fig .
[ fig : stability ] . from this stability diagram one concludes that , for oblate traps ( @xmath16 ) , the bigger the trap aspect ratio , the bigger values of the dipole parameter @xmath109 can be stabilized .
in fact , we found numerically that fermions form stable configurations for @xmath110 irrespective of the strength of their dipole interaction ( an analogous effect was observed for dipolar bosons by santos _
@xcite ) . on the other hand
, one might naively expect an exactly opposite effect in prolate traps ( @xmath15 ) where attractive interactions should dominate .
this is not quite true indeed for moderate trap aspect ratios ( @xmath111 ) the critical value of the dipole parameter is smaller , but in traps that are very soft ( @xmath112 ) in the @xmath51 direction of rotational symmetry we observe an increase of the critical value of @xmath109 ( see the inset in fig . [
fig : stability ] ) .
this can be understood with the help of the following argument .
we note that the dipole - dipole energy term vanishes for a uniform density distribution .
as the trap is made softer in the polarization direction , the shape of the cloud along the soft axis becomes more and more uniform , contributing to the interaction energy to a lesser extent ( this argument was also used to interpret our earlier results for bosons interacting via contact and dipole - dipole forces , see @xcite ) .
the dependence of @xmath98 and @xmath99 on @xmath13 , shown in fig .
[ fig : cloudparams ] , is consistent with this argument .
we see that @xmath98 decreases with decreasing @xmath13 whereas @xmath99 increases .
accordingly , the cloud gets stretched along the @xmath51 axis of symmetry , and the diameter of the circular cross section in the @xmath113 plane ( @xmath114 ) is reduced . at the center of the cloud , we thus have a relatively large volume of ( almost ) constant density , and the inhomogeneous parts of the cloud are relatively far apart . taken together
, these geometric features lead to a rather small dipole - dipole interaction energy . whereas the stability of the system considered
can be well understood , the spatial behavior of the fermionic cloud , especially near the collapse , seems to be much less intuitive . as we approach the critical parameter values , the aspect ratio @xmath98 of the cloud decreases and the cloud becomes elongated in the attractive @xmath51 direction .
this type of behavior is general in the sense that it does not depend on the trap aspect ratio , see fig .
[ fig : gammaofd ] .
it is only for traps with @xmath110 that @xmath98 reaches an asymptotic value ( dependent on @xmath13 ) for extremely large values of @xmath109 .
the dependence of the dipolar energy @xmath17 on the dipole parameter @xmath109 and the trap geometry is also of interest as it is the quantity responsible for the ( in)stability of the system , see fig .
[ fig : eddofd ] . for all prolate traps ,
@xmath17 remains negative approaching some critical value at the collapse point . for @xmath115
, the dipolar energy can be positive for moderate dipole parameters , but if their values are large enough , @xmath17 turns negative indicating the collapse . for @xmath110 , @xmath17 is always positive and increases as a function of @xmath109 . finally , we take a look at the total tfd energy , see fig . [
fig : etfdofd ] . for @xmath115 ,
the system becomes unstable at the critical value of @xmath109 .
consistent with fig .
[ fig : stability ] , we observe that larger values of @xmath109 are supported for @xmath116 and @xmath14 than for @xmath117 .
let us now discuss the dipole parameter @xmath109 and give some typical values for it .
owing to the @xmath2 dependence , one can locate the experimental system in various regions of the stability diagram of fig .
[ fig : stability ] not only by choosing ( or inducing , as proposed for bosons @xcite ) a specific value of @xmath118 , but also , to some extent , by varying @xmath2 .
one could also exploit the @xmath12 dependence of @xmath119 , which is particularly relevant for optical traps with tight confinement @xcite . for the parameters of the fermionic chromium isotope , @xmath120 hz and @xmath121 , one obtains @xmath122 , which is a very small value . therefore , the conclusion for atoms possessing even relatively large magnetic dipole moments is that irrespective of their number and the trap frequency they will always remain stable against collapse .
the following amusing analogy offers a good reason for this observation @xcite .
let us compare the characteristic sizes of the noninteracting fermi gas and the bose - condensed atomic gas interacting via a repulsive contact potential .
in the thomas - fermi approximation the appropriate quantities , in units of @xmath123 , read : @xmath124 for fermions and @xmath125 for bosons , where @xmath126 is the scattering length . by equating the two sizes
one can calculate the effective , @xmath2 dependent , scattering length due to the exclusion - principle induced repulsion : @xmath127 . for typical numbers of atoms , @xmath128 ,
this @xmath126 is huge on the scale set by typical scattering lengths for bosonic atoms ( @xmath129 ) .
once we realize how strong is the repulsion that originates in the fermi statistics , we understand that small atomic magnetic dipoles can hardly have a noticeable effect on the behavior of dipolar fermionic gases .
however , for polar molecules the situation is different . for a trap frequency of @xmath120hz and @xmath121 molecules of mass @xmath130 a.m.u .
( the typical mass of an alkaline dimer ) and a typical electric dipole moment of 1 debye , one reaches @xmath131 , which may very well put the system into the unstable regime
see fig .
[ fig : stability ] .
in this situation , a sufficiently large @xmath13 value will stabilize the system . in order to assess the quality of our variational results we have computed exact numerical solutions of eq.([eq : nonlininteq ] ) .
this equation was solved for the density distribution @xmath94 self - consistently starting from the known analytical result ( [ eq : n0 ] ) for a non - interacting ( @xmath78 )
fermi gas in a trap @xcite and slowly increasing the dipole parameter @xmath109 . for each value of @xmath109
the solution was iterated until convergence was reached .
then , the value of the dipole parameter was slightly increased . in order to compute the dipole ( integral ) term
we note that it has the form of a convolution .
thus , it can be conveniently evaluated in the fourier space where it is a simple product of the fourier transforms of the density ( computed numerically with the aid of an fft ) and the interaction potential , the latter being known analytically @xcite : @xmath132 where @xmath66 is the angle between the fourier variable @xmath133 and the @xmath51 direction . in order to assure that the integral term is evaluated accurately we used a gaussian distribution for comparison and chose the grid parameters accordingly . our numerical calculations ,
performed in three dimensions , were quite demanding so we limited their use to a check of the main features of the stability diagram in fig .
[ fig : stability ] .
the solutions obtained satisfy the virial relations ( [ eq : scale5 ] ) very well , and total energies obtained numerically are always below the corresponding values from the variational analysis .
the critical value of the dipole parameter for a spherical trap is @xmath134 as compared to @xmath135 obtained by the variational calculation .
we also confirmed the effect of increase of the critical interaction strength that we found , in the gaussian approximation , for prolate traps : for @xmath136 the critical value of @xmath109 is @xmath137 as compared to @xmath138 obtained analytically in the gaussian approximation .
now two remarks about possible extensions of our work are in order .
firstly , the results presented in this paper describe the situation of a dipolar fermionic gas at the temperature @xmath139 .
an interesting subject of study would be extension of our theory to finite temperatures .
secondly , there exists a parallel approach in the thomas - fermi model , namely the one in the momentum space @xcite .
as many experiments with cold gases yield their momental characteristics , investigation of this alternative approach also presents an attractive theoretical task .
e. would like to thank w. schleich for his support in ulm where part of this work was done .
e. is grateful for the kind hospitality extended to him in warsaw .
k. g. acknowledges support by polish kbn grant no 2 p03b 057 15 .
k. r. and k. g. are supported by the subsidy of the foundation for polish science .
part of the results has been obtained using computers at the interdisciplinary centre for mathematical and computational modeling ( icm ) at warsaw university .
m. houbiers _ et al .
a * 56 * , 4864 ( 1997 ) ; d. v. efremov , m. s. marenko , m. a. baranov , m. yu .
kagan , cond - mat/9911169 ; m. a. baranov and d. s. petrov , phys .
a * 58 * , r801 ( 1998 ) ; g. bruun , y. castin , r. dum , and k. burnett , eur .
j. d * 7 * , 433 ( 1999 ) . | we investigate the properties of a degenerate dilute gas of neutral fermionic particles in a harmonic trap that interact via dipole - dipole forces .
we employ the semiclassical thomas - fermi method and discuss the dirac correction to the interaction energy .
a nearly analytic as well as an exact numerical minimization of the thomas - fermi - dirac energy functional are performed in order to obtain the density distribution .
we determine the stability of the system as a function of the interaction strength , the particle number , and the trap geometry .
we find that there are interaction strengths and particle numbers for which the gas can not be trapped stably in a spherically symmetric trap , but both prolate and oblate traps will work successfully . |
now - a - days uranium and thorium isotopes have attracted a great attention in nuclear physics due to the thermally fissile nature of some of its isotopes@xcite .
these thermally fissile materials have tremendous importance in energy production . till date , the known thermally fissile nuclei are @xmath4u , @xmath5u and @xmath6pu . out
of which only @xmath5u has a long life time and the only thermally fissile isotope available in nature @xcite .
thus , presently it is an important area of research to look for any other thermally fissile nuclei apart from @xmath4u , @xmath5u and @xmath6pu . recently ,
satpathy et al .
@xcite showed that uranium and thorium isotopes with neutron number n=154 - 172 have thermally fissile property .
they performed a calculation with a typical example of @xmath7u that this nucleus has a low fission barrier with a significantly large barrier width , which makes it stable against the spontaneous fission .
apart from the thermally fissile nature , these nuclei also play an important role in the nucleosynthesis in the stellar evolution .
as these nuclei are stable against spontaneous fission , thus the prominent decay modes may be the emission of @xmath0- , @xmath1- and @xmath8-particles from the neutron - rich thermally fissile ( uranium and thorium ) isotopes . to measure the stability of these neutron - rich u and th isotopes
, we investigated the @xmath0- and @xmath1- decay properties of these nuclei .
also , we extend our calculations to estimate the binding energy , root mean square radii , quadrupole moments and other structural properties . from last three decades , the relativistic mean field ( rmf ) formalism is a formidable theory in describing the finite nuclear properties throughout the periodic chart and infinite nuclear matter properties concerned with the cosmic dense object like neutron star . in the same line rmf theory is also good enough to study the clusterization @xcite , @xmath0-decay @xcite , and @xmath1-decay of nuclei .
the presence of cluster in heavy nuclei like , @xmath9ra , @xmath10u , @xmath6pu and @xmath11 cm has been studied using rmf formalism @xcite .
it gives a clear prediction of @xmath0-like ( n = z ) matter at the central part for heavy nuclei and @xmath8-like structure ( n = z and @xmath12 ) for light mass nuclei @xcite .
the proton emission as well as the cluster decay phenomena are well studied using rmf formalism with m3y @xcite , lr3y @xcite and nlr3y@xcite nucleon - nucleon potentials in the framework of single and double folding models , respectively . here
, we used the relativistic mean field ( rmf ) formalism with the well known nl3 parameter set @xcite for all our calculations .
the paper is organized as follows : the rmf formalism is outlined briefly in section [ sec2 ] .
the importance of pairing correlation and inclusion with bcs approximation are also given in this section .
the results obtained from our calculations for binding energy , basis selection , potential energy surface ( pes ) diagrams and the evaluation of single - particle levels are discussed in section [ sec4 ] .
the @xmath13- and @xmath14-values are calculated in section [ sec9 ] . in this section ,
various decay modes are discussed using empirical formula and limitation of the model is also given same section . finally , a brief summary and concluding remarks are given in the last section [ sec14 ] .
in present manuscript , we used the axially deformed relativistic mean field formalism to calculate various nuclear phenomena .
the meson - nucleon interaction is given by @xcite @xmath15 where , @xmath16 is the dirac spinor and meson fields are denoted by @xmath17,@xmath18 and @xmath19 for @xmath17,@xmath20 and @xmath21 meson respectively . the electromagnetic interaction between the proton
is denoted by photon field @xmath22 .
@xmath23 , @xmath24 , @xmath25 and @xmath26 are the coupling constants for the @xmath17,@xmath20 and @xmath21 meson and photon field respectively .
the strength of the self coupling @xmath27 meson ( @xmath28 and @xmath29 ) are denoted by @xmath30 and @xmath31 , along with @xmath32 as the non - linear coupling constant for @xmath20 meson .
the nucleon mass is scripted as m , where the @xmath17 , @xmath20 , and @xmath21 meson masses are @xmath33 , @xmath34 and @xmath35 respectively . from the classical euler - lagrangian equation
, we get the dirac - equation and klein- gordan equation for the nucleon and meson field respectively . the dirac - equation for the nucleon
is solved by expanding the dirac spinor into lower and upper component , while the mean field equation for the bosons are solved in deformed harmonic oscillator basis with @xmath36 as the deformation parameter
. the nucleon equation along with different meson equation form a coupled set of equation , which can be solved by iterative method .
various types of densities such as baryon ( vector ) , scalar , isovector and proton ( charge ) densities are given as @xmath37 \rho_s(r ) & = & \sum_i \psi_i^\dagger(r ) \gamma_0 \psi_i(r ) \ , , \label{eqfn7 } \\[3 mm ] \rho_3 ( r ) & = & \sum_i \psi_i^\dagger(r ) \tau_3 \psi_i(r ) \ , , \label{eqfn8 } \\[3 mm ] \rho_{\rm p}(r ) & = & \sum_i \psi_i^\dagger(r ) \left ( \frac{1 -\tau_3}{2 } \right ) \psi_i(r ) \ , .
\label{eqfn9 } % \\[3mm]\end{aligned}\ ] ]
the calculations are simplified under the shadow of various symmetries like conservation of parity , no - sea approximation and time reversal symmetry , which kills all spatial components of the meson fields and the anti - particle states contribution to nuclear observable .
the center of mass correction is calculated with the non - relativistic approximation , which gives @xmath38 ( in mev ) .
the quadrupole deformation parameter @xmath39 is calculated from the resulting quadropole moments of the proton and neutron . the binding energy and charge radius
are given by well known relation @xcite .
in nuclear structure physics , the pairing correlation has an indispensable role in open shell nuclei .
the priority of the pairing correlation escalates with mass number a. it also plays a crucial role for the understanding of deformation of heavy nuclei . because of the limited pair near the fermi surface , it has a nominal effect for light mass nuclei on both bulk and single - particle properties . in the present case
, we consider only t=1 channel of pairing correlation , i.e. , pairing between proton - proton and neutron - neutron . in such case , a nucleon of quantum state @xmath40 pairs with another nucleon having same @xmath41 value with quantum state @xmath42 , which is the time reversal partner of other .
the philosophy of bcs pairing is same both in nuclear and atomic domain .
the first evidence of the pairing energy came from the even - odd mass staggering of isotopes . in mean field formalism the violation of particle number is account of pairing correlation .
the rmf lagrangian density only accommodates term like @xmath43 ( density ) and no term of the form @xmath44 or @xmath45 .
the inclusion of pairing correlation of the form @xmath46 or @xmath44 violates the particle number conservation @xcite .
thus , a constant gap bcs - type simple prescription is adopted in our calculations to take care of the pairing correlation for open shell nuclei . the general expression for pairing interaction to the total energy in terms of occupation probabilities @xmath47 and @xmath48 is written as @xcite : @xmath49 ^ 2,\ ] ] with @xmath50 pairing force constant .
the variational approach with respect to the occupation number @xmath47 gives the bcs equation @xcite : @xmath51 with @xmath52 .
the densities with occupation number is defined as : @xmath53.\ ] ] for the pairing gap ( @xmath54 ) of proton and neutron is taken from the phenomenological formula of madland and nix @xcite : @xmath55 where , @xmath56 , @xmath57 mev , @xmath58 , and @xmath59 .
the chemical potentials @xmath60 and @xmath61 are determined by the particle numbers for neutrons and protons .
the pairing energy of the nucleons using equation ( 7 ) and ( 8) can be written as : @xmath62 in constant pairing gap calculation , for a particular value of pairing gap @xmath54 and force constant @xmath63 , the pairing energy @xmath64 diverges , if it is extended to an infinite configuration space .
in fact , in all realistic calculations with finite range forces , the contribution of states of large momenta above the fermi surface ( for a particular nucleus ) to @xmath54 decreases with energy .
therefore , the pairing window in all the equations are extended upto the level @xmath65 as a function of the single particle energy .
the factor 2 has been determined so as to reproduce the pairing correlation energy for neutrons in @xmath66sn using gogny force @xcite .
we notice that recently karatzikos et al .
@xcite has been shown that if it is adjusted a constant pairing window for a particular deformation then it may leads to errors at different energy solution ( different state solution ) .
however , this kind of approach have not taken into account in our calculations , as we have adjusted to reproduce the pairing as a whole for @xmath67 nucleus .
it is a tough task to compute the binding energy and quadrupole moment of odd - n or odd - z or both n and z numbers are odd ( odd - even , even - odd , or odd - odd ) nuclei . to do this
, one needs to include the additional time - odd term , as is done in the shf hamiltonian @xcite , or empirically the pairing force in order to take care the effect of odd - neutron or odd - proton @xcite . in an odd - even or odd - odd nucleus
, the time reversal symmetry gets violated in the mean field models . in our rmf calculations , we neglect the space components of the vector fields , which are odd under time reversal and parity .
these are important in the determination of magnetic moments @xcite but have a very small effects on bulk properties such as binding energies or quadrupole deformations , and they can be neglected@xcite in the present context . here , for the odd - z or odd - n calculations , we employ the pauli blocking approximation , which restores the time - reversal symmetry . in this approach , one pair of conjugate states , @xmath68 , is taken out of the pairing scheme .
the odd particle stays in one of these states , and its corresponding conjugate state remains empty . in principle , one has to block in turn different states around the fermi level to find the one that gives the lowest energy configuration of the odd nucleus . for odd - odd nuclei , one needs to block both the odd neutron and odd proton .
in this section , we evaluate our results for binding energy , rms radii , quadrupole deformation parameter for recently predicted thermally fissile isotopes of th and u. these nuclei are quite heavy and needed a large number of oscillator basis , which takes considerable time for computation .
we spent few lines in the first subsection of this section to describe how to select basis space and the results and discussions are followed subsequently . the dirac equation for fermions ( proton and neutron ) and the equation of motion for bosons ( @xmath27 , @xmath69 , @xmath21 and @xmath70 ) obtained from the rmf lagrangian are solved self - consistently using an iterative methods .
these equations are solved in an axially deformed harmonic oscillator expansion basis @xmath71 and @xmath72 for fermionic and bosonic wavefunction , respectively . for heavy nuclei ,
a large number of basis space @xmath71 and @xmath72 are needed to get a converged solution . to reduce the computational time without compromising the convergence of the solution
, we have to choose an optimal number of model space for both fermion and boson fields . to choose optimal values for @xmath71 and @xmath72 , we select @xmath73th as a test case and increase the basis quanta from 8 to 20 step by step
. the obtained results of binding energy , charge radii and quadrupole deformation parameter are shown in fig .
[ bea1 ] . from our calculations
, we notice an increment of 200 mev in binding energy while going from @xmath748 to 10 .
this increment in energy decreases while going to higher oscillator basis .
for example , change in energy is @xmath75 mev with a change of @xmath76 from 14 to 20 and the increment in @xmath77 values are 0.12 fm respectively . keeping in mind the increase in convergence time for larger quanta as well as the size of the nuclei considered , we have finalized to use @xmath78 in our calculations to get a suitable convergent results , which is the current accuracy of the present rmf models . ) and quadrupole deformation parameter ( @xmath79 ) are given with bosonic and fermionic basis . ] to be sure about the predictivity of our model , first of all we calculate the binding energies ( be ) , charge radii @xmath77 and quadrupole deformation parameter @xmath39 for some of the known cases .
we have compared our results with the experimental data wherever available or with the finite range droplet model ( frdm ) of mller et al . @xcite .
the results are displayed in tables [ tab1 ] and [ tab2 ] . from the tables
, it is obvious that the calculated binding energies are comparable with the frdm as well as experimental values .
a further inspection of the tables reveal that the frdm results are more closer to the data .
this may be due to the fitting of the frdm parameters for almost all known data .
however , in case of most rmf parametrizations , the constants are determined by using few spherical nuclei data along with certain nuclear matter properties .
thus the prediction of the rmf results are considered to be reasonable , but not excellent .
ren et al .
@xcite have reported that the ground state of several superheavy nuclei are highly deformed states .
since , these are very heavy isotopes , the general assumption is that the ground state most probably remains in deformed configuration ( liquid drop picture ) .
when these nuclei excited either by a thermal neutron or by any other means , it s intrinsic excited state becomes extra - ordinarily deformed and attains the scission point before it goes to fission .
this can also be easily realized from the potential energy surface ( pes ) curve .
our calculations agree with the prediction of ren et al . for other superheavy region of the mass table .
however , this conclusion is contradicted by @xcite . according to him , the ground state of superheavy nuclei either spherical or normally deformed .
[ cols="^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^ " , ] [ tab3 ] usually , in an axially deformed constraint calculation for a nucleus , we see two maxima in the pes diagram , ( i ) prolate and ( ii ) oblate or spherical . however , in some cases , more than two maxima are also seen .
if the ground state energy is distinctly more than other maxima , then the nucleus has a well defined ground state configuration . on the other hand ,
if the difference in binding energy between two or three maxima is negligible , then the nucleus is in shape co - existence configuration .
in such a case , a configuration mixing calculation is needed to determine the ground state solution of the nucleus , which is beyond the scope of the present calculation .
it is to be noted here that in a constraint calculation , the maximum binding energy ( major peak in the pes diagram ) corresponds to the ground state configuration and all other solutions ( minor peaks in the pes curve ) are the intrinsic excited states .
the fission barrier @xmath80 is an important quantity to study the properties of fission reaction .
we calculate the fission barrier from the pes curve for some selected even - even nuclei , which are displayed in table [ tab3 ] . from the table
, it can be seen that the fission barrier for @xmath81th comes out to be 5.69 mev comparable to the frdm and experimental values of @xmath80 = 7.43 and 6.50 mev , respectively .
similarly , the calculated @xmath80 of @xmath10u is 5.65 mev , which also agree well with the experimental data 5.40 mev . in some cases ,
the fission barrier height is 1@xmath822 mev lower or higher than the experimental data .
the double - humped fission barrier in all these cases are reproduced .
similar type of calculations are also done in refs .
@xcite . in nuclei like @xmath2th and @xmath3u
, we find three maxima . among these maxima , two of them
are found at normal deformation ( spherical and normal prolate ) , but the third one is situated far away , i.e. , at relatively large quadrupole deformation . with a careful inspection , one can also see that one of them ( mostly the peak nearer to the spherical region ) is not strongly pronounced and can be ignored in certain cases . this third maximum separate the second barrier with a depth of 1 - 2 mev , responsible for the formation of resonance state , which are observed experimentally@xcite .
some of the uranium isotopes @xmath83u , the ground states are predicted to be spherical in rmf formalism agreeing with the frdm results .
the other isotopes of the series @xmath84u are found to be prolate ground state matching with the experimental data .
similarly , the thorium nuclei @xmath85th are spherical in shape and @xmath86th are prolate ground configuration .
in addition to these shapes , we also notice sallow regions in the pes curves of both th and u isotopes .
these fluctuation in the pes curves could be due to the limitation of mean field approximation and one needs a theory beyond mean field to over come such fluctuations .
for example , the generator coordinate method or random phase approximation could be some improved formalism to take care of such effects @xcite . beyond the second hump , we find the pes curve goes down and down , which never ups again .
this is the process of the liquid drop gets more and more elongation and reaches to the fission stage .
the pes curve , from which it starts downing is marked the scission points which are shown by the black dot in some of the pes curves of figs .
[ bea2 ] and [ bea3 ] . in this subsection
, we evaluate the neutron and proton single - particle energy levels for some selected nilssion orbits with different values of deformation parameter @xmath79 using the constraint calculations .
the results are given in figs .
[ bea4 ] and [ bea5 ] , explain the origin of the shape change along the @xmath0-decay chains of the thorium and uranium isotopes .
the positive parity orbit is the solid line , negative parity orbit is dash line and the dotted line ( red colour ) indicates the fermi energy for @xmath10th and @xmath87u .
th as a function of quadrupole deformation parameter @xmath79 .
the fermi levels are denoted by thick dotted(red ) curve . ] u nucleus . ] for small z nuclei , the electrostatic repulsion is very weak but at higher value of z ( superheavy nuclei ) , the electrostatic repulsion is much stronger that the nuclear liquid drop becomes unstable to surface distortion @xcite and fission . in such nucleus , the single - particle density is very large and the energy separation is small , which determines the shell stabilizes the unstable coulomb repulsion .
this effect is clear for heavy elements approaching n=126 with the gap between 3p@xmath88 and 1i@xmath89 of about 2 - 3 mev , in the neutron single - particle of @xmath87u and @xmath10th . in both the figures ,
the neutron single particle energy level 1i@xmath90 lies between 2f@xmath91 and 2f@xmath92 creating a distinct shell gap at n=114 . in @xmath10th and @xmath87u , with increasing deformation the opposite parity levels of 2g@xmath93 and 1j@xmath94 come closer to each other , which are far apart in the spherical solution .
this gives rise to the parity doublet phenomena @xcite .
in this section , we will discuss about various mode of decays encounter by superheavy nuclei both in the @xmath1-stability line as well as away from it .
this is important , because the utility of superheavy and mostly the nuclei which are away from stability lines depend very much on their life time .
for example , we do not get @xmath4u and @xmath6pu in nature , because of their short life time , although these two nuclei are extremely useful for energy production .
that is why @xmath5u is the most necessary isotope in the uranium series for its thermally fissile nature in the energy production in fission process both for civilian as well as military use .
the common mode of instability for such heavy nuclei are spontaneous fission , @xmath0- , @xmath1- and @xmath8-decays .
all these decays depend on the neutron to proton ratio as well as the number of nucleons present in the nucleus . in the previous papers@xcite
, we have analyzed the densities of nuclei in a more detailed manner . from this analysis
, we concluded that there is no visible cluster either in the ground or in the excited intrinsic states .
the possible clusterizations are the @xmath0-like matter at the interior and neutron - rich matter at the exterior region of the normal and neutron - rich superheavy nuclei , respectively .
thus , the possible mode of decays are the @xmath0-decay for @xmath1-stable nuclei and @xmath95-decay for neutron - rich isotopes .
to estimate the stability of such nuclei , we have to calculate the @xmath0-decay @xmath96 and the @xmath1-decay @xmath97 half - lives times .
to calculate the @xmath0-decay half - life @xmath100 , one has to know the @xmath13 energies of the nucleus .
this can be estimated by knowing the binding energies ( be ) of the parents , daughter and the binding energy of the @xmath0-particle , i.e. , the be of @xmath101he .
the binding energies are obtained from experimental data wherever available and from other mass formulae as well as relativistic mean field lagrangian as we have discussed earlier in this paper @xcite .
the @xmath13 energy is evaluated by using the relation : @xmath102 here , @xmath103 , @xmath104 and @xmath105 are the binding energies of the parent , daughter and @xmath106he nuclei ( be= 28.296 mev ) with neutron number n and proton number z. and half - life time @xmath107 of the @xmath0-decay chain for th isotopes are calculated using rmf , frmd @xcite , eldm @xcite and compared with the experiment @xcite . ]
knowing the @xmath108values of nuclei , we roughly estimate the @xmath0-decay half - lives @xmath109 of various nuclei using the phenomenological formula of viola and seaborg @xcite : @xmath110 the value of the parameters @xmath111 , @xmath112 , @xmath113 and @xmath114 are taken from the recent modified parametrizations of sobiczewski et al .
@xcite , which are @xmath111 = 1.66175 ; @xmath112 = 8.5166 ; @xmath113 = 0.20228 ; @xmath114 = 33.9069 . the quantity @xmath115 accounts for the hindrances associated with the odd proton and neutron numbers as given by viola and seaborg @xcite , namely @xmath116 the @xmath108values obtained from rmf calculations for th and u isotopes are shown in figs . [ bea11 ] and [ bea12 ] .
our results also compared with other theoretical predictions @xcite and experimental data @xcite .
the agreement of rmf results with others as well as with experiment is pretty well .
although , the agreement in @xmath108value is quite good , one has to note that the @xmath117 values may vary a lot , because of the exponential factor in it .
that is why it is better to compare @xmath118 instead of @xmath117 .
these values are compared in the right panel of figs .
[ bea11 ] and [ bea12 ] .
we notice , our prediction matches well with other calculations as well as experimental data .
further , a careful analysis of @xmath119 ( in seconds ) for even - even thorium , the @xmath108value decreases with increase of mass number a of parent nucleus . the @xmath13 energy of th isotopes given by duarte et al .
@xcite deviates a lot , when mass of the parent nucleus reaches to a=230 .
the corresponding @xmath119 increases almost monotonically linearly with an increase of mass number of the same nucleus .
the experimental values of @xmath119 deviate a lot in the heavy mass region , ( with parent nuclei 234 - 238 ) .
similar situation is found in case of uranium isotopes also which are shown in fig .
[ bea12 ] .
as we have discussed , the prominent mode of instability of neutron - rich th and u nuclei is the @xmath1-decay , and we have given an estimation of such decay in this subsection .
actually , the @xmath1-decay life time should be evaluated in a microscopic level , but in this paper , it is beyond the scope .
here we have used the empirical formula of fiset and nix @xcite , which is defined as : -decay half life for th and u isotopes are calculated using the formula of fiset and nix @xcite [ eq . ( 24 ) ] . the ground state binding energies are taken from frdm @xcite , inm @xcite and rmf models . ]
@xmath120 similar to the @xmath0-decay , we evaluate the @xmath14-value for th and u series using the relation @xmath121 and @xmath122 . here , @xmath123 is the average density of states in the daughter nucleus ( e@xmath124 @xmath125 number of states within 1 mev of ground state ) . to evaluate the bulk properties , such as binding energy of odd - z nuclei
, we used the pauli blocking prescription as discussed in section [ sec2 ] .
the obtained results are displayed in fig .
[ bea15 ] for both th and u isotopes . from the figure ,
it is clear that for neutron - rich th and u nuclei , the prominent mode of decay is @xmath1-decay .
this means , once the neutron - rich thermally fissile isotope is formed by some artificial mean in laboratory or naturally in supernovae explosion , immediately it undergoes @xmath1-decay . in our rough estimation ,
the life time of @xmath126th and @xmath127u , which are the nuclei of interest has tens of seconds .
if this prediction of time period is acceptable , then in nuclear physics scale , is reasonably a good time for further use of the nuclei .
it is worthy to mention here that thermally fissile isotopes of th and u series are with neutron number n=154 - 172 keeping n=164 in the middle of the island .
so , in case of the short life time of @xmath126th and @xmath127u , one can choose a lighter isotope of the series for practical utility . before drawing the concluding remarks , it is important to mention few points about the limitations of our present approach .
when we compare our calculated results with the experimental data , although we get satisfactory results , some time we do not get excellent agreement and the main possible reasons for the discrepancy of rmf with experimental values are given as : + ( 1 ) in rmf formalism we are working in the mean field approximation of the meson field . in this approximation , we are neglecting the vacuum fluctuation , which is an indispensable part of the relativistic formalism . in calculating the nucleonic dynamics
, we are neglecting the negative energy solution that means , we are working in the no sea approximation @xcite .
it is already discussed that the no - sea approximation and quantum fluctuation can improve the results upto a maximum of 20% @xcite for very light - nuclei .
therefore , the mean field is not a good approach for the light region of the periodic table .
however , for the heavy masses , this mean field approach is quite good and can be used for any practical purpose . + ( 2 ) in order to solve the nuclear many body system , here we used the hartee formalism and neglect fock term , which corresponds to the exchange correlation .
+ ( 3 ) to take care of the pairing correlation , we have used bcs type pairing approach .
this gives good results for the nuclei near the @xmath1-stability line , but it fails to incorporate properly the pairing correlation for the nuclei away from the @xmath1-stability line and superheavy nuclei @xcite . thus a better approach like hartree - fock - bogoliubov @xcite type pairing co - relation is more suitable for the present region .
+ ( 4 ) parametrization plays an important role in improvising the results .
the constants in rmf parametrizations , are determined by fixing the experimental data for few spherical nuclei .
we expect that the results may be improved by re - fitting the force parameters for more number of nuclei , including the deformed isotopes .
+ ( 5 ) the basic assumption in the rmf theory is that two nucleons interact with each other through the exchange of various mesons .
there is no direct inclusion of 3-body or higher body effects .
this effect is taken care partially by including the self - coupling of mesons and in recent relativistic approach various cross - couplings are added because of their importance .
+ ( 6 ) although , there are various mesons are observed experimentally , few of them are taken into account in the nucleon - nucleon interaction .
contribution of some of them are prohibited due to symmetry reason and many are neglected due to their negligible contributions , because of heavy mass .
however , some of them has substantial contribution to the properties of nuclei , specially when the neutron - proton asymmetry is more , such as @xmath128-meson @xcite .
+ ( 7 ) it is to be noted that the origin of @xmath0-decay or @xmath8-decay phenomena are purely quantum mechanical process .
thus the quantum tunneling plays an important role in such decay processes .
the deviation of experimental @xmath0-decay life time from the calculated results obtained by the empirical formula may not be suitable for such heavy nuclei , which are away from the stability line and more involved quantum mechanical treatment is needed for such cases .
in summary , we did a thorough structural study of the recently predicted thermally fissile isotopes of th and u series in the framework of relativistic mean field theory .
although there are certain limitations of the present approach , the qualitative results will remain unchanged even if the draw - back of the model taken into account .
the heavier isotopes of these two nuclei bear various shapes including very large prolate deformation at high excited configurations .
the change in single - particle orbits along the line of quadrupole deformation are analyzed and found parity doublet states in some cases . using an empirical estimation , we find that the neutron - rich isotopes of these thermally fissile nuclei are predicted to be stable against @xmath0- and @xmath8-decays .
the spontaneous fission also does not occur , because the presence of large number of neutrons makes the fission barrier broader . however , these nuclei are highly @xmath1-unstable .
our calculation predicts that the @xmath1-life time is about tens of seconds for @xmath126th and @xmath127u and this time increases for nuclei with less neutron number , but thermally fissile .
this finite life time of these thermally fissile isotopes could be very useful for energy production in nuclear reactor technology .
if these neutron - rich nuclei use as nuclear fuel , the reactor will achieve critical condition much faster than the normal nuclear fuel , because of the release of large number of neutrons during the fission process .
99 l. satpathy , s. k. patra and r. k. choudhury , pramana j. phys . * 70 * , 87 ( 2008 ) .
p. arumugam , b. k. sharma , s. k. patra and r. k. gupta , phys .
c * 71 * , 064308 ( 2005 ) .
birbikram singh , b. b. sahu and s. k. patra , phys .
c * 83 * , 064601 ( 2011 ) .
s. k. patra , raj .
k. gupta , b. k. sharma , p. d. stevenson and w. greiner , j. phys .
g. * 34 * , 2073 ( 2007 ) .
b. k. sharma , p. arumugam , s. k. patra , p. d. stevenson , r. k. gupta and w. greiner , j. phys .
g. * 32 * , l1 ( 2006 ) .
g. r. satchler and w. g. love , phys . rep . *
55 * , 183 ( 1979 ) .
birbikram singh , m. bhuyan , s. k. patra and raj . k. gupta , j. phys .
g. * 39 * , 025101 ( 2012 ) .
b. b. sahu , s. k. singh , m. bhuyan , s. k. biswal and s. k. patra , phys .
c * 89 * , 034614 ( 2014 ) .
g. a. lalazissis , j. knig and p. ring , phys .
c * 55 * , 540 ( 1997 ) .
s. k. patra and c. r. praharaj , phys .
c * 44 * , 2552 ( 1991 ) .
j. d. walecka , ann . phys .
* 83 * , 491 ( 1974 ) .
b. d. serot and j. d. walecka , adv .
* 16 * , 1 ( 1986 ) . c. j. horowitz and b. d. serot , nucl .
a * 368 * , 503 ( 1981 ) .
j. boguta and a. r. bodmer , nucl .
a * 292 * , 413 ( 1977 ) . c. e. price and g. e. walker , phys .
c * 36 * , 354 ( 1987 ) .
p. g. blunden and m. j. iqbal , phys .
b * 196 * , 295 ( 1987 ) .
p. g. reinhard , rep .
. phys . * 52 * , 439 ( 1989 ) .
y. k. gambhir , p. ring and a. thimet , ann . of phys .
* 198 * , 132 ( 1990 ) .
s. k. patra , phys .
c * 48 * , 1449 ( 1993 ) .
m. a. preston and r. k. bhaduri , _ structure of nucleus , addison - wesley publishing company _ , ch .
8 , page 309 ( 1982 ) .
d. g. madland and j. r. nix , nucl .
phys . a * 476 * , 1 ( 1981 ) .
j. decharg and d. gogny , phys . rev .
c * 21 * , 1568 ( 1980 ) .
s. karatzikos , a. v. afanasjev , g. a. lalazissis and p. ring , phys .
b * 689 * , 72 ( 2010 ) .
j. r. stone and p. g. reinhard , prog .
58 * , 587 ( 2007 ) .
f. tondeur , s. goriely , j. m. pearson , and m. onsi , phys .
c * 62 * , 024308 ( 2000 ) .
u. hofmann and p. ring , phys .
b * 214 * , 307 ( 1988 ) .
g. a. lalzissis , d. vretenar , and p. ring , nucl .
phys . a * 650 * , 133 ( 1999 ) .
p. mller , j. r. nix and k. l. kratz , at .
data and nucl .
data tables * 59 * , 185 ( 1995 ) .
p. mller , j. r. nix , w. d. myers and w. j. swiatecki , at .
data and nucl .
data tables * 66 * , 131 ( 1997 ) . http://www.nndc.bnl.gov/nudat2/ i. angeli and k. p. marinova , at .
data and nucl .
data tables * 99 * , 69 ( 2013 ) .
zhongzhou ren and hiroshi toki , nucl .
a * 689 * , 691 ( 2001 ) .
zhongzhou ren , ding - han chen , fei tai , h. y. zhang , and w. q. shen , phys .
c * 67 * , 064302 ( 2003 ) . z. patyk and a. sobiczewski , nucl .
phys . a * 533 * , 132 ( 1991 ) .
s. bj@xmath129rnholm and j. e. lynn , rev .
mod . phys . * 52 * , 715 ( 1980 ) .
s. k. patra , f. h. bhat , r. n. panda , p. arumugam and r. k. gupta , phys .
c * 79 * , 044303 ( 2009 ) .
h. flocard , p. quentin and d. vautherin , phys .
b * 46 * , 304 ( 1973 ) .
w. koepf and p. ring , phys .
b * 212 * , 397 ( 1988 ) .
j. fink , v. blum , p. g. reinhard , j. a. maruhn and w. greiner , phys .
b * 218 * , 277 ( 1989 ) .
d. hirata , h. toki , i. tanihata and p. ring , phys .
b * 314 * , 168 ( 1993 ) .
j. meng , h. toki , s. g. zhou , s. q. zhang , w. h. long and l. s. geng , prog .
. phys . * 57 * , 470 ( 2006 ) .
bing - nan lu , en - guang zhao and shan - gui zhou , phys . rev .
c * 85 * , 011301(r ) ( 2012 ) .
bing - nan lu , jie zhao , en - guang zhao and shan - gui zhou , phys . rev .
c * 89 * , 014323 ( 2014 ) .
jie zhao , bing - nan lu , dario vretenar , en - guang zhao and shan - gui zhou , phys . rev .
c * 91 * , 014321 ( 2015 ) .
b. b. back , h. c. britt , j. d. garrett , and o. hansen , phys .
* 28 * , 1707 ( 1972 ) ; j. blons , c. mazur , d. paya , m. ribrag , and h. weigmann , ibid . *
41 * , 1282 ( 1978 ) .
d. m. brink and a. weiguny , phys .
b * 26 * , 497 ( 1968 ) . myers and w. d. swiatecki , nucl .
* 81 * , 1 ( 1966 ) .
s. k. singh , c. r. praharaj and s. k. patra , cen .
* 12 * , 42 ( 2014 ) .
bharat kumar , s. k. singh and s. k. patra , int .
* 24 * , 1550017 ( 2015 ) .
p. g. thirolf and d. habs , prog .
. phys . * 49 * , 325 ( 2002 ) .
s. k. patra and c. r. praharaj , j. phys .
g * 23 * , 939 ( 1997 ) .
s. b. duarte , o. a. p. tavares , f. guzman and a. dimarco , at . data and and nucl .
data table * 80 * , 235 ( 2009 ) . v. e. viola jr . and g. t. seaborg , j. inorg .
* 28 * , 741 ( 1966 ) .
a. sobiczewski , z. patyk and s. c. cwiok , phys . lett . *
224 * , 1 ( 1989 ) .
e. o. fiset and j. r. nix , nucl .
a * 193 * , 647 ( 1972 ) .
r. c. nayak and l. satpathy , at .
data and nucl .
data tables * 73 * , 213 ( 1999 ) .
p. g. reinhard , m. rufa , j. maruhn , w. greiner and j. friedrich , z. phys .
a * 323 * , 13 ( 1986 ) .
z. y. zhu , h. j. mang and p. ring , phys .
b * 254 * , 325 ( 1991 ) . p. ring and p. schuck , the nuclear many - body problem , springer - verlag , berlin , 1980 .
p. ring , prog .
* 37 * , 193 ( 1996 ) .
s. kubis and m. kutschera , phys .
b * 399 * , 191 ( 1997 ) .
s. k. singh , s. k. biswal , m. bhuyan , and s. k. patra , phys .
c * 89 * , 044001 ( 2014 ) . | the properties of recently predicted thermally fissile th and u isotopes are studied within the framework of relativistic mean field ( rmf ) approach using axially deformed basis .
we calculated the ground , first intrinsic excited state for highly neutron - rich thorium and uranium isotopes .
the possible modes of decay like @xmath0-decay and @xmath1-decay are analyzed .
we found that the neutron - rich isotopes are stable against @xmath0-decay , however they are very much unstable against @xmath1-decay .
the life time of these nuclei predicted to be tens of second against @xmath1-decay .
if these nuclei utilize before their decay time , a lots of energy can be produced with the help of multi - fragmentation fission .
also , these nuclei have a great implication in astrophysical point of view . in some cases
, we found the isomeric states with energy range from 2 to 3 mev and three maxima in the potential energy surface of @xmath2th and @xmath3u isotopes . |
the peierls barrier @xmath1 for the monotone twist diffeomorphism @xmath0 is a function which can be thought of as a dislocation energy .
it measures to which extent the stationary configuration @xmath6 of rotation symbol @xmath7 , subject to the constraint @xmath8 , is not minimal . in @xcite , mather established the modulus of continuity for @xmath9 with respect to the parameter @xmath7 , and by applying this property , he gave destruction results for invariant circles under arbitrary small perturbations ( @xcite ) .
for further research , we need more information and properties about the peierls barriers . in this paper , we prove the hlder continuity of peierls barriers with respect to the parameter @xmath0 , which generalize j. mather s results in @xcite@xcite . this paper is organized as follows : in section 2 and section 3 , we introduce the definition of peierls barrier and some basic properties in aubry - mather theory .
our main results are theorem [ main theorem 1 ] and theorem [ main theorem 2 ] in section 4 . in section 5
, we give an open and dense result as an application example . for convenience ,
we denote by @xmath10 ( mod 1 ) the standard coordinate of @xmath11 and @xmath12 the corresponding coordinate of its universal cover @xmath13 .
we will let @xmath14 denote the standard coordinates of @xmath15 and @xmath16 the corresponding coordinates of the universal cover @xmath17 .
the dynamical properties of exact area - preserving monotone twist diffeomorphisms of an infinite cylinder @xmath15 have been studied by mather ( @xcite@xcite ) and by bangert ( @xcite ) . in the following ,
we refer to these papers for the definitions and results that we ll need .
we call @xmath0 an exact area - preserving monotone twist diffeomorphism if @xmath18 , @xmath19 is a diffeomorphism satisfying the following conditions : 1
. @xmath20 .
the 1-form @xmath21 on @xmath15 is exact .
3 . ( positive monotone twist )
@xmath22 , for all @xmath14 .
@xmath0 twists the cylinder infinitely at either hand . to express this condition
, we consider a lift @xmath23 of @xmath0 to the universal cover @xmath17 , @xmath24 the condition means that for fixed @xmath12 , @xmath25 the positive monotone twist condition has its geometrical meaning .
consider a point @xmath26 and denote by @xmath27 the vertical vector at @xmath28 .
let @xmath29 denote the angle between @xmath30 and @xmath31 ( count in the clockwise direction ) .
so the positive monotone twist condition means that @xmath32 everywhere .
+ we denote by @xmath33 the class of exact area - preserving monotone twist diffeomorphisms .
let @xmath34 .
although @xmath35 , most of our results can be generalized to @xmath33 without any difficulty .
this is because our main results concern what happens in a compact region @xmath36 of @xmath15 .
thus , for all @xmath37 , there exists @xmath38 and @xmath39 such that @xmath40 . if @xmath37 and @xmath23 is a lift of @xmath0 to @xmath17 such that @xmath41 , then there exists a @xmath42 generating function @xmath43 such that @xmath44 and @xmath45 where @xmath46 and @xmath47 denote the partial differential derivatives of @xmath48 with respect to @xmath12 and @xmath49 . for @xmath37
, the generating function satisfies the following conditions @xmath50 , + ( @xmath51 ) @xmath52 , for all @xmath53 + ( @xmath54 ) @xmath55 uniformly in @xmath12 .
+ ( @xmath56 ) if @xmath57 , then @xmath58 + ( @xmath59 ) if @xmath60 and @xmath61 are both minimal segments ( see 2 ) and are distinct , then @xmath62 moreover , we add two further conditions to @xmath48 which was firstly introduced by mather .
+ ( @xmath63 ) there exists a positive continuous function @xmath64 such that for @xmath65 and @xmath66 , @xmath67 + ( @xmath68 ) there exists a positive number @xmath69 such that @xmath70 @xmath71 conditions @xmath50 were firstly introduced by bangert in @xcite , and conditions @xmath72 can be implied by conditions @xmath73 ( see @xcite ) . for @xmath74
, it s not hard to observe that we can take @xmath75 in @xmath76 , the inequality `` @xmath77 '' can be replaced by `` = '' , and @xmath78 is satisfied with @xmath79 notice that if @xmath48 is a generating function , @xmath80 is still a generating function . in next section , we ll introduce some basic results on the theory of minimal configurations , which was developed by aubry and le daeron @xcite and mather . however , bangert generalized this theory where @xmath48 is not necessary differentiable and only satisfies @xmath50 ( see @xcite ) .
in bangert s set up , the variational principle @xmath48 need not be differentiable , which is very useful for us in our proof of theorem [ main theorem 1 ] and theorem [ main theorem 2 ] .
therefore , in this section , unless otherwise specified , we assume that @xmath48 is only continuous and satisfies the conditions @xmath50 .
let @xmath81 be bi - infinite sequences of real numbers with the product topology and let @xmath82 be an element in @xmath83 .
we extend @xmath48 to a finite segment @xmath84 , @xmath85 . the segment @xmath86 is called _ minimal segment _ with respect to @xmath48 if @xmath87 for all @xmath88 with @xmath89 and @xmath90 .
@xmath82 is called _ minimal configuration _ if every finite segment of @xmath12 is minimal .
we denote by @xmath91 the set of all minimal configurations of @xmath48 .
if @xmath92 , a segment @xmath86 is called _ stationary _ if @xmath93 the stationary configurations @xmath94 of @xmath48 correspond to the orbits @xmath95 of @xmath23 , where @xmath96 .
thus , the minimal configurations @xmath97 of @xmath48 correspond to a class of minimal orbits of the lift @xmath23 of @xmath0 , we denote by @xmath98 the set of all such minimal orbits . in the following
, we will give some dynamical properties for the minimal configurations for @xmath48 , or equivalently , the minimal orbits of @xmath23 .
for proofs , we refer to @xcite . given a configuration @xmath82 ,
we join @xmath99 and @xmath100 by a line segment in @xmath101 , the union of all such line segments is a piecewise linear curve in @xmath101 , which we call _ the aubry graph _ of @xmath12 .
we say that configurations @xmath12 and @xmath102 _ cross _ if their aubry graph cross .
we say @xmath103 if @xmath104 for all @xmath105 .
similarly , we can define @xmath106 and @xmath107 .
we say that @xmath12 and @xmath102 are _ comparable _ if @xmath103 or @xmath107 or @xmath106 . by condition @xmath108 , we know that any two minimal configurations either cross or are comparable . moreover , we say @xmath12 and @xmath102 are _ @xmath109 _ ( resp .
_ @xmath110 _ ) if @xmath111 ( resp .
@xmath112 ) .
[ aubry crossing lemma ] let @xmath12 and @xmath102 be @xmath48-minimal configurations , then they cross at most once .
if @xmath12 and @xmath102 coincide at some @xmath113 , i.e. @xmath114 , then they cross at @xmath105 .
notice that aubry s crossing lemma can be proved by condition @xmath115 .
if @xmath12 is a minimal configuration , then @xmath116 exists . the number @xmath117 is called _ the rotation number _ of @xmath12 .
in addition , the rotation function @xmath118 is continuous and surjective , and @xmath119 is proper , where @xmath120 .
if @xmath121 and @xmath122 , where @xmath123 and @xmath124 are relatively prime integers , then @xmath12 must satisfy one of the three relations ( see @xcite ) : a. @xmath125 for all @xmath105 .
b. @xmath126 for all @xmath105 . c. @xmath127 for all @xmath105 .
this leads us to introduce the _ symbol space _ @xmath128 , where @xmath129 denotes the set of all symbols @xmath130 and @xmath131 is defined similarly .
the symbol space has an obvious order so that @xmath132 .
we provide @xmath133 with the order topology , i.e. the set of intervals ( @xmath134)=@xmath135 is a basis for this topology .
we also define the projection map @xmath136 , @xmath137 obviously , the map @xmath138 is weakly order preserving . for more details ,
see ( @xcite ,
3 ) . from now on ,
if @xmath139 , we denote by @xmath140 the set of @xmath121 with @xmath141 .
let @xmath142 denotes the set of @xmath121 with @xmath122 and @xmath143 holds .
let @xmath144 denotes the set of @xmath121 with @xmath122 and @xmath143 or @xmath145 hold .
similarly , we also denote by @xmath146 the set of @xmath121 with @xmath122 and @xmath147 or @xmath143 hold .
notice that for each @xmath148 , @xmath149 is a non - empty closed set and totally ordered .
equivalently , for the rotation symbol @xmath148 , we can denote by @xmath150 the set of all @xmath7-minimal orbits of @xmath37 .
[ rotation estimates ] let @xmath151 be a minimal configuration , we have that @xmath152 for all @xmath153 . [ bangert ] let @xmath154 be a minimal segment , there exists @xmath155 so that @xmath156 for all @xmath157 by the totally ordered property , the projection @xmath158 , @xmath120 is a homeomorphism of @xmath149 onto its image .
we denote by @xmath159 , then it s a closed subset of @xmath13 and invariant under the translation @xmath160 ( see @xcite ) .
now , we begin to introduce the definition of the peierls barrier function @xmath161 , it measures to which extent the stationary configuration @xmath162 , subject to the condition @xmath163 , is not minimal .
besides bangert s conditions @xmath50 , we will assume @xmath48 satisfies @xmath76 and @xmath78 .
since @xmath164 is an open set , it is a union of open intervals @xmath165 , where @xmath166 . _
the peierls barrier _ is defined as follows .
let @xmath148 and @xmath167 , in the case @xmath168 , we define @xmath169 .
in the case @xmath170 , @xmath171 belongs to a complementary interval @xmath165 to @xmath172 .
denote by @xmath173 the minimal configurations of rotation symbol @xmath7 satisfying @xmath174 and let @xmath175 we define @xmath176 where the minimum is taken over the set of all configurations satisfying @xmath177 and in the case @xmath178 , the minimum is taken under the additional periodicity constraint @xmath179 .
notice that @xmath180 is well defined and finite , since @xmath48 is lipschitz and @xmath181(see @xcite ) .
it is worth pointing out that @xmath180 is non - negative and 1-periodic , i.e. @xmath182 now , let s recall some important properties of the peierls barrier which are important for our main results in section 4 . [
continuous of modulus ] let @xmath48 be a continuous real valued function satisfying @xmath183 and @xmath78 , then 1 .
@xmath184 2 .
@xmath185 in the case @xmath186 and + @xmath187 in the case @xmath188 , where @xmath138 is ( [ underlying number ] ) .
the proof strongly relies on aubry s crossing lemma .
proposition [ continuous of modulus ] ( 1 ) was firstly proved by mather in ( @xcite , theorem 7.1 ) .
the proof of ( 2 ) could be found in @xcite , theorem 2.2 , where the author proved that if @xmath186 , then @xmath189 where the constant @xmath190 .
similarly , if @xmath188 , then @xmath191 where @xmath190 .
[ continuous at irrational number ] the map @xmath192 is continuous at any irrational number @xmath7 , uniformly in @xmath171 . in general , @xmath193 is not continuous at rational symbol @xmath178 .
the reason for not restricting our attention to @xmath42 function @xmath48 is that we consider not only the generating functions with respect to @xmath37 , but also the class of functions generated by the conjunction operation .
let @xmath194 and @xmath195 be two continuous functions satisfying the conditions @xmath183 and @xmath78 , _ the conjunction _ is defined in the following way : @xmath196 notice that even when both @xmath194 and @xmath197 are smooth , the conjunction @xmath198 need not be smooth .
however , we still have the following useful property . [ conjunction ]
if @xmath194 and @xmath195 are two continuous functions satisfying the conditions @xmath183 and @xmath78 , then @xmath198 satisfies @xmath183 and @xmath78 with the same @xmath69 .
therefore , one can still define the minimal configurations and peierls barriers with respect to @xmath198 . given a function @xmath48
as descried above and a rational number @xmath199 ( in lowest terms ) , we define the following conjunction @xmath200 where @xmath201 denotes the @xmath202-fold conjunction of @xmath48 with itself . it is easy to verify the following equivalent relations about the minimal configurations and the peierls barrier functions .
[ equivalence of conjunction ] if @xmath48 satisfies @xmath183 and @xmath78 , and @xmath203 is the conjunction ( [ conjunct itself ] ) , then @xmath204 where the rotation symbol @xmath205 or @xmath206 whose denominator @xmath207 is divisible by @xmath202 .
let @xmath0 be an exact area - preserving monotone twist diffeomorphism and let @xmath23 be a lift of @xmath0 which satisfies @xmath208 , and we set @xmath209 . for the rotation interval
@xmath210 $ ] , there is a compact annulus @xmath211 $ ] with sufficiently large @xmath212 so that the minimal orbits of @xmath0 satisfies @xmath213.\ ] ] the space @xmath214 is provided with the norm : @xmath215 [ main theorem 1 ] let @xmath37 , @xmath210 $ ] and @xmath216 be as shown above .
there exist positive numbers @xmath217 and @xmath218 such that for any @xmath219 , if @xmath220 , then the corresponding peierls barrier functions satisfy @xmath221 where @xmath222 , with @xmath223 $ ] . 1 .
the projection @xmath138 is defined in ( [ underlying number ] ) .
the conclusion of theorem [ main theorem 1 ] also holds for @xmath224 perturbations .
the novelty here is that the constant @xmath225 does not depend on @xmath171 and the @xmath3-hlder regularity is uniform in @xmath171 , which could be derived from our proof below , see ( [ final 3 ] ) .
3 . for rational symbol @xmath178 , the conclusion of theorem [ main theorem 1 ] does nt hold in general since the map @xmath226 is not continuous at rational symbol . in order to prove theorem [ main theorem 1 ] , we need the following lemmas .
[ lemma 1 ] under the same assumptions of theorem [ main theorem 1 ] , let @xmath48 , @xmath227 be generating functions of @xmath0 and @xmath228 respectively , satisfying the condition @xmath229 .
there exist positive numbers @xmath230 and @xmath231 , such that if @xmath232 , then @xmath233 where @xmath234 .
take @xmath235 small enough such that @xmath236 , then the lifts @xmath237\times[-k , k])}=\|f'-f\|_{c^1({\mathbf{a}}_k)}.\ ] ] since for all @xmath238 , @xmath239 and @xmath240 , one can deduce that @xmath241)}=\|\bar{f'}-\bar{f}\|_{c^1([0,1]\times[-k , k])}=\|f'-f\|_{c^1({\mathbf{a}}_k)}.\ ] ] by adding a constant we can assume @xmath242 . by choosing suitably large @xmath36 , we can assume @xmath243 . from the assumption ( [ basic setting ] ) , for all @xmath7 with @xmath244 $ ] and @xmath245
, we have @xmath246 , i.e. @xmath247 since @xmath248 and @xmath23 are sufficiently close , then for all @xmath249 , we have @xmath250 , i.e. @xmath251 from now on , we only consider @xmath252 restricted on the set @xmath253 ( see figure 2 ) . in the following
, we set the bounded region @xmath254\times{\mathbb{r}})$ ] .
+ by the condition @xmath255 , for all @xmath256 , @xmath257 because @xmath258 are @xmath42 , so for @xmath259 , @xmath260 and @xmath261 [ [ step-1 ] ] * step 1 * + + + + + + + + firstly , let s estimate @xmath262 . by ( [ generating ] ) , we have @xmath263 and by ( [ generating of i_1 ] ) , @xmath264 if we set @xmath265}\frac{\partial \bar{f_1}}{\partial y}(x , y)$ ] , then @xmath266 since @xmath267 and @xmath268 because @xmath269 are sufficiently close , so it s easy to compute that @xmath270 ) and ( [ boundness of i_1 ] ) , @xmath271)}\geq a|\partial_1h'(t,0)-\partial_1h(t,0)|\end{aligned}\ ] ] and by ( [ equivalence ] ) , @xmath272 ) } dt\leq\int_0^x \frac{1}{a}\|f'-f\|_{c^1({\mathbf{a}}_k)}dt\leq\frac{1}{a}\|f'-f\|_{c^1({\mathbf{a}}_k)},\end{aligned}\ ] ] since @xmath273 .
[ [ step-2 ] ] * step 2 * + + + + + + + + secondly , let s estimate @xmath274 . for each fixed @xmath12 , let @xmath275 since @xmath276 , we could choose a positive number @xmath277 such that @xmath278\times[-k , k].\ ] ] because @xmath269 are sufficiently close , so it s easy to compute that for @xmath279 , @xmath280 where @xmath281 are the inverse function of @xmath282 respectively .
then , by ( [ equivalence ] ) , ( [ phi ] ) , ( [ boundness of i_2 ] ) , we can conclude that for @xmath279 , @xmath283\times[-k , k])}=\frac{1}{b}\|f'-f\|_{c^1({\mathbf{a}}_k)}. \end{split}\end{aligned}\ ] ] by ( [ generating ] ) , we get @xmath284 and by ( [ i_2 ] ) and ( [ generating of i_2 ] ) , one can deduce that @xmath285\times[-k , k])}~|(\phi'_x)^{-1}(t)-(\phi_x)^{-1}(t)|\\ & \leq ( 1+\frac{l}{b})\|f'-f\|_{c^1({\mathbf{a}}_k ) } \end{split}\end{aligned}\ ] ] where @xmath286\times[-k , k])}$ ] .
+ furthermore , @xmath287}(~|\bar{f_1}(x , k)|+|\bar{f_1}(x ,-
k)|~ ) ( 1+\frac{l}{b})\|f'-f\|_{c^1({\mathbf{a}}_k)}\\ & \leq 2\|\bar{f}\|_{c^1([0,1]\times[-k , k])}(1+\frac{l}{b})\|f'-f\|_{c^1({\mathbf{a}}_k ) } \end{split}\end{aligned}\ ] ] finally , by ( [ results of lemma 1 ] ) , ( [ perturbation of h ] ) , ( [ estimate of i_1 ] ) and ( [ estimates of i_2 ] ) , we get that @xmath233 with the constant @xmath288 only depending on @xmath36 and @xmath0
. the following lemma give an equivalent definition of the peierls barrier for the rational rotation symbol .
[ lemma 3 ] let @xmath48 be a continuous real valued function satisfying @xmath183 and @xmath78 , then for rational rotation symbol @xmath289(in lowest terms ) , the peierls barrier has an equivalent defintion : @xmath290 take @xmath291 , where @xmath165 is a complementary interval to @xmath292 .
let @xmath293 satisfying @xmath294 be the periodic minimal configurations in @xmath295 . comparing it with the definition ( [ the definiton of barrier ] )
, we only need to prove that the minimal segment @xmath296 @xmath297 which achieves the minimum in the definition ( [ the definiton of barrier ] ) satisfies the constraint @xmath298 in fact , we claim that the aubry graphs of @xmath299 do not cross .
since @xmath300 are minimal configurations , @xmath301 is a minimal segment , the claim is an easy consequence according to aubry s crossing lemma ( lemma [ aubry crossing lemma ] ) .
[ lemma 4 ] let @xmath48 be a continuous real valued function satisfying @xmath183 and @xmath78 , then the peierls barriers have the following properties : 1 . if @xmath302 , then @xmath303 2 . if @xmath302 , then @xmath304 notice that this lemma is not a direct conclusion of proposition [ continuous of modulus ] .
lemma [ lemma 4 ] ( 1 ) was firstly claimed and proved by mather in ( @xcite , ( 4.1 ) , ( 4.4a ) , ( 4.4b ) ) , lemma [ lemma 4 ] ( 2 ) can be proved by the same approach . for symbol simplicity ,
we denote by @xmath305 [ lemma 5 ] assume that @xmath252 are the generating functions described in lemma [ lemma 1 ] .
given @xmath306 $ ] , @xmath123 and @xmath124 are relatively prime , and let @xmath307 ( resp .
@xmath308 ) be the conjunction ( [ conjunct itself ] ) of @xmath48 ( resp .
@xmath227 ) , we have the following estimates : 1 .
@xmath309 2 . for @xmath310 , @xmath311 ( 1 ) .
in fact , take @xmath312 and let @xmath313 be the minimal segment which achieves the minimum of @xmath203 , i.e. @xmath314 then , @xmath315 next , we need to show that @xmath316 recall that , by proposition [ bangert ] , there exists @xmath155 such that @xmath317 then , we obtain @xmath318 and by proposition [ bangert ] again , @xmath319 notice that we have assumed that the constant @xmath36 is large enough , thus ( [ hh ] ) holds . by ( [ h ] ) and ( [ hh ] )
, we deduce that @xmath320 similarly , we can prove that @xmath321 this completes the proof of ( 1 ) .
firstly , we claim that the definition of @xmath322 only depends on @xmath203 restricted on @xmath323 .
take @xmath291 , where @xmath165 is a complementary interval to @xmath324 .
let @xmath293 satisfying @xmath325 be the @xmath326-minimal configurations of @xmath327 , @xmath328 where the minimum is taken over the set of all configurations satisfying @xmath329 and @xmath330 .
since @xmath331 , by the totally ordered property of @xmath300 , we get @xmath332 thus , @xmath333 which proves our claim .
similarly , we can prove this for @xmath334 .
on the other hand , by lemma [ lemma 3 ] , we have @xmath335 it follows that @xmath336 and @xmath337 firstly , let s estimate @xmath338 .
assume that @xmath339 is the minimal segment satisfying @xmath340 we derive from proposition [ bangert ] that there exists @xmath155 such that @xmath341 and since @xmath342 , we obtain @xmath343 consequently , @xmath344 then , @xmath345 where @xmath346 .
similarly , we can prove @xmath347 we conclude that , by ( [ part 1 ] ) and ( [ part 2 ] ) , @xmath348 secondly , let s estimate @xmath349 .
in fact , it can be similarly estimated as @xmath338 , @xmath350 ) , ( [ i2222 ] ) and lemma [ lemma 5 ] ( 1 ) that @xmath351 which completes the proof of ( 2 ) . because our proof of theorem [ main theorem 1 ] needs the technique of rational approximation , so we introduce the following lemma . [ lemma 2 ] given @xmath352 and @xmath353 , then there exists a rational number @xmath354 , @xmath123 and @xmath124 are relatively prime , such that @xmath355 it can be easily proved by pigeon hole principle .
indeed , we firstly assume that @xmath139 . for every @xmath356 , we can find a integer @xmath357 such that @xmath358 then @xmath359 are @xmath360 distinct points in the interval @xmath361 since @xmath139 . by pigeon hole principle ,
there exist two points @xmath362 such that @xmath363 , i.e. @xmath364 thus , @xmath365 which completes the proof for all @xmath139 . for @xmath366
, it can be proved similarly . to simplify notations ,
let s set @xmath367 because we only concern what happens in the compact region @xmath216 , so we can assume that the generating function @xmath48 of @xmath0 satisfies @xmath183 and @xmath78 , with some @xmath368 depending only on @xmath36 and @xmath0 . in view of the geometrical meaning of @xmath69 , we know that , restricted on the compact region @xmath216 , @xmath369 turns every vertical vector to the right by an angle by at least @xmath370 ( @xmath371 ) .
thus there exists @xmath372 such that for the perturbation @xmath228 of @xmath0 , if @xmath373 , the diffeomorphism @xmath374 turns every vertical vector to the right by an angle by at least @xmath375 where @xmath376 and @xmath377 . from now on ,
we set @xmath378 .
so if @xmath379 , the generating function @xmath227 of @xmath228 satisfying @xmath183 and @xmath380 with @xmath381 meanwhile , by lemma [ lemma 1 ] , we choose the generating functions such that @xmath229 and @xmath382 according to the proof of lemma [ lemma 1 ] , for all rotation symbol @xmath383 with @xmath244 $ ] , the minimal configurations @xmath384 thus the definition of peierls barrier functions @xmath385 only depend on @xmath386 let s begin our proof .
firstly , we approximate @xmath7 by rational number .
in fact , by lemma [ lemma 2 ] , for @xmath387 $ ] , we could find a rational number @xmath354 ( in lowest terms ) such that @xmath388 assume that @xmath186 ( the case @xmath188 is similar ) . by proposition [ continuous of modulus ] , proposition [ equivalence of conjunction ] and the estimate ( [ approximation ] ) , we have @xmath389 next , we only need to give the estimate of @xmath390 .
in fact , we take @xmath391 $ ] , and by lemma [ lemma 4 ] , lemma [ lemma 5 ] ( 2 ) , we obtain @xmath392 we derive from ( [ approximation ] ) and @xmath391 $ ] that @xmath393 finally , combining ( [ final 1 ] ) with ( [ final 2 ] ) , @xmath394 where @xmath395 which only depends on @xmath396 . we know that @xmath36 depends on @xmath397 ,
so @xmath218 .
this completes the proof . in @xcite
, moser showed that any monotone twist diffeomorphism on the cylinder @xmath15 can be regarded as the time-1 map of a periodic lagrangian system . in
what follows , we specialize to the lagrangians with one and a half degrees of freedom .
firstly , let s briefly recall some basic notions and results of mather theory . for proofs and details
, we refer to @xcite , @xcite .
let @xmath398 be a @xmath399 lagrangian satisfying tonelli conditions : 1 .
convexity : for each @xmath400 , @xmath401 is strictly convex in @xmath402 coordinate .
superlinearity : @xmath403 3 .
completeness : all solutions of the euler - lagrange equation @xmath404 are well defined for the whole @xmath405 .
let @xmath406 $ ] be an interval , and @xmath407 be an absolutely continuous curve , we denote by @xmath408 the action of @xmath409 .
an absolutely curve @xmath410 is called a _ minimizer _ or _
action minimizing curve _ if @xmath411 we call @xmath412 a _ globally minimizing curve _ if for all @xmath413 is a minimizer on @xmath414 $ ] .
notice that the minimizer satisfies the euler - lagrange equation .
let @xmath415 be the space of euler - lagrangian flow invariant probability measures on @xmath416 .
to each @xmath417 , note that @xmath418=0 for each exact 1-form @xmath419 . therefore , given @xmath420 and a closed 1-form @xmath421 $ ]
, we can define _ mather s @xmath422 function _ @xmath423 it s easy to be checked that @xmath424 is finite everywhere , convex and superlinear .
we associate to @xmath417 its rotation vector @xmath425 in the following sense : @xmath426\rangle=\int_{tm\times{\mathbb{s}^1}}\eta_c d\mu,\quad \forall c\in h^1({\mathbb{s}^1},{\mathbb{r}}).\ ] ] so we can define _ mather s @xmath370 function _ : @xmath427 @xmath370 is finite , convex , superlinear and @xmath370 is the legendre - fenchel dual of @xmath422 .
[ alpha and beta ] if @xmath428 is a tonelli lagrangian , then 1 .
the function @xmath429 is strictly convex and differentiable at all @xmath430 2 .
the function @xmath431 is differentiable everywhere .
next , we introduce the generalization of peierls barrier to several degrees of freedom . for each @xmath420 and @xmath432 , we define a function @xmath433 , @xmath434 @xmath435,{\mathbb{s}^1})}}\int_0^n ( l-\eta_c)(d\gamma(s),s)ds .\ ] ] then , following mather , we introduce the barrier function on @xmath436 @xmath437 this function is useful in the construction of connecting orbits ( see @xcite ) . by proposition [ alpha and beta ]
, @xmath438 exists for every @xmath420 , and the flat piece of graph @xmath422 has rational slope .
[ barrier equivalence ] let @xmath401 be a tonelli lagrangian whose time-1 map is an area - preserving twist diffeomorphism , then 1 . for @xmath139 , there exists a unique @xmath439 such that @xmath440 , and @xmath441 2 . for rational number @xmath354 ( in lowest terms ) , let @xmath442 , @xmath443 , then @xmath444 @xmath445 $ ] corresponds to the flat of graph @xmath422 with rational slope @xmath354 .
* now , let @xmath446 be a tonelli lagrangian whose time-1 map @xmath447 is an exact area - preserving monotone twist diffeomorphism . for the rotation interval
@xmath210 $ ] , there is a compact annulus @xmath211 $ ] with sufficiently large @xmath448 so that the minimal orbits of @xmath447 satisfies @xmath449.\ ] ] + the space @xmath450 is provided with the norm : @xmath451 * @xmath452 , by abuse of notation , we use the same symbol @xmath453 to denote the real number in @xmath13 or the closed 1-form @xmath454 of @xmath455 .
* let @xmath456 denote the barrier function ( [ barrier h ] ) associated to the lagrangian @xmath457 .
* by proposition [ alpha and beta ] and [ barrier equivalence ] , for irrational number @xmath458 , we denote by @xmath459 the unique number satisfying @xmath460 in addition , we can define @xmath461 for rational numbers as proposition [ barrier equivalence ] ( 2 ) .
similarly , let the lagrangian @xmath462 be a perturbation of @xmath401 , one can also define @xmath463 in the same way .
[ main theorem 2 ] let @xmath464,~{\mathbf{a}}_{2k}$ ] be as shown above .
there exist constants @xmath465 and @xmath466 such that for any lagrangian @xmath462 , if @xmath467 and @xmath468 $ ] , we have , 1 . for @xmath139 ,
@xmath469 2 . for rational number @xmath7 ,
@xmath470 and @xmath471 for the lagrangian of many degrees of freedom , the barrier function @xmath472 could also be defined , one may ask whether there are similar results in this case . in general , it is not true and there re counterexamples which could be found in @xcite .
however , under additional assumptions , such as nearly - integrable lagrangians of arbitrary degrees of freedom , we can also obtain some similar results , see @xcite for details .
the time-1 map @xmath447 of the lagrangian @xmath401 is an exact area - preserving monotone twist diffeomorphism .
then , there exists @xmath473 such that if @xmath467 , the time-1 map @xmath474 of @xmath462 is also a monotone twist diffeomorphism satisfying @xmath475 with the constant @xmath476 .
therefore , by proposition [ barrier equivalence ] , it s not hard to observe that theorem [ main theorem 2 ] is an easy consequence of ( [ 22222 ] ) and theorem [ main theorem 1 ] .
the destruction of invariant circles or lagrangian tori ( converse kam theory ) is an interesting and important topic in dynamic systems ( see for example , @xcite
@xcite @xcite @xcite ) . in this section , by applying theorem [ main theorem 1 ] , we give an open and dense property about the destruction of invariant circles . a real number @xmath139 is _ diophantine _ if there exist constants @xmath477 and @xmath478 such that @xmath479 a real number is _ liouville _ if it s not diophantine .
[ application ] let @xmath480 be the set of all @xmath224 exact area - preserving monotone twist diffeomorphisms and let @xmath7 be a liouville number .
then there exists a set @xmath481 which is open and dense in @xmath482 in the @xmath224 topology , such that for all @xmath483 , there is no homotopically non - trivial @xmath484invariant circle with rotation number @xmath7 .
* case @xmath485 * : @xmath7 is irrational .
denote by @xmath481 the set of all @xmath224 exact area - preserving monotone twist diffeomorphisms that do nt admit any homotopically non - trivial invariant circles with rotation number @xmath7 .
we only need to prove that the set @xmath481 is open and dense in @xmath482 . given an exact area - preserving monotone twist diffeomorphism @xmath486 , we assume that @xmath0 admits a homotopically non - trivial @xmath484invariant circle with rotation number @xmath7 . then by theorem 2.1 in @xcite , for any neighbourhood @xmath487 of @xmath0 in @xmath482 , we could find @xmath488 which does not admit any homotopically non - trivial @xmath489invariant circle of rotation number @xmath7 .
so @xmath490 , which proves that @xmath481 is a dense set in @xmath482 .
on the other hand , we know that the peierls barrier @xmath491 if and only if there exists a homotopically non - trivial @xmath484invariant circle of rotation number @xmath7 ( see @xcite ) . take @xmath483 , then there exists a point @xmath492 such that @xmath493 . by theorem [ main theorem 1 ] , we have @xmath494 thus , there exists a small neighbourhood @xmath495 of @xmath0 in @xmath482 such that for all @xmath496 , we have @xmath497 then @xmath498 , which proves that @xmath481 is an open set . * case @xmath499 * : @xmath500 is rational .
there is no homotopically non - trivial @xmath484invariant circle of rotation number @xmath354 if and only if the peierls barrier @xmath501 and @xmath502 .
denote by @xmath481 the set of all @xmath224 exact area - preserving monotone twist diffeomorphisms that do nt admit any homotopically non - trivial invariant circles with rotation number @xmath354 .
we need to prove that @xmath481 is open and dense .
in fact , the denseness of @xmath481 was proved by mather in @xcite or @xcite , and the proof of openness is similar with case @xmath503 by theorem [ main theorem 1 ] .
similar results also hold for analytic topology .
notice that in analytic situations , we need an additional condition on the irrational number @xmath7 , i.e. , @xmath504 where @xmath505 is the sequence of approximants given by the continued fraction expansion of @xmath7 ( see@xcite ) . a set is called _ residual _ if it is a countable intersection of open and dense sets .
there exists a residual set @xmath506 in the @xmath224 topology which satisfies : for each @xmath507 , there is an open and dense set @xmath508 such that for all @xmath509 , @xmath0 does not admit any homotopically non - trivial invariant circle with rotation number @xmath7 .
let @xmath510 denotes the set of all rational numbers in @xmath13 . by theorem
[ application ] , we obtain an open and dense set @xmath511 such that for all @xmath512 , there is no homotopically non - trivial invariant circle with rotation number @xmath513 .
we set @xmath514 , it is a residual set in @xmath482 in the @xmath224 topology .
take @xmath515 , then we obtain that the peierls barriers @xmath516 and @xmath517 for all @xmath360 . we deduce from proposition [ continuous of modulus ] that @xmath518 is right - continuous at the rotation symbol @xmath519 and left - continuous at the rotation symbol @xmath520
thus , there exists an open interval @xmath521 such that @xmath522 if we set @xmath523 , then it s open and dense since @xmath510 is dense in @xmath13 , and @xmath524 which means that there is no homotopically non - trivial @xmath525-invariant circle with rotation number @xmath526 this completes our proof .
the authors were supported by national basic research program of china ( 973 program ) ( grant no .
2013cb834100 ) , national natural science foundation of china ( grant no . 11631006 , grant no .
11201222 ) and a program papd of jiangsu province , china . | let @xmath0 be an exact area - preserving monotone twist diffeomorphism of the infinite cylinder and @xmath1 be the associated peierls barrier . in this paper , we give the hlder regularity of @xmath1 with respect to the parameter @xmath0 .
in fact , we prove that if the rotation symbol @xmath2 , then @xmath1 is @xmath3-hlder continuous in @xmath0 , i.e. @xmath4 where @xmath5 is a constant .
similar results also hold for the lagrangians with one and a half degrees of freedom . as application
, we give an open and dense result about the breakup of invariant circles . |
it is well - known that the unit cotangent bundle of any riemannian @xmath1-manifold @xmath2 has a canonical contact structure .
let @xmath3 be a @xmath4-manifold and @xmath5 a @xmath6-map .
then @xmath7 is called a _ @xmath8wave@xmath9 front _ if it is the projection of a legendrian immersion into the unit cotangent bundle of @xmath10 .
now let @xmath11 be a front , where @xmath12 is the space form of constant curvature @xmath13 .
then the associated parallel front @xmath14 , i.e. the surface that is equi - distant from @xmath7 at a distance @xmath15 ( called the _ parallel surfaces _ of @xmath16 ) , is well - defined .
moreover , if @xmath7 is a flat immersion , so is @xmath17 for @xmath15 close to zero . using this fact
, we shall define a flat front : a front @xmath7 is _ flat _ in a neighborhood of @xmath18 if either 1 .
@xmath19 is a regular point of @xmath7 and the gaussian curvature of @xmath7 near @xmath19 vanishes , or 2 .
@xmath19 is a singular point of @xmath7 and @xmath17 is a flat immersion around @xmath19 for all @xmath20 close to zero .
a front @xmath21 is called a _ flat front _ if it is flat everywhere on @xmath3 . for the case
@xmath22 , there have appeared several articles concerning the singularities of developable surfaces in @xmath23 .
in particular , izumiya and takeuchi @xcite proved that the set of developable surfaces whose singularities are only cuspidal edges , swallowtails or cuspidal cross caps are open and dense in the set of non - cylindrical developable surfaces , where @xmath24 represents a cuspidal edge , @xmath25 a swallowtail , and @xmath26 a cuspidal cross cap . recently
, geometric inequalities for complete flat fronts in hyperbolic @xmath0-space and complete maximal surfaces with certain singularities in minkowski @xmath0-space were found in @xcite and @xcite .
we also note that kitagawa has made a deep investigation of flat tori in the @xmath0-sphere ( @xcite , @xcite , @xcite ) .
the study of global properties of surfaces with singularities is a newly - developing research area in differential geometry . in this paper
, we shall investigate singularities of flat surfaces in the hyperbolic @xmath0-space @xmath27 .
the geometry of flat fronts in @xmath28 has been investigated in @xcite , @xcite .
in particular , an analogue of the osserman inequality for minimal surfaces in @xmath23 was given in @xcite .
like the case of constant mean curvature one surfaces in @xmath28 , flat surfaces have a representation formula in terms of holomorphic data , found by j. a. glvez , a. martnez and f. miln @xcite : let @xmath29 and @xmath30 be holomorphic @xmath31-forms on a simply - connected riemann surface @xmath3 such that @xmath32 is positive definite .
then there exists a holomorphic immersion @xmath33 such that @xmath34 and its projection to @xmath28 gives a flat front @xmath35 in @xmath28 , where we regard @xmath28 as @xmath36 moreover , any simply - connected flat front has such a representation with respect to the complex structure induced by the second fundamental form ( see @xcite , @xcite and @xcite ) .
we call @xmath29 and @xmath30 in the _ canonical forms _ of @xmath7 . in section [ sec : local ] of this paper , we will show the following : [ thm : criterion ] let @xmath37 be a flat front with canonical forms @xmath38 , where @xmath39 is a local complex coordinate . 1 .
a point @xmath18 is a singular point if and only if @xmath40 holds .
the image of @xmath7 around a singular point @xmath19 is locally diffeomorphic to a cuspidal edge if and only if @xmath41 hold at @xmath19 , where @xmath42 .
the image of @xmath7 around a singular point @xmath19 is locally diffeomorphic to a swallowtail if and only if @xmath43 hold at @xmath19 , where @xmath44 is the schwarzian derivative @xmath45 of the function @xmath46 with respect to @xmath39 , that is , @xmath47 consequently , cuspidal edges and swallowtails are stable under perturbations of @xmath48 .
it is well - known that generic fronts ( which might not be flat ) admit only cuspidal edges or swallowtails ( see arnold ( * ? ? ?
* section 21.6 ) ) .
however , density of the set of such fronts within the smaller set of flat fronts does not immediately follow . using theorem [ thm : criterion ]
, we shall prove a similar assertion for flat fronts in @xmath28 ( theorem [ thm : genericity ] ) .
moreover , we shall prove the following global result in section [ sec : global ] .
a front @xmath37 is called complete if there exist a compact set @xmath49 and a symmetric @xmath4-tensor @xmath50 on @xmath3 such that @xmath50 is identically @xmath51 outside @xmath52 and @xmath53 is a complete riemannian metric of @xmath3 , where @xmath54 is the first fundamental form of @xmath7 .
[ thm : global ] let @xmath55 be a complete flat front which is not a covering of an hourglass ( hourglasses are rotationally symmetric , see example [ ex : revolution ] ) , and let @xmath56 be the family of parallel fronts of @xmath7 . then , except for only finitely many values of @xmath15 , all the singular points of @xmath17 are locally diffeomorphic to cuspidal edges or swallowtails .
the image of the singular points under an hourglass is a single point in @xmath28 , namely the so - called `` cone - like singularity '' ( see figure [ fig : revolution ] in section [ sec : example ] ) , and any parallel front of an hourglass has the same singularity .
thus the assumption of theorem [ thm : global ] is necessary .
fronts which admit only cuspidal edges and swallowtails are called a - mersions , and their topological properties have been investigated by langevin , levitt and rosenberg @xcite .
the above theorem implies that complete flat fronts in @xmath28 are generically included in this category .
the union of singular sets for the entire parallel family of a given flat front is called a _ caustic_.
we note that roitman @xcite very recently studied the geometric properties of flat surfaces , motivated by a classical result of l. bianchi ( see section [ sec : caustic ] ) .
to prove theorem [ thm : criterion ] , we shall give criteria for a singular point on a front to be a cuspidal edge or a swallowtail , as follows : let @xmath10 be a riemannian @xmath0-manifold , @xmath57 a domain on @xmath58 , and let @xmath59 be a @xmath6-map with a singular point @xmath60 .
then there exist three functions @xmath61 such that @xmath62 where @xmath63 is a local coordinate system of @xmath10 .
the rank of a map defined by @xmath64 does not depend on the choice of local coordinate @xmath63 nor on the choice of coordinate @xmath65 .
now we assume @xmath7 is a front .
a singular point @xmath60 of @xmath7 is called _ non - degenerate _ if the jacobian matrix of @xmath66 is of rank one at @xmath19 .
there exists a regular curve near a non - degenerate singular point @xmath19 @xmath67 ( called a singular curve ) such that @xmath68 , and so that the image of @xmath69 coincides with the set of singularities of @xmath7 near @xmath19 .
the tangential direction of @xmath70 is called the _ singular direction _ , and a non - zero vector @xmath71 such that @xmath72 represents the _ null direction_. for each point @xmath70 , vectors in the null direction @xmath73
are uniquely determined up to non - zero scalar multiplication .
[ prop : criterion - front ] let @xmath10 be a riemannian @xmath0-manifold and @xmath74 be a non - degenerate singular point of a front @xmath75 . 1 .
the germ of the front @xmath7 at @xmath19 is locally diffeomorphic to a cuspidal edge if and only if @xmath76 is not proportional to @xmath77 , where @xmath78 .
the germ of the front @xmath7 at @xmath19 is locally diffeomorphic to a swallowtail if and only if @xmath76 is proportional to @xmath77 and @xmath79 we shall prove this proposition in section [ sec : criterion ] .
these criteria are useful in other situations . in fact
, this proposition is applicable for the study of singularities of maximal surfaces in minkowski space ( see @xcite ) .
the authors thank shyuichi izumiya and go - o ishikawa for fruitful discussions and valuable comments , especially ishikawa for pointing out an error in a preliminary version of our paper .
first , we recall well - known properties for singular points from @xcite .
let @xmath80 be a @xmath81-function defined on an open interval @xmath82 containing the origin such that @xmath83 .
then @xmath84 has an _ @xmath85-singularity _ at @xmath51 if @xmath86 where @xmath87 and @xmath88 . here
, we shall consider the cases @xmath89 and @xmath0 .
let @xmath90 be an open subset of @xmath91 containing the origin @xmath92 .
a map @xmath93 is called an _ unfolding _ of @xmath84 if @xmath94 holds .
moreover , if @xmath84 has an @xmath85-singularity at @xmath51 and the matrix @xmath95 is of rank @xmath96 , then @xmath97 is called a _
versal unfolding _ of @xmath84 , where , for example , @xmath98 the set @xmath99 is called the _ discriminant set _ of @xmath97 .
the following fact is useful : [ fact : descriminant ] suppose @xmath100 has an @xmath85-singularity ( @xmath89 or @xmath0 ) at @xmath51 and @xmath101 a versal unfolding of @xmath84
. then 1 .
@xmath102 is locally diffeomorphic to a cuspidal edge at @xmath92 if @xmath89 .
2 . @xmath102 is locally diffeomorphic to a swallowtail at @xmath92 if @xmath103 .
let @xmath10 be a riemannian @xmath0-manifold and @xmath104 the unit cotangent bundle .
a @xmath81-map @xmath105 is called a ( _ wave _ ) _ front _ if there exists a legendrian immersion @xmath106 such that @xmath107 , where @xmath108 is the projection .
we call @xmath109 the _ legendrian lift _ of @xmath7
. we shall use the following lemma , first pointed out by zakalyukin @xcite : [ fact : z ] let @xmath110 be a neighborhood of the origin , and let @xmath111 be fronts .
suppose that @xmath112 is a singular point of @xmath113 and the set of regular points of @xmath113 is dense in @xmath57 for each @xmath114 .
then the following two statements are equivalent : 1 .
[ item : fact - z-1 ] there exist neighborhoods @xmath115 of the origin @xmath112 and a local diffeomorphism on @xmath23 which maps the image @xmath116 to @xmath117 , namely the image of @xmath118 is locally diffeomorphic to that of @xmath119 .
[ item : fact - z-2 ] there exists a local diffeomorphism @xmath120 on @xmath121 and a local contact diffeomorphism @xmath97 on @xmath122 which sends fibers to fibers such that @xmath123 , namely the lift @xmath124 is _ legendrian equivalent _ to the lift @xmath125 . we shall prove the lemma in the appendix .
again , we shall return to the general setting : since any contact structure is locally equivalent to the canonical contact structure on @xmath126 , we may restrict our attention to fronts in the euclidean @xmath0-space @xmath23 .
let @xmath127 be a domain in @xmath121 and @xmath128 a front . identifying the unit cotangent bundle @xmath129 with the unit tangent bundle @xmath130
, there exists a unit vector field @xmath131 such that the legendrian lift @xmath109 is expressed as @xmath132 . since @xmath133 is legendrian , @xmath134 hold , where @xmath135 is the euclidean inner product of @xmath23 .
we call @xmath136 the _ unit normal vector field _ of the front @xmath7
. then there exists a @xmath81-function @xmath137 such that @xmath138 where @xmath139 denotes the cross product of @xmath23 . obviously , @xmath140 is a singular point of @xmath7 if and only if @xmath141 .
[ prop : non - deg ] a singular point @xmath60 of a front @xmath128 is non - degenerate if and only if @xmath142 holds at @xmath19 .
differentiating at @xmath19 , we have @xmath143 .
this implies that the rank of @xmath144 at @xmath19 is at most 1 , and that @xmath145 is equivalent to @xmath146 .
therefore , @xmath145 if and only if @xmath147 has rank 1 at @xmath19 , that is , the map @xmath66 in the introduction has rank 1 at @xmath19 .
let @xmath19 be a non - degenerate singular point of a front @xmath148 . since the set of singular points is the set @xmath149 , proposition [ prop : non - deg ] implies that the set of singular points is parametrized by a smooth curve @xmath150 in a neighborhood of @xmath19 , so that @xmath68 .
we call the curve @xmath70 a _ singular curve _ passing through @xmath19 , and the direction @xmath77 the _ singular direction _ at the singular point @xmath19 , where the dot denotes the derivative with respect to @xmath15 . since @xmath19 is a non - degenerate singular point , so is any point @xmath70 for sufficiently small @xmath15
. then there exists a unique direction @xmath151 up to scalar multiplication such that @xmath152 for each @xmath15 .
we call @xmath73 , which is smooth in @xmath15 , the _ null direction_. [ def : c - s - type ] let @xmath19 be a non - degenerate singular point of a front @xmath128 , @xmath70 the singular curve with @xmath68 , and @xmath73 the null direction
. then 1 .
@xmath19 is of _ type c _ if @xmath76 is not proportional to @xmath77 .
@xmath19 is of _ type s _ if @xmath76 is proportional to @xmath77 and @xmath153 holds , where @xmath154 and @xmath73 are considered as column vectors in @xmath121 .
this definition does not depend on the choices of @xmath69 and @xmath155 .
[ ex : canonical - singularities ] the map
@xmath156 gives a cuspidal edge along the @xmath39-axis .
the null direction is perpendicular to the @xmath39-axis , and it has a type c singularity at @xmath112 . the map @xmath157 gives a swallowtail at @xmath112 .
the singular curve is @xmath158 , and the null direction is parallel to the @xmath159-axis .
so @xmath112 is a singularity of type s. the above @xmath160 , @xmath161 satisfy that @xmath162 , @xmath163 and the derivative @xmath164 vanishes identically on the singular curve .
we shall now prove that any front @xmath165 can be given such a parameterization @xmath166 near a non - degenerate singular point , as follows : we assume that the origin @xmath112 of the @xmath167-plane is an arbitrarily given non - degenerate singular point of @xmath7 , namely @xmath168 and set @xmath169 then we have : [ prop : normalization ] suppose that @xmath112 is a non - degenerate singular point of a front @xmath170 .
then there is a diffeomorphism @xmath171 with @xmath172 and a rotation at the origin @xmath173 such that @xmath174 satisfies the following properties : 1 .
[ item : normalize-1 ] @xmath162 , 2 .
[ item : normalize-3 ] @xmath163 , 3 .
[ item : normalize-4 ] the derivative @xmath164 vanishes identically along the singular curve .
in particular @xmath175 holds .
[ item : normalize-5 ] if @xmath112 is of type c , the tangent vector @xmath176 can be chosen to be the singular direction at the origin of @xmath177 .
let @xmath70 be the singular curve passing through @xmath112 .
the null direction @xmath73 can be extended to a vector field @xmath178 on @xmath57 , that is , @xmath179 on the other hand , we take a vector @xmath180 which is not proportional to @xmath76 and satisfies @xmath181 if @xmath112 is of type c , we choose @xmath182 to be proportional to @xmath77 .
then there exists a vector field @xmath183 on @xmath57 such that @xmath184 the vector fields @xmath183 and @xmath178 are linearly independent in a neighborhood of the origin . hence by a lemma in @xcite , there exists a new coordinate system @xmath185 such that @xmath186 and @xmath187 ( resp .
@xmath188 ) is proportional to @xmath183 ( resp .
@xmath178 ) . scaling @xmath183 and @xmath189
, we may assume @xmath190 without loss of generality .
from now on , we use the coordinates @xmath185 .
however , for notational simplicity , we drop the overhead tilde s and write @xmath185 as just @xmath65 .
so we may assume : * the derivative @xmath191 vanishes identically on the singular curve @xmath70 . *
if @xmath112 is of type c , the tangent vector @xmath192 points in the singular direction at the origin .
since @xmath193 has unit length , we can take a rotation at the origin @xmath194 which maps @xmath193 to @xmath195 , and set @xmath196 then we have @xmath197 we set @xmath198 since @xmath199 there exists a function @xmath200 such that @xmath201 and @xmath202 , namely , @xmath203 then by @xmath204 @xmath166 gives a new coordinate system .
we now set @xmath205 then [ item : normalize-3 ] follows immediately . by differentiating
, we have @xmath206 and we get @xmath207 thus , by differentiating , we have @xmath208 which implies [ item : normalize-1 ] . on the other hand , @xmath209 since @xmath210 vanishes on @xmath69 , so does @xmath211 .
thus we have @xmath212 by differentiating @xmath213 with respect to @xmath159 , we have @xmath214 here , @xmath215 does not vanish near @xmath112 , since @xmath216
. then we have @xmath217 and thus @xmath164 vanishes on the singular curve , which proves [ item : normalize-4 ] . if @xmath112 is of type c , then @xmath195 is proportional to the singular direction of @xmath218 . since @xmath162
, we have [ item : normalize-5 ] . in the proof above
, @xmath219 gives the unit normal vector field of the ( normalized ) front @xmath220 .
in addition to the case of surfaces , we shall define fronts for plane curves .
[ def : planar - front ]
let @xmath221 be an interval .
a map @xmath222 is called a _ ( planar ) front _ if there exists a map @xmath223 such that @xmath224 is perpendicular to @xmath225 and @xmath226 is an immersion .
a point @xmath227 with @xmath228 is called a singular point of the planar front @xmath225 , where @xmath229 . at a singular point @xmath230 ,
@xmath231 holds by definition .
a planar front is a projection of a legendrian immersion in the unit cotangent bundle @xmath232 with respect to the canonical contact structure . in the cases of @xmath233 and @xmath161 in example [ ex : canonical - singularities ]
, one can easily check that their slices @xmath234 perpendicular to the @xmath39-axis give planar fronts .
the tangent line of @xmath235 is given by @xmath236 where @xmath237 is the unit normal vector of @xmath235 .
then @xmath238 is the envelope of this family of tangent lines , and the discriminant set @xmath239 characterizes the image of @xmath233 and @xmath161 . according to this observation , we shall prove that @xmath220 as in proposition [ prop : normalization ] , which has type c or type s singularities at @xmath112 , also satisfies that 1 .
the slice perpendicular to the @xmath39-axis gives a planar front , 2 .
the set @xmath239 of @xmath97 given by is a discriminant set and is locally diffeomorphic to a cuspidal edge or a swallowtail , by applying fact [ fact : descriminant ] .
now we shall prove ( a ) for non - degenerate singular points as follows : [ prop : slice ] let @xmath112 be a non - degenerate singular point of the front @xmath240 satisfying [ item : normalize-3 ] and [ item : normalize-4 ] in proposition [ prop : normalization ] .
then there exists an @xmath241 such that the map defined by @xmath242 is a planar front .
moreover , @xmath159 is a singular point of @xmath238 if @xmath166 is a singular point of @xmath218 . to prove proposition [ prop : slice ] ,
we need the following : [ lem : after ] under the assumptions of proposition [ prop : slice ] , the derivative @xmath243 is non - zero and perpendicular to @xmath244 , where @xmath245 is the unit normal vector field of @xmath218 .
since @xmath218 is a front , it follows from [ item : normalize-4 ] in proposition [ prop : normalization ] that @xmath243 does not vanish .
since @xmath246 , we have @xmath247 since @xmath248 by [ item : normalize-4 ] in proposition [ prop : normalization ] , we have @xmath249 thus , @xmath250 which is the desired conclusion .
we fix @xmath39 and let @xmath251 then @xmath225 is a map into the @xmath252-plane . by [ item : normalize-3 ] in proposition
[ prop : normalization ] , we have @xmath253 and @xmath254 this implies that a singular point of @xmath255 is a singular point of @xmath218 . on the other hand , we set @xmath256 since @xmath257 @xmath224 is a well - defined unit vector field near @xmath112 .
moreover , @xmath258 where we used the fact that @xmath163 .
thus , @xmath224 is a normal vector of @xmath225 . by lemma [ lem : after ] , we have @xmath259 , and @xmath260 hence @xmath261 for sufficiently small @xmath166 , and the map @xmath262 is an immersion . in this section
, we shall prove proposition [ prop : criterion - front ] in the introduction .
as pointed out in the beginning of the previous section , it is sufficient to prove the assertion for fronts in the euclidean @xmath0-space @xmath23 .
the idea of the proof is as follows : let @xmath112 be a non - degenerate singular point of a front @xmath263 . then by proposition [ prop :
normalization ] , we have a normalized front @xmath220 .
we set @xmath264 by proposition [ prop : slice ] , there exist positive numbers @xmath265 and @xmath266 such that @xmath267 gives a planar front for @xmath268 ; that is , there exists a unit normal vector field @xmath269 such that @xmath270 if we set @xmath271 , the equation @xmath272 gives the tangent line of the planar front @xmath273 , and the image of the planar front is the envelope of these tangent lines . on the other hand ,
it is well - known that the envelope generated by a family of lines @xmath274 is given by @xmath275 .
so if we set @xmath276 the discriminant set @xmath277 coincides with the image of the front @xmath218 .
now we set @xmath278 then if @xmath279 has an @xmath85-singularity ( @xmath280 ) and @xmath97 is a versal unfolding , we can conclude ( b ) , that is , the image of @xmath218 is locally diffeomorphic to a cuspidal edge or a swallowtail , by fact [ fact : descriminant ] . according to this plan
, we shall first prove the criterion for cuspidal edges .
first , we prepare three lemmas : [ lem : cusp-1 ] @xmath281 and @xmath282 hold , where @xmath283 denotes the derivative with respect to @xmath159 .
these are easily computed from proposition [ prop : normalization ] and [ prop : slice ] .
[ lem : cusp-2 ] @xmath284 .
these are computed by differentiating and by using lemma [ lem : cusp-1 ] .
[ lem : cusp-3 ] @xmath279 has an @xmath285-singularity at @xmath286 if and only if @xmath287 . by differentiating @xmath288 and using @xmath289 , we have + @xmath290 . since @xmath282 by lemma [ lem : cusp-1 ] , @xmath291 forms an orthonormal basis for @xmath121 .
therefore , @xmath292 on the other hand , by differentiating and by using lemma [ lem : cusp-1 ] , we have @xmath293 hence @xmath84 has an @xmath294-singularity at the origin if and only if @xmath295 .
this and prove the assertion .
[ prop : c - type ] suppose @xmath112 is a non - degenerate singular point . then the germ of the image of the front is locally diffeomorphic to a cuspidal edge if and only if @xmath112 is of type c. by lemma [ fact : z ] , local diffeomorphic equivalence between singular points on fronts implies legendrian equivalence .
since a cuspidal edge itself is of type c , any singular point locally diffeomorphic to a cuspidal edge is of type c. conversely , we shall show that a singularity of type c is locally diffeomorphic to a cuspidal edge . to prove this ,
it is sufficient to show that @xmath84 as above has an @xmath294-singularity and @xmath97 is versal . by lemma [ lem : cusp-2 ] and
, we have @xmath296 this matrix is of rank @xmath4 , since @xmath297 and @xmath298 are linearly independent .
next we prove that @xmath84 has an @xmath294-singularity .
we set @xmath299 where @xmath245 is the unit normal vector of the front @xmath218 . here
@xmath300 on the singular curve . since we have assumed that @xmath112 is of type c , [ item : normalize-5 ] in proposition [ prop : normalization ] implies that @xmath301 is the singular direction at the origin
so we have @xmath302 on the other hand , @xmath303 holds at @xmath112 , because @xmath304 .
since @xmath305 is parallel to the @xmath252-plane , we have @xmath306 in particular we have @xmath307 , and by lemma [ lem : cusp-3 ] , @xmath279 has an @xmath294-singularity at @xmath286 .
next , we prove the criterion for swallowtails : [ prop : s - type ] suppose @xmath112 is a non - degenerate singular point
. then the germ of the image of the front is locally diffeomorphic to a swallowtail if and only if @xmath112 is of type s. to prove this , we prepare a lemma : [ lem : swallow-1 ] suppose @xmath112 is a non - degenerate singular point of @xmath220 , but not of type c. then 1 .
@xmath308 , in particular @xmath309 , and 2 .
@xmath310 . by [ item : normalize-4 ] of proposition
[ prop : normalization ] , @xmath164 vanishes identically on the singular curve . since @xmath112 is not of type c , the singular direction is equal to the null direction @xmath311 , thus @xmath308 . in particular
, we have @xmath309 . differentiating by @xmath159 and @xmath39 and substituting the relation @xmath312 we have @xmath313 since @xmath314 , we have @xmath315 . since @xmath316 is orthogonal to @xmath1 , @xmath317 is proportional to @xmath316 . to show @xmath318 , it is sufficient to show @xmath319 .
moreover , @xmath320 vanishes identically , so @xmath319 is equivalent to @xmath321 .
differentiating @xmath322 with respect to @xmath39 and using the relation @xmath323 , we have @xmath324 since @xmath112 is not of type c , @xmath311 is the singular direction . in particular
@xmath325 holds .
since @xmath142 at @xmath112 , we have @xmath326 . hence by
, we have @xmath321 . for the same reason as in the proof of proposition [ prop :
c - type ] , any singular point locally diffeomorphic to a swallowtail is of type s. conversely , we shall show that a singularity of type s is locally diffeomorphic to a swallowtail . to prove this , it is sufficient to show that @xmath84 as above has an @xmath327-singularity and @xmath97 is versal . by lemma [ lem : cusp-2 ] and
, we have @xmath328 by lemma [ lem : swallow-1 ] , @xmath329 , and then the rank of this matrix is @xmath0 . by lemma [ lem : cusp-1 ] and lemma [
lem : swallow-1 ] , we have @xmath330 by differentiating @xmath288 twice , we have @xmath331 thus , @xmath279 has an @xmath327-singularity if and only if @xmath332 , which is equivalent to @xmath333 , since @xmath163 .
since @xmath334 and the singular curve @xmath69 is given by @xmath335 , the singular direction is given by @xmath336 on the other hand , the null direction @xmath155 is given by @xmath337 since @xmath112 is of type s , we have @xmath338 by the definition of @xmath339 , we have @xmath340 here , since @xmath341 is proportional to @xmath342 , we have @xmath343 thus , we have @xmath344 then we get @xmath345 using @xmath346 and , we have @xmath347 which proves @xmath333 .
izumiya and takeuchi @xcite gave criteria for the singularities of a non - cylindrical flat ruled front in the euclidean @xmath0-space @xmath23 to be cuspidal edges , swallowtails and cuspidal cross caps .
one can prove this for the case of cuspidal edges and swallowtails by directly applying our criteria .
in this section , we give a proof of theorem [ thm : criterion ] in the introduction , and show that , generically , singular points of flat fronts are cuspidal edges or swallowtails .
we denote by @xmath348 the minkowski @xmath349-space with the inner product @xmath135 of signature @xmath350 .
the hyperbolic @xmath0-space @xmath28 is considered as the upper half component of the two sheet hyperboloid in @xmath348 with the metric induced by @xmath135 . identifying @xmath348 with @xmath351 , the set of @xmath352-hermitian matrices , as @xmath353 one has @xmath354 for @xmath355 , and @xmath28
is represented as @xmath356 the tangent space of @xmath28 at @xmath357 is the set of vectors in @xmath348 which are perpendicular to @xmath19 : @xmath358 we define a bilinear , skew - symmetric product @xmath139 as @xmath359 where @xmath360 , @xmath361 and @xmath19 are considered as matrices in @xmath351 , and the products of the right - hand side are matrix multiplications .
it is easy to show that @xmath362 is a vector in @xmath363 and perpendicular to both @xmath360 and @xmath361 .
we call `` @xmath139 '' the _ cross product _ of @xmath28 .
let @xmath3 be an oriented simply - connected riemannian @xmath4-manifold , and let @xmath364 be a front whose legendrian lift is @xmath365 identifying @xmath366 with @xmath367 , we can write @xmath133 , where @xmath368 is a unit vector in @xmath369 such that @xmath370 holds for each @xmath18 .
we call @xmath136 the _ unit vector field _ of the front @xmath7 .
suppose that @xmath7 is flat , then there is a ( unique ) complex structure on @xmath3 and a holomorphic legendrian immersion @xmath371 such that @xmath7 and @xmath109 are projections of @xmath372 , where being a holomorphic legendrian map means that @xmath373 is off - diagonal .
in particular , @xmath35 , with @xmath28 considered to be as in .
( see @xcite , @xcite and @xcite for details . )
if we set @xmath374 the first and the second fundamental forms @xmath54 and @xmath375 are given by @xmath376 we call @xmath29 and @xmath30 the _ canonical forms _ of the front @xmath7 .
the holomorphic @xmath4-differential @xmath377 which appears in the @xmath378-part of @xmath54 , is called the _ hopf differential _ of @xmath7 . by definition ,
the umbilic points of the front @xmath7 coincide with the zeroes of @xmath379 .
we remark that the @xmath380-part of the first fundamental form @xmath381 is positive definite on @xmath3 .
conversely , the following assertion holds ( see @xcite for the front case and @xcite for the regular case ) : let @xmath29 and @xmath30 be holomorphic @xmath31-forms on a simply - connected riemann surface @xmath3 such that @xmath32 is positive definite .
then the solution of the ordinary differential equation @xmath382 gives a holomorphic legendrian immersion of @xmath3 into @xmath383 , where @xmath384 is a base point , and its projection into @xmath28 gives a flat front .
conversely , any flat front is locally expressed in this manner .
[ rem : parallel ] if we identify @xmath28 with the upper - half component of the hyperboloid in minkowski @xmath349-space @xmath348 , the parallel surface of @xmath7 is written as @xmath385 where @xmath15 is the signed distance from @xmath7 and @xmath136 is the unit normal vector of @xmath7 in @xmath28 .
as pointed out in @xcite and @xcite , @xmath386 then the canonical forms @xmath387 and @xmath388 of @xmath17 are written as @xmath389 let @xmath37 be a flat front .
then , on a neighborhood of @xmath19 , we can take a holomorphic legendrian immersion @xmath372 as in .
since @xmath390 is positive definite , it holds that either @xmath391 or @xmath392 .
so , by and the fact that @xmath35 , we have @xmath393 thus , if we write @xmath394 and @xmath395 in a complex coordinate @xmath39 , we have @xmath396 and then @xmath397 where `` @xmath139 '' is the cross product as in .
thus , the singular set is the set of zeroes of the function @xmath398 then @xmath19 is a singular point if and only if @xmath399 hence ( 1 ) is proven .
since @xmath7 is a front , @xmath400 as in is positive definite .
hence @xmath401 holds on a singular point @xmath19 .
moreover , at a singular point @xmath19 , we have @xmath402 hence a singular point @xmath19 is non - degenerate if and only if @xmath403 holds at @xmath19 .
let @xmath19 be a non - degenerate singular point , that is , and hold at @xmath19 .
let @xmath70 be a singular curve such that @xmath68 . since @xmath404 holds on @xmath70
, @xmath405 holds on @xmath70 , because of , where @xmath406 is the hermitian inner product on @xmath407 .
hence @xmath408 gives the singular direction .
thus , by a suitable choice of the parameter @xmath15 , the singular curve @xmath70 can be parametrized as @xmath409 the first fundamental form @xmath54 is written as @xmath410 on the curve @xmath70 .
now we set @xmath411 .
since @xmath412 , there exists a holomorphic function @xmath413 defined on a neighborhood of @xmath19 such that @xmath414 . since @xmath415 on the singular curve @xmath70
, we have @xmath416 thus the null direction @xmath73 is given by @xmath417 so we have @xmath418 here , by proposition [ prop : criterion - front ] , @xmath19 is diffeomorphic to a cuspidal edge if @xmath419 at @xmath420 .
hence , we have ( 2 ) .
next , let us prove ( 3 ) .
using and , we can compute that @xmath421.\ ] ] hence , if @xmath422 at @xmath19 , then @xmath423 this proves ( 3 ) , because of proposition [ prop : criterion - front ] .
[ rem : descriptiong ] we set the two hyperbolic gauss maps to be @xmath424 as shown in @xcite , we have the following expression @xmath425 where @xmath426 determines which member of the parallel family @xmath17 of @xmath7 we have . in this @xmath427-construction of flat fronts , it is convenient to rewrite the conditions in theorem [ thm : criterion ] in terms of @xmath427 .
we have the following identities , which will be useful for an application of theorem [ thm : criterion ] ( see @xcite ) : @xmath428 where @xmath429 , @xmath430 , @xmath431 is as in and @xmath432 represents the schwarzian derivative of @xmath433 with respect to @xmath39 : @xmath434 as an application of theorem [ thm : criterion ] , we shall now show that generic singularities of flat fronts are cuspidal edges or swallowtails .
let @xmath57 be a simply - connected domain in @xmath407 , and @xmath435 the set of holomorphic functions on @xmath57 .
then , for each @xmath436 , we can construct a flat front @xmath437 which is represented by a pair of holomorphic @xmath31-forms @xmath438 .
conversely , if @xmath19 is not an umbilic point ( i.e. @xmath439 ) , both @xmath440 and @xmath441 are not equal to zero , and we can choose a complex coordinate such that @xmath442 and @xmath443 .
thus , any flat front is locally congruent to some @xmath444 , except in neighborhoods of umbilic points .
we remark that an umbilic point can not be a singular point , since @xmath390 is positive definite . by theorem [ thm :
criterion ] , we have the following 1 .
the zeroes of @xmath445 correspond to singular sets .
moreover , a singular point @xmath60 is non - degenerate if and only if @xmath446 .
2 . a singular point @xmath19 is a cuspidal edge if and only if it is non - degenerate and @xmath447 .
3 . a singular point @xmath19 is a swallowtail if and only if it is non - degenerate , @xmath448 and @xmath449\neq 0 $ ] .
we let @xmath450 be the space of @xmath96-jets of holomorphic functions on @xmath57 .
then @xmath450 is canonically identified with the product space @xmath451 : @xmath452 in particular , @xmath450 can be considered as a @xmath6-manifold of dimension @xmath453 as well as a complex manifold of dimension @xmath454 . for a compact set @xmath455 of @xmath57 and an open subset @xmath456 in @xmath450 , we set @xmath457_k:=\{h\in \mathcal o(u)\,;\ , { j^kh}(k)\subset o\}.\ ] ] let @xmath458 be the topology of @xmath435 generated by such @xmath459_k$ ] , which is called _ the compact open @xmath460-topology . _
if @xmath461 is the canonical projection , it can be easily seen that @xmath462 is a continuous map and @xmath463_{k+1}=[k , o]_k$ ] . in particular , @xmath464 holds , and @xmath465 gives a topology on @xmath435 , called the _
compact open @xmath6-topology_. a holomorphic function @xmath436 is an interior point of a given subset @xmath466 ( with respect to the compact open @xmath6-topology ) if and only if there exist a non - negative integer @xmath467 and a finite number of compact sets @xmath468 in @xmath57 and open subsets @xmath469 such that @xmath470_l \subset \mathcal{v}\;.\ ] ] now we give a topology on the family of flat fronts @xmath471 induced from the compact open @xmath6-topology on @xmath435 .
we shall prove the following : [ thm : genericity ] let @xmath455 be an arbitrary compact set of @xmath57 and @xmath472 the subset of @xmath473 which consists of @xmath444 whose singular points on @xmath455 are locally diffeomorphic to cuspidal edges or swallowtails .
then @xmath472 is an open and dense subset of @xmath474 .
generic properties of @xmath6-maps are usually described in terms of the whitney @xmath6-topology ( cf .
@xcite ) , because it is suitable for the technique of multiplying by a cut - off function .
however , generic properties of analytic functions are different in the whitney @xmath81-topology . in the above theorem
, we use the compact open @xmath6-topology .
the two topologies are the same when the source space is compact .
however , they are different on @xmath435 .
in fact , when the source space is non - compact , the compact open @xmath6-topology satisfies the second axiom of countability , but the whitney @xmath6-topology on @xmath435 does not satisfy even the first axiom of countability and can not be treated by sequence convergence .
we do not know if the set @xmath475 ( which consists of @xmath444 whose singular points on @xmath57 are locally diffeomorphic to cuspidal edges or swallowtails ) is an open dense subset with respect to the whitney @xmath6-topology .
we set @xmath476 { \operatorname{re}}\left[e^{-h(p)}\left(h''(p)-\frac{1}{2}h'(p)^2\right)\right]=0 \end{array } \right .
\right\}.\end{aligned}\ ] ] then @xmath477 and @xmath478 are both closed subsets of @xmath479 .
the set given by @xmath480_2\ ] ] corresponds to @xmath472 under the identification @xmath481 , which is by definition an open subset in @xmath435 .
so it is sufficient to show that @xmath482 is a dense subset .
obviously , @xmath477 is a real closed submanifold of @xmath479 with codimension three .
we remark that @xmath483 is an open submanifold of @xmath479 .
the following lemma holds : [ submfd ] @xmath484 is a submanifold of @xmath479 with codimension three .
we define a @xmath6-map @xmath485 by @xmath486 \right)\\ & = \left(u , e^{-u/2}(v_1\cos\frac{v}2-u_1\sin\frac{v}2 ) , \right . \\ & \phantom{=(u , e^{-u/2}(v_1\cos } \left .
\frac{e^{-u}}2\left((-u_1 ^ 2+v_1 ^ 2 + 2u_2)\cos v+2(-u_1v_1+v_2)\sin v \right ) \right),\end{aligned}\ ] ] where we set @xmath487 then @xmath488 gives the canonical coordinate system on @xmath479 . by a direct calculation
, we have @xmath489 .
we show that @xmath490 is a regular value of @xmath491 . to determine the rank of the jacobian matrix of @xmath491 at any point in @xmath492
, we calculate the derivative of @xmath491 with respect to @xmath493 and @xmath494 : @xmath495 then we have @xmath496 we now suppose @xmath497 .
then @xmath498 hold and thus @xmath499 then and @xmath500 imply that @xmath501 , namely , that @xmath488 belongs to @xmath477 .
hence @xmath502 is of rank @xmath0 at @xmath492 in @xmath483 . by the implicit function theorem
, @xmath484 is a submanifold of codimension @xmath0 .
we shall prove that @xmath482 is a dense subset .
we now fix a function @xmath436 .
let @xmath503 be the set of polynomials of degree at most @xmath4 in @xmath39 and define a map as follows @xmath504 obviously the map @xmath433 is a diffeomorphism .
in particular , @xmath505 and @xmath506 are both submanifolds of dimension @xmath507 diffeomorphic to @xmath477 and @xmath484 respectively .
let @xmath508 be the canonical projection . since @xmath503 is a @xmath6-manifold of dimension @xmath509 , sard
s theorem yields that @xmath510 and @xmath511 are measure zero sets in @xmath503 .
thus @xmath512 is also a measure zero set .
thus there is a sequence @xmath513 in @xmath503 such that @xmath514 converges to the zero polynomial and @xmath515 we set @xmath516 .
then @xmath517 , that is , @xmath518_2
\subset \hat s(k).\ ] ]
let @xmath519 be a distance function on @xmath520 which is compatible with respect to its topology .
then a sequence @xmath521 in @xmath522 _ converges to @xmath413 uniformly _ on a given compact subset @xmath455 of @xmath57 if for any @xmath523 , there exists a positive integer @xmath524 such that @xmath525 holds .
we remark that @xmath526 holds .
since the difference @xmath527 is only a polynomial @xmath514 of degree at most @xmath4 converging to the zero polynomial , one can easily check that for each non - negative integer @xmath467 , @xmath528 converges to @xmath529 uniformly on any compact subset of @xmath57 .
let @xmath530 be an open neighborhood of @xmath120 in @xmath435 .
then by the definition of the compact open @xmath6-topology , there exist a non - negative integer @xmath467 , a finite number of compact sets @xmath468 of @xmath57 , and open subsets @xmath531 of @xmath532 such that @xmath533_{l}\subset \mathcal{v}.\ ] ] we set @xmath534 note that @xmath535 is a compact set .
since @xmath528 converges to @xmath529 uniformly on any compact subset of @xmath57 , there exists an integer @xmath536 such that @xmath537 on the other hand , @xmath538 then @xmath539 this implies that @xmath540 if @xmath541 .
thus @xmath542_{l}$ ] holds for all @xmath543 and @xmath544_{l}\subset \mathcal v \qquad ( n\ge n_0 ) \ ; .\ ] ] since @xmath545 , this implies that @xmath482 is a dense subset .
in this section , we shall give a proof of theorem [ thm : global ] in the introduction .
let @xmath37 be a flat front defined on a riemann surface @xmath3 . in this section
, we do not assume that @xmath3 is simply - connected .
thus the holomorphic lift @xmath372 of @xmath7 is defined only on the universal cover @xmath546 of @xmath3 : @xmath547 and then the canonical forms @xmath29 and @xmath30 as in are holomorphic @xmath31-forms defined on @xmath546 . note that the first and second fundamental forms as in , the hopf differential as in , and the @xmath380-part @xmath400 of the first fundamental form are all well - defined on @xmath3 .
moreover @xmath400 is positive definite on @xmath3 .
we have that @xmath548 from now on , we assume @xmath7 is _ complete _ , that is , there exist a compact set @xmath49 and a symmetric @xmath4-tensor @xmath50 such that @xmath50 is identically zero outside @xmath52 and @xmath53 is a complete riemannian metric ( see @xcite ) .
we remark that @xmath7 is complete if and only if ( see @xcite ) 1 . [
item : complete-1 ] the @xmath380-part @xmath400 of the first fundamental form is complete ( in this case , we say that @xmath7 is `` weakly - complete '' ) , 2 .
[ item : complete-2 ] @xmath400 has finite total absolute curvature , and 3 .
[ item : complete-3 ] the singular set is a compact set of @xmath3 . in the proof of theorem [ thm : global ] , we use only properties [ item : complete-1 ] and [ item : complete-2 ] ; that is , the conclusion of theorem [ thm : global ] holds for weakly - complete flat fronts such that @xmath400 has finite absolute total curvature . by completeness , we know that there exist a compact riemann surface @xmath549 and a finite number of points @xmath550 in @xmath549 such that @xmath551 ( ( * ? ? ? * lemma 3.3 ) , see also @xcite ) .
we call the points @xmath552 the _ ends _ of @xmath7 . moreover , as shown in ( * ? ? ?
* lemma 2 ) , the hopf differential @xmath379 can be extended meromorphically on @xmath549 , and at each end @xmath553 there exists a complex coordinate @xmath39 around @xmath553 such that @xmath554 and the canonical forms are written as @xmath555 where @xmath556 and @xmath557 are holomorphic functions in @xmath39 which do not vanish at the origin .
since @xmath558 and @xmath559 do not depend on the choice of complex coordinates , we denote @xmath560 these are the orders of the pseudometrics @xmath561 and @xmath562 , respectively . by
, we have @xmath563 where , by convention , @xmath564 if @xmath565 .
since @xmath7 is complete , @xmath400 is a complete riemannian metric on @xmath3 ( * ? ? ?
* corollary 3.4 ) .
thus , we have @xmath566 [ def : cylindrical ] an end @xmath553 is called _ cylindrical _ if @xmath567 let @xmath433 and @xmath568 be the hyperbolic gauss maps of @xmath7
. then @xmath433 and @xmath568 are both meromorphic functions on @xmath3 , and @xmath569 for all @xmath18 .
[ fact : regular ] at an end @xmath553 , the following properties are equivalent : 1 .
@xmath433 is meromorphic at @xmath553 .
@xmath568 is meromorphic at @xmath553 .
@xmath570 , that is , @xmath379 has at most a pole of order @xmath4 at @xmath553 .
[ def : regular ] an end @xmath553 is called _ regular _ if the three properties in fact [ fact : regular ] hold .
otherwise , @xmath553 is called _ irregular_. [ rem : cylindrical ] the ends of the hyperbolic cylinders are regular and cylindrical . as a special case of (
* theorem 6 ) , a regular cylindrical end is asymptotic to a finite cover to a hyperbolic cylinder . an _ umbilic point _
@xmath571 is a zero of the hopf differential @xmath379 .
when @xmath379 is identically zero , that is , @xmath7 is totally umbilic , @xmath7 represents the horosphere . in this section
, we assume that @xmath7 is not totally umbilic .
since @xmath379 is meromorphic on the compact riemann surface @xmath549 , the number of umbilic points is finite . as @xmath400 is positive definite at @xmath572 , implies that either @xmath573 holds at each umbilic point @xmath572 . using a local complex coordinate @xmath39 , we write @xmath574
let @xmath549 be a compact riemann surface and @xmath575 a complete flat front which is not totally umbilic .
using the canonical forms @xmath29 and @xmath30 in , we define @xmath576 though @xmath577 might be defined only on the universal cover of @xmath3 , implies that @xmath578 is well - defined on @xmath3 .
moreover , by , @xmath578 can be extended on @xmath549 as a continuous map @xmath579.\ ] ] as seen in section [ sec : local ] , the set of singular points of the flat front @xmath7 is given by @xmath580 using a local expression as in , we define @xmath581 where @xmath430 , @xmath582 and @xmath431 is as in .
[ lem : criterion - global ] the quantities in are independent on the choice of complex coordinate .
in particular , @xmath183 is a meromorphic @xmath0-form on @xmath549 , and both @xmath583 and @xmath584 are meromorphic functions on @xmath549 .
since @xmath578 is well - defined on @xmath3 , @xmath585 is well - defined on @xmath3 , and then so is its @xmath586-part .
hence @xmath587 is a meromorphic @xmath31-form on @xmath3 .
moreover , by and , @xmath588 is a meromorphic @xmath31-form on @xmath549 . since @xmath379 is a meromorphic @xmath4-differential on @xmath549
, @xmath589 is a meromorphic @xmath0-differential . as the symmetric product @xmath590 is a meromorphic @xmath4-differential
, @xmath591 is a meromorphic function on @xmath549 .
though the schwarzian derivative as in depends on the choice of complex coordinates , the difference of two schwarzian derivatives is considered as a meromorphic @xmath4-differential ; that is , if we write @xmath592 in the complex coordinate @xmath39 , @xmath593\,dz^2\ ] ] is independent of the choice of a coordinate @xmath39 , as a meromorphic @xmath4-differential . here , by
, @xmath594 holds .
this shows that @xmath584 is a well - defined meromorphic function on @xmath3 .
moreover , by and the definition , @xmath584 is meromorphic at each end . using the invariants of , we define @xmath595 z_0(f)&:=\{p\in m^2\,;\,\xi(p)=0\},\\[6pt ] z_c(f)&:=\{p\in m^2\,;\,{\operatorname{im}}\sqrt{\zeta_c(p)}=0\},\\[6pt ] z_s(f)&:=\{p\in m^2\,;\,{\operatorname{re}}\zeta_s(p)=0\}. \end{array}\ ] ] though @xmath596 is multi - valued on @xmath3 , the condition @xmath597 is unambiguous .
then by theorem [ thm : criterion ] , the following hold : * @xmath18 is a singular point if and only if @xmath598 .
* @xmath599 is a non - degenerate singular point if and only if @xmath600 .
* @xmath599 is a cuspidal edge if and only if @xmath601 .
* @xmath599 is a swallowtail if and only if @xmath602 .
* a singular point @xmath599 is neither a cuspidal edge nor a swallowtail if and only if @xmath603 here we denote the complementary set by the upper suffix @xmath604 . since @xmath605 ( resp .
@xmath606 ) describes a criterion for a singular point to be a cuspidal edge ( resp .
a swallowtail ) , we use the lower suffix `` @xmath13 '' ( resp .
`` @xmath607 '' ) .
the sets @xmath608 , @xmath605 and @xmath606 are the same for all the parallel fronts of @xmath7 ; that is , if @xmath609 is the family of parallel fronts of @xmath7 , then we have : [ lem : parallel - criterion ] @xmath610 by , we have the first assertion .
though the remaining parts can be proved by direct calculations , we give an alternative proof : let @xmath433 and @xmath568 be the hyperbolic gauss maps of @xmath7 .
then by , we have @xmath611 and @xmath584 is written as in . since the hyperbolic gauss maps and the hopf differential are independent of the choice of parallel front @xmath17 , we have the conclusion . by a direct calculation using the formulas in the proof of lemma [ lem : parallel - criterion ] and
, we have @xmath612 using this , we can prove that : [ prop : revolution ] let @xmath7 be a complete flat front which is not totally umbilic .
then the function @xmath583 is constant if and only if @xmath7 is a covering of a front of revolution . if @xmath613 , @xmath614 holds on @xmath3 .
hence @xmath577 is constant . in this case
, one can conclude that @xmath7 is a covering of a hyperbolic cylinder , which is a surface of revolution .
on the other hand , assume @xmath583 is a non - zero constant . by
, @xmath615 can only have simple poles .
then by the definition of @xmath583 in , the order of @xmath379 is at least @xmath616 .
thus , by fact [ fact : regular ] , all ends must be regular . by
, @xmath617 holds . then by
, we have @xmath618 with respect to any complex coordinate @xmath39 .
then it holds that @xmath619 where @xmath620 denotes the mbius transformation . here , the group @xmath383 acts isometrically on @xmath28 as @xmath621 where we consider @xmath28 as in . under the isometry ,
the hyperbolic gauss maps transform as @xmath622 .
hence we may assume @xmath623 in is a jordan normal form .
when @xmath623 is diagonal , we have @xmath624 , where @xmath558 is constant . here , since @xmath7 is a flat front , @xmath433 and @xmath568 have no common branch points ( see @xcite ) .
thus @xmath433 has no branch point , and then we can take @xmath625 as a local coordinate .
hence @xmath7 is locally congruent to a front of revolution ( see example [ ex : revolution ] in section [ sec : example ] ) .
thus we have the conclusion .
if @xmath623 is not diagonal , the eigenvalue of @xmath623 is @xmath626 , which is a double root .
then we have @xmath627 .
since the ends of @xmath7 are the points where @xmath628 ( ( * ? ? ?
* lemma 4.10 ) ) , the ends are common poles of @xmath433 and @xmath568 . in this case , by we have @xmath629
. then the @xmath630 at a pole @xmath19 of @xmath433 is less than or equal to @xmath631 , which contradicts the fact that all ends are regular .
let @xmath632 be a complete flat front which is not totally umbilic , and @xmath56 its parallel family . for simplicity ,
we write @xmath633 a point @xmath18 is a singular point of @xmath17 which is neither a cuspidal edge nor a swallowtail if and only if @xmath634 then by lemma [ lem : parallel - criterion ] , @xmath17 admits such a singular point if and only if @xmath635since @xmath183 in is a meromorphic @xmath0-differential on the compact riemann surface @xmath549 and @xmath636 is the set of zeroes of @xmath183 , @xmath636 is a finite set of points . thus , to prove theorem [ thm : global ] , it is sufficient to show the following proposition : [ prop : global ] let @xmath7 be a complete flat front such that @xmath583 defined in is not constant
. then @xmath637 is a finite set . before proving this proposition
, we shall give a proof of theorem [ thm : global ] .
assume a complete flat front @xmath7 is a front of revolution .
such a flat front is a horosphere , a finite cover of a hyperbolic cylinder , a snowman , or an hourglass ( see example [ ex : revolution ] in section [ sec : example ] ) . among these
, the horospheres and hyperbolic cylinders do not have singular points , and all singularities of the snowman are cuspidal edges .
since we assumed @xmath7 is not a cover of an hourglass , we have the conclusion for the case of fronts of revolution .
next , we assume @xmath7 is not a front of revolution . then by proposition [ prop : revolution ] , @xmath583 is non - constant .
hence by proposition [ prop : global ] , @xmath638 is a finite set . on the other hand ,
the parallel front @xmath17 admits a singular point which is neither a cuspidal edge nor a swallowtail if and only if holds .
hence we have the conclusion . to prove proposition [ prop : global ] , we need the following lemma : [ lem : accumulate ] let @xmath7 be a complete flat front as in with non - constant @xmath583 .
assume @xmath639 accumulates at a point @xmath640
. then 1 .
@xmath19 is a non - umbilic point in @xmath3 or an irregular cylindrical end , and 2 .
there exists a neighborhood @xmath57 of @xmath19 such that the number of connected components of @xmath641 is finite , and each connected component is a level set of @xmath578 .
this lemma will be proven in section [ sec : proof - of - lemma ] later . using these
, we shall prove proposition [ prop : global ] .
assume @xmath642 then there exists an infinite sequence @xmath643 such that @xmath644 ( @xmath645 ) are mutually distinct .
since @xmath549 is compact , we can take a subsequence of @xmath646 which converges to @xmath647 .
thus by lemma [ lem : accumulate ] , @xmath648 is finite .
this is a contradiction because the @xmath644 @xmath649 are mutually distinct .
let @xmath19 be an accumulation point of @xmath639 , and take a sequence @xmath650 consisting of mutually distinct points in @xmath639 such that @xmath651 as @xmath652 .
we show the first assertion of the lemma by way of contradiction : we assume * @xmath18 is an umbilic point , or * @xmath653 is an end which is not an irregular cylindrical end , and set @xmath654 if @xmath19 is an umbilic point , @xmath655 holds because of .
if @xmath19 is an end , @xmath656 holds when @xmath19 is cylindrical .
so , we consider two cases : * @xmath657 , that is , @xmath19 is an umbilic point or a non - cylindrical end . *
@xmath656 , that is , @xmath19 is a cylindrical end . in this case
, @xmath19 is a regular end because of our assumption .
then by fact [ fact : regular ] and , we have @xmath658 .
[ [ case-1 ] ] _ case 1 : _
+ + + + + + + + + we assume @xmath657 .
if we take a complex coordinate @xmath39 around @xmath19 such that @xmath659 , we can write @xmath660 where @xmath661 denotes a higher - order term . on the other hand , the hopf differential @xmath379 is written as @xmath662 thus , it follows from that @xmath663 we assume @xmath664 . then by , and , we have @xmath665 let @xmath666
. then @xmath667 tends to the origin as @xmath652 .
since @xmath668 , @xmath669 holds . since @xmath670 , there exist sequences @xmath671 and @xmath672 of real numbers such that @xmath673 and @xmath674 as @xmath652 . here , by and
, we have @xmath675 giving a contradiction .
then the case @xmath664 is impossible .
assume @xmath676 .
in this case , is written as @xmath677 . then by the assumption that @xmath583 is non - constant
, there exists a positive integer @xmath467 such that @xmath678 in this case , by , we have @xmath679 here @xmath680 holds on a sequence @xmath681 such that @xmath682 as @xmath652 , and @xmath670 .
hence implies that @xmath683 .
thus , we have @xmath684 where @xmath685 as @xmath652 . again , and contradict each other .
[ [ case-2 ] ] _ case 2 : _
+ + + + + + + + + we assume @xmath658 . taking a complex coordinate @xmath39 such that @xmath659 , we can write @xmath686 .
denote by @xmath687 the branch order of @xmath433 at @xmath688 .
( for example , if @xmath689 , the branch order of @xmath433 at @xmath688 is @xmath519 . ) since @xmath690 , implies that @xmath691 since @xmath658 , it holds that @xmath692 .
hence we have @xmath693 .
thus , we can write @xmath694 where @xmath695 is an integer and @xmath696 . thus , @xmath697 as in case 1 , we set @xmath666
. then we have @xmath698 where @xmath699 .
this is impossible .
hence in any case , @xmath639 does not accumulate at @xmath51 .
we consider two cases .
[ [ case-1 - 1 ] ] _ case 1 : _
+ + + + + + + + + suppose that @xmath639 accumulates at a non - umbilic point @xmath18 , i.e. @xmath700 .
take a complex coordinate @xmath39 around @xmath19 with @xmath659 .
since @xmath701 , there exists a holomorphic function @xmath702 defined on a neighborhood of the origin such that @xmath703 on the other hand , both @xmath704 and @xmath705 have neither a zero nor a pole at @xmath688 , so by , @xmath596 is a holomorphic function near @xmath688 .
since @xmath583 is not a constant , there exists a positive integer @xmath467 such that @xmath706 , where @xmath696 . then by the weierstrass preparation theorem
, we can choose a coordinate @xmath39 such that @xmath707 moreover , replacing @xmath708 by @xmath39 , we can set @xmath709 here , since @xmath710 accumulates at @xmath51 , @xmath711 in must be real , and then @xmath712 holds . on the other hand , by , @xmath713 we identify a neighborhood of @xmath19 with a neighborhood of the origin of @xmath39-plane . since @xmath639 accumulates to the origin , we can take a sequence @xmath643 such that @xmath682 as @xmath652 .
then by and , there exists a sequence @xmath714 such that @xmath715 and @xmath716 hold . subtracting from ,
we have @xmath717 on the other hand , subtracting multiplied by @xmath718 from multiplied by @xmath467 , we have @xmath719 here , since @xmath715 , we deduce that @xmath720 substituting these into , we have @xmath721 ; that is , @xmath722 .
since @xmath723 holds for @xmath645 , the imaginary part of @xmath724 vanishes identically on @xmath725 , namely , @xmath726 similarly , one can prove that @xmath727 thus , on a neighborhood of the origin , @xmath639 is the imaginary axis in the @xmath39-plane .
next , we shall prove that the imaginary axis is a level set of @xmath578 . by , and , @xmath728 holds on the imaginary axis .
thus we have @xmath729 as seen in in section [ sec : local ] , the tangent vector field of a level set of @xmath578 is represented as @xmath730 on the other hand , @xmath731 without loss of generality , @xmath732 holds on @xmath733 , where @xmath57 is a neighborhood of the origin , because a zero of @xmath734 is isolated in @xmath549 . then the tangent vector of the level set of @xmath578 at a point on the imaginary axis is parallel to the imaginary axis . hence the level set passing through a point of the imaginary axis is the imaginary axis .
that is , @xmath639 coincides with the imaginary axis , which is a level set of @xmath578 .
[ [ case-2 - 1 ] ] _ case 2 : _
+ + + + + + + + + suppose now that @xmath639 accumulates at an irregular cylindrical end @xmath19 .
let @xmath39 be a complex coordinate with @xmath659 . by irregularity , @xmath735 holds . without loss of generality
, we may assume @xmath630 is an even number .
in fact , if we set @xmath736 , that is , we take the double cover of a neighborhood of @xmath19 , the order @xmath379 at the origin with respect to the coordinate @xmath159 will be an even number .
hence , we assume @xmath737 where @xmath738 is an integer .
the second equality holds because @xmath19 is a cylindrical end .
since @xmath379 has even order at the origin , @xmath739 is a meromorphic function on a neighborhood of @xmath51 .
more precisely , we can write @xmath740 since @xmath741 , implies that @xmath742 , that is , there exists an integer @xmath467 ( @xmath743 ) such that @xmath744 @xmath745 . then by the weierstrass preparation theorem , we can choose a coordinate @xmath39 such that @xmath746 then by , @xmath584 is written as @xmath747 as @xmath643 is a sequence with @xmath682 , we have @xmath748 where @xmath715 . subtracting from , and multiplied by @xmath749 from multiplied by @xmath467 , we have @xmath750 since @xmath715 , this yields @xmath751 , and then @xmath752 for sufficiently large @xmath1 .
thus , we have @xmath753 let @xmath754 then @xmath755 is a set consisting of a finite number of lines in the @xmath39-plane through the origin , and by , each @xmath667 lies on some @xmath756 .
hence there exists a subset @xmath757 such that each @xmath756 ( @xmath758 ) contains an infinite number of elements of @xmath646 .
we fix @xmath758
. then we can take a subsequence @xmath759 of @xmath646 such that @xmath760 and @xmath761 .
since @xmath762 and @xmath763 , we have @xmath764 this shows that , on a neighborhood of the origin , @xmath639 coincides with the set of lines @xmath765 .
next , we show that @xmath756 ( @xmath758 ) is a level set of @xmath578 for each @xmath766 . by and , @xmath767 holds on @xmath756 .
hence we have @xmath768 at any point in @xmath756 , the argument of the tangent vector of the level set of @xmath578 is @xmath769 and then , the tangent vector is proportional to the line @xmath756 .
hence each @xmath756 ( @xmath758 ) is a level set of @xmath578 .
thus , we have the conclusion .
[ rem : conical ] let @xmath653 be an accumulation point of @xmath639 . then by the second part of lemma [ lem : accumulate ] , @xmath639 is a level set of @xmath578 in a neighborhood of @xmath19 ; that is , by taking a suitable parallel front , we may assume that a component of @xmath639 is a part of the singular set .
since the null direction and the singular direction coincide at each point in @xmath639 , the image of such a singular set is a single point in @xmath28 .
if the point @xmath19 is not an end , such a singularity seems to be a so - called _ cone - like singularity _ , see , for example , the hourglass in example [ ex : revolution ] .
another example is as follows : set @xmath770 on @xmath407 .
then by solving , we have a flat front @xmath771 .
the singular set of @xmath7 contains the imaginary axis , which coincides with @xmath639 ( see figure [ fig : conical ] ) .
however , this example is not complete because the canonical forms have an essential singularity at @xmath772 . ,
in @xcite , roitman investigated the caustic of ( the parallel family of ) a flat front , which is considered as the locus of singular points of the fronts in the parallel family . in this section
, we discuss caustics of flat fronts from our point of view .
let @xmath773 be a simply connected domain and @xmath774 a flat front without umbilic points .
we denote by @xmath29 and @xmath30 the canonical forms of @xmath7 , and @xmath775 , as in the previous section .
since @xmath7 has no umbilic points , @xmath577 does not take values @xmath51 and @xmath776 .
for a point @xmath777 , we denote by @xmath778 and @xmath779 the principal curvatures of @xmath7 at @xmath39 .
then by , we have @xmath780 since @xmath781 , @xmath782 holds .
then there exists a real number @xmath783 such that @xmath784 , which is called the _ radius of curvature_. by , we have @xmath785 the _ caustic _ @xmath786 of @xmath7 is defined as @xmath787 where @xmath348 is the minkowski @xmath349-space and @xmath136 is the unit normal vector of @xmath7 .
in other words , @xmath786 is the locus of the centers of the principal curvature @xmath778 of @xmath7 .
let @xmath788 be the holomorphic lift of the front @xmath7 .
then @xmath7 and the unit normal vector @xmath136 are given by @xmath789 thus , the caustic of @xmath7 is @xmath790 e_f^*\\ & = e_f\begin{pmatrix}e^{r_1 } & 0 \\ 0 & e^{-r_1}\end{pmatrix}e_f^ * = e_f\begin{pmatrix } |\rho|^{1/2 } & 0 \\ 0 & |\rho|^{-1/2 } \end{pmatrix}e_f^*.\end{aligned}\ ] ] hence if we set @xmath791 we have @xmath792 where @xmath793 since @xmath57 contains no umbilic points of @xmath7 , both @xmath29 and @xmath30 have no zeroes . thus @xmath794 and @xmath795 have no common zero , which implies that : [ lem : caustic - front ] the caustic @xmath796 of a flat front @xmath774 without umbilic points is a flat front with canonical forms @xmath794 and @xmath795 as in .
moreover , we have @xmath797 where @xmath433 and @xmath568 are the hyperbolic gauss maps of @xmath7 . in particular , the hyperbolic gauss maps @xmath798 of @xmath786 are given by @xmath799 if @xmath39 is a singular point of @xmath7 , @xmath800 holds because @xmath801 .
therefore , the caustic of a parallel family @xmath56 of flat fronts is the locus of singular points of the fronts @xmath17 for @xmath802 .
since the parallel family has a common caustic , the sets @xmath636 , @xmath710 and @xmath803 in section [ sec : global ] can be considered as well - defined on the caustic . in particular
, we have the following : [ prop : sing - caustic ] let @xmath774 be a flat front without umbilic points , and with caustic @xmath786 , where @xmath773 is a simply connected domain . then 1 .
a point @xmath60 is a singular point of the caustic @xmath786 if and only if @xmath804 .
2 . a point @xmath804 is a non - degenerate singular point of the caustic if and only if @xmath805 holds at @xmath19 , where @xmath433 and @xmath568 are the hyperbolic gauss maps of @xmath7 .
a point @xmath804 where @xmath805 is a cuspidal edge of the caustic if and only if @xmath806 . in other words ,
the locus of the cuspidal edges of @xmath609 is the set of regular points of the caustic .
furthermore , the locus of the swallowtails of @xmath56 is the set of cuspidal edges on the caustic , except the points at which @xmath807 .
a point @xmath60 is a singular point of @xmath786 if and only if @xmath808 . by ,
this is equivalent to @xmath809 = |\hat\omega\hat\theta|{\operatorname{im}}\sqrt{\zeta_c}.\ ] ] hence the first assertion holds . in this case
, @xmath19 is a degenerate singular point of @xmath786 if and only if @xmath810 then by , we have the second assertion .
finally , if @xmath19 is a non - degenerate singular point of @xmath786 , @xmath19 is a cuspidal edge if and only if @xmath811 here , by direct calculation , @xmath812 holds . since @xmath813 is a positive real number if @xmath814 , holds if and
only if @xmath815 .
here we give examples that reaffirm the properties of singularities in theorems [ thm : criterion ] and [ thm : global ] .
we make examples of flat fronts by choosing hyperbolic gauss maps @xmath433 and @xmath568 as follows : let @xmath433 and @xmath568 be meromorphic functions on a compact riemann surface @xmath549 such that @xmath433 is not identically equal to @xmath568 , and let @xmath816 if the period condition @xmath817 holds for any loop @xmath69 on @xmath3 , we have the parallel family of a complete flat front @xmath818 by substituting @xmath433 and @xmath568 in the representation formula in remark [ rem : descriptiong ] with @xmath819 .
moreover , by , we have @xmath820 for details , see @xcite .
[ ex : cylinder ] let @xmath821 and @xmath822 on @xmath823 .
then @xmath824 and @xmath825 .
so @xmath826 if and only if @xmath420 and then all points of the front are singular . when @xmath420 , @xmath827 identically , and the surface degenerates to a single geodesic line .
when @xmath828 , we have a cylinder with no singularities ( figure [ fig : revolution ] ( a ) ) . [ cols="^,^,^,^ " , ] + _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the upper row shows ( from left to right ) the three cases @xmath829 , @xmath830 and @xmath831 for @xmath832 in example [ ex : nnoid ] . the lower row shows ( from left to right ) the three cases @xmath833 , @xmath834 and @xmath835 for @xmath836 and @xmath837 in example [ ex : mn ] . _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ [ ex : mn ] in this case , @xmath838 , and we have @xmath839 ends .
then @xmath840 for small values of @xmath841 and @xmath1 , we can easily investigate the singularities . for @xmath836 and @xmath837 ( resp .
@xmath842 ) , for all @xmath15 , all singularities are always cuspidal edges except two ( resp .
four ) swallowtails when @xmath833 ( resp .
@xmath843 ) and at one ( resp .
two ) degenerate singularity ( resp .
singularities ) when @xmath834 ( resp .
@xmath844 ) . as the value @xmath845 increases through @xmath846 ( resp .
@xmath847 ) , the two ( resp .
four ) swallowtails come together into a single ( resp .
two ) degenerate singularity ( resp
. singularities ) and then disappear , leaving only cuspidal edges .
surfaces for @xmath836 and @xmath837 and @xmath0 , and their corresponding caustics , are shown in figure [ fig:12 ] . for @xmath848 and @xmath842 , and for all @xmath15 ,
the singular points are always cuspidal edges or swallowtails .
all singular points are cuspidal edges , except for one swallowtail when @xmath849 . when @xmath850 , there are no swallowtails .
as @xmath15 increases to @xmath51 , the swallowtail moves out to an end and disappears when @xmath420 .
for @xmath420 , the singular set is the line @xmath851 , and hence the cuspidal edge travels out to the end @xmath772 .
in this appendix we prove lemma [ fact : z ] , as only a sketch of the proof given in @xcite .
the authors hope this will help readers who are not familiar with singularity theory .
[ fact : a1 ] let @xmath852 be an immersion of an @xmath1-manifold @xmath853 .
then for each point @xmath854 , there exists a neighborhood @xmath57 of @xmath19 such that the restriction @xmath855 is an embedding .
[ fact : a2 ] let @xmath856 and @xmath857 be neighborhoods of the origin @xmath858 in @xmath859 . let @xmath860 @xmath861 be two embeddings such that @xmath862 and @xmath863 . then there exists a local diffeomorphism @xmath864 such that @xmath865 holds .
it is well - known that a front can be considered as a projection of a legendre immersion @xmath866 , where @xmath57 is a domain in @xmath121 and @xmath867 is the projective cotangent bundle .
the canonical contact structure of the unit cotangent bundle @xmath129 is the pull - back of that of @xmath867 .
we remark that this contact structure on @xmath867 does not depend on the riemannian metric on @xmath23 ( see @xcite ) .
so we have the following : [ prop : a4 ] let @xmath871 @xmath861 be two fronts satisfying @xmath863 , whose associated legendrian immersions @xmath872 are embeddings , where @xmath873 are neighborhoods of the origin @xmath858 in @xmath121 .
suppose that there exists a relatively compact neighborhood @xmath874 of @xmath858 @xmath861 such that 1 .
[ item : propa4 - 1 ] the closure @xmath875 is contained in @xmath873 for @xmath876 .
[ item : propa4 - 2 ] the set of regular points of @xmath877 in @xmath875 is dense in @xmath875 @xmath861 .
[ item : propa4 - 3 ] @xmath878 [ item : fact - z-1 ] follows from [ item : fact - z-2 ] immediately .
so it is sufficient to show [ item : fact - z-1 ] implies [ item : fact - z-2 ] .
by fact [ fact : a3 ] , we may assume @xmath880 . without loss of generality we may assume that @xmath881 and @xmath882 are relatively compact and @xmath883 .
by fact a.1 , we may assume that the associated legendrian immersion @xmath884 is an embedding .
since @xmath881 and @xmath882 are relatively compact , we have @xmath885 thus by proposition [ prop : a4 ] , we have @xmath879 , in particular we have @xmath886 by fact [ fact : a2 ] , there exists a local diffeomorphism @xmath887 on @xmath121 such that @xmath888 , which proves the assertion .
suppose that @xmath892 is not a finite set .
without loss of generality , we can take a sequence @xmath650 such that @xmath893 moreover , by taking a subsequence we may assume @xmath650 converges to a point @xmath894 . then by continuity , we have @xmath895 . since @xmath890 , @xmath19 is a regular point of @xmath877 .
thus , there exists a neighborhood @xmath177 of @xmath19 such that @xmath896 is an embedding , which contradicts @xmath897 since @xmath898 for sufficiently large @xmath1 .
we fix @xmath899 arbitrarily . by the previous lemma , we may set @xmath900 we identify @xmath126 with @xmath901
. then @xmath902 ( @xmath876 ) is considered as a map into @xmath903 , and there exist unit vectors @xmath904 and @xmath905 such that @xmath906 since @xmath124 and @xmath125 are embeddings , @xmath904 ( resp .
@xmath907 ) are mutually distinct .
thus the image of @xmath877 at @xmath711 consists of a mutually transversal finite number of components of surfaces . since @xmath908 , we can conclude that @xmath909 and @xmath910 for a suitable permutation of @xmath911 .
now we set @xmath912 by , we have @xmath913 then by the continuity of @xmath124 and @xmath125 , we have @xmath914 thus it is sufficient to show that @xmath915 is dense in @xmath875 . in fact , suppose that @xmath916 has an interior point . by the assumption
( 2 ) of proposition a.4 , there exists an open subset @xmath917 such that @xmath918 and @xmath877 is an immersion on @xmath919 .
take a point @xmath920 .
let @xmath921 be the tangent plane ( as a two dimensional affine plane in @xmath23 ) of the regular surface @xmath922 at @xmath923 , and @xmath924 the orthogonal projection .
since @xmath925 has a regular point , @xmath926 contains an interior point . on the other hand , by sard s theorem ,
the critical value set @xmath927 of @xmath925 is a measure zero set . since @xmath928 , this makes a contradiction .
hence @xmath916 does not have any interior points . since @xmath929 @xmath930 is dense in @xmath875 .
99 v. i. arnold , s. m. gusein - zade and a. n. varchenko , singularities of differentiable maps , vol .
@xmath31 , monographs in math .
* 82 * , birkhuser ( 1985 ) .
j. w. bruce and p. j. giblin , curves and singularities , cambridge university press , ( 1984 ) .
j. a. glvez , a. martnez and f. miln , _
flat surfaces in hyperbolic @xmath0-space _ , math . ann .
* 316 * ( 2000 ) , 419435 .
m. golubitsky and v. guillemin , stable mappings and their singularities , gtm * 14 * , springer - verlag ( 1973 ) .
s. izumiya , _ on legendrian singularities _ ,
* 101 * ( 1987 ) , 748752
. s. izumiya and n. takeuchi , _ geometry of ruled surfaces _
, applicable math . in the golden age ( 2003 ) , 305338 .
y. kitagawa , _ periodicity of the asymptotic curves on flat tori in @xmath931 _ , j. math .
japan * 40 * ( 1988 ) 457476 .
y. kitagawa , _ embedded flat tori in the unit @xmath0-sphere _ , j. math .
japan * 47 * ( 1995 ) 275296 .
y. kitagawa , _ isometric deformation of flat tori in the @xmath0-sphere with nonconstant mean curvature _ , thoku math . j.
* 52 * ( 2000 ) 283298 .
s. kobayashi and k. nomizu , foundations of differential geometry i , john wiley & sons inc .
m. kokubu , w. rossman , m. umehara and k. yamada , _ flat fronts in hyperbolic @xmath0-space ii _ , in preparation .
m. kokubu , m. umehara and k. yamada , _ an elementary proof of small s formula for null curves in @xmath932 and an analogue for legendrian curves in @xmath932 _ , osaka j. math .
* 40 * ( 2003 ) , 697715 .
m. kokubu , m. umehara and k. yamada , _ flat fronts in hyperbolic @xmath0-space _ , pacific j. math .
* 216 * ( 2004 ) , 149175 .
p. roitman , _
flat surfaces in hyperbolic @xmath0-space as normal surfaces to a congruence of geodesics _ , preprint .
r. langevin , g. levitt and h. rosenberg , _
classes dhomotopie de surfaces avec rebroussements et queues daronde
dans @xmath933 _ , canad . j. math . * 47 * ( 1995 ) , 544572 .
m. umehara and k. yamada , _
maximal surfaces with singularities in minkowski @xmath0-space _ , to appear in hokkaido math .
j. v. m. zakalyukin , _ reconstructions of fronts and caustics depending on a parameter and versality of mappings _
, j. soviet math . 27
( 1984 ) 27132735 . | it is well - known that the unit cotangent bundle of any riemannian manifold has a canonical contact structure . a surface in a riemannian @xmath0-manifold
is called a ( wave ) front if it is the projection of a legendrian immersion into the unit cotangent bundle .
we shall give easily - computable criteria for a singular point on a front to be a cuspidal edge or a swallowtail . using this
, we shall prove that generically flat fronts in the hyperbolic @xmath0-space admit only cuspidal edges and swallowtails .
moreover , we will show that every complete flat front ( which is not rotationally symmetric ) has associated parallel surfaces whose singularities consist of only cuspidal edges and swallowtails . |
we start our appendices with the following discussion , which is a summary of the work done in deriving the device - independent guessing probability ( digp ) @xcite . a probability distribution that is the outcome distribution for some measurement on a quantum state
is called a quantum distribution .
for example , a distribution @xmath135 with elements @xmath136 is quantum if there exist at least one positive semi - definite hermitian unit trace matrix @xmath137 and at least one set of positive semi - definite hermitian matrices @xmath138 , @xmath139 satisfying @xmath140 and @xmath141 such that @xmath142 .
we will often abuse notation and refer to a distribution by its elements @xmath136 when there is no confusion in doing so .
the set @xmath143 of quantum distributions is closed and convex and a distribution in @xmath143 that can not be decomposed as a convex combination of other distributions is called _ extremal _ in @xmath143 . for a non - extremal distribution @xmath144
there is in general more than one possible convex decomposition .
a non - extremal distribution @xmath136 with a convex decomposition @xmath145 can be constructed by sampling the different distributions @xmath146 with probability @xmath147 . in this case knowledge about the convex decomposition chosen changes the ability of an eavesdropper to correctly guess the outcomes @xmath148 and/or @xmath149 . without knowledge of the decomposition , or for extremal distributions ,
the probability of correctly guessing the outcome of measurement @xmath150 is @xmath151 , the probability of the most likely outcome .
with knowledge of the decomposition @xmath145 , the probability is larger or equal to @xmath151 @xmath152 for a given observed non - extremal distribution it is possible that it was produced by an agent eve that has larger predictive power than an agent which only observes the outcomes . the maximal probability for the agent eve to correctly guess an outcome @xmath149 of measurement @xmath150 given a distribution @xmath153 and a free choice of decomposition is the digp @xmath154 @xmath155 where @xmath156 is labelling the convex decompositions of @xmath157 in terms of extremal distributions @xmath146 . for any open interval of @xmath143
the function @xmath154 is a concave function @xcite .
therefore this kind of maximization is called a _
concave roof _ construction .
the guessing probability as a function on the space of probability distributions is not everywhere continuous .
an example of this is that the family of bell - inequalities described in of ref .
@xcite certifies one bit of randomness for measurements on a state with arbitrarily little entanglement . the probability distribution corresponding to such a state and the measurements in eq .
[ eq : ibeta2 ] has @xmath158 and is at the same time arbitrarily close to a distribution corresponding to measurements on a product state with @xmath159 , i.e. , a distribution which can be prepared by a local deterministic procedure .
there is thus a discontinuity where the guessing probability jumps from @xmath63 to @xmath160 .
the key to understanding this discontinuity is that the local deterministic distribution is not extremal while the quantum distribution in the neighbouring point is extremal . as seen in eq .
[ noos ] , the guessing probability is given by different functions depending on whether a distribution can be decomposed into other distributions or not , i.e. , if it is extremal or not .
this means discontinuities can appear at the boundary between extremal points and non - extremal points .
we will now show that discontinuities can _ only _ appear at such boundaries between extremal and non - extremal points in the boundary @xmath161 of the quantum set @xmath32 . to do this we use the property of the guessing probability described in eq .
[ noos ] , together with some general properties of concave functions and in particular concave roof constructions .
we want to show that the following propositions are true : [ prop1 ] the function @xmath154 on the set of quantum distributions @xmath32 is continuous in the interior of @xmath32 .
[ prop2 ] the function @xmath154 is continuous in any extremal point of @xmath32 .
proposition [ prop1 ] is trivial .
the guessing probability @xmath154 is concave by definition and any concave function is continuous on an open subset of its domain @xcite .
in particular this means that @xmath154 is continuous in the interior of @xmath32 .
to address proposition [ prop2 ] we consider the restriction @xmath162 of @xmath154 to the boundary @xmath161 of the quantum set .
first we note that the function @xmath163 by definition is continuous on any open set of extremal points since @xmath164 is a continuous function .
next we observe that the boundary @xmath165 can be decomposed into a collection of open sets of extremal points and a collection @xmath166 of closed connected possibly overlapping sets where each set is the closure of a maximal open connected subset .
a maximal open connected subset @xmath167 of the non - extremal points is an open set such that any other open connected set of non - extremal points which contains @xmath167 is @xmath167 itself .
therefore , each set @xmath168 is the convex hull of the set of extremal points in its closure .
any closed set @xmath168 has a boundary @xmath169 with the rest of @xmath165 which can be decomposed in the same way into open sets of extremal points and closed connected sets @xmath170 that are closures of maximal open connected sets of non - extremal points .
the boundary @xmath171 of @xmath170 with the rest of @xmath169 is in turn decomposable in the same way .
continuing this successive decomposition of the boundary @xmath172 we will eventually reach sets @xmath173 that are one dimensional simplexes , or alternatively sets with only extremal points in the boundary . on sets of these two types @xmath154
is a continuous function . to see this we introduce the following terminology , and use a theorem from ref .
@xcite .
a function for which all discontinuities are such that the function takes the higher value at a closed set and the lower value at an open set is called _ upper semi - continuous_. the function @xmath174 defined on a closed convex set @xmath175 can be viewed as an extension of @xmath176 to the interior of @xmath175 . this extension is called the _ concave roof extension_. [ theo1]let c be a compact set and @xmath177 be the convex hull of c. if @xmath178 is bounded , upper semi - continuous , and concave on c , then the concave roof extension @xmath179 of f to k is upper semi - continuous @xcite .
the guessing probability is bounded and concave by definition . if the boundary of @xmath175 has only extremal points it follows that @xmath176 is continuous in @xmath180 and by theorem [ theo1 ] @xmath174 is upper semi - continuous on @xmath175 . moreover , since @xmath174 is concave it can not have an upper semi - continuous discontinuity between the boundary and the interior .
if @xmath175 is a one - dimensional simplex we can , if necessary , restrict the domain of the guessing probability to a one dimensional subspace and make the same argument .
next we consider discontinuities between @xmath175 and an open set of extremal points .
[ theo2]any discontinuity of @xmath154 between a closed set and an open set of extremal points is upper semi - continuous .
if the boundary point of the closed set is extremal the @xmath154 is continuous since @xmath151 is continuous .
next consider a non - extremal boundary point of the closed set .
@xmath154 in the non - extremal point is always greater or equal to @xmath181 by eq .
thus any discontinuity is upper semi - continuous . if there is a discontinuity of @xmath154 on the boundary of @xmath175 it is , by lemma [ theo2 ] , upper semi - continuous and at a set of non - extremal points . by repeated application of theorem [ theo1 ] and lemma [ theo2 ]
we can conclude that @xmath163 is upper semi - continuous on @xmath165 and that @xmath154 is upper semi - continuous on @xmath182 . since @xmath154 is concave there can not be an upper semi - continuous discontinuity between the boundary @xmath165 and the interior of @xmath183 .
thus the only discontinuities are between non - extremal points in closed subsets of @xmath165 and extremal points in open subsets of @xmath165 .
we have described the guessing probability as a function on set of quantum distributions , but it is sometimes useful to consider it as a function of the violation of some given bell inequality @xmath184 .
a bell expression is a linear function on the space of distributions and the set of distributions for which it takes a given value @xmath185 is a hyper - plane @xmath186 .
the different values of the bell expression thus defines a family of parallel hyperplanes . on each hyperplane
@xmath186 we can consider the restriction @xmath187 of @xmath154 to the intersection of @xmath186 with @xmath183 and take its maximum @xmath188 on this intersection .
this maximum is the highest probability for eve to guess the outcome of @xmath150 for any distribution @xmath189 such that @xmath190 .
the function @xmath188 can have a discontinuity at @xmath191 only if @xmath192 intersects with a point in @xmath183 at which @xmath193 is discontinuous .
let us consider a bell expression @xmath184 and its maximal value @xmath194 on @xmath143 . if the intersection of @xmath195 and @xmath143 is a single extremal point it follows from propositions [ prop1 ] and [ prop2 ] that there is a @xmath196 such that for the range @xmath197 for which @xmath188 is a continuous function of @xmath185 .
if the intersection of @xmath195 and @xmath143 contains more than one extremal point it also contains a set of non - extremal points of @xmath198 and @xmath154 could have a discontinuity between this set and an open set of extremal points .
this discontinuity could lead to a discontinuity of the function @xmath188 at @xmath194 .
let us consider a sequence of measurements @xmath199 chosen by bob and denote @xmath200 .
the convex decomposition of the observed outcome distribution that gives eve optimal probability to correctly guess the sequence of outcomes @xmath201 of the measurements @xmath202 is a function of @xmath203 .
the guessing probability @xmath204 is thus given by @xmath205 where the extremal distributions @xmath206 and weights @xmath207 of the optimal convex decomposition are functions of @xmath203 as indicated by the index @xmath208 .
let us assume that a term which appears in the convex combination is @xmath209 thus we assume that it corresponds to the most probable sequence of outcomes @xmath210 for a specific distribution indexed by @xmath208 .
given that eve has chosen the optimal convex decomposition for guessing the outcomes of @xmath211 we consider her probability of correctly guessing the outcome of @xmath212 for @xmath213 given a particular sequence of previous outcomes @xmath214 .
it is given by @xmath215 where @xmath216 is the probability that the distribution indexed by @xmath208 will be sampled given the sequence of previous outcomes @xmath214 @xmath217 the probability in eq .
[ klut ] is larger or equal to 1/2 but is lower or equal to @xmath218 , the maximal probability that eve could guess the outcome of @xmath212 correctly given that she had chosen an optimal strategy for this and not the optimal strategy for guessing the outcomes of the sequence @xmath211 .
thus if @xmath218 is close to @xmath63 so is the expression in eq .
we want to prove that @xmath219 can be made arbitrarily close to @xmath127 by making @xmath218 sufficiently close to 1/2 for each @xmath220 .
the proof relies on the fact that if a convex combination of a collection of numbers @xmath221 equals @xmath148 , i.e. , @xmath222 where @xmath223 , and if @xmath224 for each @xmath124 , it follows that for every @xmath124 either @xmath225 or @xmath226 . from this
follows that when @xmath218 is very close to 1/2 either @xmath227 in eq .
[ klut ] is very close to 1/2 or @xmath216 is very close to zero for each @xmath208 . to see this more clearly we construct the following bound @xmath228 where we used @xmath229 for each @xmath230 and @xmath231 .
it follows that @xmath232,\end{aligned}\ ] ] and given eq .
this implies @xmath233}.\end{aligned}\ ] ] thus for sufficiently small @xmath234 either @xmath235 can be made arbitrarily small , or the probability @xmath236 that the distribution labelled by @xmath208 is sampled when @xmath212 is measured is made arbitrarily small .
the argument can be made for any @xmath237 . for @xmath67
, it follows that either @xmath238 is made arbitrarily close to @xmath63 or @xmath207 is made arbitrarily close to @xmath125 . for @xmath100
, it follows that either @xmath239 is made arbitrarily close to @xmath63 or @xmath240 is made arbitrarily close to zero .
given the second option and that @xmath238 is made arbitrarily close to @xmath63 it is implied that that @xmath241 is made arbitrarily close to 0 .
if on the other hand @xmath238 is not very close to @xmath63 it follows that @xmath207 is made arbitrarily close to zero by the preceding argument . by induction
it is clear that either the term in eq .
[ kutt ] satisfies that @xmath242 can be made arbitrarily close to @xmath243 or alternatively @xmath207 is made arbitrarily small . since the same is true for every @xmath208 in eq .
[ gutt2 ] it follows that @xmath219 can be made arbitrarily close to @xmath243 .
we start once again with the @xmath66 inequality : @xmath244 with @xmath245 . in @xcite
it was proved that the maximal quantum value @xmath246 for this inequality was given by : @xmath247 as the authors explained , in order to maximize the bound of the inequality , one can restrict the space of quantum states to pure two qubit states @xcite , meaning that the maximal value achievable with quantum measurements on quantum states can always be achieved with ( two - outcomes projective ) measurements on a pure , two - qubit ( entangled ) state . fixing the basis , we can work with the state : + given this form of the state and by parametrising the measurements of alice and bob , one can verify numerically that : @xmath249 in the range @xmath250 , i.e. @xmath251 , the whole range of interest where the state is entangled . from , it is easy to obtain an upper bound on the expectation value : latexmath:[\[\label{numupper } where one inserts the _ observed _ value of the inequality @xmath66 , obtained by the observed probability distribution @xmath253 , in this equation . in terms of the guessing probability , this gives : @xmath254 + for example , one can insert the maximal quantum value @xmath255 in or in and get that @xmath256 or @xmath257 , which coincides with the certification of one perfect local random bit for input @xmath258 on bob s side for the maximal violation of @xmath66 .
our bound is thus tight at the maximal violation of the inequality .
since the probability distribution of maximal violation is unique , the point is necessarily an extreme point @xcite , so we can directly use the observed guessing probability @xmath253 to bound the eavesdropper s predictive power ( as an extreme point allows only for one decomposition : itself ) .
+ + if we now want to use our function to bound the guessing probability _ inside _ the set ( not only at the point of maximal violation ) , and following the arguments of @xcite , one can check that the function @xmath259 bounding the @xmath260 is a concave function of its variable @xmath66 : @xmath261 + where we used that both the numerator and denominator are positive from @xmath262 .
the bound can thus be extended to the points that do not necessarily violate maximally the inequality , and our bound @xmath259 can be used in our protocol for unbounded randomness certification from a single pair of qubits .
+ + we will now give some graphs of this upper bound on @xmath260 from the region where the state is maximally entangled ( @xmath263 ) , reproducing the chsh scenario , to the one where it is only slightly entangled ( @xmath264 ) .
this should provide one with an intuition of how close quantitatively to the maximal violation of @xmath265 the observed bound @xmath66 should be in order to get close to one perfect local bit of randomness ( @xmath266 ) for a state with a given angle @xmath58 . | unpredictability , or randomness , of the outcomes of measurements made on an entangled state can be _ certified _ provided that the statistics violate a bell inequality . in the standard bell scenario where each party performs a single measurement on its share of the system
, only a finite amount of at most @xmath0 bits of randomness can be certified from a pair of qudits . in our work
, it is shown that this fundamental limitation can be overcome using sequences of ( non - projective ) measurements on the same system .
more precisely , we prove that one can certify _ any _ amount of random bits from a pair of qubits in a pure state as resource , even if it is arbitrarily weakly entangled .
in addition , this certification is achieved by near - maximal violation of a particular bell inequality for each measurement in the sequence . _
introduction_.bell s theorem @xcite has shown that the predictions of quantum mechanics demonstrate non - locality .
that is , they can not be described by a theory in which there are objective properties of a system prior to measurement that satisfy the no - signalling principle ( sometimes referred to as local realism " ) .
if one requires the no - signalling principle to be satisfied at the operational level then it can be shown that the outcomes of measurements demonstrating non - locality must be unpredictable @xcite .
this unpredictability , or randomness , is not the result of ignorance about the system preparation but is _ intrinsic _ to the theory .
although this connection between quantum non - locality ( via bell s theorem ) and the existence of intrinsic randomness is well know @xcite it was analysed in a quantitative way only recently @xcite .
it was shown how to use non - locality ( probability distributions that violate a bell inequality ) to _ certify _ the unpredictability of the outcomes of certain physical processes .
this was termed _ device independent randomness certification _
, as the certification only relies on the statistical properties of the outcomes and not on how they where produced .
the development of information protocols exploiting this certified form of randomness , such as device - independent randomness expansion @xcite and amplification protocols @xcite , then followed .
entanglement is a necessary resource for quantum non - locality , that in turn is required for randomness certification .
it is thus crucial to understand qualitatively and quantitatively how these three fundamental quantities relate one to another . in our work ,
we focus on asking how much certifiable randomness can be obtained from a single entangled state as a resource .
progress has been made in this direction for entangled states shared between two parties , alice ( @xmath1 ) and bob ( @xmath2 ) , in the standard scenario where each party makes a single measurement on his share of the system and then discards it .
an argument adapted from ref .
@xcite shows that either of the two parties , alice or bob can certify at most @xmath3 bits of randomness @xcite , where @xmath4 is the dimension of the local hilbert space the state lives in , which in turn implies a bound of @xmath5 bits when the two outputs are combined .
this demonstrates a fundamental limitation for device - independent randomness certification in this standard scenario .
the main goal of our work is to show that this limitation on the amount of certifiable random bits from one quantum state can be lifted . to do this
we will consider the sequential scenario , where sequences of measurements can be applied to each local system .
our main result is to prove that an unbounded amount of random bits can be certified in this scenario . to gain intuition , consider the following set - up where , contrary to the device - independent approach followed here , the functioning of a device can be entirely trusted .
the device consists of a quantum state prepared in the pauli-@xmath6 , or @xmath7 eigenstate @xmath8 followed by a measurement in the pauli-@xmath9 , or @xmath10
basis @xmath11 .
the outcome of this measurement is random and if the device then makes another measurement on the state , this time in the pauli-@xmath6 basis , it gives yet another random outcome . in this fashion of alternating between the two orthogonal bases , one can potentially obtain an unbounded number of random bits from one qubit .
the limitation of this procedure for producing random numbers is that one can not distinguish this device from a classical one with pre - programmed outcomes a _ local model _ for the outcomes if one does not trust the functioning of the device completely .
clearly we can not _ certify _ any randomness from a single system ( in a device - independent manner ) as in the above example , since one needs non - locality for this purpose .
but is it possible to build a scheme , that exploits non - locality and makes use of this idea of measuring the state repeatedly , to overcome the bound on the amount of certifiable randomness that one can obtain from a single entangled quantum system ?
to do so , the main obstacle comes from the fact that the local measurements needed to generate the random outcomes destroy the entanglement present in the state ( and non - locality in the correlations ) .
thus , one of the challenges is to come up with non - destructive measurements that still produce non - locality but retain some entanglement in the post - measurement state . in this way
, the state can still be used as a resource for subsequent measurements .
bell tests with sequences of measurements have received less attention than the standard ones with single measurement in the literature despite the novel features in this scenario @xcite . in our work we show that they prove useful in the task of randomness certification , which also provides another example @xcite where general measurements can overcome limitations of projective ones .
more precisely , we describe a scheme where any number @xmath12 of random bits can be certified using a sequence of @xmath13 consecutive measurements on the same system .
this work thus shows that the bound of @xmath14 random bits in the standard scenario can be overcome in the sequential scenario .
moreover , this unbounded randomness is certified by a near - maximal violation of a particular bell inequality for each measurement in the sequence .
_ the sequential measurements scenario._before presenting our results , let us introduce the scenario we work in .
we carry over many of the features from the standard scenario except now we allow party bob to make multiple dichotomic measurements in a sequence on his share of the state .
one can visualise this as in fig .
[ fig : scen ] where bob is split up into several bobs , each one corresponding to a measurement made on the state and labelled by @xmath15 , @xmath16 , where @xmath17 is the total number of measurements made in the sequence .
each @xmath15 makes one measurement and the post - measurement state is sent to @xmath18 .
we organize the bobs such that @xmath15 is doing his measurement _ before _
@xmath19 for @xmath20 .
thus in principle @xmath19 can receive the information about the inputs and outputs of previous measurements @xmath15 for all @xmath20 . to quantify the randomness produced in the setup
, we put the above scenario in the setting of _ non - local guessing games _
( e.g. refs .
@xcite ) . let us consider an additional adversary eve ( @xmath21 ) who is in possession of a quantum system potentially correlated to the one of @xmath1 and @xmath2 .
the global state is denoted @xmath22 .
we assume that at each round of the experiment @xmath21 is the one preparing the state @xmath22 and distributes @xmath23 to @xmath1 and @xmath2 .
this state will be used to make the measurements in the sequence and the aim of @xmath21 is to try to guess @xmath2 s outcomes by using measurements on her share of the state @xmath22 .
the parties @xmath1 and @xmath2 , having no knowledge about the state or the real measurements made on it , see their respective devices as black boxes that receive some classical input @xmath24 and @xmath25 , @xmath26 , respectively , and that generate a classical output @xmath27 and @xmath28 , @xmath29 , respectively ( see fig .
[ fig : scen ] ) .
they generate statistics from multiple runs of the experiment to obtain the observed probability distribution @xmath30 with elements @xmath31 .
this distribution @xmath30 lives inside the set of quantum correlations @xmath32 obtained from measurements on quantum states in a sequence as we described .
this set is convex and thus can be described in terms of its extreme points , denoted @xmath33 , and any @xmath30 can be written as @xmath34 , where @xmath35 and every @xmath36 . from studying the outcomes statistics _
only _ we can bound @xmath21 s predictive power by allowing her to have complete knowledge of how @xmath30 is decomposed into extreme points , i.e. she knows the probability distribution @xmath37 over extreme points @xmath33 .
this predictive power is quantified via the _ device - independent guessing probability _ ( digp ) @xcite where we fix the particular input string @xmath38 for which @xmath21 has to guess the outputs @xmath39 .
the digp , denoted by @xmath40 , is then calculated as the optimal solution to the following optimization problem @xcite : @xmath41 the operational meaning of this quantity is clear : eve has a complete description of the observed correlations in terms of extreme points .
she then guesses the most probable outcome for each extreme point .
one can recover the formalism of the standard scenario by simply considering that @xmath42 and @xmath43 . to quantify the amount of bits of randomness that is certified
, we use the _ min entropy _ @xmath44 which returns @xmath12 bits of randomness if @xmath45 . the amount of bits of randomness quantified in this way is the figure of merit in this work and our goal is to obtain as many bits as possible from a single system . finally , let us note that the problem in can be further relaxed to an optimization where instead of insisting on @xmath34 , we only impose the constraint that the observed statistics @xmath30 give a particular bell inequality violation @xcite .
the optimal solution to this new problem is an upper bound to the optimal solution of .
crucially , this relaxation still gives good bounds as we show in the following discussion .
_ the ingredients._alice and bob share the pure two - qubit state @xmath46 that for all @xmath470,\pi/2[$ ] is entangled . any pure ( entangled ) two qubit state can be written in this form up to some local change of basis . in ref .
@xcite , a family of bell inequalities was introduced : @xmath48 where @xmath49^{1/2}$ ] , @xmath50 and @xmath51 for @xmath52 , @xmath53 .
the maximal quantum value of the inequality @xmath54 is obtained by measuring the state with : @xmath55 where @xmath56 .
this family of inequalities has the following two useful properties : ( i ) it is maximally violated by the state @xmath57 with angle @xmath58 and , ( ii ) when maximally violated , this inequality certifies one bit of local randomness on bob s side for his second measurement choice @xmath59 : @xmath60 @xcite .
these observations are possible because the maximal violation is _ uniquely _ achieved by the probability distribution @xmath61 that arises from the previously - described state and measurements and .
therefore , for the maximal violation , @xmath62 in and the guessing probability for input choice @xmath59 is equal to @xmath63 .
however , we may not in general get correlations that maximally violate our bell inequality but give a violation that is only close to maximal . in appendices
[ app1:pggen ] , [ app2:continuityproof ] , and [ app6:contatpextunique ] we show how to make conclusions about the guessing probability for non - maximal violations . in particular , we show that for _ any _ bell inequality with a unique point of maximal violation , the guessing probability is a continuous function of the value of the inequality close to the maximal violation .
this implies in the particular case we are studying that : @xmath64 in appendix [ app5:numbound ] , we also provide a numerical upper bound on the guessing probability @xmath65 by a concave function of the value of @xmath66 .
so far we have reviewed and rigorously expanded on the results in ref .
@xcite showing how much randomness can be obtained from a pure entangled two - qubit state for dichotomic measurements . as mentioned , we can obtain more local randomness by considering measurements with more outcomes but it is ultimately limited to two bits @xcite . to go beyond this restrictive limit we work in a more general bell scenario that allows for sequences of measurements ( see fig.[fig : scen ] ) made on the state . if @xmath67 performs a projective measurement on @xmath68 the post - measurement state now shared between alice and @xmath69 is separable and thus useless for randomness production .
consequently , one needs to consider non - projective measurements to retain some entanglement in the system for the subsequent measurements . for this purpose ,
let us introduce the following two - outcome quantum measurement ( written in the formalism of kraus operators ) : @xmath70 corresponding to the two outcomes @xmath71 .
this measurement @xmath72 can be understood as a noisy version of the projective measurement @xmath73 .
it varies from being projective ( for @xmath74 ) to being non - interacting ( for @xmath75 ) .
one can verify that measuring an entangled state for @xmath760,\pi/4]$ ] ( non - projective measurement ) the post - measurement state still retains some entanglement , irrespectively of the outcome .
therefore , by tuning the parameter @xmath77 we are able to vary the destruction of the entanglement of the state at the gain of extracting information from it ( cf .
ref .
@xcite ) . intuitively , the closer to being a projective measurement , the lower the entanglement in the post - measurement state , but the bigger the violation of the initial bell inequality . _
_ a scheme for unbounded randomness certification.__we now combine the previous observations to demonstrate our main result .
first , let us recall that , as shown in @xcite , one can obtain one bit of randomness from any pure entangled two qubit state , irrespectively of the amount of entanglement in it .
moreover , one can verify that approximately one random bit can be certified if the measurements are close to the ones in ( in the sense that @xmath78 for @xmath79 ) since @xmath66 is then close to @xmath80 .
second , the measurement in eq .
is only close - to - projective for @xmath77 close to zero and leaves entanglement in the post - measurement state between alice and bob which is thus still useful for randomness certification . by repeated use of these two properties
we can certify the production an unbounded amount of random bits from a single pair of entangled qubits .
we now formally describe this process in which alice makes a single measurement on her share of the state , whereas bob makes a sequence of @xmath17 measurements on his .
each @xmath15 chooses between measurements of @xmath81 and @xmath82 for inputs @xmath83 and @xmath84 respectively , with outcomes @xmath29 .
the parameter @xmath85 is fixed before the beginning of the experiment .
the initial entangled state shared between alice and bob , before @xmath67 s measurement , is @xmath86 ( see with @xmath87 ) .
if the first non - projective measurement of the operator @xmath88 is made by @xmath67 on the initial state @xmath86 , the post - measurement state is of the form @xmath89 where @xmath90 and @xmath91 and the two unitaries , @xmath92 and @xmath93 , and angle @xmath940,\pi/4]$ ] depend on the first outcome @xmath95 and the angles @xmath96 and @xmath97 .
after his measurement , @xmath67 applies the unitary @xmath98 , conditioned on his outcome @xmath99 , on the post - measurement state going to @xmath69 .
this allows @xmath100 to use the same two measurements @xmath101 and @xmath81 independently of the outcome @xmath99 since the unitary @xmath102 is cancelled in .
this last procedure will be applied by each @xmath15 after his measurement , before sending the post - measurement state to the next @xmath18 . if the system passed through _ only _ the non - projective measurements , the state received by @xmath15 can be one of @xmath103 potential states , depending on all of the previous @xmath19 s ( @xmath104 ) outcomes ( one for each combination @xmath105 of outcomes obtained by the previous @xmath19 , these can be computed _ before _ the beginning of the experiment ) .
any of these states can be written as : @xmath106,\ ] ] where the angles @xmath107 and the matrix @xmath108 both depend on the outcomes @xmath109 , on the initial angle @xmath96 and the angles @xmath110 of the previous @xmath19 s with @xmath104 . in the notation
, we will always omit the dependence on the angles @xmath96 and @xmath111 since these are fixed _ before _ the beginning of the experiment .
for each of these different potential states with angle @xmath107 , alice adds two measurements to her input choices , where for @xmath112 , these are measurements of the observables @xmath113 which are defined as @xmath114(u_a^{\vec{b}_{i-1}})^{\dagger},\ ] ] where @xmath115 , depending on the specific state @xmath116 .
we are now ready to describe how the scheme certifies randomness .
the measurement operator @xmath82 can be made arbitrarily close to @xmath73 by choosing @xmath117 sufficiently small .
this brings the outcome statistics for measurements @xmath118 on bob s side and @xmath119 on alice s side on the state in eq .
, arbitrarily close to the statistics for the measurements in eq . and a state of the form in eq .
, for @xmath120 . therefore , the inequality @xmath121 for alice and @xmath15 as defined in can be made arbitrarily close to its maximal violation .
this in turn guarantees that the guessing probability , @xmath122 can be made arbitrarily close to @xmath63 .
note that this guessing probability does not only describe the instances when alice chooses the measurements @xmath123 .
since eve does not know alice s measurement choices in advance she can not use a strategy that gives higher predictive power for the instances when alice chooses other measurements .
finally , by making @xmath122 sufficiently close to @xmath63 for each @xmath124 ( by choosing each @xmath85 sufficiently close to @xmath125 ) the digp @xmath126 can be made arbitrarily close to @xmath127 ( see appendix [ app4:arbitrarysequ ] for a proof ) . at the end
, bob can produce @xmath12 random bits by a suitably chosen sequence @xmath82 , @xmath16 , of @xmath128 measurements .
the certification only requires that each @xmath15 occasionally chooses the projective measurement @xmath81 so that the whole statistics can be obtained .
note that bob can choose @xmath81 with probability @xmath129 and @xmath82 with probability @xmath130 for @xmath129 as close to zero as he wants .
to summarise the idea of the scheme , the post - measurement state after @xmath131 , given a sequence of non - projective measurements , is of the form of . with some probability alice
chooses measurements @xmath132 which give outcome statistics that allow randomness certification of the @xmath124th bit using the @xmath121 inequality .
therefore , we can certify randomness for each measurement @xmath15 in the sequence at the expense of increasing the number of measurements that alice chooses from .
finally , also remark that the value of _ each _ inequality @xmath121 between each @xmath15 and a can be made as close as wanted to the maximal value @xmath133 . _
_ conclusion.__we have presented a scheme for certifying an unbounded amount of random bits from a single pair of entangled qubits where one of the qubits is subjected to a sequence of measurements .
the measurements do not completely destroy the entanglement but map the state to another pure entangled two - qubit state ( with reduced entanglement ) .
our main result made use of the fact that every measurement in bob s sequence generated an almost - maximally non - local output distribution ( in the sense of violating some bell inequality almost maximally ) .
a similar study was made in ref .
@xcite , where a sequence of non - local correlations was obtained from a single pair of qubits . nevertheless , it was still an open question to know whether it was possible to get a sequence of ( arbitrarily close to ) maximally non - local correlations , and our results show that this is possible , a property that could be of further use for many other device - independent quantum information tasks .
our results proves that in the sequential scenario one can certify an unbounded amount of randomness .
however , our scheme requires an exponentially increasing number of measurements on alice s side ( @xmath134 measurement choices for @xmath17 measurements in the sequence ) and is not robust to noise .
all these issues deserve further investigation , for instance for the construction of device - independent randomness _ expansion _ protocols in the sequential scenario .
finally , on a more fundamental level , our results offer new insights in the relation between entanglement , non - locality and randomness .
we showed that a single pair of pure entangled qubits is a potentially unbounded source of certifiable random bits ( and of non - local correlations ) if we perform sequences of measurements on it . _
_ acknowledgements.__this work is supported by the erc cog qitbox and adg osyris , the eu ip siqs , the john templeton foundation , the generalitat de catalunya ( sgr 875 ) and the spanish project foqus ( fis2013 - 46768-p ) .
m.j . also acknowledges support from the marie curie cofund action through the icfonest program .
mjh acknowledges support from the epsrc ( through the nqit quantum hub ) and the fqxi large grant _ thermodynamic vs information theoretic entropies in probabilistic theories_. 99 j. s. bell , physics ( ny ) * 1 * ( 3 ) , 195 - 200 ( 1964 ) .
s. popescu and d. rohrlich , found .
phys .
* 24 * , 379 ( 1994 ) .
ll .
masanes , a. acn , and n. gisin , phys .
rev .
a. * 73 * , 012112 ( 2006 ) .
n. brunner , d. cavalcanti , s. pironio , v. scarani , and s. wehner , rev .
mod .
phys . * 86 * , 419 ( 2014 ) .
s. pironio _ et al .
_ , nature ( london ) * 464 * , 1021 ( 2010 ) .
r. colbeck , phd thesis university of cambridge , arxiv:0911.3814 ( 2006 ) .
u. vazirani , and t. vidick , phil .
trans .
r. soc .
a * 370 * , 3432 ( 2012 ) .
r. colbeck , and r. renner , nature physics * 8 * , 450 - 453 ( 2012 ) .
r. gallego , l. masanes , g. de la torre , c. dhara , l. aolita , and a. acn , nature communications * 4 * , 2654 ( 2013 ) .
g. m. dariano , p. lo presti and p. perinotti , j. phys .
a : math . gen . * 38 * , 5979 ( 2005 ) .
a. acn , s. pironio , t. vrtesi and p. wittek , arxiv:1505.03837 ( 2015 ) .
a. acn , s. massar and s. pironio , phys .
rev .
lett . * 108 * , 100402 ( 2012 ) .
g. de la torre , m. j. hoban , c. dhara , g. prettico , a. acn , phys .
rev .
lett . * 114 * , 160502 ( 2015 ) . o. nieto - silleras , s. pironio and j. silman , new j. phys . * 16 * , 013035 ( 2014 ) .
r. gallego , l. e. wrflinger , r. chaves , a. acn , and m. navascus , new j. phys . * 16 * 033037 ( 2014 ) .
s. popescu , phys .
rev .
lett . * 74 * , 2619 ( 1995 ) .
r. silva , n. gisin , y. guryanova and s. popescu , phys .
rev .
lett . * 114 * , 250401 ( 2015 ) .
r. t. rockafellar , _ convex analysis _ , princeton press , ( 1970 ) ,
theorem 10.1 .
o. bucicovschi and j. lebl , j. convex anal .
* 20 * , 1113 ( 2013 ) . |
differential cryptanalysis is the first statistical attack proposed for breaking iterated block ciphers .
its publication @xcite then gave rise to numerous works which investigate the security offered by different types of functions regarding differential attacks .
this security is quantified by the so - called _ differential uniformity _ of the substitution box used in the cipher @xcite .
most notably , finding appropriate s - boxes which guarantee that the cipher using them resist differential attacks is a major topic for the last twenty years , see e.g. @xcite .
power functions , _
i.e. _ , monomial functions , form a class of suitable candidates since they usually have a lower implementation cost in hardware . also , their particular algebraic structure makes the determination of their differential properties easier . however , there are only a few power functions for which we can prove that they have a low differential uniformity .
up to equivalence , there are two large families of such functions : a subclass of the quadratic power functions ( a.k.a .
gold functions ) and a subclass of the so - called kasami functions . both of these families contain some permutations which are apn over @xmath1 for odd @xmath10 and differentially @xmath11-uniform for even @xmath10 . the other known power functions with a low differential uniformity correspond to `` sporadic '' cases in the sense that the corresponding exponents vary with @xmath10 @xcite and they do not belong to a large class : they correspond to the exponents defined by welch @xcite , by niho @xcite , by dobbertin @xcite , by bracken and leander @xcite , and to the inverse function @xcite . it is worth noticing that some of these functions seem to have different structures because they do not share the same differential spectrum . for instance , for a quadratic power function or a kasami function , the differential spectrum has only two values , _
i.e. _ , the number of occurrences of each differential belongs to @xmath12 for some @xmath13 @xcite .
the inverse function has a very different behavior since its differential spectrum has three values , namely @xmath14 , @xmath15 and @xmath11 and , for each input difference , there is exactly one differential which is satisfied four times .
however , when classifying all functions with a low differential uniformity , it can be noticed that the family of all power functions @xmath8 over @xmath1 , with @xmath16 , contains several functions with a low differential uniformity .
most notably , it includes the cube function and the inverse function , and also @xmath17 for @xmath10 odd , which is the inverse of a quadratic function . at a first glance
, this family of exponents may be of very small relevance because the involved functions have distinct differential spectra .
then , they are expected to have distinct structures .
for this reason , one of the motivations of our study was to determine whether some link could be established between the differential properties of the cube function and of the inverse function .
our work then answers positively to this question since it exhibits a general relationship between the differential spectra of @xmath8 and @xmath18 over @xmath1 .
we also determine the complete differential spectra of some other exponents in this family .
the rest of the paper is organized as follows .
section [ sec - boo ] recalls some definitions and some general properties of the differential spectrum of monomial functions .
section [ sec - ds ] then focuses on the differential spectra of the monomials @xmath8 .
first , the differential spectrum of any such function is shown to be determined by the number of roots of a family of linear polynomials .
then , we exhibit a symmetry property for the exponents in this family : it is proved that the differential spectra of @xmath8 and @xmath18 over @xmath1 are closely related . in section [ sec - spec ] , we determine the whole differential spectrum of @xmath7 over @xmath1 .
it is expressed by means of some kloosterman sums , and explicitly computed using the work of carlitz @xcite .
we then derive the differential spectra of @xmath19 .
further , we study the functions @xmath20 and @xmath21 .
we finally end up with some conclusions .
any function @xmath23 from @xmath1 into @xmath1 can be expressed as a univariate polynomial in @xmath24 $ ] .
the _ univariate degree _ of the polynomial @xmath23 is , as usual , the maximal integer value of its exponents .
the _ algebraic degree _ of @xmath23 is the maximal hamming weight of its exponents : @xmath25 where @xmath26 and the _ hamming weight _ is calculated as follows : @xmath27 in this paper , we will identify a polynomial of @xmath24 $ ] with the corresponding function over @xmath1 .
for instance , @xmath28 $ ] is called a _ permutation polynomial _ of @xmath1 if the function @xmath29 is a permutation of @xmath1 .
boolean functions are also involved in this paper and are generally of the form @xmath30 where @xmath31 is any function from @xmath1 into @xmath1 and where @xmath32 denotes the _ absolute trace _ on @xmath1 , _
@xmath33 in the whole paper , @xmath34 denotes the cardinality of any set @xmath35 . the resistance of a cipher to differential attacks and to its variants is quantified by some properties of the _ derivatives _ of its s(ubstitution)-box , in the sense of the following definition .
it is worth noticing that this definition is general : it deals with functions from @xmath36 into @xmath37 for any @xmath38 .
[ de1 ] let @xmath23 be a function from @xmath1 into @xmath37 .
for any @xmath39 , the _ derivative of @xmath23 with respect to @xmath40 _ is the function @xmath41 from @xmath1 into @xmath37 defined by @xmath42 the resistance to differential cryptanalysis is related to the following quantities , introduced by nyberg and knudsen @xcite .
[ de2 ] let @xmath23 be a function from @xmath1 into @xmath1 .
for any @xmath40 and @xmath4 in @xmath1 , we denote @xmath43 then , the _ differential uniformity _ of @xmath23 is @xmath44 those functions for which @xmath45 are said to be _ almost perfect nonlinear ( apn)_. in this paper , we focus on the case where the s - box is a power function , _
i.e. , _ a monomial function on @xmath1 . in other words , @xmath46 over @xmath1 , which will be denoted by @xmath47 when necessary . in the case of such a power function
, the differential properties can be analyzed more easily since , for any nonzero @xmath39 , the equation @xmath48 can be written @xmath49 implying that @xmath50 then , when @xmath51 is a monomial function , the differential characteristics of @xmath23 are determined by the values @xmath52 , @xmath53 . from now on , this quantity @xmath54 is denoted by @xmath55 . since @xmath56 the _ differential spectrum _ of @xmath23 can be defined as follows .
[ de : specdiff ] let @xmath46 be a power function on @xmath1 .
we denote by @xmath57 the number of output differences @xmath4 that occur @xmath58 times : @xmath59 the _ differential spectrum _ of @xmath47 is the set of @xmath60 : @xmath61 . with same notation , we have the following equalities .
they are well - known but we indicate the proof for clarity .
[ ident ] @xmath62 where @xmath63 for @xmath58 odd .
the first equality is obviously deduced from ( [ deomega ] ) . and , for @xmath64
, @xmath65 equals the number of @xmath66 such that @xmath67 for some @xmath4 .
thus , any @xmath68 is counted in the second sum .
the differential spectrum of the power function @xmath46 over @xmath1 is also related to the weight enumerator of the cyclic code of length @xmath69 with defining set @xmath70 @xcite .
in particular , the number of codewords with hamming weight @xmath71 and @xmath11 in this cyclic code can be derived from the differential spectrum of @xmath23 ( see e.g. corollary 1 in @xcite ) .
a power function @xmath23 is said to be _ differentially @xmath15-valued _ if and only if for any @xmath72 , we have @xmath73 ( and then only two @xmath57 in @xmath74 do not vanish ) .
it is known that @xmath75 for some @xmath76 ( see an extensive study in ( * ? ? ? *
section 5 ) ) .
note that apn functions are differentially @xmath15-valued with @xmath77 .
there are some basic transformations which preserve @xmath74 .
let @xmath78 and @xmath79 over @xmath1 .
if there exists @xmath80 such that @xmath81 or if @xmath82 , then @xmath83 has the same differential spectrum as @xmath47 . in this section , @xmath78 and notation
is as in [ sec - difpo ]
. studying @xmath84 for special values of @xmath4 may give us as least a lower bound on @xmath85 .
so we first focus on @xmath86 .
[ delt0 ] let @xmath87 be such that @xmath88 .
then @xmath89 is such that @xmath90 .
in particular @xmath91 if and only if @xmath92 .
note that @xmath91 if and only if @xmath47 is a permutation .
obviously , @xmath68 is a solution of @xmath93 if and only if @xmath94 since @xmath95 is a permutation over @xmath96 .
as there are exactly @xmath97 such nonzero @xmath98 , the proof is completed .
there is an immediate consequence of lemma [ delt0 ] for specific values of @xmath87 .
[ delprime ] let @xmath99 such that @xmath87 divides @xmath100 .
then @xmath101 . in particular ,
if @xmath102 with @xmath103 then @xmath104 .
since @xmath105 , @xmath106 from lemma [ delt0 ] .
but the polynomial @xmath107 has degree @xmath108 for any @xmath4 , so that @xmath109 .
we conclude that @xmath110 .
now , let @xmath102 with @xmath103 .
then @xmath111 so that @xmath112 . as previously we conclude that @xmath113 .
[ ex1 ] if @xmath114 then @xmath115 for any even @xmath10 .
+ if @xmath116 then @xmath117 for @xmath118 for all @xmath119 .
+ if @xmath120 then @xmath121 for @xmath122 for all @xmath119 .
the previous remarks combined with our simulation results point out that @xmath86 and @xmath123 play a very particular role in the differential spectra of power functions .
this leads us to investigate the properties of the differential spectrum restricted to the values @xmath55 with @xmath124 .
[ loc - apn ] let @xmath23 be a power function on @xmath1 .
we say that @xmath23 has the _ same restricted differential spectrum as an apn function _
when @xmath125 for the sake of simplicity , we will say that @xmath23 is _ locally - apn_. this definition obviously generalizes the apn property .
for instance , the _ inverse _ function over @xmath1 is locally - apn for any @xmath10 , while it is apn for odd @xmath10 only .
another infinite class of locally - apn functions is exhibited in section [ sec - n/2 ] .
from now on , we investigate the differential spectra of the following specific monomial functions @xmath127 note that such a function has algebraic degree @xmath128 .
in this section , we first give some general properties .
[ th : 2^t-1 ] let @xmath129 over @xmath1 with @xmath130 .
then , @xmath131 consequently , for any @xmath132 , @xmath55 is the number of roots in @xmath133 of the linear polynomial @xmath134 and we have @xmath135 for some @xmath136 with @xmath137 .
to prove ( [ eq : u1 ] ) we simply compute @xmath138 thus , @xmath123 is directly deduced and it corresponds to the number of roots of @xmath139 .
let @xmath132 .
then @xmath140 is a solution of @xmath141 if and only if it is a solution of @xmath142 or equivalently if it is a root of the linear polynomial @xmath143 the values @xmath144 and @xmath145 are counted in @xmath123 ( as solutions of @xmath146 ) , while @xmath147 for any @xmath4 .
so , we get that , if @xmath148 , the number of roots of @xmath149 in @xmath1 is equal to @xmath150 . because the set of all roots of a linear polynomial is a linear space , we deduce that @xmath151 moreover , by raising @xmath149 to the @xmath152-th power , we get that any root of @xmath149 is also a root of @xmath153 with @xmath154 .
this then implies that @xmath155 with @xmath156 . finally , for @xmath157 , @xmath158 , implying that @xmath159 , which naturally corresponds to lemma [ delt0 ] .
[ reminv ] as a first easy corollary , we recover the following well - known form of the differential spectrum of the inverse function , @xmath160 over @xmath1 .
actually , the previous theorem applied to @xmath161 leads to @xmath92 and @xmath162 when @xmath10 is odd and @xmath163 when @xmath10 is even . for all @xmath124 , @xmath164
therefore , we have * if @xmath10 is odd , @xmath165 and @xmath166 , @xmath167 ; * if @xmath10 is even , @xmath168 and @xmath169 , @xmath170 .
clearly @xmath171 is locally - apn for any @xmath10 , as we previously noticed ( see definition [ loc - apn ] ) .
the following corollary is a direct consequence of theorem [ th : 2^t-1 ] .
[ coro : delta ] let @xmath129 over @xmath1 with @xmath130 .
then , its differential uniformity is of the form either @xmath172 or @xmath173 for some @xmath174 .
moreover , if @xmath175 for some @xmath176 , then this value appears only once in the differential spectrum , _ i.e. _ , @xmath177 , and it corresponds to the value of @xmath123 , implying @xmath178 . in theorem
[ th : 2^t-1 ] , we exhibited some tools for the computation of the differential spectra of functions @xmath0 .
the problem boils down _ to the determination of the roots of a linear polynomial _ whose coefficients depend on @xmath72 .
there are equivalent formulations that we are going to develop now .
the first one is obtained by introducing another class of linear polynomials over @xmath1 . for any subspace @xmath35 of @xmath1 ( where @xmath1 is identified with @xmath179 ) , we define its _ dual _ as follows : @xmath180 also , we denote by @xmath181 the image set of any function @xmath23 . [ le : sym ] let @xmath182 and @xmath183 .
let us consider the linear applications @xmath184 then the dual of @xmath185 is the set of all @xmath186 satisfying @xmath187 where @xmath188 note that @xmath189 is called the _ adjoint _
application of @xmath190 . by definition ,
@xmath191 consists of all @xmath186 such that @xmath192 for all @xmath193 .
we have @xmath194 hence @xmath186 belongs to the dual of the image of @xmath190 if and only if @xmath195 , _
i.e. _ , @xmath186 is a root of @xmath189 , completing the proof .
the following theorem gives an equivalent formulation of the quantity @xmath136 which is presented in theorem [ th : 2^t-1 ] .
[ thadj ] notation is as in lemma [ le : sym ] .
then @xmath196 consequently , this dimension can be determined by solving @xmath197 or equivalently by solving @xmath198 let @xmath199 be the dimension of the image set of @xmath190 .
it is well - known that @xmath200 . on the other hand
, lemma [ le : sym ] shows that @xmath186 is in the dual of the image of @xmath190 if and only if @xmath187 .
we deduce that @xmath201 completing the proof .
now , we discuss a different point of view , using an equivalent linear system .
[ thsys ] for any @xmath202 , we define the following equations : @xmath203 let @xmath204 be the number of solutions of @xmath205 in @xmath206 .
let @xmath207 be the number of solutions in @xmath208 of the system @xmath209 then @xmath210 .
we simply write @xmath211 which is equal to @xmath212 we are looking at the number of solutions of @xmath205 which are not in @xmath213 .
so , it is equivalent to compute the number of nonzero solutions @xmath214 of @xmath215 such that the equation @xmath216 has solutions .
this last condition holds if and only if @xmath217 , providing two distinct solutions @xmath218 such that @xmath219 , completing the proof .
[ val - b ] in theorem [ thsys ] , @xmath4 takes any value while @xmath149 is defined for @xmath220 in theorem [ th : 2^t-1 ] . for all @xmath220
, we have clearly @xmath221 .
if @xmath222 , @xmath223 and the number of roots of @xmath224 in @xmath1 is equal to @xmath225 therefore , we have @xmath226 .
recall that @xmath129 . now
, we are going to examine some symmetries between the differential spectra of @xmath227 and @xmath228 where @xmath182 and @xmath183 . in the list of properties below ,
notation is conserved as soon it is defined .
recall that @xmath229 is the adjoint polynomial of @xmath230 .
thus , both polynomials have a kernel with the same dimension ( see lemma [ le : sym ] and theorem [ thadj ] ) .
it is worth noticing that this dimension is at least @xmath231 since @xmath232 . in this section
we want to prove the following theorem .
[ thmain ] for any @xmath233 with @xmath234 , we define @xmath235 then , for any @xmath236 with @xmath237 and for any @xmath58 , we have @xmath238 .
we begin by some lemmas .
the next one will not be used for the proof of theorem [ thmain ] but clarifies some arguments . [
le : rac ] let @xmath239 and @xmath130 .
then there are exactly two elements , @xmath240 and @xmath241 with @xmath242 , such that @xmath243 for @xmath244 .
in particular , @xmath245 for @xmath246 .
let @xmath40 be fixed and let us consider the equation @xmath247 for some @xmath4 : @xmath248 there is @xmath4 such that this equation is satisfied if and only if @xmath249 which holds for any @xmath40 .
thus , for any nonzero @xmath40 there are exactly two solutions , say @xmath240 and @xmath241 whose sum equals @xmath250 . to complete the proof
, we observe that @xmath251 .
[ le : eqv ] let @xmath236 with @xmath237 .
let @xmath252 be the permutation of @xmath253 defined by @xmath254 then , for any @xmath255 in @xmath253 , @xmath256 satisfies @xmath257 first , we clearly have that @xmath252 is a permutation of @xmath253 .
indeed , @xmath258 and one can define the inverse of @xmath252 as follows : @xmath259 actually , @xmath260 and it can be checked that @xmath261 then , by using that @xmath262 and @xmath263 , we deduce that @xmath264 [ le : eq1 ] let @xmath236 with @xmath237 .
let @xmath72 and let @xmath239 such that @xmath247
. then @xmath265 , where @xmath266 .
recall that @xmath267 . we know that for any @xmath268 there is @xmath269 such that @xmath247
this is because @xmath270 ( see theorem [ thadj ] ) and @xmath271 is included in the kernel of @xmath190 .
moreover , @xmath272 if and only if @xmath273 .
we treat the case @xmath274 separately , a case where @xmath275 for @xmath276 only . in this case ,
lemma [ le : eqv ] leads to @xmath277 too where @xmath278 , since @xmath279 .
and we have for @xmath157 @xmath280 and for @xmath222 @xmath281 thus , we conclude : for @xmath274 , if @xmath4 is such that @xmath245 then @xmath278 and @xmath282 where the last equality comes from theorem [ thadj ] .
now , we suppose that @xmath283 . with @xmath284 , the equation
@xmath285 is equivalent to @xmath286 which is @xmath287 we can set @xmath288 since @xmath289 is equivalent to @xmath290 we have proved that @xmath285 is equivalent to @xmath291 then , @xmath292 .
but @xmath293 by theorem [ thadj ] , completing the proof .
_ proof of theorem [ thmain]_. recall that @xmath294 then , we want to show that , for any @xmath58 , @xmath295 . for any @xmath234 and for any @xmath296 , we define @xmath297 from theorem [ thadj ] , we know that @xmath298
. then , @xmath299 for any @xmath300 there are @xmath301 nonzero @xmath40 in @xmath302 and then @xmath301 pairs @xmath255 , for a fixed @xmath4 , in @xmath303 so that @xmath304 we use lemma [ le : eqv ] . recall that @xmath252 is the permutation of @xmath305 defined by @xmath306 then , we have @xmath307 indeed , any @xmath308 is as follows specified from @xmath255 .
we have @xmath309 from lemma [ le : eqv ] .
moreover , according to lemma [ le : eq1 ] , @xmath265 , where @xmath310 is calculated from @xmath40 and @xmath4 , for any @xmath40 such that @xmath247 .
in other terms , to any pair @xmath311 corresponds a unique pair @xmath312 .
we finally get that @xmath313 and it directly follows from ( [ eq0 ] ) that @xmath314 , completing the proof .
@xmath315 + now we are going to explain theorem [ thmain ] , in terms of the differential spectra of @xmath227 and @xmath228 , @xmath236 with @xmath237 . actually , we can deduce from the previous theorem that both functions @xmath227 and @xmath228 have the same _ restricted differential spectrum _ , _
i.e. _ the multisets @xmath316 are the same for both functions .
[ cormain ] we denote by @xmath317 , @xmath72 , the quantities @xmath55 corresponding to @xmath318 .
then , for any @xmath236 with @xmath237 , we have @xmath319 and we have equality between both following multisets : @xmath320 moreover , @xmath227 and @xmath228 have the same differential spectrum if and only if @xmath321 which can hold for odd @xmath10 only . in any case , @xmath227 is locally - apn if and only if @xmath228 is locally - apn . since @xmath183
, we clearly have @xmath322 thus , applying theorem [ th : 2^t-1 ] , we get @xmath323 and @xmath324 moreover , we have @xmath325 implying that @xmath326 we deduce from theorem [ thmain ] that @xmath327 equality ( [ msets ] ) is then a direct consequence of theorem [ th : 2^t-1 ] , since @xmath328 now , we note that @xmath329 if and only if @xmath330 .
thus , @xmath227 and @xmath228 have the same differential spectrum if and only if @xmath329 . since @xmath331 this holds if and only if @xmath332 .
it can not hold when @xmath10 is even , because in this case either @xmath333 or @xmath128 is even too . using definition [ loc - apn ] ,
the last statement is obviously derived .
the previous result implies that , if @xmath227 is apn over @xmath1 , then @xmath228 is locally - apn .
moreover , the differential spectrum of @xmath228 can be completely determined as shown by the following corollary .
[ cr : apn ] let @xmath10 and @xmath334 be two integers such that @xmath335 is apn over @xmath1 .
let @xmath336 .
then , * if @xmath10 is odd , both @xmath227 and @xmath228 are apn permutations ; * if @xmath10 is even , @xmath227 is not a permutation and @xmath228 is a differentially @xmath11-uniform permutation ( locally - apn ) with the following differential spectrum : @xmath337 , @xmath338 and @xmath169 . from theorem
[ th : 2^t-1 ] , we deduce that , if @xmath23 is apn , then @xmath339 for all @xmath340 ; moreover , @xmath341 and @xmath342 since @xmath343 and @xmath344 .
if @xmath10 is odd , @xmath345 is then the only possible value , implying that @xmath346 .
it follows that @xmath347 , @xmath348 and @xmath349 for all @xmath340 . in other words , both @xmath227 and @xmath228 are apn permutations . if @xmath10 is even , it is well - known that @xmath227 is not a permutation ( see e.g. @xcite ) .
more precisely , we have here @xmath350 since @xmath128 and @xmath351 can not be both coprime with @xmath10 .
then , we deduce that @xmath347 and @xmath352 .
the differential spectrum of @xmath228 directly follows from corollary [ cormain ] .
notation is as in corollary [ cr : apn ] . for @xmath353
, we have @xmath354 .
it is well - known that @xmath355 is an apn function over @xmath1 for any @xmath10 . since @xmath356 , @xmath357 is equivalent to the inverse function and it is also well - known that the inverse function is apn for odd @xmath10 . for even @xmath10 , @xmath168 and
the differential spectrum is computed in remark [ reminv ] .
[ coro3 ] let @xmath10 and @xmath334 be two integers such that @xmath335 is differentially @xmath11-uniform .
then , @xmath10 is even and @xmath227 is a permutation with the following differential spectrum : @xmath337 , @xmath338 and @xmath169 .
moreover , for @xmath336 , @xmath228 is apn . from corollary [ coro : delta ]
, we deduce that @xmath358 implies @xmath359 and @xmath337 .
in particular , @xmath10 is even .
since @xmath360 and @xmath361 can not be both equal to @xmath15 , we also deduce that that @xmath227 is a permutation .
its differential spectrum is then derived from lemma [ ident ]
. moreover , we have @xmath362 and @xmath348 , implying that @xmath228 is apn .
in this section , we apply the results of section [ sec - ds ] to the study of the differential spectrum of @xmath335 , for special values of @xmath128 .
we first focus on @xmath364 over @xmath1 , _
i.e. _ , @xmath365 . in this case
, we determine the complete differential spectrum of the function .
moreover , thanks to the work of carlitz @xcite , we emphasize that this spectrum is related to some kloosterman sums defined as follows .
[ prop : kloos ] ( * ? ? ?
* formula ( 6.8 ) ) let @xmath366 be the kloosterman sum @xmath367 extended to @xmath14 assuming that @xmath368 for @xmath144 .
then , @xmath369 [ th : x^7 ] let @xmath364 over @xmath1 with @xmath370 .
then , its differential spectrum is given by : * if @xmath10 is odd , @xmath371 * if @xmath10 is even , @xmath372 where @xmath366 is the kloosterman sum defined as in proposition [ prop : kloos ] .
in particular , @xmath373 is differentially @xmath374-uniform for all @xmath375 . to prove this theorem , we need some preliminary results .
we first recall some basic results on cubic equations .
[ cubic ] @xcite the cubic equation @xmath376 , where @xmath39 and @xmath377 has a unique solution in @xmath1 if and only if @xmath378 .
in particular , if it has three distinct roots in @xmath1 , then @xmath379 .
[ cubeq ] ( * ? ? ? * appendix ) let @xmath380 and @xmath381 then , we have for odd @xmath10 @xmath382 and for even @xmath10 @xmath383 now we are going to solve the equations @xmath384 ( see theorem [ th : 2^t-1 ] ) by solving a system of equations , including a cubic equation , thanks to the equivalence presented in theorem [ thsys ] .
[ th : cubeq2 ] let @xmath385 the number @xmath386 of @xmath387 such that @xmath149 has no roots in @xmath96 is given by @xmath388 where @xmath366 is the kloosterman sum defined as in proposition [ prop : kloos ] .
let @xmath132 . according to theorem [ thsys ] we know that the number ( denoted by @xmath204 ) of roots in @xmath133 of @xmath149 is twice the number of roots in @xmath208 of the following system where @xmath278 : @xmath389 since @xmath390 , @xmath391 for @xmath392 .
then , for any @xmath393 , the following situations may occur : * @xmath394 has no root in @xmath1 . in this case ,
* @xmath394 has a unique root @xmath396 . from lemma [ cubic ] , this occurs if and only if @xmath397 . in this case , @xmath398 if @xmath399 and @xmath400 if @xmath217
. * @xmath401 has three roots @xmath402 . since these roots are roots of a linear polynomial of degree @xmath11 then @xmath403 , implying @xmath404 .
then , at least one @xmath405 is such that @xmath406 .
it follows that , in this case , @xmath204 is either @xmath374 or @xmath15 .
let us now define @xmath407 from the previous discussion , we have @xmath408 where the last equality comes from proposition [ cubeq ] .
let us now compute the value of @xmath409 .
@xmath410 by using that @xmath411 .
but , we have @xmath412 implying that @xmath413 therefore , @xmath414 now , we clearly have that @xmath415 if and only if @xmath416 . moreover
, two distinct elements @xmath417 and @xmath418 in @xmath133 with @xmath419 and @xmath420 satisfy @xmath421 ( otherwise , @xmath401 with @xmath422 has at least @xmath15 roots in @xmath1 ) . therefore , we deduce that @xmath423 if @xmath10 is odd , we deduce that @xmath424 if @xmath10 is even , we deduce that @xmath425 on the other hand , by definition of the kloosterman sum @xmath366 , we have @xmath426 thus , @xmath427 we then deduce that , for any @xmath10 , @xmath428 it follows that @xmath429 ( proof of theorem [ th : x^7 ] ) in accordance with lemma [ ident ] , we obtain the differential spectrum of @xmath373 as soon as we are able to solve the following system : @xmath430 now , we apply theorem [ th : 2^t-1 ] and we recall first that @xmath431 for any @xmath432 .
moreover , we know that @xmath433 as defined in theorem [ th : cubeq2 ] . since @xmath365
, @xmath434 equals @xmath231 for odd @xmath10 and @xmath15 otherwise .
then , if @xmath10 is even then @xmath163 else @xmath162
. thus , @xmath337 for even @xmath10 and @xmath435 otherwise . from the second equation of ( [ syseq7 ] )
, we get @xmath436 and using the first equation of ( [ syseq7 ] ) @xmath437 leading to @xmath438 finally , we deduce from theorem [ th : cubeq2 ] that , for odd @xmath10 , @xmath439 and for even @xmath10 @xmath440 finally , it can be proved that @xmath441 for any @xmath375 , implying that @xmath373 is differentially @xmath374-uniform .
actually , it has been proved in ( * ? ? ?
3.4 ) that @xmath442 implying that @xmath443 when @xmath444 .
it is worth noticing that @xmath373 is apn when @xmath445 since its inverse is the quadratic apn permutation @xmath446 .
when @xmath447 , @xmath373 is locally - apn , and not apn , since it corresponds to the inverse function over @xmath448 . by combining the previous theorem and corollary [ cormain ] , we deduce the differential spectrum of @xmath449 over @xmath1 .
[ coro5 ] let @xmath449 over @xmath1 with @xmath375 .
then , we have : * if @xmath450 , @xmath451 is differentially @xmath374-uniform and for any @xmath53 , @xmath452
. moreover , its differential spectrum is given by : @xmath453 * if @xmath71 divides @xmath10 , @xmath451 is differentially @xmath454-uniform and for any @xmath53 , @xmath455 .
moreover , its differential spectrum is given by : @xmath456 let @xmath457 denote the differential spectrum of @xmath373 over @xmath1 .
we apply corollary [ cormain ] ( with @xmath458 ) .
then , if @xmath459 , @xmath460 and @xmath461 . otherwise , @xmath462 and @xmath463 .
moreover , in both cases , @xmath464 for @xmath10 even and @xmath465 for @xmath10 odd .
it follows that , * for @xmath459 , @xmath10 odd , we have @xmath466 and + @xmath467 . then , @xmath468 for all @xmath58 ; * for @xmath459 , @xmath10 even , we have @xmath469 and + @xmath470 . then , @xmath471 , @xmath472 and @xmath473 . * for @xmath474 , @xmath10 odd , we have @xmath475 and + @xmath476 . then , @xmath477 , @xmath478 , @xmath479 and @xmath480 . * for @xmath474 , @xmath10 even , we have @xmath481 and + @xmath482
. then , @xmath477 , @xmath478 , @xmath472 , @xmath483 and @xmath484 .
the result finally follows from theorem [ th : x^7 ] .
the minimum distance of the cyclic code of length @xmath100 with defining set @xmath485 has been studied by van lint and wilson in @xcite .
more precisely , they have proved that this code has minimum distance at most @xmath11 for @xmath375 .
the previous corollary recovers this result and also provides the exact number of codewords of weight @xmath71 and @xmath11 in this code .
let @xmath486 ( resp .
@xmath487 ) denote the number of codewords of hamming weight @xmath71 ( resp . of hamming weight @xmath11 ) in the binary cyclic code of length @xmath100 with defining set @xmath485 .
then , we have * if @xmath10 is odd @xmath488 * if @xmath10 is even @xmath489 let @xmath46 over @xmath1 and let @xmath55 , @xmath53 , denote the number of solutions @xmath68 of @xmath490 it is known from proposition 2 and lemma 2 in @xcite that the number of codewords of weight @xmath71 and @xmath11 in the cyclic code of length @xmath69 with defining set @xmath491 is given by @xmath492.\end{aligned}\ ] ] therefore , we have @xmath493\;.\end{aligned}\ ] ] this formula was proved in corollary 1 of @xcite , but only in the particular case where @xmath494 . for @xmath120 , theorem [ th : x^7 ] implies that @xmath495 then , the values of @xmath486 and @xmath487 are deduced from the expression of @xmath496 given in theorem [ th : x^7 ] .
we are going to determine the differential uniformity of @xmath227 for @xmath498 .
we first consider the case where @xmath10 is even .
note that in this case , @xmath227 is not a permutation since @xmath499 .
[ th : n/2 ] let @xmath10 be an even integer , @xmath500 and @xmath501 for @xmath502 . then @xmath503 is locally - apn .
more precisely @xmath504 moreover , the differential spectrum of @xmath503 is : @xmath505 if @xmath506 then @xmath507 @xmath505 if @xmath508 , @xmath509 from theorem [ th : 2^t-1 ] , we obtain directly @xmath510 .
also , @xmath162 if @xmath128 is even and @xmath163 otherwise .
now , for all @xmath124 , we have to determine the number of roots in @xmath1 of @xmath511 or , equivalently , the number of roots of @xmath512 if @xmath68 is a root of @xmath149 then @xmath513 .
so , @xmath384 implies @xmath514 thus , we get a linear polynomial of degree @xmath11 which has at least the roots @xmath14 and @xmath231 .
hence , this polynomial has @xmath515 roots where @xmath515 is either @xmath11 or @xmath15 , including @xmath144 and @xmath145 .
therefore , for any @xmath268 , @xmath516 since @xmath517 .
we deduce that @xmath503 is localy - apn .
we also proved that @xmath63 unless @xmath518 when @xmath128 is even and @xmath519 otherwise .
moreover @xmath520 . according to lemma [ ident ] , we have for @xmath128 even : @xmath521 and @xmath522 so , we get @xmath523 and conclude with @xmath524 .
we proceed similarly for odd @xmath128 , with the following equalities derived from lemma [ ident ] : @xmath525 and we directly deduce a property on the corresponding class of linear polynomials .
let @xmath526 and let @xmath527 denote the absolute trace on @xmath528 .
consider the polynomials over @xmath1 : @xmath529 then , for any @xmath530 , these polynomials have either @xmath15 or @xmath11 roots in @xmath1 .
the first one has @xmath11 roots if and only if @xmath531 with @xmath532 , where @xmath533 is the cyclic subgroup of @xmath1 of order @xmath534 .
let @xmath535 .
we define @xmath536 using ( [ eqth7 ] ) , we get : @xmath537 to be clear , we summarize the situation : + if @xmath384 , @xmath538 , then @xmath539 ; + when @xmath539 , @xmath538 , one can have @xmath540 ; + if @xmath539 for @xmath541 only , this holds for @xmath542 too .
we consider the case where @xmath384 has more than the two solutions @xmath14 and @xmath231 .
the equation @xmath539 has two solutions ( not in @xmath271 ) if and only if @xmath543 with @xmath544 . but
@xmath545 for all @xmath4 , since @xmath546 . and , @xmath547 that is : @xmath548 is not in the cyclic subgroup @xmath533 of order @xmath534 of @xmath208 .
on the other hand , if @xmath539 then @xmath549 and we get @xmath550 we compute this last expression by replacing the value of @xmath551 : @xmath552 where @xmath527 is the absolute trace on @xmath528 .
we conclude that @xmath384 if and only if @xmath553 , with @xmath554 .
according to corollary [ cormain ] , the differential spectrum of @xmath555 determines the differential spectrum of @xmath556 let @xmath10 be an even integer @xmath500 and @xmath557 for @xmath502 .
then , @xmath558 is locally - apn .
it is differentially @xmath559-uniform and its differential spectrum is @xmath560 moreover , @xmath558 is a permutation if and only if @xmath506 . first , since @xmath526 , we have @xmath561 if @xmath128 is even ( _ i.e. _ , @xmath506 ) and @xmath562 if @xmath128 is odd ( _ i.e. _ , @xmath508 ) . here
@xmath563 .
let @xmath564 ( resp .
@xmath565 ) denote the differential spectrum of @xmath227 ( resp .
@xmath558 ) over @xmath1 . * for @xmath506
, we have @xmath566 and @xmath567 .
thus , @xmath568 , @xmath479 , @xmath569 and @xmath570 .
* for @xmath508 , we have @xmath571 and @xmath572 .
thus , @xmath483 , @xmath472 , @xmath569 and @xmath570 .
the differential spectrum of @xmath558 is then directly deduced by combining the previous formulas with the values of @xmath573 computed in theorem [ th : n/2 ] . in the case where @xmath10 is odd
, the differential uniformity of @xmath227 , with @xmath574 , can also be determined .
let @xmath10 be an odd integer , @xmath575 .
let @xmath576 with @xmath577 .
then , @xmath227 is a permutation and for all @xmath530 we have @xmath431 . moreover * if @xmath578 , then @xmath579 , and the differential spectrum satisfies @xmath63 for all @xmath580 and @xmath477 . * if @xmath581 , then @xmath582 and the differential spectrum satisfies @xmath63 for all @xmath583 . from theorem
[ th : 2^t-1 ] , we have @xmath584 ; moreover , if @xmath71 divides @xmath10 then @xmath585 else @xmath162 .
now , for all @xmath124 , we have to determine the number of roots in @xmath1 of @xmath586 or , equivalently , the number of roots of @xmath587 set @xmath588 and @xmath589 . if @xmath68 is a root of @xmath149 then @xmath513
so , @xmath384 implies @xmath590 since @xmath591 has degree @xmath454 , it has either @xmath454 or @xmath11 or @xmath15 solutions . in other terms , @xmath431 .
in this work , we point out that the family of all power functions @xmath592 has interesting differential properties .
the study of these properties led us to introduce locally - apn functions , as a generalization of the differential spectrum of the inverse function .
in particular , we give several results about the functions with a low differential uniformity within family ( [ famil ] ) .
there are classes of functions @xmath227 such that @xmath593 .
it is the case for the functions @xmath373 over @xmath1 ( see theorem [ th : x^7 ] ) . the functions such that @xmath594 can be differentially @xmath11-uniform for even @xmath10 only ( see corollary [ coro3 ] ) .
we have shown that , for exponents of the form @xmath595 , the apn property imposes many conditions of the value of @xmath128.in particular , it is easy to prove , using theorem [ th : 2^t-1 ] that such exponent must satisfy @xmath596 for even @xmath10 and @xmath597 for odd @xmath10 .
another condition can be derived from the recent result by aubry and rodier @xcite who proved the following theorem . from theorem [ aubrod ] , we know that @xmath594 implies @xmath606 we consider now the function @xmath228 .
note that , from theorem [ th : 2^t-1 ] , @xmath594 implies @xmath607 and @xmath608 .
moreover , we obtain directly from corollary [ cormain ] : we now concentrate on apn functions belonging to the family ( [ famil ] ) .
some are well - known as the inverse permutation for @xmath10 odd ( @xmath161 ) and the quadratic function @xmath614 ( @xmath353 ) .
there is also the function @xmath227 for @xmath615 with @xmath10 odd , because this function is the inverse of the quadratic function @xmath616 .
recall that @xmath617 is an apn function over @xmath1 if and only if @xmath618 and we have obviously @xmath619 ( for odd @xmath10 ) .
we conjecture that these three functions are the only apn functions within family ( [ famil ] ) .
if the previous conjecture holds then there are some consequences for the functions of ( [ famil ] ) which are differentially @xmath11-uniform . from corollary [ coro3 ] , we can say that such a function @xmath227 is a function over @xmath1 with @xmath10 even .
moreover @xmath228 , @xmath183 , is apn .
if the conjecture holds then @xmath621 ( @xmath161 ) is the only one possibility .
so , in this case we could conclude that _ the inverse function is the only one differentially @xmath11-uniform function of family ( [ famil])_. y. aubry and f. rodier , `` differentially @xmath11-uniform functions , '' , in _ arithmetic , geometry , cryptography and coding theory 2009 _ , ser .
contemporary mathematics , vol .
521 , ams , 2010 , pp . 18 .
http://arxiv.org/abs/0907.1734v1 .
c. blondeau , a. canteaut , and p. charpin , `` differential properties of power functions , '' _ int .
j. inform . and coding theory _ , vol . 1 , no . 2 , pp .
149170 , 2010 , special issue dedicated to vera pless . c. bracken , e. byrne , n. markin , and g. mcguire , `` new families of quadratic almost perfect nonlinear trinomials and multinomials , '' _ finite fields and their applications _
14 , no . 3 , pp .
703714 , 2008 .
k. browning , j. dillon , m. mcquistan , and a. wolfe , `` an apn permutation in dimension six , '' in _ finite fields : theory and applications - fq9 _ , ser .
contemporary mathematics , vol .
518.1em plus 0.5em minus 0.4em ams , 2010 , pp .
3342 .
a. canteaut , p. charpin , and h. dobbertin , `` binary @xmath622-sequences with three - valued crosscorrelation : a proof of welch conjecture , '' _ ieee transactions on information theory _ , vol .
46 , no . 1 , pp .
48 , jan .
2000 .
, `` almost perfect nonlinear power functions on @xmath623 : a new class for @xmath10 divisible by 5 , '' in _ proceedings of finite fields and applications fq5_.1em plus 0.5em minus 0.4emaugsburg , germany : springer - verlag , 2000 , pp . 113121 .
f. hernando and g. mcguire , `` proof of a conjecture on the sequence of exceptional numbers , classifying cyclic codes and apn functions , '' _ journal of algebra _ , 2011 , to appear .
[ online ] .
available : http://arxiv.org/abs/0903.2016 k. nyberg , `` differentially uniform mappings for cryptography , '' in _ advances in cryptology - eurocrypt93 _ , ser .
lecture notes in computer science , vol .
765.1em plus 0.5em minus 0.4emspringer - verlag , 1993 , pp .
k. nyberg and l. knudsen , `` provable security against differential cryptanalysis , '' in _ advances in cryptology - crypto92 _ , ser .
lecture notes in computer science , vol .
740.1em plus 0.5em minus 0.4em springer - verlag , 1993 , pp .
van lint and r.m .
wilson , `` binary cyclic codes generated by @xmath624 , '' _ ieee transactions on information theory _ , vol .
32 , no . 2 , p. 283 , march 1986 | we provide an extensive study of the differential properties of the functions @xmath0 over @xmath1 , for @xmath2 .
we notably show that the differential spectra of these functions are determined by the number of roots of the linear polynomials @xmath3 where @xmath4 varies in @xmath1.we prove a strong relationship between the differential spectra of @xmath0 and @xmath5 for @xmath6 . as a direct consequence ,
this result enlightens a connection between the differential properties of the cube function and of the inverse function .
we also determine the complete differential spectra of @xmath7 by means of the value of some kloosterman sums , and of @xmath8 for @xmath9 .
* keywords . *
differential cryptanalysis , block cipher , s - box , power function , monomial , differential uniformity , apn function , permutation , linear polynomial , kloosterman sum , cyclic codes . |
in the last decade , systematic measurements of the cosmic microwave background anisotropies as well as large - scale structure of the universe have led to the establishment of the `` standard cosmological model '' ( e.g. , @xcite ) .
the universe is close to a flat geometry , and is filled with the hypothetical cold dark matter ( cdm ) particles , together with a small fraction of baryons , which serve as the seeds of structure formation of the universe .
the most striking feature in the standard cosmological model is that the energy contents of the universe is dominated by the mysterious energy component called dark energy , which is supposed to drive the late - time cosmic acceleration discovered by the observation of distant supernovae ( e.g. , @xcite ) .
currently , our understanding of the nature of dark energy is still lacking . although the observation is roughly consistent with cosmological constant and with no evidence for time dependence of dark energy , long - distance modifications of general relativity have been proposed alternative to the dark energy and these reconcile with the observation of late - time acceleration ( see @xcite for reviews ) .
while a fully consistent model of modified gravity has not yet been constructed ( see @xcite for popular models ) , a possibility of break - down of general relativity still remains and should be tested .
to understand deeply the nature of dark energy or origin of cosmic acceleration , a further observational study is definitely important .
there are two comprehensive ways to distinguish between many models of dark energy and discriminate the dark energy from modified gravity .
one is to precisely measure the expansion history of the universe , and the other is to observe the growth of structure . among various observational techniques ,
baryon acoustic oscillations ( baos ) imprinted on the matter power spectrum or two - point correlation function can be used as a standard ruler to measure the cosmic expansion history ( e.g. , @xcite , see also for recent bao measurements ) .
the characteristic scale of baos , which is determined by the sound horizon scale of primeval baryon - photon fluid at the last scattering surface @xcite , is thought to be a robust measure and it lies on the linear to quasi - linear regimes of the gravitational clustering of large - scale structure @xcite . with a percent - level determination of the characteristic scale of baos , the expansion history can be tightly constrained , and the equation - of - state parameter of the dark energy , @xmath3 , defined by the ratio of pressure to energy density of dark energy , would be precisely determined within the precision of a few % level @xcite .
this is the basic reason why most of the planned and ongoing galaxy redshift surveys aim at precisely measuring the baos ( e.g. , @xcite ) . while the robustness of the baos as a standard ruler has been repeatedly stated and emphasized in the literature , in order to pursue an order - of - magnitude improvement , a precise theoretical modeling of baos definitely plays an essential role for precision measurement of bao scale , and it needs to be investigated taking account of the various systematic effects . among these , the non - linear clustering and redshift - space distortion effects as well as the galaxy biasing can not be neglected , and affect the characteristic scale , although their effects are basically moderate at the relevant wavenumber , @xmath4mpc@xmath5 .
recently , several analytic approaches to deal with the non - linear clustering have been developed , complementary to the n - body simulations .
in contrast to the standard analytical calculation with perturbation theory ( pt ) , these have been formulated in a non - perturbative way with techniques resumming a class of infinite series of higher - order corrections in perturbative calculation .
thanks to its non - perturbative formulation , the applicable range of the prediction is expected to be greatly improved , and the non - linear evolution of baryon acoustic oscillations would be accurately described with a percent - level precision .
the purpose of this paper is to investigate the viability of this analytic approach , focusing on a specific improved treatment . in the previous paper @xcite ,
we have applied a non - linear statistical method , which is widely accepted in the statistical theory of turbulence @xcite , to the cosmological perturbation theory of large - scale structure .
we have derived the non - perturbative expressions for the power spectrum , coupled with non - linear propagator , which effectively contain the information on the infinite series of higher - order corrections in the standard pt expansion .
based on this formalism , the analytic treatment of the non - perturbative expression is developed employing the born approximation , and the leading - order calculation of power spectrum is compared with n - body simulations in real space @xcite , finding that a percent - level agreement is achieved in a mildly non - linear regime ( see also @xcite ) . here
, we extend the analysis to those including the next - to - leading order corrections of born approximation .
in addition to the power spectrum , we will consider the two - point correlation function , paying a special attention on the baryon acoustic peak , i.e. , a fourier counterpart of baos in power spectrum .
further , we also discuss the non - linear clustering in redshift space , and the predictions of improved pt are compared with n - body results , combining a non - linear model of redshift - space distortion .
we examine how well the present non - linear model accurately describe the systematic effects on baos and/or baryon acoustic peak .
this paper is organized as follows . in sec .
[ sec : preliminaries ] , we briefly mention the basic equations for cosmological pt as our fundamental basis to deal with the non - linear gravitational clustering .
we then discuss in some details in sec . [
sec : cla ] how to compute the non - linear power spectrum or two - point correlation functions .
starting from the discussions on standard treatment of perturbative calculation and its non - perturbative reformulation called renormalized pt , we introduce the closure approximation , which gives a consistent non - perturbative scheme to treat the infinite series of renormalized pt expansions , and obtain a closed set of non - perturbative expressions for power spectrum . based on this , we present a perturbative treatment of the closed set of equations while keeping important non - perturbative properties .
section [ sec : pt_vs_n - body ] gives the main result of this paper , in which a detailed comparison between improved pt calculation and n - body simulation is made , especially focusing on the non - linear evolution of baos .
we compute the power spectrum and two - point correlation function in both real and redshift spaces , and investigate the accuracy of both predictions by comparing improved pt with n - body results .
finally , section [ sec : conclusion ] is devoted to the discussion and conclusion .
throughout the paper , we consider the evolution of cold dark matter ( cdm ) plus baryon systems neglecting the tiny fraction of ( massive ) neutrinos . owing to the single - stream approximation of the collisionless boltzmann equation , which is thought to be a quite accurate approximation on large scales , the evolution of the cdm plus baryon system can be treated as the irrotational and pressureless fluid system whose governing equations are continuity and euler equations in addition to the poisson equation ( see ref .
@xcite for review ) . in the fourier representation , these equations are further reduced to a more compact form .
let us introduce the two - component vector ( e.g.,@xcite ) : @xmath6 where the subscript @xmath7 selects the density and the velocity components of cdm plus baryons , with @xmath8 and @xmath9 , where @xmath10 and @xmath11 are the scale factor of the universe and the hubble parameter , respectively . the function @xmath12 is given by @xmath13 and the quantity @xmath14 being the linear growth factor .
then , in terms of the new time variable @xmath15 , the evolution equation for the vector quantity @xmath16 becomes @xmath17 \phi_b({\mbox{\boldmath$k$}};\eta)=\int\frac{d^3{\mbox{\boldmath$k$}}_1\,d^3{\mbox{\boldmath$k$}}_2}{(2\pi)^3 } \delta_d({\mbox{\boldmath$k$}}-{\mbox{\boldmath$k$}}_1-{\mbox{\boldmath$k$}}_2)\,\gamma_{abc}({\mbox{\boldmath$k$}}_1,{\mbox{\boldmath$k$}}_2)\ , \phi_b({\mbox{\boldmath$k$}}_1;\eta)\,\phi_c({\mbox{\boldmath$k$}}_2;\eta ) , \label{eq : vec_fluid_eq}\ ] ] where @xmath18 is the dirac delta function .
here and in what follows , we use the summation convention that the repetition of the same subscripts indicates the sum over the whole vector components .
the time - dependent matrix @xmath19 is given by @xmath20 the quantity @xmath21 is the density parameter of cdm plus baryons at a given time .
each component of the vertex function @xmath22 becomes @xmath23 note that the formal solution of @xmath24 can be obtained from eq .
( [ eq : vec_fluid_eq ] ) and is expressed as ( e.g. , @xcite ) @xmath25 here , the quantity @xmath26 is the constant vector which specifies the initial condition ( see next section ) , and the quantity @xmath27 denotes the linear propagator satisfying the following equation : @xmath28 g_{bc}(\eta,\eta')=0 , \label{eq : linear_prop}\ ] ] with the boundary condition @xmath29 .
the quantity @xmath30 is the random density field given at an early time @xmath31 , which is assumed to obey the gaussian statistic .
the power spectrum of density field is defined as @xmath32 eq . ( [ eq : vec_fluid_eq ] ) or ( [ eq : formal_sol ] ) is the fundamental building block of large - scale structure , and the three quantities @xmath22 , @xmath27 and @xmath33 introduced here constitute the basic pieces of standard pt . the graphical representation of them is shown in fig .
[ fig : diagram_basic ] ( see also ref .
) with boundary condition @xmath29 . the explicit expression of vertex function @xmath22 is given by eq . ( [ eq : def_gamma ] ) .
[ fig : diagram_basic],title="fig:",width=151 ] ) with boundary condition @xmath29 .
the explicit expression of vertex function @xmath22 is given by eq .
( [ eq : def_gamma ] ) .
[ fig : diagram_basic],title="fig:",width=151 ] ) with boundary condition @xmath29 .
the explicit expression of vertex function @xmath22 is given by eq .
( [ eq : def_gamma ] ) .
[ fig : diagram_basic],title="fig:",width=188 ] in this paper , we are especially concerned with the non - linear evolution of the two - point statistics , defined as the ensemble average of @xmath24 : @xmath34 in the above , there are four types of power spectra , @xmath35 , @xmath36 , @xmath37 and @xmath38 , which respectively correspond to the auto- and cross - power spectra , @xmath39 , @xmath40 , @xmath41 and @xmath42 .
note that in general we have @xmath43 unless @xmath44 .
consider how to compute the power spectrum based on the analytic treatment . in the standard treatment of the perturbation theory
, we first assume that the field @xmath24 is a small perturbed quantity and it is expanded as @xmath45 the explicit functional form of the quantity @xmath46 is systematically derived through the order - by - order treatment of eq .
( [ eq : vec_fluid_eq ] ) . substituting the above expansion into the definition ( [ eq : def_pk ] ) and evaluating it perturbatively ,
the power spectrum @xmath47 , shortly abbreviated as @xmath48 , is schematically expressed as @xmath49 where we chose @xmath50 , which implies that the growing - mode solution is imposed at the initial condition .
the function @xmath51 is the linear power spectrum given at an early time , obtained from the first - order quantity @xmath52 ( see eq .
( [ eq : def_p_0 ] for definition ) . the subsequent terms @xmath53 and @xmath54 represent the corrections to the linear - order perturbation , arising from the higher - order quantities , @xmath55 , @xmath56 , @xmath57 . in terms of the basic pieces of the diagrams shown in fig . [
fig : diagram_basic ] , the corrections @xmath58 and @xmath59 can be diagrammatically written as the one - loop and two - loop diagrams , i.e. , connected diagrams including one and two closed loops ( e.g. , see fig.5 in ref .
@xcite ) , and they are roughly proportional to @xmath60 and @xmath61 , where @xmath62 .
the explicit expressions for the power spectra together with the solutions of higher - order perturbation are summarized in appendix [ app : spt ] .
it should be noted that in the standard pt expansion , the positivity of the perturbative corrections is not guaranteed .
as we show later , the one- and two - loop contributions change the sign depending on the scale , and the absolute values of their amplitudes become comparable at lower redshift . in this respect ,
the standard pt has a poor convergence property , and the improvement of pt predictions may not be always guaranteed even including the higher - order corrections .
by contrast , renormalized pt ) . ] re - organizes the naive expansions of the standard pt by introducing the non - perturbative statistical quantities @xcite . in terms of these quantities , partial resummation of the naive expansion series is made , and the resultant convergence of the expansions is dramatically improved . in the renormalized pt , the power spectrum @xmath48 is expressed in the form as @xmath63 with @xmath31 being the time at which initial condition is imposed . here , @xmath64 is the power spectrum given at an early time @xmath31 .
the quantity @xmath65 is one of the non - perturbative statistical quantities called non - linear propagator , together with the non - linear power spectrum .
it is defined by @xmath66 where @xmath8 stands for a functional derivative .
the propagator @xmath65 describes the influence of an infinitesimal disturbance for @xmath67 on @xmath68 , and it coincides with the linear propagator @xmath27 in the limit @xmath69 .
note that there is another non - perturbative statistical quantity called full vertex , which is the non - linear counterpart of the vertex function @xmath22 @xcite . in the expression ( [ eq : rpt_expansion ] )
, the term @xmath70 represents the corrections coming from the loop diagrams .
in contrast to the standard pt , the loop diagrams in @xmath70 are whole _
irreducible _ , as the result of renormalization or re - organization .
further , each of the irreducible diagrams consists of the non - perturbative quantities of non - linear power spectrum , non - linear propagator and full vertex . in this respect ,
renormalized pt is a fully non - perturbative formulation , and even the expansions truncated at some levels still contain the higher - order effects of non - linear gravitational evolution .
this is the basic reason why the convergence properties in the renormalized pt are expected to be improved . as a trade - off , however , a straightforward application of renormalized pt seems difficult because of its non - perturbative formulation . while the term @xmath71 collects only the irreducible diagrams , it is expressed as an infinite sum of the loop diagrams , each of which involve the non - linear power spectrum itself . in practice
, the approximation or simplification is needed to evaluate the expressions ( [ eq : rpt_expansion ] ) , which we will discuss in next subsection . in this subsection , taking a great advantage of the formulation of renormalized pt
, we discuss how to approximately treat eq .
( [ eq : rpt_expansion ] ) without losing its non - perturbative aspect as much as possible . in the framework of renormalized pt
, the non - perturbative effects on the power spectrum are largely attributed to the non - linear propagator .
thus , it seems essential to give a framework to treat both the non - linear propagator and power spectrum on an equal footing . as it has been pointed out by ref .
@xcite , a similar kind of the renormalized expansion to the power spectrum ( [ eq : rpt_expansion ] ) can be made for the non - linear propagator : @xmath72 where the term @xmath73 represents the mode - coupling correction , which is also made of the infinite sum of irreducible loop diagrams . in order to give a self - consistent treatment for both eqs .
( [ eq : rpt_expansion ] ) and ( [ eq : rpt_expansion2 ] ) , a simple but transparent approach is to first ( i ) adopt the tree - level approximation of the full vertex function , and to ( ii ) apply the truncation procedure to the mode - coupling terms .
this treatment has been frequently used in the statistical theory of turbulence in order to deal with the navier - stokes equation , and is called _ closure approximation _ @xcite . in the first approximation ( i ) , the full vertex function is simply replaced with the linear - order one , i.e. , @xmath22 defined in eq .
( [ eq : def_gamma ] ) .
as for the truncation ( ii ) , the simplest choice is to keep the one - loop renormalized diagram only , and to discard all other contributions . with this approximation , the mode - coupling terms in @xmath74 and @xmath65
are simply described by @xmath75 and @xmath76 .
the analytical expressions for the one - loop contributions becomes @xcite @xmath77 the integrand in @xmath78 contain the function @xmath79 , which represents the non - linear mode - coupling between different fourier modes , given by @xmath80 note that the mode - coupling function @xmath81 possesses the following symmetry : @xmath82 .
the corresponding diagrams to the integral expressions for power spectrum and non - linear propagator , i.e. , eqs .
( [ eq : rpt_expansion ] ) and ( [ eq : rpt_expansion2 ] ) with mode - coupling terms ( [ eq : p_mc_cla ] ) and ( [ eq : g_mc_cla ] ) , are shown in fig .
[ fig : diagram_cla ] . and @xmath83 .
in the renormalized pt , the mode - coupling term is expressed as an infinite sum of the irreducible loop corrections . truncating the infinite sum at one - loop order and adopting the tree - level approximation of the full vertex function , we obtain the closed system of power spectrum and propagator , as shown in the figure .
[ fig : diagram_cla],width=529 ] and @xmath83 . in the renormalized pt
, the mode - coupling term is expressed as an infinite sum of the irreducible loop corrections . truncating the infinite sum at one - loop order and adopting the tree - level approximation of the full vertex function , we obtain the closed system of power spectrum and propagator , as shown in the figure .
[ fig : diagram_cla],width=529 ] it is worth mentioning that the integral equations ( [ eq : rpt_expansion ] ) and ( [ eq : rpt_expansion2 ] ) with truncated mode - coupling terms ( [ eq : p_mc_cla ] ) and ( [ eq : g_mc_cla ] ) can be recast in the form of the integro - differential equations , and both the power spectrum and non - linear propagator can be computed by solving the evolution equations .
this forward treatment seems especially suited for the full non - linear treatment of closure approximation and would be faster than directly treating the integral equations .
numerical algorithm to solve evolution equations , together with preliminary results , is presented in details in ref .
@xcite ( see also @xcite ) . in the present paper
, we are especially concerned with the evolution of baos around @xmath84mpc@xmath5 , where the non - linearity of gravitational clustering is rather mild , and the analytical treatment even involving some approximations is still useful . here , employing the born approximation , we analytically evaluate the integral equations ( [ eq : rpt_expansion ] ) and ( [ eq : p_mc_cla ] ) @xcite .
a fully numerical study on baos without born approximation will be discussed in a separate paper .
the born approximation is the iterative approximation scheme in which the leading - order solutions are first obtained by replacing the quantities in the non - linear integral terms with linear - order ones .
the solutions can be improved by repeating the iterative substitution of the leading - order solutions into the non - linear integral terms .
consider the time evolution of the power spectrum started from the time @xmath31 .
for a sufficiently small value of @xmath31 , the early - time evolution of power spectrum is well - approximated by the linear theory .
assuming the growing - mode initial condition , we have @xmath85 with @xmath86 . then , substituting eq .
( [ eq : initial_pk ] ) into ( [ eq : rpt_expansion ] ) , the iterative evaluation of the the integral equations ( [ eq : rpt_expansion ] ) with ( [ eq : p_mc_cla ] ) by the born approximation leads to @xcite @xmath87 where we define @xmath88 .
the terms @xmath89 and @xmath90 respectively represent the leading- and next - to - leading order results of the born approximation to the mode - coupling term ( [ eq : p_mc_cla ] ) . the explicit expressions become @xmath91 the kernels @xmath92 and @xmath93 are respectively given by @xmath94 the diagram corresponding to the above expressions is shown in fig . [
fig : born ] .
note that in deriving the expression ( [ eq : pk_born ] ) , we do not expand the propagators @xmath65 and their non - perturbative properties still hold . in order to evaluate eq .
( [ eq : pk_born ] ) , we use the analytic solution of @xmath65 derived in ref .
@xcite , where the non - linear propagator was constructed approximately by matching the asymptotic behaviors at low- and high - k modes , based on eqs .
( [ eq : rpt_expansion2 ] ) with ( [ eq : g_mc_cla ] ) .
the resultant analytic solution behaves like @xmath95 at @xmath96 , where the quantity @xmath97 is the bessel function with its argument @xmath98 , and the velocity dispersion @xmath99 is approximately described by the linear theory , i.e. , @xmath100 .
note that the final results of the power spectrum are a little bit sensitive to the high-@xmath101 behavior of the propagator , and a naive application of the approximate solution leads to a slight shift in the amplitude of power spectrum . while this is not serious at all for the leading - order calculation
, it amounts a percent - level shift when we consider the higher - order correction , @xmath102 . as discussed by crocce & scoccimarro ( 2008 ) , one possible reason for this may be a small contribution from the sub - leading corrections in the propagator . in order to remedy the effect of small corrections ,
we follow the method proposed by ref .
we define @xmath103^{1/2},\ ] ] where @xmath104 means the non - linear matter power spectrum .
then , the sub - leading correction can be corrected by simply multiplying the factor @xmath105 by @xmath99 , i.e. , @xmath106 .
note that this treatment is only applied to the propagator in the lowest - order term in eq .
( [ eq : pk_born ] ) , which most sensitively affects the power spectrum amplitude on small scales . for simplicity ,
we use ` halofit ` @xcite to compute @xmath104 and adopt the cutoff wavenumber , @xmath107 , where @xmath108 is the non - linear scale defined by ref . @xcite . in the rest of this paper
, we present the results for the analytic treatment based on the expression ( [ eq : pk_born ] ) . in computing the mode - coupling terms @xmath89 and @xmath90
, we must first evaluate the functions @xmath92 and @xmath93 for a given set of arguments , which involve the one- and two - dimensional integrals over time @xmath109 .
we use the gaussian quadrature for these time integrations . as for the momentum integrals in the mode - coupling terms , thanks to the symmetry of the functions @xmath92 and @xmath93 , the multi - dimensional integrals in @xmath89 and @xmath90 can be reduced to the two- and four - dimensional integrals , respectively .
we use the gaussian quadratures for the momentum integral in @xmath89 .
the four - dimensional momentum integration in the mode - coupling term @xmath90 is performed with monte carlo technique of quasi - random sampling using the library , ` cuba`@xcite .
finally , we note that the formulation and analytic treatment presented here have several distinctions and similarities to the other non - perturbative calculations proposed recently . in appendix
[ app : comparison ] , we compare the present work with a subset of these treatments , and discuss how the approach developed here is complementary to or expands on these studies .
in this section , particularly focusing on the baos , we compare the improved pt predictions from the analytic treatment of closure approximation with results of n - body simulations .
we use a publicly available cosmological n - body code , ` gagdet2 ` @xcite .
we ran two sets of simulations , ` wmap3 ` and ` wmap5 ` , in which we adopt the standard lambda cdm model with cosmological parameters determined from the wmap3 and wmap5 , respectively @xcite .
the ` wmap3 ` run is basically the same n - body run as described in ref .
@xcite , and a quantitative comparison between the leading - order results of improved pt and simulations has been previously made .
we basically use the results of ` wmap3 ` run to check the consistency of the present calculations with the previous work .
the ` wmap3 ` run is also helpful to cross - check the convergence properties in the new simulation , ` wmap5 ` , which increase the number of realizations to @xmath110 .
table [ tab : n - body_params ] summarizes the parameters used in the simulations .
the initial conditions were created with the ` 2lpt ` code @xcite at initial redshift @xmath111 , based on the linear transfer function calculated from ` camb ` @xcite .
the number of meshes used in the particle - mesh computation is @xmath112 .
we adopt a softening length of @xmath113mpc for tree forces . we store three output redshifts for ` wmap3 ` run , whereas we select four output redshifts for ` wmap5 ` run ; @xmath114 , @xmath115 , and @xmath116 ( ` wmap3 ` ) : @xmath114 , @xmath117 , @xmath115 , and @xmath118 ( ` wmap5 ` ) . using these outputs , we compute the power spectrum and two - point correlation function in both real and redshift spaces .
the calculation of the matter power spectrum adopted here is basically the same treatment as in ref .
the standard method to compute the power spectrum is to square the fourier transform of the density field and to take an average over realizations and fourier modes .
this is given by @xmath119 where @xmath120 and @xmath121 are the number of fourier modes in the @xmath122-th wavenumber bin and the number of realizations , and @xmath123 and @xmath124 are the minimum and the maximum wavenumber of the @xmath122-th bin , respectively .
the quantity @xmath125 means the density field in fourier space obtained from the @xmath126-th realization data .
we use the cloud - in - cells interpolation for the density assignment of particles onto a @xmath112 mesh , and correct the window function .
note that the power spectra measured from the standard treatment above suffer from the effect of finite - mode sampling discussed by ref .
the resultant power spectrum deviates from the prediction for the ideal ensemble average , and exhibits the anomalous growth of power spectrum amplitude on large scales . in order to reduce the effect of finite mode sampling at @xmath127mpc@xmath5 ,
we multiply the measured power spectrum by the ratio , @xmath128 , where the quantity @xmath129 is calculated from the perturbation theory up to the third - order in density field , and @xmath130 is the input linear power spectrum extrapolated to a given output redshift . note that in computing @xmath129 , we use the gaussian - sampled density field used to generate the initial condition of each n - body run . with this treatment , the individual random nature of each n - body run
is weakened , and the errors associated with anomalous growth is reduced with that in redshift space . to be precise ,
we compute the multipole moments of the redshift - space power spectrum , and the ratio , @xmath131 , is multiplied for each multipole spectrum ( see sec . [
subsubsec : pk_in_red ] ) . ] . for the estimation of two - point correlation function
, we adopt the grid - based calculation using the fast fourier transformation ( fft ) . in this treatment ,
similar to the power spectrum analysis , we first compute the square of the density field on each grid of fourier space . then , applying the inverse fourier transformation , we take the average over realization and distance , and obtain the two - point correlation function .
schematically , this is expressed as @xmath132 , \label{eq : estimator_xi}\ ] ] where the operation @xmath133 stands for the inverse fft of the squared density field on each grid .
note here that @xmath134 is simply chosen at the center of the @xmath122-th radial bin , i.e. , @xmath135 .
( [ eq : estimator_xi ] ) usually suffers from the ambiguity of the zero - point normalization in the amplitude of two - point correlation function , because of the lack of the low-@xmath101 powers due to the finite boxsize of the simulations . with the @xmath112 grids and the boxsize of @xmath136gpc , however , we can safely evaluate the two - point correlation function around the baryon acoustic peak .
comparison between different computational methods , together with convergence check of this method , is presented in appendix [ app : computing_tpcf ] . finally ,
similar to the estimation of power spectrum , the finite - mode sampling also affects the calculation of the two - point correlation function .
we thus correct it by subtracting and adding the extrapolated linear density field as , @xmath137 , where @xmath138 is the correlation function estimated from the gaussian density field , and @xmath139 is the linear theory prediction of two - point correlation function .
.[tab : n - body_params ] parameters of n - body simulations [ cols="<,^,^,^,^,^,^,^,^,^,^",options="header " , ] is separately plotted . in left panel ,
one - loop and two - loop corrections in the standard pt , @xmath140 and @xmath141 , are plotted , while in right panel , the mode - coupling corrections @xmath142 and @xmath143 in the improved pt given at eqs .
( [ eq : cla_mc1 ] ) and ( [ eq : cla_mc2 ] ) respectively shown ( labeled as mc1 and mc2 ) , together with the first term in eq .
( [ eq : pk_born ] ) ( labeled as g@xmath144p@xmath145 ) .
note that dashed lines indicate the negative values .
[ fig : convergence],title="fig:",width=359 ] is separately plotted . in left panel , one - loop and two - loop corrections in the standard pt , @xmath140 and @xmath141 ,
are plotted , while in right panel , the mode - coupling corrections @xmath142 and @xmath143 in the improved pt given at eqs .
( [ eq : cla_mc1 ] ) and ( [ eq : cla_mc2 ] ) respectively shown ( labeled as mc1 and mc2 ) , together with the first term in eq .
( [ eq : pk_born ] ) ( labeled as g@xmath144p@xmath145 ) .
note that dashed lines indicate the negative values .
[ fig : convergence],title="fig:",width=359 ] before addressing a quantitative comparison between n - body simulation and improved pt , we first discuss the convergence properties of the improved pt , and consider how well the calculation based on the improved pt does improve the prediction compared to the standard pt .
[ fig : convergence ] plots the overall behaviors of the non - linear power spectrum of density fluctuation , @xmath146 , given at @xmath147 , adopting the ` wmap3 ` cosmological parameters .
in left panel , the results of standard pt are shown , and the contributions to the total power spectrum up to the two - loop diagrams are separately plotted .
on the other hand , right panel shows the results of improved pt .
we plot the contributions up to the second - order born approximation labeled as mc1 and mc2 . in fig .
[ fig : convergence ] , there are clear distinctions between standard and improved pts . while the loop corrections in standard pt change their signs depending on the scales and exhibit an oscillatory feature , the corrections coming from the born approximation in the improved pt are all positive and mostly the smooth function of @xmath101 .
further , the higher - order corrections in the improved pt have a remarkable scale - dependent property compared to those in the standard pt ; their contributions are well - localized around some characteristic wavenumbers , and they are shifted to the higher @xmath101 modes as increasing the order of pt .
these trends clearly indicate that the improved pt with closure approximation has a better convergence property .
qualitative behaviors of the higher - order corrections quite resemble the predictions of rpt by crocce & scoccimarro ( 2008 ) @xcite .
now , let us focus on the behavior of baos , and discuss how the convergence properties seen in fig .
[ fig : convergence ] affect the predictions of bao features . in fig .
[ fig : ratio_pk_real ] , adopting the ` wmap3 ` cosmological parameters , we plot the ratio , @xmath148 , where the function @xmath149 is the linear power spectrum from the smooth transfer function neglecting the bao feature in ref .
@xcite . in left panel ,
n - body simulations are compared with the leading - order results of pt predictions , i.e. , standard pt including the one - loop correction ( dashed ) , and improved pt with first - order born correction ( solid ) .
apart from the wiggle structure , the amplitude of standard pt prediction monotonically increases with wavenumber @xmath101 , and tends to overestimate the results of n - body simulations . on the other hand
, the amplitude of improved pt prediction rapidly falls off at a certain wavenumber , and the deviation from n - body results becomes significant .
however , a closer look at the behavior on large scales reveals that improved pt prediction gives a better agreement with simulation .
the results are indeed consistent with the previous findings in ref .
the situation becomes more impressive when we add the next - to - leading order corrections .
as shown in right panel , the improved pt gets the power on smaller scales , and reproduces the n - body results in a wider range of wavenumber .
by contrast , the prediction of standard pt depicted as dashed lines seems a little bit subtle . compared to the one - loop results , the amplitudes of the standard pt prediction including the two - loop correction are slightly reduced , and the agreement with n - body simulation seems apparently improved a bit at higher redshift . at lower redshift @xmath147 , however , the correction coming from the two - loop order becomes significant , and the prediction eventually underestimates the simulation .
the reason for these behaviors basically comes from the competition between positive and negative contributions of the one - loop and two - loop corrections , respectively ( see left panel of fig . [
fig : convergence ] ) .
these are consistent with those findings in ref .
@xcite ( see fig . 1 of their paper ) . , given at redshifts @xmath114(top ) , @xmath115(middle ) and @xmath116(bottom ) .
cosmological parameters used in the wmap3 simulations are adopted to compute the power spectrum from standard pt and improved pt , and the results are compared with n - body simulations ( symbols with error - bars ) .
the reference spectrum @xmath149 is calculated from the no - wiggle formula of the linear transfer function in ref .
@xcite . in each panel ,
dotted , dashed and solid lines represent the linear , standard pt and improved pt results , respectively . in left panel ,
leading - order results of standard pt and improved pt are shown , while in right panel , the results including the higher - order corrections are plotted .
[ fig : ratio_pk_real],title="fig:",width=302 ] , given at redshifts @xmath114(top ) , @xmath115(middle ) and @xmath116(bottom ) .
cosmological parameters used in the wmap3 simulations are adopted to compute the power spectrum from standard pt and improved pt , and the results are compared with n - body simulations ( symbols with error - bars ) .
the reference spectrum @xmath149 is calculated from the no - wiggle formula of the linear transfer function in ref .
@xcite . in each panel ,
dotted , dashed and solid lines represent the linear , standard pt and improved pt results , respectively . in left panel , leading - order results of standard pt and improved pt are shown , while in right panel , the results including the higher - order corrections are plotted .
[ fig : ratio_pk_real],title="fig:",width=302 ] /p_{\rm no\mbox{-}wiggle}(k)$ ] .
left panel shows the results for standard pt up to the two - loop order .
right panel presents the case of improved pt including the corrections up to the second - order born approximation of the mode - coupling term . in both panels , vertical arrows represent the wavenumbers @xmath150 of standard and improved pt ( from left to right ) , below which the leading - order pt predictions reproduce the n - body simulations well within @xmath151 accuracy ( see text in details ) .
[ fig : ratio_pk_real2],title="fig:",width=302 ] /p_{\rm
no\mbox{-}wiggle}(k)$ ] .
left panel shows the results for standard pt up to the two - loop order .
right panel presents the case of improved pt including the corrections up to the second - order born approximation of the mode - coupling term . in both panels , vertical arrows represent the wavenumbers @xmath150 of standard and improved pt ( from left to right ) , below which the leading - order pt predictions reproduce the n - body simulations well within @xmath151 accuracy ( see text in details ) .
[ fig : ratio_pk_real2],title="fig:",width=302 ] in fig .
[ fig : ratio_pk_real2 ] , to clarify the range of agreement in more quantitative ways , we plot the fractional difference divided by the smoothed reference spectra , @xmath152/p_{\rm no\mbox{-}wiggle}$ ] , where the quantity @xmath153 implies the standard and improved pt predictions in left and right panels , respectively . here , the vertical arrows represent the maximum wavenumber @xmath150 , below which the leading - order predictions of standard or improved pt reproduce the n - body results quite well within the @xmath151 accuracy . according to nishimichi et al .
@xcite , this has been determined by the detailed comparison between models and simulations , and is empirically characterized by solving the following equation : @xmath154 with @xmath155 for the one - loop standard pt , and @xmath156 for the improved pt up to the first - order born correction . comparing these convergence regimes of the leading - order calculation with results of fractional differences , fig . [ fig : ratio_pk_real2 ] shows that the inclusion of higher - order terms does not always improve the prediction in the standard pt treatment .
by contrast , the improved pt calculation does improve the predictions , and the range of agreement between n - body simulations and the predictions becomes wider . in fig .
[ fig : ratio_pk2_alpha ] , we plot the results for the ` wmap5 ` simulations , which have relatively large value of @xmath157 compared to the wmap3 run ( see table [ tab : n - body_params ] ) . left and right panels respectively plot
the ratio of power spectrum amplitude and the fractional difference between n - body results and improved pt predictions . with the @xmath110 runs of n - body simulations , the errors in the power spectrum amplitude
are greatly reduced , and it is clearly shown that the predictions of improved pt including the higher - order corrections almost coincide with the n - body results beyond the convergence regime of the leading - order calculations ( indicated by vertical arrows ) , and achieve a sub - percent accuracy . from this plot ,
the maximum wavenumber @xmath150 at each redshift can be estimated by comparing the predictions with n - body results as @xmath158mpc@xmath5(z=0.5 ) , @xmath159mpc@xmath5(z=1 ) , @xmath160mpc@xmath5(z=2 ) and @xmath161mpc@xmath5(z=3 ) .
these values roughly match those determined from the criterion ( [ eq : k_criterion ] ) with the constant @xmath162 .
although we did not store the @xmath147 data of ` wmap5 ` run to compare with analytic prediction , eq .
( [ eq : k_criterion ] ) using this constant value implies that the maximum wavenumber for improved pt becomes @xmath163mpc@xmath5 , which contrasts with the one for the one - loop prediction of standard pt , @xmath164mpc@xmath5 .
thus , the improved pt including up to the second - order born approximation is expected to be still accurate at @xmath147 , and it can cover the major part of the baos . a more detailed comparison at low redshift including other analytic prescriptions can be found in ref .
@xcite . , @xmath117 , @xmath115 and @xmath118 are shown .
the improved pt predictions plotted here include the corrections up to the second - order born approximation of the mode - coupling term , @xmath165 .
left : ratio of power spectrum to the smoothed reference spectra , @xmath148 .
solid and dotted lines are improved pt and linear theory predictions , respectively .
right : difference between n - body and improved pt results normalized by the no - wiggle formula , @xmath152/p_{\rm no\mbox{-}wiggle}(k)$ ] . in each panel ,
vertical arrows represent the wavenumber @xmath150 for the leading - order predictions of standard and improved pt ( from left to right ) .
[ fig : ratio_pk2_alpha],title="fig:",width=302 ] , @xmath117 , @xmath115 and @xmath118 are shown .
the improved pt predictions plotted here include the corrections up to the second - order born approximation of the mode - coupling term , @xmath165 .
left : ratio of power spectrum to the smoothed reference spectra , @xmath148 .
solid and dotted lines are improved pt and linear theory predictions , respectively .
right : difference between n - body and improved pt results normalized by the no - wiggle formula , @xmath152/p_{\rm no\mbox{-}wiggle}(k)$ ] . in each panel ,
vertical arrows represent the wavenumber @xmath150 for the leading - order predictions of standard and improved pt ( from left to right ) .
[ fig : ratio_pk2_alpha],title="fig:",width=302 ] /\xi_{\rm pt}(r)$ ] . in both panels ,
the symbols with error - bars indicate the n - body results averaged over the @xmath110 realizations in which the effect of finite - mode sampling is corrected : @xmath166 ( open stars ) , @xmath115 ( open squares ) , @xmath117 ( filled triangles ) , and @xmath167 ( crosses ) . , width=415 ] having confirmed the excellent properties of the improved pt , we turn to focus on the baryon acoustic peak in the two - point correlation function .
the two - point correlation function can be computed from the power spectrum as @xmath168 top panel of fig .
[ fig : xi_real ] shows the two - point correlation functions around the baryon acoustic peak at different redshifts @xmath169 and @xmath167 ( from top to bottom ) in the case adopting the @xmath170 cosmological parameters . also , lower panel plots the fractional differences between n - body and improved pt results , i.e. , @xmath171/\xi_{\rm pt}(r)$ ] .
after the correction of finite - mode sampling , the error - bars in n - body simulations are greatly reduced , and the deviation of the n - body results from linear theory predictions ( depicted as dotted lines ) is clearly seen . as decreasing the redshift ,
the baryon acoustic peaks become smeared and the position of the peak are slightly shifted to a smaller scale .
these trends can be accurately described by the leading - order calculation of improved pt , and the agreement between n - body results and the predictions is excellent .
the fractional error in amplitude is well within a few percent , except for a large separation beyond the location of baryon acoustic peak , where the accuracy of n - body results tends to be worsen due to the limited simulation boxsize .
note that the corrections coming from the higher - order born approximation do not alter the behaviors at @xmath172mpc , and their amplitudes are negligibly small compared to the error - bars of n - body simulations .
thus the leading - order prediction seems robust for describing the baryon acoustic peak .
it has been recently suggested by several authors that the smearing effect on baryon acoustic peak is mostly attributed to the random motion of mass distribution @xcite , and it is approximately described by the convolution of the gaussian smoothing function ( e.g. , @xcite ) . in the language of improved pt , this effect corresponds to the disappearance of the memory of initial condition , which is encoded in the non - linear propagator .
strictly speaking , the asymptotic behavior of the non - linear propagator is not a gaussian form in closure approximation , although the damping behavior manifestly exhibits in the approximate solution of non - linear propagator .
hence , the prediction for the two - point correlation function seems robust against the high-@xmath101 behavior of the non - linear propagator .
finally , it should be noted that the standard pt prediction fails to converge the integral in eq .
( [ eq : def_xi ] ) , because of the high-@xmath101 behavior of the power spectrum .
this is true even when including the higher - order correction of two - loop order .
thus , the successful results of improved pt prediction may be regarded as an outcome of non - perturbative property . in practical observation with galaxy redshift surveys
, the observed galaxy distribution is inevitably distorted due to the peculiar velocity of each galaxy .
the so - called redshift - space distortion is known to alter the shape of the power spectrum in two different ways ( e.g. , @xcite ) .
one is the apparent enhancement of the clustering signal called kaiser effect @xcite , which originates from the bulk motion of mass distribution falling into the massive halos .
another important effect is the finger - of - god ( fog ) effect , which effectively suppresses the power spectrum amplitude on small scales by the virialized random motion of the mass residing at a halos .
although a rigorous non - perturbative treatment of the redshift - space distortion is difficult , these two effects has been phenomenologically modeled as ( e.g. , @xcite ) @xmath173 where @xmath174 is the cosine of the angle between the line - of - sight direction and the fourier mode @xmath175 , and @xmath176 is the logarithmic derivative of linear growth factor , defined as @xmath177 .
the function @xmath178 represents the damping function which mimics the fog effect , and it asymptotically approaches unity in the @xmath69 limit , where the linear - theory formula by kaiser is recovered .
recently , scoccimarro @xcite proposed an improved version of the model ( [ eq : redshift_model_old ] ) to properly take account of the non - linear evolution of density and velocity fields on the kaiser effect ( see also @xcite ) .
this is expressed as @xmath179 \exp\{-(f\mu k\,\sigma_{\rm v})^2\}. \label{eq : redshift_model}\ ] ] here , the quantity @xmath99 is the one - dimensional velocity dispersion given by @xmath180 in what follows , we adopt the model ( [ eq : redshift_model ] ) to calculate the redshift - space power spectrum .
although this model is still phenomenological and may not be regarded as the best one , a comparison between the model predictions and n - body simulations shows that the prediction based on the model ( [ eq : redshift_model ] ) gives a better result .
taking eq .
( [ eq : redshift_model ] ) as a canonical model of the redshift - space distortion , we will investigate the extent to which the model ( [ eq : redshift_model ] ) faithfully reproduces the n - body results well , and discuss how to improve the model prescription .
, from the wmap5 simulations .
the reference spectrum @xmath181 is calculated from the no - wiggle approximation of the linear transfer function , and the linear theory of the kaiser effect is taken into account .
left panel shows the monopole power spectra ( @xmath182 ) , and the right panel shows the quadrupole spectra ( @xmath183 ) .
solid and dashed lines represent the results from the improved pt adopting the model of redshift - space distortion ( [ eq : redshift_model ] ) .
to plot the results , the linear theory was used to compute @xmath99 in dashed lines , while in solid lines , @xmath99 was determined by fitting the predictions to the n - body simulations . in each panel ,
vertical arrow indicates the maximum wavenumber @xmath150 for improved pt prediction including up to the second - order born approximation , which has been estimated from fig .
[ fig : ratio_pk2_alpha ] ( see sec .
[ subsubsec : real_pk ] for definition of @xmath150 ) .
, title="fig:",width=302 ] , from the wmap5 simulations .
the reference spectrum @xmath181 is calculated from the no - wiggle approximation of the linear transfer function , and the linear theory of the kaiser effect is taken into account .
left panel shows the monopole power spectra ( @xmath182 ) , and the right panel shows the quadrupole spectra ( @xmath183 ) .
solid and dashed lines represent the results from the improved pt adopting the model of redshift - space distortion ( [ eq : redshift_model ] ) . to plot the results ,
the linear theory was used to compute @xmath99 in dashed lines , while in solid lines , @xmath99 was determined by fitting the predictions to the n - body simulations . in each panel ,
vertical arrow indicates the maximum wavenumber @xmath150 for improved pt prediction including up to the second - order born approximation , which has been estimated from fig .
[ fig : ratio_pk2_alpha ] ( see sec .
[ subsubsec : real_pk ] for definition of @xmath150 ) .
, title="fig:",width=302 ] .
while the solid lines represent the linear theory prediction , the open squares indicate the results obtained by fitting the model ( [ eq : redshift_model ] ) to the monopole and quadrupole spectra of n - body simulations ( see fig .
[ fig : ratio_pk2_red ] ) .
[ fig : sigma_v],width=340 ] / p_{\ell,{\rm no\mbox{-}wiggle}}^{\rm(s)}(k)$ ] .
the left and right panels respectively represent the results from monopole and quadrupole power spectra .
note that the improved pt predictions are computed based on the model ( [ eq : redshift_model ] ) adopting the fitted value of @xmath99 . for comparison ,
the statistical errors limited by the cosmic variance of the survey volumes roughly corresponding to those of wfmos - like survey@xcite and boss@xcite are shown as shaded regions in panels of @xmath114 , @xmath184 and @xmath166 , assuming respectively the survey volumes of @xmath185 , @xmath186 and @xmath187 .
note that in each panel , vertical arrow indicates the maximum wavenumber @xmath150 determined from fig .
[ fig : ratio_pk2_alpha ] by comparison between n - body and improved pt results .
[ fig : ratio_pk2_error_red],title="fig:",width=302 ] / p_{\ell,{\rm no\mbox{-}wiggle}}^{\rm(s)}(k)$ ] . the left and right panels respectively represent the results from monopole and quadrupole power spectra .
note that the improved pt predictions are computed based on the model ( [ eq : redshift_model ] ) adopting the fitted value of @xmath99 . for comparison ,
the statistical errors limited by the cosmic variance of the survey volumes roughly corresponding to those of wfmos - like survey@xcite and boss@xcite are shown as shaded regions in panels of @xmath114 , @xmath184 and @xmath166 , assuming respectively the survey volumes of @xmath185 , @xmath186 and @xmath187 .
note that in each panel , vertical arrow indicates the maximum wavenumber @xmath150 determined from fig .
[ fig : ratio_pk2_alpha ] by comparison between n - body and improved pt results .
[ fig : ratio_pk2_error_red],title="fig:",width=302 ] for a quantitative comparison of model prediction with n - body simulation , we compute the multipole moments of the two - dimensional power spectrum @xmath188 : @xmath189 with @xmath190 being the legendre polynomials . substituting the model ( [ eq : redshift_model ] ) into the above , the monopole , quadrupole and hexadecapole contribution to the redshift - space power spectrum are analytically expressed as @xmath191 where the function @xmath192 is defined by @xmath193.\ ] ] the quantity @xmath194 is the incomplete gamma function of the first kind @xmath195 with its argument @xmath196 .
[ fig : ratio_pk2_red ] shows the monopole ( left ) and quadrupole ( right ) moments of the redshift - space power spectra at different redshifts , obtained from the ` wmap5 ` simulations .
we do not plot here the hexadecapole contributions , because the power spectrum estimated from the n - body simulations is still noisy even with the 30 realizations . in each panel of fig .
[ fig : ratio_pk2_red ] , the dashed lines indicate the improved pt predictions based on the model ( [ eq : redshift_model ] ) , where the corrections up to the second - order born approximations are included .
note that the velocity dispersion @xmath99 is computed from the linear theory .
clearly , the predictions all underestimate the n - body results , and the agreement between predictions and n - body simulations is restricted to a quite narrow range on large scales . as a reference
, we also show the maximum wavenumber @xmath150 of the improved pt prediction ( vertical arrows ) , in which we include the corrections up to the second - order born approximation in real space ( see fig . [
fig : ratio_pk2_alpha ] ) .
the reason why the prediction generically underestimates the n - body simulations would be partly attributed to the calculation of the velocity dispersion @xmath99 using the linear theory .
it has been advocated by several authors that the suppression of power spectrum by fog effect is originated from the non - linear structure of virialized halos , and thereby the linear theory estimation of @xmath99 may be inappropriate . in this respect ,
we admittedly regard @xmath99 as an uncontrollable parameter , which should be determined by fitting the predictions to n - body results .
the solid lines in each panel of fig . [ fig : ratio_pk2_red ] show the results of redshift - space spectra adopting the fitted values of @xmath99 . in estimating @xmath99 , both the monopole and quadrupole spectra
were fitted to the n - body results in the range of @xmath197 .
[ fig : sigma_v ] summarizes the fitted results of @xmath99 , which significantly deviate from the linear theory prediction at higher redshifts . then , apparently , overall agreement between prediction and simulation becomes fairly improved , although as a trade - off , small discrepancy manifests at low - k mode , where the n - body results rather agree well with the prediction adopting @xmath99 calculated from linear theory . in fig .
[ fig : ratio_pk2_error_red ] , left and right panels respectively plot the fractional differences of the monopole and quadrupole moments between the model predictions and n - body simulations . except for the narrow range of low - k modes ,
a percent - level agreement is almost achieved for the monopole power spectrum .
this is true at least within the convergence regime calibrated in real space ( see vertical arrows in fig .
[ fig : ratio_pk2_error_red ] ) .
however , the fractional error of the quadrupole power spectrum still exhibits a little bit large discrepancy , signaling the fact that the model ( [ eq : redshift_model ] ) misses something important for higher - multipole moment of redshift - space distortion . to see the significance of this deviation in practice , in fig .
[ fig : ratio_pk2_error_red ] , the expected @xmath198 errors limited by the cosmic variance , @xmath199 , are shown , depicted as the shaded region . here
, we specifically consider the ground - based bao surveys like wfmos survey@xcite and boss@xcite , assuming the survey volumes of @xmath200 at @xmath114 and @xmath201 at @xmath184 for wfmos survey , and @xmath202 at @xmath166 for boss , @xmath203 and @xmath204 . here
, we only consider the low-@xmath205 measurement with survey depth @xmath206 . ] . based on the approximation
that the density field is well - described by a gaussian random field , the cosmic - variance error @xmath199 can be estimated as @xmath207 ^ 2=\frac{2}{n_k}\,\sigma_{p,\ell}^2(k ) , \label{eq : var_pk}\ ] ] where the quantity @xmath208 is the number of fourier modes within a given bin at @xmath101 , and is given by @xmath209 .
the function @xmath210 is @xmath211 the expression ( [ eq : var_pk ] ) with ( [ eq : sigma_p_ell ] ) is a generalization of the cosmic - variance error in real space ( e.g. , @xcite ) to the multipole moments in redshift space .
note that the error @xmath199 depends on the bin width @xmath212 , for which we simply adopt the same bin size as used in the power spectrum analysis of n - body data .
the analytic estimate of @xmath213 based on eq .
( [ eq : var_pk ] ) is roughly consistent with the statistical errors estimated from the n - body data of 30 realizations .
comparison between the cosmic - variance errors and fractional differences shows that the discrepancy seen in the quadrupole power spectrum is definitely large , and it eventually exceeds the statistical error at large @xmath101 modes .
since this has happened inside the valid range of the improved pt calibrated in real space ( indicated as vertical arrows ) , we conclude that the current model prediction with ( [ eq : redshift_model ] ) is insufficient to describe the higher - multipole moments of baos , and a more elaborate work on the models of redshift - space distortion is needed for upcoming bao measurement .
finally , we discuss the correlation functions in redshift space .
similar to the power spectrum , we apply the multipole expansion to the anisotropic two - point correlation function as @xmath214 with @xmath215 .
the multipole moment of the correlation function , @xmath216 , is directly related to the fourier counterpart , @xmath217 through @xmath218 fig . [
fig : xi2_red ] shows the monopole ( left ) , quadrupole ( middle ) and hexadecapole ( right ) moments of correlation function . in each panel ,
the n - body results are compared with the predictions from linear theory ( dotted ) and the leading - order calculation of improved pt ( solid ) adopting the model ( [ eq : redshift_model ] ) with linear theory prediction of @xmath99 .
note that the predictions of improved pt are hardly changed by including the higher - order corrections and/or using the fitted value of @xmath99 , at least around the baryon acoustic peak , and the systematic differences between including and ignoring the corrections are well within the error - bars of n - body simulations .
as anticipated from the results in real space , the baryon acoustic peaks in the monopole moment tend to be smeared as decreasing redshift , but the effect seems little bit stronger than those in real space .
this is due to the additional effect coming from the redshift distortion .
although no prominent signal of the baos exists in the higher - multipole moments , the same tendencies can be seen in the quadrupole and hexadecapole moments .
the improved pt calculations are broadly consistent with n - body results , but small discrepancies manifest around the baryon acoustic peak and trough . lower panels of fig .
[ fig : xi2_red ] showing the fractional differences imply that these are at most @xmath219 effect in amplitude , except for the hexadecapole case with large error - bars of simulation .
it is interesting to note that no noticeable redshift dependence appears in the fractional differences , indicating that the discrepancies may be attributed to the model of redshift - space distortion .
furthermore , it turns out that these are well within the cosmic - variance errors of the ground - based bao measurement , indicated as shaded region . assuming that the underlying density field is well described by a gaussian random field , the cosmic variance for the multipole correlation functions @xmath220 can be written as ( see @xcite for cosmic - variance errors in real space ) @xmath221 ^ 2 = \frac{2}{v}\,\int \frac{dk\,k^2}{2\pi^2}\,\{j_\ell(ks)\,\sigma_{p,\ell}(k)\}^2 \label{eq : delta_xi}\ ] ] with @xmath210 given by eq .
( [ eq : sigma_p_ell ] ) .
note that the analytic estimation of cosmic - variance errors @xmath222 shown in fig .
[ fig : xi2_red ] reproduce the n - body results quite well .
hence , compared to the power spectrum in redshift space , the correlation functions obtained from the n - body simulation and analytic calculation can have a better agreement . presumably , this is because the acoustic peak structure in the correlation function is mostly attributed to the low - k behavior of the baos , and the power spectrum at low - k modes is accurately described by the model ( [ eq : redshift_model ] ) with the linear theory prediction of @xmath99 . in other words ,
the baryon acoustic peak would be robust against the non - linear effects at high - k modes ( see also @xcite ) .
this implies that even the prediction at the current level is sufficient to characterize the acoustic peak in the correlation function , and it can be used as an accurate theoretical template for future precision bao measurement .
note , however , that the measured amplitudes of the two - point correlation function are strongly correlated between different scales . in practice , not only the diagonal component but also the off - diagonal components of the covariance of the correlation function must be considered for a reliable estimation of cosmological distance , and a more careful study is needed . .
the solid and dotted lines are the predictions from the improved pt based on the model ( [ eq : redshift_model ] ) and linear theory , respectively .
note that only the leading - order born approximation to the the mode coupling term is included in the improved pt .
@xmath166 ( red ) ; @xmath184 ( magenta ) ; @xmath223 ( cyan ) ; @xmath114 ( green ) . for comparison , the statistical errors limited by the cosmic variance of the survey volumes @xmath185 , @xmath186 and @xmath187 are estimated from eq .
( [ eq : delta_xi ] ) , and are depicted as shaded regions around the n - body results at @xmath114 , @xmath184 and @xmath166 , respectively . the cosmic - variance error for hexadecapole is not shown here because of the large scatter .
bottom : fractional differences of the results between n - body simulations and improved pt predictions , @xmath224/\xi_{\rm pt}(s)$ ] for different redshifts at @xmath166 ( open stars ) , @xmath184 ( open squares ) , @xmath223 ( filled triangles ) , and @xmath114 ( crosses ) .
, title="fig:",width=245 ] .
the solid and dotted lines are the predictions from the improved pt based on the model ( [ eq : redshift_model ] ) and linear theory , respectively .
note that only the leading - order born approximation to the the mode coupling term is included in the improved pt .
@xmath166 ( red ) ; @xmath184 ( magenta ) ; @xmath223 ( cyan ) ; @xmath114 ( green ) . for comparison ,
the statistical errors limited by the cosmic variance of the survey volumes @xmath185 , @xmath186 and @xmath187 are estimated from eq .
( [ eq : delta_xi ] ) , and are depicted as shaded regions around the n - body results at @xmath114 , @xmath184 and @xmath166 , respectively . the cosmic - variance error for hexadecapole is not shown here because of the large scatter .
bottom : fractional differences of the results between n - body simulations and improved pt predictions , @xmath224/\xi_{\rm pt}(s)$ ] for different redshifts at @xmath166 ( open stars ) , @xmath184 ( open squares ) , @xmath223 ( filled triangles ) , and @xmath114 ( crosses ) .
, title="fig:",width=245 ] .
the solid and dotted lines are the predictions from the improved pt based on the model ( [ eq : redshift_model ] ) and linear theory , respectively .
note that only the leading - order born approximation to the the mode coupling term is included in the improved pt .
@xmath166 ( red ) ; @xmath184 ( magenta ) ; @xmath223 ( cyan ) ; @xmath114 ( green ) . for comparison ,
the statistical errors limited by the cosmic variance of the survey volumes @xmath185 , @xmath186 and @xmath187 are estimated from eq .
( [ eq : delta_xi ] ) , and are depicted as shaded regions around the n - body results at @xmath114 , @xmath184 and @xmath166 , respectively . the cosmic - variance error for hexadecapole is not shown here because of the large scatter .
bottom : fractional differences of the results between n - body simulations and improved pt predictions , @xmath224/\xi_{\rm pt}(s)$ ] for different redshifts at @xmath166 ( open stars ) , @xmath184 ( open squares ) , @xmath223 ( filled triangles ) , and @xmath114 ( crosses ) .
, title="fig:",width=245 ]
in this paper , we have presented the improved pt calculations of the matter power spectrum and two - point correlation function in real and redshift spaces . based on the closure approximation of the renormalized pt treatment , a closed set of the non - perturbative expressions for power spectrum and propagator
is obtained .
the resultant expression includes the effect of resummation for a class of loop diagrams at infinite order , and thereby the convergence of higher - order contributions is expected to be improved . employing the born approximation , we have analytically calculated the non - linear power spectrum , and compared the convergence properties of improved pt with those of standard pt by explicitly computing the higher - order corrections .
we have also made a detailed comparison between the improved pt result and n - body simulations . with a large boxsize and many realization data of n - body simulations ,
the statistical errors of two - point statistics are greatly reduced by the correction of the effect of finite - mode sampling , and this enables us to investigate the convergence check of numerical and analytic calculations at a percent level .
then , specifically focusing on the behaviors of baos , the power spectrum and two - point correlation functions are calculated in both real and redshift spaces . in redshift space
, the effect of redshift - space distortion which changes the clustering pattern of mass distribution should be incorporated into the improved pt predictions . in this paper , adopting the model proposed by ref .
@xcite ( eq . ( [ eq : redshift_model ] ) ) , we have quantified the extent to which the current model description faithfully reproduces the n - body results , and clarified the key ingredients toward an improved prescription of redshift - space distortion .
our important findings are summarized as follows : * the improved pt expansion based on the born approximation has better convergence properties , in marked contrast with the standard pt expansion .
the corrections coming from the mode - coupling term are well - localized positive functions of wavenumbers , and their contributions tend to be shifted to a higher @xmath101 region as increasing the order of perturbation .
thus , the inclusion of higher - order corrections stably improves the prediction , and the range of agreement with n - body results becomes wider in wavenumber . * in real - space power spectrum , the improved pt prediction including up to the second - order born correction seems essential for modeling bao precisely .
we estimated the maximum wavenumber @xmath150 , below which the results of both the n - body simulation and improved pt calculation converge well within the @xmath151 accuracy .
the resultant value of @xmath150 can be summarized as eq .
( [ eq : k_criterion ] ) with the constant value @xmath225 , which provides a way to estimate @xmath150 in a cosmology independent manner . on the other hand ,
if we consider the two - point correlation function in real space , the leading - order calculation turns out to be sufficiently accurate , and no higher - order correction is needed to describe the non - linear evolution of baryon acoustic peak seen in the n - body simulations . * modeling redshift - space power spectrum with eq .
( [ eq : redshift_model ] ) gives a broadly consistent result with n - body simulations , if we regard the velocity dispersion @xmath99 as a fitting parameter .
however , discrepancy between improved pt predictions and n - body results has appeared in the quadrupole power spectrum , and it becomes larger than the statistical errors limited by the cosmic variance of the survey volume @xmath226 a few @xmath0gpc@xmath1 .
this is true even in the valid range of improved pt , @xmath227 . on the other hand , while a small descrepancy has been also found in the two - point correlation , it turns out that the discrepancy is well within the cosmic - variance error , and even the leading - order prediction using the linear theory estimate of @xmath99 can be used as an accurate theoretical template for future ground - based bao measurement .
the recently proposed techniques to deal with the non - linear gravitational clustering , including the present treatment , have been greatly developed , and they would be a promising cosmological tool to precisely model the shape and amplitude of the power spectrum and/or the correlation functions in an accuracy of sub - percent level . combining the model of redshift - space distortion , we are now able to discuss the non - linear clustering in redshift space .
although the present paper is especially concerned with the analytical work , we note that the non - perturbative formulation with closure approximation is suited for forward treatment in time @xcite , in which all orders of born approximation can be fully incorporated into the predictions by numerically solving the evolution equations
. this approach would be particularly useful to study the non - linear matter power spectrum in general cosmological models , including the modified theory of gravity @xcite . finally , in practical application to the precision
bao measurements , there are several remaining issues to be addressed in the future work .
the improvement of the model of redshift - space distortion is , of course , a very important and urgent task .
the effect of galaxy biasing is also one of the key ingredients for modeling accurate theoretical template , and several attempts to take account of this effect have been recently made @xcite .
another interesting direction is to develop a fast computation of non - linear power spectrum or correlation function for an arbitrary cosmological model .
recently , the statistical sampling method for precise power spectrum emulation has been proposed @xcite . in this treatment ,
only a limited set of cosmological models can be used to predict power spectrum at the required accuracy over the prior parameter ranges .
the analytic approaches combining this method may provide a fast and reliable way to estimate the two - point statistics , and the development of this method would be valuable . we would like to thank yasushi suto and alan heavens for comments and discussion , thierry sousbie for teaching us an efficient computational method for two - point correlation function .
at is supported by a grant - in - aid for scientific research from the japan society for the promotion of science ( jsps ) ( no .
21740168 ) .
tn and ss acknowledge a support from jsps fellows .
this work was supported in part by grant - in - aid for scientific research on priority areas no .
467 `` probing the dark energy through an extremely wide and deep survey with subaru telescope '' , and jsps core - to - core program `` international research network for dark energy '' .
in this appendix , we briefly summarize the standard pt and derive a set of perturbative solutions . based on these solutions ,
we obtain the analytic expressions for power spectrum up to the two - loop order . as we mentioned in sec .
[ subsec : spt_vs_rpt ] , standard pt is the straightforward expansion of the quantity @xmath24 , and the perturbative solutions are obtained by order - by - order treatment of eq .
( [ eq : vec_fluid_eq ] ) . in order to systematically derive the solutions , the einstein - de sitter ( eds ) approximation
is often used in the literature . in the eds approximation , the matrix @xmath228 given by eq .
( [ eq : matrix_m ] ) is replaced with the one in the eds universe , i.e. , @xmath229 and @xmath230 .
this means that all the non - linear growth factors appearing in the higher - order solutions are expressed in terms of the linear growth factor @xmath14 . neglecting the contributions from the decaying mode , the resultant solution for @xmath24
is then expanded as @xmath231 the solution for each order of perturbation is expressed as @xmath232 where @xmath30 is the initial density field which we assume gaussian statistic .
the function @xmath233 is the symmetrized kernel of the @xmath122-th order solutions .
the explicit expressions for the kernel @xmath234 is obtained from the recursion relation , which can be derived by substituting the expansion ( [ eq : expand ] ) with ( [ eq : pt_sol ] ) into eq .
( [ eq : vec_fluid_eq ] ) ( e.g. , @xcite ) : @xmath235 with @xmath236 and @xmath237 . here
, the matrix @xmath238 is given by @xmath239 note that the kernel @xmath240 given above is not yet symmetric under the permutations of arguments , @xmath241 , and it should be symmetrized : @xmath242 using the perturbative solutions , the power spectrum defined by ( [ eq : def_pk ] ) is expanded as @xmath243 here , the quantity @xmath244 implies the ensemble average obtained from the @xmath126-th and @xmath122-th order perturbative solutions . in the above expression , the first term at the right - hand side is the linear power spectrum , while the second and third terms proportional to the growth factors @xmath245 and @xmath246 are respectively the so - called one - loop and two - loop corrections .
the explicit expressions for these corrections become ( e.g. , @xcite ) @xmath247 where @xmath248 is the initial power spectrum of the density field @xmath30 defined by eq .
( [ eq : def_p_0 ] ) , and we set @xmath249 .
note that the expression for one - loop power spectra can be further reduced to the one - dimensional and two - dimensional integral for @xmath250 and @xmath251 , respectively ( e.g. , @xcite ) . in the results presented in sec .
[ subsubsec : real_pk ] , we used the method of gaussian quadratures for numerical integration of one - loop power spectra . on the other hand , for the two - loop power spectra , the integration can not be simplified except for the first term in @xmath252 , and we need to directly evaluate the six - dimensional integration .
we adopted the monte - carlo integration to the two - loop power spectra .
the integration kernels for each term are generated numerically using the recursion relation ( [ eq : recursion ] ) and the condition ( [ eq : symmetrize ] ) .
[ fig : pietroni],height=219 ] in this appendix , we collect several recent works that attempt to improve the prediction of power spectrum and/or two - point correlation function , and discuss their qualitative differences .
a quantitative aspect of various analytic methods has been recently investigated in ref .
@xcite . here , we specifically comment on the approaches proposed by refs .
@xcite , which are very close to our treatment .
crocce & scoccimarro ( 2008 ) : : : first let us mention the work by ref . @xcite .
although the treatment presented in the paper are often quoted as rpt , strictly speaking , this is just the approximate treatment , which differs from the renormalized pt @xcite . as we mentioned in sec .
[ subsec : spt_vs_rpt ] , renormalized pt is the exact non - perturbative formulation without any approximations , and the power spectrum given by eq .
( [ eq : rpt_expansion ] ) is expressed as the infinite series of irreducible loop diagrams constructed from the non - linear propagator , full vertex , and non - linear power spectrum .
to make the analysis tractable , they adopted the following approximations : ( i ) the renormalized vertex is well - described by the ( linear ) vertex function ; ( ii ) the non - linear power spectra that enter into the calculation of @xmath253 are all replaced with the linear - order ones . in our language , this corresponds to the first - order born approximation .
then , using the approximate solution for propagator in ref .
@xcite , they explicitly calculated the power spectrum including the corrections up to the two - loop order .
the diagrams that they actually computed are shown in fig .
[ fig : rpt_born ]
. + compared to our analytical treatment with born approximation , there are two main differences .
one is the higher - order corrections that appear in the diagrams ( see fig . [
fig : born ] ) .
another important difference is the asymptotic behaviors in the non - linear propagator . at @xmath96 ,
the propagator used in their paper behaves like @xmath254 $ ] , which contrasts with @xmath95 in our closure approximation , where @xmath27 is the linear propagator and @xmath255 is defined by @xmath256 .
these distinctive features come from the partial resummation of a different class of higher - order terms when constructing the approximate solution of non - linear propagator ( see ref .
@xcite in details ) . despite these remarkable differences ,
it has been shown in ref .
@xcite that the leading - order calculations neglecting the higher - order terms ( two - loop diagram or second - order born correction ) can produce the same results which is indistinguishable from each other .
this is true at least on large scales , where the agreement between n - body simulations and improved pt predictions is better than a few percent .
pietroni ( 2008 ) : : : next consider the method proposed by ref .
@xcite , called time - rg method .
this method is based on the moment - based approach , and we first write down the moment equations . in general , this produces an infinite hierarchy of equations , however , ref .
@xcite assumes a vanishing trispectrum in order to truncate the hierarchy . as a result ,
a closed set of equations for power spectrum @xmath74 is obtained , which coulples with the evolution of bispectrum @xmath257 in some non - perturbative ways .
diagrammatic representation of this closed equations is shown in fig .
[ fig : pietroni ] , which can be compared with fig .
[ fig : diagram_cla ] in our treatment .
note that in the subject of statistical theory of turbulence , this truncation procedure is referred to as _ quasi - normal approximation _ ( e.g.,@xcite ) , and it is known to have several drawbacks ; positivity of the energy spectrum is not ensured , and it fails to recover the kolmogolov spectrum in the inertial range of turbulence .
+ nevertheless , the advantage of this treatment , similar to our closure approximation , is that the power spectrum can be computed numerically by solving the evolution equations .
this forward treatment seems quite efficient to bring out the non - perturbative effects incorporated into the formalism , and it has a wide applicability to include various physical effects .
recently , the formalism has been extended to deal with the effect of massive neutrinos @xcite .
valageas ( 2007 ) : : : the method proposed by ref .
@xcite is based on the path - integral formalism .
starting from the action for the cosmological fluid equation ( [ eq : vec_fluid_eq ] ) , which describes the statistical properties of the vector field @xmath24 , the large - n expansions as a technique of quantum field theory have been applied to derive the governing equations for power spectrum and propagator . in ref .
@xcite , two kinds of expansions have been presented , leading to the two different non - perturbative schemes , i.e. , steepest descent method and 2pi effective action method .
although both methods consistently reproduce the standard pt at the one - loop level , the latter includes the non - purturbative contributions which are not properly taken into account by the former method .
thus , the 2pi effective action method is expected to provide a better result .
it is interesting to note that despite the field - theoretical derivation , the resultant governing equations for the 2pi effective action method turn out to be mathematically equivalent to those obtained from the closure approximation @xcite .
hence , the diagrammatic representation of this formalism is exactly the same as shown in fig .
[ fig : diagram_cla ] .
matsubara ( 2008a ) : : : finally , we briefly mention the treatment proposed by ref . @xcite .
this is the lagrangian - based approach , and we begin by writing down the exact expressions for matter power spectrum in terms of the displacement vectors .
the resultant expression is in the exponential form , and the purterbative expansions are then applied for the explicit calculation of the ensemble average .
while a naive expansion of the displacement vectors , together with the solutions of lagrangian pt , merely reproduces the ( standard ) eulerian pt results , ref .
@xcite has applied a partial expansion , and some of the terms have been kept in the exponential form .
this can be interpreted as the partial resummation of a class of the infinite diagrams .
the resultant expression for power spectrum is quite similar to the one - loop result of standard pt , but slightly differs from it in the sense that there appears the exponential prefactor . as a consequence , the prediction reasonably recovers the damping behavior of the baos seen in the n - body simulations , and it also explains the smearing effect on the baryon acoustic peak in the two - point correlation function .
+ one noticiable point of this method is that it is rather straightforward to generalize the calculations in real space to those in redshift space , since the displacement vectors in redshift space can be simply given by a linear mapping from those in real space .
further , the computational cost is less expensive compared to the other analytic methods .
although the validity range of this method is restricted to a narrow range of the low-@xmath101 modes , it would be very powerful for a fast compuation of the two - point correlation function .
, title="fig:",width=264 ] , title="fig:",width=264 ] , title="fig:",width=264 ] , title="fig:",width=264 ] in this paper , the grid - based calculation with fft has been used for computing the two - point correlation functions from n - body data . here
, we compare it with other computational methods and check their convergences . fig .
[ fig : convergence_xi ] shows the two - point correlation functions measured at @xmath166 from the single realization of ` wmap5 ` simulations .
upper - left panel shows the results from the direct pair - counting . for each particle , we randomly select pairs , which are accumulated for each bin of separations , allowing for oversampling .
the estimated values of two - point correlation function are then plotted for different number of samples @xmath258 : @xmath259 , @xmath260 , @xmath261 and @xmath262 . the resultant total number of pairs , @xmath263 , indicated in the panel is given by @xmath264 , with @xmath265 being the total number of particles . note that the actual number of pairs that enters into the plotted range is less than @xmath263 . on the other hand ,
upper - right panel shows the results from the grid - based pair - counting introduced by barriga & gaztaaga ( 2002 ) @xcite ( see also ref .
@xcite ) . in this method
, we first construct the density field on a grid of @xmath266 cells , and then estimate the correlation function through the pair count on grids : @xmath267 compared to the direct pair - counting , this method is computationally efficient when we store the list of neighbor particles which contribute to a given bin of separation .
we plot the results adopting the two different number of cells , @xmath268 and @xmath269 . in lower - left panel ,
the grid - based calculation with fft ( see eq .
( [ eq : estimator_xi ] ) ) is used to compute the two - point correlation function , with different numbers of cells , @xmath268 , @xmath269 , @xmath270 and @xmath112 .
note that we adopt @xmath271 in the analysis presented in sec .
[ sec : pt_vs_n - body ] .
finally , in lower - right panel , the results for three different methods with the largest number of pairs or grids are collected and compared with each other . to check the convergence
, we further evaluate the residuals from the mean values , @xmath272 , and plot the results in each panel of fig .
[ fig : convergence_xi ] . here
, the mean values @xmath273 are estimated from the ensemble average over the three different results using the largest number of pairs or grids . as increasing the numbers @xmath263 or @xmath266 , the results for three different methods all approach the mean values @xmath273 , and a few percent - level agreement is achieved over the range of our interest ( except for the vicinity of zero - crossing point , @xmath274 ) .
it is interesting to note that residuals obtained from the grid - based pair - count and fft methods almost coincide with each other and the differences are hard to distinguish , indicating that both methods are equivalent even in the practical situation .
these experiments suggest that the grid - based calculation with fft is a reliable estimation method comparable to the other methods .
it should be emphasized that the method using fft is much more efficient than other pair - count methods .
for example , using @xmath275 cores of @xmath167ghz processors , the direct pair - counting with @xmath276 takes about two weeks to get the results shown in fig .
[ fig : convergence_xi ] .
the grid - based pair - counting is computationally less expensive than the direct pair - counting , but it still needs time - consuming calculations , especially for a large number of grids . by contrast , the method using fft only requires few minutes even with @xmath271 .
this can be achieved by a single - node calculation . | we study the non - linear evolution of baryon acoustic oscillations in the matter power spectrum and correlation function from the improved perturbation theory ( pt ) . based on the framework of renormalized pt , which provides a non - perturbative way to treat the gravitational clustering of large - scale structure ,
we apply the _ closure approximation _ that truncates the infinite series of loop contributions at one - loop order , and obtain a closed set of integral equations for power spectrum and non - linear propagator .
the resultant integral expressions are basically equivalent to those previously derived in the form of evolution equations , and they keep important non - perturbative properties which can dramatically improve the prediction of non - linear power spectrum . employing the born approximation , we then derive the analytic expressions for non - linear power spectrum and the predictions are made for non - linear evolution of baryon acoustic oscillations in power spectrum and correlation function .
we find that the improved pt possesses a better convergence property compared with standard pt calculation .
a detailed comparison between improved pt results and n - body simulations shows that a percent - level agreement is achieved in a certain range in power spectrum and in a rather wider range in correlation function . combining a model of non - linear redshift - space distortion
, we also evaluate the power spectrum and correlation function in redshift space .
in contrast to the results in real space , the agreement between n - body simulations and improved pt predictions tends to be worse , and a more elaborate modeling for redshift - space distortion needs to be developed .
nevertheless , with currently existing model , we find that the prediction of correlation function has a sufficient accuracy compared with the cosmic - variance errors for future galaxy surveys with volume of a few @xmath0gpc@xmath1 at @xmath2 . |
we start with some basic definitions .
a nonempty word @xmath2 is called a _ power _ if it can be written in the form @xmath3 , for some integer @xmath4 .
otherwise @xmath2 is called _
primitive_. thus murmur is a power , but murder is primitive .
a word @xmath5 is a _ factor _ of a word @xmath2 if there exist words @xmath6 such that @xmath7 . if further @xmath8 ( resp . , @xmath9 ) , then @xmath5 is a _ prefix _ ( resp . ,
_ suffix _ ) of @xmath2 . a prefix or suffix of a word @xmath2
is called _ proper _ if it is unequal to @xmath2 .
let @xmath10 be an ordered alphabet .
we recall the usual definition of lexicographic order on the words in @xmath11 .
we write @xmath12 if either * @xmath2 is a proper prefix of @xmath13 ; or * there exist words @xmath14 and letters @xmath15 such that @xmath16 and @xmath17 .
for example , using the usual ordering of the alphabet , we have @xmath18 . as usual
, we write @xmath19 if @xmath12 or @xmath20 .
a word @xmath2 is a _ conjugate _ of a word @xmath13 if there exist words @xmath21 such that @xmath22 and @xmath23 .
thus , for example , enlist and listen are conjugates .
a word is said to be _ lyndon _ if it is primitive and lexicographically least among all its conjugates .
thus , for example , academy is lyndon , while googol and googoo are not .
a classical theorem is that a finite word is lyndon if and only if it is lexicographically less than each of its proper suffixes @xcite .
we now turn to ( right- ) infinite words .
we write an infinite word in boldface , as @xmath24 and use indexing starting at @xmath25 . for @xmath26 ,
we let @xmath27 $ ] denote the set @xmath28 .
( if @xmath29 we get the empty set . )
we let @xmath30 $ ] denote the word @xmath31 . similarly ,
@xmath32 $ ] denotes the infinite set @xmath33 and @xmath34 $ ] denotes the infinite word @xmath35 . an infinite word or sequence @xmath24 is said to be _
@xmath0-automatic _ if there is a deterministic finite automaton ( with outputs associated with the states ) that , on input @xmath1 expressed in base @xmath0 , reaches a state @xmath36 with output @xmath37 equal to @xmath38 . for more details , see @xcite or . in several previous papers ,
we have developed a technique to show that many properties of automatic sequences are decidable .
the fundamental tool is the following : let @xmath39 be a predicate associated with a @xmath0-automatic sequence @xmath40 , expressible using addition , subtraction , comparisons , logical operations , indexing into @xmath40 , and existential and universal quantifiers .
then there is a computable finite automaton accepting the base-@xmath0 representations of those @xmath1 for which @xmath39 holds .
furthermore , we can decide if @xmath39 holds for at least one @xmath1 , or for all @xmath1 , or for infinitely many @xmath1 .
if a predicate is constructed as in the previous theorem , we just say it is `` expressible '' .
any expressible predicate is decidable . as an example , we prove let @xmath40 be a @xmath0-automatic sequence .
the predicate @xmath41 defined by `` @xmath30 $ ] is primitive '' is expressible .
[ prim - ex ] ( due to luke schaeffer ) it is easy to see that a word is a power if and only if it is equal to some cyclic shift of itself , other than the trivial shift . thus a word is a power if and only if there is a @xmath42 , @xmath43 , such that @xmath44 = x[i+d .. j]$ ] and @xmath45 = x[i .. i+d-1]$ ] .
a word is primitive if there is no such @xmath42 .
let @xmath40 be a @xmath0-automatic sequence .
the predicate @xmath46 defined by `` @xmath30
< { \bf x}[m .. n]$ ] '' is expressible .
we have @xmath30 < { \bf x}[m .. n]$ ] if and only if either * @xmath47 and @xmath30 = { \bf x}[m .. m+j - i]$ ] ; or * there exists @xmath48 such that @xmath49 = { \bf x}[m .. m+t]$ ] and @xmath50 < { \bf x}[m+t+1]$ ]
. let @xmath40 be a @xmath0-automatic sequence .
the predicate @xmath51 defined by `` @xmath30 $ ] is a lyndon word '' is expressible .
it suffices to check that @xmath30 $ ] is lexicographically less than each of its proper suffixes , that is , that @xmath52 holds for all @xmath53 with @xmath54 .
we can extend the definition of lexicographic order to infinite words in the obvious way .
we can extend the definition of lyndon words to ( right- ) infinite words as follows : an infinite word @xmath24 is lyndon if it is lexicographically less than all its suffixes @xmath55 = a_j a_{j+1 } \cdots$ ] for @xmath56 .
then we have the following theorems .
let @xmath57 be a @xmath0-automatic sequence .
the predicate @xmath58 defined by @xmath34
< { \bf x}[j .. \infty]$ ] is expressible .
this is equivalent to @xmath59 such that @xmath60 = { \bf x}[j .. j+t-1]$ ] and @xmath61 < { \bf x}[j+t]$ ] .
let @xmath57 be a @xmath0-automatic sequence .
the predicate @xmath62 defined by `` @xmath34 $ ] is an infinite lyndon word '' is expressible .
this is equivalent to @xmath58 holding for all @xmath63 .
siromoney et al .
@xcite proved that every infinite word @xmath24 can be factorized uniquely in exactly one of the following two ways : * as @xmath64 where each @xmath65 is a finite lyndon word and @xmath66 ; or * as @xmath67 where @xmath65 is a finite lyndon word for @xmath68 , and @xmath69 is an infinite lyndon word , and @xmath70 .
if ( a ) holds we say that the lyndon factorization of @xmath40 is infinite ; otherwise we say it is finite .
ido and melanon @xcite gave an explicit description of the lyndon factorization of the thue - morse word @xmath71 and the period - doubling sequence ( among other things ) .
( recall that the thue - morse word is given by @xmath72 = $ ] the number of @xmath73 s in the binary expansion of @xmath1 , taken modulo @xmath74 . ) for the thue - morse word , this factorization is given by @xmath75 where each term in the factorization , after the first , is double the length of the previous .
sbold @xcite and ern generalized these results to other related automatic sequences . in this section , generalizing the work of ido , melanon , sbold , and ern , we prove that the lyndon factorization of a @xmath0-automatic sequence is itself @xmath0-automatic .
of course , we need to explain how the factorization is encoded .
the easiest and most natural way to do this is to use an infinite word over @xmath76 , where the @xmath73 s indicate the positions where a new term in the factorization begins .
thus the @xmath77th @xmath73 , for @xmath78 , appears at index @xmath79 .
for example , for the thue - morse word , this encoding is given by @xmath80 if the factorization is infinite , then there are infinitely many @xmath73 s in its encoding ; otherwise there are finitely many @xmath73 s . in order to prove the theorem
, we need a number of results .
we draw a distinction between a _ factor _ @xmath81 of @xmath40 ( which is just a word ) and an _ occurrence _ of that factor ( which specifies the exact position at which @xmath81 occurs ) .
for example , in the thue - morse word @xmath71 , the factor @xmath82 occurs as @xmath83 $ ] and @xmath84 $ ] and many other places .
we call @xmath85 $ ] and @xmath86 $ ] , and so forth , the _ occurrences _ of @xmath82 . an occurrence is said to be lyndon if the word at that position is lyndon .
we say an occurrence @xmath87 $ ] is _ inside _ an occurrence @xmath88 $ ] if @xmath89 and @xmath90 . if , in addition , either @xmath91 or @xmath92 ( or both ) , then we say @xmath93 is _ strictly inside _
these definitions are easily extended to the case where @xmath95 or @xmath96 are equal to @xmath97 , and they correspond to the predicates @xmath98 ( inside ) and @xmath99 ( strictly inside ) given below : @xmath100 an infinite lyndon factorization @xmath101 then corresponds to an infinite sequence of occurrences @xmath102 , [ i_2 .. j_2 ] , \cdots\ ] ] where @xmath103 $ ] and @xmath104 for @xmath105 , while a finite lyndon factorization @xmath106 corresponds to a finite sequence of occurrences @xmath102 , [ i_2 .. j_2 ] , \ldots , [ i_r .. j_r ] , [ i_{r+1} .. \infty]\ ] ] where @xmath103 $ ] and @xmath104 for @xmath107 .
let @xmath40 be an infinite word .
every lyndon occurrence in @xmath40 appears inside a term of the lyndon factorization of @xmath40 .
[ lyno ] we prove the result for infinite lyndon factorizations ; the result for finite factorizations is exactly analogous .
suppose the factorization is @xmath108 .
it suffices to show that no lyndon occurrence can span the boundary between two terms of the factorization .
suppose , contrary to what we want to prove , that @xmath109 is a lyndon word for some @xmath110 that is a nonempty suffix of @xmath111 ( possibly equal to @xmath111 ) , and @xmath112 that is a nonempty prefix of @xmath113 ( possibly equal to @xmath113 ) , and and @xmath26 .
( if @xmath29 then there are no @xmath65 s at all between @xmath110 and @xmath112 . )
since @xmath110 is a suffix of @xmath111 and @xmath111 is lyndon , we have @xmath114 .
on the other hand , by the lyndon factorization definition we have @xmath115 . but
@xmath112 is a prefix of @xmath113 , so just by the definition of lexicographic ordering we have @xmath116 .
putting this all together we get @xmath117 .
so @xmath118 for all words @xmath13 .
on the other hand , since @xmath119 is lyndon , it must be lexicographically less than any proper suffix for instance , @xmath112 .
so @xmath120 .
take @xmath121 to get a contradiction with the conclusion in the previous paragraph .
the occurrence @xmath27 $ ] corresponds to a term in the lyndon factorization of @xmath40 if and only if * @xmath27 $ ] is lyndon ; and * @xmath27 $ ] does not occur strictly inside any other lyndon occurrence .
[ eight ] suppose @xmath27 $ ] corresponds to a term @xmath122 in the lyndon factorization of @xmath40 .
then evidently @xmath27 $ ] is lyndon .
if it occurred strictly inside some other lyndon occurrence , say @xmath123 $ ] , then we know from theorem [ lyno ] that @xmath123 $ ] itself lies in inside some @xmath124 , so @xmath27 $ ] must lie strictly inside @xmath124 , which is clearly impossible .
now suppose @xmath27 $ ] is lyndon and does not occur strictly inside any other lyndon occurrence . from theorem [ lyno ]
@xmath27 $ ] must occur inside some term of the factorization @xmath123 $ ] .
if @xmath27 \not= [ i' .. j']$ ] then @xmath27 $ ] lies strictly inside @xmath123 $ ] , a contradiction .
so @xmath27 = [ i' .. j']$ ] and hence corresponds to a term of the factorization .
the predicate @xmath125 defined by `` @xmath27 $ ] corresponds to a term of the lyndon factorization of @xmath40 '' is expressible .
indeed , by corollary [ eight ] , the predicate @xmath125 can be defined by @xmath126 we can now prove the main result of this section . using the encoding mentioned above
, the lyndon factorization of a @xmath0-automatic sequence is itself @xmath0-automatic .
[ lf ] using the technique of , we can create an automaton that on input @xmath77 expressed in base @xmath0 , guesses @xmath95 and checks if @xmath125 holds .
if so , it outputs @xmath73 and otherwise @xmath25 . to get the last @xmath77 in the case that the lyndon factorization is finite
, we also accept @xmath77 if @xmath62 holds
. we also have let @xmath40 be a @xmath0-automatic sequence .
it is decidable if the lyndon factorization of @xmath40 is finite or infinite .
the construction given above in the proof of theorem [ lf ] produces an automaton that accepts finitely many distinct @xmath77 ( expressed in base @xmath0 ) if and only if the lyndon factorization of @xmath40 is finite .
we programmed up our method and found the lyndon factorization of the thue - morse sequence @xmath71 , the period - doubling sequence @xmath127 , the paperfolding sequence @xmath128 , and the rudin - shapiro sequence @xmath129 , and their negations .
( the results for thue - morse and the period - doubling sequence were already given in @xcite , albeit in a different form . )
recall that the period - doubling sequence is defined by @xmath130 = | { \bf t}[n+1 ] - { \bf t}[n ] |$ ] .
the paperfolding sequence @xmath131 arises from the limit of the sequence @xmath132 , where @xmath133 and @xmath134 , where @xmath135 denotes reversal and @xmath136 maps @xmath25 to @xmath73 and @xmath73 to @xmath25 .
finally , the rudin - shapiro sequence @xmath129 is defined by @xmath137 = $ ] the number of ( possibly overlapping ) occurrences of @xmath138 in the binary expansion of @xmath1 , taken modulo @xmath74 .
the results are given in the theorem below .
the occurrences corresponding to the lyndon factorization of each word is as follows : * the thue - morse sequence @xmath71 : @xmath139 , [ 3 .. 4 ] , [ 5 .. 8 ] , [ 9 .. 16 ] , [ 17 .. 32 ] , \ldots , [ 2^i + 1 .. 2^{i+1 } ] , \ldots $ ] ; * the negated thue - morse sequence @xmath140 : @xmath141 , [ 1 .. \infty]$ ] ; * the rudin - shapiro sequence @xmath129 : @xmath142,[7 .. 14],[15 .. 30 ] , \ldots , [ 2^i - 1 .. 2^{i+1 } - 2 ] , \ldots$ ] ; * the negated rudin - shapiro sequence @xmath143 : @xmath141,[1 .. 1],[2 .. 2],[3 .. 10],[11 .. 42],[43 .. 46],[47 .. 174],\ldots , [ 4^i-4^{i-1}-4^{i-2}-1 .. 4^i-4^{i-1}-2 ] , [ 4^i-4^{i-1}-1 .. 4^{i+1}-4^i-4^{i-1}-1 ] , \ldots$ ] ; * the paperfolding sequence @xmath128 : @xmath142,[7 .. 14],[15 .. 30 ] , \ldots , [ 2^i - 1 .. 2^{i+1}-2 ] , \ldots$ ] ; * the negated paperfolding sequence @xmath144 : @xmath141 , [ 1 .. 1 ] , [ 2 .. 4 ] , [ 5 .. 9 ] , [ 10 .. 20 ] , [ 21 .. 84 ] , [ 85 .. 340 ] , \ldots , [ ( 4^i - 1)/3 .. 4(4^i - 1)/3 ] , \ldots $ ] ; * the period - doubling sequence @xmath127 : @xmath141 , [ 1 .. 4 ] , [ 5 .. 20 ] , [ 21 .. 84 ] , \ldots , [ ( 4^i - 1)/3 .. 4(4^i - 1)/3 ] , \ldots$ ] ; * the negated period - doubling sequence @xmath145 : @xmath146,[2 .. 9],[10 .. 41],[42 .. 169 ] , \ldots , [ 2(4^i -1)/3 .. 2(4^{i+1}-1)/3 - 1 ] , \ldots$ ] .
there is a useful generalization of @xmath0-automatic sequences to sequences over @xmath147 , the non - negative integers .
a sequence @xmath148 over @xmath147 is called @xmath0-regular if there exist vectors @xmath110 and @xmath112 and a matrix - valued morphism @xmath149 such that @xmath150 , where @xmath2 is the base-@xmath0 representation of @xmath1 . for more details , see . the subword complexity function
@xmath151 of an infinite sequence @xmath40 counts the number of distinct length-@xmath1 factors of @xmath40 .
there are also many variations , such as counting the number of palindromic factors or unbordered factors .
if @xmath40 is @xmath0-automatic , then all three of these are @xmath0-regular sequences .
we now show that the same result holds for the number @xmath152 of primitive factors and for the number @xmath153 of lyndon factors .
we refer to these two quantities as the `` primitive complexity '' and `` lyndon complexity '' , respectively .
the function counting the number of length-@xmath1 primitive ( resp .
, lyndon ) factors of a @xmath0-automatic sequence @xmath40 is @xmath0-regular .
[ primlyn ] by the results of , it suffices to show that there is an automaton accepting the base-@xmath0 representations of pairs @xmath154 such that the number of @xmath77 s associated with each @xmath1 equals the number of primitive ( resp .
, lyndon ) factors of length @xmath1 .
to do so , it suffices to show that the predicate @xmath155 defined by `` the factor of length @xmath1 beginning at position @xmath77 is primitive ( resp . ,
lyndon ) and is the first occurrence of that factor in @xmath40 '' is expressible .
this is just @xmath156 \not= { \bf x}[j .. j+n-1],\ ] ] ( resp .
, @xmath157 \not= { \bf x}[j .. j+n-1]).\ ] ] we used our method to compute these sequences for the thue - morse sequence , and the results are given below .
let @xmath158 denote the number of lyndon factors of length @xmath1 of the thue - morse sequence .
then @xmath159 [ lyndon - tm ] let @xmath160 denote the number of primitive factors of length @xmath1 of the thue - morse sequence . then @xmath161 [ prim - tm ] we can also state a similar result for the rudin - shapiro sequence .
let @xmath162 denote the lyndon complexity of the rudin - shapiro sequence .
then @xmath163 for all @xmath1 .
this sequence is @xmath74-automatic and there is an automaton of 2444 states that generates it .
[ lyndon - rs ] the proof was carried out by machine computation , and we briefly summarize how it was done .
first , we created an automaton @xmath164 to accept all pairs of integers @xmath154 , represented in base @xmath74 , such that the factor of length @xmath1 in @xmath129 , starting at position @xmath77 , is a lyndon factor , and is the first occurrence of that factor in @xmath129 .
thus , the number of distinct integers @xmath77 associated with each @xmath1 is @xmath162 .
the automaton @xmath164 has @xmath165 states . using the techniques in
, we then used @xmath164 to create matrices @xmath166 and @xmath167 of dimension @xmath168 , and vectors @xmath169 such that @xmath170 , if @xmath13 is the base-@xmath74 representation of @xmath1 . here
if @xmath171 , then by @xmath172 we mean the product @xmath173 . from this
we then created a new automaton @xmath174 where the states are products of the form @xmath175 for binary strings @xmath13 and the transitions are on @xmath25 and @xmath73 .
this automaton was built using a breadth - first approach , using a queue to hold states whose targets on @xmath25 and @xmath73 are not yet known . from theorem [
lyndon - bounded ] in the next section , we know that @xmath162 is bounded , so that this approach must terminate .
it did so at 2444 states , and the product of the @xmath175 corresponding to each state with @xmath2 gives an integer less than or equal to @xmath176 , thus proving the desired result and also providing an automaton to compute @xmath177 .
note that the lyndon complexity functions in theorems [ lyndon - tm ] and [ lyndon - rs ] are bounded .
this will follow more generally from theorem [ lyndon - bounded ] below .
of course , the original lyndon factorization was for finite words : every finite nonempty word @xmath13 can be factored uniquely as a nonincreasing product @xmath178 of lyndon words .
we can apply this theorem to all prefixes of a @xmath0-automatic sequence .
it is then natural to wonder if a _
single _ automaton can encode _ all _ the lyndon factorizations of _ all _ finite prefixes .
the answer is yes , as the following result shows .
suppose @xmath40 is a @xmath0-automatic sequence .
then there is an automaton @xmath164 accepting @xmath179 $ is $ w_1 w_2 \cdots w_m$ } \\ & \quad\quad \text{with } \
w_m = { \bf x}[i .. n-1 ] \rbrace.\end{aligned}\ ] ] as is well - known @xcite , if @xmath178 is the lyndon factorization of @xmath13 , then @xmath124 is the lexicographically least suffix of @xmath13 .
so to accept @xmath180 we find @xmath77 such that @xmath181 < { \bf x}[j .. n-1]$ ] for @xmath182 and @xmath183 . given @xmath164 , we can find the complete factorization of any prefix @xmath184 $ ] by using this automaton to find the appropriate @xmath77 ( as described in ) and then replacing @xmath1 with @xmath77 .
we carried out this construction for the thue - morse sequence , and the result is shown below in figure [ fig1 ] .
# 1#20.3#1 [ fig1 ] in a similar manner , there is an automaton that encodes the factorization of _ every _ factor of a @xmath0-automatic sequence : suppose @xmath40 is a @xmath0-automatic sequence .
then there is an automaton @xmath174 accepting @xmath185 $ is $ w_1 w_2 \cdots w_m$ } \\ & \quad\quad \text{with $ w_m = { \bf x}[l .. n-1]$ } \rbrace.\end{aligned}\ ] ] we calculated @xmath174 for the thue - morse sequence using our method .
it is a 34-state machine and is displayed in figure [ fig2 ] . #
1#20.13#1 [ fig2 ] another quantity of interest is the number of terms in the lyndon factorization of each prefix .
let @xmath13 be a @xmath0-automatic sequence .
then the sequence @xmath186 defined by @xmath187$}\ ] ] is @xmath0-regular .
we construct an automaton to accept @xmath188 for the thue - morse sequence the corresponding sequence satisfies the relations @xmath189 for @xmath105 , which allows efficient calculation of this quantity .
a recurrent infinite word @xmath190 , where each @xmath191 is a letter , is called _ linearly recurrent with constant _
@xmath192 if , for every factor @xmath110 and its two consecutive occurrences beginning at positions @xmath77 and @xmath95 in @xmath57 with @xmath193 , we have @xmath194 . the word @xmath195 is called a _ return word _ of @xmath110 .
thus linear recurrence can be defined from the condition that every return word @xmath2 of every factor @xmath110 of @xmath57 satisfy @xmath196 .
let @xmath197 denote the set of return words of @xmath110 in @xmath57 .
[ su 26 - 08 - 2012 00:34 ] linear recurrence implies that every length-@xmath0 factor appears at least once in every factor of length @xmath198 .
[ su 26 - 08 - 2012 00:38 ] let @xmath57 be an aperiodic linearly recurrent word with constant @xmath199 .
a. if @xmath110 is a factor of @xmath57 and @xmath2 its return word , then @xmath200 .
b. the number of return words of any given factor @xmath110 of @xmath57 is @xmath201 .
[ su 26 - 08 - 2012 00:48 ] the lyndon complexity of any linearly recurrent sequence is bounded .
[ lyndon - bounded ] let @xmath57 be a linearly recurrent sequence with constant @xmath199 .
if @xmath57 is ultimately periodic , it is purely periodic because it is recurrent , and thus its lyndon complexity is bounded .
therefore assume that @xmath57 is aperiodic , and let @xmath202 .
denote @xmath203 , so that @xmath204 the left - hand side inequality in and remark [ su 26 - 08 - 2012 00:34 ] together imply that all factors in @xmath57 of length @xmath0 occur in all factors of length @xmath1 . therefore if @xmath110 is the lexicographically smallest factor of length @xmath0 , then every lyndon factor of @xmath57 of length @xmath1 must begin with @xmath110 .
since every suffix of @xmath57 that begins with @xmath110 can be factorized over @xmath197 , we conclude further that every length-@xmath1 lyndon factor of @xmath57 is a prefix of a word in @xmath205 .
the return words of @xmath110 have length at least @xmath206 by lemma [ su 26 - 08 - 2012 00:38 ] .
furthermore , the right - hand side inequality in gives @xmath207 therefore any lyndon factor of length @xmath1 is a prefix of a word in @xmath208 .
since @xmath209 by lemma [ su 26 - 08 - 2012 00:38 ] , we conclude that @xmath210 so that the lyndon complexity of @xmath57 is bounded .
let @xmath211 be a primitive morphism , and let @xmath212 be a letter - to - letter morphism .
if @xmath213 is prolongable , so that the limit @xmath214 exists for some letter @xmath215 , then the sequence @xmath216 is called _ primitive morphic_. [ la 25 - 08 - 2012 23:36 ] primitive morphic sequences are linearly recurrent .
[ to 04 - 10 - 2012 11:42 ] the lyndon complexity of any primitive morphic sequence is bounded
. follows from lemma [ la 25 - 08 - 2012 23:36 ] and theorem [ su 26 - 08 - 2012 00:48 ] .
if @xmath57 is @xmath0-automatic and primitive morphic , then its lyndon complexity is @xmath0-automatic
. follows from corollary [ to 04 - 10 - 2012 11:42 ] and theorem [ primlyn ] , because a @xmath0-regular sequence over a finite alphabet is @xmath0-automatic .
we thank luke schaeffer for suggesting the argument in the proof of theorem [ prim - ex ] .
e. charlier , n. rampersad , and j. shallit .
enumeration and decidable properties of automatic sequences . in g.
mauri and a. leporati , eds . , _ developments in language theory 15th international conference , dlt 2011 _ , vol .
6795 , pp .
165179 , springer - verlag , 2011 .
d. go , d. henshall , and j. shallit . automatic theorem - proving in combinatorics on words . in n. moreira and r. reis , eds . , _
ciaa 2012 _ , lect .
notes in comput .
, vol . 7381 , springer , 2012 , pp .
180191 .
g. melanon .
lyndon factorization of infinite words . in c.
puech and r. reischuk , eds .
, _ stacs 96 , 13th annual symposium on theoretical aspects of computer science _ , lect .
notes in comput .
, vol . 1046 , springer , 1996 , pp .
147154 .
e. rowland and j. shallit .
@xmath0-automatic sets of rational numbers . in a. h. dediu and c. martn - vide , eds .
, _ lata 2012 proceedings _ , lect .
notes in comp .
, vol . 7183 , springer , 2012 , pp .
490501 .
j. shallit .
the critical exponent is computable for automatic sequences . in p.
ambroz , s. holub , and z. msakov , eds . , _
proceedings 8th international conference words 2011 _ , vol .
63 of _ elect .
_ , pp . 231239 , 2011 . | we investigate questions related to the presence of primitive words and lyndon words in automatic and linearly recurrent sequences .
we show that the lyndon factorization of a @xmath0-automatic sequence is itself @xmath0-automatic .
we also show that the function counting the number of primitive factors ( resp .
, lyndon factors ) of length @xmath1 in a @xmath0-automatic sequence is @xmath0-regular .
finally , we show that the number of lyndon factors of a linearly recurrent sequence is bounded . |
events containing isolated same - sign dileptons are very rare in the standard model ( sm ) , but can occur quite naturally in many different new physics models , including supersymmetry ( susy ) @xcite and universal extra dimensions @xcite .
isolated same - sign lepton pairs are thus a very clean experimental signature for new physics searches . in addition , the analysis described in this talk requires significant missing transverse energy ( @xmath2 ) and hadronic activity in the form of high-@xmath3 jets .
the choice of signal regions is influenced by two experimental observations .
first , astrophysical evidence for dark matter @xcite suggests the need for a massive , weakly - interacting stable particle which gives rise to final states with @xmath2 .
second , new physics signals with observably large cross sections are likely to be the result of strong interactions , and we therefore expect them to be accompanied by significant hadronic activity . aside from these requirements ,
our searches are as independent of the particular details of new physics models as possible .
the data used in this analysis were collected in @xmath0 collisions at a center - of - mass energy of 7 @xmath1by the compact muon solenoid ( cms ) detector at the large hadron collider ( lhc ) in 2011 .
they comprise a total integrated luminosity of 0.98 .
an example susy cascade decay leading to jets , @xmath4 , and same - sign dileptons is shown in figure [ figure1 ] .
the produced gluinos or squarks decay to charged gauginos , which subsequently decay to the lightest supersymmetric particle ( lsp ) neutralino .
the mass difference between the gluinos / squarks and the charged gaugino , typically arbitrary , defines the amount of hadronic activity expected in the event .
the mass difference between the gaugino and neutralino influences the lepton @xmath5spectrum .
further , there is a range of scenarios where a large production asymmetry exists between the @xmath6 lepton and the @xmath7 leptons .
the range of mass differences motivates a variety of selection criteria in order to cover the widest possible phase space , while the possible tau production asymmetry motivates us to look specifically at events containing taus .
electrons , muons , and hadronically decaying taus are all included in the analysis .
lepton candidates are required to have @xmath8 and to be consistent with originating from the same interaction vertex .
electron candidates @xcite consist of an energy cluster in the electromagnetic calorimeter ( ecal ) matched to hits in the tracker .
the identification of electrons is based on the shape of their electromagnetic shower in the ecal as well as track - cluster matching .
the criteria are designed to maximally reject electron candidates from qcd multijet production , and are approximately 80% efficient for electrons from the decay of @xmath9 bosons .
muon candidates @xcite must be reconstructed via two separate algorithms : tracker muons are seeded from hits in the tracker and are matched to signals in the calorimeters and muon systems .
global muons are constructed from a simultaneous fit to hits in both the tracker and muon chambers .
the identification efficiency measured in data is approximately 96% for muons of all momenta
. jets and @xmath4are reconstructed via the particle flow ( pf ) technique @xcite , which takes signals in each detector component and reconstructs physics objects from them before running jet clustering algorithms .
the hadronic jets in this analysis use the anti-@xmath10clustering algorithm with a cone of @xmath11 in @xmath12 space .
hadronic tau candidates @xcite are reconstructed starting from jets .
a variable size cone of @xmath13 is defined around the leading track , and the @xmath6 decay products are required to be confined within this cone .
the analysis starts from an initial selection of two same - sign leptons with @xmath14 for muons , electrons , and taus , respectively , and @xmath8 .
two jets with @xmath15 and @xmath16 are also required , and we further require @xmath17 and @xmath18 . this initial selection is further divided into three baseline selections : the _ inclusive _ selection for the @xmath19 , @xmath20 and @xmath21 final states requires @xmath22 .
the _ high-@xmath5 _
selection for the @xmath19 , @xmath20 and @xmath21 final states requires @xmath23 .
the tau - specific selection requires at least one of the final state particles is a tau ( @xmath24 , @xmath25 , or @xmath26 ) with @xmath27 and @xmath28 .
we constrain the baseline selection regions with the following search regions : * _ high-@xmath29 , high-@xmath4 _ : @xmath30 and @xmath31 , providing a high expected sensitivity to points in the constrained minimal supersymmetric standard model ( cmssm ) @xcite with low values of @xmath32 . * _
medium-@xmath29 , high-@xmath4 _ : @xmath22 and @xmath31 , targeting models with moderate mass - splittings between squarks / gluinos and gauginos . * _
high-@xmath29 , low-@xmath4 _ : @xmath30 and @xmath33 , providing a high expected sensitivity to cmssm parameter points with high values of @xmath32 . * _
low-@xmath29 , high-@xmath4 _ : @xmath17 and @xmath34 , providing a high expected sensitivity to models predicting low hadronic activity with high @xmath4 .
figure [ fig : figure2 ] shows the events observed in data on the @xmath29-@xmath4plane for each baseline selection categories , with the dashed and dotted lines indicating the various search regions .
backgrounds are divided into categories based on how many prompt leptons they contain .
rare processes like @xmath35 and @xmath36 , @xmath37 , @xmath38 , and @xmath39 can produce actual prompt same - sign leptons .
these are evaluated from monte carlo ( mc ) simulation and found to contribute between 10% and 40% of the total background .
a 50% systematic uncertainty is applied to the value .
processes such as @xmath40 and @xmath41 produce opposite - sign prompt leptons . in cases where the charge of one of the leptons is misreconstructed , these events constitute an additional background to the search .
they are evaluated in data and found to contribute less than 10% to the total background .
events with `` fake '' leptons from jets also contribute . here
, a fake lepton can be a lepton produced in a heavy flavor quark decay as well as a light jet that is misidentified as a lepton .
this is also evaluated in data , and is found to be the dominant effect . in order to evaluate this fake rate , the analysis loosens the lepton selection criteria ( such as lepton isolation requirements ) to define a set of `` fakeable objects '' . the ratio of lepton candidates passing the full selection ( _ tight _ leptons ) to lepton candidates passing this looser selection but failing the full selection ( _ loose _ leptons ) is then determined for qcd multijet events with a jet above a given threshold .
this tight - to - loose ( tl ) ratio is shown for both electrons and muons in figure [ fig : figure10 ] .
a summary of the background predictions as well as the number of observed events is shown for each channel in the three baseline selection regions in figure [ fig : figure3 ] .
for the inclusive and high-@xmath5dilepton selections , two sets of complementary methods are used to measure the backgrounds , as indicated by the two bars associated with each channel .
these methods compare well with each other , providing mutually consistent background predictions .
table [ table1 ] shows the events observed for each selection region , along with the total background predictions .
we see no evidence of an event yield in excess of the background predictions and set 95% cl upper limits on the number of observed events using a hybrid frequentist - bayesian cl method @xcite with nuisance parameters and the signal strengh maximizing the ratio of the signal - with - background to background - only likelihoods .
these limits are indicated in the bottom row of the table . [ cols="^,^,^,^,^,^,^,^,^",options="header " , ] [ table1 ]
it is necessary to convey the information shown in table [ table1 ] in a form that can be used to test a variety of specific physics models .
this is done by using generator - level simulation studies as approximations for the models .
this was shown to be sufficiently precise to reproduce constraints on new physics models that would otherwise require the full cms detector simulation .
we take as a benchmark point the low - mass cmssm parameter point lm6 .
the efficiency of the lepton selection is shown in figure [ fig : figure16 ] as a function of @xmath5 , and the efficiency of the @xmath29and @xmath4selection is shown in figure [ fig : figure17 ] as a function of the reconstructed @xmath29and @xmath4 , respectively .
this efficiency dependence can be parametrized by error functions to determine the overall selection efficiency for new physics signals .
in order to provide a reference for other susy searches , the results are interpreted in the context of the cmssm model .
the observed upper limits on the number of signal events shown in table [ table1 ] for the high-@xmath29 , high-@xmath4search region of the high-@xmath5dilepton baseline selection are compared to the expected number of events in the cmssm model in a plane of ( @xmath42 ) for @xmath43 , @xmath44 , and @xmath45 .
all points on this plane with mean expected values in excess of this limit are interpreted as excluded at the 95% cl .
this excluded region is shown in figure [ figure5 ] .
the shaded region represents the uncertainty on the position of the limit due to uncertainty in the production cross section of cmssm .
a search for new physics was performed in the same - sign dilepton channel , including final states with electrons , muons , and taus .
all major background sources were estimated directly from data .
events with a single fake lepton were found to be the dominant background in all channels of the search with the exception of the @xmath26 channel , where two fake @xmath6 leptons were the primary background contribution .
no evidence was seen for an excess over the background prediction .
we therefore set 95% cl upper limits on the number of signal events within @xmath8 with 0.98 of data .
these limits are reported as an exclusion curve in cmssm parameter space . | the results of searches for supersymmetry in events with two same - sign isolated leptons , hadronic jets , and missing transverse energy in the final state are presented .
the searches use @xmath0 collisions at 7 @xmath1collected in 2011 by the cms experiment . |
we present here an extension of cook s elastic block cipher .
her extension @xmath4 allows to construct an elastic cipher of length from @xmath0 to @xmath1 , where @xmath0 was the message length of the fixed input length block cipher @xmath5 .
she related the security against key recovery attack of the elastic to that of the initial block cipher round function , using it as a black box .
we start from @xmath4 and apply the elastic extension to @xmath4 itself , obtaining @xmath6 and we relate its security against key recovery attacks to that of @xmath4 , and we repeat the process to build iteratively an elastic extension @xmath7 of @xmath5 which takes inputs of length @xmath8 up to @xmath9 .
+ section [ seccook ] gives a brief description of cook s idea reminding which are the key schedule requirements .
section [ seccookext ] describes our extension , showing the algorithm to encrypt a plaintext of length @xmath10 , with @xmath11 .
section [ secsecurity ] is devoted to the security of @xmath7 , first analyzing how the extension influences the diffusion properties of the round function of @xmath5 , and second giving a proof of how the security against key recovery attack of @xmath7 is reduced to that of @xmath5 .
this is made following and extending cook s proof for @xmath4 .
let @xmath12 , @xmath13 . let @xmath14 denote the set of all binary strings with length at least @xmath15 .
the function @xmath16 denotes the length of the string @xmath17 .
a _ message / plaintext space _
@xmath18 is a nonempty subset of @xmath19 for which @xmath20 implies that @xmath21 for all @xmath22 of the same length of @xmath17 .
a _ ciphertext space _ ( or _ range _ ) @xmath23 is a nonempty subset of @xmath19 . from now on , @xmath24 .
a _ key space _
@xmath25 is a nonempty set together with a probability measure on that set .
a _ pseudorandom permutation _ ( prp ) with key space @xmath25 , message space @xmath18 and range @xmath23 is a set of permutations @xmath26 where each @xmath27 .
we usually write @xmath28 .
we assume that @xmath29 depends only on @xmath16 .
a _ variable input length ( vil ) cipher _ is a prp @xmath30 .
a _ block - cipher round _ ( or simply a _ round _ ) is a key - dependent permutation of the message / ciphertext space , precisely @xmath31 . as usual
we can consider @xmath32 to denote the vector space @xmath33 .
the number @xmath15 is called the _ block length_. we consider rounds made of a round function ( which is an invertible boolean function indicated with the letter @xmath34 , and that may be applied to just some of the round input bits , as in des ) , followed by an affine map and a sum with the round key .
block - cipher ( bc ) _ is a cipher @xmath35 obtained from the composition of @xmath36 rounds , whose keys are generated from the cipher key @xmath37 by a key schedule .
we write @xmath38 with @xmath39 and @xmath37 .
call @xmath5 a block cipher with input length @xmath0 , composed by @xmath40 rounds , where the @xmath41th round is indicated by @xmath42 , with @xmath43 .
the _ cycle of a block cipher _ is a boolean function made of the least , over any key , number ( called length ) of consecutive rounds such that each bit of the cycle output is a function of at least two input bits .
we consider only ciphers that can be divided in cycles of the same length .
suppose @xmath5 has @xmath44 cycles , and let @xmath45 be @xmath5 @xmath46th cycle function ( @xmath47 where @xmath48 is the number of rounds per cycle ) .
consider only block ciphers whose round function has a whitening as last operation ( this is very common in real cipher such as aes , serpent des , ... ) .
@xmath5 has @xmath49 rounds .
we can define now @xmath4 , the _ elastic extension _ of @xmath5 [ defelastic ] is given by cook @xcite .
] an _ elastic cipher _
@xmath4 of @xmath5 is a vil cipher with message space @xmath50 with @xmath51 , constructed from @xmath5 as follows : * the number of rounds of @xmath4 is @xmath52 * @xmath4 s @xmath41th round @xmath53 has one @xmath5 s cycle as round function , whose input are the leftmost @xmath0 bits and whose last sum with the key is expanded to the rightmost @xmath54 bit string @xmath55 that is not input into the @xmath5 cycle ; the round function @xmath56 is followed by a swap and an exor : 1 .
call @xmath57 the rightmost @xmath54 bit string which is the bit - wise sum of @xmath55 with @xmath54 key bits . 2 .
choose a @xmath54 bit string @xmath58 from the output of the cycle and let its bits be the rightmost ones of @xmath53 output ; 3 .
replace @xmath58 with @xmath59 * add an initial and final whitening * add an initial and final key - dependent permutation ( as in @xcite ) between the initial and final whitening and the rounds .
we give three requirements for the key schedule taken from @xcite which can be satisfied if we use a pseudorandom generator ( e.g. rc4 as in @xcite ) :
our proposal is to consider an elastic cipher @xmath4 working with fixed length messages ( as it was a block cipher ) , and apply iteratively the elastic extension to it .
we can define a cycle of an elastic cipher exactly as a cycle for a block cipher : the _ cycle @xmath60 of an elastic cipher @xmath4 _ , with @xmath61 , is a boolean function made of the least , over any key , number of consecutive @xmath4 s rounds such that each bit of @xmath60 output is a function of at least two input bits .
now we are ready to construct @xmath6 , @xmath62 , ... , @xmath7 , simply by iterating the previous process .
the first round function of @xmath6 looks like fig.[fig4 ] , whose notation is explained in the following lines : * the subscripts @xmath63 indicate respectively the @xmath0 leftmost , the @xmath0 second leftmost and the @xmath64 rightmost bits of the input @xmath65 or of the key @xmath66 .
* @xmath67 is the round function of @xmath68 used in round @xmath41 of @xmath68 .
* @xmath69 is the cycle function of @xmath68 used as round function in round @xmath41 of @xmath70 .
* @xmath71 is a portion of bits of the key @xmath66 , where @xmath72 $ ] indicates a set of indexes . in fig.[fig4 ] we have : @xmath73 $ ] , @xmath74 $ ] , @xmath75 $ ] .
* @xmath76 is the portion @xmath77 of the input of the @xmath78th round , or equivalently the output of the @xmath41th round .
similarly for @xmath79 and @xmath80 .
( 0,-3.7992187)(10.589063,3.7992187 ) ( 5.51,2.8657813)(3.51,1.6657813 ) ( 7.51,2.2657812)(3.51,1.6657813 ) ( 5.51,1.2657813)(3.51,0.66578126 ) ( 4.51,2.8657813)(4.51,3.2657812 ) ( 6.51,2.2657812)(6.51,3.2657812 ) ( 4.51,1.2657813)(4.51,1.6657813 ) ( 4.51,1.2657813)(6.51,1.6657813 ) ( 6.51,1.2657813)(4.51,1.6657813 ) ( 8.391898,1.7531185)(8.123966,1.7531185)(8.391898,2.9257812)(8.123966,2.8280594 ) ( 8.391898,1.7531185)(7.99,1.6553966)(8.391898,0.4827338)(8.123966,0.58045566 ) ( 1.0245312,-0.32921875)@xmath81 ( 4.474531,2.5707812)@xmath82 ( 4.4645314,0.99078125)@xmath83 ( 6.474531,1.9707812)@xmath84 ( 4.4845314,3.5507812)@xmath85 ( 6.4445314,3.5907812)@xmath86 ( 8.004531,3.5307813)@xmath87 ( 5.51,0.28578126)(3.51,-0.9142187 ) ( 7.51,-0.31421876)(3.51,-0.9142187 ) ( 5.51,-1.3142188)(3.51,-1.9142188 ) ( 4.51,0.28578126)(4.51,0.68578124 ) ( 6.51,-0.31421876)(6.51,1.2657813 ) ( 4.51,-1.3142188)(4.51,-0.9142187 ) ( 4.51,-1.3142188)(6.51,-0.9142187 ) ( 6.51,-1.3142188)(4.51,-0.9142187 ) ( 4.51,-3.3342187)(4.51,-1.9142188 ) ( 6.51,-2.3142188)(6.51,-1.3342187 ) ( 4.474531,-0.00921875)@xmath88 ( 4.4645314,-1.5892187)@xmath83 ( 6.474531,-0.6092188)@xmath89 ( 6.4645314,-2.5892189)@xmath83 ( 8.51,-0.33421874)(7.51,-0.93421876 ) ( 7.974531,-0.62921876)@xmath90 ( 7.99,-0.33421874)(7.99,3.2857811 ) ( 4.4645314,-3.5692186)@xmath91 ( 6.4645314,-3.5692186)@xmath92 ( 7.884531,-3.5692186)@xmath93 ( 6.51,-3.3342187)(6.51,-2.9342186 ) ( 7.91,-1.9342188)(7.91,-0.93421876 ) ( 6.51,-2.3342187)(7.91,-1.9342188 ) ( 7.91,-2.3342187)(6.51,-1.9342188 ) ( 7.91,-3.3342187)(7.91,-2.3342187 ) ( 7.51,2.8657813)(3.51,0.46578124 ) ( 7.51,0.46578124)(3.51,-1.9342188 ) ( 7.11,-2.3342187)(5.91,-2.9342186 ) ( 9.464531,1.7907813)@xmath94 ( 3.1917381,0.8977824)(3.4839127,0.8977824)(3.1917381,-0.9742165)(3.4839127,-1.8182166 ) ( 3.1917381,0.8977824)(3.63,1.0537823)(3.1917381,2.9257812)(3.4839127,2.7697814 ) ( 2.13,-0.39495495)(2.3979259,-0.39495495)(2.13,-3.534096)(2.3979259,-3.272501 ) ( 2.13,-0.39495495)(2.5318887,-0.13335985)(2.13,3.0057812)(2.3979259,2.7441862 ) ( 2.8245313,0.87078124)@xmath95 call @xmath7 an elastic cipher of @xmath5 of _ level _ @xmath15 .
@xmath15 is also called the level of the extension @xmath7 .
@xmath7 is able to encipher blocks from length @xmath8 to @xmath9 .
we ll call @xmath96 and @xmath97 respectively the round and the cycle functions of @xmath7 .
@xmath98 ( and @xmath99 ) indicates respectively the number of round ( cycles ) @xmath7 .
@xmath100 with @xmath101 , indicates the number of rounds of @xmath102 used in @xmath7 . the level increases by 1 when the message length doubles , and in general the level increases by @xmath103 when the length passes from @xmath9 to @xmath104 , which means the level increases logarithmically in respect to the message length .
once @xmath15 is fixed , only @xmath7 takes variable inputs and has initial and final key - dependent permutations , all other @xmath105 , with @xmath106 , take as input messages of fixed length @xmath107 and do not have initial and final key - dependent permutation .
here we give explicitly the steps to encipher a message of any length greater than @xmath0 using an extended elastic cipher of a certain level , obtained from a fil block cipher @xmath5 that works on messages of length @xmath0 .
this algorithms are the definition of @xmath7 .
furthermore we give some theorems and their proofs to understand the formulas used .
we divide the enciphering algorithm in two parts : initialization of @xmath7 and encryption . for brevity , in this article
, we will omit the algorithm for decryption and @xmath108 .
the following algorithm initializes the elastic cipher @xmath7 , given a message @xmath65 of given length .
+ input : @xmath65 .
+ output : the level @xmath15 , the exponent @xmath54 , the number of rounds @xmath109 , the key length , @xmath66 a random vector of length @xmath110 . 1 .
determine the level @xmath15 of the elastic extension as the least power of 2 such that @xmath111 2 .
determine @xmath112 //
@xmath113 is the length function .
determine @xmath114 has the property that it is the least number of @xmath7 rounds such that , in @xmath7 , each input bit influences @xmath5 round function the same number of times it influences the same round function in @xmath5 .
determine the key @xmath66 : determine the key length : //
( see th.[thkeylength ] ) + @xmath115\}\cdot r_n+2^nl+2y+k_{perm}$ ] fill a @xmath110 string with random bits //
e.g. using a pseudorandom generator steps to encrypt a message @xmath65 with @xmath7 initialized .
+ input : @xmath116 .
+ output : @xmath65 encrypted .
1 . @xmath117 //
@xmath103 is a pointer to the first bit not used of the key .
2 . @xmath118}$ ] //initial whitening 3 .
@xmath119 4 .
@xmath120})$ ] //key dependent permutation , @xcite .
@xmath121 // move the pointer @xmath103 .
@xmath122 7 . for @xmath123 1 .
@xmath124}=cyclefunction(p_{[0, ... ,2^{n-1}l-1]},n-1)$ ] // cycle of @xmath125 2 .
@xmath126}=p_{[2^{n-1}l, ... ,2^{n-1}l+y-1]}\oplus k_{[k, ... ,k+y-1]}$ ] //add key 3 .
@xmath127 4 .
@xmath128}=p_{[j, ... ,j+y-1]}$ ] //@xmath55 is a temporary value for the swap steps 5 .
@xmath129}=p_{[j, ... ,j+y-1]}\oplus p_{[2^{n-1}l, ... ,2^{n-1}l+y-1]}$ ] .
@xmath46 fixed would not influence the proof of security . ]
@xmath126}=t_{[0, ... ,y-1]}$ ] // ( d)-(e)-(f ) are the swap / exor steps .
@xmath130 8 .
@xmath120})$ ] 9 .
@xmath121 10 .
@xmath118}$ ] 11 .
return @xmath65 cycle of @xmath102 , used in the encryption function .
+ input : the bit string @xmath17 , the level @xmath132 of the cycle function ( @xmath17 has length @xmath133 ) .
+ output : the @xmath133 bit string @xmath17 processed by @xmath134 . 1 .
if @xmath135 then 1 . for @xmath136 do 1 .
@xmath137 2 .
@xmath138 2 .
return m 2 . if @xmath139 then 1 .
@xmath140}$ ] 2 .
@xmath141}$ ] 3 .
for @xmath142 do 1 .
@xmath143 2 .
@xmath144}$ ] 3 .
@xmath145 4 .
@xmath146 5 .
@xmath147 6 .
@xmath148 4 .
return @xmath149 @xmath98 does not represent how many rounds of level @xmath132 are in @xmath7 , but how many rounds of level @xmath15 are in @xmath7 . to know how many times the cycle function ( or equivalently the round function ) of @xmath102 is used in @xmath7 , then we need the following : [ number of rounds @xmath150 in @xmath7][thnumroundsr0 ] the number of rounds of level @xmath132 in @xmath7 is : @xmath152 in particular , which means polynomially ( in our case linearly ) with the input length @xmath9 . ]
if @xmath135 : @xmath153 the following theorem , beside giving an explicit formula for the key bits needed for encryption shows that cook s third requirement for the key schedule is satisfied because the key expansion rate grows polynomially with the input length .
let s call @xmath154 the function that gives the number of bits needed for the key of the cycle of level @xmath15 , or simply the key length of cycle @xmath97 .
[ thkeylength ] if we consider @xmath7 with initial and final whitening and key - dependent permutation , the number of bits needed as key bits is : @xmath155\}\cdot r_n+2^nl+2y+k_{perm}\ ] ]
in her paper , @xcite , cook shows that if complete diffusion occurs after @xmath3 cycles in @xmath5 ( the fixed length block cipher ) , then it occurs after at most @xmath2 rounds in @xmath4 ( the elastic version of @xmath5 ) .
we extend this proof for @xmath7 .
[ thcompldiff ] if complete diffusion occurs after @xmath3 cycles in @xmath125 ( an elastic cipher working with length message @xmath8 ) , then it occurs after at most @xmath2 rounds in @xmath7(the elastic version of @xmath125 ) . according to @xcite ,
complete diffusion is achieved in @xmath3 rounds _ if every single bit input to a @xmath156-bit block cipher influences the value in all @xmath156 bits after @xmath3 rounds _ , where we say that a @xmath3 rounds input bit @xmath157 influences a @xmath3 rounds output bit @xmath158 if _ changing @xmath157 , while holding all other @xmath159 input bits constant , causes @xmath158 to change with probability @xmath160 , when the probability is taken over all possible values of other input bits and the key bits are held constant_. + first notice that whitening does not impact diffusion , so , for the purposes of our proof , we can view the elastic expansion without them .
+ we can see that each of the first @xmath161 input bits @xmath158 that influences a bit in position @xmath46 of the output of @xmath162 will influence one ( or more ) bit in position @xmath46 of the output of @xmath96 , and if @xmath157 influences a bit exored after the cycle , then @xmath157 will also influence a output bit of @xmath96 in position between @xmath8 and @xmath163 .
+ so , if we have complete diffusion in @xmath125 after @xmath3 rounds , then , in @xmath7 , the first @xmath8 input bits will keep influencing all @xmath164 output bits after @xmath3 rounds .
+ in @xmath7 , the rightmost @xmath54 bits of the input do not influence the first cycle , but because of the swap step they influence the second cycle and since then they will influence all the cycles and will reach complete diffusion after @xmath2 cycles .
we now illustrate how the elastic expansion impacts on the probabilities that a bit of the output of a round function changes , given that one input bit changed while keeping all other input bits fixed , and with constant key bits .
+ consider a black box @xmath165 that works with a @xmath0 bit string ( which represents our cycle function ) .
we call the input bit string @xmath166 and the output bit string @xmath167 , and we denote a single input bit ( in position @xmath41 ) @xmath168 and a single output bit ( in position @xmath46 ) @xmath169 , where @xmath170 .
consider the following experiment @xmath77 : 1 .
initialize four counters @xmath171 .
2 . choose position @xmath172 for an input bit @xmath168 and an output bit @xmath173 .
3 . fix all other input bit to some value .
fix @xmath174 , calculate @xmath175 and take note of the value assumed by @xmath173 .
fix @xmath176 , calculate @xmath175 and take note of the value assumed by @xmath173 . 6 .
if @xmath173 was equal to 0 both in point 4 and 5 then @xmath177 .
+ if @xmath173 was equal to 1 both in point 4 and 5 then @xmath178 .
+ if @xmath173 was equal to 0 in point 4 and to 1 in point 5 then @xmath179 .
+ if @xmath173 was equal to 1 in point 4 and to 0 in point 5 then @xmath180 .
repeat the experiment for all other @xmath181 possible values of the input . denote the number of times we had a change in the bit @xmath173 with @xmath182 , and denote the number of times the value in @xmath46th position remained equal with @xmath183 .
+ at the end of the experiment , changing the bit in position _
i _ of the input produces a change in the bit in position @xmath46 of the output with probability @xmath184 and does not produce a change with probability @xmath185 .
+ the ideal situation for a cycle function is when @xmath186 . +
now consider a second experiment @xmath187 , where we have the same black box @xmath165 as in experiment @xmath77 , and where we add some new steps , which are the elastic steps that influence diffusion , namely the exor and swap steps .
as in fig.[fig10 ] @xmath188 .
( 0,-1.6792188)(9.449062,1.6792188 ) ( 6.57,0.8257812)(3.37,0.22578125 ) ( 4.97,-0.17421874)(4.17,-0.77421874 ) ( 4.97,0.8257812)(4.97,1.2257812 ) ( 4.57,-0.17421874)(4.57,0.22578125 ) ( 4.57,-0.17421874)(6.97,0.22578125 ) ( 6.97,0.22578125)(6.97,1.2257812 ) ( 6.97,-0.17421874)(4.57,0.22578125 ) ( 6.97,-1.1742188)(6.97,-0.17421874 ) ( 4.57,-1.1742188)(4.57,-0.77421874 ) ( 4.9245315,1.4507812)@xmath168 ( 4.5445313,-0.46921876)@xmath83 ( 5.0445314,0.51078125)@xmath165 ( 6.9445314,1.4707812)@xmath189 ( 7.094531,-1.3892188)@xmath190 ( 4.3145313,-1.4492188)@xmath191 now the input bit string has @xmath192 bits .
+ we want to repeat the experiment with this new scheme to see how the elastic expansion acts on the probability of a change in one single output bit given a variation in one single input bit . + first consider the output of @xmath165 .
we observe that repeating the experiment @xmath193 times the output string right after the black box , and @xmath194 will indicate its @xmath46th bit ] will give @xmath195 , where @xmath196 denotes the number of times the @xmath46th output bit passed from @xmath36 to @xmath197 in the experiment @xmath198 , with @xmath199 .
this is because we are actually repeating the experiment @xmath77 @xmath200 times , with the leftmost @xmath54 bits not influencing the function @xmath165 .
this gives that @xmath201 .
+ now consider the output of the entire experiments . because last @xmath54 bits do not influence the black box function , they do not influence the part of the final output of the experiment @xmath187 that does not come from the output of the last exor .
so we are interested in analyzing the bits of the output in those position @xmath46 influenced by the exor step . + through the entire experiment @xmath187 , the bit @xmath202 will be exored half of the times with 0 and half of the times with 1 ( which come from the rightmost @xmath54 bits ) .
this means that if we counted , for instance , @xmath203 times for the @xmath46th output bit of @xmath165 not changing from 0 , after the exor we will count @xmath204 times for the @xmath46th output bit of the exor not changing from 0 . with the same reasoning
we obtain the following scheme : + @xmath205 the probability a bit of the output ( in position @xmath46 ) changed given a change of one bit in the input ( in position @xmath41 ) is : @xmath206 and similarly @xmath207 .
+ this means that @xmath173 ( when @xmath46 is a position included between the positions coming from the exor output ) changes with probability @xmath208 , given a variation of @xmath168 .
+ reasoning as in theor .
[ thcompldiff ] , and with the last results , we can conclude that : [ thidealdiff ] if ideal diffusion occurs after @xmath3 cycles in @xmath125 ( an elastic cipher working with length message @xmath8 ) , then it occurs after at most @xmath2 rounds in @xmath7(the elastic version of @xmath125 ) .
the fact that after the exor each bit changes with probability 50% is a good result but it may be a weakness . because an adversary could distinguish a block cipher from the elastic cipher if the elastic round function @xmath209 does not achieve ideal diffusion after a certain number of rounds ; it would be sufficient to implement the experiment we described for a reasonable amount of input messages and see if there is an output bit sequence that change with probability @xmath210 , while all other output bit change with probability significantly different from @xmath210 .
[ thseckeyrec ] given an elastic cipher , @xmath125 of level @xmath211 ( without initial and final whitening and key - dependent permutation ) , working on @xmath8-bit blocks and its elastic version , @xmath7 , that works on @xmath212-bit blocks , where @xmath213 , if there exists an attack , @xmath214 , on @xmath7 that allows the round keys to be determined for @xmath36 consecutive rounds of @xmath7 using @xmath215 operation , then there exists an attack @xmath216 on @xmath125 with @xmath36 cycles that finds the expanded key for @xmath125 and that uses @xmath217 , assuming there are no message - dependent expanded key , meaning any expanded - key bits utilized in @xmath125 depend only on the key and do not vary across plaintext or ciphertext inputs .
in particular , if @xmath214 is polynomial then @xmath216 is polynomial .
we show how the attack @xmath214 can be converted to an attack to @xmath125 in polynomial time .
+ suppose we want to find a set of cycles keys @xmath218 that enciphered the set of @xmath197 messages @xmath219 in the ciphertexts @xmath220 using the elastic cipher @xmath125 working with blocks of length @xmath8 . + first of all notice that we put an exponent @xmath46 to @xmath25 , because there may be more than one set of cycle keys enciphering the same couples @xmath221 for @xmath222 .
one set is sufficient for us . to find the set of keys @xmath223 build an attack as follows .
+ consider the sets : @xmath224 and @xmath225 , where @xmath226 is the bit string @xmath227 followed by @xmath54 zeros ( we could choose any fixed string of @xmath54 bits instead of all zeros ) , and @xmath228 is the bit string @xmath229 followed by any string of @xmath54 bits .
+ the elements of the sets @xmath230 and @xmath231 fit as input and output for a @xmath36-rounds reduced elastic extension @xmath7 of @xmath125 .
+ so we can apply the attack @xmath214 to the set of pairs @xmath232 and obtain a set @xmath233 of @xmath234 sets @xmath235 @xmath236 of @xmath36 round keys for @xmath7 in polynomial time in @xmath0 and @xmath36 .
view this in fig.[fig12 ] .
( 0,-3.8692188)(5.2490625,3.8692188 ) ( 1.89,3.0157812)(0.49,2.4157813 ) ( 1.89,2.4157813)(0.49,1.8157812 ) ( 1.49,2.4157813)(0.89,1.8157812 ) ( 1.49,1.4157813)(0.89,0.81578124 ) ( 2.49,2.4157813)(1.89,1.8157812 ) ( 2.49,3.0157812)(0.49,0.81578124 ) ( 1.09,3.0157812)(1.09,3.4157813 ) ( 2.09,3.0157812)(2.09,3.4157813 ) ( 1.09,1.4157813)(1.09,1.8157812 ) ( 1.09,0.41578126)(1.09,0.81578124 ) ( 1.09,1.4157813)(2.09,1.8157812 ) ( 2.09,1.4157813)(1.09,1.8157812 ) ( 2.09,0.41578126)(2.09,1.4157813 ) ( 2.49,2.8157814)(2.89,2.8157814 ) ( 2.49,2.2157812)(2.89,2.2157812 ) ( 1.89,-0.7842187)(0.49,-1.3842187 ) ( 1.89,-1.3842187)(0.49,-1.9842187 ) ( 1.49,-1.3842187)(0.89,-1.9842187 ) ( 1.49,-2.3842187)(0.89,-2.9842188 ) ( 2.49,-1.3842187)(1.89,-1.9842187 ) ( 2.49,-0.7842187)(0.49,-2.9842188 ) ( 1.09,-0.7842187)(1.09,-0.18421875 ) ( 2.09,-0.7842187)(2.09,-0.18421875 ) ( 1.09,-2.3842187)(1.09,-1.9842187 ) ( 1.09,-3.3842187)(1.09,-2.9842188 ) ( 1.09,-2.3842187)(2.09,-1.9842187 ) ( 2.09,-2.3842187)(1.09,-1.9842187 ) ( 2.09,-3.3842187)(2.09,-2.3842187 ) ( 2.49,-0.9842188)(2.89,-0.9842188 ) ( 2.49,-1.5842187)(2.89,-1.5842187 ) ( 1.0645312,3.6607811)@xmath237 ( 2.0645313,3.6607811)@xmath238 ( 1.0645312,2.7207813)@xmath239 ( 3.5045313,2.7407813)@xmath240 ( 1.1345313,-1.0792187)@xmath241 ( 3.5045314,-0.99921876)@xmath242 ( 3.3545313,2.2407813)@xmath243 ( 3.3845313,-1.6392188)@xmath244 ( 2.2045312,2.1407812)@xmath245 ( 1.1445312,1.1207813)@xmath83 ( 2.2045313,-1.6792188)@xmath246 ( 1.1245313,-2.6792188)@xmath83 ( 0.9845312,0.28078124)@xmath247 ( 2.0445313,0.22078125)@xmath248 ( 3.0245314,-3.5992188)@xmath249 ( 0.9945313,-3.6392188)@xmath250 from @xmath233 we choose one of the possible set of round keys , namely @xmath251
. we can find @xmath252 in the following way .
the last operation in @xmath239 is a whitening of @xmath8 bits .
call this key bit string @xmath253}^j$ ] , then set : @xmath254}^j=\\ k_{c_{n-1,1}[0,\ldots,2^{n-1}l-1]}^{j ' } \oplus ( 0 \| ... \| 0 \| k_{w_1[2^{n-1}l, ...
,2^{n-1}l+y-1]}^{j ' } \| 0 \| ...
\| 0)\end{gathered}\ ] ] ( 0,-1.12)(15.989062,1.12 ) ( 5.403333,1.12)(1.21,0.52 ) ( 5.403333,0.52)(1.21,-0.08 ) ( 10.63,1.12)(6.41,0.52 ) ( 10.63,0.52)(6.41,-0.08 ) ( 14.31,0.52)(10.59,-0.08 ) ( 10.61,-0.48)(6.41,-1.12 ) ( 8.623333,-0.44)(8.623333,-0.04 ) ( 8.63,-0.48)(11.07,-0.08 ) ( 5.63,0.32)(6.31,0.32 ) ( 5.63,0.52)(6.31,0.52 ) ( 5.63,0.72)(6.31,0.72 ) ( 7.2166667,-0.48)(7.2166667,-1.08 ) ( 9.996667,-0.5)(9.996667,-1.1 ) ( 3.2845314,0.825)@xmath239 ( 3.2945313,0.225)@xmath255}^j$ ] ( 8.574532,0.205)@xmath255}^{j'}$ ] ( 12.394532,0.225)@xmath256}^{j'}$ ] ( 8.544531,0.845)@xmath239 ( 8.544531,-0.775)@xmath83 ( 10.304531,-0.775)@xmath257 ( 6.804531,-0.795)@xmath238 once we got @xmath255}^j$ ] we put the remaining bits of @xmath252 equal to the remaining bits of @xmath240 , obtaining the desired cycle key @xmath252 . with this key
we determine : @xmath258 where @xmath259 is the cycle function obtained using the cycle key @xmath252 .
+ notice that we can sum the last @xmath54 key bits of the whitening step because the message to which they are exored is known all zeros in our case ; this is the reason why we can not do the same thing with the next cycle key but we have to restart a new attack as we are going to explain .
now we repeat the process mounting the attack to a @xmath260 round reduced extension @xmath7 of @xmath125 , using the couples @xmath261 and we find one by one the cycle keys @xmath252 , with @xmath262 .
the attack is done applying @xmath214 @xmath36 times with @xmath197 pairs of ( plaintext , ciphertext ) and running one cycle of @xmath125 for each pair , for a total of @xmath263 application of @xmath162 .
if @xmath214 requires to know the output of each round of the reduced round version of @xmath7 , then @xmath264 rounds of @xmath7 have to be computed . if we call @xmath215 the time to run the attack , then the total number of operation required for the reduction is @xmath265 . in her theorem
, cook also considers the time to check that an expanded key found by @xmath214 adheres to the key schedule of @xmath5 . for us
it is not necessary since the key schedule of @xmath125 is a pseudorandom generator and as such , any bit sequence must fit .
cook , d. : elastic block ciphers .
thesis , columbia university , ( july 2006 ) .
cook , d. , keromytis , a. , yung , m. : elastic block ciphers : the basic design . in asiaccs
07 : proceedings of the 2nd acm symposium on information , computer and communications security , pages 350 - 352 , new york , ny , usa , ( 2007 ) .
cook , d. , keromytis , a. , yung , m. : elastic block ciphers in practice : methods , security and instantiations . ? ? ? .
cook , d. , keromytis , a. , yung , m. : the security of elastic block ciphers against key - recovery attacks . in proceedings of information security ,
lncs 4779 , springer - verlag , ( 2007 ) . | given a block cipher of length @xmath0 cook s elastic cipher allows to encrypt messages of variable length from @xmath0 to @xmath1 . given some conditions on the key schedule , cook s elastic cipher is secure against any key recovery attack if the underlying block cipher is , and it achieves complete diffusion in at most @xmath2 rounds if the underlying block cipher achieves it in @xmath3 rounds .
we extend cook s construction inductively , obtaining an elastic cipher for any message length greater than @xmath0 with the same properties of security as cook s elastic cipher . |
unified models of active galactic nuclei ( agn ) propose that all types of agn are fundamentally the same although the presence of a dusty molecular torus prevents the observation of the broad line region for particular line of sights . in this picture type 1 agn ( e.g. seyfert 1/qso ) are those for which the nucleus can be directly observed , while in type 2 agn ( e.g. seyfert 2 ) it is obscured by the torus ( e.g.lawrence and elvis , 1982 and antonucci , 1993 ) .
strong support for this model has been given by x - ray , near - ir and mid - ir observations showing that type 2 agn are characterized by strong absorption whilst type 1 agn are relatively unabsorbed ( turner 1997 , alonso - herrero , ward , kotilainen , 1997 , maiolino 1998 and clavel 2000 ) .
in this framework the infrared radiation is interpreted in terms of thermal emission from hot dust grains heated by the high energy central source emission ( optical to x - ray continuum , see e.g. granato , danese , franceschini 1997 ) .
barcons ( 1995 hereafter b95 ) investigated the 12@xmath5 m to hard x - ray ( hx , 2 to 10 kev ) flux ratios of a sample of nearby agn and galaxies using _ iras _ and _
heao1 a-2 _ observations .
they found that type 1 agn have larger hx/12@xmath5 m flux ratios than type 2 agn whilst normal galaxies have very weak hx emission .
this implies high column densities ( log(n@xmath10)@xmath1122 @xmath12 ) in the type 2 objects , supporting the basic unified model hypothesis .
the study of b95 was necessarily restricted to nearby objects .
the investigation of this flux ratio to deeper fluxes and higher redshifts will probe the properties of luminous sources such as qsos and broad absorption line qsos ( balqsos ) and provide clues to the constancy of agn activity in the universe .
the _ asca _ and _ bepposax _ x - ray satellites have been successful in determining the hx properties of agn and have resolved @xmath030% of the hx background into discrete sources , mostly type 1 and type 2 agn ( e.g. ueda 1998 , 1999 , cagnoni 1998 , fiore 1999 and akiyama 2000 ) .
the recent launch of the _ chandra _ and _ xmm - newton _ x - ray satellites have now made it possible to probe to fainter fluxes at hx energies .
recent _ chandra _ observations ( e.g. mushotzky 2000 , fiore 2000a , brandt 2000 , giacconi 2000 and hornschemeier 2000 ) have uncovered a number of hx emitting , but optically apparently normal galaxies .
the hx slopes of the sources are sufficiently hard that , when combined with the hx emitting agn sources , they can account for @xmath1175% of the hx background ( e.g. mushotzky 2000 , giacconi 2000 and hornschemeier 2001 ) .
comparisons to deep sub - mm surveys have shown that these hx sources are almost always unassociated with sub - mm sources ( e.g. hornschemeier 2000 ; severgnini 2000 ; fabian 2000 , barger 2000 ) , which instead appear to be star forming systems . due to the expected association between hx and mir emission a closer correlation should be found in the mid - ir band .
we present here _ bepposax _ hard x - ray observations of the s1 region of the european large area _
_ survey ( elais ) at 15@xmath5 m .
elais ( oliver 2000 ) was the largest program undertaken by _ iso _ and , covering an area of @xmath012 sq .
deg , is the largest 15@xmath5 m ( serjeant 2000 ) and 90@xmath5 m ( efstathiou 2000 ) survey before _ sirtf _ and _ first _ come into play . in this paper
we focus on observations in the southern s1 region .
in addition to the _ iso _ observations this region has been covered in the u , b , r and i ( la franca in prep .
, heraudeau in prep . ) optical bands and the 1.4 ghz radio ( gruppioni 1999 ) band . selected objects have been observed spectroscopically in the optical ( la franca in prep .
, gruppioni in prep . ) and photometrically in the near - ir ( heraudeau in prep . ) .
the field of view of the _ bepposax _ mecs ( 25 arcminutes of radius ; boella et al , 1997a , b ) matches well with that of the elais sub - fields , covering a considerably larger area than both _
chandra _ and _ xmm - newton _
( see fig .
it is thus ideal for a shallow x - ray survey of elais fields . throughout this paper @xmath13=50 kms@xmath14 ,
@xmath15=0 and @xmath16=0.5 are used .
the _ bepposax _ observations were taken with both the lecs and mecs instruments .
the lecs observations are not considered here due to their lower sensitivity and much poorer psf below 2 kev .
the mecs observations ( boella 1997b ) , from 1.3 to 10 kev , cover 5 of the 9 sub - fields in the s1 region of elais ( see fig .
the total area covered is @xmath01.7 sq.deg . , although the sensitivity of mecs varies with off - axis distance with the most sensitive observations covering 0.7 sq.deg .
the on - source exposure time for these 5 pointings was on average @xmath036 ksec , corresponding to a flux limit of @xmath010@xmath1 @xmath17 on - axis ( see table 1 ) .
the _ bepposax _ positional accuracy is dependent on many factors ( see fiore 2000b for details ) leading to a 95% error radius of one arcmin at off - axis distances @xmath612 arcmin and 1.5 arcmin at greater distances .
slightly better accuracies have been achieved in those fields where we have been able to identify more than one source at another wavelength and perform a positional shift of the x - ray sources to match the more accurate optical positions ( see table 1 ) .
as discussed in the introduction , the characteristics of the elais selection and the ir counts have been presented by oliver et al .
( 2000 ) and serjeant .
the survey has covered an area of @xmath0 12 deg@xmath18 down to @xmath01 mjy in the 15@xmath5 m lw3 filter ( referred to here as mir , cersarksky et al , 1996 ) . in this paper
we focus on observations in the s1 region centered at @xmath19(2000 ) = 00@xmath20 34@xmath21 44.4@xmath22 , @xmath23(2000 ) = -43@xmath24 28@xmath25 12@xmath26 , which covers an area of about @xmath27 .
about 100 sources of the preliminary analysis of elais ( serjeant 2000 ) down to [email protected] mags have been spectroscopically identified during three spectroscopic identification campaigns in 1998 ( la franca in prep .
, gruppioni in prep ) .
the _ bepposax _ observations were cleaned and linearized using the sdc on - line analysis .
the mecs2 and mecs3 images were co - added together and binned by a factor of 2 .
source detection was performed using ximage with the interactive routine _
sosta_. when measuring the counts of a source a box of 13 pixels ( each pixel corresponds to 8 arcsecs ) was used .
this was found to maximize the signal to noise .
as the psf is much wider than the positional uncertainty , the effect on the flux from poor centroiding with weak sources is negligible .
the net counts were automatically corrected for vignetting , sampling dead time and the point spread function . for each source
the background was measured in large apertures at nearby orthogonal sky positions .
the conversion from counts to fluxes was determined using a power law of @xmath19=0.7 ( @xmath28 ) .
we experimented with a range of column densities ( 20@xmath6log(n@xmath29)@xmath623 ) but found small differences ( @xmath010% ) in the conversion factors as compared to the uncertainty in the counts of the x - ray sources and the flux uncertainties of the mir sources , which are of the order of 30% ( serjeant 2000 ) .
the overall conclusions are not affected by these uncertainties .
we initially found 13 sources at the snr@xmath113 poisson level in either the 1.3 to 10 kev ( t ) or 2 to 10 kev ( hx ) band although 2 of these sources could be unreliable because of possible source confusion ( sources s2 and s4 ) , and are included in a supplementary table ( see tables 2 and 3 and fig .
1 ) . based on the source box size
we would expect @xmath61 spurious source due to background fluctuations . considering the relatively small cosmic volume sampled ,
the snr@xmath30@xmath113 number counts of sources are reasonably consistent with those found by giommi , perri and fiore ( 2000 ) , see fig .
we have added to the supplementary list , table 3 , 2 sources just below the detection threshold , with snr@xmath31=2.9 , but which are close ( @xmath060 arcsec ) to the position of an elais 15@xmath5 m qso .
these sources are used in all the further analysis .
the softness ratios of the sources were determined following the procedure of fiore ( 2000b ) with the ratio defined as ( s - h)/(s+h ) where s = count rate in 1.3 to 4.5 kev band and h = count rate in 4.5 to 10 kev band , see fig .
3 and tables 2 and 3 .
the distribution of ratios is consistent with that found in the _ bepposax _ hellas survey ( i.e. @xmath321@xmath6(s - h)/(s+h)@xmath61 , fiore 2000b ) .
we have positionally correlated the hx sources to the mir sources finding 6 matches ( table 4 , fig .
one further match is made with a 2.6@xmath8 _ isocam _ source .
this source is included in the further analysis although it should be noted that this is not a significant _ isocam _ detection .
the separation between the x - ray and optical centroids are @xmath6 60.5 arcsecs in all cases .
the optical counterparts of the _ isocam _ sources have magnitudes of r@xmath620 mags .
this optical magnitude limit is consistent with that found for @xmath3310@xmath1 @xmath34 , the depth of this survey , in other hx surveys ( e.g. schmidt et al , 1998 , akiyama et al , 2000 and hornschemeier et al , 2001 ) .
the surface density of the _ isocam _ sources is @xmath0150 sources / deg@xmath18 , and therefore the chance of a mis - association with a _
bepposax _ error box ( 60 arcsec radius ) is 13% . in previous _
bepposax _ and _ asca _ surveys ( e.g. akiyama , 2000 and la franca 2000 ) sources have been matched to optical counterparts . in these studies
the high optical source density often results in @xmath111 possible optical counterpart within the large _ bepposax _ and _ asca _ error boxes .
the relatively low source density of mir in this study results in less ambigious source matching although we are implictly assuming that an _ isocam _ source within a _ bepposax _ error box is associated with the hx source .
this association is expected within the unified model of agn ( antonucci , 1993 ) and is found from both observational and theoretical tests of this model ( e.g. b95 , granato , danese and franceschini , 1997 , krabbe , boker and maiolino , 2000 and alonso - herrero , 2001 ) .
these x - ray and infrared observations are two orders of magnitude deeper than those of b95 ( see fig .
we measured 3@xmath8 x - ray upper limits for all the spectroscopically classified mir sources , labeled in fig 1 .
this was performed by measuring the counts at an mir source position in the same fashion as for the detected sources .
the measured upper limits depend on the off - axis distance and range from 10@xmath1 @xmath17 on - axis to 3 - 5x10@xmath1 @xmath17 off - axis .
in this discussion we comment on the hx - mir source identifications and compare the observed hx - mir properties of the _ bepposax _ and _ isocam _ objects to those of other samples previously reported in the literature ( e.g. b95 and elvis 1994 , hereafter e94 ) .
as already discussed we found 7 mir sources ( including the 2.6@xmath8 _ isocam _ source ) within the 16 error - boxes ( @xmath060 arcsec , 95% significance , see section 2 ) of the _ bepposax _ detections .
we have obtained optical spectroscopy for 6 of these sources ( la franca in prep . and
gruppioni in prep . ) , see table 4 .
as the optical spectroscopic campaign targetted those sources with [email protected] mags , we did not obtain a spectroscopic identification for the brightest source .
we find 4 normal broad emission line qsos , with [email protected] and @xmath36 erg s@xmath14 and 1 broad absorption line qso ( balqso ) at @xmath7=2.2 and @xmath37=6x10@xmath38 erg s@xmath14 .
the 2.6@xmath8 _ isocam _ source is associated with a galaxy at @xmath7=0.325 and has @xmath37@xmath39 erg s@xmath14 if the association is real . based on the distribution of object types found in other _
bepposax _ and _ asca _ surveys ( e.g. akiyama et al , 2000 and la franca 2000 ) we would expect to find two type 2 sources in our study .
we find none which seems surprising but is not significant ( 1@xmath8 ) given the small number of sources .
one of the two low redshift qsos ( @xmath401 ) is radio detected ( source 3 , @xmath7=0.559 , @xmath41 = 1.5 mjy , gruppioni 1999 ) .
balqsos are rarely detected in the x - ray , particularly in the soft band , possibly due to large absorption of the x - ray emission ( gallagher 1999 ) .
this is the first time a high redshift balqso has been detected in both the mir and hx .
the unclassified source 8 is a bright optical source with an extended optical profile and is possibly detected at 90@xmath5 m ( @xmath42 = 81 mjy , efsthathiou 2000 ) .
its soft hx ratio ( section 4.2 ) is consistent with either an agn or a thermal bremsstrahlung emitting source .
its high hx / mir flux ratio ( section 4.3 ) suggests the former and it is probably a nearby seyfert galaxy .
the detection of a normal galaxy seems surprising although many apparently normal galaxies have been detected in recent _ chandra _ surveys ( e.g. mushotzky 2000 , fiore 2000a , brandt 2000 , giacconi 2000 and hornschemeier 2000 ) although none have such a high x - ray luminosity and many tend to have harder x - ray spectral slopes .
this object has an hx / mir flux ratio , x - ray spectral slope and luminosity similar to qsos although lacks the typical qso broad optical emission lines .
other interesting possibilities are that it is a bl lac object , although the lack of radio emission and a strong calcium break suggest against this , or a dominant cd cluster galaxy with the x - ray emission coming from the cluster medium .
the expected log(hx / mir ) ratio for this latter possibility is @[email protected] ( e.g. edge and stewart , 1991 and bregman , mcnamara and oconnell , 1990 ) . due to the uncertain nature of this source
these possibilities are not further pursued .
we have compared the softness ratios ( see fiore 2000b ) to those expected for various hx seds , see fig .
. the softness ratios of the qsos , although extremely uncertain , show a variety of soft and hard sources 0.5@xmath6(s - h)/(s+h)@xmath6 - 0.5 , similar to those found in the _ bepposax _ hellas survey ( fiore 2000b ) . we have compared the observed hx / mir flux ratios to those expected for a variety of object types
b95 statistically studied the correlation between hx and 12@xmath5 m for the _ iras _
12@xmath5 m sample of agn and emission line galaxies ( rush , malkan , spinoglio , 1993 , hereafter rms ) .
the 12@xmath5 m sample represents the properties of mir sources in the local universe whilst the redshifts of these _ isocam_/_bepposax _ objects are substantially higher . therefore in order to compare this study to that of b95 it is necessary to take into account the effects of redshift and the construction of spectral energy distributions ( seds ) are necessary . in the generation of seds
we have decided to take an empirical approach due to the uncertain contribution of agn and galactic activity at mid - ir wavelengths ( e.g. see the detailed modelling of cen a which is clearly an agn although its infrared emission appears to be dominated by galactic processes , alexander , 1999 ) . to do this
we firstly evaluated and constructed hx and infrared seds of qsos , seyfert 2 and hii galaxies and then normalized the x - ray sed to reproduce the hx / mir flux ratios of the local value at @xmath440 as determined by e94 for qsos , and by b95 for seyfert 2 and hii galaxies .
only a few seyfert 2s were detected at hx energies by b95 although the hx / mir flux ratio distribution is very similar to the _ iras _
60@xmath5 m selected , hx detected 16 seyfert 2 galaxies in alexander ( 2001 ) ( log(hx / mir)[email protected] ) .
the b95 hx / mir flux ratio for the hii galaxies is an upper limit .
the hx seds were taken from pompilio , la franca and matt ( 2000 ) for the cases of qso and seyfert 2 ; the qso sed is a two part power law whilst the seyfert 2 sed uses this same spectrum convolved with the seyfert 2 absorbing column density distribution found by maiolino ( 1998 ) .
the seyfert 2 galaxies in the maiolino study were not mir selected .
however , the column density distribution is very similar to that found for far - ir selected seyfert 2 galaxies ( i.e. log(@xmath45)@xmath022 to 25 @xmath12 , alexander , 2001 ) .
the hii galaxy hx sed was determined assuming a bremsstrahlung thermal spectrum with kt=5.8 kev , as found for the starburst galaxy ngc253 ( cappi 1999 ) .
the seyfert 2 and hii galaxy infrared seds were determined using the xu ( 1998 ) empirical algorithm which takes the four _ iras _ band fluxes to predict an overall 2 to 120@xmath5 m sed assuming three basic components : agn , starburst and cirrus .
the @xmath612@xmath5 m emission is predicted from the _ iras _ colours and produced using agn , starburst and cirrus observational templates which include pah emission and dust absorption features .
the sample used to determine these seds was the rms sample as classified by alexander and aussel ( 2000 ) .
these seds do not account for starlight , which can potentially be a large contributor for @xmath46@xmath67@xmath5 m and would lead to underestimates of the hx / mir ratios .
therefore we have only used these seds for z@xmath61 , which is sufficient for the non - qso sources . for the qsos
we have used the empirically determined sed of e94 which includes the contribution from starlight and gives a good match to the low redshift optically selected pg qsos ( schmidt and green , 1983 ) . as this sed
is derived mostly from low redshift sources , we are assuming little sed evolution for z@xmath63 ( see section 4.3.1 for discussion ) .
the hx / mir flux ratios of the _ bepposax_/_iso _ sample are shown in fig .
the errorbars associated with the seds take into account the statistical spread in the hx / mir ratios measured locally by b95 . in the case of seyfert 2s and
qsos the effect of redshift is to increase the hx flux with respect to the mir flux .
the difference with redshift is not dramatic for either object type because the assumed hx column densities are quite low , although see section 4.3.2 for the case of a compton thick source ( log(n@xmath10)@xmath1124 @xmath12 ) .
the hii galaxies are difficult to detect in the hx at any redshift but particularly at high redshift where the hx k - correction is positive . in the case of a higher temperature of the x - ray emitting gas than that assumed here
, the hx / mir ratio will stay constant to higher redshifts , dropping off significantly at the exponential cut - off energy ( i.e. at @xmath47 , where @xmath48 is the bremsstrahlung temperature in kev ) .
the flux ratio distribution of the hx / mir upper limits are shown in fig .
7b . the mean hx upper limit and log(hx / mir ) are 3.3x10@xmath1 @xmath34 and @xmath325.1 respectively .
to detect all the sources requires significantly deeper x - ray observations .
hx observations of the northern elais fields will be carried out by _ chandra _ and _ xmm - newton _ to limiting fluxes of @xmath03x10@xmath49 and 10@xmath50 @xmath34 respectively .
assuming the mean flux and hx / mir flux ratio of this _ bepposax _ survey , the _ chandra _ and _ xmm - newton _ surveys should reach mean flux ratios of log(hx / mir)@[email protected] and log(hx / mir)@[email protected] although clearly some objects with lower ratios will also be detected .
survey ( two acis - i 16 arcmin pointings ) covers a small area but should detect all the agn and some hii galaxies within the field of view whilst the _ xmm - newton _ survey covers a similar area to this survey and should detect all the qsos , most of the seyfert 2 galaxies and a few hii galaxies . the majority of the hx - mir associated sources are qsos .
the observed qso hx / mir flux ratios appear consistent with that of the median e94 qso sed .
the e94 uvsx sample includes 47 quasars selected to have at least 300 counts in the einstein ipc and with v@xmath617 mags to be observable by iue ; 29 members are radio quiet and 18 radio loud .
the sample is consequently biased toward objects at low redshifts ( although 7 objects have @xmath511 , 4 @xmath512 , 1 @xmath513 ) , of moderate luminosity and with high x - ray to optical flux ratios . in order to increase the statistics of our comparison we added to the mir sample qsos from andreani , franceschini and granato ( 1999 ) with @xmath52200 mjy , and created as a comparison an optically selected sample using 12@xmath5 m detections or upper limits of the pg optically selected qsos from sanders ( 1989 ) . to increase the number of objects at high redshift
we have taken those objects from neugebauer ( 1986 ) with z@xmath112 .
we have determined which of these are balqsos using the list of junkkarinen , hewitt and burbidge ( 1991 ) .
the data are shown in fig .
the mean log(hx / mir ) ratios of the low redshift qsos ( [email protected] ) are @[email protected] and @[email protected] for the mir and optically selected objects respectively .
these flux ratios are consistent with each other and with the e94 sed .
this suggests that mir selected qsos of r@xmath620 mags come from the same population as optically selected qsos and that there is little sed evolution for z@xmath63 . as a further and more definitive test
we have compared the log(b / mir ) ratios of all the mir detected qsos in the s1 region to those of the pg qso sample ( schmidt and green , 1983 and sanders et al , 1989 ) , see fig .
the advantage of this comparison is that we can use the full qso set from both the elais and pg surveys
. the mir qsos are consistent with the flux ratios of the e94 sed and the pg qsos .
this further strengthens the evidence that mir selected qsos of r@xmath620 mags come from the same population as optically selected qsos .
although the statistics are still poor , low redshift balqsos seem to be located in a different region of the hx / mir plane to qsos , see fig .
this is consistent with that expected from an absorbed source .
the low - redshift qso , that lies close to the balqsos in the hx / mir plane shown in fig .
8 , is a narrow line seyfert 1 galaxy ( izw1 ) , an unusually red source ( e.g. see e94 ) which may be different to the general qso population ( e.g. see brandt and gallagher , 2000 ) .
our balqso has a hx / mir flux ratio similar to that found for normal qsos , apparently in contradiction with that found for low redshift balqsos .
this difference could be due to the negative hx k - correction effect of an absorbed hx spectrum , see fig .
we can test whether the observed hx / mir flux ratio is compatible with that expected for an absorbed hx source by constructing a possible balqso sed .
we have chosen to use the compton thick hx spectrum of the type 2 qso object iras09104 + 4109 ( franceschini 2000 ) and the qso infrared sed of e94 , normalised to the hx / mir ratio of iras09104 + 4109 .
the positive effect of redshift on the hx emission can clearly be seen ( see fig .
8) showing that a high redshift hx absorbed source can have a similar hx / mir flux ratio to an unabsorbed source . therefore
, in contrast to low redshift balqsos , high redshift balqsos should be comparatively easy to detect in the hx band , allowing a determination of their true fraction in the high - z qso population .
we have presented shallow _ bepposax _ observations , reaching an on - axis ( @xmath00.7 sq .
deg ) 2 - 10 kev sensitivity of @xmath010@xmath1 @xmath53 , of the southern s1 region of elais , reaching a 15@xmath5 m sensitivity of 1 mjy .
this is the first hx analysis of an _ isocam _ survey .
we have constructed hx and infrared seds to determine the expected flux ratios for qsos and balqsos up to @xmath7=3.2 and for seyfert 2s and normal galaxies up to @xmath7=1.0 .
our main findings are : \(i ) we detect 13 sources with snr@xmath113 in the 1.3 - 10 kev or 2 - 10 kev x - ray bands and a further 2 sources with less reliable detections that have positions close ( @xmath060 arcsec ) to qsos .
the number densities of the snr@xmath33 sources are consistent with the _ asca _ and _ bepposax _ logn - logs function .
\(ii ) 6 of these sources have a reliable _ isocam _ counterpart and one further source has a less reliable ( 2.6@xmath8 ) _ isocam _ counterpart .
we have optical spectroscopic classifications for 6 of these sources finding 4 qsos , 1 balqso at @xmath7=2.2 and 1 apparently normal galaxy ( the 2.6@xmath8 _ isocam _ source ) .
the unclassified object has x - ray and photometric properties consistent with that of a nearby seyfert galaxy .
the galaxy has properties suggesting either an unusual qso or a galaxy cluster .
\(iii ) the qsos cover a wide redshift range ( 0.4@[email protected] ) , and have hx / mir flux ratios consistent with those found for nearby _ iras _ and optically selected pg qsos and the qso sed of elvis et al ( 1994 ) . by further comparing the b / mir flux ratios of the mir qsos to those of the blue band selected pg sample , we suggest that mir selected qsos of r@xmath620 mags come from the same population as optically selected qsos .
\(iv ) the high redshift balqso has a hx / mir ratio similar to that of qsos , but different to that found for low redshift balqsos .
this difference can be explained as the negative k - correction effect of an absorbed x - ray spectrum observed at high redshift .
this suggests that high redshift balqsos should be comparatively easy to detect in the hx band , allowing the true fraction of balqsos in the high redshift qso population to be determined .
we acknowledge the ec tmr network ( fmrx - ct960068 ) for financial support and murst cofin-98 - 032 and asi contracts for partial support .
dma additionally thanks the nsf careers grant ast-998 3783 for post - doctoral support .
this research has made use of saxdas ( sax data analysis system ) linearized and cleaned event files ( rev.2.0 ) produced at the _ bepposax _ science data center and the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with nasa .
we gratefully thank belinda wilkes , the referee , for an efficient and thorough reading of this manuscript .
we further thank cong xu for providing the code to determine the mid - ir seds used in this paper and sarah gallagher , ann hornschemeier and niel brandt for valuable comments on earlier drafts of this paper .
akiyama , m. , 2000 , apj , 532 , 700 alexander , d.m . ,
efstathiou , a. , hough , j.h . ,
aitken , d.k . , lutz , d. , roche , p.f . ,
sturm , e. 1999 , mnras , 310 , 78 alexander , d.m . ,
aussel , h. , 2000 , in _ iso _ surveys of a dusty universe , eds .
lemke , d. , stickel , m. , wilke , k. ( astro - ph/0002200 ) alexander , d.m .
, 2001 , mnras , 320 , l15 alonso - herrero , a. , ward , m.j . ,
kotilainen , j.k
. , 1997 , mnras , 288,977 alonso - herrero , a. , quillen , a.c . , simpson , c. , efstathiou , a. , ward , m.j .
2001 , aj , accepted andreani , p. , franceschini , a. , granato , g.l . , 1999 ,
mnras , 306 , 161 antonucci , r. 1993 , ara&a , 31 , 473 barcons , x. , franceschini , a. , de zotti , g. , danese , l. , miyaji , t. , 1995 , apj , 455 , 480 ( b95 ) barger , a.j . , cowie , l.l . ,
mushotzky , r.f . ,
richards , e.a . , 2000 ,
aj , submitted ( astro - ph/0007175 ) boella , g. , butler , r. c. , perola , g. c. , piro , l. , scarsi , l. , bleeker , j. a. m. , 1997a , a&as , 122 , 299 boella , g. , 1997b , a&a , 122 , 327 brandt , w.n . , gallagher , s.c . , 2000 , joint mpe , aip , eso workshop on nls1s ( astro - ph/0005140 ) brandt , w.n . , 2000 , aj , 119 , 2349 bregman , j.n . , mcnamara , b.r . , oconnell , r.w . , 1990 , apj , 351 , 406 cagnoni , i. , della ceca , r. , maccacaro , t. , 1998 , apj , 493 , 54 cappi , m. , 1999 , a&a , 350 , 777 ceballos , m.t . , barcons , x. , 1996 , mnras , 282 , 493 cesarsky , c.j .
, 1996 , a&a , 315 , l32 clavel , j. , 2000 , a&a , 357 , 839 della ceca , r. , 1990 , apjs , 72 , 471 edge , a.c . , stewart , g.c . , 1991 , mnras , 252 , 428 efstathiou , a. , 2000 , mnras , 319 , 1176 elvis , m. , 1994 , apjs , 95 , 1 ( e94 ) fabian , a.c . , smail , i. , iwasawa , k. , 2000 , mnras , 315 , l8 fiore , f. , la franca , f. , giommi , p. , elvis , m. , matt , g. , comastri , a. , molendi , s. , gioia , i. , 1999 , mnras , 306 , l55 fiore , f. , la franca , f. , vignali , c. , comastri , s. , matt , g. , perola , g.c .
, cappi , m. , elvis , m. , nicastro , f. , 2000a , new astronomy , 5 , 143 fiore , f. , 2000b , mnras , submitted franceschini , a. , bassani , l. , cappi , m. , granato , g.l . ,
malaguti , g. , palazzi , e. , persic , m. , 2000 , a&a , 353 , 910 gallagher , s.c . ,
brandt , w.n .
, sambruna , r.m . ,
mathur , s. , yamasaki , n. , 1999 , apj , 519 , 549 giacconi , r. , 2000 , apj , submitted ( astro - ph/0007240 ) giommi , p. , perri , m. , fiore , f. , 2000 , a&a , 362 , 799 granato , g.l . , danese , l. , franceschini , a. , 1997 , apj , 486 , 147 gruppioni , c. , 1999 , mnras , 305 , 297 hornschemeier , a.e .
, 2000 , apj , 541 , 49 hornschemeier , a.e . , 2001 , apj ( astro - ph/0101494 ) junkkarinen , v. , hewitt , a. , burbidge , g. , 1991 , apjs , 77 , 203 krabbe , a. , boker , t. , maiolino , r. 2000 , apj , submitted ( astro - ph/0010119 ) la franca , f. , 2000 , mnras , submitted lawrence , a. , elvis , m. , 1982 , apj , 256 , 410 lawson , a.j . , turner , m.j.l . , williams , o.r . , stewart , g.c . , saxton , r.d .
1992 , mnras , 259 , 743 lawson , a.j . ,
turner , m.j.l . , 1997 , mnras , 288 , 920 maiolino , r. , et al . 1998 , a&a , 338 , 781 mushotzky , r.f . ,
cowie , l.l . ,
barger , a.j . ,
arnaud , k.a . , 2000 , nature , 404 , 459 neugebauer , g. , soifer , b.t . ,
miley , g.k . ,
clegg , p.e .
, 1986 , apj , 308 , 815 oliver , s. , 2000 , mnras , 316 , 749 pompilio , f. , la franca , f. , matt , g. , 2000 , a&a , 353 , 440 reeves , j.n . ,
turner , m.j.l . ,
ohashi , t. , kii , t. , 1997 , mnras , 292 , 468 rush , b. , malkan , m. , spinoglio , l. , 1993 , apjs , 89 , 1 ( rms ) sambruna , r.m . , barr , p. , giommi , p. , maraschi , l. , tagliaferri , g. , treves , a. , 1994 , apjs , 95 , 371 sanders , d.b . ,
phinney , e.s . , neugebauer , g. , soifer , b.t . , matthews , k. , 1989 , apj , 347 , 29 saxton , r. , 1993 , mnras , 262 , 63 schmidt , m. , green , r.f . , 1983 , apj , 269 , 352 schmidt , m. , , 1998 , a&a , 329 , 495 serjeant , s. , 2000 , mnras , 316 , 768 severgnini , p. , maiolino , r. , salvati , m. , 2000 , a&a , 360 , 457 turner , m.j.l . , 1990 , mnras , 244 , 310 turner , t.j . , pounds , k. , 1989 , mnras , 240 , 833 turner , t.j , george , i.m . , nandra , k. & mushotzky , r.f .
1997 , apj , 488 , 164 ueda , y. , 1998 , nature , 391 , 866 ueda , y. , takahashi , t. , ishisaki , y. , ohashi , t. , makishima , k. , 1999 , apj , 524 , l11 williams , o.r .
, 1992 , apj , 389 , 157 xu , c. , 1998 , apj , 508 , 576 : col .
( 1 ) _ bepposax _ source number ; ( 2 ) _ bepposax _ position ( j2000 ) , see text for positional accuracy ; ( 3 ) log hard x - ray ( 2 - 10 kev ) luminosity ( @xmath54 ) ; ( 4 ) offset from the optical source position ( arcsec ) ; ( 5 ) 15@xmath5 m flux density ( mjy ) from serjeant et al ( 2000 ) , the superscript ( a ) refers to the 2.6@xmath8 _ isocam _ source , see section 3.3 ; ( 6 ) log 15@xmath5 m luminosity ( @xmath54 ) ; ( 7 ) hx / mir flux ratio ; ( 8) optical position ( j2000 ) ; ( 9 ) , ( 10 ) apparent and absolute r band magnitudes ( mags ) from la franca et al ( in prep . ) ; ( 11 ) redshift ; ( 12 ) source classification from la franca et al ( in prep . ) and gruppioni et al ( in prep . ) | we present _ bepposax _ observations of the southern s1 region in the european large area _ iso
_ survey ( elais ) .
these observations cover an area of @xmath01.7 sq .
deg . and
reach an on - axis ( @xmath00.7 sq .
deg ) 2 - 10 kev ( hx ) sensitivity of @xmath010@xmath1 @xmath2 .
this is the first hx analysis of an _ isocam _ survey .
we detect 9 sources with a signal to noise ratio snr@xmath33 , 4 additional sources with a 1.3 to 10 kev ( t ) snr@xmath43 and 2 additional sources which appear to be associated with qsos with [email protected] .
the number densities of the snr@xmath33 sources are consistent with the _ asca _ and _ bepposax _ logn - logs functions .
six _
bepposax _
sources have reliable _ isocam _
15@xmath5 m counterparts within @xmath060 arcsec .
all these _ isocam _ sources have optical counterparts of r@xmath620 mags .
five of these sources have been previously optically classified giving 4 qsos and 1 balqso at @xmath7=2.2 .
the remaining unclassified source has x - ray and photometric properties consistent with that of a nearby seyfert galaxy .
one further hx source has a 2.6@xmath8 _ isocam _ counterpart associated with a galaxy at z=0.325 . if this _
isocam _ source is real , the hx / mir properties suggest either an unusual qso or a cd cluster galaxy .
we have constructed mir and hx spectral energy distributions to compute the expected hx / mir ratios for these classes of objects up to @xmath7=3.2 and assess the hx / mir survey depth .
the balqso has an observed x - ray softness ratio and hx / mir flux ratio similar to qsos but different to those found for low redshift balqsos .
this difference can be explained in terms of absorption , and suggests that high redshift balqsos should be comparatively easy to detect in the hx band , allowing their true fraction in the high redshift qso population to be determined .
the qsos cover a wide redshift range ( [email protected] ) and have hx / mir flux ratios consistent with those found for nearby _ iras _ and optically selected pg qsos .
this suggests that mir selected qsos of r@xmath620 mags come from the same population as optically selected qsos .
we confirm this with a comparison of the b / mir flux ratios of mir and blue band selected qsos . |
the galactic plane survey ( gps , * ? ? ? * ) is part of the ukirt infrared deep sky survey ( ukidss , * ? ? ?
the gps has covered the entire northern and equatorial galactic plane that is accessible to ukirt , surveying 1868 deg@xmath2 and providing @xmath3 photometry for @xmath41 billion stars .
the survey includes two epochs of k band photometry , with the aims of ( 1 ) finding examples of brief and rarely observed phases of stellar evolution via detection of high amplitude variability and ( 2 ) providing a proper motion catalogue .
there has not previously been a panoramic survey for variability in the galactic plane at near - infrared wavelengths , so even a 2 epoch survey represents an opportunity for new discoveries .
this study provides a foretaste of the results that will come from the multi - epoch survey vista variables in the via lactea ( vvv , * ? ? ?
* ) in the southern plane .
high amplitude variability in the near - infrared can be produced by several physical phenomena . evolved giant and supergiant stars such as r cor bor stars , mira variables and oh / ir stars
are known to display large variability at infrared wavelengths ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
active galactic nuclei ( agn ) can show up as point sources in the search through the gps catalogues , and these objects are also variable in the near - infrared @xcite .
other known classes of infrared variable stars include symbiotic stars @xcite , cataclysmic variables / novae @xcite and classical be stars @xcite ( see also * ? ? ?
* for a discussion on classes of near - infrared variable stars ) .
eruptive pre - main - sequence ( pms ) are also among stars where high amplitude variability has been observed .
this type of variable stars suffer episodic outbursts due to an abrupt increase in the mass accretion rate on to the central star , by a factor of up to 1000 .
the sudden rises in luminosity , of up to 6 magnitudes , can persist from months to @xmath4100 yr @xcite
. theoretical models of pms evolution of sun - like stars generally assume that they accrete matter from their circumstellar discs in a continuous fashion ( see discussion of some of the recent models in * ? ? ?
they are assumed to gradually descend hayashi tracks on the hertzsprung - russell ( hr ) diagram until they approach the main sequence .
however , two problems with this simple picture have persisted for @xmath420 years .
the first is that hr diagrams of pms clusters typically show a wide scatter about the best fitting isochrone .
the second is the `` luminosity problem '' ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) which is that the typical luminosities of pms stars in clusters are lower than expected ( @xmath41 l@xmath5 ) for sun - like stars that should be above the main sequence .
if eruptive variability is common amongst pms stars it can resolve both problems : low average luminosity would be due to the great majority of the cycle being spent in the slowly accreting state , whilst the scatter would be due to the lingering effects of the more dramatic accretion events on the location of each star in the hr diagram @xcite .
if correct , this has serious consequences for the body of mass estimates of pms stars in the literature .
@xcite estimate that masses could be overestimated by @xmath6 to @xmath7 for brown dwarfs and low - mass - stars with @xmath8 .
these eruptive ysos have been classically divided into two sub - classes ( see e.g. * ? ? ?
* ) : ( 1 ) fu orionis variables ( fuors ) , which show large increases in flux followed by a slow decay over @xmath910 yrs . while in outburst they present strong co absorption at 2.29 @xmath10 m with h@xmath11o absorption also present in their near - infrared spectrum . ( 2 ) exor variables that have recurrent short - lived ( @xmath121 yr ) outbursts and quiescent periods of 5 - 10 yrs .
the ir spectrum is dominated by strong br@xmath13 in emission , with co bands showing in emission while in outburst and as absorption features while in quiescent states ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
both classes commonly power outflows seen as herbig - haro objects and near ir h@xmath14 emission , although a direct connection between outbursts and the propagation of the jets is unclear .
@xcite show that v2494 cyg would be the first fuor object to show such direct connection .
these ( optically defined ) classifications struggle to include some intermediate objects and exclude younger protostars that have higher accretion rates but are too deeply embedded in circumstellar matter to be observed in the optical .
just three optically invisible protostars have been confidently identified as eruptive variables , with @xmath15 bandpass variability in excess of 2 magnitudes : oo ser @xcite , v2775 ori ( = [ ctf93]216 - 2 ) @xcite and gm cha ( = iso - cha i 192 ) @xcite .
a recent study of protostars in cygnus ob7 found two more candidates that are currently under investigation , @xcite , whilst five more possible outbursting protostars are identified in @xcite . a few others have been proposed as embedded fuor - like objects based only on their spectral characteristics , e.g. pp13s @xcite and ar6a+b @xcite . at present
only @xmath410 fuors and 14 - 18 exors are known @xcite but spectroscopic and h@xmath14 studies of pms stars suggest that eruptive variability may be common or universal ( see e.g * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
most pms stars have been observed to show a low level near - infrared variability due to various physical processes ( such as rotation or hot spots , see e.g. * ? ? ?
* ; * ? ? ?
* ) , but with a mean peak - to - trough variability of 0.15 - 0.17 mag in the k band ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , with variations only very rarely exceeding 1 mag in k. ysos are also known to vary at mid - infrared wavelengths ( see e.g * ? ? ? * ) and the same physical processes are not expected to produce variability larger than @xmath16 mag , whereas variations larger than @xmath17 mag are usually associated with eruptive variability ( see * ? ? ? * and references therein )
. a higher incidence of @xmath15 bandpass variations @xmath91 mag ( @xmath18 ) has recently been reported in class i ysos in the braid nebula within cygnus ob7 @xcite , which might be expected if fluctuations in the accretion rate are higher in these less evolved sources , which have a higher average accretion rate than the class ii and class iii ysos that dominate the other studies .
two sources from @xcite and one source from @xcite were identified as possible eruptive variables , pending spectroscopic follow up .
we note that variable extinction along the line of sight can produce large change in the magnitudes of ysos , such as the ones observed in ux ori - like or aa tau - like variables ( see e.g. , * ? ? ?
* ; * ? ? ?
* ) .
this paper presents initial results of an ongoing project to perform an unbiased search for high amplitude near - infrared variables in the 2-epoch gps data
. the main focus of this study will be the analysis of ysos , specifically likely young eruptive variable stars of the fuor / exor class .
the paper is divided as follows : section 2 describes the methodology of the search for high amplitude variables in the 5th and 7th data releases ( hereafter dr5 and dr7 respectively ) of gps and how these objects represent a very small fraction of the stars in the studied sample of the survey ( @xmath4 1 every 262000 stars ) .
we then go on to describe the locations and properties of the objects found in our searches .
in addition we discuss how ysos are likely the dominant population of high - amplitude ( @xmath19 mag ) infrared variables .
starting from section 3 we focus on the detailed analysis of objects arising from dr5 of gps ( the detailed analysis of dr7 objects will be the subject of a future publication ) . therefore section 3 describes follow up observations of dr5 objects ; in section 4 we detail the results of follow up spectroscopy and photometry , in section 5 we present a brief discussion and section 6 presents a discussion on variable stars not associated with star forming regions .
we searched the gps for high amplitude infrared variables identified via the two epochs of @xmath15 band photometry , which provide a minimium time baseline of 2 years .
we used sql queries of the wfcam science archive ( wsa , * ? ? ? * ) to select stars ( objects with image profile classifier _
mergedclass=-1 _ ) that varied by @xmath20 mag , having @xmath21 mag in at least one epoch and for which no serious post - processing photometric errors are found ( @xmath22 ) .
the magnitude cut was designed to reduce the number of false positives , given that the faintest stars in crowded fields are the most likely to have incorrectly deblended fluxes .
the cut used `` apermag3 '' , the default 2@xmath23 diameter aperture . in order to further decrease the number of contaminant sources we also made additional cuts on the ellipticity of the detections ( @xmath24 ) and in astrometric offsets between the observations (
the detection in both epochs were required to be within @xmath25 from each other ) .
the queries returned a large number of candidates and in order to remove false positives , images of each candidate were inspected visually .
most false positives arose from bad pixels , diffraction spikes from bright stars or a small number of fields with bad astrometry .
two epoch data were only available for a small proportion of the galactic plane at the time of the searches , as indicated in figure [ fig : gps_cov ] .
the figure shows that the area covered is mostly located at galactic latitudes @xmath26 . in other words , two epoch data for the mid - plane were largely excluded from the available data releases .
this is for historical scheduling reasons .
this spatial bias in the initial results reported here will be eliminated in future searches as the dataset becomes more complete .
the fact that the mid - plane is not covered , implies that relatively few star formation regions or young stellar populations with similarly small scale heights will be included in our search . to be more specific ,
most star formation regions at heliocentric distances , @xmath27>@xmath28 kpc will be excluded , assuming a scale height of @xmath29 pc for molecular clouds located near @xmath30=@xmath31 ( e.g. * ? ? ?
in addition , massive sfrs are also excluded ( ob stars scale height of @xmath32 pc , e.g. * ? ? ? * ; * ? ? ? * ) .
table [ table : par ] presents the parameters for the variables selected in the searches to be described below .
column @xmath17 gives the source number sorted by ascending right ascension of the objects .
column @xmath33 presents the original designation given to the sources by the authors .
this designation is maintained throughout the remaining of this work .
column @xmath28 corresponds to the full ukidss gps designation for the source .
coordinates for the objects are given in columns @xmath34 and @xmath35 .
the @xmath15 band magnitude of the source is presented in column @xmath36 , where @xmath37 represents the @xmath15 band epoch with contemporaneous @xmath38 and @xmath39 photometry represents the earlier epoch , but this is not true for all gps sources . ] .
the corresponding @xmath40 and @xmath41 colours are given in columns @xmath42 and @xmath43 respectively .
column @xmath44 gives the @xmath15 band magnitude difference between the 2 gps epochs , @xmath45 .
column @xmath46 presents @xmath47 , the absolute value of the peak - to - trough difference when using all of the available data from this work and the literature .
finally , the number of epochs used in the calculation of the latter , @xmath48 , is given in column @xmath49 , whilst the data release from which the objects arise is given in column @xmath50 .
the initial search was performed in the first tranche of two epoch gps variability data , which was released as part of dr5 , on 28th of august 2009 and covered 31 deg@xmath2 in the first galactic quadrant .
the search yielded 672 candidates .
after inspection of the images to remove false positives , only 17 stars remained , with magnitude variations up to @xmath51 mag ( where @xmath52 is the absolute difference between the two gps k band magnitudes ) .
we conducted a second , similar search for extreme infrared variables in dr7 , which was released on the 10th of september 2010 and covered 155 deg@xmath2 , thus delivering an extra 124 deg@xmath2 of two epoch data ( the dr5 catalogue is embedded within the dr7 release ) .
the query returned , after removal of candidates from dr5 , 3365 candidates .
the removal of false positives by visual inspection of the images yielded 28 real variables .
adding the results from the dr5 and dr7 searches yields a total of 45 high amplitude infrared variables .
we note that the amplitude of the variations are always observed to be much larger than individual errors .
the amplitudes observed at 2.2 @xmath10 m are also larger than the expected difference that arises from comparing different photometric passbands , e.g. ukidss @xmath15 vs 2mass @xmath53 , when ancillary data is available for the stars .
just to emphasize how rare these events are within the surveyed area of the galactic plane , we note that when excluding the @xmath19 mag condition in our searches , @xmath54 million stars are selected from the dr7 catalogue ( which includes the dr5 release ) .
this implies that high amplitude variability occurs in @xmath4 1 out of 262000 stars in gps .
the colours and magnitudes of the 45 stars found in the analysis are shown in figures [ dr57cmd ] and [ nircdr57 ] , and listed in table [ table : par ] .
two of the 17 variables in dr5 show blue colours .
one of them was identified via simbad as nova sct 2003 @xcite corresponding to gpsv17 in table [ table : par ] .
the other 16 stars from dr5 are unknown in the literature , except gpsv13 , which is listed in the @xcite catalogue of h@xmath55 emission line stars . also , some have previous detections in the deep near infrared survey of the southern sky ( denis ) or the 2 micron all sky survey ( 2mass ) .
two stars arising from dr7 , gpsv38 and gpsv45 , present exceptionally red @xmath41 colours and @xmath56 mag .
these correspond to the known oh / ir stars , glmp755 @xcite and iras 18396 - 0807 @xcite .
oh / ir stars typically represent the most extreme oxygen - rich asymptotic giant branch ( agb ) stars , with mass loss rates greater than 10@xmath57 m@xmath5yr@xmath58 .
they are thought to be the extension towards longer periods and higher luminosities of mira variables @xcite , being more evolved and in some cases perhaps in the process of becoming planetary nebulae . given the higher mass loss rate , these objects are deeply embedded , being only detectable towards near - infrared wavelengths . thus these stars will likely be a source of `` contamination '' in our search for eruptive pms variables ( see discussion below ) .
however , these are still interesting objects .
the mid - infrared colours of two sources detected here put them at the red end of the sequence of colours predicted for o - rich agb stars with increasing mass loss @xcite in the work of @xcite , implying that these stars are about to leave the sequence to become planetary nebulae . from the dr5 sample ,
eight of the stars appear very red in colour - magnitude diagrams ( cmd , see left panel of fig .
[ dr57cmd ] ) , i.e. redder than the giant branch plotted for stars in the vicinity .
five of the remaining stars also appear quite red , being projected against the giant branch .
these 13 `` red '' stars have colours indicating a @xmath15 band excess due to hot dust in our colour - colour diagrams , a common property of pre - main - sequence ( pms ) stars with accretion discs .
in fact , most of them are consistent with being reddened t tauri stars ( fig .
[ nircdr57 ] ) .
we note that 4 of these are located in the boundary between reddened t tauri and main sequence stars , however some of their properties favour the t tauri interpretation .
three of them ( gpsv9 , gpsv3 and gpsv15 ) have larger k - band excesses , which could point to the objects being class i ysos .
the results of the dr7 search are similar to those of dr5 , in that most of the variables have redder ( @xmath38-@xmath15 ) colours than normal dwarf branch or giant branch stars in a 6@xmath596 arcmin field centered on the variable objects ( see fig .
[ dr57cmd ] ) .
there are some extreme cases : gpsv28 is extremely red and it may correspond to iras source iras 20226 + 4206 .
gemini nifs spectroscopy also shows that the object has a spatially extended h@xmath11 outflow ( contreras pea et al .
, in prep . ) .
figure [ nircdr57 ] shows the near - infrared colours of the selected candidates .
many of the objects fall in the region dominated by disc emission , i.e. between the reddened sequence of main sequence stars and the t tauri locus of @xcite .
some candidates show a larger excess , which would usually suggest classification as a class i yso , i.e. a disc @xmath60 envelope system . strong evidence to support the yso interpretation is given by the apparent high concentration of our sample towards sfrs .
we performed the search for a possible association of the sources using simbad and the @xcite catalog of sfrs .
the main condition for the stars to be flagged as likely associated with a sfr was that evidence of star formation would show up in a 300@xmath23 radius search , centred on the star , in simbad ( i.e. pms stars , dark clouds , hii regions ) and the @xcite catalog .
four stars failed this main condition but are still flagged as likely associated with sfrs for the following reasons : * gpsv5 and gpsv10 are within the large concentration of high amplitude variables in an area of 1 deg@xmath2 that is coincident with the serpens ob2 association ( see below ) and have colours consistent with a t tauri interpretation ( see fig .
3 ) . in addition
gpsv5 is close to a bright 8 @xmath10 m emission feature in figure [ fig : serp ] .
gpsv10 ( discussed in more detail in section [ rc : sec ] ) also shows an sed that is consistent with a yso classification .
* gpsv26 is part of the sample that is within the cygnus x star forming complex ( see below ) .
its near infrared colours and sed are also consistent with a yso interpretation ( see figure [ app : ysos_seds1 ] in appendix a ) .
* gpsv19 is only 318@xmath23 from sfr g35.20 - 1.75 in the @xcite catalog ( see below ) .
its colour is consistent with a yso classification and it is an unusually red object compared to its local @xmath61 field . the lack of mid - infrared data for the object does not allow us to confirm this interpretation through its sed .
however , the spectrum of the object shows h@xmath11 emission at @xmath62 m , probably associated with a molecular outflow , and br@xmath13 emission , both of which support a yso classification ( contreras pea et al , in prep . ) .
+ using our search criteria , it was found that 12 of the 17 dr5 variables present an apparent spatial association with known star forming regions .
gpsv1 through 11 are located in an area of 1 deg@xmath2 that is coincident with the serpens ob2 association , a loose association of pms stars located @xmath42 kpc away @xcite and which contains several molecular clouds .
the candidates do not appear to be located in the regions of most recent massive star formation that can be identified by the bright 8 @xmath10 m emission in figure [ fig : serp ] .
two objects , gpsv3 and gpsv9 , are actually within the 8 @xmath10 m nebulosity and they have the reddest near - infrared colours of the serpens ob2 objects .
however , they do not seem to be class i objects ( see section [ subsec : ev ] ) .
the variable object gpsv16 is found to be within @xmath63 of the sfr listed as g71.52 - 0.39 @xcite in the @xcite catalog ( fig.[fig : serp ] ) .
we find 17/28 dr7 variables likely associated with a known sfr .
three stars ( gpsv19,gpsv21 and gpsv22 ) are likely associated with sfr g35.20 - 1.75 @xcite , gpsv42 is in sfr g26.35 + 1.83 @xcite and gpsv44 in sfr g28.96 + 3.54 @xcite .
twelve of our candidates are distributed along the cygnus x star forming complex ( fig .
[ fig : serp ] ) , which is known to contain several rich and complex sfrs .
the presence of fuor outbursts would not be new for this region ( see e.g. * ? ? ?
* ; * ? ? ?
we note that 4 other stars fall in the yso region in colour - colour plots but lack a spatial association with known sfrs .
these are gpsv15 , 36 , 40 and 43 .
as discussed in section [ subsec : ev ] , gpsv15 displays several characteristics associated with eruptive ysos , and dr7 objects gpsv36 and 43 show seds similar to ysos ( see appendix [ app : dr7 ] ) .
gpsv40 is a faint object , and very red in its local cmd .
however , mid - infrared data are not available to confirm its yso nature . in order to verify the significance of this spatial association with sfrs
, we performed a monte carlo selection of many samples of 45 real gps sources in the 2 epoch area , using the same cuts as the variable star selection , except for @xmath0 mag .
we find that only 10% of randomly chosen sources would be within 300@xmath23 of a known star forming region , i.e. 4 - 5 stars in our sample would be near sfrs by chance .
however , as noted above , 4 stars in our sample have characteristics of ysos but are not spatially associated with known sfrs .
this indicates that chance associations have little effect on the proportion of variables ( 29/45 ) that would be classified as ysos primarily on the basis of spatial association . in view of this , we believe that the strong concentration of high amplitude variables in sfrs found in our study is sufficient evidence on its own that most of the high amplitude variables are ysos .
confirmation for the majority of the dr5 objects ( based on seds and spectra ) is discussed at length in section [ sec : res ] . supporting evidence for the dr7 objects , in the form of seds ,
is given in appendix [ app : dr7 ] .
spectra for a subset of the dr7 objects will be included in a future paper focussing mainly on the cygnus x region ( contreras pea et al .
, in prep . ) .
preliminary inspection indicates that most of them show emission or absorption features that support a yso classification . despite the strong evidence that the sample is dominated by ysos
, we still need to discuss other classes of high amplitude variable stars that could have the observed red colours and may account for those objects which are not in known sfrs .
the faint k band magnitudes of the stars in our sample suggests that they are too faint to be r cor bor variables or luminous agb stars , such as mira - type variables . to verify this
, we first derive the approximate extinction towards background objects in the local cmd of each of our sources . to do this
, we noted the typical @xmath40 colour of the reddest sources in the local field ( excluding any extreme outliers ) and then calculated an estimate of the maximum extinction that they could have by assuming that they have the intrinsic colours of @xmath640.1 , corresponding to early b type stars .
we then derive the distance at which a typical agb star ( assuming typical @xmath65 mag , @xmath66 mag , * ? ? ?
* ) would have to be located in order to have the apparent magnitude of the individual variable source being considered ( using the local @xmath67 estimated before ) .
in each case a typical agb star would have to be located well beyond the galactic disc edge at galactocentric radius @xmath68 kpc ( see e.g. * ? ? ?
* ) , indicating that there are no typical mira variables in the sample .
agb stars with high mass loss rates , such as oh / ir stars or extreme carbon stars ( see e.g. * ? ? ?
* ) , are deeply embedded in circumstellar material and can suffer high extinction .
these stars may only be observable at near - infrared wavelengths and appear fainter than typical agb stars ( see e.g. , * ? ? ?
. therefore they can show up in our sample and in fact two of them are present in the dr7 sample , as discussed above .
@xcite study a complete sample of nearby dust - obscured agb stars , spread over @xmath69 of the sky , that are undergoing heavy mass loss ( @xmath70yr@xmath58 ) .
they find 30 stars within 1 kpc from the sun ( the completeness limit of the study ) with a scale height of 236 pc .
this implies a space density of 9.35 kpc@xmath71 .
most agb stars are located at @xmath7210 kpc ( ishihara et al.2011 ) . for a typical inner disc sight line at @xmath73
this corresponds to a maximum heliocentric distance , @xmath74 kpc .
( beyond this distance , the galactic latitudes of our survey ( @xmath75 ) would place most sources a few scale heights above the plane , so very few agb stars should be found ) .
an area of 1 deg@xmath76 along this @xmath73 sightline encompasses a conical volume of 1.1 kpc@xmath77 out to d=14 kpc . after allowing for the effect of scale height on the source density
this is equivalent to an effective volume of 0.12 kpc@xmath78 in the mid - plane .
then , 1.12 dust - enshrouded agb stars deg@xmath79 should be contained in this volume .
however , we expect to detect only a small fraction of them in a 2-epoch survey .
in fact , we find that only @xmath80 should be detected , based on a monte carlo analysis using a large sample of randomly generated sinusoidal curves with different amplitudes and periods , similar to the sample of @xcite , when randomly selecting two epochs separated by 3 years .
in addition , only 2/30 stars in the @xcite sample would have similar magnitudes to the brightest stars in our gps sample , if they are located at @xmath7210 kpc ( the remaining stars would almost certainly saturate in ukidss even at these large distances ) .
these two stars correspond to the first and third highest mass losing agb stars in the @xcite sample ( @xmath81yr@xmath58 ) .
agb stars with this extremely high mass loss rate , such as the oh / ir stars in the dr7 sample and the extreme carbon stars in @xcite , show a steep rise of their seds towards the mid - infrared , characterised by a very red @xmath82 $ ] colour , displaying @xmath82>9 $ ] mag . when we apply the expected detection ratio ( 0.27 ) and the fraction of agb stars with high mass loss rates that would show up in our survey ( 2/30 ) we end up with 3.1 dust - enshrouded agb stars in the area covered by our study .
this number is remarkably close to our actual detection of two oh / ir stars in dr7 .
only two objects from the remainder of the sample have @xmath82 $ ] colours similar to extreme agb stars ( gpsv3 and gpsv15 ) but we are confident that they are ysos rather than agb stars , see section [ sec : erup_3_15 ] . despite this , it is reasonable to expect that we will find more very red agb stars in future data releases , some of which may be previously unknown objects .
symbiotic variable stars can be located in the same area as most of our variable stars in the jhk colour - colour diagram ( see figure 2 in * ? ? ?
the d - type symbiotic stars in particular have the fairly red _ ( h@xmath64k ) _ colours found among most of our sample
. however , d - type symbiotic stars have agb star companions , most of which can be expected to saturate in the gps k band data ( see the earlier discussion in this section ) .
@xcite note that the space density of symbiotic stars is very uncertain but it is clear that they are a small galactic population . as an illustration of their rarity , @xcite selected a large number of possible symbiotic stars based on an h@xmath55 and broad bandpass colour selection .
spectroscopic follow up @xcite showed that most of their candidates were in fact t tauri stars , despite their attempt to reduce contamination by avoiding spatially clustered candidates ( symbiotic variables , like agb stars are expected to be found in isolation ) .
it is unlikely therefore that there are many symbiotic stars in our sample but it is conceivable that some of the isolated objects with bluer _ ( h@xmath64k )
_ colours ( after allowing for extinction ) might be s - type symbiotic stars .
active galactic nuclei ( agn ) can also show up in our searches as point sources , with a wide range of near infrared colours .
@xcite finds an expected density of @xmath4 40 quasars down to @xmath83 mag in their study of 12.5 deg@xmath2 of the early data release of the ukidss large area survey ( las , * ? ? ?
* ) , which implies @xmath4 470 quasars in the area covered in our study
. however , the fraction of high - amplitude ( @xmath19 mag ) variable agns appears to be much lower than 1@xmath1 .
for example , none of the 116 quasars in the sample of @xcite displays @xmath19 mag .
@xcite searched for near - infrared counterparts of agns in the ukidss large area survey ( dr6 , covering @xmath4 1850 deg@xmath2 ) and 2mass catalogues .
they identified 1920 agn via @xmath15 band variability between the two epochs ( ukidss and 2mass ) , using several catalogue data quality cuts to remove most erroneous variables .
their sample included 15 candidate high amplitude agns ( @xmath0 mag ) and they provided the coordinates to us at our request .
only 9 of these 15 have a point source profile - classification ( @xmath84 ) in ukidss , and visual inspection reveals that only 6 of these 9 are real .
the approximate completeness limit of their study only reaches to about @xmath85 mag ( although some objects are still detected at fainter magnitudes ) . to compare with our gps sample
, we must first correct for the fact that the @xcite high amplitude variable sample will not include most agn that were fainter at the less sensitive 2mass epoch and then allow for our fainter magnitude limit .
we find a factor of 1.6 for the first correction ( using the proportion of highly variable agn that were fainter in ukidss rather than in 2mass ) and we conservatively estimate a factor of 4 more quasars arises from our @xmath83 mag limit , using table 7 of @xcite . after applying these corrections and scaling to our survey area
we find that we should expect @xmath86 high - amplitude agns in the area covered in dr7 .
we note that the las covers areas of low extinction , so if we allow for the higher extinction in the galactic plane ( typically 0 to 1 mag in the @xmath15 band ) then this estimate should be considered an upper limit .
when we add the results of our searches of dr5 and dr7 , we find that 29 out of 45 variables are likely associated with sfrs , thus representing @xmath466@xmath1 of our sample .
as discussed above , 4 other stars show characteristics of ysos but do not seem to be associated with a known star forming region ( a discussion on this point is presented below ) .
thus , the fraction of variables in sfrs could increase . in view of the spatial bias against star formation regions in the mid - plane in this initial sample ( see section [ sec : search ] )
it is clear that ysos must dominate the galactic population of high amplitude infrared variables with @xmath0 mag , at least in the magnitude range that this search has explored ( @xmath87 to 16 mag ) .
the question then arises as to whether ysos dominate the total population of high amplitude infrared variables .
inspection of the general catalogue of variable stars ( gcvs , * ? ? ?
* ) shows that mira - type variables are the commonest type at bright magnitudes .
these very luminous stars are generally saturated in the gps , except for a tiny minority with exceptionally high extinction and mass loss rate , such as the two oh / ir stars in our dr7 sample .
we note that the high variability oh / ir phase is only expected to last 1700 yr @xcite , thus we do not expect to find many of these type of variables .
the gcvs data indicate that 67% of miras have @xmath88 mag .
( note that the @xmath15 to @xmath89 amplitude ratio has an average value of 0.2 , see @xcite ) .
@xcite calculated a space density of 265 kpc@xmath71 for a complete sample of agb stars within 1 kpc of the sun , with a scale height of 330 pc .
@xcite performed the search for agb stars complete out to 8 kpc and they found a similar space density for sources located at galactocentric radii @xmath90 kpc , after correcting for colour - based incompleteness ( see figure 9 of that work ) .
they also showed that most o - rich agb stars , which dominate the numbers at @xmath91 , are located at galactocentric radii @xmath7210 kpc .
a typical inner disc sight line at @xmath73 with an area of 1 deg@xmath76 encompasses a volume of 1.11 kpc@xmath71 at @xmath7210 kpc . after allowing for the effect of scale height
, the effective volume comes to 0.21 kpc@xmath71 at these distances .
this volume should contain @xmath9250 agb stars deg@xmath79 .
only a small fraction of agb stars are expected to be mira variables , which are at the end of the thermally pulsating phase of their evolution .
@xcite give the fraction as 8% . combining this with the 67% fraction of miras with @xmath88 mag
then implies a source density of @xmath922.7 deg@xmath79 for typical sight lines in quadrants 1 and 4 of the milky way .
this is broadly consistent with an estimate of @xmath93 deg@xmath79 in the galactic plane obtained from inspection of the gcvs : the gcvs only includes mira variables with known periods so we expect it to provide a somewhat smaller space density .
the 29 ysos imply a source density of @xmath94 = 0.19 deg@xmath79 .
however , the true number is higher because : ( 1 ) with only two epochs most high amplitude variables will be missed ; ( 2 ) the source density rises towards our magnitude cut ( see fig . [ histgps ] ) , indicating that most distant , low luminosity pms variables will be undetected ; and ( 3 ) the present dataset excludes the mid - plane and is therefore strongly biased against sfrs ( see section [ sec : search ] ) .
the @xcite catalogue indicates that item ( 3 ) excludes 60% of galactic sfrs ( and the true fraction is undoubtedly higher since the catalogue is incomplete for distant sfrs in the mid - plane ) .
the effect of item ( 2 ) can not be reliably quantified , but it is likely to raise the source density by at least an order of magnitude .
e.g. typical members of our sample with @xmath15 = 14.8 mag , @xmath27=1.4 to 2 kpc ( sources in cygnus x or serpens ob2 ) would be below our @xmath15=16 magnitude cut at @xmath95 kpc .
the overwhelming majority of galactic sfrs lie at greater distances than this .
the study of the spiral structure in the galaxy by @xcite is said to be complete for star forming complexes with excitation parameters , u , brighter than 60 pc cm@xmath79 .
we find that 85% of these are found to be at @xmath97 kpc .
also , most pms stars are less luminous than those in our sample but presumably can still have high amplitude variability , judging by the rising incidence to faint magnitudes .
finally , item ( 1 ) is believed to cause approximately @xmath98 incompleteness : our analysis of high - amplitude variables arising from the 2010 - 2012 vvv data ( with at least 14 epochs of k band observations for each variable ) reveals that only @xmath4 25% of such variables would be detected when comparing the magnitudes of two epochs separated by @xmath4 2 years . adopting factors of 4 , 6.7 and 2.5 for items 1 , 2 and 3 respectively
, our observed surface density of 0.19 deg@xmath79 rises to 12.7 deg@xmath79 .
if we assume that 5 ysos in our sample are actually chance spatial associations of other types of variable ( see @xmath992.2 ) , and also remove the 4 stars that did not pass our 300@xmath23 criterion for sfr association but were still included , the surface density still remains high at 8.7 deg@xmath79 .
this figure is of course highly uncertain , and the figure of 2.7 deg@xmath79 for miras in quadrant 1 sight lines is also quite uncertain ( by perhaps a factor of two given that the space density of agb stars has a @xmath450% uncertainty in ishihara et al.(2011 ) and the error on the 8% fraction of agb stars that are mira variables is not clear in the literature ) .
nonetheless , we conclude that pms high amplitude infrared variables are likely to be the commonest type of high amplitude infrared variable in most parts of the milky way , except in the bulge where agb stars have higher densities .
the true incidence will no doubt be established more precisely in the near future using data from vvv @xcite .
optical surveys have been used effectively to identify low mass pms stars even in regions where molecular gas , a common tracer for sfrs , has dissipated @xcite .
although the near - ir variability of such objects has been less thoroughly studied , previous surveys show that the majority of ysos are variable at these wavelengths due to different physical processes ( see e.g. * ? ? ?
however , peak to peak amplitudes have been found to be low , with a mean of 0.17 mag in the work of e.g. @xcite . only a small fraction show larger amplitudes , e.g. only @xmath100 of stars with discs show @xmath101 mag in the sample of @xcite .
the fraction seems remarkably higher ( @xmath102 ) in the recent time series study of class i protostars @xcite ( see section [ sec : intro ] ) which suggests that infrared variability surveys can most easily pick out the parts of molecular clouds where star formation has occurred very recently .
@xcite also shows that the fraction of high - amplitude variables increases with extending time baselines .
it is possible then that high - amplitude variability in the infrared , studied over a long baseline such as the one in gps , can also effectively trace low mass pms stars and thus evidence of star formation .
this could be supported by the finding of the four objects with yso colours and not associated with known sfrs .
only marginal and inconclusive evidence for any association to star formation could be found for three of these sources .
the projected location of gpsv15 is at the edge of the dark cloud ldn 667 @xcite , gpsv36 is within 2 arcmin of reflection nebula gn 18.08.8 @xcite and gpsv43 is located in a highly reddened field near the edge of the w40 hii region @xcite . a search for low - amplitude near - infrared variability in the gps near these sources does not reveal the presence of a pms association , however a more statistically thorough study is necessary to firmly discard the presence of such an association .
the high variability of the candidates , along with the observed colours and the possible association with sfrs for many of them , makes the stars strong candidates for eruptive variable classification .
this study will focus on the analysis of the eruptive variable candidates arising from dr5 of gps ; the analysis of dr7 objects will be presented on a later paper ( contreras pea et al .
, in prep ) .
@l@l@c@c@c@c@c@c@c@c@c@c@ n@xmath103 & object i d & gps designation & @xmath55 & @xmath104 & @xmath37 & @xmath105 & @xmath106 & @xmath45 & @xmath47 & @xmath48 & dr + & & & ( j2000 ) & ( j2000 ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & & + 1 & gpsv36 & ugps [email protected] & 18:11:40.30 & -15:29:54.27 & 12.20(0.02 ) & 2.56(0.02 ) & 1.05(0.02 ) & 1.43 & 1.76 & 4 & dr7 + 2 & gpsv37 & ugps [email protected] & 18:12:19.43 & -15:03:04.30 & 15.80(0.04 ) & 1.14(0.04 ) & 0.30(0.04 ) & 1.14 & 1.14 & 2 & dr7 + 3 & gpsv1 & ugps [email protected] & 18:14:52.94 & -11:57:48.77 & 15.11(0.02 ) & 2.72(0.04 ) & 1.21(0.03 ) & 3.75 & 3.75 & 3 & dr5 + 4 & gpsv8 & ugps [email protected] & 18:15:56.91 & -11:41:13.82 & 12.79(0.02 ) & 1.94(0.02 ) & 0.75(0.02 ) & -1.07 & 1.17 & 6 & dr5 + 5 & gpsv10 & ugps [email protected] & 18:16:12.83 & -11:38:09.17 & 13.99(0.02 ) & 3.36(0.03 ) & 1.56(0.02 ) & -1.28 & 1.50 & 6 & dr5 + 6 & gpsv7 & ugps [email protected] & 18:17:07.12 & -11:51:42.55 & 14.15(0.02 ) & 2.71(0.02 ) & 1.29(0.02 ) & -1.00 & 1.00 & 4 & dr5 + 7 & gpsv9 & ugps [email protected] & 18:17:17.38 & -11:58:54.30 & 16.43(0.05 ) & @xmath92.30 & @xmath92.56 & -1.16 & 1.16 & 4 & dr5 + 8 & gpsv5 & ugps [email protected] & 18:17:27.11 & -11:35:31.49 & 15.53(0.02 ) & 2.08(0.04 ) & 0.77(0.03 ) & 1.05 & 1.05 & 4 & dr5 + 9 & gpsv4 & ugps [email protected] & 18:17:53.51 & -12:01:16.61 & 15.53(0.02 ) & 2.01(0.04 ) & 0.95(0.04 ) & 1.07 & 1.07 & 4 & dr5 + 10 & gpsv38 & ugps [email protected] & 18:17:58.57 & -09:18:31.40 & 14.38(0.02 ) & @xmath95.43 & @xmath94.62 & -2.68 & 2.68 & 4 & dr7 + 11 & gpsv11 & ugps [email protected] & 18:18:02.14 & -11:54:24.85 & 15.84(0.03 ) & 3.55(0.18 ) & 1.69(0.07 ) & -1.34 & 2.62 & 5 & dr5 + 12 & gpsv6 & ugps [email protected] & 18:18:09.24 & -11:55:19.61 & 16.37(0.05 ) & 1.88(0.08 ) & 0.67(0.07 ) & -1.00 & 1.00 & 4 & dr5 + 13 & gpsv3 & ugps [email protected] & 18:18:44.74 & -11:36:51.92 & 12.47(0.02 ) & @xmath97.00 & 3.56(0.02 ) & 1.28 & 1.72 & 6 & dr5 + 14 & gpsv2 & ugps [email protected] & 18:19:33.64 & -11:28:00.80 & 14.47(0.02 ) & 1.95(0.02 ) & 0.72(0.02 ) & 1.94 & 2.45 & 6 & dr5 + 15 & gpsv39 & ugps [email protected] & 18:21:26.31 & -12:12:24.75 & 15.12(0.02 ) & 1.03(0.02 ) & 0.35(0.03 ) & 1.11 & 1.11 & 2 & dr7 + 16 & gpsv40 & ugps [email protected] & 18:22:06.71 & -12:32:54.95 & 16.67(0.09 ) & 3.39(0.32 ) & 1.78(0.21 ) & -1.10 & 1.10 & 2 & dr7 + 17 & gpsv41 & ugps [email protected] & 18:26:42.18 & -06:27:47.30 & 17.31(0.11 ) & 1.28(0.13 ) & 0.38(0.14 ) & -1.34 & 1.34 & 2 & dr7 + 18 & gpsv42 & ugps [email protected] & 18:31:35.31 & -02:14:24.21 & 13.99(0.02 ) & 3.55(0.02 ) & 1.30(0.02 ) & 1.45 & 1.45 & 3 & dr7 + 19 & gpsv43@xmath107 & ugps [email protected] & 18:32:37.67 & -02:31:02.70 & 14.60(0.02 ) & @xmath92.06 & 1.90(0.02 ) & -1.06 & 1.18 & 4 & dr7 + 20 & gpsv44 & ugps [email protected] & 18:33:21.45 & -04:57:26.06 & 13.46(0.02 ) & 4.66(0.04 ) & 2.18(0.02 ) & 2.30 & 2.30 & 2 & dr7 + 21 & gpsv45 & ugps [email protected] & 18:42:21.94 & -08:04:59.68 & 11.37(0.02 ) & 7.78(0.16 ) & 5.11(0.03 ) & 2.01 & 2.01 & 4 & dr7 + 22 & gpsv18 & ugps [email protected] & 18:48:34.53 & -05:35:18.36 & 15.18(0.02 ) & 1.83(0.02 ) & 0.60(0.02 ) & -1.10 & 1.10 & 2 & dr7 + 23 & gpsv17 & ugps [email protected] & 18:49:37.60 & -09:33:50.94 & 17.77(0.16 ) & 0.36(0.17 ) & 0.10(0.18 ) & -2.35 & 2.35 & 2 & dr5 + 24 & gpsv12 & ugps [email protected] & 18:53:18.82 & -09:43:29.15 & 15.77(0.03 ) & 0.78(0.03 ) & 0.16(0.03 ) & 1.10 & 1.29 & 3 & dr5 + 25 & gpsv19 & ugps j190003.3 + 010528.7 & 19:00:03.34 & 01:05:28.73 & 14.19(0.02 ) & 5.27(0.18 ) & 2.27(0.03 ) & 2.08 & 2.08 & 2 & dr7 + 26 & gpsv20 & ugps [email protected] & 19:00:28.02 & -06:49:18.77 & 13.52(0.02 ) & 0.54(0.02 ) & 0.16(0.02 ) & 1.04 & 1.04 & 4 & dr7 + 27 & gpsv13 & ugps j190122.5 + 115203.2 & 19:01:22.50 & 11:52:03.24 & 12.92(0.02 ) & 0.29(0.02 ) & 0.14(0.02 ) & -1.02 & 1.23 & 4 & dr5 + 28 & gpsv21 & ugps j190146.8 + 011349.3 & 19:01:46.89 & 01:13:49.37 & 17.27(0.09 ) & @xmath92.53 & 1.86(0.18 ) & -1.81 & 1.81 & 2 & dr7 + 29 & gpsv22 & ugps j190147.0.021228.0 & 19:01:47.07 & 01:12:28.08 & 13.40(0.02 ) & 4.89(0.04 ) & 1.86(0.02 ) & 1.72 & 1.72 & 2 & dr7 + 30 & gpsv15 & ugps j190332.1 + 120557.2 & 19:03:32.11 & 12:05:57.25 & 11.21(0.02 ) & 6.23(0.02 ) & 2.93(0.02 ) & 1.03 & 1.38 & 4 & dr5 + 31 & gpsv14 & ugps j190815.0 + 124021.4 & 19:08:15.06 & 12:40:21.44 & 15.98(0.03 ) & 0.88(0.03 ) & 0.25(0.03 ) & 2.08 & 2.08 & 2 & dr5 + 32 & gpsv23 & ugps j191557.0.022906.7 & 19:15:57.05 & 01:29:06.77 & 15.78(0.02 ) & 0.61(0.02 ) & 0.16(0.02 ) & 1.01 & 1.63 & 3 & dr7 + 33 & gpsv16 & ugps j201310.3 + 333128.1 & 20:13:10.31 & 33:31:28.17 & 13.46(0.02 ) & 3.20(0.02 ) & 1.47(0.02 ) & 2.21 & 2.21 & 3 & dr5 + 34 & gpsv24 & ugps j201613.5 + 415436.5 & 20:16:13.50 & 41:54:36.51 & 14.83(0.02 ) & 2.73(0.02 ) & 0.99(0.02 ) & 1.07 & 1.07 & 2 & dr7 + 35 & gpsv25 & ugps j201924.8 + 410503.0 & 20:19:24.83 & 41:05:03.07 & 14.82(0.02 ) & 2.33(0.02 ) & 0.97(0.02 ) & 1.22 & 1.22 & 2 & dr7 + 36 & gpsv26 & ugps j202038.9 + 405606.8 & 20:20:38.91 & 40:56:06.86 & 14.48(0.02 ) & 3.13(0.02 ) & 1.19(0.02 ) & 1.36 & 1.48 & 3 & dr7 + 37 & gpsv27 & ugps j202117.2 + 403825.1 & 20:21:17.28 & 40:38:25.18 & 15.73(0.02 ) & 3.28(0.08 ) & 1.15(0.03 ) & 1.14 & 1.14 & 3 & dr7 + 38 & gpsv28 & ugps j202421.3 + 421605.6 & 20:24:21.37 & 42:16:05.63 & 11.88(0.02 ) & @xmath97.92 & 2.62(0.02 ) & 1.00 & 1.56 & 3 & dr7 + 39 & gpsv29 & ugps j202504.2 + 403332.6 & 20:25:04.23 & 40:33:32.63 & 15.76(0.02 ) & 1.14(0.02 ) & 0.38(0.02 ) & 1.42 & 1.42 & 2 & dr7 + 40 & gpsv30 & ugps j202605.3 + 420932.9 & 20:26:05.36 & 42:09:32.94 & 15.38(0.02 ) & @xmath94.42 & 2.86(0.11 ) & 1.51 & 1.51 & 2 & dr7 + 41 & gpsv31 & ugps j202759.3 + 415255.7 & 20:27:59.34 & 41:52:55.79 & 16.60(0.05 ) & @xmath93.20 & 1.37(0.09 ) & -1.16 & 1.16 & 2 & dr7 + 42 & gpsv32 & ugps j203117.5 + 413242.5 & 20:31:17.58 & 41:32:42.52 & 15.87(0.03 ) & 2.86(0.10 ) & 1.06(0.04 ) & -1.00 & 1.00 & 2 & dr7 + 43 & gpsv33 & ugps j203352.9 + 413844.9 & 20:33:52.90 & 41:38:44.94 & 16.41(0.04 ) & 2.59(0.14 ) & 0.81(0.07 ) & -1.13 & 1.13 & 2 & dr7 + 44 & gpsv34 & ugps j203427.1 + 421353.3 & 20:34:27.13 & 42:13:53.31 & 11.48(0.02 ) & 4.58(0.02 ) & 1.82(0.02 ) & 1.74 & 2.11 & 3 & dr7 + 45 & gpsv35 & ugps j204142.0 + 393852.0 & 20:41:42.01 & 39:38:52.03 & 16.84(0.05 ) & @xmath92.96 & @xmath92.16 & -1.79 & 2.79 & 3 & dr7 + + band magnitude difference from gps ( _ top _ ) and when taking into account all of the available data ( _ middle _ ) .
the middle graph shows an increase of the observed amplitude of the variability when more epochs over a extended baseline are available .
_ bottom _ mean gps @xmath15 magnitude distribution for the 45 objects in our sample.,title="fig : " ] + band magnitude difference from gps ( _ top _ ) and when taking into account all of the available data ( _ middle _ ) .
the middle graph shows an increase of the observed amplitude of the variability when more epochs over a extended baseline are available .
_ bottom _ mean gps @xmath15 magnitude distribution for the 45 objects in our sample.,title="fig : " ] spectroscopic and photometric follow - up observations of 15 of the dr5 candidates were obtained in visitor mode during two consecutive nights in june 30th / july 1st 2010 .
the data were acquired at the very large telescope ( vlt ) in cerro paranal , chile with the isaac imager and spectrograph .
observations were carried out with the short - wavelength ( 1 - 2.5 @xmath10 m ) arm of isaac , equipped with a hawaii rockwell 1024@xmath591024 array the imaging mode has a scale of 0.147@xmath23/px with a field of view of 152@xmath108152@xmath23 .
low - resolution ( 0.8@xmath23slit , r=@xmath109 700 ) @xmath53 spectra were obtained in two different positions along the direction of the slit ( positions a and b ) , in an abba sequence with individual integration times of 120 s. in order to remove telluric lines , main sequence f - type stars were observed consecutively at similar airmass with the same instrumental setup .
filament lamp images were taken for purposes of flat - fielding of the data , while xenon - argon lamp images were acquired for wavelength calibration and to model the slit curvature .
in addition a set of 12 startrace images were downloaded from the eso archive to correct for the tilt of isaac spectra .
flat - fielding , wavelength calibration , tilt correction and extraction of the spectrum for each object and the corresponding f - type calibrator were performed with the usual tasks in the noao / twodspec package in iraf .
finally , the calibrator spectrum was divided by a blackbody curve of similar temperature to that of the star and used to correct the target spectrum .
photometric observations for each target consisted of 12 images in @xmath110 , @xmath39 and @xmath53 with typical exposure times of 5 s. each image was observed in a different position within a 15@xmath23 wide jitter box .
the seeing at @xmath53 varied between 0.65 - 1.1@xmath23 for the first night and 1.0 - 2.04@xmath23 for the second night of observations .
dark images and sky - flats were obtained for bias subtraction and to correct for pixel - to - pixel variations in the detector response .
the latter consisted of @xmath4 10 - 15 images in filters @xmath110 and @xmath39 with exposure times of @xmath35 s. @xmath53 images were not taken for this project so suitable images were acquired from the eso archive , taking care to select images that were observed with the same instrumental setup , exposure times and a similar observing date .
source magnitudes were obtained through aperture photometry , performed on final sky - subtracted images , using apphot in iraf .
the calibration of the final magnitudes was obtained with the process described below .
firstly , a set of 180 local standards from the iraf aperture photometry results were selected , corresponding to bright stars , with low photometric errors , no processing errors and with no close companions .
cross - match with
the ukidss gps catalog of the same area covered by isaac , yielded typically @xmath4 60 - 100 matches . by taking the average ( after removal of outlying sources ) of @xmath111
, we obtain the offset that will be applied to the isaac instrumental magnitudes .
this process is sufficiently accurate for both @xmath110 and @xmath39 , given the similar @xmath112 and @xmath113 for the ukidss and isaac filters , however , @xmath114 differs from @xmath115 , being similar to 2mass @xmath116 filter profile . to study the differences between ukidss and 2mass @xmath15 , we cross - matched our local standards with the 2mass catalog for the same area , finding around 20 - 40 stars .
the differences between ukidss and 2mass are plotted in fig .
[ uk2 m ] , here it can be seen that magnitudes are similar in both systems , especially for brighter sources .
thus the selection of either system for the calibration would not have a severe effect on the calibrated magnitudes .
this was found to be true for each of our objects . given the greater number of ukidss stars found , and the greater precision of ukidss magnitudes for the fainter stars , we decide to use ukidss photometry for the @xmath53 calibration .
a magnitude correction , however , needs to be applied for our sample , especially the case of the reddest stars where most of its flux falls in the @xmath15-band .
using the reduced spectrum of the object , a correction factor was derived by comparing the ratio of the total flux of the star in ukidss and isaac @xmath15 filters , to the same ratio for a blackbody spectrum at t = 9700 k ( which is appropriate for these vega - based magnitudes ) .
the correction factor is then given by @xmath117 this correction was largest for the reddest stars in our sample but never exceeded 0.05 mag . follow up of the two reddest objects in the dr5 sample , gpsv3 and gpsv15 , was performed on may 8th , 2012 with the fire spectrograph mounted on the magellan baade telescope at las campanas observatory , chile .
the observations were carried out in the high - throughput prism mode which provides a continuous coverage from @xmath118 at low resolution of @xmath119 with a 0.6@xmath23 slit .
the objects were observed in the usual abba pattern along the slit , with individual exposures of 148 s and at airmasses of 1.1 and 1.4 for gpsv3 and gpsv15 respectively .
observations of spectroscopic calibrators were carried out for purposes of telluric correction and flux calibration ; the calibrators were observed at similar airmass as the objects and with the same instrumental setup .
quartz lamp images were obtained for flat fielding of the data , they consisted of high - voltage ( 2.2v ) images that provide data for the z / j bands but saturate the red end of the spectrum , and low - voltage(1.1v ) that generate counts in h / k but are too faint for the blue end .
additionally , nenear arc lamp images were acquired for wavelength calibration .
the images were reduced using the longslit package of the firehose software . using the nenear lamp image
, firehose generates an arc solution , which had typical uncertainties of 0.4 pixels or @xmath4 2.7 @xmath120 .
the second step consisted of creating a flat image ; this is done by combining the 2.2v ( blue ) and 1.1v ( red ) with a smooth weighting function spread over @xmath4 150 pixels centered on a transition pixel defined by the user . finally the software traces and extracts the spectrum of the object .
telluric correction and flux calibration were performed in the standard mode using noao / twodspec package in iraf , in a similar process to that described in section [ isaac : dr ] . unfortunately , the selected exposure times saturated part of the k - band data of both objects , more noticeably in gpsv15 .
however , the data still provide useful information for our analysis , particularly at the shorter wavelengths not covered by the isaac or nifs spectra .
additional k - band spectroscopic data of four variables , namely gpsv3 , gpsv8 , gpsv15 and gpsv16 , were acquired in august 2012 using the near - infrared integral field spectrometer ( nifs ) of gemini north telescope , mauna kea , hawaii .
the instrument has a field of view of @xmath121 with a pixel scale of @xmath122 across image slices and @xmath123 within image slices and a resolving power @xmath124 at 2.2 @xmath10 m .
the observations are part of the queue mode programme gn-2012b - q-102 .
observations of gpsv16 consisted of 4 integrations of 8 minutes , 2 of which were on source . the other , brighter objects were observed with two 8 minute integrations : one on source and one sky offset .
main sequence f - type stars were also observed for telluric correction and relative flux calibration . for each object a standard set of calibration data were acquired , consisting of flat images , xear lamp and its corresponding dark images and ronchi calibration mask to correct for the spatial distortion of integral field unit .
data reduction was performed in the standard way using the tasks of the gemini / nifs package in iraf . with the aim of investigating the presence of molecular outflows around eruptive variables candidates in serpens ob2 , near - infrared @xmath15 band and 2.12 @xmath10 m ( 1 - 0)s(1 ) h@xmath11 narrow band imaging data were obtained on august 7th 2012 using the wide field camera ( wfcam , * ? ? ?
* ) mounted on ukirt .
the instrument has four rockwell hawaii ii arrays ( 2048@xmath592048 pixels ) separated by 90% of a detector width , with a pixel scale of 0.4@xmath125pixel .
the standard 2@xmath592 tiling pattern was used to cover a continuous @xmath126 region containing 10 of the 11 eruptive variables candidates in the serpens ob2 association .
the total integration times were 2 minutes in @xmath15 and 25 minutes at 2.12 @xmath10 m .
the observations consisted of 24 images in @xmath15 , with exposure times of 5 s ( @xmath15 ) and 20 images at 2.12 @xmath10 m , with 75 s exposures .
a 2@xmath592 microstepping pattern was used to fully sample the spatial resolution in both filters , yielding a 0.2@xmath23 pixel scale in the reduced images .
data reduction and photometry were done by the cambridge astronomical survey unit ( casu ) using the standard reduction pipeline .
difference ( @xmath127 ) images are built in order to reveal emission line regions possibly associated with h@xmath11 jets and outflows .
the continuum subtracted images are built with the standard procedure ( see e.g. * ? ? ? * ) .
vphas@xmath60 ( drew et al , in prep . )
is an eso public survey , collecting @xmath128 , @xmath129 , @xmath130 , @xmath131 and narrowband @xmath132 photometry of the southern galactic plane and bulge .
it is in execution on the vlt survey telescope ( vst ) , using the 1 square - degree imager , omegacam .
the camera is constructed around a 4@xmath598 mosaic of 2k@xmath594k ccds .
the pixel size projects to 0.21@xmath23 on the sky , permitting good sampling of the 0.8 - 1.0@xmath23 point - spread function achieved .
the exposure times for the data used here were @xmath128 : 150 s , @xmath129 : 30 s , @xmath130,@xmath131 : 25 s , and @xmath132 120 s. to ensure more uniform seeing conditions , finally , across the full filter set , @xmath128 , @xmath129 , @xmath130 exposures are obtained as a contemporaneous group , separately from @xmath133 , @xmath134 and @xmath132 exposures . at each field pointing , each filter is exposed twice ( or three times , in @xmath132 ) at different offsets to compensate for gaps between the ccds making up the mosaic .
the data are reduced and calibrated by the cambridge astronomical survey unit ( casu ) , on the basis of nightly standards . in principle
this should yield an accuracy of @xmath1350.03 magnitudes .
the scale set should be regarded as provisional , as it is too early to establish a global calibration .
all magnitudes and colours are referred to vega as the zero - magnitude , zero - colour reference object .
the serpens ob2 region , relevant to this study , was observed as part of vphas@xmath60 on the nights of june 10th 2012 ( @xmath130 , @xmath131 and @xmath132 ) and november 11th 2012 ( @xmath128 , @xmath129 , @xmath130 ) . in addition to ukidss and isaac near - infrared photometry , we search for detections within 1@xmath23 of our objects in other near- and mid- ir surveys in the ipac science archive , namely _
spitzer-_glimpse3d @xcite , _
spitzer-_glimpse360 @xcite , wise @xcite , msx6c @xcite , _ akari _ @xcite , 2mass @xcite and denis @xcite .
we also searched for counterparts in the initial data release of iphas @xcite .
we note that all of our sample lies outside the area of the _ spitzer-_glimpse survey of the mid - plane but the 11 sources in the serpens ob2 region are included within glimpse3d , whilst gpsv16 is included within glimpse360 , a post - cryogenic _
mission which only uses filters i1
( @xmath136 $ ] ) and i2 ( @xmath137 $ ] ) @xcite .
@l@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@c@ & & & & & & & & & & + filter & r@xmath138 & i@xmath138 & h@xmath139 & i & j & k@xmath140 & j & h & k@xmath140 & j@xmath107 & h@xmath107 & k@xmath141 & k@xmath142 & j & h & k@xmath140 & k & @xmath143 $ ] & @xmath136 $ ] & @xmath137 $ ] & @xmath144 $ ] & @xmath145 $ ] & @xmath146 $ ] & @xmath147 $ ] & @xmath148 $ ] & @xmath149 $ ] & @xmath150 $ ] & @xmath151 $ ] & @xmath152 $ ] & @xmath153 $ ] & @xmath154 $ ] + & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) & ( mag ) + 5@xmath155 limit & _ 21.8 _ & _ 21 _ & _ 20.8 _ & _ 18.5 _ & _ 16.5 _ & _ 14.5 _ & _ 17 _ & _ 16.3 _ & _ 15.5 _ & _ 19.8 _ & _ 19 _ & _ 18.1 _ & _ 18.1 _ & _ 21 _ & _ 19.9 _ & _ 19.5 _ & _ 17.3 _ & _ 17.4 _ & _ 15.4 _ & _ 14.9 _ & _ 13.6 _ & _ 13 _ & _ 16.6 _ & _ 15.6 _ & _ 11.3 _ & _ 8.0 _ & _ 6.9 _ & _ 3.5 _ & _ 3.3 _ & _ 7.4 _ & _ 4.2 _ + & & & & _ nd_&_nd_&_nd _ & _ nd _ & _ nd _ & _ nd _ & 17.83 & 16.32 & 15.11 & 18.86 : & 17.98 & 16.49 & 15.38 & & & 15.26 : & 14.47 : & 13.33 : & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ + & & & & & & & & & & ( 0.04 ) & ( 0.02 ) & ( 0.02 ) & ( 0.55 ) & ( 0.05 ) & ( 0.03 ) & ( 0.04 ) & & & ( 0.16 ) & ( 0.35 ) & ( 0.34 ) & & & & & & & & & &
+ gpsv2 & 20.92 : & 19.06 & 20.04:&18.14&16.06&13.98&15.72&14.59&13.96&16.43&15.19&14.47&16.41&17.79&16.06&15.08&14.30&14.28 & 13.11 & 12.28 & 11.40 & 10.42 & 13.61 & 11.53 & 7.92 & 6.00 : & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ + & ( 0.10 ) & ( 0.04 ) & ( 0.12 ) & ( 0.18 ) & ( 0.24 ) & ( 0.22 ) & ( 0.08 ) & ( 0.07 ) & ( 0.06 ) & ( 0.02 ) & ( 0.02 ) & ( 0.02 ) & ( 0.06 ) & ( 0.05 ) & ( 0.03 ) & ( 0.04)&(0.02 ) & ( 0.02 ) & ( 0.08 ) & ( 0.06 ) & ( 0.07 ) & ( 0.05 ) & ( 0.07 ) & ( 0.04 ) & ( 0.05 ) & ( 0.18)& & & & & + gpsv3 & & & & _ nd_&_nd_&12.28&16.62&16.39&13.28&_nd_&16.03&12.47&13.75&_nd_&17.24&13.81&12.09 & 12.51 & 9.25 & 7.60 & 6.30 & 5.30 & 9.27 & 7.02 & 4.44 & 3.29 : & 4.32 & 3.43 : & 3.39:&4.77 & 3.39 + & & & & & & ( 0.13 ) & & & ( 0.04 ) & & ( 0.02 ) & ( 0.02 ) & ( 0.02 ) & & ( 0.04 ) & ( 0.05 ) & ( 0.02 ) & ( 0.02 ) & ( 0.05 ) & ( 0.05 ) & ( 0.03 ) & ( 0.03 ) & ( 0.03 ) & ( 0.02 ) & ( 0.04 ) & ( 0.16 ) & ( 0.11 ) & ( 0.18 ) & ( 0.20 ) & ( 0.47 ) & ( 0.11 ) + gpsv4 & & & & _ nd_&_nd_&_nd_&_nd_&_nd_&_nd_&17.54&16.48&15.53 & 16.60 & 17.80 & 16.87 & 16.26 & 16.42 & 16.42 & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ + & & & & & & & & & & ( 0.04 ) & ( 0.03 ) & ( 0.02 ) & ( 0.07 ) & ( 0.06 ) & ( 0.04 ) & ( 0.06 ) & ( 0.02 ) & ( 0.04 ) & & & & & & & & & & & & &
+ gpsv5 & & & & _ nd_&_nd_&_nd_&_nd_&_nd_&_nd_&17.61&16.30&15.53&16.58 & 18.19&16.85&15.97&15.55&15.62 & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ + & & & & & & & & & & ( 0.03 ) & ( 0.02 ) & ( 0.03 ) & ( 0.07 ) & ( 0.05 ) & ( 0.04 ) & ( 0.05 ) & ( 0.02 ) & ( 0.02 ) & & & & & & & & & & & & & + gpsv6 & & & & _ nd_&_nd_&_nd_&_nd_&_nd _ & _ nd _ & 18.25 & 17.04 & 16.37 & 15.36 & 17.98 & 16.89 & 16.18 & 15.77 & 15.80 & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ + & & & & & & & & & & ( 0.07 ) & ( 0.04 ) & ( 0.05 ) & ( 0.03 ) & ( 0.04 ) & ( 0.03 ) & ( 0.04 ) & ( 0.02 ) & ( 0.02 ) & & & & & & & & & & & & & + gpsv7 & & & & _ nd_&_nd_&_nd _ & _ nd _ & _ nd _ & _ nd _ & 16.86 & 15.44 & 14.15 & 13.15 & 16.53 & 15.00 & 13.82 & 13.30 & 13.48&11.82 & 11.33&10.96 & 9.80 & 11.20 & 10.71 & 8.42 & 5.20&_nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ + & & & & & & & & & & ( 0.02 ) & ( 0.02 ) & ( 0.02 ) & ( 0.02 ) & ( 0.04 ) & ( 0.03 ) & ( 0.04 ) & ( 0.02 ) & ( 0.02 ) & ( 0.07 ) & ( 0.10 ) & ( 0.09 ) & ( 0.04 ) & ( 0.03 ) & ( 0.03 ) & ( 0.08 ) & & & & & & + gpsv8 & 19.85 & 18.16 & 18.85&16.70&14.29&12.03 & 14.59 & 13.12 & 12.11 & 14.72 & 13.54 & 12.79 & 11.72 & 14.71 & 13.53 & 12.89 & 12.88 & 12.96&11.28 & 10.49 & 9.80 & 8.72&12.06&11.08 & 7.87 & 5.62:&_nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ + & ( 0.04 ) & ( 0.02 ) & ( 0.04 ) & ( 0.11 ) & ( 0.11 ) & ( 0.10 ) & ( 0.04 ) & ( 0.03 ) & ( 0.03 ) & ( 0.02 ) & ( 0.02 ) & ( 0.02 ) & ( 0.02 ) & ( 0.04 ) & ( 0.03 ) & ( 0.04 ) & ( 0.02 ) & ( 0.02 ) & ( 0.05 ) & ( 0.05 ) & ( 0.05 ) & ( 0.03 ) & ( 0.04 ) & ( 0.03 ) & ( 0.06 ) & ( 0.47 ) & & & & & + gpsv9 & & & & _ nd_&_nd_&_nd_&_nd_&_nd_&_nd_&_nd_&_nd_&16.43&15.28&_nd_&_nd_&16.19&15.77&16.08&13.52:&12.73:&_nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ + & & & & & & & & & & & & ( 0.05 ) & ( 0.02 ) & & & ( 0.05 ) & ( 0.02 ) & ( 0.03 ) & ( 0.19 ) & ( 0.16 ) & & & & & & & & & & &
+ gpsv10 & & & & _ nd_&_nd_&13.04 & 16.56&14.60&13.10&17.36&15.55&13.99&12.71&17.15&15.39&14.01&14.21&14.36&11.05 & 10.40 & 9.71&8.75 & 12.16 & 10.78 & 8.52 & 4.16 & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ + & & & & & & ( 0.16 ) & & ( 0.08 ) & ( 0.05 ) & ( 0.03 ) & ( 0.02 ) & ( 0.02 ) & ( 0.02 ) & ( 0.04 ) & ( 0.03 ) & ( 0.04 ) & ( 0.02 ) & ( 0.02 ) & ( 0.03 ) & ( 0.06 ) & ( 0.03 ) & ( 0.03 ) & ( 0.04 ) & ( 0.03 ) & ( 0.10 ) & ( 0.08 ) & & & & &
+ gpsv11 & & & & _ nd_&16.06&_nd_&16.05&14.80&14.39&19.38:&17.53&15.84&14.49&20.28:&18.31:&16.99&14.37&14.40&13.02&12.45&11.51:&10.30:&_nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ + & & & & & ( 0.24 ) & & ( 0.13 ) & ( 0.13 ) & ( 0.13 ) & ( 0.18 ) & ( 0.06 ) & ( 0.03 ) & ( 0.02 ) & ( 0.2 ) & ( 0.12 ) & ( 0.08 ) & ( 0.02 ) & ( 0.02 ) & ( 0.09 ) & ( 0.11 ) & ( 0.16 ) & ( 0.19 ) & ( 0.08 ) & ( 0.06 ) & ( 0.15 ) & & & & & &
+ gpsv12 & & & & _ nd_&_nd_&_nd_&16.26:&16.75&16.36&16.55&15.93&15.77&16.87&16.32&15.77&15.58 & & & & & & & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ + & & & & & & & ( 0.12 ) & & & ( 0.02 ) & ( 0.02 ) & ( 0.03 ) & ( 0.08 ) & ( 0.04 ) & ( 0.03 ) & ( 0.04 ) & & & & & & & & & & & & & & &
+ gpsv13 & 13.98 & 13.38 & 13.45&_nc_&_nc_&_nc_&12.46&12.13&11.90&13.21&13.06&12.92&11.90&13.33&13.21&13.12 & & & & & & & 13.23 & 13.29 & 12.30&8.66&_nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ + & ( 0.02 ) & ( 0.02 ) & ( 0.02 ) & & & & ( 0.02 ) & ( 0.02 ) & ( 0.02 ) & ( 0.02 ) & ( 0.02 ) & ( 0.02 ) & ( 0.02 ) & ( 0.03 ) & ( 0.02 ) & ( 0.04 ) & & & & & & & ( 0.03 ) & ( 0.04 ) & & & & & & &
+ gpsv14 & & & & & & & _ nd_&_nd_&_nd_&16.88&16.23&15.99 & 18.07 : & & & & & & & & & & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ + & & & & & & & & & & ( 0.02 ) & ( 0.02 ) & ( 0.03 ) & ( 0.23 ) & & & & & & & & & & & & & & & & & &
+ gpsv15 & & & & & & & 17.21&14.50&11.61&17.45&14.14&11.21&12.24&18.29&15.39&12.59 & & & & & & & 8.41 & 6.43 & 3.72 & 2.51 & 4.07 & 2.87:&2.95:&4.07 & 2.67 + & & & & & & & & ( 0.04 ) & ( 0.02 ) & ( 0.02 ) & ( 0.02 ) & ( 0.02 ) & ( 0.02 ) & ( 0.04 ) & ( 0.02 ) & ( 0.03 ) & & & & & & & ( 0.02 ) & ( 0.02 ) & ( 0.02 ) & ( 0.02 ) & ( 0.11 ) & ( 0.17 ) & ( 0.20 ) & ( 0.25 ) & ( 0.21 ) + gpsv16 & & & & & & & _ nd_&_nd_&_nd_&16.66&14.93&13.46&15.67&16.90&15.12&13.83&&&11.73&10.75&&&12.20 & 10.80 & 7.97 & 5.25 & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ + & & & & & & & & & & ( 0.02 ) & ( 0.02 ) & ( 0.02 ) & ( 0.03 ) & ( 0.07 ) & ( 0.04 ) & ( 0.04 ) & & & ( 0.03 ) & ( 0.03 ) & & & ( 0.03 ) & ( 0.02 ) & ( 0.03 ) & ( 0.04 ) & & & & &
+ gpsv17 & & & & _ nd_&_nd_&_nd_&_nd_&_nd_&_nd_&18.14&17.88&17.78 : & 15.43 & & & & & & & & & & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ & _ nd _ + & & & & & & & & & & ( 0.05 ) & ( 0.08 ) & ( 0.16 ) & ( 0.02 ) & & & & & & & & & & & & & & & & & & + + + + + + +
table [ table : photom ] shows the results of the 2010 isaac and 2012 ukirt photometry , along with the magnitudes found for targets in other near- and mid - ir surveys .
the errors are typically small ( @xmath156 mag ) , although given the faintness of some of our objects and significant source confusion in some fields , some of the detections in 2mass , and wise have moderate to poor signal to noise ratios ( snr ) .
detections with snr @xmath157 or errors larger than the typical values are marked with `` : '' in table [ table : photom ] .
we also note that even though errors in denis photometry are in the range @xmath158 mag , the photometric quality flag had a value of 100 for all of the detections in table [ table : photom ] and are considered to be reliable measurements . only one measurement is left out of the table and it corresponds to a vphas@xmath60 @xmath129 detection of gpsv8 with @xmath159 mag . _
spitzer _ glimpse detections yield more evidence that most of our objects are pms stars .
figure [ irac : ysos ] shows the location of our sample on colour - colour diagrams often used for yso classification when only near - infrared and irac 3.6 - 8.0 @xmath10 m fluxes are available ( see e.g. * ? ? ?
* ) . for comparison
we show the sample of ysos in young stellar clusters of @xcite , the classification scheme shown in the diagrams is the same as the one used by these authors . due to the high variability of our sample
, the location of the objects will be affected in the two lower panels , which use @xmath39 and @xmath15 photometry .
the latter are not contemporary with glimpse3d ( @xmath4 2006 ) , glimpse360 ( @xmath4 2010 - 2012 ) nor wise ( @xmath4 2010 ) .
therefore we use the closest epoch with simultaneous observations in these two bands , i.e. gps @xmath39 and @xmath37 ( @xmath4 2008 ) for the serpens objects , and vlt / isaac @xmath39 and @xmath15 ( 2010 ) for both gpsv15 and gpsv16 .
figure [ irac : ysos ] shows that most of our objects fall in what appears to be an area of transition between class i to class ii objects , especially on the diagrams based purely on _ spitzer _ photometry .
two sources stand out as they are located in the region where class i objects are found : these are gpsv3 and gpsv15
. however , the latter is not covered by _ spitzer _
surveys and filters @xmath160 and @xmath161 of wise are used instead .
the use of these filters should not have much effect on location in the diagrams as differences between wise and irac filters are only observed for @xmath162 magnitudes and @xmath163 magnitudes @xcite , which is fainter than the measured magnitudes for gpsv15 .
spectral energy distributions are constructed with the help of data from public surveys
. fits to stars in table [ table : sedfits ] were performed with the sed fitting tool of @xcite , which explores the parameter space of class 0 to class iii ysos at all system inclinations . in order to perform the fits we did not include every available data point , and in general we only used the information arising from ukidss @xmath164 , _ spitzer _ @xmath165 and _ wise _ @xmath166 .
we note that these are not contemporaneous observations , so the fits could be unreliable due to the variability of our objects .
in addition we are not certain whether @xcite models can accurately describe highly variable ysos , since their discs may have an unusual structure .
the distances are set to vary between 1 - 5 kpc and @xmath167 within 0 - 30 mag .
we note that varying the distances in the given range instead of using the known distances for the serpens objects or gpsv16 , had a negligible effect on the measured parameters , i.e. the change in the values were never larger than their estimated errors .
the weighted mean values and standard deviation of the parameters ( in logarithmic scale ) shown in table [ table : sedfits ] were determined using the models for which @xmath168 ( as suggested in * ? ? ?
* ) , where @xmath169 is the @xmath170 of the best - fitting model and @xmath171 represents the number of data points used in the fitting process .
the last column of table [ table : sedfits ] shows the number of models that fulfilled this condition for each of the objects in our sample .
the discussion on the results of the fits can be found in the analysis of the individual objects presented below .
given the faintness of many of our objects , the low resolution vlt / isaac spectra were of insufficient quality to clearly detect characteristics of eruptive variables .
the subsequent gemini / nifs and magellan / fire spectroscopy included only relatively bright sources whose isaac spectra suggested a possible eruptive variable classification .
the following analysis is then divided between stars where spectral characteristics of eruptive variables could be found and those for which there is only non - spectroscopic evidence for a yso or eruptive pms variable classification .
four stars are defined as likely to be eruptive variables .
these correspond to gpsv3 , gpsv15 , gpsv8 and gpsv16 . [ cols="<,^,^,^,^,^,^ " , ] these two objects are the reddest stars in our dr5 sample ( see fig . [ dr57cmd ] ) , with gpsv3 in fact not being detected in @xmath38 band in any of the near - infrared surveys nor in our isaac imaging .
both stars also show a large ( @xmath39-@xmath15 ) excess in colour - colour diagrams ( fig .
[ nircdr57 ] ) , beyond what is observed for classical t tauri stars and in agreement with what is expected for deeply embedded class i ysos .
the colours are also similar to those of the deeply embedded fuor - like objects pp13s and ar6b @xcite , and the deeply embedded outburst sources oo ser and gm cha @xcite . the @xmath15 band light curves of both objects show similar behaviour , with repetitive increases in magnitude ( fig .
[ lc_seds : erup ] ) .
their isaac spectra ( fig .
[ sp : erup ] ) also show a remarkable resemblance , where we observe a featureless continuum that rises steeply toward longer wavelengths .
this is similar to the aforementioned embedded object oo ser , and not too dissimilar to pp13s and ar6b , which differ only in that they show a strong @xmath172=2 - 0 co absorption trough in the 2.3 @xmath10 m region ( see fig . 5 in
the same steep rise is observed in the fire spectra of the sources , where gpsv15 is detected in @xmath39 band but with no sign of the h@xmath11o absorption that is observed in a number of fuor and fuor - like stars .
+ nifs higher resolution spectra reveal a more complicated structure in the co region .
the spectra of the objects only show clearly the @xmath173 bandhead ( stronger in gpsv15 ) with strong absorption that can be identified with the low @xmath38-lines of the @xmath174 and @xmath30 branches of the @xmath173 transition .
the weakness of the bandheads , relative to the lower j transitions , indicates that we are observing gas that is somewhat cooler than is typical of m - type stellar atmospheres , so that only the lower energy levels are populated .
moreover , the spectra in figure [ sp : erup ] strongly indicate that the photosphere is obscured from view and that the k band spectra are dominated by circumstellar matter .
this is seen especially in the magellan / fire spectra ( upper right panel of fig .
[ sp : erup ] ) , with a lack of flux at wavelengths shorter than @xmath175 ( where the stellar component would be expected to contribute to the observed flux of the system ) and steeply rising flux at longer wavelengths , in agreement with where a disc following a t @xmath176 k blackbody distribution is expected to begin contributing to the system .
we removed the slope of the higher resolution spectra by fitting a straight line to the 2.18 - 2.28 @xmath10 m region and extrapolating towards the co region .
the results are shown in fig .
[ nifs_co ] . in gpsv15
we see numerous features rising above the normalised continuum level in between the co absorption lines . this could be due to emission from the rapidly rotating hot portion of the inner disc , which would show as the wings of rotationally broadened emission lines in between the stronger but narrower absorption lines .
rotation velocities would be of the order of 200 km / s within a few stellar radii of a low mass yso , which is slightly higher than the nifs velocity resolution of 60 km / s . since features in excess of the continuum are not observed in between every pair of adjacent co absorption lines we can not be certain of this interpretation .
however , the wavelengths where we see no excess above the continuum are mostly located close to one of the higher order bandheads ( 3 - 1 , 4 - 2 and 5 - 3 ) where the close proximity of numerous co absorption lines disfavours observation of weak rotationally broadened emission features , at this spectral resolution .
observations at higher spectral resolution are therefore desirable in order to resolve the profile of rotationally broadened lines from the disc and verify their existence .
the presence of any such features in gpsv3 is marginal and strongly dependent on the choice of wavelengths used for the removal of the slope of the continuum .
however , several emission lines are observed in the high resolution spectrum of gpsv3 ( see fig .
[ nifs_co ] , right panel ) .
these lines can be identified with atomic features observed in the photospheric absorption spectra of k - m dwarf stars ( see e.g. , * ? ? ?
specifically we detect lines of mg i , na i , ca i , fe i , si i , al i , and perhaps s i and c i , most of which have rarely been reported in emission .
a temperature inversion is required to observe lines in emission , which could occur in a hot disc or wind near the star , making these features of circumstellar origin .
one of the strongest lines is observed at @xmath177 ( also observed in gpsv15 , see fig .
[ nifs_co ] ) , this is identified as the @xmath178 transition of mg i @xcite .
emission at this wavelength has also been observed in a sample of embedded protostars by @xcite but the authors mark this feature as unknown in their analysis .
evidence for the presence of a hot inner disc in gpsv3 is given by the observed emission of the na i doublet revealed in the nifs spectra .
this characteristic is observed in exors when in outburst phase and is explained as arising from a hot inner disc ( see e.g. , * ? ? ?
however , it is unclear whether we are observing emission arising from a hot inner disc , wind or a combination of both . in the case of gpsv15 ,
blueshifted absorption of @xmath4 40 km s@xmath58 is observed in the co bandhead and individual lines . blueshifted co absorption
has been observed previously in the fu ori objects v1057 cyg @xcite and ar6a @xcite . in the latter
, it was suggested that the blueshift might be caused by rotational broadening , as described by @xcite , but this was unclear . in v1057 cyg ,
the observed ( 50 km s@xmath58 ) blueshifted absorption was interpreted by @xcite as arising from an ejected dense and low temperature shell ( t@xmath4 620 k ) .
we note that the 40 km s@xmath58 blueshift seen in gpsv15 is too small to significantly affect our discussion concerning the presence of rotationally broadened emission features in between the absorption features . in summary , the observed co structure of gpsv15
could be explained by an emission component from an inner disc , combined with absorption by a cooler gas arising in an ejected shell .
it is also conceivable that cooler parts of the disc at larger radii are contributing to the absorption ( which could also be the case in gpsv3 ) .
however , the presence of both emission and absorption components in an individual object is more readily understood in terms of two separate structures .
the seds of gpsv3 and gpsv15 are exceptionally red in the 1 to 5@xmath179 m region , with @xmath38-w2@xmath911 mag .
this is consistent with a class i yso classification but at longer wavelengths their seds become roughly flat ( from @xmath45 - 12 @xmath10 m ) and then decline towards 23 @xmath10 m , as measured by _
akari _ and wise .
the sed flattening and lack of cold dust emission is consistent with an accretion disc .
we find that for gpsv15 , only a relatively small number of models could be used to estimate the values of the parameters in table [ table : sedfits ] . in addition
the @xmath180 per data point for the best - fitting model of gpsv3 was found to be larger than the derived values for the remaining objects .
thus , the parameters derived by the fits for gpsv3 and gpsv15 , shown in table [ table : sedfits ] can be considered as highly unreliable .
however the @xcite fits support our impression from inspection of the seds that gpsv3 and gpsv15 are not systems with optically thick envelopes given the lack of emission of cold dust at longer wavelengths .
even though both sources show clear yso characteristics , the fact that gpsv15 is not associated with any star forming region raises the doubt that these stars could actually be highly reddened post - agb stars , or oh / ir stars similar to those mentioned in section [ sec : search_res ] . using k band photometry and _ akari 9 @xmath10 m _
observations @xcite were able to determine mass loss rates and distances for carbon- and oxygen - rich agb stars in the galaxy .
the [ k]-[9 ] colours for both gpsv3 and gpsv15 would correspond to agb stars with very large mass loss rates , as high as 10@xmath181 m@xmath5yr@xmath58 according to fig .
a.2 in @xcite .
however , using these values and their observed fluxes at 9 @xmath10 m in eq .
a.2 of @xcite , yields distances larger than 25 kpc from the sun , which at their galactic latitude ( @xmath182 ) would put them 800 pc above the galactic plane and well beyond the galactic disc edge ( @xmath183 kpc , see e.g. * ? ? ?
* ) , thus ruling out a post - agb classification for both gpsv3 and gpsv15 .
k band photometry of gpsv8 shows that the star was brighter during 2mass and gps @xmath184 observations in 1999 and 2005 but faded by the epoch of gps @xmath37 in 2008 and remained at that state during isaac and ukirt observations in 2010 and 2012 ( see fig .
[ lc_seds : erup ] ) .
the fading is also apparent when comparing _
@xmath185 ( @xmath186 ) with _ wise _ @xmath187 ( @xmath4 2010 ) photometry , where the difference of both @xmath188,@xmath189 are larger than the typical difference of @xmath4 0.2 mag found for field stars with a similar magnitude to gpsv8 . the isaac spectrum of gpsv8 ( fig .
[ sp : erup ] ) shows strong br@xmath13 emission along with co absorption , with na i and ca i also in absorption .
the same spectroscopic characteristics can be observed in the recent 2012 nifs spectroscopic data .
they are typical of exor variables in the quiescent state @xcite and consistent with the k band light curve .
gpsv8 is also detected in vphas@xmath60 .
direct comparison of its @xmath190 colours places it in the @xmath132 emitters region of @xcite .
we do note that their actual selection of @xmath132 emitters depended on the magnitude of the source and field where it was located .
the aforementioned colours also place it near the unreddened mid m star colours in figure 6 of @xcite .
however this classification is not possible given its observed @xmath191 colours and the sed of the star does not agree with a mid m star classification .
the r@xmath138-h@xmath55 colour implies emission with maximum equivalent width of 40 - 50 , for the case of a heavily veiled yso or object with a continuum that is free of tio bands . with near - infrared br@xmath13 emission as well and a low mass class ii yso fit with the @xcite models ( see table [ table : sedfits ] ) an accreting yso scenario for gpsv8 is well supported .
the light curve of gpsv16 ( see fig . [ lc_seds : erup ] ) supports an outburst scenario with the star still close to maximum brightness at the moment of our isaac observations ( 2010 ) .
further evidence to support this scenario is given by the fact that the object is not apparent in the 2mass ( 1999 ) images .
the isaac spectrum shows br@xmath13 emission as well as co in emission .
these emission features are much more clearly seen in the 2012 nifs spectrum of the source ( fig .
[ sp : erup ] ) .
these are characteristics of an exor variable in outburst .
the fact that the spectrum of gpsv16 has not changed noticeably between the 2010 and 2012 observations suggests that the star was either at a bright state or been fortuitously observed during two separate outbursts .
the sed of gpsv16 shows a strong rise towards 2.2 @xmath10 m and a flat distribution towards longer wavelengths , similar to what is observed in classical t tauri stars ( ctts ) . supporting a yso scenario is the fact that the star is located at a projected distance of @xmath4 170@xmath23 from the star forming region g71.52 - 0.39 of the @xcite catalog ( see figure [ fig : serp ] ) , and although it is not conclusive this could be evidence for a possible association . sed fitting of gpsv16 with the tool of @xcite suggests a classification as low mass class ii ysos ( see table [ table : sedfits ] ) .
in addition to gpsv3 and gpsv8 , the rest of gpsv1 to gpsv11 are also located in the serpens ob2 region .
the seds of gpsv1 , gpsv2 , gpsv7 , gpsv10 and gpsv11 are consistent with them being ysos , rising until typically 3.6 @xmath10 m and then showing flatter seds towards longer wavelengths .
we note the cases of gpsv7 and gpsv10 , which show a dip in their seds in the mid - infrared before rising towards @xmath192 m .
ysos with this type of sed have been interpreted as having optically thin or evacuated inner holes and an optically thick edge at a radius of @xmath410 au @xcite . alternatively
, extinction of the hot inner parts of the disc by an edge - on system orientation can produce a similar effect .
we note that the glimpse3d catalogue photometry flags indicate that contamination from a nearby source may have affected the photometry of gpsv7 .
the same source is probably responsible for the non - detection in 2mass and denis and may well have affected the wise data . from the aforementioned sources , only gpsv7 shows interesting spectral features in our low resolution isaac spectra , with co absorption bandheads in the 2.3 @xmath10 m region ( fig .
[ lc : ysos ] ) and no sign of br@xmath13 emission . however , its peak - to - trough amplitude in @xmath15 barely exceeds 1 magnitude , making its eruptive nature questionable .
gpsv1 and gpsv11 correspond to the most interesting cases .
the magnitude of gpsv1 rose by 3.75 mag between the two epochs of gps , then declined by 0.26 mag in our isaac photometry .
the fact that the star is not visible in the 2mass image provides some support for a possible recent outburst scenario .
unfortunately even at its brightest state the star is still faint for our isaac spectroscopic observations ( @xmath193 15.4 mag ) and no characteristic features for fuor / exors could be identified from the spectrum of this object .
however , the flat slope of the spectrum rules out a normal main sequence classification and , as noted above , the sed rises steeply towards 5.8 @xmath10 m , consistent with a yso interpretation . if the variability is due to a yso outburst , the rate of decline between gps @xmath194 and isaac is @xmath40.01 mag / month and the star would return to its pre - outburst magnitude ( assuming the first epoch of ukidss gps photometry as the pre - outburst state ) in @xmath4 30 yrs .
this falls in the expected timescale for an fuor outburst ( see e.g. * ? ? ?
gpsv11 shows similar magnitudes in both 2mass ( 1999 ) and the first epoch of gps ( 2005 ) , followed by a strong decline towards the gps k@xmath195 ( 2008 ) and isaac ( 2010 ) epochs of @xmath4 2.5 magnitudes .
the recent ukirt observations ( 2012 ) show that the object has returned to a magnitude level similar to that of 2mass .
the magnitude change in this source is probably not related to variable extinction along the line of sight ( see below ) .
it is unfortunate however that the spectrum , also dominated by noise , makes it impossible to observe any spectral characteristics that could clarify the nature of the object , however the observed slope in the spectrum of gpsv11 goes against a main sequence star classification which supports the yso scenario for this star . .
spectral features usually found for eruptive variable stars are shown only for gpsv7 , the only object where co absorption bands can be identified .
spectra for the other four objects are dominated by noise . ]
gpsv2 and gpsv10 are also highly variable objects with variations of 2.4 and 1.5 mag respectively .
however , the variability of the former is more likely related to variable extinction in the line of sight ( see below ) . similar to gpsv8 , gpsv2 is also detected in vphas@xmath60 and is located on the lower envelope of the @xmath132 emitters distribution from @xcite .
the case for line emission is also marginal when comparing to figure 6 in @xcite .
the @xmath196 and @xmath197 could be compatible with an unreddened mid - m star .
the slope of the observed k band spectrum is incompatible with this scenario .
the observed sed also goes against the mid m classification .
then the @xmath132 equivalent width could be 30 - 40 if the underlying sed is free of tio bands .
this supports a yso classification .
the slope of the isaac spectrum and the sed of gpsv10 are inconsistent with a normal main sequence star and support the yso classification .
gpsv9 is a very red ( and faint ) object .
its relatively low observed amplitude of 1.16 magnitudes makes an eruptive yso classification questionable , similar to objects gpsv4 , gpsv5 and gpsv6 .
however , the apparent spatial association of these objects with serpens ob2 , makes them likely ysos .
spectra of the four candidates did not help in the classification of these objects .
we note that for two of the objects discussed as probable eruptive variables above , gpsv8 and gpsv15 , peak - to - trough k - band amplitudes do not exceed @xmath41.5 mag , bringing some conflict with one of the main characteristics used to define eruptive variables , which is an increase of brightness larger than at least @xmath42 mag .
band amplitudes of the ( new detections ) exor sample from @xcite ( see their table 1 ) show a limit of @xmath198 mag , with the majority of the sample having @xmath199 mag . ]
although this is probably due to the incomplete coverage of the light curves , we want to discard other possible physical mechanisms that could be causing large variability .
colour - colour plot comparing the 2mass ( _ blue _ ) , gps ( _ green _ ) and isaac ( _ red _ ) colours for some of the objects in our sample .
the errors are plotted only for objects that present significant uncertainties on their measurements.,title="fig : " ] + colour - colour plot comparing the 2mass ( _ blue _ ) , gps ( _ green _ ) and isaac ( _ red _ ) colours for some of the objects in our sample .
the errors are plotted only for objects that present significant uncertainties on their measurements.,title="fig : " ] several physical mechanisms can explain the near - infrared variability observed in ysos .
some of these mechanisms are cool spots induced by magnetic activity , hotspots formed by the accretion flow ( see e.g. * ? ? ?
* ) , magnetic field interaction between the disc and the star @xcite and accretion driven wind and outflow @xcite .
however , these mechanisms often produce short - term variability with amplitudes that are not expected to exceed 1 magnitude in k band .
changes in the extinction along the line of sight can produce larger changes in magnitude and on longer timescales ( see e.g. the long - lasting deepening of aa tau , * ? ? ?
* ) . in this case
the colour of the star would be expected to move along the reddening vector in colour - colour plots .
the fact that we use @xmath200 mag to select our candidates already discards some of the mechanisms stated before .
the changes in colour of the sources discussed above ( fig .
[ chcol : erup ] ) can provide further information on the nature of the objects ; exor variables for example are expected to have bluer colours as their magnitude increases . however , this is not observed in our sample .
we note that due to the limited coverage we are not able to directly compare colours from quiescent and bright states .
we can only discard variable extinction as a possible explanation because the stars do not move along the reddening vectors , except in the case of gpsv2 .
@xcite find 1235 variable stars in their study of the orion molecular cloud .
visual inspection of their light curves yields 20 variables with @xmath15 peak - to - trough amplitudes @xmath9 1 mag .
the variability of most of these stars is of the order of days and is likely to be explained by the physical processes discussed above .
however , a handful of stars present large variability ( up to 1.7 mag ) on longer timescales , driven by a few points in the light curves and resembling what one would expect to observe in eruptive variables .
we note that @xcite do not claim to have observed any fu ori type outbursts . comparing short - term variables ( _ dashed line _ ) with long - term and more likely to be accretion - related variables ( _ dotted line _ ) . ]
we want to determine the likelihood of detecting any of the 20 high - amplitude @xcite variables , by just observing two epochs and as a function of a defined amplitude cut . in order to do so
, we selected two random points of the light curve of each star in the high amplitude @xcite sample , determine the magnitude difference between the two points and marked a detection if the difference was larger than the defined amplitude cut . the process was iterated a few thousand times .
the likelihood of the detection of each star is then defined as the number of detections divided by the number of iterations .
the sum of the expected detections over the 20 variables from the @xcite sample was multiplied by the ratio between the number of ob stars in serpens ob2 and those within the region of orion studied by @xcite .
the described process yields the number of variable stars we would expect to have selected in serpens ob2 as a function of the amplitude cut ( see fig . [
lc : carp ] ) .
even though there are a number of caveats to this analysis , we find that for an amplitude cut of @xmath200 mag , we should detect 7 variables , which is remarkably close to our actual number ( 11 ) . at this amplitude cut ,
a large part of our sample would be formed by stars which variability is explained by other processes than abrupt changes in the accretion rate .
however , we observe that for @xmath201 mag , the sample is mostly formed by stars which vary over longer timescales driven by a few points in their light curves ( dotted line in fig . [ lc : carp ] ) , and thus more likely associated with accretion - related variability .
even though our analysis uses sources in serpens ob2 , we can assume that the observed photometric changes in ysos in our sample which present @xmath201 mag are more likely related to this type of variability .
the visual inspection of the ukirt @xmath127 difference images reveals the presence of only one pair of bright h@xmath11 molecular outflows in the area of serpens covered by our study .
we identify them as the previously discovered emission features mho2201 and mho2202 , which are associated with the deeply embedded luminous yso iras 18151 - 1208 @xcite .
there is also widespread diffuse h@xmath11 emission in parts of the region , but no other jet - like features that can be associated with a single star .
this suggests that the observations were not deep enough to detect jets associated with low mass ysos .
no @xmath202 outflows could be observed near the eruptive variable candidates gpsv2 to gpsv11 .
similarly , there is no sign of an outflow in the nifs integral field data for the four objects spectroscopically verified as eruptive variables ( gpsv3 , gpsv8 , gpsv15 and gpsv16 ) .
most fu orionis objects are known to be associated with herbig - haro objects @xcite and drive molecular outflows @xcite , whilst exors generally do not show evidence of shocks as a major mechanism of excitation @xcite .
fu orionis itself , although having massive winds , does not drive a molecular outflow @xcite which could be explained by the lack of organization of magnetic fields required for jet launching and collimation in the binary scenario of @xcite or by previous outflows having cleared up the remnant envelope evidencing the lack of swept up material which constitutes the molecular outflows @xcite . from the objects associated with the deeply embedded subclass of eruptive ysos ,
oo ser is associated with faint h@xmath11 shock fronts @xcite , corresponding to mho 3245 in the catalog of @xcite .
gm cha is the driving source of a co outflow and shows a @xmath15 band spectrum dominated by h@xmath11 emission with several shocked h@xmath11 emission knots observed along the co flow @xcite .
finally , v2775 ori is not associated with any outflows nor h@xmath11 shock emission .
given the limitation imposed by the sensitivity of the dataset and the fact that outflows are not ubiquitous in eruptive variable ysos , the lack of observable h@xmath11 emission near our somewhat more distant candidates is not a strong argument against their eruptive variable classification .
although four stars in our sample show strong characteristics of eruptive variability , it is difficult to classify them as either of the two known classes of this kind of variables .
gpsv3 and gpsv15 show similar spectra to known deeply - embedded fuor - like objects and that of oo ser .
however , the amplitude and time - scale of the variation , with what appear to be repetitive outbursts , are not in line with the observed behaviour of oo ser itself and the outbursting protostar [ cft93]216 - 2 @xcite .
in fact , the photometric behaviour of both candidates is similar to that observed for deeply embedded objects v371 ser @xcite and source 90 in the study from @xcite of sfr 173.58 + 2.45 .
the latter has large infrared colours and it is the driving source of an h@xmath11 molecular outflow .
its variability was first thought to be of the fu orionis kind but this classification was discarded because of the shorter time scale of the variation .
@xcite still conclude that its variability probably arises from variable accretion .
v371 ser is classified as a class i object , also driving a h@xmath11 molecular outflow .
its variability is periodic ( @xmath203 ) and is thought to be explained by periodic variation of the accretion rate , modulated by the presence of a close companion @xcite .
a binary companion in an elliptic orbit is a possible explanation for the fuor and exor phenomena , given that some repetitive exors have been seen ( though not fuors ) .
@xcite proposed that v371 ser represents an extension of eruptive variability to shorter periods .
the spectrum and photometry of gpsv8 from 2010 - 2012 is consistent with an exor during quiescent states .
however , the star appears to have been at a bright state during 1998 - 2005 ( although this conclusion is based on only two epochs of observations ) , if so the time - scale is inconsistent with exor variables .
the spectrum of gpsv16 shows strong emission lines ( br@xmath13 , co ) during 2010 and 2012 observations , which is usually observed in exor variables during bright states .
however , the photometric data of the object would point towards an fuor type outburst given the longer timescale of the variation .
a number of stars in which the observed eruptive variability is thought to be accretion related have shown characteristics that can be associated with both sub - classes of eruptive variables .
e.g. v1647 ori , v2775 ori and oo ser .
these objects have been proposed to be intermediate objects between exors and fuors or being the extension of the phenomenon towards lower luminosities .
this raises the scenario where eruptive variability can not be simply divided in two distinctive sub - classes but it is instead formed by a continuum of outburst events of different luminosities , amplitudes and time - scales , but triggered by a similar physical mechanism ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
thus , our results seem consistent with this scenario , where gpsv3 and gpsv15 would be new additions to the deeply - embedded extension of the fuor phenomenon .
future photometric monitoring of the sources will allow us to better constrain the timescales of the observed variability . also ,
spectroscopic analysis of sources in later data releases from gps ( dr7,dr8 ) , with addition of future searches in the multiple @xmath15 band epochs data from vvv will help us in our effort to characterise and understand the subclasses of eruptive variables .
five stars from our dr5 sample are marked as variables not associated with sfrs .
gpsv15 has already been discussed at length and has been classified as a probable eruptive variable in our analysis . from the remaining four stars , gpsv13 and gpsv17
correspond to blue objects , with gpsv17 already being identified as nova sct 2003 .
the location of gpsv12 in its local cmd is consistent with a late type dwarf or a distant giant in the galactic halo , whilst gpsv14 seems to be projected against the dwarf sequence on its cmd in figure [ dr57cmd ] .
high amplitude variability can be produced by a great number of physical mechanisms .
inspection of figure 2 of @xcite shows that symbiotic stars , classical be stars and cataclysmic variables , among others , can be found at the location of the objects in near - infrared colour - colour plots .
we also note that algol - type eclipsing variables show large variability at optical wavelengths and while amplitudes at near - infrared wavelengths are lower , it is conceivable that we could detect an extreme case of this type of variable stars .
not much information is available for objects gpsv12 and gpsv14 .
the latter was not included in the vlt / isaac follow up and only has magnitude information from the ukidss gps data .
gpsv12 was faint at the time of our isaac observations and the spectrum did not help to draw any conclusion regarding its classification .
gpsv12 could plausibly be an agn ( see @xmath992.2 ) but the higher amplitude of variability in gpsv14 ( 2.1 mag ) makes this interpretation unlikely for that object .
+ gpsv13 is one of the two blue dr5 sources and it is also one of the brightest variable stars in our sample .
its 1999 2mass k@xmath140 magnitude is very similar to the first k band observation of gps in 2005 .
the star was fainter at the second gps epoch ( 2008 ) and continues fading towards our vlt / isaac observations ( 2010 ) .
the spectrum of the star at this last epoch shows br@xmath13 in absorption ( see figure [ figs : gps13 ] ) .
the object was found to be classified as an h@xmath55 emitter in the @xcite catalogue .
the iphas observation for this object took place in 2004/2005 during the bright state of the object .
figure [ figs : gps13 ] shows the optical spectra of gpsv13 collected in june 2009 at the 1.5 m fred laurence whipple observatory ( flwo ) tillinghast telescope using the fast spectrograph for the tillinghast telescope ( fast , * ? ? ?
* ) , with a spectral resolution of @xmath204 , covering the wavelength range 3500@xmath647500 .
the equivalent width ( ew ) of the h@xmath55 emission is 9.7 , and the spectral typing of the object yields a b5 spectral type based on the ew of the hei 4471 , and just resolved but weak mgii 4481 absorption lines .
the reddening is estimated to be a@xmath205 . assuming a b5v classification yields a distance of @xmath4 2.8 kpc .
the 2004/2005 @xmath190 colours of gpsv13 allow us to estimate the ew of the h@xmath55 emission as @xmath420 - 22 , which is higher than the measured value of 2009 and consistent with the photometric behaviour of the object under the classical be interpretation described below .
classical be stars show variability over long timescales that can be attributed to the dissipation of the disc @xcite . during this process
, the star will go from having an emission line spectrum to a normal b type star spectrum ( see e.g. * ? ? ?
* ; * ? ? ?
this process could explain the observed behaviour of the spectrum of gpsv13 .
in addition , dissipation of the disc could explain the large ( 1.2 mag ) decline in brightness in the k band from 2005 to 2010 .
inspection of figure 9 of @xcite shows that the dissipation of a pole - on disc in the models of these authors just reproduce the observed amplitude of the change .
this scenario also explains the apparent lack of ir excess in the 2010 wise w1 , w2 magnitudes .
an alternative scenario would be that gpsv13 is a fairly evolved herbig ae / be star with a far ir excess not detectable in wise data .
however the star is not close to any known sfr , there is no observable nebulosity in the wise colour image of the region and the estimated distance to the object implies that gpsv13 would be @xmath4 160 pc above the galactic plane , at least three molecular cloud scale heights for an inner galaxy object .
figure 2 in @xcite compares the location of classical be and herbig ae / be stars in near - infrared colour - colour plots , the authors find that these types of variables seem to occupy different places in these diagrams .
the observed colours of gpsv13 would place the object in the classical be region .
also , its sed resembles those of classical be stars from the sample of @xcite , see [ figs : gps13 ] .
thus the classical be scenario seems the most likely explanation for gpsv13 .
if gpsv13 is in fact a classical be star , it would correspond to one of the most ( if not the most ) extreme variable star of this class . to our knowledge
classical be stars have not been observed to vary by more than 1 magnitude ( see e.g. , * ? ? ?
* ; * ? ? ?
we have presented the results of the first panoramic search for high - amplitude near - infrared variables in the galactic plane .
we summarize our principal results as follows .
* the search for high - amplitude infrared variables in data releases dr5 and dr7 of gps yields 45 stars which display @xmath0 mag .
the sample included one known nova and two oh / ir stars .
most notably , two thirds of our candidates are located within known areas of star formation and are likely ysos .
four other stars also show characteristics of ysos but do not appear to be close to areas of star formation .
thus we discuss the possibility of high - amplitude ir variables as a tracer of star formation . +
* the high yso fraction found in our study implies an average source density of 0.19 deg@xmath79 .
we conclude that this number would be likely to rise to @xmath413 deg@xmath79 in the mid - plane , if we had sufficient sensitivity to detect them across the galaxy , after allowing for the effects of 2-epoch sampling and scale height upon our sample .
this would make pms stars the commonest type of high amplitude variable in the near - infrared .
+ * considering the dr5 sample , we assign an eruptive variable classification to four stars based on their observed spectral characteristics and their resemblance to known members of the fuor / exor class .
two of these are likely new members of the deeply embedded subclass of eruptive variables .
nine stars likely associated with the serpens ob2 association are confirmed as having yso characteristics , but an eruptive variable classification can not be confirmed . in this context ,
physical mechanisms that could explain ir variability in ysos are discussed .
variable extinction along the line of sight is found to be the likely cause of the variation for only one of our variable objects . +
* the comparison with the variable yso sample of @xcite allows us to estimate that the variability in stars with @xmath201 mag in two epochs of observation separated by a few years is most likely related to longer timescale variability like that observed in eruptive variables . +
* stars confirmed as eruptive variables show characteristics that could be associated with either of the two known classes of these type of variable stars ( exors or fuors ) .
this is found to be consistent with recent studies that suggest that the episodic accretion phenomena is comprised of a continuum of events with different properties but triggered by a similar physical mechanism . +
* variable stars in the serpens ob2 region do not show any signs of molecular outflows .
+ * four stars in the dr5 sample appear not to be associated with star forming regions .
we argue in favour of a classical be classification for one of such objects .
if so , the star would be one of the most variable classical be stars .
this paper is based in part on observations obtained in programme gn-2012b - q-102 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 ( usa ) , the science and technology facilities council ( uk ) , the national research council ( canada ) , conicyt ( chile ) , the australian research council ( australia ) .
cnpq ( brazil ) and conicet ( argentina ) .
this paper is also based on observations taken as part of programme 085.c-0352(a ) with the very large telescope in cerro paranal , which is operated by the european southern observatory .
we thank our colleagues in the spitzer legacy survey of the cygnus - x complex for their work in providing the public data for many of the dr7 objects .
we also thank our colleagues in the glimpse360 spitzer legacy survey team for providing the irac data for gpsv16 . this research has made use of the simbad database , operated at cds , strasbourg , france .
this research has made use of the nasa/ ipac infrared science archive , which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
we thank the referee , jacco van loon , for a thorough and helpful report .
we would also like to thank j. najita for the very helpful discussions on some of our co spectra and d. steeghs for his assistance in providing the fast data . c. contreras pea
is supported by a university of hertfordshire phd studentship .
99 alves de oliveira c. , casali m. , 2008 , a&a , 485 , 155 ashok n. m. , bhatt h. c. , kulkarni p. v. , joshi s. c. , 1984 , mnras , 211 , 471 aspin c. , sandell g. , 2001 , mnras , 328 , 751 aspin c. , reipurth b. , 2003 , aj , 126 , 2936 avedisova v. s. , 2002 , arep , 46 , 193 bans a. , knigl a. , 2012 , apj , 758 , 100 baraffe i. , chabrier g. , gallardo j. , 2009 , apj , 702 , l27 baraffe i. , vorobyov e. , chabrier g. , 2012 , apj , 756 , 118 bedijn p. j. , 1987 , a&a , 186 , 136 benjamin r. a. , et al . , 2003 , pasp , 115 , 953 bessell m. s. , brett j. m. , 1988 , pasp , 100 , 1134 bouvier j. , grankin k. , ellerbroek l. , bouy h. , barrado d. , 2013 , arxiv , arxiv:1304.1487 briceo c. , calvet n. , hernndez j. , vivas a. k. , hartmann l. , downes j. j. , berlind p. , 2005
, aj , 129 , 907 caratti o garatti a. , et al . , 2011 ,
a&a , 526 , l1 caratti o garatti a. , et al . , 2012 , a&a , 538 , a64 carciofi a. c. , bjorkman j. e. , 2006 , apj , 639 , 1081 cardelli j. a. , clayton g. c. , mathis j. s. , 1989 , apj , 345 , 245 carpenter j. m. , hillenbrand l. a. , skrutskie m. f. , 2001 , aj , 121 , 3160 casali m. , et al .
, 2007 , a&a , 467 , 777 castelli f. , kurucz r. l. , 2004 , astro , arxiv : astro - ph/0405087 catelan m. , et al . , 2013 , arxiv , arxiv:1310.1996 chapman n. l. , mundy l. g. , lai s .- p . ,
evans n. j. , ii , 2009 , apj , 690 , 496 cioni m .-
l. , marquette j .- b .
, loup c. , azzopardi m. , habing h. j. , lasserre t. , lesquoy e. , 2001 , a&a , 377 , 945 cioni m .- r .
l. , et al .
, 2013 , a&a , 549 , a29 connelley m. s. , greene t. p. , 2010
, aj , 140 , 1214 corradi r. l. m. , et al . , 2008 , a&a , 480 , 409 corradi r. l. m. , et al . , 2010 , a&a , 509 , a41 crutcher r. m. , chu y. h. , 1982 , assl , 93 , 53 cutri r. m. , et al . , 2012 ,
wise.rept , 1 davis c. j. , gell r. , khanzadyan t. , smith m. d. , jenness t. , 2010 , a&a , 511 , a24 davis c. j. , et al . , 2011 ,
a&a , 528 , a3 drr m. , chini r. , haas m. , lemke r. , nrnberger d. , 2013 , a&a , 553 , a48 dougherty s. m. , taylor a. r. , 1994 , mnras , 269 , 1123 drew j. e. , et al . , 2005 , mnras , 362 , 753 dutra c. m. , bica e. , 2002 , a&a , 383 , 631 egan m. p. , et al . , 2003 ,
ycat , 5114 , 0 dye s. , et al .
, 2006 , mnras , 372 , 1227 elias f. , cabrera - cao j. , alfaro e. j. , 2006 , aj , 131 , 2700 enya k. , yoshii y. , kobayashi y. , minezaki t. , suganuma m. , tomita h. , peterson b. a. , 2002 , apjs , 141 , 45 epchtein n. , et al . , 1994 , ap&ss , 217 , 3 evans n. j. , ii , balkum s. , levreault r. m. , hartmann l. , kenyon s. , 1994 , apj , 424 , 793 evans n. j. , ii , et al . , 2009 , apjs , 181 , 321 fabricant d. , cheimets p. , caldwell n. , geary j. , 1998 , pasp , 110 , 79 fedele d. , van den ancker m. e. , petr - gotzens m. g. , rafanelli p. , 2007
, a&a , 472 , 207 forbes d. , 2000 , aj , 120 , 2594 gibb e. l. , rettig t. w. , brittain s. d. , wasikowski d. , simon t. , vacca w. d. , cushing m. c. , kulesa c. , 2006 , apj , 641 , 383 gonzlez - solares e. , walton n. , irwin m. , lewis j. , rixon g. , greimel r. , drew j. , 2008 , aspc , 394 , 197 grinin v. p. , kozlova o. v. , natta a. , ilyin i. , tuominen i. , rostopchina a. n. , shakhovskoy d. n. , 2001 , a&a , 379 , 482 gutermuth r. a. , megeath s. t. , myers p. c. , allen l. e. , pipher j. l. , fazio g. g. , 2009 , apjs , 184 , 18 hambly n. c. , et al . , 2008 , mnras , 384 , 637 hartmann l. , kenyon s. j. , 1996 , ara&a , 34 , 207 hartmann l. , hinkle k. , calvet n. , 2004 , apj , 609 , 906 hartmann l. , megeath s. t. , allen l. , luhman k. , calvet n. , dalessio p. , franco - hernandez r. , fazio g. , 2005 , apj , 629 , 881 hashimoto o. , 1994 , a&as , 107 , 445 herbig g. h. , 1989 , esoc , 33 , 233 hernndez j. , calvet n. , hartmann l. , briceo c. , sicilia - aguilar a. , berlind p. , 2005
, aj , 129 , 856 hodapp k .- w . , hora j. l. , rayner j. t. , pickles a. j. , ladd e. f. , 1996 , apj , 468 , 861 hodapp k. w. , chini r. , watermann r. , lemke r. , 2012 , apj , 744 , 56 hunter d. a. , thronson h. a. , jr . , wilton c. , 1990 , aj , 100 , 1915 ioannidis g. , froebrich d. , 2012 , mnras , 425 , 1380 ishihara d. , kaneda h. , onaka t. , ita y. , matsuura m. , matsunaga n. , 2011 , a&a , 534 , a79 jimnez - esteban f. m. , garca - lario p. , engels d. , perea caldern j. v. , 2006 , a&a , 446 , 773 kenyon s. j. , hartmann l. w. , strom k. m. , strom s. e. , 1990 , aj , 99 , 869 kleinmann s. g. , hall d. n. b. , 1986 , apjs , 62 , 501 knapp g. r. , 1987 , pasp , 99 , 1134 kspl . , brahm p. , prusti t. , acosta - pulido j. , hony s. , mor a. , siebenmorgen r. , 2007 , a&a , 470 , 211 kspl .
, et al . , 2011 , a&a , 527 , a133 kouzuma s. , yamaoka h. , 2012 , apj , 747 , 14 lamm m. h. , bailer - jones c. a. l. , mundt r. , herbst w. , scholz a. , 2004 , a&a , 417 , 557 lawrence a. , et al .
, 2007 , mnras , 379 , 1599 lebzelter t. , wood p. r. , 2005 , a&a , 441 , 1117 lewis b. m. , 2000 , apj , 533 , 959 lorenzetti d. , larionov v. m. , giannini t. , arkharov a. a. , antoniucci s. , nisini b. , di paola a. , 2009 , apj , 693 , 1056 lorenzetti d. , et al . , 2012 , apj , 749 , 188 lucas p. w. , et al . , 2008 , mnras , 391 , 136 lynds b. t. , 1962 , apjs , 7 ,
1 maddox n. , hewett p. c. , warren s. j. , croom s. m. , 2008 , mnras , 386 , 1605 magakian t. y. , 2003 , a&a , 399 , 141 magakian t. y. , et al . , 2013 , mnras , 432 , 2685 mcswain m. v. , huang w. , gies d. r. , grundstrom e. d. , townsend r. h. d. , 2008 , apj , 672 , 590 mcswain m. v. , huang w. , gies d. r. , 2009 , apj , 700 , 1216 meyer m. r. , calvet n. , hillenbrand l. a. , 1997 , aj , 114 , 288 minniti d. , et al . , 2010 , newa , 15 , 433 minniti d. , saito r. k. , alonso - garca j. , lucas p. w. , hempel m. , 2011 , apj , 733 , l43 murakami h. , et al . , 2007 , pasj , 59 , 369 najita j. , carr j. s. , glassgold a. e. , shu f. h. , tokunaga a. t. , 1996 , apj , 462 , 919 nakano s. , sato h. , nishimura h. , nakamura t. , wakuda s. , yamaoka h. , pearce a. , 2003 , iauc , 8190 , 1 olivier e. a. , whitelock p. , marang f. , 2001 , mnras , 326 , 490 ortiz r. , maciel w. j. , 1996 , a&a , 313 , 180 parihar p. , messina s. , distefano e. , shantikumar n. s. , medhi b. j. , 2009 , mnras , 400 , 603 persi p. , tapia m. , gmez m. , whitney b. a. , marenzi a. r. , roth m. , 2007 , aj , 133 , 1690 porter j. m. , rivinius t. , 2003 , pasp , 115 , 1153 quireza c. , rood r. t. , bania t. m. , balser d. s. , maciel w. j. , 2006 , apj , 653 , 1226 reed b. c. , 2000 , aj , 120 , 314 reipurth b. , aspin c. , 1997 , aj , 114 , 2700 reipurth b. , aspin c. , 2004 , apj , 606 , l119 reipurth b. , aspin c. , 2010 , vaoa.conf , 19 rice t. s. , wolk s. j. , aspin c. , 2012 , apj , 755 , 65 robitaille t. p. , whitney b. a. , indebetouw r. , wood k. , 2007 , apjs , 169 , 328 robitaille t. p. , et al .
, 2008 , aj , 136 , 2413 romanova m. m. , ustyugova g. v. , koldoba a. v. , lovelace r. v. e. , 2013 , mnras , 430 , 699 russeil d. , 2003 , a&a , 397 , 133 saito r. k. , et al . , 2012 , a&a , 537 , a107 saito r. k. , et al . , 2013 , a&a , 554 , a123 samus n. n. , kazarovets e. v. , kireeva n. n. , pastukhova e. n. , durlevich o. v. , 2010 , oap , 23 , 102 scholz a. , xu x. , jayawardhana r. , wood k. , eislffel j. , quinn c. , 2009 , mnras , 398 , 873 scholz a. , 2012 , mnras , 420 , 1495 scholz a. , froebrich d. , wood k. , 2013 , mnras , 430 , 2910 scoville n. z. , yun m. s. , sanders d. b. , clemens d. p. , waller w. h. , 1987 , apjs , 63 , 821 skrutskie m. f. , et al . , 2006 ,
aj , 131 , 1163 soszyski i. , et al . , 2005 , aca , 55 , 331 stamatellos d. , whitworth a. p. , hubber d. a. , 2012 , mnras , 427 , 1182 tisserand p. , et al . , 2004 , a&a , 424 , 245 van loon j. t. , zijlstra a. a. , whitelock p. a. , waters l. b. f. m. , loup c. , trams n. r. , 1997 , a&a , 325 , 585 varricatt w. p. , davis c. j. , adamson a. j. , 2005 , mnras , 359 , 2 varricatt w. p. , davis c. j. , ramsay s. , todd s. p. , 2010 ,
mnras , 404 , 661 vijh u. p. , et al . , 2009 , aj , 137 , 3139 volk k. , kwok s. , langill p. p. , 1992 ,
apj , 391 , 285 wallace l. , hinkle k. , 1997 , apjs , 111 , 445 whitelock p. , feast m. , catchpole r. , 1991 , mnras , 248 , 276 whitelock p. , feast m. , 2000 , mnras , 319 , 759 whitelock p. a. , feast m. w. , van leeuwen f. , 2008 , mnras , 386 , 313 whitney b. , et al . , 2011 ,
aas , 43,#241.16 witham a. r. , knigge c. , drew j. e. , greimel r. , steeghs d. , gnsicke b. t. , groot p. j. , mampaso a. , 2008 , mnras , 384 , 1277 wright e. l. , et al .
, 2010 , aj , 140 , 1868 zhang b. , zheng x. w. , reid m. j. , menten k. m. , xu y. , moscadelli l. , brunthaler a. , 2009 , apj , 693 , 419
the spectral energy distributions for the 17 objects from dr7 associated with sfrs ( section [ sec : conc_ext ] ) are presented in figure [ app : ysos_seds1 ] .
in addition we present the seds for the three objects that have similar characteristics to ysos in colour - colour plots but are not located near a known star formation region .
these are gpsv36 , 40 and 43 .
the seds are constructed using the 2 epoch ukidss gps dr7 information ( 2005,2008 ) and public data , when available , from the spitzer legacy survey , cygnus - x ( 2007 - 2008 observations ) , and/or wise ( 2010 observations ) .
some of the objects that lack mid - infrared data , or for which an infrared excess is not clear , are compared to an m5v ( 3240 k ) @xcite model atmosphere , and the same model reddened to @xmath67=10 and 20 mag .
@xmath67=20 is near the upper end of the possible interstellar extinction column values for any of the variables ( determined in @xmath992.2 using the local cmd for the field around each source ) .
the fact that the near- and mid - infrared observations are not contemporaneous does not allow to clearly classify some of the objects as ysos .
this is observed in gpsv32 where if we only take into account the first epoch of gps and the spitzer data , its sed resembles that of an m - dwarf with @xmath206 mag of extinction .
this changes if we take the second gps epoch and spitzer data , where the sed would point to an infrared excess .
the same issue arises in gpsv29 .
it is also difficult to classify as ysos objects in our sample without mid - infrared data . in every case
, these objects could point to either infrared excess ( not being observed ) or correspond to highly extinct m - dwarfs . as an example we compare the object gpsv34 ( a spectroscopically confirmed yso , contreras pea et al .
, in prep . ) with the same @xcite model atmospheres in figure [ app : ysos_seds1 ] .
we see that the near infrared data from the second gps epoch is well fitted by a m - dwarf with @xmath206 mag of extinction .
however , the mid - infrared data allows us to confirm its probable yso nature .
two of the objects with gps data only are confirmed via its spectra in the analysis to be presented in an upcoming study ( contreras pea et al .
, in prep . )
, these are gpsv19 and gpsv22 . even though we can not confirm every object in sfr as a yso , the seds shown here and the spectral characteristics for a sample of them ( contreras pea et al .
, in prep . )
point to such classification for the majority of them .
this supports our conclusion that finding high amplitude variables close to areas of star formation is sufficient evidence that most of these stars are ysos . | we present initial results of the first panoramic search for high - amplitude near - infrared variability in the galactic plane .
we analyse the widely separated two - epoch k - band photometry in the 5th and 7th data releases of the ukidss galactic plane survey .
we find 45 stars with @xmath0 mag , including 2 previously known oh / ir stars and a nova . even though the mid - plane is not yet included in the dataset , we find the majority ( 66@xmath1 ) of our sample to be within known star forming regions ( sfrs ) , with two large concentrations in the serpens ob2 association ( 11 stars ) and the cygnus - x complex ( 12 stars ) .
sources in sfrs show spectral energy distributions ( seds ) that support classification as young stellar objects ( ysos ) .
this indicates that ysos dominate the galactic population of high amplitude infrared variable stars at low luminosities and therefore likely dominate the total high amplitude population .
spectroscopic follow up of the dr5 sample shows at least four stars with clear characteristics of eruptive pre - main - sequence variables , two of which are deeply embedded .
our results support the recent concept of eruptive variability comprising a continuum of outburst events with different timescales and luminosities , but triggered by a similar physical mechanism involving unsteady accretion . also , we find what appears to be one of the most variable classical be stars .
[ firstpage ] infrared : stars stars : low - mass stars : pre - main - sequence stars : agb and post - agb stars : protostars stars : variables : t tauri , herbig ae / be . |
the quantum mechanical geometric phase was discovered by berry in 198384@xcite .
the context was unitary evolution governed by the schrdinger equation in the adiabatic approximation , i.e. , with a hermitian hamiltonian possessing a ` gentle ' time - dependence . assuming that as an operator the hamiltonian is cyclic , i.e. , it returns to its original form after a certain interval of time ( during which there are no level crossings ) , the approximate solutions to the schrdinger equation are also cyclic .
the geometric phase is then seen explicitly in these solutions at the end of the cycle . immediately after berry s discovery , it was pointed out by barry simon@xcite that the geometric phase expresses the non - integrability , or anholonomy , of a natural ` rule of parallel transport ' ( a connection ) in a principal fibre bundle , with structure group @xmath0 , which occurs in the framework of quantum mechanics .
this was therefore a characterisation of this phase in the language of differential geometry .
the ensuing years saw two streams of work relating to the geometric phase .
one consisted of extensions of berry s original work , in the sense of relaxing the conditions under which the phase is definable . the other consisted of interesting earlier results which could be reinterpreted as instances of this phase , and so as precursors to it .
we recall three significant efforts of the first kind .
aharonov and anandan@xcite showed that the adiabatic condition is not necessary given a cyclic solution to the schrdinger equation involving any ( time - dependent ) hamiltonian , one can reconstruct a corresponding geometric phase .
this was followed by the work of samuel and bhandari@xcite , in which the cyclic condition on a solution was also dispensed with .
given a solution of the schrdinger equation involving any ( time - dependent ) hamiltonian , over any stretch of time , one can in a simple way extend it to a closed or cyclic solution , then use the aharonov - anandan method to identify a geometric phase . both these extensions of berry s original framework used quantum mechanical notions , specifically the schrdinger equation .
the third step in the direction of increasing generality was taken by mukunda and simon@xcite : the geometric phase is entirely kinematical in content , not requiring a hamiltonian operator and associated schrdinger equation .
it is determined once one is given a ( sufficiently smooth ) curve of unit vectors in any complex hilbert space , without reference to any specifically quantum mechanical notions .
( the relevant expressions and definitions are recalled below ) . turning to the efforts of the second kind , within quantum mechanics we may cite the bohm - aharonov effect ( already dealt with by berry in his original work ) , and the clarification of the connection between bargmann invariants and geometric phases@xcite . beyond these , it is interesting that many instances of the geometric phase have been identified within classical ( wave ) optics the gouy phase from 1890@xcite ; the work by rytov , and vladimirskii@xcite in 1938 and 1941 on the behavior of light polarisation in the short wavelength limit of wave optics ; and pancharatnam s studies in 1956@xcite on light polarisation and an associated phase .
these classical optics examples will be reviewed briefly in the next section .
it will be seen that even in these situations the basic idea of a complex linear vector space carrying a hermitian inner product , usually regarded as characteristic of quantum mechanics , is essential to be able to identify ` classical geometric phases ' . in this way , the deep link between bargmann invariants and geometric phases , and the connection to berry s original discovery , are always kept in evidence . with this background ,
we now describe briefly the kinematic approach to the geometric phase .
let @xmath1 be a complex hilbert space of any dimension , with vectors @xmath2 and inner product @xmath3@xcite . in quantum mechanics @xmath4
is related to a probability ; in classical wave optics @xmath5 generally stands for light intensity which can , but need not be , normalised .
we next denote by @xmath6 the unit sphere in @xmath1 : @xmath7 the group @xmath0 of complex phase factors acts on @xmath6 ( also on @xmath1 ) in a natural way : @xmath8 the quotient @xmath9 , i.e. , collections or equivalence classes of vectors @xmath10 differing only by phases , forms the ` ray space ' @xmath11 : @xmath12 whereas @xmath6 is a subset of @xmath1 , _ the ray space @xmath13 is not _
: the @xmath14 relationship is that there is a projection map @xmath15 from the former to the latter : @xmath16 referring to an earlier comment , @xmath6 is a @xmath0 principal fibre bundle over the base @xmath11 . in the quantum mechanics context , points in @xmath11 correspond one - to - one to physical pure states . in this framework , given any continuous piecewise once differentiable parametrised curve @xmath17 in @xmath6 , @xmath18 with image @xmath19 in @xmath11 , @xmath20 \nonumber\\ & = \{\rho(s ) = \rho(\psi(s ) ) = \psi(s)\psi(s)^\dag\,|\ , s_1\le s \le s_2\}\subset\mathcal{r}\,,\end{aligned}\ ] ] the geometric phase @xmath21 $ ] is defined : @xmath22 & = \varphi_{\rm tot}[\mathcal{c } ] -\varphi_{\rm dyn}[\mathcal{c}]\,,\nonumber\\ \varphi_{\rm tot}[\mathcal{c } ] & = \arg(\psi(s_1),\psi(s_2)),\nonumber\\ \varphi_{\rm dyn}[\mathcal{c } ] & = \text{im } \int_{s_1}^{s_2 } ds \left(\psi(s ) , \frac{d\psi(s ) } { ds}\right)\,.\end{aligned}\ ] ] as indicated , @xmath21 $ ] is a functional of @xmath23 , while @xmath24 and @xmath25 are both functionals of @xmath26 .
an important consequence of this definition is a result involving the so - called bargmann invariants@xcite .
the simplest such invariant involves three pairwise nonorthogonal vectors @xmath27 and is the expression @xmath28 which is in general complex . the fact that @xmath29 = @xmath30 for all real @xmath31 shows that @xmath32 lives in @xmath13 rather than in @xmath33 .
indeed bargmann , during the course of his famous proof of wigner s theorem , introduced @xmath32 simply to point out this gauge - invariance property and to indicate that it could be used to distinguish between unitaries and antiunitaries : while @xmath32 is invariant under unitaries , its argument changes signature under antiunitaries .
it is in @xcite that this object introduced by bargmann almost in passing was elevated to become the basis of a complete kinematic theory of geometric phase .
it is worth emphasizing that the framework described above , based on the triplet of spaces @xmath44 , supports the geometric phase concept in a simple and direct way .
though suggested by the formal ( complex linear space ) structure of quantum mechanics , it can be used in other situations as well , such as classical wave optics .
we adopt this viewpoint in trying to define geometric phases in various physical , particularly classical optical , situations .
before we outline the organization of the material of this paper , it may be useful to add an extended remark by way of pointing to the precise context of this work .
there has been considerable interest in recent times to understand the interplay between the spatial degree of freedom ( coherence ) and polarisation degree of _ electromagnetic _ beams [ see , for instance , @xcite and references therein ] .
it is equally important to understand the behaviour of this interplay as the maxwell beam passes through an optical system .
indeed , it turns out that the _ defining _ properties of the age old mueller matrix can not be correctly enumerated without consideration of this interplay or entanglement@xcite .
a lens of focal length @xmath45 relates the output field amplitude @xmath46 ( just after the lens plane ) to the input @xmath47 ( just before the lens plane ) , @xmath48 being cartesian variables in the transverse plane , through @xmath49 but when it comes to vector waves , it is clear that the same transformation applied to every ( cartesian ) component of the electric field vector @xmath50 will not map solutions of maxwell equations at the input plane to solutions at the output , for such a democratic action on the electric field components _ does not _ respect the transversality condition @xmath51 . since this condition is a _ constraint _ connecting the spatial degrees of freedom to the polarization degree , it would be respected only if the spatial modulation @xmath52 is accompanied by ` appropriate ' _ local rotations _ of the electric field components ( local polarization)@xcite .
let us arrange the components of the electric and magnetic field amplitude vectors @xmath50 , @xmath53 in a transverse plane @xmath54 = constant into a six - component electromagnetic vector @xmath55 the approach of@xcite rooted at the very poincar symmetry of the maxwell system of equations led to this fundamental result : if @xmath56 is the amplitude transmittance function of an optical system in scalar fourier optics [ a lens , for instance , has @xmath57 , then the action @xmath58 where @xmath59 are a pair of @xmath60 _ numerical _ matrices arising from the structure of the poincar group@xcite , _ does take _
solutions @xmath61 of maxwell s equations to solutions @xmath62 .
that is , the matrices @xmath59 effect on the components of @xmath63 the correct local rotations alluded to above@xcite .
this result readily leads to fourier optics for maxwell beams@xcite and to electromagnetic gaussian beams@xcite , resulting in a straight forward description of not only the longitudinal component but also the cross - polarisation component@xcite .
it is in respect of this result that the late henri bacry anticipated : `` it is highly probable that a rigorously gauge theory will be developed in a near future '' , the local rotations referred to above constituting `` an @xmath64 gauge group''@xcite .
the work presented here is only the first step of an ambitious programme constituting our attempt towards a possible realization of this anticipation . while the earlier formulation of fourier optics for maxwell beams@xcite concentrated on paraxial propagation about a _
fixed direction _
, the present work aims at laying a global and structurally robust skeleton in the _ space of directions _ , handling satisfactorily the well known topological obstructions . in the sequel
, we plan to adapt suitably the methods of @xcite in local patches of the space of directions , and then ` stitch ' together the patches in a smooth manner to arrive at the general case .
the contents of this paper are organised as follows .
section ii gives brief accounts of three applications of the geometric phase concept to classical optical situations : the gouy phase in scalar paraxial wave optics ; the pancharatnam study of phases in pure polarization optics with fixed propagation direction ; and the behaviour of polarisation in the eikonal or ray limit of maxwell s equations in a transparent medium with given refractive index function .
the pancharatnam case uses the poincar sphere @xmath65 of polarisation states , for a fixed direction of propagation , while the ray case uses the sphere of propagation directions @xmath66 . in all these cases ,
the use of the basic quantum mechanical framework is highlighted .
section iii builds on the last example of section ii in two ways the generalisation from the unphysical case of a single ray to a physical beam of finite cross - sectional area made up of a narrow bundle of nearly parallel rays ; and the inclusion of polarization gadgets in the path of the beam .
once again the quantum mechanical framework proves adequate , and now both spheres @xmath65 , @xmath66 come into the picture .
section iv takes up certain global features of the sphere of directions @xmath66 , and builds on a recent suggestion@xcite that passage to the complex extension of the tangent planes to @xmath66 removes an obstruction which exists in the real domain .
( the work of@xcite was motivated in part by earlier works of@xcite . ) using elementary group theoretical arguments , based on the groups @xmath64 and @xmath67 , a particularly simple global basis of complex orthonormal vector fields tangent to @xmath66 is constructed .
section v uses the constructions of section iv to study again the beams of section iii and their geometric phases : in appropriate situations , the complete geometric phase separates into a contribution from @xmath65 and another from @xmath66 .
the final section vi contains some concluding remarks , while the appendix compares the present framework for handling geometric phases with that proposed in @xcite .
in this section we review three situations in classical optics displaying geometric phases , presenting only the essential details .
the first concerns scalar wave optics , the other two include polarisation .
quantum mechanical notation is used when convenient@xcite .
we deal with the scalar optical wave field in free space , with fixed ( angular ) frequency @xmath68 , wave number @xmath69 and wavelength @xmath70 . in the paraxial approximation to the helmholtz equation , with the positive @xmath54-axis as the propagation direction
, we obtain the paraxial wave equation in two transverse dimensions : @xmath71 where @xmath72 , reminiscent of @xmath73 .
( the exponential factor @xmath74 has been omitted in @xmath75 ) .
this is formally similar to the schrdinger equation in quantum mechanics for a free nonrelativistic particle of unit mass in two dimensions , with @xmath76 in place of @xmath73 , and with the longitudinal variable z playing the role of ` time ' .
the `` hamiltonian operator '' @xmath77 for eq.([eq2.1 ] ) is @xmath78 if we restrict to one transverse dimension , we have the simpler paraxial wave equation @xmath79 our focus is on ` centred ' gaussian solutions to this paraxial wave equation , and their phases . based on group theoretical considerations , it is convenient to parametrise normalised centred gaussians by a complex variable @xmath80 with negative imaginary part , i.e. , lying in the lower half complex plane : @xmath81 then the centred gaussian solution to eq.([eq2.3 ] ) , with width @xmath82 in the ` waist ' plane @xmath83 , is : @xmath84 here , @xmath85 is the evolving gouy phase .
it is the argument of @xmath86 ( on - axis phase at the plane @xmath87 constant ) and ` jumps ' by @xmath88 ( by @xmath89 for @xmath90 transverse dimensions ) across the waist plane : @xmath91 that the parameter @xmath80 lives in the lower half plane is a consequence of our taking monochromatic time - dependence in the usual form @xmath92 .
had it been taken in the form @xmath93 , as some authors do , then @xmath80 would live in the upper half plane .
the evolution of gaussian beams through first order systems described by abcd - matrix is governed by the well known kogelnik abcd - law@xcite , @xmath94 of which the particular case @xmath95 , corresponding to free propagation from @xmath96 to @xmath97 [ i.e. , @xmath98 , is already quoted in eq .. it may be noted in passing that the abcd - law has been generalized to partially coherent gaussian beams , the so - called gaussian schell - model beams , in@xcite and to arbitrary beams in@xcite .
the argument of @xmath129 on the right hand side in the result ( [ eq2.14 ] ) is a quadrilateral in the ` ray space ' @xmath11 , with geodesic sides , as it should be .
the geodesics needed here are to be constructed in the manner of eq.([eq1.9 ] ) at the vector space or wave amplitude level , followed by projection @xmath15 to @xmath11 . a quite subtle analysis@xcite ( here omitted )
shows that in the present instance ( and some others of interest ) we can use ` geodesics ' drawn within the manifold of centred gaussian amplitudes ( which may be called ` constrained geodesics ' ) , and the basic connection ( [ eq1.11 ] ) , ( [ eq2.7 ] ) between bargmann invariants and geometric phases continues to be valid .
next , as the definition ( [ eq1.7 ] ) of geometric phases shows , for practical calculations one can choose any convenient ` lift ' @xmath17 of @xmath19 at the level of hilbert space vectors or wave amplitudes , obeying @xmath130 $ ] . in particular if @xmath17 is chosen to be a closed loop ( the quadrilateral @xmath131 in @xmath11 is of course closed ) the piece @xmath132 $ ] in eq.([eq1.7 ] ) vanishes and we are left with @xmath21=-\varphi_{\rm dyn}[\mathcal{c}]$ ] . beyond this
, one can use the phase freedom at each point along @xmath17 to assume , in the present case , that @xmath17 is a closed loop within the space of centred gaussian wave functions @xmath124 .
it can then be pictured or drawn as a closed curve in the lower half of the complex @xmath80 plane .
one must only ensure that the ` vertices ' are chosen properly , so as to project onto the vertices specified in @xmath11 in ( [ eq2.14 ] ) , and the connecting curves represent ` constrained geodesics ' properly . when all this is done , the result is as shown in fig.1 .
alternatively , the line integral in eq .
equals@xcite the negative of one - fourth of the ( hyberbolic ) area of the enclosed quadrilateral , the abcd - law being a signature of the natural lobachevskian hyperbolic geometry with metric @xmath147 underlying the manifold of gaussian states , the lower half @xmath80-plane .
the area itself is given by the ` hyperbolic deficiency ' which , for a ( geodesic ) quadrilateral , equals @xmath148 minus sum of the interior angles .
the interior angle vanishes at @xmath149 as well as at @xmath150 . for the other two angles @xmath151 we see from fig.1 that @xmath152 and @xmath153 .
thus the deficiency equals @xmath154 , leading to a geometric phase of @xmath155 .
one thus recovers @xmath156 well known in the context of laser beams , now as a geometric phase .
we appreciate that this demonstration of the link between gouy and geometric phases is fully within the @xmath157 framework of quantum mechanics used in the quantum kinematic approach@xcite to geometric phase , briefly recapitulated in section 1 .
now we include the polarization degree of freedom , and to begin with consider the extreme case when it is the only variable .
with given frequency @xmath68 and wave number @xmath69 , we _ fix also the direction of propagation _ to be the positive @xmath54-axis , and consider plane waves in various states of pure polarization .
the analysis again falls perfectly into the quantum mechanical @xmath157 scheme .
dropping the standard factor @xmath74 , at each @xmath54 the electric field is a complex two - component vector in the transverse x - y plane , @xmath158 .
the hilbert space @xmath1 for this case is then @xmath159 of dimension two . as is well known , the spaces @xmath6 and @xmath11 are @xmath160 and @xmath65 respectively , unit spheres in real four and three - dimensional euclidean spaces , the latter being the poincar sphere of pure polarization states : @xmath161 given @xmath162 at some @xmath54 , the corresponding pure polarization state is represented by a point @xmath163 computed as follows : @xmath164 where the @xmath165 s are the standard quantum mechanical pauli matrices . under free propagation governed by the free maxwell equations , the amplitude @xmath162 and the polarization state @xmath166 are both constant : @xmath166 is stationary on @xmath65 .
more generally , we imagine the plane wave passing through transparent linear intensity preserving polarization gadgets which act on @xmath162 and alter the polarization state @xmath166 .
these are placed at various locations ( lumped ) or over various stretches ( distributed ) along the @xmath54-axis , separated from one another by intervals of free propagation .
the effect of such gadgets on @xmath162 is again governed by maxwell s equations for propagation of the field through suitable transparent material media .
as our interest is only in the behaviour of the polarization state @xmath166 , the intensity being held constant , we can represent each polarization gadget by a corresponding element of the two - dimensional unitary unimodular group @xmath67@xcite the additional @xmath0 phase in the full unitary group @xmath167 is not relevant for this purpose . with this physical picture in place , let us write @xmath168 and @xmath169 for the field and the polarization state at position @xmath54 along the propagation axis : @xmath170 then @xmath168 evolves according to the schrdinger - like equation @xmath171 where @xmath172 is a real three - dimensional vector and @xmath173 is the ` hamiltonian '
. correspondingly for @xmath169 we have @xmath174 thus while @xmath168 undergoes a gradually unfolding @xmath67 transformation , @xmath169 experiences a gradual rotation belonging to @xmath64@xcite .
( free propagation stretches correspond to @xmath175 , and hence to @xmath176 ; for lumped elements like a quarter or half wave plate @xmath172 is a dirac delta function ) . over a finite stretch @xmath96 to @xmath97
, we have : @xmath177 with @xmath178 determining @xmath179 through the well known @xmath180 homomorphism@xcite .
if @xmath173 and @xmath172 are constant from @xmath96 to @xmath97 , say @xmath77 and @xmath181 respectively , we have @xmath182 where @xmath183 is the right handed rotation about axis @xmath184 by amount @xmath185@xcite .
then over such a stretch @xmath186 and as @xmath187 constant as well , we see that @xmath169 moves on a latitude circle in a plane perpendicular to @xmath181 . in case @xmath188 ,
@xmath169 moves on the great circle perpendicular to @xmath184 , the equator with respect to @xmath184 ; and as then @xmath189 , _ such stretches contribute zero dynamical phases_. a cyclic evolution in this pancharatnam situation carries the electric field over some curve @xmath190 ( assuming for simplicity @xmath191 ) , say from @xmath192 at @xmath96 to @xmath193 at @xmath97 .
then @xmath169 describes a closed loop @xmath194 . by eq.([eq1.7 ] ) , the associated geometric phase can be readily computed , and it turns out to be very simply related to the geometry of @xmath65 : @xmath195 & = \varphi_{\rm tot}[\mathcal{c}]-\varphi_{\rm dyn}[\mathcal{c}]\nonumber\\ & = \arg \boldsymbol{e}^{(1)^\dag } \boldsymbol{e}^{(2)}-\text{im } \int_{z_1}^{z_2 } dz\,\boldsymbol{e}(z)^\dag \frac{d\boldsymbol{e}(z ) } { dz}\nonumber\\ & = \theta + \int_{z_1}^{z_2 } dz\ , \boldsymbol{e}(z)^\dag h(z)\boldsymbol{e}(z)\nonumber\\ & = \frac{1}{2 } \omega[c_{\rm pol}]\,,\end{aligned}\ ] ] where @xmath196 $ ] is the solid angle ( with sign ) subtended by @xmath197 at the origin of @xmath65 .
the third situation we consider from the geometric phase perspective is one that has been studied for a long time on account of its obvious physical relevance .
it is the short wave length or eikonal or ray
limit of maxwell s equations , leading to differential equations for light rays in a given transparent medium , plus the law for evolution of the electric field along them@xcite .
we first recall the basic equations resulting from the eikonal limit , then some important previous work , and finally consider the situation from the geometric phase perspective . in comparison to the previous pancharatnam case , in the eikonal limit the propagation ( ray ) direction
is allowed to vary while , in a sense to be clarified later , the polarisation state stays constant .
we consider maxwell s equations for propagating electric ( and magnetic ) fields in a transparent nonconducting non - magnetic material medium characterised by a time - independent isotropic refractive index function @xmath203 . to leading order
, the eikonal limit gives a system of second order ordinary differential equations whose solutions are rays in the medium : @xmath204 each solution @xmath205 ( for given initial conditions ) determines a ray @xmath206 , a curve in physical three - dimensional euclidean space .
here @xmath101 is arc length measured along @xmath206 from some starting point on @xmath206 .
_ we hereafter work with some definite @xmath206_. as a space curve , @xmath206 is characterized by the following vectors and scalars defined pointwise along it , the dot denoting derivative with respect to @xmath101 : @xmath207 at each @xmath208 , @xmath209 is a right handed orthonormal triad , unique at generic points with nonzero curvature ; it is locally determined by @xmath210 and @xmath211 .
formally these vectors obey the ` equations of motion ' @xmath212 the first equation ( which is actually trivial ) means that @xmath213 obeys the minimal fermi - walker transport law@xcite , while @xmath214 and @xmath215 do not do so .
next we consider the evolution of the electric field @xmath216 along @xmath206 .
this comes from the next to leading order terms in the eikonal limit of maxwell s equations , and when expressed in terms of the normalised electric field @xmath217 we have again the fermi - walker transport law along with the transversality condition : @xmath218 that both @xmath219 and @xmath217 obey the fermi - walker law is consistent with the need to maintain the transversality condition @xmath220 along @xmath206 . at each @xmath208 , @xmath221 and @xmath222
span the transverse plane perpendicular to @xmath219 there .
if we introduce another orthonormal basis in this plane , @xmath223 , @xmath224 , obeying the fermi - walker transport law like @xmath217 , we have the evolution equations @xmath225 as initial condition we take @xmath226 at some @xmath227 .
then the pair @xmath228 rotates steadily with respect to the pair @xmath229 at a rate given by the torsion : @xmath230 now @xmath231 are constants along @xmath206 : @xmath232 all the three - dimensional vectors @xmath233 have corresponding components with respect to some fixed global cartesian frame in space .
the representation ( [ eq2.34 ] ) identifies @xmath217 at each @xmath208 with a ` vector ' @xmath234 in the two - dimensional complex linear space @xmath235 . using this we can represent the polarization state at @xmath208 by a point @xmath236 on the poincar sphere @xmath65 : @xmath237 as long as no polarization gadgets are placed anywhere on @xmath206 , the @xmath238 are constants , so the polarization state represented by @xmath239 is also constant :
only the propagation direction @xmath219 varies .
this is to be compared with the pancharatnam situation : under free propagation , both propagation direction @xmath240 and polarization state @xmath163 are constant .
if polarization gadgets are placed along the axis , @xmath240 ( by definition ) stays constant , while @xmath169 moves on @xmath65 . in the present context
we can say the cyclic case occurs when the choice of a ` later ' point @xmath241 is such that @xmath242 , @xmath243 , @xmath244 are the same as @xmath245 , @xmath246 , @xmath247 respectively at the initial point @xmath248 .
this happens if @xmath249 the behaviours of input linear and circular polarizations are then particularly simple .
the linear case corresponds to real @xmath238 ; then @xmath250 is a vector in space with real cartesian components all along @xmath206 . from eqs.([eq2.32],[eq2.33 ] ) we can relate @xmath251 to @xmath252 as follows : @xmath253 the two transverse planes at @xmath254 , @xmath255 on @xmath206 are parallel to one another , and @xmath251 is obtained from @xmath256 by a right handed rotation by angle @xmath257 about @xmath245 .
a detailed calculation shows that @xmath257 has a geometrical meaning . over the range @xmath258 , the unit tangent @xmath219 to @xmath206 describes a _ closed loop _
@xmath259 on the sphere of directions @xmath66 : @xmath260 then we have the result that the integrated torsion is ( the negative of ) the _ solid angle
_ subtended by @xmath259 at the centre of @xmath66 : @xmath261\,.\end{aligned}\ ] ] in the cases of circular polarisations , we get phase shifts rather than a rotation in space . in these cases , @xmath217 is a complex three - vector at all @xmath205 on @xmath206 : @xmath262}\boldsymbol{\psi}(s_1)\,.\end{aligned}\ ] ] these results on the behaviours of polarization in the ray limit of maxwell s equations were obtained very early by rytov and by vladimirskii@xcite .
in particular , rytov showed that the phase difference between rcp and lcp evolves at a rate proportional to the torsion ; while vladimirskii showed that the spatial rotation experienced in the cyclic case for linear polarization is essentially by the solid angle @xmath263 $ ] . to cast the above discussion into the geometric phase format of section i
, it is useful to write the evolution equation ( [ eq2.30 ] ) for the ( normalised ) electric field in a schrdinger - like form with a suitable hermitian hamiltonian operator .
we view @xmath217 ( referred to axes fixed in space ) as a ( normalised ) element of @xmath264 , which is the hilbert space in the present context , and find : @xmath265 thus @xmath266 is a pure imaginary antisymmetric @xmath267 matrix .
the transversality condition @xmath268 is to be added as a constraint consistent with the evolution .
the definition ( [ eq1.7 ] ) allows us to define a geometric phase for any @xmath269 and @xmath270 , and we find that _ due to transversality the dynamical phase always vanishes_. bringing in the spaces @xmath271 and @xmath272 , the complex two - dimensional projective space appropriate to @xmath264 , we have : @xmath273 & = c
\subset r_4\,:\nonumber\\ \varphi_g[c ] & = \varphi_{\rm tot}[\mathcal{c}]-\varphi_{\rm dyn}[\mathcal{c}]\,,\nonumber\\ \varphi_{\rm tot}[\mathcal{c } ] & = \arg(\boldsymbol{\psi}(s_1)^\dag \boldsymbol{\psi}(s_2))\,,\nonumber\\ \varphi_{\rm dyn}[\mathcal{c}]&= \text{im } \int_{s_1}^{s_2 } ds\,\boldsymbol{\psi}(s)^\dag \frac{d\boldsymbol{\psi}(s ) } { ds } \nonumber\\ & = \text{im}\left(-i \int_{s_1}^{s_2 } ds \,\boldsymbol{\psi}(s)^\dag h(s ) \boldsymbol{\psi}(s)\right)=0,\,\nonumber\\ \text{i.e.},\qquad\quad \varphi_g[c ] & = \varphi_{\rm tot}[\mathcal{c}]\,.\end{aligned}\ ] ] here we recognize that @xmath17 _ can not be drawn freely in @xmath274 because of the transversality condition , so in this way it is constrained by _ @xmath206 . to illustrate the above ,
let us quote some particular cases in a table , eq.(2.46 ) : + @xmath275}\\ \hline \text{free } & \text{linear } & \text{real } & 0 ~\text{or}~
\pi\\ \gamma~ \text{cyclic , eq.\,(\ref{eq2.36 } ) } & \text{rcp / lcp } & ~~\boldsymbol{\psi}(s_2)=e^{\mp i\omega[c_{\rm dir}]}\boldsymbol{\psi}(s_1)~~ & ~\;\mp \,\omega[c_{\rm dir}]~\;\\ \hline \end{array}\ ] ] the distinction between @xmath276 and @xmath277 should be kept in mind .
the brief reviews presented in the previous section show that the pancharatnam situation and the ray optic limit are mutually complementary . in the former only the polarization state changes , while in the latter only the propagation direction changes .
now we try to cover the ( important ) middle ground between them .
we endeavour to build up a physical picture , based ultimately on maxwell s equations , with the motivation to arrive at geometric phases in the framework of section i. as in the eikonal limit , we consider light traveling through a transparent non - magnetic stationary medium with refractive index @xmath203 .
we recall that the concept of a single ray is not physically meaningful and can not be realised .
the eikonal limit of maxwell s equations leads at first to a first order partial differential equation in three - dimensional space for the eikonal , a function @xmath278 .
a particular eikonal @xmath278 leads to a continuous family or succession of wave fronts over each of which @xmath278 is constant , and which taken together cover some region of physical space .
rays are then lines drawn in this region , orthogonal at each point to the wave front passing through that point .
these rays are solutions to eq.([eq2.27 ] ) .
thus one eikonal @xmath278 determines a corresponding succession of wave fronts and in turn one family @xmath279 of rays .
there is only one wavefront , and only one ray belonging to @xmath279 , through each point in the relevant region .
it is in this sense that single rays are not directly physically realisable .
the best that we can do is to consider a narrow or well collimated ( i.e. , nearly parallel ) bundle of nearby rays with some nonzero cross - sectional area which may vary along the bundle [ consequently the wavefronts along the bundle , correspondingly limited in their spatial extent , are nearly planar ] . calling this a beam , at each ` point ' along it we have some finite spatially limited wavefront . in this picture
we have in mind some @xmath206 obeying eq.([eq2.27 ] ) acting as the ` backbone ' of the beam . at each location @xmath280
, we have a propagation direction @xmath219 , a spatially limited ` plane wave ' perpendicular to @xmath219 , and a transverse electric field @xmath281 .
thus we arrive at a physical picture of a continuous succession of limited plane waves each at a spatial location @xmath205 , with propagation direction @xmath219 and in some polarization state .
now we can go a step further and allow the wavelength to be finite , as long as it is much smaller than all other physically relevant dimensions , including the linear dimensions of the limited plane wave elements . in this way we motivate the passage from a physically unrealisable ray to a realisable beam by a process of ` thickening ' of the former . in the sequel
, the spatial locations @xmath205 of successive plane wave elements of the beam will sometimes be omitted .
the parameter @xmath101 continues to be distance measured along the beam from some initial point , increasing at each location in the direction of @xmath219 . for a beam propagating ` freely ' in the medium in this way , the evolution equation for @xmath217 is eq.([eq2.30 ] )
this , as we have seen , is a consequence of maxwell s equations in the medium , and can be put into the schrdinger - like form ( [ eq2.41 ] ) with a hermitian hamiltonian operator . the solution eq.([eq2.34 ] ) with constant @xmath234 implies a constant polarization state @xmath282 given in eq.([eq2.35 ] ) .
we can now go another step further and imagine placing various polarisation gadgets over ( short ) stretches of the beam , equivalently of @xmath206 , where @xmath234 varies as function of @xmath101 , governed by a ` polarisation hamiltonian ' as in eq.([eq2.20 ] ) .
thus we arrive at new evolution equations for @xmath217 based on the following ingredients : @xmath283 we have now written @xmath284 for the ` direction ' part of the hamiltonian , appearing in eq.([eq2.41 ] ) ; it is completely determined by the local geometrical properties of @xmath206 .
the other contribution to the evolution of @xmath217 is from the ` polarization ' part of the hamiltonian , as in eq.([eq2.20 ] ) , written now as @xmath285 .
this controls the evolution of the local two - component transverse description of @xmath217 resolved along @xmath223 . the complete evolution equation for @xmath217 is easily found to be schrdinger - like , with a hamiltonian which is a ( complex ) hermitian @xmath267 matrix : @xmath286 the implied evolution equation for @xmath287 is as in eq.([eq2.21 ] ) : @xmath288 over portions of the beam free of polarization gadgets , where @xmath289
, the local properties of @xmath206 determine the propagation , and the polarization state is constant .
passage through gadgets leads to changing @xmath290 and @xmath291 .
both kinds of changes in @xmath217 are ultimately traced back to maxwells equations ; and the complete evolution equation @xmath292 respects the transversality condition @xmath293 . in all of this ,
the separation of effects due to change in beam direction and those due to presence of polarisation gadgets , is essentially unambiguous .
let us now bring in geometric phase considerations .
as in the ray case in section ii(c ) , we are able to use the basic quantum mechanical @xmath157 framework with @xmath264 , @xmath294 , and @xmath295 which is of real dimension four . for the calculation of dynamical phases
we need the result @xmath296 so _ there is a contribution only from the presence of polarisation gadgets _ [ this was to be expected since we have arranged the ` evolution in direction ' to be of vanishing dynamical phase ] . for general @xmath269 and @xmath270 with initial and final spatial positions @xmath255 , @xmath254 on the beam
we define : @xmath297 & = c \subset \mathcal{r}_4\,.\end{aligned}\ ] ] ( it is implicit that @xmath217 is located in space at @xmath205 and is transverse , so as in section ii it can not be drawn arbitrarily in @xmath274 ) .
then we have : @xmath298 & = \varphi_{\rm tot}[\mathcal{c}]- \varphi_{\rm dyn}[\mathcal{c}]\,,\nonumber\\ \varphi_{\rm tot}[\mathcal{c } ] & = \arg(\boldsymbol{\psi}(s_1)^\dag \boldsymbol{\psi}(s_2))\,,\nonumber\\ \varphi_{\rm dyn}[\mathcal{c } ] & = \text{im}\int_{s_1}^{s_2}ds~\left(\boldsymbol{\psi}(s),\ , \frac{d } { ds } \boldsymbol{\psi}(s)\right)\nonumber\\ & = -\frac{1}{2 } \int_{s_1}^{s_2}ds \ , \boldsymbol{a}(s)\cdot\hat{\boldsymbol{n}}(s)\,.\end{aligned}\ ] ] we may remark finally that while the geometric phase in the present physical situation is always defined by eq.([eq3.6 ] ) , it is only in a quite special situation that we get a simple final expression ( [ eq3.12 ] ) , in a way combining the pancharatnam result ( [ eq2.25 ] ) and the pure ray result ( [ eq2.43 ] ) .
what needs to be stressed however is that the separation of the contributions from the sphere of directions @xmath66 and from the poincar ( polarization ) sphere @xmath65 is essentially unambiguous . in the pancharatnam limit
, @xmath259 shrinks to a point and we recover ( [ eq2.25 ] ) ; while in the pure ray limit with no polarisation gadgets , it is @xmath197 that shrinks to a point and we get back ( [ eq2.43 ] ) for circular polarisations .
the situation analysed in the previous section from the geometric phase point of view is that of a ( narrow well collimated ) light beam of fixed frequency travelling in physical space through a given transparent medium , encountering various polarisation gadgets on its way .
the path of the beam is based on a ray @xmath314 obeying eq.([eq2.27 ] ) for a given refractive index function @xmath203 . from @xmath206
we obtain a particular one - dimensional curve @xmath315 , the two - dimensional sphere of directions .
the ray @xmath206 also gives a preferred choice of a real orthonormal basis @xmath316 in the transverse plane at each @xmath208 , perpendicular to @xmath219 there . by resolving the normalised complex transverse electric field @xmath317 with respect to this basis , we are able to describe it by a normalized complex two - component column vector @xmath290 , leading to the representation of the polarization state by a point @xmath318 . in particular ,
real @xmath317 implies real @xmath290 and vice versa , corresponding to linear polarisations .
the choice of @xmath206 thus provides both @xmath319 , and @xmath316 .
we can regard the latter as a preferred real orthonormal basis in the real tangent plane @xmath320 , for each @xmath321 . as a result ,
the geometric phase contributions from beam direction and beam polarisation are essentially unambiguously separated .
let us now view the problem from another _ more global perspective , not immediately related to a ray or to a picture embedded in physical space_. we take the sphere @xmath66 of plane wave propagation directions as starting point , writing @xmath322 for points on it ( instead of @xmath219 obtained from @xmath206 as upto now ) .
each @xmath322 is the unit vector in the direction of a wave vector @xmath240 associated with a possible ( spatially limited ) propagating plane wave .
we now ask if there is a way to choose a real orthonormal basis @xmath323 in the real tangent plane @xmath324 , well defined and varying smoothly with @xmath322 for all @xmath325 .
since this question is posed prior to the possible choice of a ray @xmath206 , even if such @xmath323 exist , it need have nothing to do with the @xmath316 later supplied by a ray @xmath206 at a point on it where @xmath326 . as we have seen , it is @xmath316 which has specific advantages from a physical point of view , which may be absent with @xmath323 .
it is however a known fact from differential geometry that such choices of @xmath323 for all @xmath327 do not exist .
this is expressed by saying that the sphere @xmath66 is not parallelizable@xcite as a real four - dimensional manifold the tangent bundle @xmath328 is not ( homeomorphic to ) the product @xmath329 .
a useful way to display this circumstance , suited for further developments , is as follows . in real three - dimensional euclidean space
let us choose a right handed cartesian system of axes with origin @xmath330 , and with @xmath331 , @xmath332 , the unit vectors along the coordinate axes .
points on the unit sphere @xmath66 with centre at @xmath330 will be written @xmath333 , @xmath334 , @xmath335 .
it is necessary to define two subsets @xmath336 , @xmath337 of @xmath66 whose union gives @xmath66 but which have a nontrivial ( indeed , substantial ) overlap : @xmath338 we need the action of proper rotations , elements of the rotation group @xmath64 , on @xmath306 .
the right handed rotation about axis @xmath339 by angle @xmath185 corresponds to the @xmath340 matrix @xmath341 for any @xmath342 , there are infinitely many rotations carrying @xmath343 to @xmath322
. however there is no way to choose one such rotation for each @xmath322 , such that it is globally well - defined and varies smoothly with @xmath322 for all @xmath327 . over @xmath336 , which is @xmath66 with just one point ( the south pole ) removed
, a convenient choice does exist : @xmath344 this is well defined at @xmath345 but not at @xmath346 .
acting on @xmath347 at the north pole , we get a real orthonormal basis for the tangent plane @xmath348 when @xmath349 : @xmath350 over @xmath337 a similar choice is : @xmath351 now this is well defined at @xmath346 but not at @xmath345 . acting on @xmath347 at the north pole
we get a different real orthonormal basis for @xmath348 when @xmath352 : @xmath353 in the overlap , which is all of @xmath66 with just the north and south poles removed , we have connecting or ` transition ' formulae : @xmath354 there are now two equally good ways to express the nonparallelizable nature of @xmath355 ( i ) it is not possible to extend the definition of @xmath356 ( respectively @xmath357 ) to cover the south pole @xmath346 ( respectively the north pole @xmath345 ) possessing smooth behaviour for all @xmath358 ; ( ii ) the real orthonormal bases @xmath359 , @xmath360 for @xmath348 over @xmath336 , @xmath337 respectively can not be modified in any way to yield a real orthonormal basis for @xmath348 varying smoothly with @xmath322 all over @xmath66 .
a more formal statement is this : it is impossible to find two smoothly varying angles @xmath361 over @xmath336 , @xmath337 respectively such that the transition group element @xmath362 in eq.([eq4.7 ] ) can be factorised as @xmath363 for , if such choices were possible , then @xmath356 @xmath364 , @xmath365 @xmath366 would carry @xmath343 to @xmath322 and be smoothly defined for all @xmath367 .
now the ` topological obstruction ' described above is in the _ real domain _ ,
i.e. , viewing each tangent plane @xmath348 as a real two - dimensional vector space @xmath368 .
it has however been pointed out recently that if one complexifies each @xmath348 into a complex two - dimensional vector space @xmath369 , then the obstruction vanishes@xcite : it is possible to choose orthonormal bases for these complexified tangent spaces in a globally smooth manner .
there is naturally considerable freedom in such choices ; we describe now a group theory based choice which seems natural and minimal in some sense .
this requires the use of the group @xmath67 ( which is a double cover of @xmath64 , though this property is not used in the @xmath67 version of eq.(4.8 ) established below ) .
what we will show is that _ the factorisation attempted in eq.([eq4.8 ] ) is possible if on the right hand side we allow for elements from @xmath67_. the defining representation of @xmath67 is @xmath370 group composition being matrix multiplication .
the axis - angle description of @xmath67 elements is @xmath371 the two - to - one mapping @xmath180 respecting the composition laws , the homomorphism , is : @xmath372 the rotations @xmath356 , @xmath373 defined in eqs.([eq4.3],[eq4.5 ] ) are images , in the sense of this mapping , of elements @xmath374 , @xmath375 in @xmath67 respectively : @xmath376 now the overlap transition rule ( [ eq4.7 ] ) involves the subgroup of elements @xmath377 , which happen to ` coincide ' with elements @xmath378 in the following sense : @xmath379 it now turns out that _ within _ @xmath67 a factorisation of the form ( [ eq4.8 ] ) _ is possible _ : @xmath380 the structures of @xmath381 , @xmath382 are suggested by those of @xmath383 , @xmath384 in eq.([eq4.12 ] ) : in the latter we make the cyclic changes @xmath385 , and replace @xmath386 by @xmath387 . if we use eq.([eq4.14 ] ) in eq.([eq4.13 ] ) and then in eq.([eq4.7 ] ) we see that @xmath388 is a globally well - defined and smoothly varying matrix in @xmath389 with the property @xmath390 here , the group @xmath389 is the three - dimensional extension of @xmath67 in eq.([eq4.9 ] ) , and consists of @xmath340 unitary unimodular matrices . the subset _
( not subgroup ) _ of @xmath389 carrying @xmath343 to @xmath322 is easy to characterise : @xmath391 ( this decomposition is however not unique on account of the shared elements ( [ eq4.13 ] ) ) . and indeed @xmath392 is of this form , and becomes after simplification : @xmath393 if we act with @xmath392 on @xmath394 , @xmath395 at the north pole we obtain a globally well defined and smooth complex orthonormal basis for @xmath396 all over @xmath306 . to distinguish these vectors from the real ones encountered up to now we write them as @xmath397 , @xmath224
they are simply related to @xmath398 , @xmath399 of eqs.([eq4.4 ] , [ eq4.6 ] ) : @xmath400 while it should be evident _ a priori _ that , given the existence of such global complex @xmath397 all over @xmath66 , there should be considerable freedom in their choice , the specific form of @xmath392 in eq.([eq4.18 ] ) suggests that the choice ( [ eq4.19 ] ) is specially simple . in particular , for the one - parameter family of @xmath401 with fixed @xmath386 the connection between @xmath397 and @xmath402
is through ( portion of ) a _ one - parameter subgroup _ of @xmath67 .
hereafter we always use @xmath403 given above .
any ( complex ) three - vector @xmath404 orthogonal to @xmath322 , thus belonging to @xmath405 , can be expanded as @xmath406 with respect to @xmath403 , and _ as a convention _ , we can regard @xmath404 at @xmath322 and @xmath407 at @xmath408 as ` the same ' if @xmath409 in this sense we see that the union of @xmath405 over all @xmath322 is a cartesian product , which as discussed earlier is not true for the usual tangent bundle in the real domain : @xmath410 the most obvious use of the above result is in the following context .
suppose @xmath404 is a complex vector - valued transverse function of @xmath322 , which for concreteness we regard as an element of a hilbert space @xmath1 as follows : @xmath411 with @xmath412 the solid angle over @xmath66 .
then we can expand @xmath404 in the basis ( [ eq4.19 ] ) and have : @xmath413 this shows that @xmath1 is the tensor product @xmath414 where @xmath415 is the hilbert space of ( scalar ) complex square integrable functions over @xmath66 , and @xmath416 is the two - dimensional complex hilbert space ( appropriate for the ` polarization qubit ' ) .
if @xmath417 is a transverse electric field amplitude of a plane wave with propagation direction @xmath322 , using the expansion ( [ eq4.20 ] ) we may attempt to represent its polarisation state by a point on the poincar sphere @xmath65 in the ` usual ' way : @xmath418 however this is not in general the ` usual representation ' of polarisation states in the sense that , for instance , linear polarisations corresponding to real @xmath417 ( upto overall phases ) need not imply real @xmath234 , so @xmath236 may not lie on the equator of @xmath65 in the 1 - 2 plane . indeed , for @xmath419 , from eqs.([eq4.7],[eq4.19 ] ) we have : @xmath420 using these in eq.([eq4.26 ] ) we find that states of rcp and lcp are represented on @xmath65 by the diametrically opposite points @xmath421 , @xmath422 , not by the usual north and south poles @xmath423 .
correspondingly linear polarization states lie on the great circle on @xmath65 in the plane orthogonal to @xmath424 .
we consider applications of the global results of the previous section to the calculation of geometric phases . in the treatment in sections
ii , iii the starting point was a ray @xmath206 in a given transparent medium , based on which a beam passing through polarisation gadgets was then considered . from this , a curve @xmath277 was obtained , as in eq.([eq2.38 ] ) .
the normalised electric field along the beam was then used to define a curve @xmath425 in the quantum mechanical @xmath157 framework , eqs.([eq2.42 ] , [ eq3.5 ] ) , with @xmath264 . at all stages the validity of maxwell s equations was kept in mind . in @xcite , however , a curve @xmath277 is taken as the starting point for the discussion of geometric phases for beams with varying direction and polarisation state . from the point of view developed by us , this would mean that in principle , for a chosen @xmath277 to be physically realisable , we must imagine a transparent medium with suitable refractive index function @xmath203 , and a ray @xmath206 in this medium , such that a beam traveling along @xmath206 reproduces @xmath259 as we follow @xmath426 along @xmath206 .
all this as well as the validity of maxwell s equations will be implicitly assumed in what follows . as an illustration ,
let us consider the case where @xmath259 is a closed loop , i.e. , @xmath443 .
let us further assume that @xmath251 differs from @xmath256 just by a phase @xmath444 so that @xmath445 is closed .
since in any case @xmath397 s are determined by @xmath322 , these assumptions mean that @xmath446 thus @xmath447 describes a closed loop @xmath308 , and the geometric phase ( [ eq5.6 ] ) becomes : @xmath448 & = \theta - { \rm im}\int_{s_1}^{s_2 } ds\,\boldsymbol{z}(s)^\dag\dot{\boldsymbol{z}}(s ) - \int_{s_1}^{s_2 } ds\ , \boldsymbol{z}(a)^\dag h(s)\boldsymbol{z}(s).\end{aligned}\ ] ] comparing the first two terms with eq.([eq2.25 ] ) we see that they reproduce exactly @xmath310 $ ] , and the net result is @xmath449 = \frac{1}{2}\omega[c_{\rm pol } ] - \int_{s_1}^{s_2 } ds \ , \boldsymbol{z}(s)^\dag h(s)\boldsymbol{z}(s)\,.\end{aligned}\ ] ] further simplification of the second term seems not possible on general grounds , unless one has some information on the way @xmath217 varies with @xmath101 as @xmath439 traces the loop @xmath259 .
the separation of @xmath21 $ ] into the two terms on the right in eq.([eq5.10 ] ) corresponds to the use of the @xmath403 as a basis for @xmath405 at each @xmath322 .
a change from @xmath403 to some other globally smooth basis would alter both terms , while preserving the value of @xmath21 $ ] .
this could possibly limit the direct physical meaning we may ascribe to , say , @xmath450 $ ] on the right hand side .
we hope to have shown that in all geometric phase considerations in the domain of classical optics , the mathematical framework of quantum mechanics is adequate and flexible enough to provide a basis for the entire analysis .
this is so in scalar wave , pure polarisation , as well as beam propagation problems .
we have attempted to provide a clear physical picture of the situations being considered , fully tracing the phenomena ultimately to maxwell s equations in every case .
the relevance of global topological aspects when discussing propagation direction and polarisation state simultaneously was pointed out in @xcite . in our treatment
we have addressed these using elementary group theoretical arguments relevant to the situation leading , in our view , to particularly simple and elegant results .
the approach of this work now needs to be extended to other situations where , in place of a narrow beam endowed with polarisation properties , an extended polarised wave field in space is contemplated .
this and other similar extensions will be taken up elsewhere .
throughout this paper we have tried to show that the standard @xmath157 structure of quantum mechanics , with @xmath6 a @xmath0 principal fibre bundle over base @xmath11 , can be used under all circumstances to handle geometric phases in classical optical situations . in @xcite
a somewhat different structure has been used .
we describe here briefly the connection between the two approaches . for geometric phases associated with light beams
we have used the complex three - dimensional hilbert space @xmath451 with inner product ; the unit sphere @xmath271 of real dimension five ; and the ray space @xmath452 of real dimension four . here
@xmath274 is a @xmath0 principal fibre bundle over base @xmath453 .
we now define characteristic subsets of these spaces as follows : @xmath454 in more detail in the case of @xmath455 we have : @xmath456 for two points @xmath322 , @xmath457 , we find easily : @xmath458 consisting of essentially real @xmath162 corresponding to linear polarisations . in contradistinction ,
the total and base spaces used in @xcite are @xmath459 defined as : @xmath460 these too are of real dimensions five and four respectively , and @xmath461 is a @xmath0 principal fibre bundle over base @xmath462 .
it is easy to see that the second statement in eq .
leads to related cartesian product structures for @xmath463 and @xmath464 : @xmath465 in our treatment , as mentioned above , we use uniformly @xmath466 ( the sphere of normalised vectors in @xmath467 ) rather than @xmath461 , and the associated projective space @xmath468 rather than @xmath462 .
it is important to recognize that @xmath469 , and @xmath470 .
writing @xmath471 for general points in @xmath461 : @xmath472 since @xmath473 , the map @xmath474 is well - defined : @xmath475 however this is a _ many - to - one map_. given @xmath476 , @xmath477 is not unique as : @xmath478 thus , while both @xmath461 and @xmath274 are real five - dimensional manifolds , we do not have a one - to - one map between them , so they are not identical spaces . in a similar way
, it can be checked that @xmath462 and @xmath453 are nonidentical . in @xcite ,
geometric phases are defined for smooth closed curves @xmath479 , with images @xmath480 .
such a curve @xmath481 in parametrised form may be written as @xmath482 with suitable end point conditions . in our approach , since as seen in eq .
the map @xmath483 is well - defined , we can pass from @xmath484 to @xmath485 in an unambiguous manner : @xmath486 and then use eq . to define the geometric phase in the kinematic approach .
this is similar to the way in which in sections 2 and 3 we take the electric field vector along a ray or a beam and use it to obtain a smooth curve in @xmath466 for which a geometric phase can be defined using the kinematic approach .
the expression for the phase given in@xcite is the same as in our treatment , which stays entirely within the standard @xmath157 structure of quantum mechanics .
inner products in hilbert spaces will be generally written as @xmath487 rather than as @xmath488 in dirac notation . for two - component complex column vectors
@xmath489 we sometimes write @xmath490 in place of @xmath491 . for three - component complex column vectors , described by cartesian components in physical space for instance , we write @xmath492 or @xmath493 or @xmath494 as convenient .
simon , s. simon , n. mukunda , f. gori , m. santarsiero , r. borghi , and r. simon , `` a complete characterization of pre - mueller and mueller matrices in polarization optics '' , jour .
a * 27 * , 188199 ( 2010 ) .
a. v. tavrov , y. miyamoto , t. kwabata , and m. takeda , `` generalized algorithm for the unified analysis and simultaneous evaluation of geometrical spinredirection phase and pancharatnam phase in complex interferometric system '' , jour .
a * 17 * , 154 - 161 ( 2000 )
. it may be useful to recall that for a spherical triangle on @xmath495 , the corresponding solid angle ( subtended at the centre of the sphere ) is the ` spherical excess ' , i.e. , the amount by which the sum of the three internal angles exceeds @xmath15 .
this excess occurs because @xmath495 possesses positive curvature . | we present a study of geometric phases in classical wave and polarisation optics using the basic mathematical framework of quantum mechanics .
important physical situations taken from scalar wave optics , pure polarisation optics , and the behaviour of polarisation in the eikonal or ray limit of maxwell s equations in a transparent medium are considered .
the case of a beam of light whose propagation direction and polarisation state are both subject to change is dealt with , attention being paid to the validity of maxwell s equations at all stages .
global topological aspects of the space of all propagation directions are discussed using elementary group theoretical ideas , and the effects on geometric phases are elucidated .
+ _ ocis codes _ : ( 030.1640 ) coherence ; ( 350.5500 ) propagation ; ( 260.5430 ) polarization |
in the neighbourhood of a second order phase transition the behaviour of various quantities is governed by power laws .
for example the correlation length diverges as @xmath8 where @xmath9 is the reduced temperature , @xmath10 and @xmath11 are the amplitudes in the high and the low temperature phase , respectively , and @xmath12 is the critical exponent of the correlation length .
@xmath13 is the critical temperature , where the phase transition occurs .-
transition of @xmath2he we shall denote the critical temperature by @xmath14 . ] the specific heat behaves as @xmath15 where @xmath16 and @xmath17 are the amplitudes in the high and the low temperature phase , respectively , and @xmath18 is an analytic background , which has to be taken into account here , since the critical exponent @xmath19 of the specific heat is negative for the three - dimensional xy universality class .
a universality class is characterised by the dimension of the system , the range of the interaction and the symmetry of the order parameter .
critical phenomena can be understood in the framework of the renormalization group ( rg ) . for reviews on critical phenomena and the renormalization group
see e.g. @xcite .
the xy universality class in three dimensions with short range interactions is of particular interest , since the @xmath0-transition of @xmath2he shares this universality class .
results for critical exponents and amplitude ratios such as @xmath20 obtained at the @xmath0-transition , are more precise than those from experiments on other systems .
it is an interesting question how the critical behaviour is modified by a confining geometry .
if the system is finite in all directions , thermodynamic functions have to be analytic functions .
i.e. a singular behaviour like eqs .
( [ xipower],[cpower ] ) is excluded . as a remnant of such singularities there remains a peak in the neighbourhood of the transition . with increasing linear extension the hight of the peak increases and the temperature of the maximum approaches the critical temperature .
this behaviour is described by the theory of finite size scaling ( fss ) . for reviews
see @xcite . in the present work we study thin films .
thin films are finite in one direction and infinite in the other two directions . in this case
singular behaviour is still possible .
however the associated phase transition belongs to the two - dimensional universality class .
i.e. in the case of @xmath21 symmetry , a kosterlitz - thouless ( kt ) transition @xcite is expected . in our recent work @xcite
we have focused on the study of this transition and the scaling of the transition temperature with the thickness of the film . here
we investigate the behaviour of the specific heat of thin films in the neighbourhood of the @xmath0-transition .
the specific heat has been studied in a number of experiments on thin films of fluid @xmath2he and @xmath22he-@xmath2he mixtures near the @xmath0-transition . for recent reviews
see @xcite .
the physics of thin films is governed by the ratio @xmath23 , where @xmath24 is a correlation length of the bulk system and @xmath7 the thickness of the thin film . for @xmath25 the behaviour of the film
is essentially given by the thermodynamic limit of the three - dimensional system .
in the critical region , when @xmath24 gets close to @xmath7 or even larger , the behaviour deviates from the three - dimensional one and is characterised by universal functions of @xmath23 .
in particular , the behaviour of the specific heat can be described by the universal scaling function @xmath26 a^{-1 } |t|^{\alpha}= \tilde g_{2,r}(l_0/\xi ) = g_{2,r}(t [ l_{0}/\xi_0]^{1/\nu})\ ] ] where @xmath27 is the specific heat of the three - dimensional thermodynamic limit and @xmath28 the specific heat of a film of thickness @xmath7 . as @xmath29 either @xmath30 or @xmath31 can be taken and analogously as @xmath32 either @xmath10 or @xmath11 . in the last step of the equation
we have used eq .
( [ xipower ] ) .
following rg - theory , the analytic background @xmath18 is the same for the bulk and the thin film .
therefore it cancels in the difference that is considered here .
alternatively one might consider @xmath33 a^{-1 } |t|^{\alpha}= \tilde g_{1,r}(l_0/\xi ) = g_{1,r}(t [ l_{0}/\xi_0]^{1/\nu})\ ] ] where @xmath34 is chosen such that @xmath35 in the high temperature phase .
multiplying eqs .
( [ sc1a],[sc1b ] ) by @xmath36^{1/\nu})^{-\alpha}$ ] one arrives at @xmath37 a^{-1 } \xi_0^{\alpha/\nu } l_0^{-\alpha/\nu}= \tilde f_{2,r}(l_0/\xi ) = f_{2,r}(t [ l_{0}/\xi_0]^{1/\nu})\ ] ] and @xmath38 a^{-1 } \xi_0^{\alpha/\nu } l_0^{-\alpha/\nu}= \tilde f_{1,r}(l_0/\xi ) = f_{1,r}(t [ l_{0}/\xi_0]^{1/\nu})\ ] ] respectively .
often in the literature the factors @xmath39 and @xmath40 are omitted and @xmath41 is used as argument of the scaling function .
this poses no problem as long as films of different thicknesses of the same system are considered .
however , comparing films of e.g. @xmath2he at different pressures or @xmath22he-@xmath2he mixtures at different concentrations of @xmath22he this fact has to be taken into account .
the same holds for the comparison of such experimental results with those obtained from lattice models or field theory . in the following
we shall use the notation @xmath42 and @xmath43 .
these universal scaling functions have been determined by a number of experiments on @xmath2he and mixtures of @xmath22he and @xmath2he . in the high temperature phase ,
the data follow nicely the prediction of finite size scaling as can be seen e.g. from figure 14 of @xcite . in this figure , @xmath44 |t|^{\alpha}$ ]
is plotted as a function of @xmath45^{1/\nu}$ ] for films of @xmath2he at vapour pressure of thicknesses 483 @xmath46 up to @xmath47 m . the data for different thicknesses fall nicely on top of each other . in their figure 20
the authors of @xcite have plotted data for the low temperature phase in an analogous way . up to @xmath48 the data for different thicknesses fall nicely on top of each other .
however for larger values of @xmath49 the data start to scatter .
this is most pronounced at @xmath50 , where the function assumes a minimum .
there is a factor of about @xmath51 between the value for the thinnest and the value for the thickest film . from @xmath52
up to @xmath53 there is a factor of about two between the thinnest and the thickest of the films .
note that in figure 20 of @xcite @xmath7 is given in @xmath46 and @xmath54 . in the case of superfluid helium
the order parameter is a complex wavefunction .
this wave function vanishes at the boundaries of the film . in order to mimic this in theoretical models , dirichlet
boundary conditions with vanishing field are employed .
using such boundary conditions , the scaling function @xmath55 has been calculated by using the @xmath56-expansion to o@xmath57 @xcite .
the coefficients of o@xmath58 and o@xmath57 are numerically of similar size
. therefore one should not expect quantitatively accurate results for @xmath55 obtained this way .
both @xmath59 and @xmath55 have been computed by using perturbation theory in three dimensions fixed @xcite in one - loop approximation . in the high temperature phase and at the critical point of the bulk system ,
the experimental results are fairly well reproduced .
however in the low temperature phase , in particular close to the kt transition , no accurate predictions can be obtained .
also monte carlo simulations of the standard xy model on a simple cubic lattice have been performed to determine the specific heat of thin films . for a precise definition of the xy model see below . in @xcite
staggered boundary conditions have been used to obtain a vanishing order parameter at the boundaries .
the authors of @xcite have employed free ( in their notation `` open '' ) boundary conditions as we do in the present work . in both cases
the authors have computed the finite size scaling function @xmath59 . in @xcite
the authors have simulated lattices of a thickness up to @xmath60 lattice units , while in @xcite the thicknesses @xmath61 , @xmath62 and @xmath4 have been studied .
the results of @xcite and @xcite for @xmath59 agree .
there is also a reasonable match with experiments on helium films .
the purpose of the present paper is to compute the finite size scaling function @xmath55 for the first time by using monte carlo simulations of a lattice model .
furthermore we carefully study corrections to scaling , allowing us to quantify the error of our result for the finite size scaling function . for finite systems we expect that the leading corrections are @xmath63 , irrespective of the type of the boundary conditions @xcite .
the numerical value of the correction exponent is @xmath64 for the xy universality class in three dimensions @xcite ; similar results are obtained with field - theoretic methods ; see e.g. @xcite . in order to avoid these corrections , we study an improved model . in improved models the amplitude of corrections @xmath63 vanishes or in practise ,
it is so small that its effect can be ignored .
the precise definition of the model that we have simulated is given below .
on top of the restricted geometry , free boundary conditions introduce new physical effects . for a discussion see e.g. reviews on surface critical phenomena @xcite .
in fact , free boundary conditions lead to additional corrections to scaling .
the leading one is @xmath6 @xcite ; it can be cast in the form @xmath65 .
by @xmath66 to account for surface corrections , was first discussed by capehart and fisher @xcite in the context of the surface susceptibility of ising films . ] in @xcite we have obtained the accurate numerical estimate @xmath67 for the model that we simulate here .
this paper is organized as follows : in the next section we define the lattice model that we have simulated and the observables that we have computed . in section three
we discuss how corrections caused by the free boundary conditions affect the finite size scaling behaviour of the specific heat .
next we discuss the details of our simulations .
based on these simulations we compute the scaling functions @xmath59 and @xmath55 .
we compare our results with those from experiments on thin films of @xmath2he , field theoretic methods and previous monte carlo simulations .
we study the two component @xmath1 model on a simple cubic lattice .
we label the sites of the lattice by @xmath68
. the components of @xmath69 might assume the values @xmath70 .
we simulate lattices of the size @xmath71 and @xmath72 . in 1 and
2-direction we employ periodic boundary conditions and free boundary conditions in 0-direction .
this means that the sites with @xmath73 and @xmath74 have only five nearest neighbours .
this type of boundary conditions could be interpreted as dirichlet boundary conditions with @xmath75 as value of the field at @xmath76 and @xmath77 . note that viewed this way , the thickness of the film is @xmath78 rather than @xmath7 .
this provides a natural explanation of the result @xmath67 obtained in @xcite and might be a good starting point for a field theoretic calculation of @xmath79 .
the hamiltonian of the two - component @xmath1 model , for a vanishing external field , is given by @xmath80 \ ; , \ ] ] where the field variable @xmath81 is a vector with two real components .
@xmath82 denotes a pair of nearest neighbour sites on the lattice .
the partition function is given by @xmath83 \
, \exp(-{\cal h}).\ ] ] note that following the conventions of our previous work , e.g. @xcite , we have absorbed the inverse temperature @xmath84 into the hamiltonian . in the limit @xmath85 the field variables are fixed to unit length ; i.e. the xy model is recovered .
for @xmath86 we get the exactly solvable gaussian model . for @xmath87
the model undergoes a second order phase transition that belongs to the xy universality class .
numerically , using monte carlo simulations and high - temperature series expansions , it has been shown that there is a value @xmath88 , where leading corrections to scaling vanish .
numerical estimates of @xmath89 given in the literature are @xmath90 @xcite , @xmath91 @xcite and most recently @xmath92 @xcite .
the inverse of the critical temperature @xmath93 has been determined accurately for several values of @xmath0 using finite size scaling ( fss ) @xcite .
we shall perform our simulations at @xmath94 , since for this value of @xmath0 comprehensive monte carlo studies of the three - dimensional system in the low and the high temperature phase have been performed @xcite . at @xmath94 one gets @xmath95 @xcite .
since @xmath94 is not exactly equal to @xmath89 , there are still corrections @xmath96 , although with a small amplitude . in fact , following @xcite , it should be by at least a factor 20 smaller than for the standard xy model .
first we should note that in eq .
( [ hamiltonian ] ) @xmath84 does not multiply the second term . therefore , strictly speaking
, @xmath84 is not the inverse temperature .
however , in order to study universal quantities it is not crucial how the transition line in the @xmath84-@xmath0 plane is crossed , as long as this path is smooth and not tangent to the transition line . here ,
following computational convenience , we vary @xmath84 at fixed @xmath0 .
correspondingly we define the energy density as the derivative of the reduced free energy density with respect to @xmath84 .
furthermore we multiply by @xmath97 to get positive numbers : @xmath98 it follows @xmath99 we then define the specific heat as the derivative of the energy density with respect to @xmath84 : @xmath100 it is easy to show that @xmath101 ^ 2 \right \rangle - \left \langle \sum_{<x , y > } \vec{\phi}_x
\cdot \vec{\phi}_y \right \rangle^2 \right ) \;.\ ] ] the second moment correlation length @xmath102 and the transversal correlation length @xmath103 of the bulk system are used to set the scale in the high and the low temperature phase , respectively . here
we shall use the results given in @xcite . for completeness
we recall the definitions of the two correlation lengths .
the second moment correlation length in k - direction is defined by @xmath104 where @xmath105 is the magnetic susceptibility and @xmath106 is the fourier transform of the correlation function at the lowest non - zero momentum in k - direction .
note that in the high temperature phase there is little difference between @xmath102 and the exponential correlation length @xmath107 which is defined by the asymptotic decay of the two - point correlation function .
following @xcite : @xmath108 for the thermodynamic limit of the three - dimensional system .
model we use the convention @xmath109 . ] in @xcite we find for @xmath110 by fitting the data for the second moment correlation length for @xmath111 : @xmath112 \;\;,\ ] ] where @xmath113 .
the helicity modulus @xmath114 gives the reaction of the system under a torsion . to define the helicity modulus we consider a system , where rotated boundary conditions are introduced in one direction (
e.g. the 1-direction ) : for @xmath115 and @xmath116 the term @xmath117 in the hamiltonian is replaced by @xmath118 the helicity modulus is then given by @xmath119 note that we have skipped a factor of @xmath120 compared with the standard definition @xcite
. defined this way , the helicity modulus has the dimension of an inverse length . in the literature @xmath121
is referred to as transversal correlation length .
fitting the data for the helicity modulus at @xmath110 , given in table 2 of @xcite , up to @xmath122 , we get @xmath123 with @xmath113 .
in this section we discuss the finite size scaling behaviour of the specific heat of thin films .
the free energy density of the bulk system is given by @xmath124 where @xmath125 is the analytic background , @xmath126 is an analytic correction and @xmath127 , @xmath128 and @xmath129 are non - analytic corrections . in order to simplify the notation ,
we have omitted subscripts @xmath130 that indicate the phase .
numerical values of the corrections exponents are @xmath131 @xcite and @xmath132 @xcite . note that the correction amplitudes @xmath133 , @xmath134 , @xmath135 are small for the @xmath1 model at @xmath110 , while @xmath136 and @xmath137 should assume generic values . in the following discussion we shall , for simplicity , ignore these corrections to scaling . inserting @xmath138 into eq .
( [ bulkfree ] ) we arrive at @xmath139 \xi(t)^{-d } + b(t)\ ] ] where we have used the hyperscaling relation @xmath140 , where @xmath141 is the dimension of the system .
note that @xmath142 is universal .
the free energy density of a thin film with periodic boundary conditions with the thickness @xmath7 is given by @xcite @xmath143^{1/\nu } ) + b(t)\ ] ] where @xmath141 is the dimension of the system and @xmath125 in eq .
( [ periodicfree ] ) is the same function as in eq .
( [ bulkfree ] ) .
also @xmath144 is an analytic function at @xmath145 .
there might be a singularity at some @xmath146 related with the effectively two - dimensional transition . in order to eliminate the analytic background
one considers the difference @xmath147 ( l_0/\xi(t))^{d } - q_p(t [ l_0/\xi_0]^{1/\nu } ) \right ] \nonumber \\ & = & l_0^{-d } p_p(t [ l_0/\xi_0]^{1/\nu } ) \;\;.\end{aligned}\ ] ] the specific heat is defined as minus the second derivative of @xmath148 with respect to an analytic function of @xmath149 . using our definitions ,
it is minus the second derivative with respect to @xmath150 itself . applied to eq .
( [ bulkfree ] ) we arrive at @xmath151 for the amplitude of the specific heat . taking minus the second derivative of eq .
( [ xxxp ] ) with respect to @xmath149 we arrive at @xmath152^{2/\nu } p_p''(t [ l_0/\xi_0]^{1/\nu } ) = \nonumber \\ - h(0)^{-2 } \tilde a [ l_0/\xi_0]^{\alpha/\nu } [ \tilde a \xi_0^d]^{-1 } p_p''(t [ l_0/\xi_0]^{1/\nu } ) & = & a^{-1 } [ l_0/\xi_0]^{\alpha/\nu } f_{2,p , r}(t [ l_0/\xi_0]^{1/\nu } ) \;\;.\end{aligned}\ ] ] in the case of free boundary conditions , which are studied in the present work , boundary effects have to be taken into account . in @xcite
we have numerically shown that leading corrections can be accounted for by replacing @xmath7 by @xmath153 , where we have obtained the estimate @xmath67 .
hence @xmath154^{1/\nu } ) + b(t ) + c(t ) l_{0,eff}^{-1 } \;\;.\ ] ] the additional term @xmath155 gives a correction of the analytic background caused by the free boundary conditions .
written this way it allows that the correction to the background is given by an effective thickness that is different from @xmath156 .
the prefactor @xmath157 in front of @xmath158 corrects the volume that is used to compute the free energy density . taking the same steps as in the case of periodic boundary conditions we arrive at @xmath159 l_0^{-\alpha/\nu}= a \xi_0^{-\alpha/\nu } f_{2,r}(t [ l_{0,eff}/\xi_0]^{1/\nu } ) -w(t ) l_{0,eff}^{-1-\alpha/\nu}\ ] ] alternatively in the literature one considers @xmath160 l_0^{-\alpha/\nu}= a \xi_0^{-\alpha/\nu } f_{1,r}(t [ l_{0,eff}/\xi_0]^{1/\nu } ) + w(t ) l_{0,eff}^{-1-\alpha/\nu}\ ] ] where @xmath35 or better @xmath161 for @xmath34 in the high temperature phase .
here we study the neighbourhood of the critical point .
therefore we shall approximate @xmath162 . to simplify the notation we shall write @xmath163 instead of @xmath164 in the following . note that for large @xmath165^{1/\nu}$ ] the difference @xmath166 $ ] becomes small compared with @xmath167 or @xmath168 .
on the other hand , corrections due to the boundary are virtually independent on @xmath165^{1/\nu}$ ] .
therefore corrections due to the boundary might lead to large relative errors for large values of @xmath165^{1/\nu}$ ] .
in this subsection we shall consider systems with periodic boundary conditions in all three directions and the linear size @xmath169 . in the high temperature phase , corrections to the thermodynamic limit decay exponentially with increasing lattice size . in practice
it turns out that for @xmath170 these corrections are much smaller than the statistical error that we reach here . since
a goldstone mode is present in the low temperature phase of the system @xcite corrections to the thermodynamic limit decay only with some power of the linear lattice size .
in particular for the energy density and the specific heat , we expect that , to leading order , the correction decays @xmath171 .
therefore rather large lattices are needed to get a good approximation of the thermodynamic limit .
one should note that the estimator ( [ edef ] ) of the energy density is self - averaging , while the one ( [ cdef2 ] ) for the specific heat is not .
since the lattices have to be rather large to avoid sizable finite size effects , it turns out that the specific heat can be most efficiently determined by fitting the energy density in same range of @xmath84 . to this end , we have computed the energy density for a large number of @xmath84-values . as starting point
we have taken the data given in tables 2 and 5 of @xcite .
these were supplemented by a rather large number of new simulations to obtain a dense grid of @xmath84-values . in particular , we have simulated the @xmath172 lattice in the range @xmath173 in steps of @xmath174 , for @xmath175 in steps of @xmath176 and for @xmath177 , @xmath178 , @xmath179 , @xmath180 , @xmath181 , @xmath182 , @xmath183 , @xmath184 , @xmath185 , @xmath186 and @xmath187 .
we simulated the @xmath188 lattice at @xmath189 , @xmath190 , @xmath191 , @xmath192 and @xmath193 , the @xmath194 lattice at @xmath195 , @xmath196 , @xmath197 , @xmath198 , @xmath199 , @xmath200 and @xmath201 and finally the @xmath202 lattice at @xmath203 , @xmath204 , @xmath205 , and @xmath196 . in the case of
@xmath206 we typically performed @xmath207 measurements and @xmath208 for the larger lattice sizes . for each of these measurements
a metropolis sweep , several overrelaxation sweeps and single cluster @xcite updates were performed . for a discussion of the monte carlo algorithm see @xcite . in total , these simulations took a little less than one year of cpu time on one core of a 2218 opteron processor ( 2.60 ghz ) . in @xcite
the results for the thermodynamic limit in the low temperature phase were obtained by fitting the data of several lattice sizes with the ansatz @xmath209 . in the case of the simulations that we have added here ,
we have checked that the @xmath210 corrections are sufficiently small to be ignored . in the neighbourhood of the transition ,
we have fitted the energy density with the ansatz @xmath211 where @xmath212 , @xmath213 , @xmath95 and @xmath214 @xcite are input and @xmath215 , @xmath216 and @xmath217 are the 5 free parameters of the fit . from the finite size scaling behaviour of @xmath218 systems with periodic boundary conditions in all directions at the critical point we find @xcite : @xmath219 for the non - singular part of the energy density and @xmath220 for the non - singular part of the specific heat at @xmath110 .
we did not include a term with the exponent @xmath221 into the ansatz ( [ critical ] ) .
we expect that it is effectively taken into account by the last two terms in ( [ critical ] )
. note that the main purpose of fitting the energy density with ansatz ( [ critical ] ) is to interpolate our data in a large range of @xmath84-values .
after some preliminary studies , we decided to fit the energy density in the range @xmath222 using the ansatz ( [ critical ] ) . in total
we have 98 data points in this interval and we get @xmath223d.o.f .
@xmath224 for our fit .
the results for the fit parameters are @xmath225 , @xmath226 , @xmath227 , @xmath228 and @xmath229 .
first we have computed the universal combination @xmath230 from the result of the fit .
note that @xmath231 .
using the central values for the input parameters we obtain @xmath232 , where we have taken into account the covariance of @xmath233 and @xmath234 .
furthermore we have checked the dependence of our result on the input parameters .
in fact , the dependence on the value of @xmath19 is quite small . taking the preferred value of the experiment on the space shuttle @xcite
@xmath235 we get @xmath236 .
we have also checked the effect of the error of the other input parameters .
it turns out that the uncertainty of @xmath213 has the largest effect on @xmath237 : replacing @xmath238 by @xmath239 we get @xmath240 . as our final result
we quote @xmath241\ ] ] where in @xmath242 we give the statistical error and in @xmath243 $ ] the sum of all errors due to the uncertainty of the input parameters of the fit .
note that the present result is compatible with the final result @xmath244 given in @xcite . in figure [ c3d ]
we have plotted the specific heat obtained from the fit of the energy density using ansatz ( [ critical ] ) .
computing the statistical error , correlations among the fit - parameters are properly taken into account . in this plot , errors can not be resolved . in order to make the errors visible we have plotted in figure [ ploterror ] the statistical error and the difference between the results obtained by using the central values of the input parameters and results where we have replaced one of the central values by the central value plus the error .
as one might expect , the differences diverge in the neighbourhood of the critical point . in the low temperature phase
the largest uncertainty is due to the error of @xmath213 . in particular
, going to the upper boundary @xmath245 of our fit interval , the uncertainty induced by the error of @xmath213 rapidly increases .
therefore we decided to use the values for the specific heat obtained from the fit ( [ critical ] ) only up to @xmath246 . for larger values of @xmath84
we follow an alternative approach as discussed below . in the case of the high temperature phase
we shall use the results obtained from the fit ( [ critical ] ) down to @xmath247 . in order to complete the computation of the specific heat
, we have fitted our data for the energy density with the ansatz @xmath248 where @xmath249 is identified with the specific heat at @xmath250 .
we have included all values of @xmath84 within the interval @xmath251 $ ] into the fit .
we have tested various values of @xmath252 .
our final results are taken from fits with @xmath253 . in the range
@xmath254 we have used @xmath255 .
for @xmath256 we have used @xmath257 and for @xmath258 we have used @xmath259 . in all cases @xmath223d.o.f .
is close to one . to give an impression of the accuracy that is reached , we quote @xmath260 at @xmath246 and @xmath261 at @xmath262 . for comparison with the results obtained from the fit with ansatz ( [ critical ] ) we have also performed fits with ansatz ( [ taylor ] ) in the range @xmath263 using @xmath264 .
the numerical results for the specific heat obtained from the two different approaches are compatible within error bars .
as we have mentioned already in the introduction , often in the literature the factor @xmath40 is ignored when computing the finite size scaling functions @xmath59 and @xmath55 .
therefore , in order to compare the results of different systems , we have to compute @xmath265_{system 1 } } { [ a \xi_0^{-\alpha/\nu}]_{system 2 } } \;.\ ] ] in order to fix the ratio between our definition for the specific heat of the @xmath1 model at @xmath110 and that of experiments on @xmath2he at vapour pressure , we take the ratio of our result for @xmath31 obtained in the previous subsection and that of @xcite for the three - dimensional thermodynamic limit of @xmath2he at vapor pressure . from table
ii of @xcite we read off and that used in @xcite differs by a factor of @xmath19 ] @xmath267 . using an alternative fit ansatz the authors get @xmath268 .
we regard this difference as an estimate of the possible error of @xmath269 . from the same fits ,
the authors obtain @xmath270 and @xmath271 , respectively . in order to match with these experimental numbers ,
we have taken @xmath272 obtained from fitting the energy density with ansatz ( [ critical ] ) , using @xmath235 as input . instead , using @xmath273 we arrive at @xmath274 .
i.e. the value of @xmath275 is quite insensitive on the value of @xmath19 that is assumed .
furthermore , we need the amplitude of the correlation length for @xmath2he at vapor pressure in the high and the low temperature phase . for the transversal correlation length in the low temperature phase one finds @xcite and refs . therein : @xmath276 where @xmath54 .
alternatively we can compute the amplitudes of the correlation lengths from @xmath277 using the results for the universal amplitude ratios @xmath278 @xmath279 and @xmath280 given in @xcite . to this end
, we first have to convert the results for the specific heat of the experiment from @xmath281 into @xmath282 . to this end
we need the density @xmath283 @xcite of @xmath2he at the @xmath0-transition and the boltzmann constant @xmath284 .
we arrive at @xmath285 where we have taken into account the errors of @xmath286 and of the experimental estimate of @xmath275 . in the case of the high temperature phase
we get @xmath287 where we have used @xmath288 taken from table ii of @xcite as input and the error is estimated by using @xmath289 obtained from an alternative ansatz @xcite .
now we are ready to compute the ratio @xmath290 where we can either use @xmath102 or @xmath103 .
we get @xmath291 using @xmath273 or @xmath292 using @xmath235 .
hence we arrive at @xmath293 with a relative uncertainty of about @xmath294 .
let us note that this number is only valid for @xmath2he at vapour pressure and the @xmath1 model at @xmath110 and the particular definitions of the specific heat that have been used .
first we performed simulations at the inverse transition temperature @xmath296 @xcite of the three - dimensional system . here
the correlation length of the thin film is relatively small ; we find @xmath297 .
hence already rather small ratios of @xmath298 are sufficient to approximate well the two - dimensional thermodynamic limit and therefore large values of @xmath7 can be reached .
furthermore , by construction @xmath299 .
therefore this is an ideal location to accurately study the finite size scaling behaviour and in particular the corrections caused by free boundary conditions . at the critical point of the three - dimensional system
( [ definef2 ] ) reduces to @xmath300 where @xmath301 @xcite and @xmath302 .
all numbers for the specific heat discussed in this section are determined by using eq .
( [ cdef2 ] ) .
as a first step , we have simulated the thickness @xmath3 for @xmath303 and @xmath304 .
we conclude from these simulations that , at the level of our statistical error , the two - dimensional thermodynamic limit of the specific heat is reached for @xmath305 .
based on this result we performed simulations for @xmath306 and @xmath304 with @xmath307 throughout . in all cases
, we performed @xmath308 measurements , where for each measurement we performed one metropolis sweep , two overrelaxation sweeps and wall cluster @xcite and single cluster updates .
the number of single cluster updates was chosen such that the average size of a cluster times the number of clusters is a bit less than the number of lattice sites . in total
these simulations took about 5 month of cpu time on one core of a 2218 opteron processor ( 2.60 ghz ) .
the results for the specific heat , as defined by eq .
( [ cdef2 ] ) , are summarized in table [ heatbc ] .
._[heatbc ] results for the specific heat @xmath309 obtained by using eq .
( [ cdef2 ] ) at @xmath310 , which is the estimate of @xcite for @xmath311 , for lattices of the size @xmath307 . _ [ cols= " > , > " , ] the finite size scaling function shows a clear minimum at a finite value of @xmath312 $ ] . by construction ,
the position of this minimum does not depend on @xmath163 .
for @xmath313 we get @xmath312_{min } \approx 1.452 $ ] . for @xmath314
the minimum takes the value @xmath315 and for @xmath316 the value @xmath317 .
for @xmath3 we get @xmath312_{min } \approx 1.468 $ ] and as value @xmath318 for @xmath314 and @xmath319 for @xmath320 .
we consider it as a robust result that @xmath55 assumes a negative value at its minimum .
note that the kt transition takes place at @xmath321 @xcite .
i.e. the minimum of @xmath55 is located at a temperature slightly higher than the transition temperature . in figure [ f2expl ]
we compare our result for @xmath55 with experimental ones . to this end , we plot @xmath322 l_0^{-\alpha/\nu}$ ] as a function of @xmath323 using the experimental data of @xcite for thin films of @xmath2he at vapor pressure of the thicknesses 483 , 1074 , 2113 , 5039 , 6918 and @xmath324 .
these data are taken from the web page @xcite .
the experimental data for the specific heat are given as a function of the reduced temperature @xmath325 . in order to plot them as a function of @xmath323 we use @xmath326 , eq .
( [ xi0 t ] ) . as value of the critical exponents we take @xmath327 @xcite and correspondingly @xmath273 .
furthermore we plot the results of @xcite for a film of 57 @xmath328 m as thickness .
we have rescaled these data as discussed in the previous subsection on the high temperature phase . for comparison
we give our result for @xmath55 obtained from @xmath313 and @xmath5 , taking into account boundary corrections characterized by @xmath329 and @xmath314 ( solid black lines ) . to indicate the possible errors we give in addition the results obtained for @xmath329 and @xmath316 ( dashed black lines ) .
we have multiplied our numbers by @xmath330 , eq .
( [ fuckor ] ) . here
almost all experimental results for the finite size scaling function @xmath55 are somewhat larger than ours .
there is some scattering among the experimental results .
those for the larger thicknesses of the film are closest to our @xmath55 .
there is a nice match for the position of the minimum of @xmath55 obtained from @xmath331 @xmath328 m and our result .
however @xmath55 , computed by using the experimental data , never assumes negative values .
also the dip around the minimum is much less pronounced than it is in our case .
it is beyond the scope of the present work to discuss possible sources of these discrepancies .
this requires detailed discussions with the experimentalists . here
we like to give an explicit formula for @xmath55 in the neighbourhood of @xmath332 . to this end
, we use the results for the amplitudes @xmath277 obtained in subsection [ threed ] , the amplitudes of the correlation length @xmath333 and @xmath334 and the value of @xmath335 obtained in subsection [ fsstc ] .
in addition we have extracted from our data the slope of the specific heat of thin films at the transition temperature of the three - dimensional bulk system . as result
we get in the high temperature phase : @xmath336(l_0/\xi_{2nd})^{-\alpha/\nu } + 0.65(2 ) ( l_0/\xi_{2nd})^{1/\nu } + \ldots\ ] ] where @xmath337 gives the dependence on the value of @xmath19 that is used for the analysis .
this formula gives a good approximation of @xmath55 up to about @xmath338 , where the deviation from the full result as computed above is less than 1@xmath339 . in the low temperature phase
we obtain : @xmath340(l_0/\xi_{t})^{-\alpha/\nu } - 2.43(5 ) ( l_0/\xi_{t})^{1/\nu } + \ ; \ldots\ ] ] this is a good approximation of @xmath55 up to about @xmath341 , where the deviation from the full result as computed above is less than 1@xmath339 . in the high temperature phase , for @xmath343 , the two boundaries are uncorrelated and therefore the dependence of physical quantities on @xmath7 is trivial . in the case of the specific heat we can write ( ignoring boundary corrections ) @xmath344 inserting this equation in the definition of the scaling function
@xmath55 we get @xmath345 since @xmath55 is a function of @xmath346 only and @xmath347 does not depend on @xmath7 , it follows @xmath348 following the convention in the literature @xmath349 in the low temperature phase , the situation is more complicated , due to the presence of the goldstone mode .
power like corrections are present in this case : @xmath350 + 2 c_s(\beta ) \;\;\ ; \mbox{for } \;\ ; \beta > \beta_c \;\ ; , \ ; l_0 \gg \xi_{t,3d } \;\;.\ ] ] we do not have data for a sufficiently large range of temperatures to check carefully this behaviour . assuming the correctness of eq .
( [ f2larget ] ) we read off from our data for @xmath351 : @xmath352 in the high temperature phase and @xmath353 in the low temperature phase , where @xmath354 in both case , we have taken into account the boundary corrections .
it follows that @xmath355 and @xmath356 in units of @xmath357 .
furthermore , we can compute the universal ratio @xmath358 experimental results for films of @xmath2he have been summarized by the authors of @xcite as @xmath359 in units of @xmath357 .
it follows @xmath360 our result for @xmath361 is consistent with that of experiments . in the case of @xmath362
we see a small discrepancy .
the results for @xmath363 are consistent within the error bars .
the surface specific heat has been calculated using the perturbative expansion in three dimensions fixed in the two - loop approximation by mohr and dohm @xcite .
inserting numerical values into eqs .
( 3,4 ) one gets @xmath364 and @xmath365 in units of @xmath357 @xcite .
we notice that the value for @xmath361 is in excellent agreement with our results and also close to the experimental one .
in contrast , the value for @xmath362 is clearly ruled out by us as well as by the experiment . already
the authors of @xcite have pointed out that their result for @xmath362 does not provide an accurate numerical estimate .
krech and dietrich quote as result of the @xmath56-expansion , eq .
( e6 ) of @xcite : @xmath366 + \mbox{o}(\epsilon^2 ) \right\ } \;.\ ] ] inserting @xmath367 one gets @xmath368 in order to compare with the experiments on superfluid @xmath2he at vapour pressure we insert @xmath369 and then convert from units of @xmath370 to @xmath371 .
we arrive at @xmath372 j \aa/ \mbox{mole}^{-1 } k^{-1 } \;\;\ ] ] following eisenriegler @xcite @xmath373 where we have inserted @xmath367 .
the result of eisenriegler has been extended by upton @xcite to @xmath374 . for @xmath367
one gets : @xmath375\ ] ] for a detailed discussion of the field theoretic results see the phd thesis of mohr @xcite .
in order to compute the finite size scaling function @xmath59 we have to determine the specific heat at the temperatures , where @xmath102 of the three - dimensional bulk system assumes the values @xmath376 , @xmath4 and @xmath5 or , taking into account corrections , @xmath377 , @xmath378 and @xmath379 . to this end , we have first numerically inverted eq .
( [ xires ] ) .
the corresponding values of @xmath84 are @xmath380 , @xmath381 , @xmath382 and @xmath383 , @xmath384 , @xmath385 , respectively . at these values of @xmath84
the specific heat of the three - dimensional bulk system assumes the values @xmath386 , @xmath387 , @xmath388 and @xmath389 , @xmath390 , @xmath391 , respectively .
these values are obtained by using the results of fits with the ansatz ( [ critical ] ) .
the error is dominated by the uncertainty of @xmath213 that has been used as input in eq .
( [ critical ] ) . in figure
[ f1ew ] we plot @xmath392 l_{0,eff}^{-\alpha/\nu}$ ] as a function of @xmath393 , where we have used @xmath329 and @xmath314 . in particular for the low temperature phase we see a strong dependence on @xmath7 .
we argue that this is due to analytic corrections . in the case of @xmath55
these corrections affect the specific heat of the three - dimensional bulk system and the thin film in the same way .
therefore there is only a rather small effect on the difference of the two . in the case of @xmath59 the specific heat of the three - dimensional bulk system and the thin film
are taken at different reduced temperatures .
therefore there is a quite huge effect on the difference of the two . motivated by this , in figure [ f1ewx ] , we have subtracted @xmath394 instead of @xmath167 . the coefficient @xmath395 is fitted by eye to get a reasonable collapse of the curves at low temperatures . in figure [ f1exp ]
we compare with experimental results for the finite size scaling @xmath59 . in order to compute @xmath34
we have used @xmath396 , eq .
( [ xi02nd ] ) .
we have plotted @xmath397 l_0^{-\alpha/\nu}$ ] as a function of @xmath398 , where the specific heat is given in units of @xmath281 and @xmath7 in @xmath46 . to this end
we have used the data of @xcite for thin films of @xmath2he at vapor pressure of the thicknesses 483 , 2113 , 6918 and @xmath324 .
these data are taken from the web page @xcite . for a better readability of the figure
we do not give the data for @xmath399 and @xmath400 .
in addition we give the results of @xcite . in this case , we have computed @xmath167 from the results of the fits given in ref .
@xcite . for comparison
we have taken our results for @xmath313 and @xmath5 from figure [ f1ewx ] . in order to match with the experimental results ,
we have multiplied our numbers by @xmath401 . in the high temperature phase ,
the experimental results fall nicely on top of each other .
in contrast , in the low temperature phase , in particular for temperatures below the position of the peak , we see some scattering of the experimental results . in the main our results are compatible with those of the experiments on @xmath2he films . in the high temperature phase
our result is slightly larger than the experimental one .
finally let us compare our results with those obtained from field theory and from previous monte carlo simulations . in figure 1 ( a ) of @xcite a one - loop result for the finite size scaling function @xmath59
the specific heat is given in units of @xmath281 and the length in units of @xmath46 .
the function has similar qualitative features as our result .
however the peak in the low temperature phase is much more shallow than in our case .
the maximal value is about 7 , while we get 11.9 .
the position of the peak ( @xmath402 ) slightly differs from ours ( @xmath403 ) . in @xcite
the function @xmath59 has been calculated from monte carlo simulations of the standard xy model in three dimensions .
the authors have used staggered boundary conditions in order to suppress the order parameter at the boundary .
they have simulated lattices of the thicknesses @xmath404 and @xmath405 .
they define the specific heat as the second derivative of the free energy with respect to the temperature .
they compute it using @xmath406 \;\;.\ ] ] their final result is given in figure 4 of @xcite .
see also , e.g. , figure 2 of @xcite . from the scattering of the data , we conclude that statistical errors are much larger than in our case .
therefore the authors were not able to conduct a detailed analysis of corrections to scaling as we did here .
their result looks qualitatively the same as ours .
the value of the peak seems to be somewhat larger than in our case .
also the position of the peak is slightly different ; from figure 4 of @xcite we read off @xmath407 .
the authors of @xcite have simulated the standard xy model using films of the thicknesses @xmath408 and @xmath4 . throughout , they have used @xmath409 . they have used free ( `` open '' in their notation ) boundary conditions in the short direction . their result for @xmath59
is presented in figure 2 of @xcite . from the scattering of the results it is clear that statistical errors are much larger than in our case .
the results of @xcite and @xcite are consistent .
there is also reasonable agreement with experimental results .
similar to @xcite the value of the maximum seems to be larger than in our case , also the position of the peak is slightly different ( @xmath410 ) .
also in @xcite corrections to scaling are not discussed .
we have studied the finite size scaling behaviour of the specific heat of thin films in the three - dimensional xy universality class . to this end
we have simulated the improved two - component @xmath1 model on the simple cubic lattice . in order
the get a vanishing order parameter at the boundary , which is observed in experiments on films of @xmath2he near the @xmath0-transition , we have employed free boundary conditions .
these can be interpreted as dirichlet boundary conditions with the value @xmath411 of the field at the boundary .
we discuss how leading boundary corrections affect the finite size scaling behaviour of the specific heat of thin films .
we point out that the analytic part of the specific heat might suffer from boundary corrections that are not described by @xmath412 , which characterizes the leading corrections to the singular part .
first we have performed simulations to get the energy density of the three - dimensional system for a large number of temperatures .
these simulations supplement those of @xcite . using these data we computed accurate estimates of the specific heat in the range @xmath413 of inverse temperatures .
next we have analysed in detail the finite size scaling behaviour of the specific heat of thin films at the @xmath0-transition . to this end
we have simulated films up to a thickness of @xmath414 lattice units .
our result is in nice agreement with that obtained for thin films of @xmath2he at the @xmath0-transition @xcite .
furthermore we have simulated films of the thicknesses @xmath415 and @xmath5 for a large range of inverse temperatures @xmath84 in the neighbourhood of the @xmath0-transition .
we have taken great care to obtain reliable estimates for the two - dimensional thermodynamic limit of the thin films .
using our data we have computed the finite size scaling functions @xmath59 and @xmath55 defined in the introduction .
it turns out that corrections to scaling which are caused by the free boundary conditions , have to be taken into account , to get a good collapse of the data obtained for different thicknesses of the film .
these corrections can be described , to leading approximation , by an effective thickness @xmath416 of the film . in @xcite
we have obtained @xmath67 from a finite size scaling study at the critical point of the three - dimensional system . however , a priori , this @xmath156 only applies to the singular part of the specific heat .
our analysis of the data shows that in fact the analytic part of the specific heat requires an additional correction which is @xmath6 .
the comparison of our results for the finite size scaling functions @xmath59 and @xmath55 and those obtained from experiments on films of @xmath2he at the @xmath0-transition in general show nice agreement .
we think that , in order to explain the minor discrepancies , a detailed knowledge of the experimental work is required . therefore we abstain from any speculation on the sources of these discrepancies .
we have also compared with results obtained from field theory and previous monte carlo simulations .
field theoretic calculations are of low order ; o(@xmath56 ) in the case of the @xmath56-expansion and one or two - loop in the case of the perturbative expansion in three dimensions fixed .
previous monte carlo simulations are effected by relatively large statistical errors .
corrections to finite size scaling were not discussed in these works .
this work was supported by the dfg under grants no ja 483/23 - 1 and ha 3150/2 - 1 .
the simulations were performed on the compute cluster grawp at the institute for theoretical physics of the universitt leipzig and on various computers at the physics department of the humboldt - universitt zu berlin .
i like to thank w. janke for discussions and support , v. dohm , m. kimball and j. lipa for sending data and helping me with the literature .
kosterlitz j m and thouless d j , _ ordering , metastability and phase transitions in two - dimensional systems _ 1973 j. phys .
c * 6 * 1181 ; kosterlitz j m , _ the critical properties of the two - dimensional xy model _ , 1974 j. phys .
c * 7 * 1046 campostrini m , hasenbusch m , pelissetto a , and vicari e , _ theoretical estimates of the critical exponents of the superfluid transition in he4 by lattice methods _ , 2006 phys .
b * 74 * 144506 [ arxiv : cond - mat/0605083 ] hasenbusch m , pinn k and vinti s , _
critical exponents of the 3d ising universality class from finite size scaling with standard and improved actions _ , 1999 phys .
b * 59 * 11471 [ arxiv : hep - lat/9806012 ] lipa j a , swanson d r , nissen j a , geng z k , williamson p r , stricker d a , chui t c p , israelsson u e and larson m , _ specific heat of helium confined to a 57-@xmath328 m planar geometry near the lambda point _ , 2000 , phys
. rev .
lett . * 84 * 4894 | we study the finite size scaling behaviour of the specific heat of thin films in the neighbourhood of the @xmath0-transition . to this end
we have simulated the improved two - component @xmath1 model on the simple cubic lattice .
we employ free boundary conditions in the short direction to mimic the vanishing order parameter at the boundaries of a @xmath2he film .
most of our simulations are performed for the thicknesses @xmath3 , @xmath4 and @xmath5 of the film .
it turns out that one has to take into account corrections @xmath6 to obtain a good collapse of the finite size scaling functions obtained from different @xmath7 .
our results are compared with those obtained from experiments on thin films of @xmath2he near the @xmath0-transition , from field theory and from previous monte carlo simulations .
0.5 cm 0.5 cm * the specific heat of thin films near the @xmath0-transition : a monte carlo study of an improved three - dimensional lattice model * martin hasenbusch 0.3 cm _ institut fr physik , humboldt - universitt zu berlin _ _ newtonstr .
15 , 12489 berlin , germany _ 0.1 cm _ e mail : 0.1 cm [email protected]_ 0.4 cm |
in current star formation theory , jets / outflows from a newly forming star are believed to transport significant amounts of energy and momentum away from the region of the central source ( @xcite ; @xcite ; @xcite ; @xcite ) . this can have a big influence on the way in which the stars form , because , for example , jets may drive the injection of turbulence in the parent cloud , thereby regulating the star formation rate . at the same time , they may be able to extract the excess angular momentum from the accretion disk , thus allowing the matter to drift through it and finally accrete onto the star
. jets , therefore , are considered a fundamental ingredient in the star formation process . to fully understand the mechanisms underlying the physics of these commonly observed nebulae , however , it is necessary to know the mass outflow rate , which regulates the dynamics of the flow and , therefore , is the most important input parameter for any model of flow generation and propagation .
for example , in a series of recent papers by our group , differential doppler shifts at the borders in a number of ttauri jets were reported , suggestive of rotation around the symmetry axis ( @xcite ; @xcite ; @xcite ; 2007 ) .
the combination of mass outflow rate and toroidal velocity estimates allows determination of the angular momentum transported by the flow , and thus also a comparison with the angular momentum that the associated disk has to loose in order to accrete at the observed rate .
indeed in one test case , namely the jet from rwaur , we found that the angular momentum present in the jet is at least 60 - 70% of that required to be extracted from the disk ( @xcite ) .
this is of course a result of primary importance , since it could be the first long - awaited validation of the popular idea that jets exist to remove the excess angular momentum in forming system , thus providing a solution to one of the main theoretical problems in star formation . in order to confirm this finding
, however , we have to determine the mass and angular momentum outflow rate close to the star , and in a wider sample of objects .
a good starting point is to use the _ hst_/stis spectra that were recently acquired by our group in a survey to search for rotation signatures in jets from ttauri stars .
the survey included jets from five young stars , namely th28 , dgtau , hh30 , cwtau and rwaur .
the spectra , taken with the slit placed at a few tens of arcseconds from the star , and oriented transversely to the flow direction , sampled the collimation region of the jets and allowed us to confirm the presence of differential doppler shifts at the jet borders and derive estimates of toroidal velocities ( @xcite ; 2007 ) .
these spectra , taken with the slit perpendicular to the jet axis , included several optical forbidden emission lines that can be used to diagnose the physical conditions of the gas .
thus , we have embarked in a further analysis of these data to understand how the plasma properties behave close to the star where the jet is launched , collimated and accelerated , and finally to find the mass and angular momentum outflow rate in our sample .
we point out that estimating the mass outflow rate at the jet base is more likely to exclude spurious effects , such as the addition of mass from possible entrainment of ambient gas and dust or significant disruption of the flow , which are less likely to have a dominant influence in the early stages of jet propagation .
the determination of the mass outflow rate requires a knowledge of the total gas density , which , contrary to the electron density , is not directly measurable from the observed lines .
one , therefore , has to use some method to derive the total density from the available observational data , and indeed several methods have been developed to this aim in the recent past .
for example , from a measure of the intrinsic emission line luminosity one can derive the number of emitting particles of a given species in the observed volume , which can then lead to the gas density under an assumption of abundances .
however , this method relies on an accurate prior knowledge of reddening estimates , excitation temperature , ionisation state of the given species , and filling factor , all of which bring into the calculation substantial uncertainties ( see e.g. @xcite ) .
an alternative method is via a determination of the hydrogen ionisation fraction of the emitting region .
a direct measure of the electron density can be easily obtained from the [ sii]@xmath06716,6731 doublet ( see e.g. @xcite ) . dividing
the electron density by the ionisation fraction then closely approximates the total gas density .
one approach to finding the ionisation fraction is to model the line ratios under the assumption of a definite mechanism for the gas heating .
there is not yet , however , a general consensus on the mechanism causing the jet emission although , without a doubt , the observed forbidden lines are excited collisionally .
the most widely accepted explanation is that the gas is being heated by internal shocks , although other possibilities include ambipolar diffusion , turbulent mixing - layers and compression by jet instabilities ( see e.g. @xcite and references therein ) . on the basis that the gas is shock heated
, @xcite constructed a grid of planar shock models to calculate the line ratios , and compared the results to spectra of a few stellar jets integrated along the beam . in this way
, they were able to find the most likely value of the hydrogen ionisation fraction in the flows , and hence the total density , after accounting for the shock compression .
this method , however , is model dependant , and implies heavy and lenghty calculations .
subsequently , a technique was developed whereby the ratios of the optical forbidden emission lines of s , n and o in the red wavelength range are used to infer the hydrogen ionisation and the electron temperature _ without _ having to assume a specific method of heating , but only assuming that the atomic levels are populated collisionally in the absence of ionising photons ( @xcite ; @xcite ) .
this so - called be technique is based on the recognition that , in the conditions present in stellar jets , the ionisation states of oxygen and nitrogen are tightly correlated to that of hydrogen via charge - exchange .
consequently , the line ratios can be easily modeled and compared with the observed ratios to find the most suitable values of the gas physical quantities .
in addition to the requirement of very little calculations , the technique does not require an accurate knowledge of distance or extinction .
furthermore , the results do not depend on any specific mechanism for jet formation and/or evolution , and thus can be applied to targets in very different conditions .
the _ be technique _ has been applied to jet observations of varying spectral and spatial resolution ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) , which lately included high angular resolution observations of the dgtau , rwaur , lkh@xmath1233 jets using _
hst_/stis multiple - slit configuration , ( @xcite ; @xcite ) as well as sub - arcsecond ground - based data of the dgtau and rwaur jets obtained with adaptive optics ( @xcite ; @xcite ) .
very recently , @xcite used a slightly different version of the _ be technique _ to diagnose the physical condition of the hh30 jet from _
hst_/stis ` slitless ' spectra .
in all cases , the method has brought very valuable information on the excitation of the jet gas allowing , for instance , constraints on models for jet heating , and allowing confirmation that shocks are , as expected , the most likely cause of the thermal behaviour of the plasma . in most cases ,
it has also been possible to estimate the mass and momentum flux in the flow from the derived total densities .
this in turn has facilitated a discourse on the dynamical relationship between optical jets and coaxial molecular flows ( see , e.g. @xcite ) .
furthermore , as mentioned above , in the test case of the rw aur jet , using the results of the _ be technique _ and doppler gradient measurements , we could give an estimate of the angular momentum transported by the flow .
for all these reasons , we decided to use the _ be technique _ to determine the mass and angular momentum flux in our sample of jets observed at high angular resolution .
the stis data included all the optical lines necessary to apply this procedure , which has proven to be very well suited to the diagnostics of large datasets in extended jet surveys , as is the case of this paper . in the following ,
we describe the application of the method to our spectra ( section[observationsanddataanalysis ] ) , and the obtained results ( section[results ] ) . given the form of our input , we present the results for the physical conditions as position - velocity ( pv ) maps , in which the electron density , the ionisation fraction , the electron temperature and the total density are resolved both as a function of velocity and distance from the jet axis . from these values , we derived estimates of the mass and angular momentum outflow rates of the jet close to the launch point .
our findings are discussed in section[discussion ] .
_ hst_/stis spectroscopic observations at optical wavelengths ( proposal i d 9435 ) were made at the base of jets from several ttauri stars ( see table[targets ] ) .
these data formed the basis of two recently published papers which reported measurements of gradients in the doppler shift of the jet radial velocity profile transverse to the flow direction ( @xcite ; 2007 ) . in those papers ,
the possibility that the measurments can be interpreted as indications of jet rotation was investigated .
the adopted observing procedure involved centering the _ hst_/stis slit on the ttauri star , rotating the slit to a position angle perpendicular to the jet axis , and then offsetting the slit to a position along the jet which is a fraction of an arcsecond from the source , as illustrated in figure[orientation ] .
the jets were observed with slit offsets of 0@xmath23 in the cases of th28 , dgtau , rwaur , but 0@xmath22 for cwtau ( due to known lack of line emission at 0@xmath23 @xcite ) , and 0@xmath26 for hh30 .
( the larger offset for hh30 was deemed more appropriate since the star and the jet base are obscured by the system s edge - on disk . )
these angular distances correspond to deprojected distances along each jet of 52 , 68 , 58 , 37 and 84au respectively .
the ccd detector was used with the g750 m grating , centred on 6581 , and a slit of aperture [email protected]@xmath4 .
spectral sampling was 0.554pixel@xmath5 , corresponding to a radial velocity sampling of @xmath625kms@xmath5pixel@xmath5 , and spatial sampling was 0.@xmath705pixel@xmath5 .
resolution is twice the sampling .
long exposures were made of individual jet targets , each of @xmath62000s to 2700s ( for further details see @xcite ; 2007 ) .
the data were calibrated through the standard _ hst _ pipeline , subtraction of any reflected stellar continuum was performed , and signals from defective pixels were removed . in total
, this yielded eight transverse jet spectra , which included h@xmath1 , [ ] @xmath06300,6363 , [ ] @xmath06548,6583 and [ ] @xmath06716,6731 .
data analysis is not reported for the faintest targets , namely the approaching jets from th28 , rwaur and lkh@xmath1321 , due to lack of sufficient signal - to - noise for the [ ] emission , which is fundamental to the diagnostic procedure used here .
an indepth description of the technique can be found in several previously published papers on the method and its application ( @xcite ; @xcite ; @xcite ) .
briefly , the _ be technique _ exploits the fact that in low excitation conditions , such that the plasma is partially neutral and provided that no strong sources of energetic photons are present , the ionisation state of oxygen and nitrogen is dominated by charge exchange with hydrogen .
( justification for this assumption was subsequently confirmed by @xcite . ) in such a plasma , all the sulfur can be considered singly ionised .
young stellar jets satisfy these conditions ( if except is made for the so - called irradiated jets @xcite ) , because they have low electron temperatures and effectively no photoionisation , as the stellar source is usually a low mass star with little production of ionising radiation . under these assumptions ,
the ionisation state of both oxygen and nitrogen can be expressed as a function of the hydrogen ionisation fraction and the electron temperature . combining this with the fact that the [ ] @xmath86731/@xmath86716 is a well - known increasing function of the electron density ( until it reaches the critical density of [ ] , i.e. @xmath62 10@xmath9@xmath10 ) means that the intensity ratio of any two of these observed emission lines is a _ known _ function of electron density , hydrogen ionisation fraction and electron temperature . in practice ,
the _ be technique _ consists in calculating a series of model emission line ratios , and in comparing the result with the observed ratios , to determine the most suitable values of the physical quantities in the emitting gas .
in particular , the sulfur doublet ratio is used to determine the electron density , while the ionisation fraction and the electron temperature are found by combining the ratio of nitrogen to oxygen lines , [ ] / [ ] and the ratio of the oxygen to sulfur lines , [ ] / [ ] , at the determined electron density ratio .
in fact , the first ratio is predominantly sensitive to the ionisation fraction ( [ ] / [ ] increases with the ionisation fraction ) , while the second is comparably sensitive to both ionisation fraction and electron temperature ( [ ] / [ ] increases with increasing electron temperature , and with decreasing electron density , for densities above the critical density ) .
once the electron density and ionisation are determined , it is possible to calculate the total hydrogen density in the emitting gas .
the application of the technique is straightforward , but it had to be automated for the treatment of large datasets .
this has been done recently upon the analysis of _
spectra of the lkh@xmath1233 jet acquired by our group .
a numerical code was written ( hereafter the _ be code _ ) which algebraically inverts the functions generated by the line ratios .
the code was succesfully tested on the lkh@xmath1233 data , and the results were published by @xcite .
we applied this same code to our jet spectra , but we have updated some aspects to account for the fact that our data is in a higher excitation regime , i.e. close to the jet launch point . to compute the fractional population in the atomic levels we use a 5-level code by @xcite
( see @xcite appendix a for details ) .
we use solar abundances from @xcite .
we update the collision strengths , partially following @xcite : @xmath11 ( ) from @xcite ; @xmath11 ( ) from @xcite ; @xmath11 ( ) from @xcite .
all the other atomic parameters ( collisional ionisation rates , recombination rates , dielectronic recombination rates , charge exchange rates ) are those specified in @xcite . before calculating the line ratios , a spurious shift in space generated by the instrument optics was determined for each line and removed ( see @xcite ; 2007 for full details ) .
emission lines were resampled in velocity so that differences in dispersion were accounted for , and so ratios of emission at the same velocity could be obtained .
we considered the brightest components in the [ ] and [ ] emission line doublets , the other being always one third of the brightest component as dictated by atomic physics . in calculating the electron density from the [ ] doublet , values of the [ ] @xmath86731/[]@xmath86716 ratio reaching 2.15 indicate saturation of the doublet ratio such that it is not sensitive to higher densities .
these positions were thus given a lower limit on electron density of 2.5@xmath12@xmath10 determined by the [ ] critical density .
meanwhile , values of the same ratio barely reaching 0.69 cross the lower limit in density sensitivity , and these positions were given an upper limit on electron density of 50@xmath10 . in applying the _ be code _ to our jet spectra , we obtained results for electron density , electron temperature and ionisation fraction .
the electron density was then divided by the ionisation fraction to yield an approximation of the total hydrogen density .
while each jet was observed to emit in all forbidden emission lines , it was found that the code does not converge at certain pixel positions .
the problem was found to arise in cases where jet material is travelling with a broad range of velocities , and the various lines peak at different velocity components , i.e. specifically for the dgtau jet .
the [ ] emission is more intense at lower velocities .
therefore , at higher velocities , the [ ] / [ ] ratio is often too low to allow convergence to a single value of the algebraic inversion performed by the _
be code_. in an attempt to overcome this problem , therefore , we eased the constraints on the algebraic calculations .
we take the electron density derived from the [ ] doublet , as before .
then , instead of using the _ be code _ , we predict [ ] / [ ] and [ ] / [ ] ratios from a mesh of ionisation and temperature values . comparing the predicted and observed ratios
, we find the point where the sum of the differences squared is at a minimum .
this directs us to which combination of temperature and ionisation values best matches the observed line ratios .
this method of using lighter constraints , which we will call the _ be - lite code _ , produced output which was confirmed to be exactly the same as the output from the original _ be code _
, when applied to a jet with a narrow range of velocities , i.e. the th28 jet .
the _ be - lite code _ was then applied to dgtau .
the original _ be code _ was found to represent only 40@xmath13 of the output of the _ be - lite code _ in the analysis of the dgtau jet .
the mesh covered a temperature range from 0.210@xmath9k to 310@xmath9k in steps of 400k , and an ionisation range from 0.005 to 0.5 in steps of 0.005 .
upon final application of the appropriate code to each target , the analysis of th28 and dgtau gave the best results .
this is because these data provide high signal - to - noise in all emission lines .
the analysis of the other targets was more limited , due to insufficient signal - to - noise in [ ] emission , which is less than or equal to that of the other lines in all cases .
all targets examined were found to emit in the forbidden emission line doublets [ ] @xmath06300,6363 , [ ] @xmath06548,6583 and [ ] @xmath06716,6731 . in the adopted observing mode ,
the slit does not collect light from the star .
in fact , the slit ( of width 01 ) is placed perpendicular to the jet usually at 03 from the source , and the instrument spatial line spread function has a half - width at zero - maximum of 015 in the optical regime .
therefore , we can be confident that the observed emission lines originate in jet material and are not contaminated by stellar emission .
position - velocity contour maps of the brightest emission line in each doublet are shown in figures [ th28red_pvs ] , [ dgtau_pvs ] , [ hh30_pvs ] , [ cwtau_pvs ] , [ rwaurred_pvs ] .
position - velocity images of the ratios of these lines are shown as a guide to the distribution of relative fluxes , figures [ th28red_ratios ] , [ dgtau_ratios ] , [ hh30_ratios ] , [ cwtau_ratios ] , [ rwaurred_ratios ] .
for ease of comparison , ratio maps are presented such that they scale positively with the relavant map of physical parameters .
( recall from section[betechnique ] that [ ] @xmath86731/@xmath86716 ] increases with electron density , [ ] / [ ] increases with ionisation fraction , [ ] / [ ] increases with increasing electron temperature and with decreasing electron density . )
position - velocity images of jet physical parameters are shown in figures [ th28red_diagn ] , [ dgtau_diagn ] , [ hh30_diagn ] , [ cwtau_diagn ] and [ rwaurred_diagn ] .
these figures represent output from the _ be code _ applied to th28 , hh30 , cwtau and rwaur , and the _ be - lite code _ applied to dgtau .
the typical values are summarised in table[diagn_table ] .
the overall results indicate that , at the base of jets of the examined ttauri stars , the electron density is high enough to saturate the [ ] doublet ( i.e. @xmath14210@xmath15@xmath10 ) . for all targets , the doublet reaches saturation in many points . where it does not reach saturation , the electron density is always at least above 10@xmath15@xmath10
( it should be borne in mind that ranges for the other parameters are often derived for regions of saturated [ ] doublet ratios . )
the typical temperature in this region of the jet is high , and always in the range 1 - 210@xmath15k ( except for rwaur where it only reaches 510@xmath16k , although this dataset is the poorest ) .
the ionisation level for all targets is in the range 0.03 - 0.3 , where dgtau and rwaur noticeably define the lower end of this range .
the total hydrogen density at the jet base lies in the range 510@xmath15 - 510@xmath17@xmath10 , although this is only a lower limit for saturated regions .
since optically thin forbidden line emission is integrated through the jet , the results represent physical conditions seen in projection .
this causes a smoothing of any gradients in the results . since
such jets are shown to have an onion - like structure , i.e. parameters vary with each layer from the jet axis @xcite , the borders of the flow ( i.e. regions farthest from the jet axis ) are less affected by the projection effect , because the line of sight does not cut through as many `` layers of the onion '' .
the bipolar jet from th28 was identified by @xcite .
this system is almost in the plane of the sky ( 10@xmath18 ) , with a low radial velocity jet ( @xmath630kms@xmath5 ) , and a narrow range of velocities within the jet .
low resolution spectra of this jet have been analysed by @xcite , who find a fall off in electron density with distance along the jet already identified by @xcite , and values for electron density , temperature and ionisation in the region of 10@xmath19@xmath10 , 210@xmath9k and 0.07 - 0.61 respectively .
these values pertain to a distance of 2@xmath7 - 11@xmath7 from the driving source . of all our targets , the results for the th28 receding jet present the clearest indications of the plasma physical conditions .
this is due to good signal - to - noise in all forbidden emission lines , comparable jet width in all emission lines , and a similar narrow range of velocities in all lines .
the position - velocity contour plots , figure[th28red_pvs ] , show that the jet emission is not symmetric .
this is abundantly evident in the [ ] plot , but is also supported by fainter [ ] plot .
the analysis reveals an asymmetry also in electron density .
although there is not much spread in the jet velocity , this asymmetry is clearly evidenced by both lower and higher velocity gas .
the electron density on one side is a lower limit as it has reached the critical density for the [ ] doublet . aside from the obvious asymmetry
, there is no other clear trend in electron density , which lies typically in the range of 0.5 - 1.5 10@xmath15@xmath10 , in the region where the sulfur doublet does not indicate saturation .
there is , however , a trend in the temperature . values reach 310@xmath9k at a thin front located at the peak red - shifted velocity , and then drop to about 210@xmath9k at lower velocities material .
the temperature drops further still moving from on - axis to the jet borders to 110@xmath9k .
thus , along the jet axis , a spine of higher temperatures in the region of 210@xmath9k is decipherable .
also noteable is the decrease in temperature with increasing electron density on one side of the jet .
this inverse relationship is noticeable to a lesser degree over the rest of the jet s cross - section .
the ionisation fraction reaches 0.5 for higher red - shifted velocities .
otherwise , the ionisation appears constant over the whole jet cross - section , with a typical value for this target of 0.3 .
the combination of electron density and ionisation fraction leads to an overall jet hydrogen density of typically 3 10@xmath15@xmath10 .
the asymmetry in electron density is reflected giving a lower limit on the hydrogen density in a few positions , reaching 1.4 10@xmath17@xmath10 .
the behaviour of jet physical parameters which we have identified in these high resolution data shows some possible characteristics typical of those predicted by resolved shock models , e.g. @xcite . according to these authors ,
a resolved shock is represented by high temperature which decreases with distance behind the shock , a corresponding increase in ionisation , and an increasing electron density as the gas is compressed behind the shock .
the magnitude of the values depends on the speed of the shock . in our results
, we see these parameters in velocity space and so if a shock is observed we would expect to see a high temperature peak at lower red - shifted velocities .
but such a trend is not clearly identified , although we so see high ionisation at higher velocities , as would be expected of a shock resolved in velocity space .
we also do nt see any trend of density in velocity space , but an asymmetry in the spatial direction .
@xcite note an agreement with such expectations for temperature and ionisation trends in their analysis of low spectral resolution data for hh30 , but remark on the obvious lack of agreement between the observations and models in that the density is expected to increase in the post - shock region but is not observed to do so .
dgtau was one of the first ttauri stars to be associated with an optical jet @xcite .
this is one of the brightest jets from low mass young stars , and consequently one of the best studied amoung young stellar jets .
previous studies of the gas physics , with a view to determining the heating mechanism , were conducted by @xcite based on sub - arcsecond spectro - imaging data taken from the ground .
gas parameters were reported as a function of distance from the source and in three velocity intervals .
the datapoints closest to the star are at 0@xmath24 .
also , bacciotti et al .
( 2000 ; 2002 ) reported physical parameters derived from _ hst_/stis data of initial portion of this jet with the slit parallel to the flow direction .
the contour plots , figures [ dgtau_pvs ] , reveal that this jet emits in two clearly identifiable velocity bins , the higher velocity component ( hvc ) and the lower velocity component ( lvc ) .
[ ] is clearly more collimated and traces high velocities , while [ ] is spatially extended and traces low velocities .
[ ] traces both lvc and hvc with its peak at high velocities , and demonstrates that the lvc is indeed less collimated than the hvc .
the nominal boundary between hvc and lvc was determined to be -100kms@xmath5 , by two - gaussian deblending of the [ ] @xmath86300 line profile .
note that the [ ] @xmath86300 emission is blue - shifted to the edge of the ccd .
therefore , for this emission line , we do not detect emission at velocities beyond -312kms@xmath5 for instrumental reasons . as a direct result of the large velocity dispersion within this jet , and slightly difference velocity ranges for different lines , the results for dgtau presented here are produced with the _ be - lite code_.
immediately , we note the saturation of the [ ] doublet in many points , both in the hvc and in the lvc . because of this saturation , it is not possible to define a more detailed trend for electron density as a function of velocity .
spatially , however , we can clearly see that the electron density is higher close to the jet axis then at the jet borders , in agreement with @xcite .
a similar spatial trend in other parameters could not be established since [ ] was not observed in the jet borders .
a typical electron temperature of 2.510@xmath9k was found for the hvc and 0.510@xmath9k for the lvc respectively .
the overall results for ionisation show the observed region of the jet , at 0@xmath23 from the source , to be of low ionisation with values reaching only 0.02 in the hvc and 0.05 for the lvc .
recall that we report the electron density in the lvc as a lower limit , implying that the temperature and ionisation derived at these positions should be treated with some caution , because using the critical density instead of the true density gives high values of temperature and low values of ionisation .
we then find a total hydrogen density to be quite scattered , with a tendancy for the hvc to be more dense .
typical values reach @xmath610@xmath20 in the hvc and 510@xmath21@xmath10 in the lvc .
spatially , no trend in the temperature , ionisation and total density was found .
our results agree with trends close to the star reported in the literature .
@xcite report the same low ionisation levels in the same position along the jet as our observations .
they also find the same trend in the ionisation being higher for the lvc than the hvc , in agreement with our results .
however , they find higher electron density for the hvc than the lvc , whereas we find saturation in both cases . finally , it is important to highlight that the high electron density which we find at the base of the jet is not found further along the jet @xcite .
@xcite identified this jet , upon observing emission line nebulosity in the hl tau complex in taurus .
this jet is in the plane of the sky , with a low radial velocity ( @xmath6 - 5kms@xmath5 ) and narrow range of velocity within the jet .
hst images of this system @xcite showed for the first time a disk - jet perpendicular configuration in young stars .
previous diagnostic studies focused on this target include @xcite and @xcite , both of which are based on data of high spatial resolution but low spectral resolution .
there is no division of hvc and lvc in the case of hh30 , since the internal velocity dispersion is small .
datapoints here are fewer , with respect to the previous two targets , due to the low level of [ ] emission .
hh30 shows a scatter in the distribution of electron density , but with all values above @xmath610@xmath15@xmath10 .
its temperature range is scattered within the range 0.4 - 0.810@xmath15k , and its ionisation lies in the range 0.08 - 0.15 .
the total hydrogen density is thus typically 1.510@xmath17@xmath10 .
@xcite presented a similar analysis of hh30 , but in one dimension along the jet and low spectral resolution , showing how each parameter varies with distance from the source .
close to the source , they report an electron temperature of 2 10@xmath9k , and a lower limit on electron density due to saturation of the [ ] doublet ratio .
they report that the jet ionisation fraction rapidly rises from 0.065 at 0@xmath22 to 0.1 at 0@xmath24 , and then slowly increases up to 0.140 within 2@xmath7 from the source .
our results are in good agreement with these values for the examined position .
more recently , @xcite conducted a study of the hh30 jet using hst slitless spectroscopy .
this method relies on the fact that choosing a wide ( 2@xmath7 ) slit ensures the jet is unresolved spectrally , but produces sufficient dispersion to separate images of the jet in the different optical emission line .
the data represents the jet to 4(600au ) along the flow , and in two epochs .
hence , the analysis shows how jet parameters vary with distance and time .
our results are in excellent agreement with those reported at 03 from the source in both epochs ( i.e. log electron density @xmath64.6 , electron temperature @xmath65000k , ionisation fraction @xmath60.1 ) .
the jet from this source was discovered in optical images by @xcite .
although this jet is well studied both through high resolution imaging and spectroscopic observations ( e.g. @xcite ; @xcite ; @xcite ) , no previous analysis of the gas physics has been published for this target .
cwtau very clearly exhibits a separation of jet emission into a hvc and lvc , with the nominal boundary between the two measured as -50kms@xmath5 using two - gaussian deblending . as with hh30 ,
points are few due to the level of [ ] emission . from the available datapoints ,
both velocity components appear saturated in electron density with some unsaturated points in the lvc at typically 0.810@xmath15@xmath10 .
values are found in electron temperature of 0.810@xmath22k and ionisation was typically of 0.2 - 0.3 .
total hydrogen density then became 810@xmath22@xmath10 .
rwaur is a complex triple star system . the highly collimated jet from rwaur a was identified by @xcite .
previous studies of the gas physics have been conducted @xcite , and gas parameters reported as a function of distance from the source and in three velocity intervals . in that study ,
the datapoints closest to the star are at 0@xmath24 .
the analysis of the rwaur jet yields just a few datapoints in the diagnostic maps , due to the poor [ ] emission . the [ ] doublet begins to saturate and leaves a scattered pattern which can be vaguely observed to increase with increasing red - shifted velocities .
the derived temperature reaches 0.510@xmath15k . like the dgtau jet
, rwaur also indicates low ionisation at the observed location along the jet , with a maximum of 0.07 .
our results are in agreement with previous studies @xcite , which give upper limits at the jet base to be @xmath610@xmath15k for the temperature , @xmath610@xmath23 for the ionisation fraction , @xmath610@xmath15@xmath10 for the electron density .
the resulting total hydrogen density reaches 510@xmath17@xmath10 .
sources of error on the estimated parameters depend on the flux measurement error , the effects due to differential extinction , the choice of elemental abundances .
the error on all fluxes is calculated as the rms of the background , after cleaning cosmic rays and defective pixels .
this varies according to target , but remains relatively constant across all emission lines for a given target spectrum . from this
we obtain the three sigma threshold reported as the contour floor in each contour plot of the emission lines .
errors on each of the derived physical parameters due to uncertainty on the line flux measurement were found to be on the order of 30@xmath13 for points of good signal - to - noise .
the amount of dust extinction towards the various positions along these jets is unknown , and may even vary along the beam of a single object ( see @xcite ; @xcite ) .
the extinction towards the targets in this study could not be calculated , since the emission lines observed are too close to each other in wavelength .
therefore , we have not accounted for the effects of extinction by dereddening our emission line fluxes .
differential reddening , i.e. the variation of extinction with wavelength , does affect flux ratios . however , since we are dealing with emission lines spanning a small range in wavelength , the effect is not expected to be significant .
previous studies confirm that this is indeed the case .
@xcite examined the effects of differential reddening on their study of hh30 .
they find that the uncertainty introduced by an extinction of a@xmath24@xmath61 is not significant compared to the uncertainties in the relative abundances , and hence choose to ignore the effects in their analysis .
@xcite report that , assuming a fiducial extinction of a@xmath24@xmath63 , variations in relative flux values reveal errors in the final jet parameters of amounts no larger than the measurement errors , being at most 10@xmath13 for the ionisation fraction and 15@xmath13 for the temperature . for each of our sources
, the level of extinction is a@xmath24@xmath253 ( @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) . furthermore
, the visual extinction of ttauri stars is expected to sharply decrease moving away from the source , and hence should be much lower than 3 at 0@xmath23 - 0@xmath26 .
therefore , we neglect the error introduced by the fact that we are not dereddening the fluxes prior to the analysis .
lastly , the _ be code
_ relies on an assumption of elemental abundances for the determination of temperature and ionisation .
solar abundances are adopted , but may not accurately reflect conditions in star forming clouds .
@xcite show that , using abundances determined for the interstellar medium in orion , the values of ionisation and temperature inferred using the _ be technique _ are within 15@xmath13 of those obtained assuming the most recent determinations of solar abundances .
since our errors arising from flux measurements are in the region of 30@xmath13 , we determine that only an insignificant error is introduced to our results by adopting solar abundances , instead of the actual abundances for the considered star forming regions ( i.e. taurus - aurigae and lupus ) .
as discussed in section[introduction ] , the fundamental parameter to be derived from the diagnostic analysis is the mass outflow rate of the flow .
mass flux was determined from an average of the total hydrogen density , @xmath26 , the poloidal velocity of the emission peak for [ ] , @xmath27 , and the jet radius , @xmath28 , in order to get an average value integrated over the full jet cross - section .
the formula used was @xmath29 , where @xmath30 is the mean molecular weight ( @xmath61.24 ) and @xmath31 is the mass of a proton . the jet poloidal velocity was determined from radial velocity measurements as @xmath32 . for hh30 , this value is taken from proper motion measurements close to the source @xcite , since the inclination angle is so small that poloidal velocities can not be determined from radial velocity measurements without incorporating considerable error .
the jet cross - section was determined by assuming that the fwhm of the [ oi]@xmath86300 emission can be taken as the jet diameter , since it traces both hvc and lvc . in cases where the jet material is clearly divided into hvc and lvc ,
i.e. for dgtau and cwtau , the mass outflow rate was estimated for each component separately .
the estimates are reported in table[massflux ] .
since the measurement uncertainties on the quantities in the calculation are at the 30@xmath13 level , the uncertainty on the calculation of the mass outflow rate leads to a factor of a few , similar to previously published estimations .
we note that any mass outflow rate estimate conducted in this way can represent an upper limit since , not only does this method assumes a homogenous distribution of material within the jet , but also the densest regions dominate the observed emission leading to a bias towards higher densities .
however , close to the star we do expect a homogenity and high density .
conversely , the positions where the [ ] doublet saturates indicate a lower limit on the electron density and hence on the total density measurement . therefore , considering these counter - acting factors , the mass outflow rate is more of an average indication , rather than an upper or lower limit .
angular momentum flux estimates were obtained by assuming the radial velocities and mass outflow rates were indicative of gas physics over the full jet cross - section .
values were determined by combining the mass outflow rate with toroidal velocities derived from the differences in radial velocity across the jet .
the average toroidal velocity of the jet is calculated from the transverse radial velocity gradient ( @xcite ; 2007 ) and the jet inclination angle as @xmath33 .
the radius from the jet axis was taken as half the fwhm , as measured for the mass outflow rate calculation . since we approximate the mass outflow rate and toroidal velocity , and presume them to be unvarying over the full cross - sectional area of the jet , we are satisfied with a simplified formula for the integration of the angular momentum outflow rate , @xmath34 . for hh30 ,
no radial velocity difference were reported above the error bars @xcite , and so no angular momentum outflow rate was calculated .
the results are reported in table[massflux ] . for this jet ,
the mass and angular momentum outflow rates in this lobe are of the order of those previously reported for ttauri jets , although on the low side , with values of 1.2 10@xmath35m@xmath36yr@xmath5 and 2.5 10@xmath37m@xmath36yr@xmath5aukms@xmath5 respectively .
the asymmetry in electron density identified in this target translates to a mass outflow rate which is doubled on one side of the jet axis compared with the other side .
values were calculated for two semi - circular areas ( seen in projection ) either side of the central pixel row which represents the jet axis . for the calculation ,
the poloidal velocity of the emission peak was used , along with the average hydrogen density either side of the axis .
this yielded estimates of 0.4 and 0.9 10@xmath35m@xmath36yr@xmath5 for the two semi - circles .
furthermore , the latter value relies heavily on regions where the electron density is high enough to cause saturation of the [ ] doublet ratio and so it is likely to represent a lower limit ( if the jet is homogenous at this position ) . for dgtau
, the jet material is clearly divided into hvc and lvc , and so the mass outflow rate was estimated for each component separately .
the mass outflow rate is higher in the hvc at 4.1 10@xmath35m@xmath36yr@xmath5 compared to 2.6 10@xmath35m@xmath36yr@xmath5 in the lvc . @xcite
take a nominal boundary of -70kms@xmath5 between the _ medium _ velocity component and lvc ( instead of -100kms@xmath5 ) , and see the trend reversed .
they find values of 5.1 and 8.3 10@xmath35m@xmath36yr@xmath5 as determined from hst / stis spectra with the slit placed parallel to the flow axis .
our total mass outflow rate of 6.7 10@xmath35m@xmath36yr@xmath5 is somewhat lower than 1.3 10@xmath38m@xmath36yr@xmath5 for the parallel slit data , but the latter assumed jet widths for each velocity component which were based on slit positions , and which transpired to be 1.5 times our fwhm measurements .
the mass accretion rate for this system is measured as 2 10@xmath37m@xmath36yr@xmath5 @xcite and so , doubling our values to account for a bipolar flow , we find an mass flow ratio of @[email protected] .
our results show that the jet angular momentum outflow rate in the approaching jet is also considerably higher in the hvc .
we found estimates of 8.4 and 4.5 10@xmath37m@xmath36yr@xmath5aukms@xmath5 for hvc and lvc respectively .
the total of 1.3 10@xmath40m@xmath36yr@xmath5aukms@xmath5 represents the highest angular momentum outflow rate of all our targets .
the value is in good agreement with 3.8 10@xmath40m@xmath36yr@xmath5aukms@xmath5 @xcite , considering the aforementioned difference in the adopted jet radii .
our high spatial resolution results show higher values of angular momentum outflow rate closer to the axis , which is traced by the hvc . in this jet hvc and lvc
are well separated and so projection effects may be lessened , given that the hvc traces the core of the jet while the lvc traces the outer borders .
the mass outflow rate of hh30 is very low at 4.0 10@xmath41m@xmath36yr@xmath5 , and it lies outside the typical range for ttauri stars of 10@xmath37 to 10@xmath35m@xmath36yr@xmath5 .
nevertheless , our result compares well with a previous determination from plateau de bure interferometric data of 1.0 10@xmath41m@xmath36yr@xmath5 @xcite .
our results are also in agreement with the estimate of 1.7 10@xmath41m@xmath36yr@xmath5 , made assuming a jet velocity of 200kms@xmath5 @xcite .
no angular momentum outflow rate could be calculated for this target due to lack of evidence of toroidal velocities @xcite . as in the case of the dgtau jet , hvc and lvc
are clearly separated in cwtau jet .
the mass outflow rate is found to be almost evenly distributed between the two components giving values of 0.4 and 0.3 10@xmath35m@xmath36yr@xmath5 for hvc and lvc respectively but here , unlike dgtau , the datapoints are few and so the result is less reliable .
when compared with the mass accretion rate of 10@xmath37m@xmath36yr@xmath5 @xcite , we obtain a ratio of @xmath60.01 . as in all cases where the sulfer doublet ratio saturates
, the mass outflow rate of the jet may be underestimated .
the angular momentum outflow rate in the approaching jet , also evenly distributed in velocity , was found to be 0.5 and 0.6 10@xmath37m@xmath36yr@xmath5aukms@xmath5 for the hvc and lvc respectively .
the total of 1.1 10@xmath37m@xmath36yr@xmath5aukms@xmath5 represents the lowest angular momentum outflow rate of all our targets . despite the lack of datapoints , a mass outflow rate of 1.7 10@xmath35m@xmath36yr@xmath5
was estimated at 0@xmath23 above the disk - plane .
this value is lower than 3.0 10@xmath35m@xmath36yr@xmath5 reported at 0@xmath22 above the disk - plane from hst / stis spectra with the slit parallel to the flow direction ( @xcite ) . as in the case of dgtau ,
the difference arises from the fact that the latter calculation adopted a jet radius according to the slit position rather than the jet fwhm .
their jet radius is 1.5 times our measurement .
our mass outflow rate result , when doubled to account for a bipolar flow , is 2@xmath13 of the mass accretion flux of 1.6 10@xmath37m@xmath36yr@xmath5 @xcite .
we estimate the angular momentum outflow rate in the receding jet to be 2.9 10@xmath37m@xmath36yr@xmath5aukms@xmath5 , compared to 1.0 10@xmath40m@xmath36yr@xmath5aukms@xmath5 estimated for the aforementioned parallel slit data @xcite .
again , the difference arises from the difference in radii .
the estimates increase to 6.0 10@xmath35m@xmath36yr@xmath5 and 2.6@xmath40m@xmath36yr@xmath5aukms@xmath5 for measurements further above the disk - plane at 0@xmath25 @xcite .
we have analysed the gas physics for several ttauri jets close to the launching point , based on high resolution _
hst_/stis spectra taken with the slit perpendicular to the flow direction .
we have applied the _ be diagnostic technique _ to the line emission spectra , to obtain maps of the electron density , temperature and ionisation fraction . to this aim , we have also adapted the _ be code _ , to allow inclusion of more datapoints in the calculations .
the new approach involves easing the algebraic constraints , which cause difficulties where jets exhibit a broad range of velocities and the various the various lines trace different velocity components , i.e. dgtau . our results represent the first survey of physical conditions at the base of t tauri jets presented in the form of position - velocity diagrams for the physical quantities .
in fact , with our high spatial and spectral resolution dataset , we resolved the jet physics as a function of velocity and distance from the jet axis , in the region just a few tens of au above the disk plane where the flow is launched .
the overall survey results indicate that , at the jet base , the plasma has a high electron density ( @xmath142 10@xmath9@xmath10 ) , high electron temperature ( 210@xmath9k ) , and low ionisation level ( 0.03 - 0.3 ) which varies considerably depending on the target .
indeed , in all cases saturation of the [ ] doublet is reached in at least some of the datapoints , and so in this region close to the launch point of the jet the electron density reported is often a lower limit .
this is different to findings further along the jet which are a factor of ten lower ( @xcite ; @xcite ) . in the case
where previous studies have been carried out , we find good agreement for values reported close to the star . in the case of th28 ( which presents the best dataset ) ,
possible shock signatures are present , thus providing a observational indications that shocks can contribute to heating the jet close to the source .
we determine the mass and angular momentum outflow rates for the jets close to their base .
estimates we determined for the mass and angular momentum outflow rates , both of which are fundamental parameters in constraining models of accretion / ejection structures , particularily if the parameters can be determined close to the jet footpoint .
values for a single jet lobe are in the range 4.0 10@xmath41 to 6.7 10@xmath35m@xmath42yr@xmath5 and 1.1 10@xmath37 to 1.3 10@xmath40m@xmath42yr@xmath5aukms@xmath5 .
again we find good agreement with the literature in cases where values were reported close to the star .
mass flow ratios were found to be @[email protected] - 0.07 where accretion rates were available in the literature ( i.e. for 3 of 5 targets ) .
this is in the range predicted by accretion - ejection models ( @xcite ; @xcite ; @xcite ) .
although we have examined a region of the jet at about 50 - 80au from the source corresponding to the collimation zone , we note that the region where the jet is formed and launched is believed to be on scales of less than 1au , a region which is currently out of the reach of present instrumentation and often obscured by infalling matter .
we await near - infrared interferometry as an opportunity to observing this zone .
* acknowledgements * the present work was supported in part by the european community s marie curie actions - human resource and mobility within the jetset ( jet simulations , experiments and theory ) network , under contract mrtn - ct-2004 - 005592 .
burrows , c. j. , stapelfeldt , k. r. , watson , a. m. , krist , j. e. , ballester , g. e. , clarke , j. t. , crisp , d. , gallagher iii , j. s. , griffiths , r. e. , hester , j. j. , hoessel , j. g. , holtzman , j. a. , mould , j. r. , scowen , p. a. , trauger , j. t. , & westphal , j. a. , 1996 , apj , * 473 * , 437 eiroa , c. , oudmaijer , r. d. , davies , j. k. , de winter , d. , garzn , f. , palacios , j. , alberdi , a. , ferlet , r. , grady , c. a. , cameron , a. , deeg , h. j. , harris , a. w. , horne , k. , mern , b. , miranda , l. f. , montesinos , b. , mora , a. , penny , a. , quirrenbach , a. , rauer , h. , schneider , j. , solano , e. , tsapras , y. , wesselius , p. r. , 2002 , , * 384 * , 1038 | we present an analysis of the gas physics at the base of jets from five t tauri stars based on high angular resolution optical spectra , using the hubble space telescope imaging spectrograph ( hst / stis ) .
the spectra refer to a region within 100 au of the star , i.e. where the collimation of the jet has just taken place .
we form pv images of the line ratios to get a global picture of the flow excitation .
we then apply a specialised diagnostic technique to find the electron density , ionisation fraction , electron temperature and total density .
our results are in the form of pv maps of the obtained quantities , in which the gas behaviour is resolved as a function of both radial velocity and distance from the jet axis .
they highlight a number of interesting physical features of the jet collimation region , including regions of extremely high density , asymmetries with respect to the axis , and possible shock signatures . finally , we estimate the jet mass and angular momentum outflow rates , both of which are fundamental parameters in constraining models of accretion / ejection structures , particularily if the parameters can be determined close to the jet footpoint . comparing mass flow rates for cases where the latter is available in the literature ( i.e. dg
tau , rw aur and cw tau ) reveals a mass ejection - to - accretion ratio of 0.01 - 0.07 .
finally , where possible ( i.e. dg tau and cw tau ) , both mass and angular momentum outflow rates have been resolved into higher and lower velocity jet material . for the clearer case of dg tau
, this revealed that the more collimated higher velocity component plays a dominant role in mass and angular momentum transport . |
let @xmath0 be a simply connected semisimple algebraic group over an algebraically closed field @xmath1 of characteristic 0 ; all @xmath0-modules considered in the following will be supposed to be rational .
an algebraic @xmath0-variety is said to be _ spherical _ if it is normal and if it contains an open @xmath10-orbit , where @xmath11 is a borel subgroup ; a subgroup @xmath12 is said to be _ spherical _ if the homogeneous space @xmath7 is so : any spherical variety can thus be regarded as an open embedding of a spherical homogeneous space , namely its open @xmath0-orbit . important classes of spherical varieties are that of toric varieties and that of symmetric varieties : toric varieties are those spherical varieties whose open orbit is an algebraic torus ; symmetric varieties are those spherical varieties whose generic stabilizer @xmath9 is such that @xmath13 , where @xmath14 is an algebraic involution and where @xmath15 is the set of its fixed points . other important classes of spherical varieties are that of flag varieties and the more general one of wonderful varieties : a _ wonderful variety _
( of _ rank _
@xmath16 ) is a smooth projective @xmath0-variety having an open @xmath0-orbit which satisfies following properties : * the complement of the open @xmath0-orbit is the union of @xmath16 smooth prime divisors having a non - empty transversal intersection ; * any orbit closure equals the intersection of the prime divisors containing it . a spherical subgroup @xmath9 is said to be _ wonderful _ if @xmath7 possesses a wonderful completion ( which is unique , if it exists ) . by @xcite
every self - normalizing symmetric subgroup is wonderful ; more generally every self - normalizing spherical subgroup is wonderful by @xcite .
however many natural examples of embeddings of a spherical homogeneous space do not need to be normal .
for instance , consider a simple @xmath0-module @xmath2 ( in which case we will call @xmath17 a _ simple projective space _ ) possessing a line @xmath18 $ ] fixed by a spherical subgroup .
then consider the orbit @xmath19 \subset { \mathbf p}(v)$ ] , which is spherical , and take its closure @xmath20 } \subset { \mathbf p}(v)$ ] , which generally is not normal ; denote @xmath21 its normalization .
the aim of this work is the study of the orbits of compactifications which arise in such a way , and as well the study of the orbits of their normalizations . in @xcite
it has been proved that any spherical subgroup which occurs as the stabilizer of a point in a simple projective space is wonderful .
if @xmath22 is the wonderful completion of @xmath19 $ ] , then the morphism @xmath19 \to x$ ] extends to @xmath22 and thus we get a morphism @xmath23 : examining such morphism we get a description of the set of orbits of @xmath5 and of @xmath21 .
moreover this leads to a combinatorial criterion to establish whether or not two orbits in @xmath22 map onto the same orbit in @xmath5 , which in particular implies that different orbits in @xmath5 are never @xmath0-equivariantly isomorphic .
our main theorem ( theorem [ teorema caso stretto ] ) is a combinatorial criterion for @xmath24 to be bijective ; this is done under the assumption that @xmath22 is _ strict _ , i.e. that all isotropy groups of @xmath22 are self - normalizing : strict wonderful varieties , introduced in @xcite , are those wondeful varieties which can be embedded in a simple projective space ; they form an important class of wonderful varieties which generalize the symmetric ones of @xcite .
the condition of bijectivity involves the double links of the dynkin diagram of @xmath0 and it is trivially fulfilled whenever @xmath0 is simply laced or @xmath22 is symmetric ; it is easily read off by the _
spherical diagram _ of @xmath22 , which is a useful tool to represent a wonderful variety starting from the dynkin diagram of @xmath0 .
main examples of strict wonderful varieties where bijectivity fails arise from the context of _ wonderful model varieties _ introduced in @xcite ; the general strict case is substantially deduced from the model case . a _ model space _ for a connected ( possibly non - simply connected ) semisimple algebraic group @xmath25 is a quasi - affine homogeneous space whose coordinate ring contains every simple @xmath25-module with multiplicity one ; model spaces were classified in @xcite , where it is introduced the _ wonderful model variety _ @xmath26 , whose orbits naturally parametrize up to isomorphism the model spaces for @xmath25 : every orbit of @xmath26 is of the shape @xmath27 , where @xmath28 is a model space , and conversely this correspondence gives a bijection up to isomorphism . in order to illustrate the above mentioned criterion of bijectivity in the case of a wonderful model variety ,
let s set up some further notation .
if @xmath29 is a dominant weight ( w.r.t . a fixed maximal torus @xmath30 and a fixed borel subgroup @xmath31 ) , define the @xmath32 of @xmath29 as the set @xmath33 where @xmath34 is the set of simple roots w.r.t .
if @xmath36 is a simple factor of type @xmath37 or @xmath38 , number the associated subset of simple roots @xmath39 starting from the extreme of the dynkin diagram of @xmath40 which contains the double link ; define moreover @xmath41 as the subsets whose element index is respectively even and odd .
if they are defined , set @xmath42 @xmath43 or set @xmath44 ( resp .
@xmath45 ) otherwise . finally ,
if @xmath40 is of type @xmath46 , number the simple roots in @xmath47 starting from the extreme of the dynkin diagram which contains a long root .
suppose that @xmath19\subset { \mathbf p}(v)$ ] is the open orbit of a wonderful model variety @xmath26 , where @xmath25 is isogenous with @xmath0 ; denote @xmath29 the highest weight of @xmath2 and set @xmath20}$ ] .
then the normalization @xmath48 is bijective if and only if the following conditions are fulfilled , for every connected component @xmath49 : * if @xmath50 is of type @xmath37 , then either @xmath51 or @xmath52 ; * if @xmath50 is of type @xmath38 , then @xmath53 ; * if @xmath50 is of type @xmath46 and @xmath54 , then @xmath55 as well .
when the generic stabilizer @xmath9 is a self - normalizing symmetric subgroup , compactifications in simple projective spaces were studied in @xcite . under this assumption ,
an explicit description of the orbits of @xmath5 was given and it was proved that these orbits are equal to those of the normalization of @xmath5 . thus our results generalize those contained in @xcite . in the case of a compactification of the adjoint group @xmath56 ( regarded as a @xmath57-symmetric variety ) obtained as the closure of the orbit of the line generated by the identity in a projective space @xmath58 ( where @xmath2 is a simple @xmath0-module ) , a complete classification of the normality and of the smoothness has been given in @xcite .
the paper is organized as follows . in section 1 ,
we set notations and preliminaries ; in section 2 we give some general results about spherical orbit closures in projective spaces ; in section 3 we recall some results from @xcite about stabilizers of points in simple projective spaces . in section 4
, we describe the orbits of the compactifications @xmath5 and @xmath21 ; in section 5 , we prove the criterion of bijectivity of the normalization map in the strict case ; in section 6 , we briefly consider the non - strict case giving some sufficient conditions of bijectivity and non - bijectivity of the normalization map .
+ _ aknowledgements .
_ i want to thank a. maffei , who proposed me the problem , for all his precious help , and p. bravi for many useful discussions on the subject .
as well , i want to thank the referees for their careful reading and useful comments .
spherical diagrams have been made with the package _ lunadiagrams _ , made by p. bravi and available at http://www.mat.uniroma1.it/@xmath59bravi/lunadiagrams .
fix a borel subgroup @xmath11 and a maximal torus @xmath35 ; denote @xmath60 the corresponding root system and @xmath61 the corresponding set of simple roots .
if @xmath62 is any subgroup , denote @xmath63 its character group ; if @xmath2 is a @xmath0-module , denote @xmath64 the set of @xmath9-eigenvectors of @xmath2 and , if @xmath65 , denote @xmath66 the subset of @xmath64 where @xmath9 acts by @xmath67 . if @xmath68 is a dominant weight , we will denote @xmath69 the simple @xmath0-module with highest weight @xmath29 .
if @xmath70 is a lattice ( i.e. a finitely generated free @xmath71-module ) , then @xmath72 denotes the dual lattice and @xmath73 denotes the rational vector space generated by @xmath70 . if @xmath74 is a cone contained in some vector space @xmath2 , then @xmath75 denotes the dual cone in the dual vector space @xmath76 .
let @xmath5 be a spherical @xmath0-variety and fix a base point @xmath77 in such a way that @xmath78 is an open subset ; denote @xmath79 .
let s introduce some data associated to @xmath5 : * @xmath80 .
* @xmath81 , its elements are called the _ colors _ of @xmath5 .
if @xmath82 is an orbit , then @xmath83 denotes the set of colors which contain @xmath84 . *
if @xmath85 is a rational discrete valuation of @xmath86 , then @xmath87 defines an element @xmath88 by @xmath89 where @xmath90 is any @xmath10-semiinvariant function of weight @xmath67 : since @xmath5 possesses an open @xmath10-orbit , such definition does not depend on the function , but only on the weight . if @xmath91 , by abuse of notation we will denote @xmath92 the image of the respective valuation @xmath93 . + a rational discrete valuation @xmath87 is said @xmath0-invariant if @xmath94 , for any @xmath95 and for any @xmath96 ; set @xmath97 the set of @xmath0-invariant rational valuations of @xmath86 .
the map @xmath98 identifies @xmath99 with a convex cone which generates @xmath100 as a vector space ( * ? ? ?
4.1 ) ; together with such embedding , @xmath97 is called the @xmath0-_invariant valuation cone of _ @xmath5 . both @xmath101 and @xmath102 , as well as
the map @xmath103 , depend only on the open orbit @xmath104 and they are the main objects of the _ luna - vust theory _ ( see @xcite ) , which classifies normal equivariant embeddings of a given spherical homogeneous space .
a spherical variety is said to be _ simple _ if it possesses only one closed orbit ; it is said to be _ toroidal _ if no color contains a closed orbit .
if a spherical homogeneous space @xmath7 possesses a complete , simple and toroidal embedding , then this is uniquely determined and it is called the _ canonical embedding _ of @xmath7 ; we will denote it @xmath105 and it dominates any simple complete embedding of @xmath7 . in general , a canonical embedding of @xmath7 exists if and only if the index of @xmath9 in its normalizer is finite , in which case @xmath9 is said to be _
if @xmath5 is a simple spherical variety with complete closed orbit @xmath84 , then the picard group @xmath106 is freely generated by the classes @xmath107 $ ] , with @xmath108 ; moreover , a divisor is generated by global sections ( resp .
ample ) if and only if it is equivalent to a linear combination of such colors with non - negative ( resp .
positive ) coefficients ( * ? ? ?
2.6 and thm .
wonderful varieties are always spherical ( see @xcite ) and a spherical variety is wonderful if and only if it is complete , toroidal , simple and smooth . a spherical subgroup which appears as the generic stabilizer of a wonderful variety is said _ wonderful_. if @xmath9 is a spherical subgroup , then the normalizer @xmath109 acts on the right on @xmath7 by @xmath110 . consider the induced action of @xmath109 on @xmath111 : the kernel of such action is called the _ spherical closure _ of @xmath9 ; if @xmath9 coincides with its spherical closure , then it is called _ spherically closed_. spherically closed subgroups are always wonderful ( * ? ? ?
7.6 ) ; a wonderful variety is said to be _ spherically closed _ if its generic stabilizer is so .
suppose now @xmath22 is a wonderful variety with open @xmath10-orbit @xmath112 and set @xmath79 ; suppose moreover that the center of @xmath0 acts trivially on @xmath22 .
denote @xmath113 the set of colors and @xmath114 the closed orbit ; denote @xmath115 the @xmath116-fixed point ( where @xmath116 denotes the opposite borel subgroup of @xmath10 with respect to @xmath117 ) .
since @xmath0 is semisimple and simply connected , @xmath118 is identified with a sublattice of @xmath119 , while @xmath120 is identified with @xmath63 : if @xmath121 , then @xmath122 will be identified with the character of @xmath116 acting on the fiber of @xmath122 over @xmath123 , while if @xmath124 , then @xmath122 will be identified with the character of @xmath9 acting on the fiber over @xmath125 .
let s introduce some more data attached to a wonderful variety @xmath22 , together with some results which can be found with more details and references in @xcite and in @xcite .
* @xmath126 ; its elements are called the _ spherical roots _ of @xmath22 .
spherical roots form a basis of the lattice @xmath127 ; they also coincide with the minimal set of generators of the free semigroup @xmath128 .
the cardinality of @xmath129 coincides with the _ rank _ of @xmath22 , i.e. with the number of @xmath0-stable prime divisors of @xmath22 , which are naturally in correspondence with spherical roots .
if @xmath130 , denote @xmath131 the corresponding @xmath0-stable prime divisor : it is a wonderful @xmath0-subvariety whose set of spherical roots is @xmath132 .
spherical roots are either positive roots or a sum of two positive roots ; the _ support _ of a spherical root @xmath133 is the set of simple roots where @xmath133 is supported . *
the _ cartan pairing _ of @xmath22 is the natural pairing @xmath134 between colors and spherical roots defined by the equality @xmath135 = \sum_{d \in { \delta } } c(d,{\sigma } ) [ d]$ ] in the picard group @xmath136 . if @xmath137 and @xmath130 , then @xmath138 .
* @xmath139 = @xmath140 is the set of colors _ moved by _
@xmath141 , where @xmath142 and where @xmath143 is the minimal parabolic subgroup associated to @xmath141 . for every @xmath142 it holds @xmath144 . * @xmath145 ; it coincides with the set of simple roots associated to the stabilizer of the open @xmath10-orbit , which is a parabolic subgroup . as well , @xmath146 coincides with the set of simple roots associated with the stabilizer of the @xmath116-fixed point @xmath123 in the closed orbit @xmath84 . *
@xmath147 is the set of simple spherical roots ; correspondingly , set @xmath148 the _ set of colors of type a_. if @xmath149 , set @xmath150 ; then for every spherical root @xmath133 it holds @xmath151 * @xmath152 ; correspondingly , set @xmath153 the _ set of colors of type 2a_. if @xmath154 , set @xmath155 ; then for every spherical root @xmath133 it holds @xmath156 * @xmath157 ; correspondingly , set @xmath158 the _ set of colors of type b_. if @xmath159 , set @xmath155 ; then for every spherical root @xmath133 it holds @xmath160 * @xmath161 is the _ spherical system _ of @xmath22 , where @xmath162 has to be thought of as an abstract set together with the pairing @xmath163 .
this is the combinatorial datum which expresses a wonderful variety : each wonderful variety is uniquely determined by its spherical system ( see @xcite ) .
there is also an abstract combinatorial definition of spherical system ( see @xcite or @xcite ) , introduced in order to obtain the classification of wonderful varieties ( _ luna s conjecture _ ) : the geometrical realizability of spherical systems has been checked in many cases ( see @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ) and recently a general proof which avoids a case - by - case approach has been proposed in @xcite . a very useful tool to represent spherical systems are _ spherical diagrams _ , obtained adding information to the dynkin diagram of @xmath60 ( see @xcite ) . * denote @xmath164 and @xmath165 the restrictions to the closed and to the open orbit ; then we get a commutative diagram @xmath166 \ar@{->}_{{\omega}}[d ] & { \mathcal x}(h ) \ar@{->}[d ] \\ { \mathcal x}(b )
\ar@{->}[r ] & { \mathcal x}(b\cap h ) } \ ] ] which identifies @xmath167 with the fiber product ( * ? ? ?
2.2.1 ) @xmath168 on the combinatorial level , the map @xmath169 is described on colors as follows ( ( * ? ? ?
* thm . 2.2 ) ) : @xmath170 where @xmath171 is the fundamental dominant weight associated to @xmath142 . *
if @xmath172 is a subset of spherical roots , then the _ localization at @xmath173 _ of @xmath22 is the @xmath0-stable subvariety @xmath174 it is a wonderful variety whose spherical system is @xmath175 , where @xmath176 .
denote @xmath177 the set of colors of @xmath178 ;
if @xmath179 and @xmath180 set @xmath181 and @xmath182 . let @xmath183 be the pullback map ; then @xmath169 factors through @xmath184 and by its combinatorial description it follows that : * * if @xmath185 then @xmath186 ( resp .
@xmath187 ) is supported on @xmath188 ( resp . on @xmath189 ) with multiplicity one , while it is not supported on @xmath189 ( resp . on @xmath188 ) ; * * if @xmath190 then @xmath186 and @xmath187 are supported on @xmath191 with multiplicity one ; * * if @xmath192 , then @xmath193 ; * * if @xmath194 , then @xmath195 ; * * if @xmath159 , then @xmath196 is supported on @xmath191 with multiplicity one and on at most one more color .
* @xmath22 is said to be _ strict _ if the stabilizer of any point is self - normalizing ; equivalently , we will say also that @xmath9 is strict .
a wonderful variety is strict if and only if it can be embedded in a simple projective space ( see @xcite ) . *
consider the following sets of spherical roots @xmath197 set @xmath198 the _ set of loose spherical roots_. loose spherical roots of the first kind are easily described , they are those of the following types ( where @xmath199 and simple roots are labelled as in bourbaki ) : * * spherical roots @xmath200 with support of type @xmath201 and with @xmath202 ; * * spherical roots @xmath203 with support of type @xmath204 .
+ for every @xmath205 , it is defined a @xmath0-equivariant automorphism @xmath206 of order 2 which fixes pointwise the @xmath0-stable divisor @xmath131 associated to @xmath133 . if @xmath207 , then @xmath208 acts trivially on @xmath113 , while if @xmath209 , then @xmath208 exchanges @xmath210 and @xmath211 and acts trivially on @xmath212 . such automorphisms commute and generate @xmath213 ( see @xcite ) .
+ by the natural identification @xmath214 , it follows that * * @xmath9 is self - normalizing if and only if @xmath215 ; * * @xmath9 is spherically closed if and only if @xmath216 ; * * @xmath9 is strict if and only if @xmath217 and @xmath215 .
+ in particular , if @xmath218 , then @xmath9 is self - normalizing if and only if it is spherically closed if and only if it is strict .
* @xmath219 denotes the set of the _ spherical roots of _ @xmath0 : its elements are the spherical roots of all possible rank one @xmath0-wonderful varieties . following the classification of the latter ( see @xcite ) ,
such set is classified for any @xmath0 ( see @xcite and @xcite ) .
let @xmath2 be a @xmath0-module and let @xmath220 be a projective variety with an open @xmath10-orbit ; denote @xmath221 the open orbit .
since @xmath5 contains finitely many @xmath10-orbits ( * ? ? ?
2.6 ) , every @xmath0-orbit in @xmath5 is spherical .
let @xmath222 be the normalization ; then @xmath21 is a complete spherical variety with the same open orbit of @xmath5 whose orbits are naturally in bijection with those of @xmath5 : [ @xcite prop
. 1 ] [ timashev ] the normalization morphism @xmath222 induces a bijection between @xmath0-orbits . if @xmath223 is an orbit , set @xmath224 the corresponding orbit .
denote @xmath225 and @xmath226 the open @xmath10-orbits ; fix base points @xmath227 and @xmath228 so that we have isomorphisms @xmath229 with @xmath230 .
let s recall a result which will be useful in the following : [ @xcite prop 5.1 and cor .
5.2 ] [ brion ] let @xmath9 be a spherical subgroup of @xmath0 .
* the algebraic group @xmath231 is diagonalizable ; moreover , if @xmath232 is the identity component of @xmath9 , then @xmath233 . *
if @xmath10 is any borel subgroup such that @xmath234 is open in @xmath0 , then @xmath109 equals the right stabilizer of @xmath234 . combing back to our situation
, then we obtain : [ stabilizzatori sobri ] @xmath235 is a normal subgroup with finite index ; in particular @xmath236 is a finite diagonalizable group .
since @xmath237 is a finite morphism , it preserves dimensions of orbits : so we have @xmath238
. then @xmath235 implies @xmath239 and we obtain @xmath240 this shows @xmath241 and the claim follows .
[ indice ] let @xmath230 be two spherical subgroups of @xmath0 with @xmath242 normal in @xmath243 ; fix a borel subgroup @xmath10 such that @xmath244 is open in @xmath0 and consider the projection @xmath245 .
then @xmath246 and @xmath247 identifies @xmath248 with a sublattice of @xmath249 such that @xmath250 first claim follows by the equality @xmath251 , which stems immediately from theorem [ brion ] ii ) .
if @xmath252 , denote @xmath253 the kernel of the restriction @xmath254 , which is surjective by the following argument : if @xmath255 is the unipotent radical , then @xmath256 and @xmath257 and the restriction @xmath258 is surjective since @xmath259 is a diagonalizable subgroup of the torus @xmath260 . by definition
, we have isomorphisms @xmath261 and @xmath262 ; thus the restriction gives a surjective homomorphism @xmath263 whose kernel is @xmath248 . on the other hand @xmath264 and @xmath265 ,
hence the equality @xmath251 implies @xmath266 therefore we get @xmath267 going back to our situation , since @xmath236 is a finite diagonalizable group , it is isomorphic to its character group and we get the following corollary .
[ indice2 ] let @xmath268 be an orbit and let @xmath269 , with @xmath230 ; then @xmath270 fix a closed orbit @xmath82 . since parabolic subgroups are self - normalizing , @xmath84 and @xmath271 are isomorphic ; from now on , we will denote both of them with the same letter @xmath84 .
let @xmath272 \in y^{b^-}$ ] be the unique fixed point by @xmath116 ( where @xmath273 is a lowest weight vector ) and let @xmath274 be a highest weight vector such that @xmath275 . if @xmath276 is the stabilizer of @xmath277 $ ] , then @xmath276 and @xmath278 are opposite parabolic subgroups ; denote @xmath279 the associated levi subgroup .
consider the affine @xmath276-stable open subset @xmath280 defined by the non - vanishing of @xmath281 and recall that there exists an affine closed @xmath282-stable subvariety @xmath283 containing @xmath284 and possessing an open @xmath285-orbit such that the multiplication morphism @xmath286 ( where @xmath287 denotes the unipotent radical of @xmath276 ) is a @xmath276-equivariant isomorphism ( * ? ? ?
1.2 ) . since @xmath288 = k[s_x]^l = k$ ]
, we get that @xmath289 possesses a unique closed @xmath282-orbit , namely the fixed point @xmath284 . if @xmath290 , denote @xmath291 and @xmath292 its closure respectively in @xmath5 and in @xmath21 .
[ chiusura colori ] let @xmath290 , let @xmath223 be an orbit and set @xmath293
. then @xmath294 if and only if @xmath295 . by lemma [ indice ]
, @xmath296 is the open @xmath10-orbit of @xmath297 .
suppose that @xmath298 and fix @xmath227 ; if @xmath299 , then we obtain @xmath300 , which implies @xmath301 .
suppose conversely that @xmath295 : then @xmath302 .
[ colori xl ] let @xmath290 ; then @xmath303 if and only if @xmath304 .
notice that @xmath305 is closed and that it contains @xmath84 if and only if @xmath291 contains @xmath84 .
if @xmath304 , then @xmath306 , thus @xmath307 is non - empty , @xmath282-stable and closed in @xmath289 : hence @xmath308 , which implies that @xmath309 .
if conversely @xmath310 , then @xmath311 and we get @xmath312 . combining previous lemmas it follows that the set of colors @xmath313 whose closure in @xmath21 contains the closed orbit @xmath84 is @xmath314 if @xmath5 possesses a unique closed orbit @xmath84 , then previous lemmas allow us to compute the _ colored cone _ of the normal embedding @xmath315 : this is the couple @xmath316 , where @xmath317 is the cone generated by the elements @xmath318 , where @xmath319 is any @xmath10-stable ( possibly @xmath0-stable ) prime divisor which contains @xmath84 ( see @xcite ) .
in fact , since @xmath21 is simple and complete , then @xmath320 contains the @xmath0-invariant valuation cone @xmath321 ( ( * ? ? ? * thm .
5.2 ) ) : therefore @xmath320 is the cone generated by @xmath321 together with @xmath322 .
let @xmath223 be an orbit .
set @xmath323 , @xmath324 and @xmath325 ; then @xmath326 and @xmath327 .
considering the rings of functions , we get a commutative diagram @xmath328 \ar@{^{(}->}[r ] \ar@{->>}[d ] & k[{\widetilde}{x}_0 ] \ar@{->>}[d ] \\
k[z_0 ] \ar@{^{(}->}[r ] & k[z'_0 ] } \ ] ] [ @xcite thm .
2.3 ] [ suriettivita semiinvarianti ] every @xmath10-semiinvariant function @xmath329^{(b)}$ ] ( resp .
@xmath330^{(b)}$ ] ) can be extended to a @xmath10-semiinvariant function @xmath331^{(b)}$ ] ( resp .
@xmath332^{(b)}$ ] ) .
if @xmath70 is a finitely generated free @xmath71-module and @xmath333 is a submonoid , then the _ saturation _ of @xmath334 in @xmath70 is the submonoid @xmath335 , where @xmath336 is the cone generated by @xmath334 .
if @xmath337 is a submonoid containing @xmath334 , then we will say that @xmath334 is _ saturated in _
@xmath338 if @xmath339 .
[ reticoli saturi ] fix an orbit @xmath223 and set @xmath293 .
* @xmath340 is the saturation of @xmath341 in @xmath127 .
* if @xmath342 , then @xmath343 is the normalization of @xmath344 . set @xmath345^{(b)}/k^*$ ] and
@xmath346^{(b)}/k^*$ ] ; in terms of colored cones , such monoids are described as follows ( ( * ? ? ?
* thm . 3.5 ) ) : @xmath347 since every @xmath10-semiinvariant function is uniquely determined by its weight up to a scalar factor , by previous theorem restriction gives an isomorphism of multiplicative monoids @xmath348^{(b ) } { \ , : \,}f{\bigr|}_{z'}\not \equiv 0 \right\ } \stackrel{\sim}{{\longrightarrow } } k[z'_0]^{(b)}:\ ] ] since the first one is saturated in @xmath349^{(b)}$ ] , we may then identify @xmath350 with a saturated submonoid of @xmath351 .
hence we may as well identify @xmath340 with a sublattice of @xmath127 : in fact every @xmath10-semiinvariant rational function on @xmath21 ( resp . on @xmath297 ) can be written as a quotient of two @xmath10-semiinvariant regular functions on @xmath352 ( resp .
on @xmath353 ) . since @xmath21 is normal , @xmath354 is saturated ; therefore , being saturated in @xmath351 , we see that @xmath355 as well is saturated . since the colored cone of a spherical variety is strictly convex ( ( *
4.1 ) ) , @xmath356 has maximal dimension : thus the equality @xmath357 implies that @xmath358 is a saturated sublattice .
hence by @xmath359 = [ k : k ' ] < \infty$ ] we get i ) .
finally , ii ) stems from the fact that a @xmath0-stable subvariety of a spherical variety is normal ( * ? ? ?
3.5 ) together with the fact that the restriction @xmath360 is finite and birational .
let @xmath2 be a simple @xmath0-module and suppose @xmath361 is a spherical orbit .
a necessary and sufficient condition so that a spherical subgroup @xmath9 arises in such a way has been given in @xcite : [ sferically closed ] a spherical subgroup occurs as the stabilizer of a point in a simple projective space if and only if it is spherically closed .
let @xmath22 be the wonderful completion of @xmath7 : then the embedding @xmath362 extends to a morphism @xmath363 and it determines a @xmath10-stable effective divisor @xmath364 which is generated by global sections .
conversely , it is possible to start with an arbitrary spherically closed wonderful variety @xmath22 together with an effective @xmath10-stable divisor @xmath365 generated by global sections and to consider then the associated morphism @xmath366 , where @xmath367 is the dual of the simple module generated by the canonical section @xmath368 . in @xcite , there have been given necessary and sufficient conditions on such a divisor @xmath365 so that @xmath369 restricts to an embedding of the open orbit @xmath7 ; the aim of this subsection is to recall some facts which lead to such conditions .
let @xmath22 be a wonderful variety with base point @xmath370 and set @xmath79
. set @xmath161 its spherical system and @xmath371 its set of colors . recall that a subset @xmath372 is said to be
_ distinguished _
if there exists @xmath373 such that @xmath374 , for every @xmath130 .
if @xmath375 is a sober subgroup and if @xmath376 is the projection , then the subset of colors @xmath377 is distinguished ; conversely , if @xmath378 is a distinguished subset , then there exists a unique wonderful subgroup @xmath379 with @xmath380 connected such that @xmath381 .
this is the content of the following theorem : [ @xcite thm .
5.4 , @xcite prop . 3.3.2 , @xcite thm .
3.1.1 ] [ corrispondenza distinta ] @xmath382 + there is an inclusion - preserving bijection as follows @xmath383 moreover , if @xmath375 is a wonderful subgroup with @xmath380 connected and if @xmath378 is the corresponding distinguished subset , then * the projection @xmath384 identifies @xmath385 with @xmath386 ; * the spherical system of the wonderful completion of @xmath387 is @xmath388 defined as follows : * * @xmath389 is the set of indecomposable elements of the free semigroup @xmath390 * * @xmath391 ; * * @xmath392 , and the pairing is obtained by restriction . in the notations of previous theorem ,
the wonderful completion of @xmath387 is denoted @xmath393 and it is called the _ quotient wonderful variety _ of @xmath22 by @xmath394 , while @xmath395 is called the _ quotient spherical system _ of @xmath396 by @xmath394 .
[ distinti saturi ] by ( * ? ? ?
* lemma 5.3 and thm .
5.4 ) together with ( * ? ? ?
* lemma 3.3.1 ) , there is an inclusion - preserving bijection between distinguished subsets @xmath378 and sober subgroups @xmath375 such that @xmath380 is connected . in (
5.6.2 ) it was proved that , in case @xmath0 is of type @xmath397 , then such a subgroup @xmath398 is necessarily wonderful ; although this was claimed in general in @xcite , a general proof ( which stems from the classification of spherical systems ) appeared only recently in ( * ? ? ?
3.3.1 ) .
suppose that @xmath375 is a sober subgroup such that @xmath380 is connected and denote @xmath378 the distinguished subset of colors which map dominantly to @xmath387 .
denote @xmath399 the canonical embedding and extend the projection @xmath400 to a morphism @xmath401 ; consider @xmath402 as a sublattice of @xmath127 .
set @xmath403 : it is a linear subspace which contains @xmath404 and which intersects the valuation cone @xmath321 in a face ; as a cone , @xmath405 is generated by this face together with @xmath404 ( * ? ? ?
* lemma 3.3.1 ) .
while the valuation cone @xmath406 is the image of @xmath321 under the quotient map @xmath407 by @xmath405 , the lattice @xmath402 is identified with a sublattice of @xmath127 as follows @xmath408 ( * ? ? ?
* lemma 5.3 and thm .
5.4 ) ; as a consequence , @xmath402 is saturated in @xmath127 . if @xmath177 is the set of colors of @xmath409 and if @xmath410 , then @xmath411 ; since the fibers of @xmath412 are complete and connected , it follows that @xmath413 = k[\phi^{-1}(m'_0)]$ ] .
considering the @xmath10-semiinvariant functions , we get then the identification of semigroups @xmath414^{(b ) } } } \ !
\raisebox{-.60ex}{\ensuremath{k^ * } } } } \simeq - { \ensuremath{\raisebox{.40ex}{\ensuremath{{\mathbf n}{\sigma } } } \
! \big / \ !
\raisebox{-.60ex}{\ensuremath{{\delta}^*}}}}.\ ] ] if @xmath409 is smooth , then such semigroup is free ; conversely , given any distinguished subset @xmath372 , the semigroup @xmath415 is free ( * ? ? ?
3.1.1 ) , which means that @xmath409 is necessarily smooth .
recall the restrictions to the closed and to the open orbit @xmath164 and @xmath416 and let @xmath417 with @xmath418 for every @xmath419 .
if @xmath420 is the canonical section , then the submodule @xmath421 generated by @xmath422 is identified with the simple module @xmath423 ( which contains a unique @xmath9-invariant line where @xmath9 acts by @xmath424 ) and we get a morphism @xmath425 define the _ support _ of @xmath365 as @xmath426 as a consequence of theorem [ corrispondenza distinta ] , we get the following corollary .
[ distinti stabilizzatori ] let @xmath22 be a wonderful variety and let @xmath427 be a divisor generated by global sections and consider the associated morphism @xmath428 .
then the correspondence of theorem [ corrispondenza distinta ] gives an inclusion - preserving bijection as follows @xmath429 let @xmath375 be a wonderful subgroup with @xmath380 connected and set @xmath372 the corresponding distinguished subset .
if @xmath430 is the wonderful embedding , then the projection @xmath384 extends to a morphism @xmath431 and pullback identifies @xmath432 with the submodule @xmath433 \subset { \mathbf z}{\delta}= \operatorname{pic}(m)$ ] .
thus the map @xmath434 factors through a map @xmath435 if and only if @xmath436 . [ @xcite ] [ coppia fedele ] let @xmath22 be a spherically closed wonderful variety and denote @xmath113 its set of colors . a divisor generated by global sections
@xmath437 is called _ faithful _ if it satisfies the following conditions : ( fd1 ) : : every non - empty distinguished subset of @xmath113 intersects @xmath438 ; ( fd2 ) : : if @xmath439 is a loose spherical root , then @xmath440 .
[ @xcite prop .
2.4.3 ] [ fedelta ] let @xmath22 be a spherically closed wonderful variety and let @xmath427 .
then the morphism @xmath441 restricts to an embedding @xmath442 if and only if @xmath365 is faithful .
fix @xmath443 a representative of the line @xmath444 and suppose that @xmath445 $ ] ; then previous corollary implies * ( fd1)*. suppose by absurd that * ( fd2 ) * fails and let @xmath446 be a loose spherical root such that @xmath447 . if @xmath448 is the corresponding automorphism , then @xmath449 exchanges @xmath450 and @xmath451 and fixes every other color @xmath452 : therefore @xmath449 fixes @xmath365 .
the action of @xmath213 on @xmath453 is defined extending by linearity the right action of @xmath231 on @xmath113 , i.e. by the action of @xmath109 on @xmath63 .
therefore , if @xmath454 is a representative of @xmath449 , then @xmath455 , i.e. @xmath456 moves the line @xmath457 $ ] in a line where @xmath9 acts by the same character : since @xmath9 is spherical , such a line is unique , thus @xmath458 $ ] which is absurd .
suppose conversely that @xmath365 is a faithful divisor . by * ( fd1 ) * it follows that @xmath459 $ ] , therefore by theorem [ brion ] we get @xmath460 \subset n_g(h)$ ] .
suppose by absurd that there exists @xmath461 { \smallsetminus}h$ ] .
then @xmath455 and the equivariant automorphism corresponding to the coset @xmath462 fixes @xmath365 : therefore by * ( fd2 ) * we get that every color @xmath463 is fixed by @xmath456 . on the other hand , since @xmath9 is spherically closed , every element in @xmath464 acts non - trivially on @xmath113 .
take @xmath142 such that @xmath456 moves @xmath465 , then we get @xmath446 and @xmath466 : therefore @xmath467 , which contradicts * ( fd2)*. [ sistema sferico stabilizzatore ] in the same hypotheses of corollary [ distinti stabilizzatori ] , suppose moreover that every distinguished subset of @xmath113 intersects @xmath438 and set @xmath468 then the spherical system of @xmath469 is @xmath175 , where @xmath470 for every loose spherical root @xmath205 , the quotient @xmath471 is easily proved to be a wonderful variety , whose spherical system is @xmath472 , where @xmath473 and where @xmath474 . if @xmath454 is a representative of the coset corresponding to @xmath208 , then @xmath475 , where @xmath476 is the subgroup generated by @xmath9 together with @xmath456 . by the first part of the proof of previous theorem it follows that @xmath476 fixes @xmath444 , thus we get a commutative diagram @xmath477 \ar@{->}[drr ] & & \\ { \ensuremath{\raisebox{.40ex}{\ensuremath{m } } \ !
\raisebox{-.60ex}{\ensuremath{{\gamma}({\sigma } ) } } } } \ar@{->}[rr ] & & { \mathbf p}(v^*_{{\omega}({\delta } ) } ) } \ ] ] consider now the quotient variety @xmath478 , where @xmath479 is the subgroup generated by the elements @xmath208 , with @xmath480 : then , by previous discussion and by proposition [ fedelta ] , it follows that @xmath478 is a spherically closed wonderful variety endowed with a faithful divisor whose associated characters are the same of @xmath365 . in the hypotheses of previous corollary ,
the assumption that every distinguished subset of colors intersects @xmath438 ( which is equivalent to assume that @xmath9 and @xmath481 have the same dimension ) involves no loss of generality : we can always reduce to that case considering , instead of @xmath22 , the quotient wonderful variety @xmath482 , where @xmath483 is the maximal distinguished subset which does not intersect @xmath438 .
let @xmath22 be a spherically closed wonderful variety with base point @xmath370 and set @xmath79 ; set @xmath161 its spherical system and @xmath371 its set of colors . if @xmath364 is a faithful divisor , set @xmath486 and consider the morphism @xmath366 .
set @xmath487 and set @xmath488 the normalization ; then we get a commutative diagram @xmath489_{\phi_{\delta } } \ar@{->>}[rr]^{\tilde{\phi}_{\delta } } & & { \widetilde{x}_{\delta}}\ar@{->>}[d]^{p } \\ & & { x_{\delta } } & \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\ ! \subset { \mathbf p}(v ) } \ ] ] if @xmath490 is the closed orbit , then by lemma [ chiusura colori ] and lemma [ colori xl ] we get that @xmath491 is canonically identified with @xmath492
. if @xmath493 is an orbit , in the following @xmath494 will denote the pullback of @xmath364 . notice that if @xmath142 then @xmath495 where @xmath496 denotes the set of simple roots associated to the closed orbit @xmath84 .
if @xmath22 is strict , then the variety @xmath484 depends only on the support of @xmath365 ( * ? ? ? * lemma 1 ) : if @xmath497 are faithful divisors on @xmath22 , then @xmath498 as will be shown by lemma [ inclusioni s(d ) ] , this is not true if @xmath22 is not strict .
[ sistema sferico orbita xlt ] let @xmath499 be an orbit and let @xmath269 ; let @xmath500 be any orbit which maps on @xmath501 and choose the stabilizers so that @xmath502 . then @xmath242 is the maximal subgroup such that @xmath503 in particular , @xmath342 if and only if @xmath504 is connected .
set @xmath505 the maximal subgroup of @xmath243 containing @xmath506 such that @xmath507 is connected .
since @xmath508 and since @xmath509 , by theorem [ brion ] we get that @xmath510 is a normal subgroup ; thus by lemma [ indice ] it follows that @xmath511 if and only if @xmath512 .
consider the inclusions @xmath513 : since @xmath514 is saturated in @xmath127 , proposition [ reticoli saturi ] shows that @xmath340 is the saturation of @xmath341 in @xmath514 . on the other hand , by remark [ distinti saturi ]
it follows that @xmath515 is saturated in @xmath514 :
since @xmath516 = [ k : k^ * ] < \infty$ ] , we get the equality @xmath512 . combining previous proposition together with (
3.3.1 ) and corollary [ sistema sferico stabilizzatore ] we get the following corollary .
let @xmath499 be an orbit and let @xmath517 with @xmath235
. then @xmath242 is a wondwerful subgroup .
if moreover @xmath22 is strict , then @xmath243 is the spherical closure of @xmath242 .
if @xmath518 is an orbit and if @xmath519 is the corresponding orbit , denote @xmath520 the sets of spherical roots of the respective wonderful completions . by corollary [ sistema sferico stabilizzatore ] there
exists a bijection between @xmath521 and @xmath522 , which associates to @xmath523 the unique @xmath524 which is proportional to @xmath525 : more precisely , if @xmath526 , then @xmath527 .
if @xmath528 , then we say that : * @xmath133 is of _ type _
@xmath529 if @xmath530 has support of type @xmath531 ; * @xmath133 is of _ type _
@xmath532 if @xmath533 has support of type @xmath531 ; * @xmath133 is of _ type _
@xmath534 if @xmath203 has support of type @xmath535 ; * @xmath133 is of _ type _
@xmath536 if @xmath537 has support of type @xmath535 .
consider a spherical root @xmath528 such that @xmath538 : following the explicit description of @xmath219 , such a root either is a simple root , or it is of type @xmath529 or it is of type @xmath534 . if @xmath539 is any orbit and if @xmath519 is the corresponding orbit , define @xmath540 to be the subset of spherical roots which have to be doubled to get the spherical roots of @xmath501 .
[ orbite non isomorfe e radici sferiche cong ] an orbit @xmath518 is not isomorphic to its corresponding orbit @xmath541
if and only if @xmath501 possesses a spherical root @xmath525 of the shape @xmath542 , where @xmath543 are pairwise distinct elements ( and where @xmath544 ) . by corollary [ sistema sferico stabilizzatore ] , @xmath501 and @xmath297 are not isomorphic if and only if @xmath545 ; suppose @xmath546 . by proposition [ sistema sferico orbita xlt ]
the wonderful completion of @xmath297 is the quotient of a wonderful subvariety @xmath547 ; therefore we can write @xmath548 , where @xmath549 are spherical roots of @xmath409 . since @xmath550 , by
the discussion preceeding the lemma @xmath551 is either a simple root , or it is of type @xmath529 or it is of type @xmath534 .
if @xmath551 is a simple root or if it is of type @xmath529 then it follows immediately that every @xmath552 is equal to one .
suppose instead that @xmath551 is of type @xmath534 ; in order to show the thesis it is enough to consider the case wherein @xmath409 is a wonderful variety whose spherical roots are all supported on a subset @xmath553 of type @xmath204 .
an easy computation shows that , if @xmath554 and if @xmath394 is any distinguished subset of colors of @xmath409 , then the quotient @xmath555 never possesses @xmath556 as a spherical root .
therefore , if @xmath557 , it must be either @xmath558 or @xmath559 and the claim follows .
as exemplified in the following sections ( example [ ex2 ] and example [ ex3 ] ) , proposition [ sistema sferico orbita xlt ] together with corollary [ sistema sferico stabilizzatore ] allow to compute explicitly the set of orbits of @xmath484 and that of @xmath485 in terms of their spherical systems .
this is further simplified by the following proposition , which shows that , given an orbit @xmath518 , there exists a minimal orbit @xmath560 mapping on @xmath501 . if @xmath561 , define @xmath562 its support over @xmath129 ; define @xmath563 [ orbita minimale ] let @xmath539 be an orbit and let @xmath560 the orbit whose closure has @xmath564 as set of spherical roots .
then @xmath565 maps on @xmath501 and and every other orbit which maps on @xmath501 contains @xmath565 in its closure .
let @xmath493 be an orbit mapping on @xmath501 and let @xmath566 be the associated set of spherical roots .
since @xmath567 , we get @xmath568 ; this shows @xmath569 , i.e. @xmath570 . in order to prove that @xmath571 it is enough to notice that @xmath572 .
[ orbite massimali ] unlike the symmetric case ( see @xcite ) , in the general spherical case there does not need to exist a maximal orbit in @xmath22 mapping on a fixed orbit @xmath518 : for instance this is shown by example [ ex2 ] and by example [ ex3 ] . since @xmath564 depends only on @xmath521 ( or equivalently on @xmath522 ) , we get the following corollaries .
two orbits @xmath573 map to the same orbit in @xmath484 if and only if @xmath574 where @xmath575 is the pullback of @xmath365 to @xmath576 and where @xmath577 is the maximal distinguished subset of colors of @xmath578 not intersecting the support of @xmath575 .
[ orbite non isomorfe ] two orbits in @xmath484 ( resp . in @xmath485 )
have different sets of spherical roots ; in particular two orbits in @xmath484 ( resp . in @xmath485 ) are never isomorphic .
if @xmath217 , then by its combinatorial description it follows that the restriction map to the closed orbit @xmath164 is injective : this means that the generic stabilizer @xmath9 never fixes two different lines in the same simple module . however ,
if @xmath579 , it could happen that a simple module @xmath17 contains two different orbits both isomorphic to the open orbit @xmath7 : previous corollary shows then that there does not exist any spherical orbit in @xmath17 containing both of them in its closure .
for instance , this occurs in the following example .
[ ex orbite sferiche doppie ] consider @xmath580 , which is a wonderful variety for @xmath581 , and fix the base point @xmath582,[0,1])$ ] so that the generic stabilizer is the maximal torus @xmath117 of diagonal matrices .
consider the simple module @xmath583_5 $ ] formed by the homogeneous polynomials of degree 5 : then @xmath584 $ ] and @xmath585 $ ] are distinct orbits in @xmath17 both isomorphic to the open orbit @xmath586 .
keeping the notations of previous section , suppose that @xmath22 is strict . following lemma is a stronger version of lemma [ orbite non isomorfe e radici sferiche cong ] .
[ orbite non isomorfe e radici sferiche ] let @xmath22 be a strict wonderful variety and let @xmath365 be a faithful divisor on it ; let @xmath518 be an orbit .
then @xmath587 if and only if there exists a spherical root @xmath523 of type @xmath532 and a spherical root @xmath588 of type @xmath589 . by lemma [ orbite non isomorfe e radici sferiche cong ]
, we may assume that @xmath297 possesses a spherical root @xmath525 of type @xmath529 or of type @xmath534 .
since @xmath218 , it is uniquely determined a spherical root @xmath588 which is of type @xmath590 ( with @xmath591 ) in the first case and of type @xmath534 in the second case .
since @xmath22 is strict , the latter can not happen ; thus we are in the first case .
suppose that @xmath592 and @xmath593 ; let @xmath594 be the short root in the support of @xmath133 . since @xmath22 is strict , @xmath595 moves a color @xmath596 , while @xmath592 implies @xmath597 for every @xmath598 : therefore @xmath599 is distinguished and by the faithfulness of @xmath365 we get @xmath600 , which implies @xmath601 .
but this is a contradiction since @xmath593 implies @xmath602 .
if @xmath130 is a spherical root of type @xmath589 , write @xmath603 , where @xmath604 are respectively the long simple root and the short simple root in the support of @xmath133 . since @xmath22 is strict , both @xmath605 and @xmath606 move exactly one color ; set @xmath607 and @xmath608 .
* if @xmath609 , then no orbit @xmath539 possesses a spherical root @xmath523 of type @xmath532 with @xmath588 .
* if @xmath610 , then there exists an orbit @xmath539 such that @xmath611 ; in particular @xmath612 and the normalization @xmath613 is not bijective .
ii ) . consider the rank one orbit @xmath493 whose unique spherical root is @xmath133 .
if @xmath616 and @xmath617 , then @xmath618 set @xmath619 and @xmath293 ; set @xmath620 the maximal distinguished subset not intersecting the support of @xmath621 . since @xmath622 and since @xmath623 is the unique color @xmath624 such that @xmath625 , we get @xmath626 which shows @xmath627 . on the other hand @xmath628 ,
thus @xmath297 is not spherically closed and @xmath629 . *
if @xmath22 is a strict wonderful symmetric variety and if @xmath365 is a faithful divisor on it , then the normalization @xmath488 is bijective .
* suppose that the dynkin diagram of @xmath0 is simply laced .
if @xmath22 is any strict wonderful variety for @xmath0 and if @xmath364 is any faithful divisor , then the normalization @xmath488 is bijective . *
if @xmath609 for every @xmath130 of type @xmath589 , then the normalization morphism @xmath613 is bijective . by the classification of symmetric varieties , we deduce that a strict wonderful symmetric variety never possesses a spherical root of type @xmath589 . then all of the claims above follow straightforward by previous lemmas .
consider the wonderful model variety @xmath22 of @xmath630 , whose spherical system is expressed by the spherical diagram @xmath631 then the divisor @xmath632 is faithful . consider the codimension one orbit @xmath493 having spherical root @xmath633 ; following proposition [ sistema sferico orbita xlt ] and corollary [ sistema sferico stabilizzatore ] , we get the following sequence of spherical diagrams @xmath634 where the first one represents the orbit @xmath635 , the second one represents the orbit @xmath636 and the third one represents the orbit @xmath637 .
suppose that @xmath493 is an orbit with set of spherical roots @xmath172 and set of colors @xmath177 . if @xmath185 , set @xmath642 ; then by the description of the pullback map @xmath643 it follows that @xmath644 where @xmath645 .
i ) . let @xmath518 be an orbit possessing @xmath646 as a spherical root ; let @xmath647 and let @xmath493 be an orbit which maps on @xmath501 .
then by proposition [ reticoli saturi ] we get that @xmath648 , while by corollary [ sistema sferico stabilizzatore ] together with theorem [ corrispondenza distinta ] we get @xmath649 ; by the remark at the beginning of the proof this implies the thesis .
ii ) . consider the rank one orbit @xmath650 whose unique spherical root is @xmath141 , set @xmath619 and @xmath293 . then @xmath648 is a loose spherical root and by the remark at the beginning of the proof we get @xmath651 , where @xmath652 is the pullback of a hyperplane section and where @xmath653 is identified with a subset of @xmath654 .
then by corollary [ sistema sferico stabilizzatore ] we get that @xmath640 .
suppose that @xmath639 .
as shown by example [ ex3 ] , if @xmath655 , then it may not exist any orbit @xmath518 possessing @xmath646 as a spherical root ; conversely , if there exists such an orbit , it may be as well @xmath655 . *
if there exists @xmath639 such that @xmath447 is non - zero , then the normalization @xmath656 is not bijective . * if the dynkin diagram of @xmath0 is simply laced and if @xmath440 for every @xmath639 , then the normalization @xmath656 is bijective .
reasoning as in lemma [ d1d2 ] and in corollary [ esempi ] , other sufficient conditions of bijectivity can be obtained imposing further conditions on the support of @xmath365 on the multiple links of the dynkin diagram of @xmath0 and on the simple spherical roots of @xmath22 .
m. brion , , notes de la session de la s. m. f. `` oprations hamiltoniennes et oprations de groupes algbriques '' , grenoble ( 1997 ) , http://www-fourier.ujf-grenoble.fr/@xmath59mbrion/spheriques.pdf , 59 pp . | let @xmath0 be a simply connected semisimple algebraic group over an algebraically closed field @xmath1 of characteristic 0 and let @xmath2 be a rational simple @xmath0-module .
if @xmath3 is a spherical orbit and if @xmath4 is its closure , then we describe the orbits of @xmath5 and those of its normalization @xmath6 . if moreover the wonderful completion of @xmath7 is strict , then we give necessary and sufficient combinatorial conditions so that the normalization morphism @xmath8 is a homeomorphism .
such conditions are trivially fulfilled if @xmath0 is simply laced or if @xmath9 is a symmetric subgroup . |
the discovery of superconductivity at @xmath8 = 26 k in lafeaso@xmath2f@xmath3 has stimulated intensive efforts on searching for new materials with higher superconducting transition temperature , @xmath8 , and revealing their unconventional nature .
@xcite until now , a few series of iron - based superconductors ( fescs ) have been discovered , with a maximum @xmath8 raised to 56 k when la is substituted by other rare - earth elements , e.g. , ce , sm , nd , pr and gd , in the @xmath17fe@xmath18o family ( @xmath17 = rare earth , @xmath18= pnictogen ) .
@xcite the fescs show strong similarities to the copper oxides : @xcite ( i ) both possess a layered crystal structure and a relative high @xmath8 ; ( ii ) superconductivity seems to be closely tied up with magnetism . on the other hand ,
significant differences have been observed between them .
for example , the parent compounds of fescs are bad metals , in contrast to the mott - insulators in the cuprates .
@xcite the superconducting order parameter of fescs is supposed to be s@xmath19-type , @xcite while the cuprates are d - wave superconductors .
@xcite moreover , the fescs show nearly isotropic upper critical fields at low temperature , @xcite but the cuprates are rather anisotropic .
@xcite in these compounds , elemental substitutions have been shown to be an efficient approach to study superconductivity and its interplay with magnetism . in @xmath17feaso ,
superconductivity can be induced by doping either electrons or holes into the @xmath17o layers , as well as doping electrons into the feas layers . upon substituting fe with co , the magnetic / structural transition
is quickly suppressed and superconductivity exists over a narrow doping range .
@xcite in the @xmath17o layers , partial substitution of gd with th suppresses the magnetic / structural transition and then gives rise to superconductivity where @xmath8 reaches a maximum of 56 k at 20% th .
@xcite similarly , a narrow superconducting region was also observed in gdfeaso@xmath20 by introducing oxygen deficiency .
@xcite furthermore , o / f substitution in cefeaso also induces superconductivity with @xmath8 raised to 41 k. @xcite on the other hand , the rare - earth elements usually form an antiferromagnetically ordered state at very low temperatures .
for instance , cefeaso sequentially undergoes two afm - type transitions upon cooling down from room temperature , one associated with fe ( @xmath21 150 k ) and the other one attributed to ce ( @xmath22 3.4 k ) .
@xcite a similar situation happens in gdfeaso ( @xmath21 130 k , @xmath23 4.2 k ) .
@xcite on the other hand , cefepo is a paramagnetic(pm ) heavy fermion metal .
@xcite recent studies demonstrated that a ferromagnetic ( fm ) state of ce @xmath1 electrons , separating the afm state from the pm heavy fermion state , develops in the intermediate doping region in cefeas@xmath2p@xmath24o .
@xcite moreover , the co - end compounds , namely @xmath17co@xmath18o , demonstrate rich magnetic properties attributed to the interactions between the @xmath0 and @xmath1 electrons .
for example , cecoaso and cecopo exhibit an enhanced sommerfeld coefficient of 200 mj / mol - k@xmath25 , and their co - ions undergo a fm transition at @xmath26 = 75 k. @xcite lacoaso and lacopo are itinerant ferromagnets with a curie temperature of @xmath27 50 k and 60 k , respectively .
@xcite complex magnetic properties were observed in smcoaso and ndcoaso , in which the co - ions first undergo a fm transition and then an afm transition upon cooling from room temperature , @xcite followed by an afm transition of the @xmath1 electrons at @xmath28 = 5 k and @xmath29 = 3.5 k. @xcite in gdcoaso , both co and gd are magnetically ordered too , @xcite but there is no systematic study so far .
these indicate that the 1111-type iron pnictides may provide us a rare system to study the interplay of @xmath0 and @xmath1 electrons , and their emergent properties .
a systematic study of the interplay between @xmath0 and @xmath1 electrons would help elucidate the nature of superconductivity and magnetism in iron pnictides . in this paper , we synthesize and measure the physical properties of cefe@xmath2co@xmath3aso and gdfe@xmath4co@xmath5aso ( @xmath30 ) , which represent two different cases of @xmath1 magnetic moments .
a complete temperature - doping phase diagram is derived for each compound , showing remarkably rich physical properties tuned by @xmath0-@xmath1 interactions .
the article is organized as follows : immediately after the introduction , we describe our experimental methods in section ii .
section iiia characterizes the crystal structure of the compounds . in section iiib ,
c and d , we present the experimental results of the electrical resistivity , magnetic properties and specific heat for cefe@xmath2co@xmath3aso and gdfe@xmath4co@xmath5aso , from which their phase diagrams are derived ( section iiie ) .
finally , a summary is given in section iv .
co@xmath3aso with @xmath31 .
the inset shows the enlarged ( * 102 * ) peaks .
( b ) and ( c ) lattice parameters as a function of nominal co content for cefe@xmath2co@xmath3aso and gdfe@xmath4co@xmath5aso , respectively .
the insets plot the unit cell volume ( left axis ) and the @xmath32 ratio ( right axis ) as a function of doping.,width=307 ] polycrystalline samples of cefe@xmath2co@xmath24aso and gdfe@xmath4co@xmath33aso ( @xmath34 ) were synthesized in evacuated quartz ampoules by a two - step solid state reaction method .
high - purity ce ( 99.8@xmath35 ) , gd ( 99.9@xmath35 ) and as ( 99.999@xmath35 ) chunks , also co@xmath36o@xmath37 ( 99.97@xmath35 ) , fe@xmath38o@xmath36 ( 99.99@xmath35 ) , fe ( 99.9@xmath35 ) and co ( 99.9@xmath35 ) powders were used as raw materials . in the following ,
we describe the methods of synthesizing the cefe@xmath2co@xmath24aso compounds ; the same holds for the gd - compounds .
dehydrated co@xmath36o@xmath37 and fe@xmath38o@xmath36 were prepared by heating the powders at 1173 k for 10 h. first , ceas precursor was prepared by heating the mixture in a sealed evacuated quartz ampoule at 1073 k for 24 h , and then at 1323 k for 48 h. then , ceas , co@xmath36o@xmath37 , fe@xmath38o@xmath36 , fe and co powders were weighed according to the stoichiometric ratio , thoroughly ground , and pressed into pellets . the entire process was carried out in an argon - filled glove box .
finally , the pellets were placed in alumina crucibles , sealed into evacuated quartz ampoules , and then slowly heated to 1323 k at a rate of 100 k / h and annealed at this temperature for one week to obtain sintered pellets .
the structure of these polycrystalline samples was characterized by powder x - ray diffraction ( xrd ) at room temperature using a panalytical xpert mrd diffractometer with cu k@xmath39 radiation and a graphite monochromator .
measurements of the dc magnetic susceptibility , electrical resistivity and specific heat were carried out in a quantum design magnetic property measurement system ( mpms-5 ) and a physical property measurement system ( ppms-9 ) , respectively .
temperature dependence of the electrical resistivity was measured by a standard four - point method .
figure 1(a ) plots several representative xrd patterns of the cefe@xmath2co@xmath24aso polycrystalline samples .
the gdfe@xmath4co@xmath33aso compounds show similar xrd patterns ( not show here ) .
the vertical bars on the bottom denote the theoretically calculated positions of the bragg diffractions for cefeaso .
all the peaks can be well indexed based on the tetragonal zrcusias structure with a space group of p4/nmm .
no obvious impurity phases are detected , indicating high quality of these samples .
the shift of the ( * 102 * ) peak toward the right ( larger @xmath40 value ) with increasing co concentration , as shown in the right inset of fig.1(a ) , reveals a contraction of the crystal lattice with increasing co - concentration .
the lattice parameters , refined by the least square fitting , are plotted as a function of nominal co - concentration in fig.1(b ) and fig.1(c ) for cefe@xmath2co@xmath24aso and gdfe@xmath4co@xmath33aso , respectively .
different from cefeas@xmath2p@xmath24o where both the @xmath41- and @xmath42-axis shrink with increasing p - content , @xcite in cefe@xmath2co@xmath24aso and gdfe@xmath4co@xmath33aso , the @xmath42-axis shrinks significantly while the @xmath41-axis increases slightly with increasing co concentration , meaning that fe / co substitution increases three dimensionality in comparison with as / p substitution .
this becomes more pronounced for @xmath43 ( ce ) and @xmath44 ( gd ) , in which regime co - electrons order ferromagnetically at low temperatures .
the shrinkage of the @xmath42-axis reduces the distance between the ceo(gdo ) and feas / coas layers , and thus enhances the hybridizations and magnetic exchange interactions between @xmath0 and @xmath1 electrons .
the inset of fig.1(b ) and fig.1(c ) shows the unit cell volume and the @xmath32 ratio versus co concentration , both of which monotonically decrease with increasing co content .
the former decreases by 2.22@xmath35 and 2.2@xmath35 at @xmath10 and @xmath11 , respectively .
the shift of the xrd patterns and the variation of lattice parameters suggest that the co atoms are successfully incorporated into the crystal lattice .
( a ) @xmath45 .
( b ) @xmath46 .
( c ) @xmath47 .
the insets of ( a ) and ( b ) expand the low temperature regime .
the inset of ( c ) shows the derivative plots of the electrical resistivity with respect to temperature for @xmath48 = 0.9 , 0.95 and 1 , from which the transition temperatures of @xmath12 and @xmath26 can be determined.,width=307 ] the temperature dependence of the normalized electrical resistivity is shown in fig.2 for cefe@xmath2co@xmath24aso ( @xmath49 ) .
the resistive anomaly observed around @xmath50 150 k in cefeaso characterizes the almost coincident sdw - transition of fe @xmath0 electrons and the tetragonal to orthorhombic structural phase transition .
@xcite with further decreasing temperature , another resistive kink appears around @xmath51 k , marking the afm transition of ce .
@xcite upon partially substituting fe with co , the transition at @xmath52 is quickly suppressed and becomes hardly visible above @xmath53 .
meanwhile , superconductivity appears around @xmath54 , and the transition temperature reaches a maximum of @xmath55 13.5 k at @xmath56 [ see the inset of fig.2(a ) ] .
note that , for @xmath57 , no signature of a magnetic transition can be identified below @xmath8 in the electrical resistivity .
this is different from that of cefeaso@xmath2f@xmath24 in which a non - zero resistive transition shows up below @xmath8 attributed to the magnetic order of ce .
@xcite however , evidence for the afm transition of ce can be inferred from the magnetic susceptibility and specific heat ( see below ) . in comparison with cefeaso@xmath2f@xmath24 , @xcite
another difference is the low - temperature semiconducting behavior observed in the underdoped cefe@xmath2co@xmath24aso ( @xmath58 ) , which may originate from a disorder effect or kondo - like scattering .
the tiny amount of co atoms in the materials may act as magnetic impurities , leading to a resistive minimum as a result of the kondo effect .
the system eventually becomes metallic with further increasing co concentration .
figure 2(b ) presents the normalized electrical resistivity for @xmath59 , which decreases monotonically with decreasing temperature . in this doping region ,
only one pronounced resistive transition is seen around 4 k , which is attributed to the afm transition of ce @xmath1 electrons , as shown by the arrow in the inset of fig.2(b ) . for @xmath43
, the electrical resistivity demonstrates stronger temperature dependence with a broad hump around 100 k , likely attributed to the enhanced @xmath0-@xmath1 hybridizations as typically found in some kondo lattice compounds .
@xcite upon further co - doping , the low - temperature plateau , observed for @xmath59 , persists as a change in slope as clearly shown in the derivatives of the electrical resistivity with respect to temperature , @xmath60 , in the inset of fig.2(c ) .
one peak corresponds to the fm transition of co ( @xmath26 ) and the other , to the afm order of ce ( @xmath12 ) , both shifting to higher temperature with further increasing co concentration .
the magnetic transition of ce is broadened on the co - rich side ; we attribute it to the polarization effect of co - ferromagnetism which will be further illustrated in terms of magnetic susceptibility and specific heat .
( a ) @xmath61 .
( b ) @xmath62 .
the insets expand the low temperature regime.,width=321 ] = 0.07 ( a ) , 0.10 ( b ) , 0.15 ( c ) .
the insets show the upper critical fields @xmath63 obtained at the 90@xmath35 ( green square ) , 50@xmath35 ( blue circle ) , 10@xmath35 ( red triangle ) of the normal resistance r@xmath64.,width=321 ] figure 3 plots the temperature dependence of the normalized electrical resistivity for gdfe@xmath4co@xmath33aso ( @xmath65 ) at zero - field . in gdfeaso ,
the coincident fe - sdw transition and the structural phase transition decreases to @xmath66 130 k. @xcite at very low temperatures , a resistive kink appears around 3.6 k , as shown by the arrow in the left inset of fig.3(a ) , below which the gd @xmath1 electrons become antiferromagnetically ordered .
@xcite the @xmath52 is quickly suppressed upon substituting fe with co and becomes imperceptible around @xmath67 .
simultaneously , superconductivity shows up around @xmath68 = 0.07 , with the transition temperature reaching a maximum of @xmath69 k around @xmath70 [ see the right inset of fig.3(a ) ] .
similar to the ce - compounds , the gd - afm transition is not visible in the electrical resistivity of the superconducting samples ( @xmath71 ) , but it can be inferred from the magnetic susceptibility ( see below ) . upon further increasing the co concentration , no superconductivity can be observed down to 3 k , and the afm transition of gd shows up around 4 k for @xmath72 . in the underdoped gdfe@xmath4co@xmath33aso ( @xmath73 ) , the electrical resistivity also shows semiconducting behavior .
figure 3(b ) presents the normalized electrical resistivity for the heavily co - doped samples ( @xmath74 ) , which decreases monotonically with decreasing temperature , showing metallic behavior above 5 k. the electrical resistivity demonstrates a stronger temperature dependence for @xmath75 , which may be attributed to spin - wave scattering from the fm co. again , the gd @xmath1 electrons undergo an afm transition below 4 k. in particular , a step - like upturn transition appears at low temperature for @xmath76 , which can be clearly seen from the upper inset of fig.3(b ) .
for example , the electrical resistivity for @xmath77 increases abruptly at 3.9 k and then decreases below 3.2 k. such a resistive upturn might be associated with a gap opening at the afm transition temperature of gd .
the subsequent resistive decrease may arise from a magnetic reorientation , which is observed in the bulk properties too ( see below ) .
this step - like transition disappears in the stoichiometric gdcoaso material , in which only a small kink associated with the afm transition of gd @xmath1 electrons is observed around @xmath16 @xmath78 3.5 k , as shown in the lower inset of fig.3(b ) .
figure 4 presents the resistive superconducting transitions for @xmath68 = 0.07 , 0.10 , and 0.15 at various magnetic fields . in zero field , all the samples show a sharp transition with @xmath79 @xmath80 12.6 k , 17.6 k and 12.2 k for @xmath68 = 0.07 , 0.10 and 0.15 , respectively .
application of a magnetic field shifts the superconducting transition to lower temperatures and significantly broadens the transition width .
the latter is likely attributed to the vortex flow as previously discussed in other fe - based superconductors .
@xcite the insets of fig.4 plot the upper critical fields @xmath63 , which were derived at various resistive drops of the superconducting transition . following the whh model :
@xmath81 $ ] , @xcite
one can estimate @xmath82@xmath83(0 ) , which gives @xmath82@xmath83(0 ) = 19 t , 32 t and 13 t for @xmath68 = 0.07 , 0.1 and 0.15 , respectively .
the coherence length @xmath84 can be calculated from the ginzburg - landau formula @xmath85 , where @xmath86 is the quantum of magnetic flux .
this yields @xmath87 and 50 @xmath88 for @xmath89 and 0.15 , respectively . for cefe@xmath2co@xmath3aso , @xmath90 .
the open and filled symbols show the data measured at zero - field cooling ( zfc ) and field cooling ( fc ) , respectively . to demonstrate the existence of an afm transition of ce below @xmath8
, we plot @xmath91 for @xmath56 in the inset , from which a kink at @xmath12 can be identified.,width=307 ] .the superconducting transition temperatures @xmath8 ( k ) for cefe@xmath2co@xmath3aso , @xmath92 , determined from 90@xmath35 ( @xmath93 ( onset ) ) , 50@xmath35 ( @xmath93 ( mid ) ) and 10@xmath35 ( @xmath93 ( zero ) ) of the normal resistivity at @xmath8 and the onset of magnetic susceptibility @xmath94 . [ cols="<,^,^,^,^ " , ] for gdfe@xmath4co@xmath5aso , @xmath95 .
the open and filled circles show the zfc and fc data , respectively .
the inset shows the case for @xmath67 , where the afm transition of gd was observed below the superconducting transition at @xmath8 = 13 k.,width=321 ] co@xmath5aso , @xmath65 .
no difference was observed for the zfc- and fc - data .
note that the curves for @xmath96 and @xmath97 have been shifted by an offset of 0.05 emu / oe@xmath98mol .
the insets plot the inverse susceptibility as a function of temperature for @xmath99 ( a ) and @xmath11 ( b ) .
the red lines are fits to the curie - weiss law and the black stars are a fit to the mean field model
. the lower inset of ( b ) shows d@xmath91/d@xmath100 versus @xmath100.,width=321 ] the low temperature magnetic susceptibility @xmath101 for the gdfe@xmath4co@xmath5aso superconducting samples , is presented in fig.10 .
bulk superconductivity was observed for @xmath71 , with the superconducting volume fraction reaching over 50@xmath35 for @xmath68 = 0.1 , 0.12 and 0.15 .
moreover , a magnetic transition of gd can be tracked around 4 k as denoted by the arrow in the inset .
the superconducting transition temperatures @xmath8 , derived from the electrical resistivity and magnetic susceptibility , are summarized in table ii .
figure 11 plots the magnetic susceptibility of the non - superconducting samples , measured in a field of @xmath102 = 0.1 t. to make it clear , a value of 0.05 emu / oe@xmath98mol is added as an offset in fig.11(a ) . for @xmath103 , the magnetic susceptibility @xmath101 shows a peak around 4 k , which corresponds to the afm transition of gd , and is nearly independent of the co doping concentration . in the high temperature range ( @xmath104 200 k )
, the magnetic susceptibility @xmath105 can be well described by the curie - weiss law . in the inset of fig.11(a ) , we plot 1/@xmath105 versus @xmath100 for gdfeaso ; the red line is a fit to the curie - weiss law . the derived effective moment and paramagnetic curie temperature are 8.3@xmath106 and -17.1 k , respectively .
this effective moment is close to but somewhat lager than the free - ion moment of gd , possibly due to the contributions of @xmath107 electrons .
@xcite as the fe /
co substitution increases beyond @xmath108 , the behavior changes dramatically .
the susceptibility , as shown in fig.11(b ) , rises sharply at a temperature that increases with @xmath68 , reaching 75 k at @xmath109 . the inverse susceptibility for @xmath109
can be also fitted by the curie - weiss law at temperatures above 200 k [ see the upper inset of fig.11(b ) ] , with a slope consistent with a magnetic moment of 7.9 @xmath106 , very close to the free - ion moment of gd .
however , the value of @xmath110 = 2.3 k , is substantially small .
the downward curvature of the inverse susceptibility is characteristic of a ferrimagnet , in which two magnetic species are coupled antiferromagnetically .
the strong dependence of the transition temperature @xmath26 with co concentration [ see the lower inset of fig.11(b ) ] suggests that the fm coupling among co - atoms , rather than strong afm coupling to the gd , is the primary driver of the transition .
we model the behavior within a mean field approach , @xcite in which the magnetization of co atoms is given by @xmath111 , and that of the gd by @xmath112 . here
@xmath113(@xmath114 ) is the curie constant for co(gd ) , @xmath115 is the fm co - co coupling constant and @xmath116 is the afm co - gd coupling .
we assume that @xmath114 follows from the full atomic moment of gd , 7.9@xmath117 , and seek a solution in which @xmath118 and @xmath119 k. the solid star symbols in the inset of fig.11(b ) show a fit to the data with @xmath120 , confirming that the afm coupling between the two species is relatively weak . despite the weak coupling ,
the large magnetic moment of gd leads to an appreciable magnetization that opposes that of the co , causing a strong decrease in the combined magnetization at low temperature .
the data in fig.12 confirm this picture , which plots the field dependence of magnetization at different temperatures with the magnetic field up to 3 t. below @xmath121 , the magnetization is strictly linear in field and independent of co concentration .
for @xmath122 , in which the co moments are ferromagnetically ordered , spontaneous magnetization was observed below the curie temperature @xmath26 and its size increases with increasing the co concentration . for gdcoaso ,
the magnetization reaches 0.39 @xmath106 with a maximum near 60 k. similar results were also reported in lacoaso , ndcoaso and smcoaso compounds with a saturated moment of 0.46@xmath106 , 0.20@xmath106 and 0.18@xmath106 respectively .
@xcite the saturated moment of co in these compounds is very small , meaning that the ferromagnetism of co is itinerant in nature .
the fit shown in fig.11 , however , suggests that the co moment is closer to 0.5@xmath123 , indicating that the spontaneous magnetization in fig.12 is reduced by the negative magnetization induced in the gd . at low temperatures ,
the spontaneous magnetization has disappeared and the magnetization is again linear in field .
this contrasts sharply with the low temperature hysteretic behavior observed in cefe@xmath2co@xmath24aso , demonstrating that the admixture of @xmath1 and @xmath0 electrons dominates in cecoaso as compared with the fully localized @xmath1electrons of gdcoaso .
we expect that the gd blocks order antiferromagnetically at low temperature and tend to align the magnetically - ordered co blocks oppositely .
this type of behavior has been observed by means of neutron scattering in ndcoaso .
@xcite in that case , the afm interactions among the nd atoms and between nd and co are sufficiently strong to reorient the fm co layers into an antiparallel alignment . at zero field for cefe@xmath2co@xmath3aso , @xmath49 .
inset of ( a ) shows the data for low co concentrations ( @xmath124 ) .
inset of ( b ) plots @xmath125 versus @xmath126 for cefeaso and cecoaso .
the dashed lines are fits to @xmath127.,width=307 ] in order to further characterize the magnetic order of ce @xmath1 electrons in cefe@xmath2co@xmath24aso , we also measured the temperature dependence of specific heat @xmath128 .
figure 13 presents the low - temperature specific heat @xmath125 of several representative doping concentrations .
one can see that the specific heat @xmath125 shows a @xmath129-type transition for @xmath130 , characteristic of a typical afm transition .
the transition temperatures are consistent with those derived from the electrical resistivity and magnetic susceptibility as shown above . in previous measurements of neutron scattering and muon spin relaxation
, it was found that there exists a strong coupling between ce-@xmath1 and fe @xmath0 electrons in cefeaso .
@xcite in this case , suppression of the sdw transition of fe may have a significant impact on the magnetic order of ce . to confirm it
, we have performed a detailed measurement of low temperature specific heat for @xmath131 . as seen in the inset of fig.13(a ) , the afm transition of ce indeed shifts to higher temperature while suppressing the sdw transition of fe by fe / co substitution , indicating a kind of competing interaction between ce @xmath1 and fe @xmath0 electrons .
it is noted that , in the electrical resistivity @xmath132 , the sdw transition is barely visible for @xmath133 .
however , the continuous increase of @xmath12 up to @xmath134 may indicate that the critical doping concentration for suppressing the sdw transition is around @xmath135 , which is compatible if we extrapolate the sdw - phase boundary in the phase diagram as shown in fig.15(a ) .
furthermore , the specific heat @xmath125 of the superconducting samples shows a pronounced magnetic transition of ce around @xmath136 4 k. one notes that the specific heat jump @xmath137 at the superconducting transition ( about 90 mj / mol - k@xmath138 assuming @xmath139 ) is too tiny to be observed in comparison with that at @xmath12 . with further increase of co content ,
the afm transition of ce depends weakly on the doping concentration up to @xmath140 . above that , the transition is significantly broadened and its maximum is slightly shifted towards higher temperatures with increasing @xmath48 , being similar to the resistive results .
such behavior is likely attributed to the co - induced polarization on ce - moments , which may take place prior to the afm order of ce at @xmath12 .
the specific heat data can be fitted by @xmath141 at temperatures above the magnetic transition of ce [ see the dashed lines in the inset of fig.13(b ) ] , from which one can estimate the sommerfeld coefficient @xmath142 .
the derived @xmath142 value is summarized in the inset of fig.15 , which is nearly a constant for @xmath143 , but then increases with further increasing @xmath48 . in comparison with cefeaso ( @xmath142 @xmath144 65 mj / mol - k@xmath138 ) , the @xmath142 value of cecoaso is significantly enhanced ( @xmath145 186 mj / mol - k@xmath25 ) , implying an increase of the hybridization between @xmath1 and @xmath0 electrons upon substituting fe with co. the large electronic specific heat coefficient and the strong temperature dependence of the electrical resistivity [ see fig.2(c ) ] suggest strong electron correlations on the co - rich side .
we note that the @xmath142 value of cecoaso is also much larger than that of lacoaso ( @xmath146 20 mj / mol - k@xmath138 ) , @xcite excluding the possibility that the enhanced @xmath142 value is due to the magnetic scattering of co. ( blue circle , left axis ) and magnetic susceptibility @xmath91 ( red square , right axis ) for gdfe@xmath4co@xmath5aso , @xmath65.,width=321 ] we also performed measurements of the low temperature specific heat for gdfe@xmath4co@xmath5aso .
figure 14 shows the magnetic susceptibility @xmath101 and the specific heat @xmath125 of several representative doping concentrations . for @xmath147 ,
observations of a @xmath115-type anomaly in specific heat and a cusp in magnetic susceptibility unambiguously identify the afm transition of gd around @xmath148 k. with increasing the co content , a second anomaly develops in the heat capacity at lower temperature for @xmath149 , but disappears again at @xmath11 .
similar evidence can be also inferred from the magnetic susceptibility @xmath101 , which is most clearly seen in @xmath121 . as already described above , the second transition was also observed in the electrical resistivity [ see fig.3(b ) ] .
all these indicate that the system ( @xmath15 ) may undergo a magnetic reorientation transition below @xmath16 attributed to the coupling between the co @xmath0 and gd @xmath1 electrons .
co@xmath3aso ( a ) and gdfe@xmath4co@xmath5aso ( b ) , plotted as a function of nominal co content .
various symbols denote different types of transition temperatures .
the lines are guides to the eyes .
the inset of ( a ) plots the sommerfeld coefficient @xmath150 as a function of co concentration for cefe@xmath2co@xmath3aso.,width=307 ] based on the above experimental data , in fig.15 we present a magnetic and superconducting phase diagram for cefe@xmath2co@xmath24aso and gdfe@xmath4co@xmath33aso ( @xmath34 ) .
cefeaso is a bad metal , in which fe @xmath0 and ce @xmath1 electrons are antiferromagnetically ordered below 150 k and 4 k , respectively . upon substituting fe with co , the magnetic order of fe @xmath0 electrons
is suppressed around @xmath151 ( marked by sdw@xmath152 ) and superconductivity appears in a narrow range of @xmath153 ( marked by sc ) , indicating that the formation of superconductivity is closely related to the magnetism of fe in these compounds . at lower temperatures ,
ce @xmath1 electrons undergo an afm transition in the entire doping - range ( marked by afm@xmath154 ) ; the nel temperature @xmath12 slightly increases with suppressing the sdw transition of fe and then remains nearly unchanged up to @xmath155 . with further increasing co -
content ( @xmath156 ) , the co @xmath0 electrons form long range fm order , the curie temperature of which eventually increases with @xmath48 , reaching @xmath157 k in cecoaso ( marked by fm@xmath158 ) . on the co - rich side , the co - ferromagnetism has a strong polarization effect on the afm order of ce ( marked by po ) .
gdfeaso is also a bad metal , with the fe @xmath0 and gd @xmath1 electrons ordered antiferromagnetically below 130 k and 4 k , respectively .
the fe - sdw transition vanishes around @xmath70 and superconductivity appears in a range of @xmath7 . at lower temperatures ,
gd is antiferromagnetically ordered with @xmath16 @xmath78 4 k ( marked by afm@xmath159 ) , being nearly independent of the fe / co substitution .
meanwhile , long range fm order of co @xmath0 electrons develops above @xmath160 , with the transition temperature reaching 75 k in gdcoaso .
unlike cecoaso , however , gdcoaso shows ferrimagnetic behavior , with clear evidence that the gd moments oppose that of co , leading to a strong suppression of the magnetization at low temperatures .
moreover , for @xmath161 , it appears that a magnetic reorientation transition occurs below @xmath16 , as noted by @xmath162 in fig.15(b ) . as shown above , the 1111-type iron pnictides exhibit remarkably rich magnetic and superconducting properties via elemental substitutions .
different from cefe@xmath2co@xmath24aso and gdfe@xmath4co@xmath33aso , superconductivity was only observed around @xmath163 in cefeas@xmath2p@xmath24o , which seems to coexist with ferromagnetism of ce below 4 k. @xcite the magnetic order of ce persists up to @xmath164 in cefeas@xmath2p@xmath24o , but it changes from an afm- to a fm - type order while the sdw transition of fe is suppressed around @xmath165 0.3 .
@xcite in contrast , the afm order of ce - ions is remarkably robust against fe / co substitution .
its ordering temperature @xmath12 increases with @xmath48 while suppressing the sdw transition of fe , followed by a nearly unchanged value .
the persistence of the ce - afm order , with a slight dependence of @xmath12 on co - doping , seems to contradict the kondo physics , which effect would be enhanced when the interlayer distance is compressed by the fe / co substitution as evidenced from the enhancement of the sommerfeld coefficient in cecoaso ( @xmath142 @xmath166 200 mj / mol - k@xmath25 ) .
@xcite this is possibly due to the polarization of co - atoms caused by spin - orbit coupling which competes with the kondo effect .
the resulting afm exchange interaction between co @xmath0 and ce @xmath1 electrons is thus anisotropic in the spin space .
the multi - orbital characteristics may also add complications to the interplay among various intralayer and interlayer exchange interactions , leading to diverse behavior in the @xmath0-@xmath167 electron pnictides .
for example , in other @xmath17co@xmath18o systems , e.g. , re = nd and sm,@xcite the interlayer @xmath168-@xmath169 exchange interaction is strong enough to realign the co moments , leading to a subsequent afm transition of co at a lower temperature .
such transitions and magnetic structures have recently been proposed from neutron scattering in ndcoaso.@xcite however , in cefe@xmath2co@xmath24aso , ce - ions carry a small magnetic moment and the material shows a relatively weak lattice shrinkage along the @xmath42-axis upon substituting fe with co ; the resulting afm exchange interaction is likely not sufficiently strong to realign the co moments . instead ,
the ce moments are partially polarized by the internal field of co. however , the large moment of gd in gdfe@xmath4co@xmath33aso is robust against the fm order of co. the coupling between the gd and co lattices leads to ferrimagnetic behavior and also possible magnetic reorientation of co at low temperatures . since the fm order of co is generally observed in the @xmath17coaso families , @xcite it is desirable to extend such measurements to other series , to study the interplay of @xmath1 and @xmath0 electrons and its influence on superconductivity and magnetism in iron pnictides .
further investigations on the co - rich side is highly important in order to clarify the magnetic structures of @xmath0 and @xmath1 electrons in these systems .
in summary , we have systematically studied the transport , magnetic and thermodynamic properties of the polycrystalline samples cefe@xmath2co@xmath24aso and gdfe@xmath4co@xmath33aso ( @xmath30 ) and presented a complete doping - temperature phase diagram . it is found that both ce- and gd - compounds exhibit similar behavior on the fe - side , but rather distinct properties on the co - side . on the fe - rich side , superconductivity occurs upon suppressing the sdw transition of fe . at lower temperatures , the rare - earth elements ( ce and gd )
undergo an afm - type magnetic transition , whose temperature is slightly increased while suppressing the sdw - transition of fe , but then becomes nearly unchanged over a wide doping regime . on the co - rich side , co electrons form a long - range fm order , the transition temperature of which increases with increasing co content and reaches @xmath170 75 k in cecoaso and gdcoaso .
evidence of a strong polarization effect of the co - ferromagnetism on ce atoms is observed on the co - rich side .
but the gd is rather robust against fe / co substitution ; it is hardly polarized even in the case that co - ions are ferromagnetically ordered . however , the coupling between gd- and co - species gives rise to ferrimagnetic behavior on the co - side , which can be well described in terms of a mean field model that considers weak afm coupling between the gd - species and the co - species .
our findings suggest that , the delicate interplay of @xmath0-@xmath1 electrons rising from the competition of electron hybridizations and magnetic exchange coupling leads to remarkably rich physics in the iron pnictides .
we would like to thank c. geibel , e. d. bauer , r. e. baumbach , f. c. zhang , z. a. xu , and g. h. cao for useful discussions .
work at zhejiang university is supported by the national basic research program of china ( nos.2009cb929104 and 2011cba00103),the national science foundation of china ( nos.10934005 , 11174245 , 11274084 ) , zhejiang provincial natural science foundation of china and the fundamental research funds for the central universities .
work at los alamos national lab was performed under the auspices of the us doe and supported in part by the los alamos ldrd program .
work at the university of texas at dallas is supported by the afosr grant ( no.fa9550-09-1-0384 ) .
h. maeter , h. luetkens , y. g. pashkevich , a. kwadrin , r. khasanov , a. amato , a. a. gusev , k. v. lamonova , d. a. chervinskii , r. klingeler , c. hess , g. behr , b. bchner , and h. h. klauss , phys .
b * 80 * , 094524 ( 2009 ) . | we study the interplay of @xmath0 and @xmath1 electrons in the iron pnictides cefe@xmath2co@xmath3aso and gdfe@xmath4co@xmath5aso , which correspond to two very different cases of @xmath1-magnetic moment .
both cefeaso and gdfeaso undergo a spin - density - wave ( sdw ) transition associated with fe @xmath0 electrons at high temperatures , which is rapidly suppressed by fe / co substitution .
superconductivity appears in a narrow doping range : @xmath6 for cefe@xmath2co@xmath3aso and @xmath7 for gdfe@xmath4co@xmath5aso , showing a maximum transition temperature @xmath8 of about 13.5 k for ce and 19 k for gd . in both compounds ,
the @xmath1-electrons form an antiferromagnetic ( afm ) order at low temperatures over the entire doping range and co @xmath0 electrons are ferromagnetically ordered on the co - rich side ; the curie temperature reaches @xmath9 75 k at @xmath10 and @xmath11 . in the ce - compounds , the nel temperature @xmath12 increases upon suppressing the sdw transition of fe and then remains nearly unchanged with further increasing co concentration up to @xmath13 ( @xmath14 4 k ) . furthermore ,
evidence of co - induced polarization on ce - moments is observed on the co - rich side . in the gd - compounds ,
the two magnetic species of gd and co are coupled antiferromagnetically to give rise to ferrimagnetic behavior in the magnetic susceptibility on the co - rich side . for @xmath15 ,
the system undergoes a possible magnetic reorientation below the nel temperature of gd ( @xmath16 ) .
our results suggest that the effects of both electron hybridizations and magnetic exchange coupling between the @xmath0-@xmath1 electrons give rise to a rich phase diagram in the rare - earth iron pnictides .
pacs number(s ) : : 74.70.xa,71.20.eh,75.20.hr |
cygnus x-2 ( discovered by bowyer et al . 1965 ) is a very well studied bright low - mass x - ray binary ( lmxb ) .
hasinger & van der klis ( 1989 ) classified the brightest lmxbs into two sub - classes , i.e. the `` z '' sources and the `` atoll '' sources , on the basis of their correlated x - ray spectral and fast timing behaviour .
cygnus x-2 was classified as a z source . in the x - ray colour - colour diagram ( cd )
z sources trace out a z shaped track and move smoothly , without jumps , through the branches . in all the six
known z sources , except gx 349 + 2 , three branches have been identified , the horizontal branch ( hb ) , the normal branch ( nb ) and the flaring branch ( fb ) . the transition between the hb and the nb is called the hard vertex , and between the nb and the fb the soft vertex .
the variation of one single parameter , i.e. the mass transfer rate to the compact object ( ) , is thought to produce the tracks ( e.g. hasinger & van der klis 1989 , lamb 1991 ) .
recent studies of exosat data of the z sources ( see kuulkers et al .
1994a , 1996a , b ; kuulkers & van der klis 1996 ) indicate that the z sources can be divided into two groups , the cyg - like sources ( cyg x-2 , gx 5 - 1 and gx 340 + 0 ) and the sco - like sources ( sco x-1 , gx 349 + 2 and gx 17 + 2 ) .
the cyg - like sources show motion of the z pattern in the colour - colour diagram and the hardness - intensity diagram ( hid ) , their hbs are horizontal , and when the sources are on their short fbs the x - ray intensity generally decreases .
the sco - like sources , instead , do not display significant motion of their z pattern in the cd , their hbs are almost vertical , and the x - ray intensity increases when they move onto their extended fbs . it was proposed ( see kuulkers & van der klis 1995b , and references therein ) that the sco - like sources are viewed face - on ( low orbital inclination ) and the cyg - like sources more edge - on ( higher orbital inclination ) , and that the motion of the cyg - like sources could be caused by accreting material , such as a precessing accretion disk getting into the line of sight .
theoretical studies ( psaltis et al . 1995 ) indicate that a difference of magnetic field strength can explain some of the differences between the cyg- and the sco - like sources . in their model ,
the cyg - like sources have a somewhat higher magnetic field strength ( b @xmath1 gauss ) than the sco - like sources ( b @xmath2 gauss ) .
their so - called unified model can explain the z tracks in the cds and hids and the difference in the z shapes between the cyg - like and sco - like sources .
however , no explanation is given for the fact that the cyg - like sources display motion of the tracks , whereas the sco - like sources do not .
cygnus x-2 shows the most pronounced motion of the z pattern in the cd and the hid of all z sources .
one of the first observations of z pattern motion in the hid was found by vrtilek et al .
( 1986 ) using einstein mpc data .
they observed a factor of two increase in intensity between different epochs but also an increase of the intensity on time scales less than a day . at that time the z track behaviour of cygnus x-2 was not yet known , so they interpreted the variations as due to orbital effects .
nowadays we know that most of the intensity variations of cygnus x-2 are due to the non - periodic motion of the source through the z track , and most likely associated with changes in .
however , not all intensity variations that were found by vrtilek et al .
( 1986 ) can be explained in this manner . in their fig . 2
, three branches are present which now can probably be identified as nbs .
these branches are shifted with respect to each other , making this in retrospect the first indication that cygnus x-2 displays motion of the z - track in the hid .
clear evidence for z track motion in the cd and the hids was found when different exosat observations of were compared with each other ( hasinger et al .
1985b , hasinger 1987 , 1988 ) .
hasinger ( 1987 ) discussed six exosat observations of cygnus x-2 .
he found that the z tracks of different observations were displaced in the hids with respect to each other .
hasinger et al .
( 1985b ) and hasinger ( 1988 ) reported a state of cygnus x-2 with very low count rates .
the count rates , especially in the high energy bands , flared and the spectrum was hard . in the cd
this observation looked like a large fb , and it is displaced to harder colours with respect to other cygnus x-2 exosat observations ( see also kuulkers et al .
llllll nr & start ( ut ) & end ( ut ) & total time ( h ) & mode & binary phase@xmath3 + 1 & 1987 - 06 - 07 15:29 & 1987 - 06 - 10 21:43 & 6.74 & mpc1 , mpc3 , pc & 0.70 - 0.03 + 2 & 1988 - 06 - 10 05:33 & 1988 - 06 - 14 04:05 & 28.80 & mpc2 , mpc3 , pc & 0.14 - 0.54 + 3 & 1988 - 10 - 06 00:50 & 1988 - 10 - 07 23:40 & 18.24 & mpc1 , mpc2 , mpc3 , pc & 0.12 - 0.30 + 4 & 1989 - 10 - 22 01:41 & 1989 - 10 - 24 01:06 & 8.07 & mpc1 , pc & 0.81 - 0.01 + 5 & 1990 - 11 - 30 16:13 & 1990 - 12 - 02 13:31 & 2.32 & mpc1 , mpc3 & 0.91 - 0.11 + 6 & 1991 - 05 - 15 01:16 & 1991 - 05 - 16 10:00 & 7.83 & mpc1 , mpc3 & 0.71 - 0.85 + 7 & 1991 - 06 - 09 18:26 & 1991 - 06 - 11 06:41 & 11.32 & mpc1 , mpc2 , mpc3 & 0.33 - 0.48 + + vritlek et al .
( 1988 ) investigated the long - term temporal variability of cygnus x-2 using six different instruments on three satellites ( oso 8 , heao 1 , and einstein ) .
three count rate states between 2 and 10 kev were found . during the lowest state the spectrum was harder than during the other two states .
the motion of the z trough the cd and hid was again reported in early ginga observations of cygnus x-2 reported by hasinger et al .
they found that the z track in the hid during the october 1988 observation was shifted to higher intensities ( by a factor two ) with respect to the july 1988 observation . in the cd the z track of the october 1988 observation was shifted downward parallel to the nb with respect to the july 1988 observation .
the shape of the z track in both the cd and the hid was different between these two observations . during the july 1988 observation
the z track performed a loop in the hid at the nb - fb transition , such that along the fb the source first rose in intensity and later decreased . during the october 1988 observation
no such loop was found .
instead , along the fb the source immediately started to decrease in intensity . in the cd of the july 1988 observations the hb is diagonal and
the fb is large . in the october 1988 cd
the hb is horizontal and the fb is very small .
kuulkers et al .
( 1996a ) reanalysed all exosat observations of cygnus x-2 .
they also reported motion of the z pattern and different shapes of the z track .
they compared their results with all previous reports on cygnus x-2 , and described the phenomenolgy in terms of three different intensity levels ( possibly part of a continuous sequence ) . in the rare low level cygnus
x-2 displays a hard spectrum at the lowest intensities .
these episodes seemed to occur at binary phases 0.8 - 0.2 ( with phase 0.0 defined as the x - ray source superior conjunction ) , however not in each orbital cycle .
vritlek et al .
( 1988 ) and kuulkers et al .
( 1996a ) discussed various possible explanations , e.g. that at these times the secondary hides its heated face and/or part of our view of the inner disk region , so that most of the observed radiation then comes from a scattering hot corona surrounding the inner accretion disk , or that a tilted disk or a hot spot at the outer disk hide much of the inner disk region from our view .
the other two intensity levels were called the medium and the high level . a periodicity in the occurence of these intensity levels could not be found .
the shape of the z track in the hids was dependent on these intensity levels .
as the source moves onto the fb during high level episodes , the intensity immediately starts to decrease and the fb in the hid is almost horizontal .
they called this a colour - independent dip . when the source moves up the fb during medium level episodes the intensity first increases and then decreases .
the fb in the hid is also pointed upwards and therefore referred to as a colour - dependent dip .
no z track was observed during the low level .
the fast timing behaviour of z sources is closely related to the position of the sources in the cd and hid ( hasinger & van der klis 1989 ) . on all branches a very - low frequency noise ( vlfn ) and a high frequency noise ( hfn ) component in the power spectra
can be identified . on the horizontal and normal branches also low frequency noise ( lfn )
can be identified which decreases in amplitude from the hb to the upper normal branch and disappears on the lower normal branch .
quasi - periodic oscillations ( qpos ) are found on the horizontal branch ( called hbos ) with a frequency from @xmath412 hz at the left end of the hb up to @xmath455 hz near the hard vertex , and then remains constant upto halfway the normal branch .
however , wijnands et al .
( 1996a ) found a decrease of the frequency of what is probably the hbo down the nb in gx 17 + 2 . on the nb qpos ( nbos )
are observed with a frequency between 57 hz .
sometimes the hbo is seen simultaneously with the nbo , indicating different mechanisms are responsible for these qpos . on the flaring branch
sometimes a qpo ( fbo ) is seen with a frequency near 7 hz at the beginning of the fb , but rapidly increasing in frequency up to @xmath5 hz further up the fb .
the 7 - 20 hz flaring branch qpo has not been found on .
however , kuulkers & van der klis ( 1995a ) detected a qpo on the fb of cygnus x-2 with a frequency of 26 hz .
this qpo was found in an intensity dip on the fb and is believed to be a different phenomenon from the 7 - 20 hz fbo .
two z sources are known to show bursts .
gx 17 + 2 ( tawara et al . 1984 ; kahn & grindlay 1984 ; sztajno et al .
1986 ; kuulkers et al .
1994b , 1996b ) shows bona fide type i burst ( thermonuclear flashes on the neutron star surface , hoffman et al . 1978 ) .
in cygnus x-2 burst - like events have been reported ( kahn & grindlay 1984 ; kuulkers et al .
kuulkers et al .
( 1995 ) detected 9 burst - like events in the archival exosat data of cygnus x-2 .
these burst - like events had a short duration ( t@xmath63 s ) , did not show evidence for cooling , and did not occur in specific regions of the z track . to date it is unclear wether these burst - like events are genuine type i bursts . in this paper
we present an uniform analysis of all available data obtained by the large area counter ( lac ) instrument on board the ginga satellite .
cygnus x-2 was observed with the ginga satellite ( makino and the astro - c team 1987 ) during 7 observations . for a log of the observations we refer to table [ obslog ] . due to earth occultations and the high background in the south atlantic anomaly ( saa )
the observations were broken up into many pieces with a length of several tens of minutes . the lac instrument ( turner et al . 1989 ) on board ginga was used in several different observation modes : the mpc1 , mpc2 , mpc3 and the pc mode . in the mpc1 and the mpc2 ( hereafter mpc1/2 ) modes the source
was observed in 48 photon energy channels and the time resolution ranged from 62.5 ms to 16 s. in the mpc3 mode the source was observed in 12 channels with a best high time resolution of 7.8 ms . in the pc mode
only four overlapping channels were used and the time resolution was 1 or 2 ms . by manipulating the coarse gain ( cg ) amplifier , which had two levels , the energy range could be set to either 0.537 kev ( low cg level ) or to 0.318.7 kev ( high cg level ) . by changing the high voltage ( hv ) levels the energy range could be extended to even higher energies ( e.g. up to 60 kev ) . due to these different cg amplifier settings , hv level settings and lower discriminator settings ( see turner et al .
1989 ) different sets of energy channels were used at different times . llllll observation period & mode & band 1 & band 2 & band 3 & energy range used for the total intensity + & & kev & kev & kev & kev + 1987 and 1988 & mpc1/2 and mpc3 & 2.34.7 & 4.77.0 & 7.018.6 & 2.318.6 + 1991 & mpc1/2 & 2.34.7 & 4.77.0 & 7.018.6 & 2.318.6 + 1991 & mpc3 & 2.44.7 & 4.77.1 & 7.118.8 & 2.418.8 + 1989 and 1990 & mpc1/2 & 2.94.8 & 4.86.7 & 6.719.0 & 2.919.0 + 1989 and 1990@xmath3 & mpc1/2 and mpc3 & 1.23.8 & 3.87.6 & 7.619.0 & 1.219.0 + 1987 & pc & 0.62.9 & 2.99.1 & 7.912.3 & 0.69.1 + 0.812.3 + 1988 & pc & 0.82.9 & 2.99.1 & 7.912.3 & 0.89.1 + 0.812.3 + 1989 & pc & 1.25.7 & 1.215.7 & 5.717.9 & 1.217.9 + 1.215.7 + + generally the satellite attitude control system kept the viewing direction within 30 arcmin from the position of , which corresponds to a collimator transmission of @xmath7 70% . the systematic error in the observed count rates due to uncertainties in the collimator transmission
is of the order of a few precent ( hertz et al .
due to , e.g. , a long period without attitude control measurements the offset angle sometimes increased causing the collimator transmission to decrease to about 40% .
the lac collimator shows energy - dependent transmission effects due to reflection of soft x - rays ( below @xmath46 kev ) from the collimator walls at large offset angles ( turner et al .
this low energy reflection was not taken into account when the intensities were corrected for collimator transmission ( sect .
[ makingcdhid ] ) , which gives rise to an overestimation of the count rates for the lowest energy channels . when we encountered data with a collimator transmission below @xmath460 % we used figs .
13 and 14 of turner et al .
( 1989 ) in order to correct the count rates below 6 kev .
usually this means that the count rates were corrected for an overestimation of 34% .
sometimes the offset angle was even higher and the collimator transmission came below @xmath420% . at such high offset angle
the collimator response is independent of incident photon energy ( turner et al .
1989 ) . however , at these times the fractional uncertainty in the collimator transmission , and thus in the corrected count rates , due to aspect jitter is @xmath7 15% ( hertz et al .
for this reason , we did not use any data with a collimator transmission @xmath8% in our analysis . for our studies of the cds and the hids we used data rebinned to a time resolution of 4 seconds , which were background substracted and corrected for deadtime and aspect using the isas analysis system .
the pc binned data were not corrected for background because no good background determinations are available for them .
the 16 s resolution mpc1 data of the november 1990 and june 1991 observations had multiple register overflows and were left outside our analysis .
the 4 s resolution mpc1 data of the june 1987 , the october 1989 , the may / june 1991 observations and the 2 s resolution mpc2 data of the june / october 1988 and the june 1991 observations had single overflows and the true count rates could be recovered . in order to make cds and hids we calculated the soft colour , the hard colour and the total intensity .
the definitions of those quantities and the energy bands used in order to calculated them , are given in table [ bands ] .
instead of the colours and intensities , usually used in the literature , we used the logarithmic value of those quantities . any process which influences the intensity in the specified energy bands by multiplication , i.e. the instrumental response
, produces only a shift in the cd and the hids . using logarithmic values of the colours and the intensity
therefore makes it easier to compare data of the same source obtained by different instruments , by simply shifting the diagrams . also , the parametrization ( sect .
[ sz ] ) in this way does not depend on the actual values of the quantities or the count rates , but only on their changes .
it can now be used not only for the cd but for the hids as well .
usually the total intensity was taken as the sum of the count rates in the same energy bands as were used to calculate the colours . however , for the pc data of the 1987 and the 1988 observations the total intensity was defined as the count rate in all four energy channels .
for the 1989 pc data the count rates in the three lowest energy channels were summed , as the fourth channel only contained background photons .
the lowest energy boundary of the 1987 pc data differs slightly from the lowest boundary in the 1988 pc data ( table [ bands ] ) .
the effect on the z in the cd and hid is negligible and therefore we discuss the 1987 and 1988 pc data together .
the 1989 pc data is discussed seperately because of the different energy boundaries ( see below ) . in order to compare the mpc1/2 data with the mpc3 data we rebinned the mpc1/2 data to 12 channels using the same energy boundaries as the mpc3 data .
although in the 1991 observations the mpc3 data had slightly different energy boundaries compared to the mpc1/2 data ( table [ bands ] ) , the difference was small and had no significant effect on the place of the z in the cd and hid .
due to the hv and cg levels of 1989 and 1990 observations ( energy range 1 - 60 kev ) the energy boundaries of their spectral channels differ from those of the other five observations . due to
the broad energy channels the energy boundaries of these observations could not be succesfully interpolated to the boundaries of the others . therefore , we discuss the 1989 and 1990 observations
seperately from the other observations . in order to get an approximately uniform analysis we took for the mpc1 mode data colours which resembles the colours used in the other observations .
hasinger et al .
( 1990 ) introduced the concept of rank number of the z track in the cd .
this concept was refined by hertz et al .
( 1992 ) and by dieters & van der klis ( 1996 ) .
here we use the method described by dieters & van der klis ( 1996 ) for the measurement of the position along the z , with the difference that we perform all operations on the logarithmic colours and intensity values ( see sect .
[ makingcdhid ] ) .
we therefore transform the two logarithmic colour coordinates ( hard colour and soft colour ) into the coordinates ( the distance along the z track ) and ( the distance perpendicular to the z track ) . we also calculated the velocity ( ) and the acceleration ( ) along the z track , as defined as @xmath9/[{\rm t}(i+1 ) - { \rm t}(i-1)],\ ] ] where ( i ) is the position in the z track on time t(i ) , and @xmath10/[{\rm t}(i+1 ) - { \rm t}(i-1)],\ ] ] respectively ( see also dieters & van der klis ( 1996 ) ) .
the length of the nb is scaled to 1 and positions on the hb and fb are normalised to the length of the nb .
the hard vertex corresponds to = 1 , and the soft vertex to = 2 .
we applied the parametrization also on the z tracks in the hid .
we obtained a different set of and values which we call and .
the distribution of or is a measure for the thickness of the branches in the cd or hid , respectively .
to quantify the thickness of the branches we calculated the sample standard deviation of the and distributions using the method discribed in the appendix . as a measure of the overall velocity and acceleration in the cd
the standard deviations of the and distribution were used , again using the method outlined in the appendix . for the power - spectral analysis we used the mpc3 and pc data .
we made ffts of 128s data segments which resulted a frequency range of 0.007864 hz ( mpc3 7.8 ms resolution data ) or 0.0078256 hz ( pc 1.95 ms resolution data ) in the power spectra .
we calculated the ffts for the energy interval 1.218.6 kev ( june 1987 , june / october 1988 and may / june 1991 mpc3 observations ) or 1.219.0 kev ( november 1990 mpc3 observation ) . for the pc mode we used the sum of the energy intervals 0.812.3 kev and 0.69.1 kev ( june 1987 ) , or 0.812.3 kev and 0.89.1 kev ( june and october 1988 ) , or 1.217.9 kev and 1.215.7 kev ( october 1989 ) .
the average level of the photon counting noise , modified by deadtime processes ( the poisson level ) , was estimated and subtracted using a counter deadtime of 206 @xmath11s ( mpc3 data ) and 16.5 @xmath11s ( pc data ) ( see mitsuda & dotani 1989 ) . in the resulting power spectra
several components can be identified : * very low frequency noise ( vlfn ) , which we fitted with @xmath12 , where @xmath13 is the frequency , @xmath14 is the power - law index and @xmath15 the normalization constant . *
low frequency noise ( lfn ) and a high frequency noise ( hfn ) , are fitted with @xmath16 , where @xmath17 is the power - law index , @xmath18 the cut - off frequency , and @xmath19 the normalization constant , for the lfn ( l ) and hfn ( h ) , respectively .
* hbo , its harmonic , nbo and fbo , are fitted with lorentzians : @xmath20 , where @xmath21 the centroid frequency , @xmath22 the full width at half maximum ( fwhm ) of the qpo and @xmath23 the normalization constant .
the fractional rms amplitudes of the various noise components were determined by integrating their contribution over the following frequency ranges ; vlfn : 0.0011 hz , lfn : 0.01100 hz , hfn : 0.01100 hz .
we note that variations in the collimator transmission could influences the vlfn ( see e.g. the exosat data of gx 17 + 2 [ kuulkers et al .
1996b ] ) .
the errors in the parameter values were calculated using an error scan through @xmath24 space using @xmath25 . due to the poor statistics it was not always possible to fit a hfn component .
when hfn was fitted we used @xmath26 because in z sources this index was found to be consistent with zero ( hasinger & van der klis 1989 , dieters & van der klis 1996 ) . the best way to examine the correlations between the timing behaviour and the position on the z
track is to select a small part of the z track , using the parametrization and determine what the corresponding timing properties are for that segment .
however , our data did not allow this procedure as we did not have enough data at high time resolution for this procedure to give sufficient statistics .
therefore we first selected parts of the data that covered a relatively small range and afterwards determined an average value and range for the each data set .
the cds and hids for all observations are presented in figs .
[ fig1][fig89oktober_pc ] .
each point represents a time interval of 120 seconds of which @xmath27% contained data . in both the cd and hid of the mpc - data ( figs .
[ fig1]a and b ) a nb can be seen .
no hb is present .
the fb is only just seen in the cd , but it is clearly visible in the hid .
the count rate decreases as soon as the source enters this fb . in the hid ,
the upper nb is approximately twice as broad as the middle and lower nb .
this broadening of the upper nb is not seen in the cd .
the pc data ( figs .
[ fig87_88_pc]a and b ) only show a nb .
the june 1987 data were previously discussed by mitsuda & dotani ( 1989 ) , however , they used 16-second data points . due to the scatter induced by poisson statistics
they could not detect the broadening of the upper nb in the hid .
this observation was already discussed by hasinger et al .
parts of the data ( from 1988 june 12 02:16 to 1988 june 13 08:49 ut and from 1988 june 14 01:29 to 03:59 ut ) were obtained when the offset angle was about 1 , giving a collimator transmission of 40 - 60% .
we corrected these data for the overestimation of the count rates in the low photon energy bands ( sect .
[ observations ] ) . in the cds of the mpc and pc data ( figs .
[ fig1]c and [ fig87_88_pc]c , respectively ) clearly a hb , a nb and an extended fb are visible .
the horizontal branch is not horizontal but diagonal .
the hids of the mpc and pc data ( figs .
[ fig1]d and [ fig87_88_pc]d , respectively ) also show all three branches .
when the source moves into the fb , the count rate first slightly increases and then decreases .
the cd and hid of the mpc data ( figs .
[ fig1]e and f ) were reported before by hasinger et al .
( 1990 ) . in the cd and hid all three branches are visible .
the hb is really horizontal and when the source enters the fb the count rate decreases immediately . in the cd and hid of the pc data ( figs .
[ fig87_88_pc]e and f , not previously reported ) three seperate areas of points are visible . comparing these figures with the mpc data of this observation , the upper and the middle area ( indicated by hv and sv , respectively ) are probably near the hard and soft vertices , respectively .
this is confirmed by the fast timing analysis ( see sect . [ results_timing ] ) .
the status of the third , lowest area ( indicated by e ) is unclear .
the fast timing properties ( see sect . [ results_timing ] and table 5 ) suggest that cygnus x-2 was near the soft vertex , either on the nb , or on the fb .
part of the data of this observation ( from 1991 may 15 01:16 to 04:53 ut ) was obtained when the offset angle was large and the collimator transmission low ( 4060% ) .
we attempted a correction to the low photon energy channels for the count rate overestimation ( sect .
[ observations ] ) . during the time of low transmission , cygnus x-2 seemed to be on the nb , as judged from the cd , but in the hid the same data points were displaced from the nb . correcting the count rates for the low - energy flux overestimation was not enough to place the points exactly on the nb in the hid . in order to show more clearly the soft vertex and the beginning of the fb ( see below ) we did not include these points in the hid , and did not use them in our further analysis . in the cd ( fig .
[ fig2]a ) the hb is short and diagonal .
no clear fb can be seen , but a full nb can . in the hid ( fig .
[ fig2]b ) the beginning of a fb can be seen , which is not made up of the points with the low collimator transmission .
the count rate increases when the source enters the fb . during part of this observation ( from 1991 june 10 10:06 to 16:02 ut )
cygnus x-2 was observed with a very large offset angle and a collimator transmission of 1020% .
we did not include these data in our analysis because of the large uncertainties in the count rates ( sect .
[ observations ] ) . the remaining data ( figs .
[ fig2]d and e ) show a clear nb , and a hint for a fb , which is most clearly seen in the hid . when examining the data at higher time resolution we see a more developed fb .
no hb is observed .
when the source moves onto the fb the count rate decreases immediately .
the fb is approximately horizontal .
notice that the very broad nb in the hid corresponds to a narrow nb in the cd . comparing the five observations we find the following . the overall intensity level , defined as the mean count rate on the nb , changed by a factor of 1.34 between the observations .
the lowest intensities were observed in june 1988 and may 1991 , the highest intensities in october 1988 and june 1991 .
intermediate intensities were observed in june 1987 . following the classification of kuulkers et al .
( 1996a ) our lowest overall intensities correspond to their `` medium level '' and our highest overall intensities to their `` high level '' .
we shall adopt this terminology in what follows . as we also observe a level intermediate between these two ,
it is obvious from our data that the medium and high levels are probably part of a continuous range instead of a discrete set .
we did not observe cygnus x-2 when it was in the so - called `` low level '' ( kuulkers et al .
we confirm the previously reported ( hasinger et al .
1990 ; kuulkers et al .
1996a and references therein ) correlation between the shape of the z in the cds and hids and the overall intensity level . in the _ medium level _
the hb is not horizontal but oriented under an @xmath445angle in the cd .
it is less steep in the hid . in the cd
the hb is quite short compared to the nb . in the hid the hb and nb are of the same length .
the nb is nearly exactly vertical in the hid , with almost no changes in intensity .
the fb is very well developed in the cd , with large colour changes . in the hid , when the source enters the fb the count rate first increases , then , further up the fb , decreases . in the _ high level _ the hbs in both cd and hid
are nearly horizontal . in the cd
the hb is again short compared to the nb , while in the hid it is approximately the same length as the nb .
the nb in the hid shows a positive correlation between hard colour and intensity .
the fb in the cd is hardly visible , with hardly any colour changes . in the hid
the fb is well developed and horizontal ( little colour changes ) , and when the source moves onto the fb the count rate immediately decreases .
_ `` intermediate '' level _ shows characteristics of both the medium and the high level . due to lack of data
nothing can be said about the orientation of the hb during the intermediate level .
the hard colour of the nb in the hid shows a positive correlation with intensity , similar to the high level , although not as clear .
the fb in the cd is hardly visible , with small colour changes , also similar to the high level .
when the source moves on the fb in the hid , which is not horizontal , this is similar to the medium level , the count rate immediately decreases ( like in the high level ) . in the cd and hids ,
the z - track shifts downward ( to softer colours ) , parallel to the nb in the cd , when the overall intensity level increases . in the medium level
it is at its highest position ( hard colour ) , in the intermediate level it is lower , and in the high level it is lowest .
when we compare the nbs in the hids we see that the branch width changes with overall intensity . in the high level the nb is broader than in the medium level .
the nb in the intermediate level is only broad in its upper part . comparing the hard colour curves ( figs .
[ fig2]c and f ) , we see that the character of the motion through the nb depends on intensity level . in the medium level
the colours change slower than in the high level .
these two new results are discussed in the next section .
the parametrization makes it possible to investigate the kinematics of the motion along the z track in more detail ( see dieters & van der klis 1996 ) . when examining as a function of time
it is evident that the source does not jump through the branches , but moves smoothly along the z track . in the high level cygnus x-2
seems to move faster up and down the normal branch than when the overall intensities are lower ( compare figs .
[ fig2]c and f ) . in order to investigate this we calculated the velocity ( ) and acceleration ( ) distributions ( sect .
[ sz ] ; dieters & van der klis 1996 ) .
we find that the distributions are symmetric about zero : there is no difference in the motion up and down the z track .
the scatter of and increases from the hb , through the nb , to the fb indicating that the source moves through the z most slowly on the hb , faster on the nb and fastest on the fb . on the nb the scatter of and increases
when the overall intensity increases . in order to quantify this we determined the standard deviations of ( ) and ( )
. on the nb , and show a strong correlation with overall intensity ( see figs .
[ standevs]a and b ) .
they increase by a factor @xmath41.4 and @xmath41.5 , respectively , when the overall intensity increases by a factor of 1.34 .
no correlations between and on the fb and the overall intensity were found . for
the hb the data do not allow a conclusion on this point . in order to examine the width of the nb in the hid we applied the parametrization on the z track in the hid ( sect .
the time series are symmetric around zero .
the lack of points on the hbs and fbs makes a definitive conclusion about the evolution of the width of the z from the hb , via the nb to the fb , impossible .
however , when two or more branches are present the z width seems to increase from the hb to the nb , and further on the fb . as mentioned earlier ( sect .
[ compare ] ) we find that the nb width in the hid increases when the overall intensity increases .
this effect is visible as a strong increase in the standard deviation of on the nb , with overall intensity ( fig .
[ standevs]c ) .
when the overall intensity increases by a factor @xmath41.34 , the standard deviation of on the nb increases by a factor @xmath41.75 . on the other branches the standard deviation of stays approximately the same at all overall intensities .
when we examine the nb width in the cds we do not find any increase in the standard deviation of with overall intensity : the nb has approximately the same width at all intensity levels in the cds . in order to be able to compare these observations roughly with the other observations , and because of the very broad photon energy channels of the mpc3 mode ( see sect .
[ makingcdhid ] ) , we made cds and hids of the mpc 1 data only . because during the november 1990 observation the mpc1 mode was used only very briefly , we combined the october 1989 ( dots ) and november 1990 ( crosses ) data in fig .
[ fig3 ] . in both the cd and the hid the hb and
the nb are clearly visible .
no fb is present .
an upward curve can be seen at the beginning of the hb during the october 1989 observation .
such an upward curve of the hb was previously reported by kuulkers et al .
( 1996a ) in the 1985 exosat observation on day 301/302 .
the colour points of the november 1990 observation are displaced with respect to the october 1989 data . in order to include the mpc3 data and to find out on wich part of the z track the source was in during the november 1990 observation
, we had to use different colours , with broader energy bands .
for comparison we did the same for the october 1989 data .
the energy bands used are given in table [ bands ] .
figure [ fig89_90_all_mpc ] shows a clear hb and nb for the october 1989 data ( dots ) , as before , and the curve upwards of the hb , although less pronounced .
the november 1990 data ( crosses ) show only a hb which is shifted in colour with respect to the october 1989 data .
no upward curve is visible in the november 1990 data . the october 1989 pc data cd and the hid ( fig . [ fig89oktober_pc ] )
show a nb , the hard vertex and parts of the hb . in the hid ,
the points ( indicated by the h ) on the hb with the lowest intensity and low hard colour are not on the hb in the cd .
they are placed on the hard vertex in the cd .
the fast timing behaviour would suggest that those points are at the left end of the hb ( see sect .
[ hbo ] ) the 1989 and 1990 observations were performed using different energy bands as compared to the other observations .
therefore , we can not directly compare their overall intensity levels .
however , if we assume that the difference in the total count rate between 2.318.6 kev and 2.919.0 kev is minor we see that the overall intensity of the october 1989 observation is about the same as for the june 1987 observations .
there is no nb available for the november 1990 observation so no direct comparison with the other observations of the mean nb intensity can be made .
however , the shift of the hb during this observation with respect to the october 1989 hb indicates that the overall intensity of the november 1990 observation was higher than the overall intensity of the october 1989 observation .
therefore , we conclude that the october 1989 observation was taken at an overall intensity level in between the medium and high level , and the november 1990 observation during a high level episode .
the results of the power spectra analysis are shown in tables 4 and 5 . since in the hb
the vlfn is low ( e.g. hasinger & van der klis 1989 ) , we were in some cases unable to measure the vlfn , at the left end of the hb . in the june 1987 and june 1988 observations we fixed the power law index of the vlfn when cygnus x-2 is on the hb .
we have high time resolution data of the same part of the z track ( especially near the soft vertex ) but during different overall intensity levels .
the comparison of these data reveals a very significant difference of the vlfn properties between the medium and high overall intensity levels in the lower part of the nb . during the medium overall intensity observations ( june 1988 and may 1991 ) we detect a weak ( 0.5 - 1% rms amplitude ) and flat ( power law index of @xmath41 ) vlfn on the lower nb and near the soft vertex . the rms amplitude and power law index increase to @xmath45.3% and @xmath41.4 , respectively , on the upper fb in the june 1988 observation .
however , during the high overall intensity observations ( october 1988 and june 1991 ) we detect a very strong ( 8.5% and 4.5% rms , respectively ) and steep ( index of 1.4 - 1.7 ) vlfn on the lower nb and near the soft vertex .
the vlfn decreases in strength to about 6.7 % rms amplitude on the fb in the october 1988 observation .
we note that this strong vlfn near the soft vertex can not be caused by changing collimator response . during the intermediate overall intensity observation ( june 1987 ) we see
vlfn with a strength of @xmath28% rms and power law index of @xmath41.1 near the soft vertex , which increases to @xmath44.2% and @xmath41.4 , respectively , on the fb .
this gradual change in the strength and steepness of the vlfn as a function of overall intensity level is clearly visible in figs .
[ power_spectra]a - i .
this figure shows the power spectra in the medium level , the intermediate level and the high level at approximately the same values .
it is remarkable that when cygnus x-2 is in the high level the vlfn near the soft vertex is strong and steep and that when the source is in the medium level the vlfn at the same values is much weaker and less steep . during intermediate levels
the vlfn is weak and flat on the nb but strong and steep on the fb .
the correlation of the amplitude and steepness of the vlfn near the soft vertex with mean overall intensity level is not strict . at the highest overall intensities ( 1991 june )
the amplitude is about the half of the one at the second highest overall intensity ( 1988 october ) , and the spectrum is less steep . due to the scarcity of high time resolution data when was on the hb or near the hard vertex and the difficulty in determining , comparisons of the lfn between the observations are difficult .
the june 1987 and 1988 observations show a lfn with an rms of @xmath44.7 % on the hb which decreases to @xmath43.0 % on the hard vertex and upper nb .
the november 1990 observation shows that the rms of the lfn increases from the beginning of the hb to the hard vertex ( from @xmath45.8 % to @xmath46.5 % rms ) . the same is seen in the october 1989 observation ( lfn rms from 4.5 % to 5.5 % ) , but in this observation the rms decreases to 3.8 % when the source moves into the upper nb . no clear correlation between the power law index ( 0.00.3 ) or the cut - off frequency ( 4.0 13 hz ) with
when lfn or nbo were fitted it was difficult to fit the hfn simultaneously , possibly due to interference between them ( see also kuulkers et al .
1994a ) . when it was necessary to fit a hfn component it had an rms between 12% and cut - off frequencies between 10 and 20 hz .
the october 1989 pc data show a strong hfn ( @xmath4 6% rms ) with a high cut - off frequency between 30 and 50 hz .
normal branch qpos ( nbos ) were detected in the medium and intermediate level observations . to our suprise
, none were detected in the high level observations ( see fig . [ power_spectra ] ) , although we had high timing data in the lower nb for them .
the rms 2@xmath29 upper limits ( @xmath30% ) are significantly lower than the rms ( 12.5% ) of the nbos in the medium and intermediate levels .
this also shows that during different overall intensity levels the rapid x - ray variability at the same position on the z track differs significantly from each other .
when nbos are seen no clear relation is detected between the amplitude of the nbo and .
however , the fwhm and the centroid frequency seems to increase from the middle of the nb to the fb .
the fwhm increases from @xmath42.4 hz to @xmath413 hz and the centroid frequency from @xmath45 hz to @xmath48 hz . in the june 1987 ,
the june 1988 and may 1991 observations it was impossible to make a distinction between a broad nbo or a hfn on the lower fb , although a nbo provided a better fit ( e.g. for the june 1987 observation @xmath24/dof = 27.5/29 for the nbo versus 31.0/30 for the hfn ) .
horizontal branch qpos are seen during all overall intensity levels , on the hb as well as in the upper nb .
taking all data into account , we find that the frequency of the hbo increases from @xmath431 hz at the left end of the hb to @xmath455 hz at the hard vertex . the pc data of the june 1987 and october 1989 observations indicate that when the source moves down the nb the hbo frequency decreases again to @xmath4 47 hz . the rms amplitude decreases from the hb to the upper nb , from @xmath44% to @xmath41% .
llllllllllll no . & date & start ( ut ) & t@xmath31 & mode & time res .
& & orbital phase@xmath32 & @xmath33 & @xmath34 & @xmath35 + & & & ( s ) & & ( s ) & & & & & ( s ) + i & 1988 - 06 - 12 & 03:56:34 & @xmath4 8 & mpc2 & 2 & 1.000 & 0.33 & @xmath36105 & 1.92 & 3.66 + ii & 1988 - 06 - 12 & 10:25:52 & @xmath4 4 & mpc2 & 2 & 0.931 & 0.36 & @xmath3694 & 3.52 & 2.00 + iii & 1988 - 06 - 13 & 10:47:27 & @xmath4 8 & mpc2 & 0.0625 & 1.229 & 0.47 & @xmath364.1 & 0.63 & 1.51 + iv & 1988 - 10 - 06 & 02:32:25 & @xmath4 2 & mpc2 & 0.0625 & 1.222 & 0.11 & @xmath3613.3 & 1.74 & 1.09 + v & 1989 - 10 - 22 & 08:04:47 & @xmath4 4 & mpc1 & 4 & 1.41.5@xmath37 & 0.84 & @xmath38 & 2.67 & 4.00 + + + as already mentioned in sect .
[ pcmodes ] , the pc data of october 1989 show several points at low intensities and high soft colour .
those points are placed in the cd near the hard vertex but in the hid near the left end of the hb .
the power spectrum corresponding to these points is shown in fig .
[ harmonic ] .
a hbo at 25 hz is clearly visible ( 3.8% rms amplitude ) and indicates that cygnus x-2 was indeed at the left end of the hb ( see fig .
[ fig89oktober_pc ] , the points marked h ) .
a second qpo is visible at @xmath449 hz with a rms amplitude of 2.5% .
hasinger et al .
( 1985a ) and hasinger ( 1987 ) found already evidence for the harmonic of the hbo in the exosat data of cygnus x-2 .
however , the frequency ratio was smaller ( [email protected] ) than expected for a second harmonic .
our frequency ratio ( [email protected] ) indicates that this feature is indeed the second harmonic of the hbo .
we found five 28 s bursts similar to the burst - like events reported by kuulkers et al .
the times and the properties are given in table [ bursts ] . in figs .
[ fig1]c and e and fig .
[ fig3]a it is indicated where in the z track the bursts occured .
the occurence of the burst seems to be uncorrelated with the overall intensity level , the orbital phase , or the location in the z track ( although no burst were detected on the fb ) . for the ratio , @xmath33 , of the average persistent flux to the time - averaged flux emitted in the bursts ( which reflects the ratio between the gravitational and nuclear burning energy per gram of accreted matter for thermonuclear bursts ) , we could only obtain lower limits ( table [ bursts ] ) , since there are many interruptions by saa and/or earth occultations .
they were calculated using the average count rate from the start of the last data gap to just before the burst .
the other typical burst parameters @xmath34 , the ratio of the mean persistent pre - burst flux and net peak burst flux , and @xmath35 , the ratio of the total integrated net burst flux and the net burst peak flux ( a representation of the burst duration ) were also calculated ( see table [ bursts ] ) .
the count rates used in order to calculate the burst parameters were corrected for deadtime , background and aspect
. burst iv was observed during time when the satellite was slewing to the source .
therefore , during that time the collimator transmission was low ( @xmath40 50 % ) and the uncertainty on the count rate large .
the count rate for burst iv was not corrected for the overestimation of the count rates in the low photon energy bands , due to the reflection of low energy photons ( below 6 kev ) agains the collimator walls ( see sect . [ observations ] ) . due to the low time resolution of the data obtained during the occurence of bursts i ,
ii and v , we did not examine the burst profiles and spectral properties of these events .
the burst profiles , for different energy bands , of bursts iii and iv are shown in figs .
[ burst88junimpc2h ] and [ burst88oktobermpc2h ] , respectively .
these bursts do not show evidence for spectral cooling as would expected for bona fide type i bursts .
instead , it is clearly visible that burst iii shows evidence for spectral hardening : the higher the energies , the broader the event .
also , the post - burst count rate is higher than the pre - burst count rate .
this effect is most prominent at the higher energies .
although these phenomena are not seen during event iv , no evidence for spectral cooling is seen either . in the nine bursts in the exosat data of cygnus x-2 hints for spectral hardening
were already found ( kuulkers et al .
1995 ) , although the poor statistics made a definite conclusion impossible . taking all bursts into account
we conclude that these events are probably not bona fide type i. we searched for qpos and periodic oscillations during burst iii and iv and found none .
statistics were insufficient to set meaning full upper limits .
we have found several characteristics in the behaviour of cygnus x-2 which seem to depend on overall intensity level . in
the following we discuss some of those characteristics in more detail and suggest possible explanations for the different overall intensity levels and associated phenomena .
we find conclusive evidence that the rapid x - ray variability at the same point on the nb differs between different observations .
we find a different kind of high timing behaviour near the soft vertex when cygnus x-2 is in the medium level and the high level ( see fig . [ power_spectra ] ) . in the medium level weak
, flat vlfn is detected and a pronounced nbo occurs . in the high level
the vlfn is stronger and steeper , and the nbo is not detectable . in the intermediate intensity level ,
the vlfn is strong and steep in the fb , but weak and flat in the nb , and the nbo is about the same strength as during the medium level .
our results are the first unambigous detection of differences in the rapid x - ray variability at a specific position in the z track ( near the soft vertex ) between different observations ( see dieters & van der klis 1996 for a study of sco x-1 with respect to this issue ) .
so far , we find no evidence for any difference between the properties of the x - ray variability during different intensity levels on other parts of the z track ( such as the hb )
. however , our data do not allow a very detailed comparison of the timing behaviour on those parts of the z track .
it is thought ( van der klis et al .
1985 , 1987 ; hasinger & van der klis 1989 ; lamb 1991 ) that variations in are responsible for the changes in the rapid x - ray variability along the z and the formation of the z track , and that for this reason the rapid x - ray variability is closely related to the position of the source on the z track . however , the shifting and shape changing of the z track in cygnus x-2 and other z sources suggest that variations in can not explain all aspects of the formation of the z track .
we , therefore , suggest that both the shape of the z track and the rapid x - ray variability are not totally determined by the mass accretion rate onto the compact object , but also partly by a so far unknown process , which may also cause the long - term intensity variations ( see sect .
[ longterm ] ) .
two other z sources , gx 5 - 1 and gx 340 + 0 , display motion of their z track in the cd ( kuulkers et al . 1994a ; kuulkers & van der klis 1996 ) , although not as pronounced as
. kuulkers et al .
( 1994a ) and kuulkers & van der klis ( 1996 ) did not find any significant changes in the rapid x - ray varability in gx 5 - 1 and gx 340 + 0 , respectively , when the z track moved through the cd and hid .
if the motion of the z track in the cd and hid in cygnus x-2 , gx 5 - 1 , gx 340 + 0 are caused by the same phenonemon , we then predict that the rapid x - ray variability in gx 5 - 1 and gx 340 + 0 changes when the z tracks of those sources move through the diagram .
the reason why so far no changes have been found is most likely due to the much lower amplitude of the motion of the z through the cd and hid of gx 5 - 1 and gx 340 + 0 , as compared to , which suggest that the amplitude of the difference of the rapid x - ray variability would also be much smaller .
moreover , in cygnus x-2 sofar only differences in the rapid x - ray variability were found near the soft vertex and not on e.g. the hb , while gx 5 - 1 and gx 340 + 0 were mainly observed in the hb and upper nb , and hardly near the soft vertex .
kuulkers et al .
( 1994a ) and kuulkers & van der klis ( 1996 ) could not make a good comparison of the rapid x - ray variability near the soft vertex when the z track of gx 5 - 1 and gx 340 + 0 , respectively , moved in the cd and hid .
van der klis ( 1991 ) suggested that the vlfn could be ( partly ) due to motion of the source along the z track . in our analysis , we found that both the vlfn fractional amplitude and the velocity of motion along the z track increase when the overall intensity level increases .
however , the strongest vlfn was not observed during the most extreme high level , but during the less extreme , slightly lower one , whereas the highest velocity was indeed observed during the most extreme high level .
therefore , we conclude that the motion along the z track is at most only partly responsible for the vlfn . as differences in nb slope in the hid can not explain this , part of the intensity variations causing the vlfn must take place perpendicular to the nb .
the increase in velocity along the z track when the overall intensity increases indicates that the spectrum changes more rapidly during high level episodes than during medium level episodes .
if the z track is traced out by changes in the mass accretion rate onto the neutron star , then an increase in the velocity of motion along the z track indicates that changes more rapidly when the overall intensity level increases .
however , it then seems unlikely that variations in can explain the shifts and changes in shape of the z track between intensity levels ( see also sect .
[ difference ] ) .
it is possible that the increase in velocity of motion along the z track is due to the same phenomenon causing the z tracks to shift and the shapes to change .
the width of the nb in the hid increases when the overall intensity increases , while the width of the nb in the cd remained approximately constant .
this indicates that the difference in the width of the nb in the hid is due to intensity variations and not due to spectral variations .
however , at the same time the velocity of motion along the z track increases , which indicates an increase in spectral variations .
so , in the hid , when the source moves perpendicular to the z track ( change in intensity ) , it also moves along the z track ( change in spectrum ) .
perhaps colours are well correlated with , but the intensity varies also due to another ( unspecified ) process that is more prominent in the high overall intensity level .
several models have been proposed ( priedhorsky & holt 1987 , and references therein ) in order to explain long - term intensity variations in x - ray binaries , e.g. long - term variations in and precessing accretion disks .
precessing neutron stars have also been proposed to explain the variations ( see priedhorsky & holt 1987 ; schwarzenberg - czerny 1992 , and references therein ) .
long - term changes in the mass accretion rate have been proposed ( see priedhorsky & holt 1987 ) to explain the long - term intensity variations in low - mass x - ray binaries other than cygnus x-2 ( e.g 4u 1820 - 30 , the rapid burster , aql x-1 , excluding her x-1 ) . however , kuulkers et al .
( 1996a ) and wijnands et al .
( 1996b ) argued that the long - term intensity variations in cygnus x-2 can not be due to variations in the mass accretion rate .
the main argument is that variations in the mass accretion rate are thought to produce motion of the source along the z track ( see e.g. hasinger & van der klis 1989 ) , while the z track is observed during several different intensity levels . in order to explain
long term intensity variations , not related to orbital variations , in high - mass x - ray binaries and her x-1 , precessing accretion disks have been proposed ( see priedhorsky & holt 1987 and references therein ) . the recent detection ( smale et al . 1996 , wijnands et al .
1996b ) of a 78 day period in the rxte , vela 5b ( see also smale & lochner 1992 ) and ariel v all sky monitor data of cygnus x-2 favours a precessing accretion disk in cygnus x-2 .
however , explaining the long - term x - ray variations of cygnus x-2 with a precessing accretion disk is not without serious contradictions .
if we assume that , during the medium overall intensity level , the emission region is blocked by ( part of ) the accretion disk and much radiation is absorbed , scattered and/or reflected , the power spectrum should be blurred and quasi - periodic oscillations ( both the nbo and the hbo ) should be harder to detect than when the emission region is not blocked by the accretion disk ( the high level ) .
however , we see exactly the opposite : no nbo is detected during high overall intensity levels .
the upper limits are significantly lower than the actually detected values during the medium levels . if we assume that not during the medium level but during the high level the emission region is blocked by the accretion disk , then the difference between the power spectra are not totally unexpected .
however , it is difficult to explain the increase of the count rate when the accretion disk is blocking our view of the emission region .
another possibility is that the accretion disk blocks , during the medium overall intensity levels , a localised emission area ( e.g. the neutron star surface ) , which is not blocked during the high levels . in this region
no nbo occurs and therefore the fractional amplitude of the nbo during the high level is diluted by the additional flux .
however , not only the strength of the nbo should be affected , but also the strength of the hbo , except when the hbo would originate from this localised area , in which case we would expect an inverse effect .
this , however , is unlikely , because during all intensity levels hbos are observed at approximately the same fractional amplitude .
therefore , it is unlikely that the increase in count rates from the medium to the high level is caused by an localised emission region , which is sometimes hidden from our view by a precessing accretion disk , or by a precessing accretion disk , which sometimes blocks part of the total x - ray radiation . kuulkers et al.(1996a ) proposed that the difference in overall count rates in the cyg - like sources is caused by anisotropic emission , the radiation being scattered preferentially into the equatorial plane by a puffed - up inner disk structure . in this model , with increased scattering in the high level , the decrease in nbo amplitude follows naturally due to light travel time effects in the scattering process .
our observations therefore support their model .
another option is that the differences between the levels are due to changes in the properties of the emission region .
a precessing neutron star could produce changes in the inner disk region . due to the precession of the neutron star the orientation of the magnetic field with respect to the accretion disk changes . according to numerical calculations ( psaltis et al .
1995 ) the strength of the magnetic field is enough for the field to have a profound effect on the spectrum of the source and possibly also on the count rate .
a change in the effective magnetic field , caused by a different orientation of the field , could maybe explain the different z tracks during the different levels .
also , the rapid x - ray variability would be affected , although it is not clear if this could explain the differences that we found .
the behaviour of the hbo should differ between the different intensity levels due to the different effective magnetic field , but no difference is found . it is also not clear why the nbo should differ between intensity levels . the decrease of the hbo frequency from the hard vertex down the nb in the june 1987 and october 1989 pc data allows us to apply the method described by wijnands et al .
( 1996a ) , in order to derive an upper limit for the equatorial magnetic field strength . assuming that the radius of the neutron star is @xmath41
cm we derive an upper limit of @xmath42 g on the star s magnetic field at the magnetic equator .
this value is similar to the value found for gx 17 + 2 ( wijnands et al .
this upper limit is consistent with numerical computations on the x - ray spectrum of cygnus x-2 ( psaltis et al .
1995 ) . as noted by wijnands et al .
( 1996a ) , a further decrease of the frequency down the nb will reduce this upper limit .
the fact that we see the hbo frequency decrease on the nb in cygnus x-2 supports the interpretation that the qpo on the nb in gx 17 + 2 , which also decreases in frequency ( wijnands et al .
1996a ) , is the hbo .
we observed cygnus x-2 over a broad range of intensities .
the source alternated between the known medium and high overall intensity levels and also sometimes in a state in between these two levels , both in intensity and other source characteristics .
we found new correlations between several characteristics of the source and the overall intensity level : * the velocity and acceleration along the normal branch increase when the overall intensity increases . during the high intensity level cygnus x-2 moves
faster and more irregularly up and down the normal branch than during lower levels . *
the width of the normal branch in the hardness - intensity diagram increases when the overall intensity increases . *
the very - low frequency noise near the soft vertex increases in amplitude and in steepness when the overall intensity increases .
* the normal branch quasi - periodic oscillation is not detectable ( @xmath40 0.9 % rms ) near the soft vertex when is in the high level . during the other levels
a nbo can be easily detected at a level of 12.5 % ( rms ) .
the following correlations were already known and are confirmed by our analysis : * the overall intensity changes with a factor of @xmath41.34 between the medium and the high overall intensity level . *
the shape of the z - track in the colour - colour diagram and hardness - intensity diagram changes when the overall intensity changes . in the colour - colour diagram
the horizontal and the flaring branch become more horizontal when the overall intensity increases . when the source enters the flaring branch during the medium overall intensity level the intensity first increases and later decreases .
when the source enters the flaring branch during the high overall intensity level the intensity immediately starts to decrease .
* the whole z - track shifts to softer colours when the overall intensity increases .
we also found several other previously unreported phenomena : * clear detection of overall intensity level episodes in - between the medium and the high overall intensity level , indicating that the different overall intensity levels are part of a continouos range instead of a discrete set . *
the frequency of the horizontal branch quasi - periodic oscillation decreases down the nb , giving a model dependent upper limit on the magnetic field strength at the magnetic equator of @xmath43 g. * detection of five bursts in the ginga data , which do not show the characteristics of bona fide type i bursts .
+ this work was supported in part by the netherlands organization for scientific research ( nwo ) under grant pgs 78 - 277 and by the netherlands foundation for research in astronomy ( astron ) under grant pgs 781 - 76 - 017 .
ek acknowledge receipts of an esa fellowship .
for the sample @xmath44 , @xmath45 , @xmath46 , @xmath47 let burington r.s . , may d.c . , 1970 , handbook of probability and statistics , with tables , second edition , mc graw - hill new york - london , p.177 bowyer s. , byram e.t . , chubb t.a . , friedman h. , 1965 , science 17 , 894 crampton d. , crowler a.p .
, 1980 , pasp 92 , 147 dieters s. , van der klis m. , 1996 , a&a submitted hasinger g. , 1987 , in : helfand d.j . , huang z. ( eds . ) the origin and evolution of neutron stars , iau 125 , p.333 hasinger g. , 1988 , in : tanaka y. ( ed . ) physics of neutron stars and black holes , p.97 hasinger g. , van der klis m. , 1989 , a&a 225 , 79 hasinger g. , langmeier a. , sztajno m. , pietsch w. , gottwald m. , 1985a , iau circ .
4153 hasinger g. , langmeier a. , pietsch w. , sztajno m. , 1985b , sp.sc.re .
40 , 233 hasinger g. , van der klis m. , ebisawa k. , dotani t. , mitsuda k. , 1990 , a&a 235 , 131 hertz p. , vaughan b. , wood k.s . , et al . , 1992 ,
apj 396 , 201 hoffman j. a. , marshall h. l. , lewin w. h. g. , 1978 , nat 271 , 630 kahn s.m .
, grindlay j.e . , 1984 ,
apj 281 , 826 kuulkers e. , van der klis m. , 1995a , a&a 303 , 801 kuulkers e. , van der klis m. 1995b , in : bhringer h. , morfill g.e .
, trmper j.e .
seventeenth texas symposium on relativistic astrophysics and cosmology , ann .
new york ac . sci .
759 , p.344 kuulkers e. , van der klis m. , 1996 , a&a 314 , 567 kuulkers e. , van der klis m. , oosterbroek t. , et al . , 1994a , a&a 289 , 795 kuulkers e. , van der klis m. , oosterbroek t. , van paradijs j. , lewin w.h.g . , 1994b , in : holt s.s
. , day c.s .
the evolution of x - ray binaries , american institute of physics , p.539 kuulkers e. , van der klis m. , van paradijs j. , 1995 , apj 450 , 748 kuulkers e. , van der klis m. , vaughan b.a .
, 1996a , a&a 311 , 197 kuulkers e. , van der klis m. , oosterbroek t. , van paradijs j. , lewin w.h.g . , 1996b , mnras in press lamb f.k . , 1991 , in : ventura j. , pines d. ( eds . ) neutron stars : theory and observations , kluwer , dordrecht , nato asi series c , 344 , p.445 makino f. , the astro - c team 1987 , ap
25 , 223 mitsuda k. , dotani t. , 1989 , pasj 41 , 557 priedhorsky w.c .
, holt s.s . , 1987 , sp.sc.re .
45 , 291 psaltis d. , lamb f.k . ,
miller g. , 1995 , apj 454 , l137 schwarzenberg - czerny a. , 1992 , a&a 260 , 268 smale a. p. , lochner j. c. , 1992 , apj 395 , 582 smale a. p. , kuulkers e. , wijnands r. a. d. , 1996 , iau circ .
6452 sztajno m. , langmeier a. , trmper j. , van paradijs j. , lewin w.h.g .
, 1986 , mnras 222 , 499 tawara y. , hirano t , kii t. , matsuoka m. , murakami t. , 1984 , pasj 36 , 861 turner m.j.l .
, thomas h.d . , patchett b.e . , et al . , 1989 ,
pasj 41 , 345 van der klis m. , 1991 , in : ventura j. , pines d. ( eds . ) neutron stars : theory and observations , kluwer , dordrecht , nato asi series c 344 , p.27 van der klis m. , jansen f. , van paradijs j. , van den heuvel e. p. j. , lewin w. h. g. , 1985 , nat 316 , 225 van der klis m. , jansen f. , van paradijs j. , et al . , 1987 , apj 313 , l19 vrtilek s.d . ,
kahn s.m . , grindlay l.e . , seward f.d . , helfand d.j . , 1986 ,
apj 307 , 698 vrtilek s.d . ,
swank j.h . ,
kelly r.l . ,
kahn s.m . , 1988 , apj 329 , 276 wijnands r. a. d. , van der klis m. , kuulkers e. , et al .
, 1996a , apj 469 , l5 wijnands r. a. d. , kuulkers e. , smale a. p. , 1996b
, apj 473 , l45 | we have analysed all available x - ray data on the low - mass x - ray binary cygnus x-2 obtained with the ginga satellite .
a detailed analysis of the spectral and fast timing behaviour of these 4 years of data provides new insights in the behaviour of this z source .
we confirm the previously observed recurrent patterns of behaviour in the x - ray colour - colour and hardness - intensity diagrams consisting of shifts and shape changes in the z track .
however , we find a continuous range of patterns rather than a discrete set .
the source behaviour in the diagrams is correlated with overall intensity , which varied by a factor of 1.34 in the ginga data .
we find that when the overall intensity increases , the mean velocity and acceleration of the motion along the normal branch of the z track increase , as well as the width of the normal branch in the hardness - intensity diagram .
contrary to previous results we find that , during different observations , when the source is at the same position in the normal branch of the z track the rapid x - ray variability differs significantly . during the kuulkers et al .
( 1996a ) `` medium '' level , a normal branch quasi - periodic oscillation is detected , which is not seen during the `` high '' overall intensity level . also , during the high overall intensity level episodes the very - low frequency noise on the lower normal branch is very strong and steep , whereas during the medium overall intensity level episodes this noise component at the same position in the z track is weak and less steep .
the explanation of the different overall intensity levels with a precessing accretion disk is difficult to reconcile with our data .
furthermore , we found that the frequency of the horizontal branch quasi - periodic oscillation decreases when cygnus x-2 enters the upper normal branch , giving a model dependent upper limit on the magnetic field strength at the magnetic equator of @xmath0 g. we also report five bursts , with durations between two and eight seconds , whose occurrence seems to be uncorrelated with location in the z track , overall intensity level or orbital phase .
the burst properties indicate that they are not regular type i bursts . |
thanks to a number of elegant neutrino oscillation experiments in the past few decades , it has been well established that neutrinos are massive and lepton flavors are mixed @xcite . in the framework of three - flavor neutrino oscillations ,
the lepton flavor mixing is described by a @xmath13 unitary matrix @xmath14 , i.e. , the pontecorvo - maki - nakagawa - sakata ( pmns ) matrix @xcite , which is conventionally parametrized in terms of three mixing angles @xmath15 and one dirac cp - violating phase @xmath16 , namely , @xmath17 where @xmath18 and @xmath19 for @xmath20 have been defined .
the global - fit analysis of solar , atmospheric , reactor and accelerator neutrino oscillation experiments @xcite yields three mixing angles @xmath21 , @xmath22 , @xmath23 and two neutrino mass - squared differences @xmath24 and @xmath25 in the case of normal mass ordering @xmath26 ( nmo ) , while @xmath21 , @xmath27 , @xmath28 and @xmath29 , @xmath30 in the case of inverted mass ordering @xmath31 ( imo ) .
see table 1 for a summary of the latest global - fit results from ref .
@xcite .
besides precision measurements of the known mixing parameters , the primary goals of ongoing and forthcoming oscillation experiments are to pin down neutrino mass ordering ( i.e. , the sign of @xmath32 ) , to measure the leptonic cp - violating phase @xmath16 , and to determine the octant of @xmath33 ( i.e. , @xmath34 or @xmath35 ) . in order to study
the experimental sensitivities and better understand future experimental results , we should pay particular attention to the mikheyev - smirnov - wolfenstein ( msw ) matter effects on the propagation of neutrino beams in a medium @xcite . roughly speaking , current and future neutrino oscillation experiments can be categorized into three different types , in which terrestrial matter effects on neutrino oscillations always play an important role .
ccccccccc & & & & + & & best - fit & & @xmath36 range & & best - fit & & @xmath36 range + @xmath37 & & @xmath38 & & @xmath39 @xmath40 & & @xmath38 & & @xmath39 @xmath40 + @xmath41 & & @xmath42 & & @xmath43 @xmath44 & & @xmath45 & & @xmath46 @xmath47 + @xmath33 & & @xmath48 & & @xmath49 @xmath50 & & @xmath51 & & @xmath52 @xmath53 + @xmath16 & & @xmath54 & & @xmath55 @xmath56 & & @xmath57 & & @xmath55 @xmath56 + + + @xmath58 & & @xmath59 & & @xmath60 @xmath61 & & @xmath59 & & @xmath60 @xmath61 + + @xmath62 & & @xmath63 & & @xmath64 @xmath65 & & @xmath66 & & @xmath67 @xmath68 + + * _ medium - baseline reactor neutrino experiments _ the reactor experiments with a baseline length @xmath69 and a neutrino - beam energy @xmath70 , such as juno @xcite and reno-50 @xcite , are sensitive to the oscillations driven by both @xmath71 and @xmath32 .
hence , they will be able to determine neutrino mass ordering and precisely measure oscillation parameters .
it has been found @xcite that the earth matter effects for juno are as large as @xmath72 , significantly affecting the determination of @xmath73 and @xmath71 , whose precisions are estimated to be @xmath74 and @xmath75 , respectively @xcite . * _ long - baseline accelerator neutrino experiments _ for the long - basline accelerator experiments t2k @xcite ( @xmath76 and @xmath77 ) , no@xmath78a @xcite ( @xmath79 and @xmath80 ) and lbnf - dune @xcite ( @xmath81 and @xmath82 ) , it is the relative sign between the matter potential for electron neutrinos ( or antineutrinos ) and @xmath32 that changes the oscillation probability of @xmath83 ( or @xmath84 ) , opening another possibility to pin down neutrino mass ordering . the difference between oscillation probabilities of neutrinos and those of antineutrinos implies leptonic cp violation , which however suffers from a contamination induced by the cp - asymmetric earth matter .
in addition , the neutrino super - beam experiments ess@xmath78sb ( @xmath85 and @xmath86 ) and moment ( @xmath87 and @xmath88 ) have also been proposed to measure the cp - violating phase with relatively low energy neutrinos and short baseline lengths @xcite . *
_ huge atmospheric neutrino experiments _ the experiments pingu @xcite , orca @xcite , and hyper - kamiokande @xcite will implement huge ice or water cherenkov detectors to precisely measure atmospheric neutrinos , for which a wide range of energies ( @xmath89 ) and baseline lengths ( @xmath90 ) should be considered .
though neutrinos and antineutrinos can not be distinguished in these experiments , the msw resonance in the earth matter occurs either in neutrino oscillations for nmo or in antineutrino oscillations for imo .
therefore , matter effects help determine neutrino mass ordering .
a @xmath36 significance can be reached at the ical detector of ino , which can also discriminate between @xmath91 and @xmath92 events @xcite . in principle , for any neutrino energy and baseline length , one can exactly calculate neutrino and antineutrino oscillation probabilities in the earth matter by numerical methods . however , it is obviously difficult in this way to reveal the underlying physics for neutrino oscillations and to fully understand the numerical results .
for this reason , two theoretical approaches have been suggested to study the earth matter effects .
first , one can establish an exact relation between the effective parameters in matter , i.e. , the mixing matrix @xmath93 ( parametrized in terms of three mixing angles @xmath94 and one cp - violating phase @xmath95 ) and three neutrino masses @xmath96 ( or mass - squared differences @xmath97 ) , and the intrinsic parameters in vacuum .
for example , @xmath98 holds exactly for a constant matter density @xcite , where the jarlskog invariant in vacuum @xcite is defined by @xmath99 $ ] with the greek and latin letters running over @xmath100 and @xmath101 , respectively , and likewise for @xmath102 and @xmath93 .
another example is the toshev relation @xmath103 @xcite in the standard parametrization . in addition
, the notion of unitarity triangles has also been introduced to describe leptonic cp violation @xcite , and the exact and approximate relations between the unitarity triangles in matter and those in vacuum have been found in refs .
although these exact relations are interesting in themselves , they are in practice not useful to directly explain experimental observations and extract fundamental oscillation parameters .
the second approach is to expand the oscillation probabilities in terms of small perturbation parameters , which can be @xmath104 and the smallest mixing angle @xmath105 @xcite .
another choice is @xmath106 , where @xmath107 is the matter potential from the coherent forward scattering of neutrinos on the background particles and defined as @xmath108 , with @xmath109 being the fermi constant , @xmath110 the number density of background electrons , and @xmath111 the neutrino energy . in the case of low neutrino energies or low matter densities ,
an expansion in @xmath112 is useful to show the corrections of matter potential to the oscillation probabilities in vacuum . in the seminal paper by freund @xcite , the analytical approximations for three - flavor neutrino oscillation probabilities
have been systematically studied , and the formulas are valid as long as the oscillation driven by @xmath71 has not developed and the corresponding msw resonance is not reached .
the latter condition corresponds to @xmath113 @xcite , namely , @xmath114 where @xmath115 is the matter density . for the earth matter ,
the electron fraction is @xmath116 and @xmath117 $ ] with @xmath118 being the avogadro s number .
although freund s formulas actually work even for @xmath119 , it has been shown in ref .
@xcite and ref . @xcite that the series expansion of @xmath120 in terms of @xmath121 is problematic in the region of low energies or small matter densities , where @xmath122 .
more accurate approximate formulas for low energies @xmath123 have been derived in ref .
@xcite by retaining @xmath124 . however , the analytical results in refs .
@xcite are not applicable for large matter effects and higher neutrino energies .
furthermore , a critical problem for the @xmath125 expansion is related to the atmospheric resonance @xmath126 , where the function @xmath127^{1/2}$ ] can not be expanded correctly . as we will show later , @xmath124 and @xmath128 are two key parameters to avoid any difficulties associated with the low - energy solar resonance and the high - energy atmospheric resonance , respectively .
in fact , analytical formulas for arbitrary neutrino energies and baseline lengths are derived in refs .
@xcite , where the resonances related to @xmath71 and @xmath32 have been treated carefully by introducing a few intermediate rotation angles for basis transformations .
thus , the analytical results can be cast into a simple and compact form , in which the eigenvalues of the zeroth - order hamiltonian and rotation angles , instead of intrinsic mixing parameters , are involved . since all the existing analytical approximations are not fully satisfactory , we are well motivated to derive a new set of analytical formulas for neutrino oscillation probabilities , which fulfills the following three criteria : 1 . they are valid for arbitrary neutrino energies and any baseline length .
such formulas are applicable to atmospheric neutrino experiments .
they are expressed in terms of intrinsic oscillation parameters , and in a simple and compact form .
any complicated formulas are not very useful in practice .
they give accurate values of oscillation probabilities , under the condition that the first two criteria are met at the same time .
for this purpose , we expand the oscillation probabilities in terms of @xmath121 , but retain the parameter that corresponds to @xmath124 ( @xmath128 ) in the case of low ( high ) energies or small ( large ) matter densities . in addition ,
an @xmath0-gauge neutrino mass - squared difference @xmath129 is introduced so as to seek an optimal value of @xmath0 that greatly simplifies approximate formulas .
the remaining part of this paper is organized as follows . in section 2
, we briefly review the basic strategy to derive analytical formulas of neutrino oscillation probabilities .
we introduce @xmath129 and demonstrate that it is suggestive of simple analytical formulas for @xmath6 .
the oscillation probabilities in the special case of @xmath6 are presented in section 3 , where the mapping between effective and intrinsic mixing parameters is also obtained as a by - product .
the accuracies of the analytical formulas are examined and compared with previous ones .
finally , we summarize our main results in section 4 . some useful formulas are listed in three appendices .
in the framework of three - flavor neutrino oscillations , the effective hamiltonian responsible for the evolution of neutrino flavor eigenstates in matter is given by @xmath130 \ ; .
\label{eq : heff_f } \ ] ] in the case of a constant matter density , i.e. , a constant value of @xmath107 , we then have two distinct ways to derive the exact oscillation probabilities .
first , one can diagonalize the effective hamiltonian by using a unitary transformation @xmath131 where @xmath96 for @xmath132 are effective neutrino masses in matter , and @xmath93 is the effective pmns matrix , which can also be parametrized in terms of three mixing angles @xmath133 and one cp - violating phase @xmath95 . in terms of these effective parameters , it is straightforward to write down the oscillation probabilities @xmath134 as follows @xmath135 \sin^2 \widetilde{f}^{}_{ji}+ 8\widetilde{j } \sum_\gamma \epsilon^{}_{\alpha \beta \gamma } \sin \widetilde{f}^{}_{21 } \sin \widetilde{f}^{}_{31 } \sin \widetilde{f}^{}_{32 } \ ; , \label{eq : pab_eff } \ ] ] where @xmath136 $ ] and @xmath137 with @xmath97 have been defined in the same manner as for neutrino oscillations in vacuum , and @xmath138 is the baseline length .
the probabilities for antineutrino oscillations @xmath139 can be obtained by replacing @xmath140 in eq .
( [ eq : pab_eff ] ) and @xmath141 everywhere in the effective parameters .
second , according to the cayley - hamilton theorem , the evolution matrix @xmath142 of neutrino flavor eigenstates is determined by three eigenvalues of the effective hamiltonian @xmath143 and the matrix elements of @xmath143 @xcite , namely , @xmath144 where @xmath145 denotes the @xmath13 unit matrix and the relevant coefficients are @xmath146 with @xmath147 being the eigenvalues of @xmath143 .
the neutrino oscillation probabilities are simply given by @xmath148 , while the results for antineutrino oscillations can be derived by changing @xmath149 and @xmath141 in the effective hamiltonian . in order to simplify the analytical formulas as much as possible ,
we tentatively introduce a generic definition of neutrino mass - squared difference @xmath150 where @xmath4 is a real and positive parameter .
it is evident that @xmath151 reduces to the conventional definitions of atmospheric neutrino mass - squared differnce @xmath152 for @xmath153 , @xmath32 for @xmath154 , and @xmath155 for @xmath156 . in the global - fit analysis of neutrino oscillation data ,
the first two definitions have been used in ref .
@xcite in the imo and nmo cases , respectively , while the last one has been implemented in ref .
@xcite for either neutrino mass ordering .
another definition @xmath157 , corresponding to @xmath158 , has been advocated by parke @xcite and demonstrated to be advantageous to reactor antineutrino experiments . although for quite a different reason , as we will show later , the introduction of @xmath151 in eq .
( [ eq : dels ] ) with @xmath158 turns out to be very useful in simplifying the approximate formulas of oscillation probabilities . with the help of @xmath151 ,
the effective hamiltonian @xmath143 can be rewritten as @xmath159 where @xmath145 is the identity matrix of rank three and @xmath160 with @xmath161 and @xmath162 .
note that the first term on the right - hand side of eq .
( [ eq : heff_new ] ) is flavor - independent and thus irrelevant for neutrino oscillations . in the formalism shown in eqs .
( [ eq : ch ] ) and ( [ eq : si ] ) , the evolution matrix is now @xmath163 and only the eigenvalues of @xmath164 need to be calculated . to find out the eigenvalues of @xmath164 , it is more convenient to convert into the mass basis in vacuum via @xmath165 , where the neutrino mass term , i.e. , the first term on the right - hand side of eq .
( [ eq : m_f ] ) , becomes diagonal .
more explicitly , we have @xcite @xmath166 where @xmath167 for @xmath132 are three elements in the first row of the pmns matrix .
therefore , it is expected that a proper choice of @xmath0 will be helpful in reducing the complexity of three eigenvalues , and thus the final oscillation probabilities . furthermore , it is interesting to notice that only the matrix elements @xmath167 for @xmath132 are involved in @xmath168 and @xmath169 , so a suitable value of @xmath0 is anticipated to be mainly associated with @xmath170 and @xmath171 , or @xmath37 in the standard parametrization . now it is time to derive the oscillation probabilities by using eqs .
( [ eq : ch ] ) and ( [ eq : si ] ) .
first of all , the eigenvalues @xmath172 ( for @xmath173 ) of @xmath164 or equivalently @xmath168 can be obtained by solving the following eigen - equation @xmath174 where the relevant coefficients are @xmath175 \right\ } \ ; , \quad \\ d & = & - \alpha \left [ \widehat{a } \eta|u_{e1}^{}|^2 + \widehat{a } \left ( \eta - 1 \right ) |u_{e2}^{}|^2 + \alpha \eta ( \eta - 1 ) \big ( 1 + \widehat{a } |u_{e3}^{}|^2 \big ) \right ] \ ; .
\nonumber \label{eq : bcd } \ ] ] the eigenvalues of the effective hamiltonian have been known for a long time @xcite , but it has recently been noticed that the results depend also on neutrino mass ordering @xcite . to be explicit , taking @xmath176 , we have @xmath177 \ ; , \nonumber\\ \lambda_{2}^ { } & = & -\frac{b}{3 } - \frac{1}{3\delta_*^ { } } \sqrt{x^2 - 3 y } \big [ z - \sqrt{3(1-z^2 _ { } ) } \big ] \ ; , \nonumber\\ \lambda_{3}^ { } & = & -\frac{b}{3 } + \frac{2}{3\delta_*^ { } } z\sqrt{x^2 - 3y } \ ; , \label{eq : lambda_nmo } \ ] ] for the nmo ; or @xmath178 \ ; , \nonumber\\ \lambda_{2}^ { } & = & -\frac{b}{3 } + \frac{1}{3\delta_*^ { } } \sqrt{x^2 - 3 y } \big [ z - \sqrt{3(1-z^2 _ { } ) } \big ] \ ; , \nonumber\\ \lambda_{3}^ { } & = & -\frac{b}{3 } - \frac{2}{3\delta_*^ { } } z\sqrt{x^2 - 3y } \label{eq : lambda_imo } \ ; , \ ] ] for the imo , where we have defined @xmath179 \ ; , \nonumber\\ y & = & \delta_{*}^{2 } \bigg\{\widehat{a}(1 - |u_{e3}^{}|^2 ) + \alpha \big[1 + \widehat{a } - \widehat{a } ( 1 - |u_{e3}^{}|^2)(\eta - \frac{|u_{e1}^{}|^2}{1 - |u_{e3}^{}|^2})\big ] + \alpha^2 ( 1 - \eta ) \bigg\ } \ ; , \nonumber\\ z & = & \cos \bigg\ { \frac13 \arccos \frac{\delta_*^ { } \big[2 x^3 - 9xy + 27\delta_*^{3 } \alpha \widehat{a } ( 1 + \alpha - \eta \alpha ) |u_{e1}^{}|^2\big ] } { 2 |\delta_*^{}|\sqrt{(x^2 -3y)^3}}\bigg\ } \ ; .
\label{eq : xyz } \ ] ] note that @xmath180 s are the eigenvalues of @xmath164 , and @xmath181 correspond to @xmath182 in the nmo case with @xmath183 , but to @xmath184 in the imo case with @xmath185 . in the latter case
, though @xmath186 is the largest eigenvalue , @xmath187 becomes negative and thus @xmath188 is the smallest one .
in addition , it is easy to verify that @xmath189 holds for either neutrino mass ordering . according to eqs .
( [ eq : ch ] ) and ( [ eq : si ] ) , it is straightforward to compute the evolution matrix @xmath190 with @xmath191 and thus the oscillation probabilities @xmath192 + 2 { \rm i } \xi_3^ { \alpha\beta } \frac{\sin\big[f_*^{}(\lambda_{2}^ { } - \lambda_{1}^{})\big]}{\lambda_{2}^ { } - \lambda_{1}^ { } } \right|^2 \ ; , \label{eq : pab_s } \ ] ] where the flavor - dependent coefficients @xmath193 ( for @xmath132 ) with @xmath121 and @xmath194 running over @xmath195 , @xmath196 and @xmath197 can readily be identified from similar equations for @xmath164 to those for @xmath143 in eqs . ( [ eq : ch ] ) and ( [ eq : si ] ) . a further exploration of the right - hand side of eq .
( [ eq : pab_s ] ) gives rise to @xmath198}{(\lambda_{2}^ { } - \lambda_{1}^{})^2 } \nonumber \\ & ~ & + 2\left\{{\rm re}[\xi^{\alpha \beta}_1 \xi^{\alpha \beta * } _ 2 ] \cos\big[f_*^ { } ( 3 \lambda_3^ { } + b ) \big ] + { \rm im}[\xi^{\alpha \beta } _ 1 \xi^{\alpha \beta*}_2 ] \sin\big[f_*^ { } ( 3 \lambda_3^ { } + b ) \big ] \right\}\cos \big[f_*^{}(\lambda_{2}^ { } - \lambda_{1}^{})\big ] \nonumber \\ & ~ & + 4\left\ { { \rm im}[\xi^{\alpha \beta}_1 \xi^{\alpha \beta * } _ 3 ] \cos\big[f_*^ { } ( 3 \lambda_3^ { } + b ) \big ] -{\rm re}[\xi^{\alpha \beta } _ 1 \xi^{\alpha \beta*}_3 ] \sin\big[f_*^ { } ( 3 \lambda_3^ { } + b ) \big]\right\ } \frac { \sin\big[f_*^{}(\lambda_{2}^ { } - \lambda_{1}^{})\big]}{\lambda_{2}^ { } - \lambda_{1}^ { } } \nonumber \\ & ~ & + 4 { \rm im}[\xi^{\alpha \beta } _ 2 \xi^{\alpha \beta*}_3 ] \cos\big[f_*^ { } ( \lambda_{2}^ { } - \lambda_{1}^{})\big ] \frac { \sin\big[f_*^ { } ( \lambda_{2}^ { } - \lambda_{1}^{})\big]}{\lambda_{2}^ { } - \lambda_{1}^{}}\ ; , \label{eq : pab_full } \ ] ] where the identity @xmath199 has been implemented .
although we will not show the exact expressions of @xmath193 s , some useful properties of them can be implemented to further simplify the oscillation probabilities and the series expansions of @xmath193 s with respect to the small parameter @xmath121 have been collected in appendix a. in the appearance channel @xmath200 with @xmath201 , the identity @xmath202 holds exactly .
therefore , it is easy to verify that @xmath203 = 0 $ ] , @xmath204 = -|\xi^{\alpha \beta}_1|^2 = - |\xi^{\alpha \beta}_2|^2 $ ] and @xmath205 = - { \rm im}[\xi^{\alpha \beta * } _ 1 \xi^{\alpha \beta}_3]$ ] . under these conditions , eq .
( [ eq : pab_full ] ) will be reduced to @xmath206}{(\lambda_{2}^ { } - \lambda_{1}^{})^2 } \nonumber \\ & ~ & + 2|\xi^{\alpha \beta}_1|^2 \left\ { 1 - \cos\big[f_*^ { } ( 3 \lambda_3^ { } + b ) \big ] \cos \big[f_*^{}(\lambda_{2}^ { } - \lambda_{1}^{})\big ] \right\}\nonumber \\ & ~ & -4{\rm re}[\xi^{\alpha \beta } _ 1 \xi^{\alpha \beta*}_3 ] \sin\big[f_*^ { } ( 3 \lambda_3^ { } + b ) \big ] \frac { \sin\big[f_*^{}(\lambda_{2}^ { } - \lambda_{1}^{})\big]}{\lambda_{2}^ { } - \lambda_{1}^ { } } \nonumber \\ & ~ & + 4 { \rm im}[\xi^{\alpha \beta } _ 1 \xi^{\alpha \beta*}_3 ] \left\ { \cos\big[f_*^ { } ( 3 \lambda_3^ { } + b ) \big ] - \cos\big[f_*^ { } ( \lambda_{2}^ { } - \lambda_{1}^{})\big ] \right\ } \frac { \sin\big[f_*^ { } ( \lambda_{2}^ { } - \lambda_{1}^{})\big]}{\lambda_{2}^ { } - \lambda_{1}^{}}\ ; , \label{eq : pab_app } \ ] ] where one can observe four different types of oscillation terms . in the disappearance channel @xmath207 , we have @xmath208 and @xmath209 , and thus arrive at @xmath210}{(\lambda_{2}^ { } - \lambda_{1}^{})^2 } \nonumber \\ & ~ & - 2\xi^{\alpha \beta}_1 \xi^{\alpha \beta}_2 \left\ { 1 - \cos\big[f_*^ { } ( 3 \lambda_3^ { } + b ) \big ] \cos \big[f_*^{}(\lambda_{2}^ { } - \lambda_{1}^{})\big ] \right\}\nonumber \\ & ~ & -4\xi^{\alpha \beta}_1 \xi^{\alpha \beta}_3 \sin\big[f_*^ { } ( 3 \lambda_3^ { } + b ) \big ] \frac { \sin\big[f_*^{}(\lambda_{2}^ { } - \lambda_{1}^{})\big]}{\lambda_{2}^ { } - \lambda_{1}^ { } } \ ; , \label{eq : pab_dis } \ ] ] in which only three oscillation terms survive . in order for the oscillation probabilities to be valid for arbitrary neutrino energies and baseline lengths , as we shall see later
, it is important to always keep those oscillation terms not expanded at all .
so far , all the analytical results in the previous section are exact . in this section
, we will first expand the eigenvalues @xmath180 s in terms of @xmath121 and derive the approximate formulas of neutrino oscillation probabilities in the general @xmath0-gauge .
simple and compact formulas in the special case of @xmath6 then emerge in an obvious way . as a by - product
, the mapping between effective and fundamental mixing parameters is also obtained .
finally , numerical verifications are carried out to show high precisions of our analytical formulas , in comparison with the exact ones .
( @xmath173 ) of the matrix @xmath164 in eq . (
[ eq : m_f ] ) shown as functions of the neutrino energy @xmath111 , where the matter density @xmath211 , the electron fraction @xmath116 and @xmath212 have been taken for illustration .
the best - fit values of neutrino oscillation parameters from table 1 have been adopted .
the left panel is for the case of nmo while the right panel for imo .
note that @xmath213 and @xmath214 are negative in the imo case , so their absolute values have been plotted together with @xmath186 in the right panel . ]
let us begin with the series expansion of three eigenvalues .
first , to clearly see the relative sizes of @xmath180 s , we have shown their exact values as functions of the neutrino energy @xmath111 in fig .
[ fig : lambda ] , where the matter density @xmath215 , the electron fraction @xmath116 and @xmath212 have been taken for illustration .
in addition , the best - fit values of neutrino oscillation parameters from table 1 are adopted . in the left panel ,
the results for the nmo are given , where one can observe a potential level crossing at @xmath216 for the solar resonance , and another one around @xmath217 for the atmospheric resonance , if a poor approximation to @xmath218 is adopted .
in the right panel , since both @xmath213 and @xmath214 in the imo case are actually negative , their absolute values are shown together with @xmath186 .
it is obvious that there is no level crossing in this case between @xmath186 and @xmath219 , but the level crossing at @xmath216 for the solar resonance still exists . for antineutrino oscillations in matter ,
as is well known , the atmospheric resonance will be present in the imo case , while absent in the nmo case .
a correct treatment of these eigenvalues in the regions of resonances is crucial to get well - behaved analytical results . in this work
we shall use @xmath121 as the only expansion parameter and deal carefully with the would - be divergences in the neighborhood of resonances and in the limiting cases ( e.g. , the vacuum oscillations with @xmath122 ) .
looking at the analytical results of @xmath180 s in eqs .
( [ eq : lambda_nmo ] ) and ( [ eq : lambda_imo ] ) , one should first expand @xmath220 and @xmath221 in terms of @xmath121 , and then insert their approximate expressions back into eqs .
( [ eq : lambda_nmo ] ) and ( [ eq : lambda_imo ] ) . after a straightforward but tedious calculation
, we finally get @xmath222 \nonumber\\ & & + \frac{\alpha ( 1 + \widehat a + 3 \widehat c)}{8 ( \widehat c')^3 } \big [ ( 2 \eta -1 ) ( 1 + \widehat a ) -3 \widehat a c^{2}_{\theta^ { } _ { 13 } } ( \eta - c^{2}_{\theta^{}_{12 } } ) \big ] \nonumber\\ & & - \frac{3 \alpha^2 \widehat a s^{2}_{2 \theta^{}_{12 } } c^{2}_{\theta^{}_{13 } } ( 1 - \widehat a - \widehat c ) } { 4 \widehat c \widehat c ' ( 1 + \widehat a + \widehat c)^2 } + \frac{3 \alpha^2 \widehat a^2 s^{2}_{2\theta^{}_{13 } } ( \eta - c^{2}_{\theta^{}_{12 } } ) ^2 } { 8 \widehat c^3 ( \widehat c ' ) } \nonumber\\ & & - \frac{\alpha^2 } { 8 \widehat c(\widehat c')^3 } \big [ 2 \widehat c ( 1 - 2 \eta ) - 3 ( \eta -c^{2}_{\theta^{}_{12 } } ) ( 1 - \widehat a c^{}_{2 \theta^{}_{13 } } - \widehat c ) \big ] \nonumber\\ & & \times \big[(1- 2 \eta ) ( 1 + \widehat a ) + 3 \widehat a c^{2}_{\theta^{}_{13 } } ( \eta - c^{2}_{\theta^{}_{12 } } ) \big ] -\frac{\alpha^2 ( 1 + \widehat a + 3 \widehat c)}{32 ( \widehat c')^5 } \nonumber\\ & & \times \bigg\ { 4 \widehat ( \widehat c')^2 ( 1 - \eta + \eta^2 ) -3 \big [ ( 1 - 2 \eta ) ( 1 + \widehat a ) + 3 \widehat a c^{2}_{\theta^{}_{13 } } ( \eta - c^{2}_{\theta^{}_{12 } } ) \big]^2 \bigg\ } \ ; , \label{eq : z } \ ] ] and @xmath223 \nonumber\\ & & + \frac{3 \alpha^2}{8 ( \widehat c')^3 } \big [ \widehat c^2 + \widehat a c^{2}_{\theta^{}_{13 } } \big[s^{2}_{2 \theta^{}_{12 } } - 2 c^{}_{2 \theta_{12}^ { } } ( \eta - c^{2}_{\theta^{}_{12 } } ) ( 1 - \widehat a ) \nonumber\\ & & + \widehat a ( 4 - 3 c^{2}_{\theta^{}_{13 } } ) ( \eta - c^{2}_{\theta^{}_{12}})^2 \big]\big ] \bigg\ } \ ; , \label{eq : xy } \ ] ] where @xmath224 and @xmath225 have been introduced also for @xmath226 . in addition , we have defined a regulator for the atmospheric resonance @xcite @xmath227 and @xmath228 . note that @xmath229 appears in the denominators and will cause divergences in the further expansions in terms of @xmath230 when @xmath231 .
therefore , we shall keep the exact form of @xmath229 in eq .
( [ eq : c_hat ] ) in our calculations of the oscillation probabilities . on the other hand , in the low - energy or vacuum limit with @xmath122
, we have learned from refs .
@xcite and @xcite that one can not expand the function @xmath124 mentioned in section 1 in terms of @xmath121 .
a further study shows that this function arises from the difference between two eigenvalues @xmath214 and @xmath213 , namely , the terms proportional to @xmath232 in eqs .
( [ eq : lambda_nmo ] ) and ( [ eq : lambda_imo ] ) .
therefore , we define @xmath233 and expand @xmath186 up to the second order of @xmath121
. then , @xmath213 and @xmath214 can be obtained from the identity @xmath199 and the definition of @xmath234 , namely , @xmath235 where the higher - order terms of @xmath236 have been omitted .
note that eq .
( [ eq : lambda_app ] ) is valid for both nmo and imo .
the corresponding coefficients @xmath237 ( for @xmath132 ) in eq . (
[ eq : lambda_app ] ) can be directly computed by making use of eqs .
( [ eq : bcd ] ) , ( [ eq : z ] ) and ( [ eq : xy ] ) .
more explicitly , we have @xmath238}{2 \widehat c } \ ; , \nonumber\\ \rho^{}_3 & = & \displaystyle ( \eta - c^{2}_{\theta^{}_{12 } } ) ^2 s^{2}_{2 \theta^{}_{13 } } \frac{\widehat a^2}{4 \widehat c^3 } -\frac{s^{2}_{2\theta^{}_{12 } } ( 1 - \widehat a - \widehat c ) ( 1 + \widehat a - \widehat c ) } { 8 \widehat c ( 1 + \widehat a + \widehat c ) } \ ; , \label{eq : rhoi } \ ] ] where one can clearly observe that the above coefficients will be greatly simplified for @xmath6 . in particular
, @xmath239 implies that the first order correction to @xmath186 is vanishing , so the leading - order results are already very precise .
additionally , at the second order , only one term is left in @xmath186 .
however , this is not the case for @xmath213 and @xmath214 , as extra contributions come from @xmath234 , which can be determined from @xmath240 \biggm\ } ^2 \nonumber\\ & & -2 \alpha ( 1 + \widehat a -\widehat c ) ( \eta + c^{2}_{\theta^{}_{12 } } -1 ) + \frac{2 \alpha^2 \widehat a^2 ( 1 - \widehat a -\widehat c ) } { \widehat c ( 1 + \widehat a + \widehat c)^3 } c^{4}_{\theta^{}_{13 } } s^{2}_{2 \theta^{}_{12 } } \nonumber\\ & & - \frac{8 \alpha^2(1 + \widehat a s^{2}_{\theta^{}_{13 } } ) ( \eta -1 ) \eta } { 1 + \widehat a + \widehat c}- \frac{4 \alpha^2 \widehat a^3 ( \eta -c^{2}_{\theta^{}_{12}})^2 } { \widehat c^3 ( 1 + \widehat a + \widehat c ) } c^{4 } _ { \theta^{}_{13 } } s^{2}_{\theta^{}_{13 } } \nonumber\\ & & - \frac{8 \alpha^2 \widehat a ( \eta - c^{2}_{\theta^{}_{12}})}{\widehat c ( 1 + \widehat a + \widehat c)^2 } ( \eta + c^{2}_{\theta^{}_{12 } } -1 ) ( 1 - \widehat a c^{}_{2 \theta^{}_{13 } } - \widehat c)c^{2}_{\theta^{}_{13 } } \ ; .
\label{eq : epsilon } \ ] ] in the derivation of eq .
( [ eq : epsilon ] ) , the identities @xmath199 and @xmath241 have been used , where both @xmath242 and @xmath243 have been given in eq .
( [ eq : bcd ] ) .
note that , instead of @xmath234 itself , @xmath244 has been expanded in @xmath121 in eq .
( [ eq : epsilon ] ) where high - order terms of @xmath236 have been neglected . as we will show in the next subsection
, @xmath234 reduces to @xmath124 in the case of @xmath6 and in the limit of @xmath122 .
therefore , @xmath234 is the parameter that we should retain in the series expansion .
having obtained @xmath180 s , we can calculate @xmath245 s according to eqs .
( [ eq : ch ] ) and ( [ eq : si ] ) .
their analytical expressions have been collected in appendix a. after inserting @xmath245 s and @xmath180 s into eqs .
( [ eq : pab_app ] ) and ( [ eq : pab_dis ] ) , we finally obtain the approximate formulas of oscillation probabilities @xmath246 ( 1 - \cos \widetilde
f_+^ { } \cos\widetilde f_-^ { } ) \nonumber\\ & & + \bigr [ \frac{1 + \widehat a - \widehat c - 2 \alpha c^{}_{2\theta^{}_{12 } } } { 4 \epsilon\widehat c^2 } s^{2}_{2\theta^{}_{13 } } - \frac{2 \alpha \widehat a ( \eta - c^{2}_{\theta_{12}^ { } } ) } { \epsilon \widehat c^4 ( 1 + \widehat a + \widehat c ) } \nonumber\\ & & \times(1 - 6 \widehat a c^{}_{2\theta_{13}^ { } } - \widehat c + \widehat a \widehat c + 5 \widehat a^2)c^{4}_{\theta_{13}^ { } } s^{2}_{\theta_{13}^{}}\bigr ] \sin\widetilde f_+^ { } \sin\widetilde f_-^ { } \nonumber\\ & & -\frac{4 \alpha^2 ( 1 - \widehat a + \widehat c ) } { \epsilon^2\widehat c ( 1 + \widehat a + \widehat c ) ^3 } s^{2}_{2\theta_{12 } ^ { } } c^{4}_{\theta_{13}^ { } } \sin^2 \widetilde f_-^ { } \ ; , \label{eq : pee } \ ] ] @xmath247 \nonumber\\ & & \times ( 1 - \cos \widetilde f_+^ { } \cos \widetilde f_-^ { } ) -\frac{2}{\epsilon } \bigr[\frac { ( 1 + \widehat a - \widehat c - 2 \alpha c^{}_{2\theta^ { } _ { 12 } } ) } { 8 \widehat c^2 } s^2_{2 \theta_{13}^ { } } s^2_{\theta_{23}^ { } } - \frac{\alpha ( 1 - \widehat a + \widehat c)}{\widehat c^2 } { \cal j } \cot \delta \nonumber \\ & & -\frac{\alpha \widehat a ( \eta - c^2_{\theta_{12}^{}})}{\widehat c^4 ( 1 + \widehat a + \widehat c ) } ( 1 - 6 \widehat a c^{}_{2\theta_{13}^ { } } - \widehat c + \widehat a \widehat c + 5 \widehat a^2 ) c^4_{\theta_{13}^ { } } s^2_{\theta_{13}^ { } } s^2_{\theta_{23}^ { } } \bigr ] \nonumber\\ & & \times\sin\widetilde f_+^ { } \sin\widetilde f_-^ { } + \biggm\{\frac{\alpha^2 ( 1 - \widehat a + \widehat c ) } { \widehat c ( 1 + \widehat a + \widehat c ) } s^{2}_{2\theta_{12}^ { } } c^{2}_{\theta_{13}^ { } } c^{2}_{\theta_{23}^ { } } + \frac{16 \alpha { \cal j \cot \delta}}{\widehat c ( 1 + \widehat a + \widehat c ) ^2}\bigr [ \alpha c^ { } _ { 2\theta_{12}^ { } } \nonumber \\ & & \times ( \widehat c + \widehat a c^2 _ { \theta_{13}^ { } } ) - \widehat a c^2 _ { \theta_{13}^{}}\bigr ] - \frac{\alpha^2 ( 1 + \widehat a)}{\widehat c(1 + \widehat a + \widehat c)^2 } s^{2}_{2\theta_{12}^ { } } s^{2}_{2 \theta_{13}^ { } } s^{2}_{\theta_{23}^ { } } \nonumber\\ & & + \frac{16 \alpha^2 \widehat a ( \eta -c^2_{\theta_{12}^ { } } ) } { \widehat c^3 ( 1 + \widehat a + \widehat c)^2 } ( 1 - 3 \widehat a c^ { } _ { 2\theta_{13}^ { } } - \widehat
c + \widehat a \widehat c + 2 \widehat a^2 ) c^2_{\theta_{13}^ { } } { \cal j } \cot\delta\biggm\}\frac{\sin^2 \widetilde f_-^{}}{\epsilon^2}\nonumber\\ & & + \frac{8 \alpha { \cal j}}{\epsilon \widehat c ( 1 + \widehat a + \widehat c ) } ( \cos \widetilde f^{}_{+ } - \cos \widetilde f^{}_{- } ) \sin \widetilde f^{}_{- } \ ; , \label{eq : pme } \ ] ] where @xmath248 \;. \label{eq : fpm } \ ] ] the expansion of @xmath249 is given to @xmath250 , but a few terms proportional to @xmath251 are kept in the coefficients in front of oscillation terms , as they may become important in some cases . for completeness , we also present the complete expression for @xmath252 in eq .
( 56 ) in appendix b. as is proved in ref .
@xcite , only two oscillation probabilities are independent , say , @xmath253 and @xmath254 .
the other probabilities can be constructed by making use of unitarity condition @xmath255 and the time - reversal transformation @xmath256 for a constant matter density . furthermore , considering that the rotation matrix in the @xmath257-@xmath258 sector commutes with the matter potential term in the effective hamiltonian , we can establish the relations @xmath259 and @xmath260 .
for this reason , only two independent appearance probabilities @xmath253 and @xmath254 are shown in this work , while @xmath261 is given as an example in the disappearance channel .
note that the oscillation probabilities are for nmo , and we can get the corresponding results by replacing @xmath234 with @xmath262 for imo . given the above approximate formulas for oscillation probabilities , we also verify that these expansions indeed reduce to those that already exist in the literature .
for example , in the low energy range , @xmath121 , @xmath112 and @xmath234 are of the same order and can be expanded simultaneously , from which one can arrive at eq .
( 4.6 ) in ref .
@xcite . on the other hand , for the high energy region with @xmath263
, one can safely expand @xmath234 in terms of @xmath121 and restore the familiar results of freund @xcite and akhmedov _ et al . _
@xcite .
as we have mentioned , the analytical expressions can be substantially simplified when @xmath264 is adopted , where all the terms proportional to @xmath265 automatically disappear .
the resultant simplified formulas will be presented and discussed in the next subsection .
here we show that the choice of @xmath264 is not only advantageous for the analytical simplicity , but also for numerical accuracy . for different choices of @xmath0 .
the best - fit values of neutrino oscillation parameters in table 1 have been adopted and a baseline of @xmath266 is employed . the left panel is for nmo and the right panel is for imo . ] to examine the influence of @xmath0 on the accuracy of analytical approximations in the oscillation probabilities , we have adopted the best - fit values of neutrino oscillation parameters in table 1 .
in addition , the matter density @xmath267 for the earth s crust and @xmath116 are taken for illustration . in order to test the numerical accuracy
, we define the absolute error of the analytical approximations of @xmath268 as @xmath269 for @xmath270 , i.e. , @xmath271 where @xmath272 is calculated by a fully numerical evolution of the neutrino flavor states .
note that an unusual baseline of @xmath273 km is employed in order to make the fine structure of oscillations more prominent .
the oscillation probabilities and their absolute errors are given in fig .
[ fig : error ] , where we can observe that the case of @xmath264 is the most accurate one for almost the entire range of neutrino energies . in fig .
[ fig : error ] do not mean the best precision but the intersection points of exact and approximate oscillation probabilities , which are caused by the modifications of oscillation frequency and amplitude in the approximate formulas . ] comparing with previous analytical approximations of the oscillation probabilities , our results are advantageous in several aspects .
first , we have included all the possible leading terms of the whole energy region .
taking the expansion terms @xmath274 and @xmath121 for instance , although @xmath274 is a higher - order term than @xmath121 near the atmospheric resonance , it is significantly enhanced in the low energy range where @xmath234 becomes small .
thus both are maintained in the expansion .
second , our analytical results keep @xmath234 and @xmath229 as independent parameters in order to avoid any divergence in the low - energy limit and near the atmospheric resonance , respectively .
third , for the first time , we have presented the analytical results with a generic @xmath0 value , which is convenient to make a comparison with the previous results .
we further show that @xmath264 is the best choice in terms of both simplicity and numerical accuracy .
if @xmath6 is fixed , we can obtain much simpler formulas for relevant oscillation parameters in matter and those for the oscillation probabilities as well .
first , let us focus on the two regulators for eliminating possible divergences . as indicated in eq .
( [ eq : c_hat ] ) , @xmath128 depends on @xmath0 implicitly through @xmath112 and @xmath151 , so its expression is not modified .
the other one is @xmath275 where eq .
( [ eq : epsilon ] ) with @xmath6 has been used . in the low - energy limit
, @xmath112 will also be a small parameter , just like @xmath121 . in this case
, it is easy to verify @xmath276 where higher - order terms of @xmath277 are omitted .
it has been found in ref .
@xcite that one can keep @xmath234 in the oscillation probabilities , whose low - energy behaviors will then be remarkably improved . in the high - energy limit , it is safe to expand @xmath234 in terms of @xmath121 , and thus we get @xmath278 note that @xmath234 in this case is not a small parameter , as the matter effects become important or even dominant , e.g. , @xmath279 . for an arbitrary neutrino energy ,
it is necessary to make use of the full result of @xmath234 in eq .
( [ eq : epsilon_new ] ) .
second , as it is useful to define the oscillation phases @xmath280 and @xmath281 in vacuum , we obtain the following analytical approximations of their counterparts in matter with the help of eq .
( [ eq : fpm ] ) , namely , @xmath282 reflecting the corrections induced by the earth matter to neutrino mass - squared differences .
given the above parameters , the oscillation probabilities for the special case of @xmath283 turn out to be @xmath284 and @xmath285 ( 1 - \cos\widetilde f_{+}^ { } \cos\widetilde f_{-}^ { } ) \nonumber \\ & & - \bigr[\frac { s^{2}_{2 \theta_{13}^ { } } s^{2}_{\theta_{23}^ { } } } { 4 \epsilon \widehat c^2 } ( 1 + \widehat a - \widehat c -2 \alpha c^{}_{2\theta_{12}^ { } } ) - \frac{2 \alpha ( 1 - \widehat a + \widehat c ) } { \epsilon \widehat c^2 } { \cal j } \cot\delta \bigr ] \sin\widetilde f_{+}^ { } \sin\widetilde f_{-}^ { } \nonumber \\ & & + \biggm\ { \frac{\alpha^2 ( 1 - \widehat a + \widehat c ) } { \widehat c ( 1 + \widehat a + \widehat c ) } s^{2}_{2\theta_{12}^ { } } c^{2}_{\theta_{13}^ { } } c^{2}_{\theta_{23}^ { } } - \frac{\alpha^2 ( 1 + \widehat a)}{\widehat c(1 + \widehat a + \widehat c)^2 } s^{2}_{2\theta_{12}^ { } } s^{2}_{2 \theta_{13}^ { } } s^{2}_{\theta_{23}^ { } } \nonumber \\ \hspace{-0.15 cm } & & \hspace{-0.15 cm } \displaystyle + \frac{16 \alpha { \cal j \cot \delta}}{\widehat c ( 1 + \widehat a + \widehat c ) ^2 } \bigr [ \alpha c^ { } _ { 2\theta_{12 } ^ { } } ( \widehat c + \widehat a c^2 _ { \theta_{13}^ { } } ) - \widehat a c^2 _ { \theta_{13}^ { } } \bigr ] \biggm\ } \frac{\sin^2 \widetilde f_{-}^ { } } { \epsilon^2 } \nonumber \\ & & - \frac{8 \alpha { \cal j } } { \epsilon \widehat c ( 1 + \widehat a + \widehat c ) } ( \cos \widetilde f^{}_{- } - \cos \widetilde f^{}_{+ } ) \sin \widetilde f^{}_{- } \ ; , \label{eq : pme_new } \ ] ] which are much simpler than the general formulas in eqs . ( [ eq : pee ] ) and ( [ eq : pme ] ) .
see also the results of @xmath254 in eq .
( 57 ) in appendix b. to carry out a systematic test of numerical accuracy of analytical approximations , we consider the absolute errors @xmath286 defined in eq .
( [ eq : deltap ] ) and the approximate results are now obtained by using the simplified formulas in the case of @xmath6 .
the numerical results of @xmath286 for a wide range of neutrino energies and baseline lengths have been shown in fig .
[ fig:2d ] , where the sizes of absolute errors are denoted by different colors .
some comments on the numerical calculations are in order : * in fig .
[ fig:2d ] , the matter density of @xmath211 with @xmath116 and the best - fit values of neutrino oscillation parameters from table 1 have been used in numerical calculations .
in addition , to avoid fast oscillations at low energies , we have averaged the oscillation probabilities over a gaussian energy resolution of @xmath72 .
the baseline lengths and neutrino energies have been set to be @xmath287 and @xmath288 , respectively .
hence , both current and future oscillation experiments as mentioned in the introduction are essentially covered . as for the atmospheric neutrinos ,
our assumption of a constant matter density renders it impossible to reveal the structure of parametric resonances @xcite .
however , it suffices to illustrate the numerical difference between our analytical formulas and the exact oscillation probabilities .
* in the lower part of each plot in fig .
[ fig:2d ] , i.e. , for @xmath289 , one can observe that the errors are always far below the level of @xmath290 .
this can be understood by noticing that the oscillations driven by @xmath71 have not yet developed for a short baseline .
the @xmath32-driven oscillations indeed take place for short baseline lengths and low neutrino energies , however , the amplitudes will be suppressed by the smallest mixing angle @xmath41 .
for higher neutrino energies , we need longer baseline lengths for the @xmath32-driven oscillations to develop .
the errors in the entire range of baseline lengths and energies are below @xmath291 , demonstrating an excellent agreement between our approximate formulas and the exact ones . * for imo on the right column of fig .
[ fig:2d ] , one can observe that the discrepancy is at most @xmath292 , as a consequence of the absence of resonances in this case .
for nmo , the region of largest errors always appears around @xmath293 and @xmath294 , where the atmospheric resonance is encountered , while around the region of the solar resonance relatively smaller errors are observed .
such a difference on the size of error at two different resonances may be attributed to the fact that @xmath295 and @xmath296 are more close to each other at the atmospheric resonance than @xmath297 and @xmath295 at the solar resonance .
notice that the approximate formulas of @xmath261 and @xmath253 in eqs .
( [ eq : pee_new ] ) and ( [ eq : pme_new ] ) , together with that of @xmath254 in eq .
( 57 ) in appendix b , are the main results of this work .
given their simplicity and high level of numerical accuracy , one may directly employ them to perform both analytical and numerical studies on neutrino oscillation phenomena in current and upcoming oscillation experiments .
we leave such applications for a future work . , where the matter density of @xmath211 with the electron fraction @xmath116 and the best - fit values of neutrino oscillation parameters in table 1 have been used . the absolute errors @xmath286 ( for @xmath298 )
have been defined in eq .
( [ eq : deltap ] ) , and the probabilities are averaged over a gaussian energy resolution of @xmath72 . ] with those newly obtained approximate formulas for oscillation probabilities , a more accurate mapping of the intrinsic mixing parameters to the effective mixing parameters in matter can actually be established as a by - product . to see this clearly
, we first re - express the exact formulas of neutrino oscillation probabilities in eq .
( [ eq : pab_eff ] ) in terms of effective parameters in matter .
starting with the disappearance channel @xmath299 , we have @xmath300 \nonumber \\ & & -2 c^{}_{2 \widetilde \theta^{}_{12 } } c^2_{\widetilde \theta^ { } _ { 13 } } s^2_{\widetilde \theta^{}_{13 } } \sin\frac { ( \widetilde \delta^{}_{31 } + \widetilde \delta^{}_{32 } ) l}{4 e } \sin(\frac{\widetilde\delta^{}_{21 } l } { 4 e } ) \nonumber\\ & & - 4 c^2_{\widetilde \theta^{}_{12 } } s^2_{\widetilde \theta^{}_{12 } } c^4_{\widetilde \theta^{}_{13 } } \sin^2 ( \frac{\widetilde \delta^{}_{21 } l}{4 e } ) \ ; , \label{eq : pee_eff } \ ] ] where @xmath301 and @xmath302 have been defined as before . comparing between @xmath261 in eq .
( [ eq : pee_eff ] ) and @xmath303 with @xmath304 in eq .
( [ eq : pab_dis ] ) , one can immediately realize @xmath305 by identifying the oscillation terms of the same kind . in the derivation of eq .
( [ eq : map_ee ] ) , we have implemented the following relations @xmath306 which are verified by using @xmath307/\delta^{}_*$ ] and @xmath97 .
thus far , the results are exact and no approximations have been made . to get more useful results for effective mixing angles @xmath308 and @xmath309
, we have to expand @xmath310 ( for @xmath132 ) with respect to @xmath121 but keep @xmath234 unchanged as before .
since @xmath261 is independent of @xmath311 and the cp - violating phase @xmath312 , one should further consider the oscillations in the appearance channels .
as an example , we study the oscillation probability in the appearance channel @xmath83 , namely , @xmath313 \nonumber\\ & & + 2 ( c^{}_{2 \widetilde \theta^{}_{12 } } c^2_{\widetilde \theta^ { } _ { 13 } } s^2_{\widetilde \theta^{}_{13 } } s^2_{\widetilde \theta^ { } _ { 23 } } + 2 \widetilde{\cal j } \cot \widetilde \delta ) \sin\frac { ( \widetilde \delta^{}_{31 } + \widetilde \delta^{}_{32 } ) l}{4 e } \sin ( \frac{\widetilde\delta^{}_{21 } l}{4 e } ) \nonumber\\ & & + 4 \big[c^2_{\widetilde \theta^{}_{12 } } s^2_{\widetilde \theta^ { } _ { 12 } } c^2_{\widetilde \theta^{}_{13 } } ( c^2_{\widetilde \theta^ { } _ { 23 } } - s^2_{\widetilde \theta^{}_{13 } } s^2_{\widetilde \theta^ { } _ { 23 } } ) + c^{}_{2 \widetilde \theta^{}_{12 } } \widetilde{\cal j } \cot \widetilde\delta \big]\sin^2 ( \frac{\widetilde \delta^{}_{21 } l}{4 e } ) \nonumber\\ & & + 4 \widetilde{\cal j } \big[\cos\frac { ( \widetilde\delta^{}_{31 } + \widetilde\delta^{}_{32 } ) l}{4 e } - \cos(\frac{\widetilde\delta^{}_{21 } l}{4 e } ) \big ] \sin^ { } ( \frac{\widetilde \delta^{}_{21 } l}{4 e } ) \ ; , \label{eq : pme_eff } \ ] ] from which additional relations between @xmath314 and the parameters @xmath315 , similar to those in eq .
( [ eq : map_ee ] ) can be found . using the series expansions of @xmath193 listed in appendix a and setting @xmath6 , we finally arrive at the mapping for three mixing angles @xmath316 and that for the jarlskog invariant @xmath317\ ; .
\label{eq : map_j } \ ] ] note that the leptonic cp violation is now described by the jarlskog invariant , and the direct relation between @xmath95 and the vacuum mixing parameters can be easily obtained using eqs .
( [ eq : map_mix ] ) and ( [ eq : map_j ] ) . in appendix c
we also list the mapping of the three mixing angles and the jarlskog invariant for a generic value of @xmath0 . as a cross check
, we further use the relations derived in eqs .
( [ eq : map_mix ] ) and ( [ eq : map_j ] ) to calculate the following oscillation probability of @xmath318 @xmath319 \nonumber\\ & & - 2 \big[c^{}_{2 \widetilde \theta^{}_{12 } } c^2_{\widetilde \theta^{}_{13 } } c^2_{\widetilde \theta^{}_{23 } } s^2_{\widetilde \theta^{}_{23 } } ( 1 + s^2_{\widetilde \theta^{}_{13 } } ) + 2 c^{}_{2 \widetilde \theta^{}_{23 } } \widetilde{\cal j } \cot \widetilde \delta \big]\sin \frac { ( \widetilde \delta^{}_{31 } + \widetilde \delta^{}_{32 } ) l } { 4 e } \sin ( \frac{\widetilde\delta^{}_{21 } l}{4 e } ) \nonumber\\ & & + 4 \big [ c^2_{\widetilde \theta^{}_{12 } } s^2_{\widetilde\theta^{}_{12 } } s^2_{\widetilde \theta^{}_{13 } } + s^2_{\widetilde \theta^{}_{13 } } c^2_{\widetilde \theta^{}_{23 } } s^2_{\widetilde \theta^{}_{23 } } - c^2_{\widetilde \theta^{}_{12 } } s^2_{\widetilde \theta^ { } _ { 12 } } c^2_{\widetilde \theta^{}_{23 } } s^2_{\widetilde \theta^ { } _ { 23 } } ( 1 + 4 s^2_{\widetilde \theta^{}_{13 } } + s^4_{\widetilde\theta^{}_{13 } } ) \nonumber\\ & & + c^{}_{\widetilde\theta^{}_{12 } } s^{}_{\widetilde \theta^ { } _ { 12 } } s^{}_{\widetilde\theta^{}_{13 } } c^{}_{\widetilde \theta^ { } _ { 23 } } s^{}_{\widetilde \theta^{}_{23 } } c^{}_{2 \widetilde \theta^{}_{12 } } c^{}_{2 \widetilde \theta^ { } _ { 23 } } ( 1 + s^{2}_{\widetilde \theta^{}_{13 } } ) \cos\widetilde \delta - 2 c^{2}_{\widetilde \theta^{}_{12 } } s^{2}_{\widetilde\theta^{}_{12 } } s^{2}_{\widetilde \theta^{}_{13 } } c^{2}_{\widetilde \theta^{}_{23 } } s^{2}_{\widetilde \theta^{}_{23 } } \cos 2 \widetilde \delta \big]\sin^2 ( \frac{\widetilde\delta^{}_{21 } l}{4 e } ) \nonumber\\ & & + 4 \widetilde{\cal j } \big[\cos\frac { ( \widetilde\delta^{}_{31 } + \widetilde\delta^{}_{32 } ) l}{4 e } - \cos(\frac{\widetilde\delta^{}_{21 } l}{4 e } ) \big ] \sin^ { } ( \frac{\widetilde \delta^{}_{21 } l}{4 e } ) \ ; . \label{eq : ptm_eff } \ ] ] it turns out that the expression in eq . ( 57 ) can be exactly reproduced , when both of them are matched to the same order of @xmath121 .
it is worth mentioning that the mapping relations for mixing angles and the jarlskog invariant have been truncated at the second order of @xmath121 and serve as excellent approximations to the exact results . for illustration
, we have calculated the effective mixing angles @xmath320 and the effective jarlskog invariant @xmath102 for different neutrino energies . as depicted in fig .
[ fig : map ] , the exact results are denoted as solid curves ( red ) , while the approximate results based on eqs .
( [ eq : map_mix ] ) and ( [ eq : map_j ] ) are represented by dashed curves ( blue ) .
one can see that our approximate results are in perfect agreement with the exact ones , and the differences between them are invisible from the plots . for comparison , the numerical results according to the mapping relations found by freund in ref .
@xcite are given as dotted curves ( green ) .
significant deviations can be observed in the figures for @xmath321 and @xmath322 , which can be explained by the divergence encountered in the low - energy region . and
the effective jarlskog invariant @xmath102 shown as functions of neutrino energies , where @xmath6 , a constant matter density of @xmath211 with the electron fraction @xmath116 , and the best - fit values of neutrino oscillation parameters in table 1 have been used . ] and @xmath323 , where @xmath6 is fixed and the other input parameters are the same as in fig . [
fig : error ] . ]
moreover , given the approximate expressions of effective mixing parameters in eqs .
( [ eq : map_mix ] ) and ( [ eq : map_j ] ) , one can insert them back into eqs .
( [ eq : pee_eff ] ) , ( [ eq : pme_eff ] ) and ( [ eq : ptm_eff ] ) and obtain a new set of oscillation probabilities , which we call @xmath324 , @xmath325 and @xmath326 , respectively . as the effective mixing parameters are expanded up to @xmath8
, these new oscillation probabilities will be more accurate in the sense that part of higher - order terms are now included . to illustrate this point
, we compute the absolute errors @xmath327 according to eq .
( [ eq : deltap ] ) and compare it with @xmath286 from fig .
[ fig : error ] in the case of @xmath6 .
the results are shown in fig .
[ fig : newp ] , where one can find @xmath327 ( blue solid curves ) are almost always one or two orders of magnitude smaller than @xmath286 ( red dashed curves ) .
in this work we have taken a deep look into analytical approximations for three - flavor neutrino oscillation probabilities in matter of a constant density and presented a new set of simple and compact formulas .
a useful definition of the @xmath0-gauge neutrino mass - squared difference @xmath1 is introduced and the calculations are performed in the series expansions of @xmath121 ( i.e. , @xmath328 ) .
the approximate oscillation probabilities are valid for arbitrary neutrino energies and any baseline length . among different choices of @xmath0 ,
it turns out that the case of @xmath158 is the best one in terms of both simplicity and numerical accuracy .
these formulas are particularly useful for the future long baseline accelerator neutrino experiments and the atmospheric neutrino experiments with the baseline lengths from @xmath329 to @xmath330 and a wide range of neutrino energies ( @xmath89 ) .
the main features of our results can be summarized as follows . *
our calculations are based on the cayley - hamilton theorem , where only the effective hamiltonian and its three eigenvalues are needed in order to derive the oscillation probabilities .
the series expansions of @xmath121 are applied to the exact expressions of the eigenvalues in eq .
( [ eq : lambda_nmo ] ) or eq .
( [ eq : lambda_imo ] ) . however , the @xmath234 parameter in the expansions of @xmath331 and @xmath332 in eq .
( [ eq : lambda_app ] ) behaves as the function @xmath120 and can not be expanded in terms of @xmath121 in the low energy range with @xmath333 .
thus , we keep @xmath234 intact in the calculations . *
our calculations employ a generic @xmath0-gauge neutrino mass - squared difference @xmath151 and derive the @xmath0-gauge oscillation probabilities as shown in eqs .
( [ eq : pee ] ) , ( [ eq : pme ] ) and ( [ eq : ptm ] ) for @xmath334 , @xmath335 and @xmath336 respectively . given the expressions of @xmath337 ( @xmath173 ) in eq .
( [ eq : rhoi ] ) and @xmath244 in eq .
( [ eq : epsilon ] ) , the analytical results of oscillation probabilities are greatly simplified for @xmath6 , where all the terms proportional to @xmath265 automatically disappear . moreover , as demonstrated in fig .
[ fig : error ] for different values of @xmath0 , the choice of @xmath6 is the most accurate one for almost the entire range of neutrino energies . *
fixing the gauge at @xmath6 , the oscillation probabilities are presented in eqs .
( [ eq : pee_new ] ) , ( [ eq : pme_new ] ) and ( [ eq : ptm_new ] ) for @xmath334 , @xmath335 and @xmath336 respectively , constituting the main results of this work . regarding the accuracy of these analytical approximations ,
a careful study is performed in fig .
[ fig:2d ] for the neutrino energies from @xmath291 gev to @xmath338 gev and the baseline length range @xmath339 .
one can observe that in the nmo case the errors in the entire range of baseline lengths and neutrino energies are below @xmath291 , while for imo below @xmath340 .
the largest errors appear in nmo around @xmath293 and @xmath294 , where the atmospheric resonance is encountered and the small energy splitting between @xmath332 and @xmath341 slows down the convergence of the series expansions .
* as a by - product a more accurate mapping of the intrinsic mixing parameters to the effective mixing parameters in matter is established in eqs .
( [ eq : map_mix ] ) and ( [ eq : map_j ] ) for three mixing angles and the jarlskog invariant , respectively . with the effective mixing parameters
, one can obtain a new set of oscillation probabilities in eqs .
( [ eq : pee_eff ] ) , ( [ eq : pme_eff ] ) and ( [ eq : ptm_eff ] ) for @xmath324 , @xmath325 and @xmath326 , respectively . the accuracy of the effective mixing parameters is proved in fig .
[ fig : map ] for the whole energy range including the regions of the solar and atmospheric resonances . for the new set of oscillation probabilities
, one can find from fig .
[ fig : newp ] that the accuracy of @xmath342 will be one or two orders of magnitude better than @xmath303 because some higher - order terms are also properly included . *
finally , in the low energy range , @xmath121 , @xmath112 and @xmath234 are of the same order and can be expanded simultaneously , from which one can arrive at eq .
( 4.6 ) in ref .
@xcite . on the other hand , for the high energy region with @xmath263
, one can safely expand @xmath234 in terms of @xmath121 and restore the familiar results of freund @xcite and akhmedov _ et al . _
@xcite for future long - baseline accelerator and atmospheric neutrino experiments , with the goals of determining the neutrino mass ordering and measuring the leptonic cp violating phase , a set of compact and simple analytical approximations of oscillation probabilities in matter is very helpful .
these analytical oscillation probabilities should be directly connected with the fundamental oscillation parameters and be valid for arbitrary neutrino energies and any baseline length . in this sense ,
our analytical approximations in this work meet all the afore - mentioned criteria and can be readily applied to future oscillation experiments .
we leave such applications for a separate work in the near future .
in this appendix , we present expressions for the @xmath245 terms , which are coefficients in front of various oscillation terms in eq .
( [ eq : pab_s ] ) . for this purpose
, we employ the cayley - hamilton theorem to express the evolution matrix @xmath343 into a form similar to eq .
( [ eq : ch ] ) with @xmath344 replaced by @xmath345 .
correspondingly , @xmath346 in eq .
( [ eq : si ] ) are now the eigenvalues of @xmath345 , i.e. , @xmath218 , and @xmath347 read as @xmath348 . then , with the explicit form of @xmath345 given in eq .
( [ eq : m_f ] ) , we are able to obtain the expressions of various @xmath349 for various oscillation channels @xmath350 , according to the definitions of @xmath349 in eq .
( [ eq : pab_s ] ) . as shown in eq .
( [ eq : pab_full ] ) , the final oscillation probabilities @xmath350 only depend on certain combinations of @xmath349 , we therefore just show the analytical expansions for those relevant ones .
for @xmath334 , we have @xmath351 , @xmath352 and @xmath353 @xmath354 @xmath355 for @xmath335 , we have @xmath356 and @xmath357 @xmath358 @xmath359 @xmath360 \nonumber \\ & & + \frac{4 \alpha^2 \widehat a ( \eta -c^2_{\theta_{12}^ { } } ) } { \widehat c^3 ( 1 + \widehat a + \widehat c)^2 } ( 1 - 3 \widehat a c^ { } _ { 2\theta_{13}^ { } } \nonumber\\ & & - \widehat c + \widehat a \widehat c + 2 \widehat a^2 ) c^2_{\theta_{13}^ { } } { \cal j } \cot\delta \;. \label{eq : xi_me4}\end{aligned}\ ] ] for @xmath318 , we have @xmath361 and @xmath362 @xmath363 \nonumber\\ & & + \frac{\alpha
\widehat a^2 ( \eta -c^2 _ { \theta_{12}^ { } } ) } { 2 \widehat c^4 ( 1 + \widehat a + \widehat c)^2 } \bigr [ ( 1 + \widehat a ) ( 7 + 7\widehat a + 5 \widehat c ) \nonumber\\ & & -2 c^2_{\theta_{13}^ { } } ( 2 + 12\widehat a + 3 \widehat c ) \bigr ] c^4_{\theta_{13}^ { } } s^2_{\theta_{13}^ { } } s^2_{2\theta_{23}^ { } } \ ; , \label{eq : xi_tm2}\end{aligned}\ ] ] @xmath364 @xmath365 \nonumber\\ & & - \frac{\alpha^2 ( 1 + \widehat a ) } { 8 \widehat c } s^{2}_{2 \theta_{12}^ { } } s^{2}_{\theta_{13}^ { } } s^{2}_{2 \theta_{23}^ { } } \cos 2 \delta \nonumber \\ & & + \frac{\alpha^2 ( \eta -c^2_{\theta_{12}^ { } } ) } { \widehat c^3 ( 1 + \widehat a + \widehat c ) } \bigr[2 ( \widehat c - 1 ) - 5 \widehat a -2 \widehat a^2 + 3 \widehat a \widehat c \nonumber\\ & & + \widehat a^2 ( \widehat a + \widehat c)(3 + 2 \widehat a ) -2 \widehat a c^2_{\theta_{13}^ { } } ( -5 + \widehat a + 2 \widehat a^2 \nonumber\\ & & + 3 \widehat c + 2 \widehat a \widehat c ) \bigr ] c^{}_{2\theta_{23}^ { } } { \cal j } \cot\delta \nonumber\\ & & + \frac{\alpha^2 ( 1 + \widehat a)}{4 \widehat c } s^{2}_{2 \theta_{12 } ^ { } } s^{2}_{\theta_{13}^ { } } + \frac{\alpha^2 ( 1 + \widehat a + \widehat c ) } { 4 \widehat c ( 1 - \widehat a + \widehat c ) } s^{2}_{\theta_{13}^ { } } s^{2}_{2 \theta_{23}^ { } } \nonumber \\ & & - \frac14 \alpha^2 s^{2}_{2\theta_{12}^ { } } s^{2}_{2\theta_{23 } ^ { } } ( t)- \frac{\alpha^2 ( \eta -c^2_{\theta_{12}^ { } } ) } { 8 \widehat c^3 } ( 1 - \widehat a - \widehat c ) \nonumber \\ & & \times ( 1 - \widehat c - 4 \widehat a c^{}_{2\theta_{13}^ { } } + 3 \widehat a^2 - \widehat a\widehat c ) c^{}_{2\theta_{12}^ { } } c^{2}_{\theta_{13}^ { } } s^{2}_{2\theta_{23}^ { } } \nonumber\\ & & + \frac{\alpha^2 \widehat a ( \eta -c^2_{\theta_{12}^ { } } ) ^2}{4 \widehat c^5 ( 1 + \widehat a + \widehat c ) } \bigm\ { - ( 1 + \widehat a ) ^2 \bigr [ 1 + 2 \widehat a + \widehat a^2 ( 3 - \widehat a)\bigr ] \nonumber\\ & & + \widehat c + \widehat a \widehat c ( 3 + \widehat a + \widehat a^3 ) -2 \widehat a^2 c^4_{\theta_{13}^ { } } ( 13 -3 \widehat a - 3 \widehat c ) \nonumber\\ & & -\widehat a c^2_{\theta_{13}^ { } } \bigr[-9 + 7 \widehat c -24 \widehat a + \widehat a \widehat c \nonumber \\ & & + \widehat a^2 ( -11 + 4 \widehat a + 4 \widehat c ) \bigr ] \bigm\ } c^4_{\theta_{13}^ { } } s^2_{2 \theta_{23}^ { } } \ ; , \label{eq : xi_tm4}\end{aligned}\ ] ] where @xmath366 + \frac{1+\widehat a}{4 \widehat c } ( 1 + 4 s^{2}_{\theta_{13}^ { } } + s^{4}_{\theta_{13}^ { } } ) \nonumber \\ & & + \frac{c^{4}_{\theta_{13}^{}}}{\widehat c^3 ( 1 + \widehat a + \widehat c ) ^3}\bigm\{-\widehat a c^{2}_{\theta_{13}^ { } } ( 3-\widehat a ) ( 1-\widehat a)^3 ( 1 - \widehat a - \widehat c ) \nonumber\\ & & + \widehat a^2 s^{2}_{2\theta_{13}^ { } } ( 3 - 3\widehat c + 6\widehat a -3 \widehat a^2 -\widehat a \widehat c ) \nonumber\\ & & - \widehat a s^{2}_{\theta_{13}^ { } } ( 1+\widehat a ) \bigr[6 ( 1 + \widehat a - \widehat c ) + \widehat a^2 ( 2 - \widehat a ) ( 1 + \widehat a + \widehat c ) \bigr]\bigm\}\;. \label{eq : t}\end{aligned}\ ] ]
in this appendix , we show the expression for the oscillation probability @xmath336 with an arbitrary value of @xmath0 for completeness : @xmath367(1 - \cos \widetilde f_+^ { } \cos \widetilde f_-^ { } ) \nonumber\\ & & - \frac{2}{\epsilon}\biggr\{\frac { ( 1 - \widehat a + \widehat c ) } { 16 \widehat c^2 ( 1 + \widehat a + \widehat c ) } ( 1 + \widehat a - \widehat c - 2 \alpha c^{}_{2\theta_{12}^ { } } ) ( c^{}_{2\theta_{13}^ { } } - \widehat a - 3 \widehat c ) c^{2}_{\theta_{13}^ { } } s^{2}_{2\theta_{23}^ { } } \nonumber\\ & & + \alpha c^{}_{2\theta_{23}^ { } } { \cal j}\cot\delta \bigr[\frac{1 + \widehat a + \widehat c } { \widehat c^2 } - \frac{8 \widehat a ( \widehat a c^{2}_{\theta_{13}^ { } } + \widehat c + \widehat a \widehat c ) } { \widehat c^2 ( 1 + \widehat a + \widehat c ) ^2 } \bigr ] + \frac{\alpha \widehat a^2 ( \eta -c^2 _ { \theta_{12}^ { } } ) } { 2 \widehat c^4 ( 1 + \widehat a + \widehat c)^2 } \nonumber\\ & & \times \bigr [ ( 1 + \widehat a ) ( 7 + 7\widehat a + 5 \widehat c)-2 c^2_{\theta_{13 } ^ { } } ( 2 + 12\widehat a + 3 \widehat c ) \bigr ] c^4_{\theta_{13}^ { } } s^2_{\theta_{13}^ { } } s^2_{2\theta_{23}^ { } } \biggr\}\sin\widetilde f_+^ { } \sin\widetilde f_-^ { } \nonumber\\ & & + \biggr\{-\frac{\widehat a ( 1 - \widehat a - \widehat c)}{\widehat c ( 1 + \widehat a + \widehat c ) } c^{4}_{\theta_{13}^ { } } s^{2 } _ { 2\theta_{23}^ { } } + \frac{\alpha ( 1 - \widehat a - \widehat c ) } { \widehat c } c^{}_{2\theta_{12}^ { } } c^{2}_{\theta_{13}^ { } } s^{2}_{2\theta_{23}^ { } } \nonumber\\ & & + \frac{\alpha \widehat a ( \eta -c^2_{\theta_{12}^ { } } ) ( 1 - \widehat a - \widehat c)}{\widehat c^3 ( 1 + \widehat a + \widehat c ) } ( 1 - 3 \widehat a c^ { } _ { 2\theta_{13}^ { } } - \widehat c + 2 \widehat a^2 ) c^4_{\theta_{13}^ { } } s^2_{2 \theta_{23}^ { } } \nonumber\\ & & + \frac{4 \alpha ( 1 - \widehat a - \widehat c- 2\widehat a^2 - 2\widehat a \widehat c ) } { \widehat c ( 1 + \widehat a + \widehat c ) } { \cal j } c^{}_{2\theta_{23}^ { } } \cot\delta \nonumber \\ & & + 4 \alpha^2 c^{}_{2\theta_{12}^ { } } c^ { } _ { 2\theta_{23}^ { } } { \cal j } \cot\delta \bigr[\frac{4 } { c^{2}_{\theta_{13}^ { } } ( 1 - \widehat a + \widehat c ) } - \frac{2}{\widehat c ( 1 + \widehat a + \widehat c ) } .\nonumber \\ & & -\frac{(1 - \widehat a - \widehat c ) ^2 ( 1 + 2 c^{2}_{\theta_{13}^ { } } + \widehat a + 3 \widehat c ) } { 2 \widehat c s^{2}_{\theta_{13}^ { } } ( 1 + \widehat a + \widehat c ) ^2 } \bigr ] - \frac{\alpha^2 ( 1 + \widehat a ) } { 2 \widehat c } s^{2}_{2 \theta_{12}^ { } } s^{2}_{\theta_{13}^ { } } s^{2}_{2 \theta_{23}^ { } } \cos 2 \delta \nonumber \\ & & + \frac{4 \alpha^2 ( \eta -c^2_{\theta_{12}^ { } } ) } { \widehat c^3 ( 1 + \widehat a + \widehat c ) } \bigr[2 ( \widehat c - 1 ) - 5 \widehat a -2 \widehat a^2 + 3 \widehat a \widehat c + \widehat a^2 ( \widehat a + \widehat c)(3 + 2 \widehat a ) \nonumber\\ & & -2 \widehat a c^2_{\theta_{13}^ { } } ( -5 + \widehat a + 2 \widehat a^2 + 3 \widehat c + 2 \widehat a \widehat c ) \bigr ] c^{}_{2\theta_{23}^ { } } { \cal j } \cot\delta \nonumber\\ & & + \frac{\alpha^2 ( 1 + \widehat a ) } { \widehat c } s^{2}_{2 \theta_{12 } ^ { } } s^{2}_{\theta_{13}^ { } } + \frac{\alpha^2 ( 1 + \widehat a + \widehat c ) } { \widehat c ( 1 - \widehat a + \widehat c ) } s^{2}_{\theta_{13}^ { } } s^{2}_{2 \theta_{23}^ { } } - \alpha^2 s^{2}_{2\theta_{12}^ { } } s^{2}_{2\theta_{23 } ^ { } } ( t)\nonumber \\ & & - \frac{\alpha^2 ( \eta -c^2_{\theta_{12}^ { } } ) } { 2 \widehat c^3 } ( 1 - \widehat a - \widehat c ) ( 1 - \widehat c - 4 \widehat a c^{}_{2\theta_{13}^ { } } + 3 \widehat a^2 - \widehat a\widehat c ) c^{}_{2\theta_{12}^ { } } c^{2}_{\theta_{13}^ { } } s^{2}_{2\theta_{23}^ { } } \nonumber\\ & & + \frac{\alpha^2 \widehat a ( \eta -c^2_{\theta_{12}^ { } } ) ^2 } { \widehat c^5 ( 1 + \widehat a + \widehat c ) } \bigm\ { - ( 1 + \widehat a ) ^2 \bigr [ 1 + 2 \widehat a + \widehat a^2 ( 3 - \widehat a)\bigr ] + \widehat c + \widehat a \widehat c ( 3 + \widehat a + \widehat a^3 ) \nonumber\\ & & -2 \widehat a^2 c^4_{\theta_{13}^ { } } ( 13 -3 \widehat a - 3 \widehat c ) -\widehat a c^2_{\theta_{13}^ { } } \bigr[-9 + 7 \widehat c -24 \widehat a + \widehat a \widehat c + \widehat a^2 ( -11 + 4 \widehat a \nonumber \\ & & + 4 \widehat c)\bigr ] \bigm\ } c^4_{\theta_{13}^ { } } s^2_{2 \theta_{23}^ { } } \biggr\}\frac{\sin^2 \widetilde f_-^{}}{\epsilon^2 } + \frac{8 \alpha { \cal j}}{\epsilon \widehat c ( 1 + \widehat a + \widehat c ) } ( \cos \widetilde f^{}_{+ } - \cos \widetilde f^{}_{- } ) \sin \widetilde f^{}_{-}\ ; , \label{eq : ptm}\end{aligned}\ ] ] where the expressions of @xmath368 and @xmath369 are shown in eqs .
( [ eq : fpm ] ) and ( [ eq : t ] ) , respectively . taking @xmath370 in eq .
( [ eq : ptm ] ) , the form of @xmath336 reduces to @xmath371 ( 1 - \cos\widetilde f_{+}^ { } \cos\widetilde f_{-}^ { } ) \nonumber \\ & & - \biggm\ { \frac { ( 1 - \widehat a + \widehat c ) } { 8 \epsilon \widehat c^2 ( 1 + \widehat a + \widehat c ) } ( 1 + \widehat a - \widehat c -2 \alpha c^{}_{2\theta_{12}^ { } } ) ( c^{}_{2\theta_{13}^ { } } - \widehat a - 3 \widehat c ) c^{2}_{\theta_{13}^ { } } s^{2}_{2\theta_{23}^ { } } \nonumber \\ & & + 2 \alpha c^{}_{2\theta_{23}^ { } } { \cal j}\cot\delta \bigr [ \frac{1 + \widehat a + \widehat c}{\epsilon \widehat c^2 } - \frac{8 \widehat a ( \widehat a c^{2}_{\theta_{13}^ { } } + \widehat c + \widehat a \widehat c)}{\epsilon \widehat c^2 ( 1 + \widehat a + \widehat c ) ^2 } \bigr ] \biggm\ } \sin\widetilde f_{+}^ { } \sin\widetilde f_{-}^ { } \nonumber \\ & & + \biggm\{-\frac{\widehat a ( 1 - \widehat a - \widehat c)}{\widehat c ( 1 + \widehat a + \widehat c ) } c^{4}_{\theta_{13}^ { } } s^{2}_{2\theta_{23}^ { } } + \frac{\alpha ( 1 - \widehat a - \widehat c ) } { \widehat c } c^{}_{2\theta_{12}^ { } } c^{2}_{\theta_{13}^ { } } s^{2}_{2\theta_{23}^ { } } + \frac{\alpha^2 ( 1 + \widehat a)}{\widehat c } s^{2}_{2 \theta_{12 } ^ { } } s^{2}_{\theta_{13}^ { } } \nonumber \\ & & + s^{2}_{\theta_{13}^ { } } s^{2}_{2 \theta_{23}^ { } } \frac{\alpha^2 ( 1 + \widehat a + \widehat c ) } { \widehat c ( 1 - \widehat a + \widehat c ) } - \alpha^2 s^{2}_{2\theta_{12}^ { } } s^{2}_{2\theta_{23}^{}}t + \frac{4 \alpha ( 1 - \widehat a - \widehat c- 2\widehat a^2 - 2\widehat a \widehat c ) } { \widehat c ( 1 + \widehat a + \widehat c ) } { \cal j } c^{}_{2\theta_{23}^{}}\cot\delta \nonumber \\ & & + 4 \alpha^2 c^{}_{2\theta_{12}^ { } } c^ { } _ { 2\theta_{23}^ { } } { \cal j } \cot\delta \bigr[\frac{4 } { c^{2}_{\theta_{13}^ { } } ( 1 - \widehat a + \widehat c ) } -\frac { ( 1 - \widehat a - \widehat c ) ^2 ( 1 + 2 c^{2}_{\theta_{13}^ { } } + \widehat a + 3 \widehat c ) } { 2 \widehat c s^{2}_{\theta_{13}^ { } } ( 1 + \widehat a + \widehat c ) ^2 } \nonumber\\ & & - \frac{2 } { \widehat c ( 1 + \widehat a + \widehat c ) } \bigr ] -\frac{\alpha^2 ( 1 + \widehat a ) } { 2 \widehat c } s^{2}_{2\theta_{12}^ { } } s^{2}_{\theta_{13}^ { } } s^{2}_{2\theta_{23}^ { } } c^{}_{2\delta } \biggm\}\frac{\sin^2 \widetilde f_{-}^ { } } { \epsilon^2 } \nonumber\\ & & - \frac{8 \alpha { \cal j } } { \epsilon \widehat c ( 1 + \widehat a + \widehat c ) } ( \cos \widetilde f^{}_{- } - \cos \widetilde f^{}_{+ } ) \sin \widetilde
f^{}_{-}\;. \label{eq : ptm_new}\end{aligned}\ ] ] the aboslute error of the above @xmath336 for a wide range of neutrino energies and baseline lengths has been shown in fig .
[ fig:2d ] .
now we show the mapping of three mixing angles and the jarlskog invariant with an arbitrary @xmath0 . comparing between eq .
( [ eq : pab_dis ] ) with @xmath372 and eq .
( [ eq : pee ] ) , or similarly between eq .
( [ eq : pab_app ] ) with @xmath373 and eq .
( [ eq : pme ] ) , one can obtain relations similar to eq .
( [ eq : map_ee ] ) but for an arbitrary @xmath0 .
then , based on the expressions for @xmath374 and @xmath375 in appendix a , we can get @xmath376 @xmath377 @xmath378 @xmath379 which reduce to the results given in eqs .
( [ eq : map_mix ] ) and ( [ eq : map_j ] ) if @xmath264 is taken .
the authors would like to thank professor zhi - zhong xing for helpful discussions and partial involvement at the early stage of this work , which was supported in part by the national natural science foundation of china under grant nos .
11135009 and 11305193 , by the strategic priority research program of the chinese academy of sciences under grant no .
xda10010100 , by the national recruitment program for young professionals and the cas center for excellence in particle physics ( ccepp ) .
99 k. a. olive _ et al . _
[ particle data group collaboration ] , `` review of particle physics , '' chin .
c * 38 * , 090001 ( 2014 ) .
b. pontecorvo , `` mesonium and anti - mesonium , '' sov .
jetp * 6 * , 429 ( 1957 ) [ zh .
fiz . * 33 * , 549 ( 1957 ) ] .
z. maki , m. nakagawa and s. sakata , `` remarks on the unified model of elementary particles , '' prog .
* 28 * , 870 ( 1962 ) .
m. c. gonzalez - garcia , m. maltoni and t. schwetz , `` updated fit to three neutrino mixing : status of leptonic cp violation , '' jhep * 1411 * , 052 ( 2014 ) [ arxiv:1409.5439 ] . m. c. gonzalez - garcia , m. maltoni and t. schwetz , `` global analyses of neutrino oscillation experiments , '' nucl .
b * 908 * , 199 ( 2016 ) [ arxiv:1512.06856 ] .
f. capozzi , e. lisi , a. marrone , d. montanino and a. palazzo , `` neutrino masses and mixings : status of known and unknown @xmath380 parameters , '' nucl .
b * 908 * , 218 ( 2016 ) [ arxiv:1601.07777 ] .
l. wolfenstein , `` neutrino oscillations in matter , '' phys .
d * 17 * , 2369 ( 1978 ) .
s. p. mikheev and a. y. smirnov , `` resonance amplification of oscillations in matter and spectroscopy of solar neutrinos , '' sov .
j. nucl .
* 42 * , 913 ( 1985 ) [ yad .
fiz . * 42 * , 1441 ( 1985 ) ] .
y. f. li , j. cao , y. wang and l. zhan , `` unambiguous determination of the neutrino mass hierarchy using reactor neutrinos , '' phys .
d * 88 * , 013008 ( 2013 ) [ arxiv:1303.6733 ] .
s. b. kim , `` new results from reno and prospects with reno-50 , '' nucl
. part .
* 265 - 266 * , 93 ( 2015 ) [ arxiv:1412.2199 ]
. y. f. li , y. wang and z. z. xing , `` terrestrial matter effects on reactor antineutrino oscillations at juno or reno-50 : how small is small ?
. phys .
c * 40 * , no .
9 , 091001 ( 2016 ) [ arxiv:1605.00900 ] .
f. an _ et al . _ [ juno collaboration ] , j. phys .
g * 43 * , no . 3 , 030401 ( 2016 ) [ arxiv:1507.05613 ] .
s. iwamoto , recent results from t2k and future prospects " , talk given at international conference on high energy physics , chicago , 2016 .
r. acciarri _ et al . _ [ dune collaboration ] , `` long - baseline neutrino facility ( lbnf ) and deep underground neutrino experiment ( dune ) : volume 2 : the physics program for dune at lbnf , '' arxiv:1512.06148 .
e. baussan _ et al .
_ [ essnusb collaboration ] , `` a very intense neutrino super beam experiment for leptonic cp violation discovery based on the european spallation source linac , '' nucl .
b * 885 * , 127 ( 2014 ) [ arxiv:1309.7022 ] .
j. cao _ et al .
_ , `` muon - decay medium - baseline neutrino beam facility , '' phys .
st accel .
beams * 17 * , 090101 ( 2014 ) [ arxiv:1401.8125 ] .
m. blennow , p. coloma and e. fernndez - martinez , `` the moment to search for cp violation , '' jhep * 1603 * , 197 ( 2016 ) [ arxiv:1511.02859 ] .
m. g. aartsen _ et al . _
[ icecube pingu collaboration ] , `` letter of intent : the precision icecube next generation upgrade ( pingu ) , '' arxiv:1401.2046 .
u. f. katz [ km3net collaboration ] , `` the orca option for km3net , '' [ arxiv:1402.1022 ] .
k. abe _ et al .
_ , `` letter of intent : the hyper - kamiokande experiment detector design and physics potential , '' arxiv:1109.3262 . s. ahmed _ et al . _
[ ical collaboration ] , `` physics potential of the ical detector at the india - based neutrino observatory ( ino ) , ''
arxiv:1505.07380 . v. a. naumov , `` three neutrino oscillations in matter , cp violation and topological phases , '' int .
j. mod .
d * 1 * , 379 ( 1992 ) .
p. f. harrison and w. g. scott , `` cp and t violation in neutrino oscillations and invariance of jarlskog s determinant to matter effects , '' phys .
b * 476 * , 349 ( 2000 ) [ hep - ph/9912435 ] .
z. z. xing , `` sum rules of neutrino masses and cp violation in the four neutrino mixing scheme , '' phys
. rev .
d * 64 * , 033005 ( 2001 ) [ hep - ph/0102021 ] . c. jarlskog ,
`` commutator of the quark mass matrices in the standard electroweak model and a measure of maximal cp violation , '' phys .
* 55 * , 1039 ( 1985 ) .
d. d. wu , `` the rephasing invariants and cp , '' phys .
d * 33 * , 860 ( 1986 ) .
s. toshev , `` on t violation in matter neutrino oscillations , '' mod .
a * 6 * , 455 ( 1991 ) .
h. fritzsch and z. z. xing , `` mass and flavor mixing schemes of quarks and leptons , '' prog .
* 45 * , 1 ( 2000 ) [ hep - ph/9912358 ] .
j. a. aguilar - saavedra and g. c. branco , `` unitarity triangles and geometrical description of cp violation with majorana neutrinos , '' phys .
d * 62 * , 096009 ( 2000 ) [ hep - ph/0007025 ]
. j. sato , `` neutrino oscillation and cp violation , '' nucl .
instrum .
a * 472 * , 434 ( 2001 ) [ hep - ph/0008056 ] .
y. farzan and a. y. smirnov , `` leptonic unitarity triangle and cp violation , '' phys .
d * 65 * , 113001 ( 2002 ) [ hep - ph/0201105 ] .
h. zhang and z. z. xing , `` leptonic unitarity triangles in matter , '' eur .
j. c * 41 * , 143 ( 2005 ) [ hep - ph/0411183 ] .
z. z. xing and h. zhang , `` reconstruction of the neutrino mixing matrix and leptonic unitarity triangles from long - baseline neutrino oscillations , '' phys .
b * 618 * , 131 ( 2005 ) [ hep - ph/0503118 ] .
h. j. he and x. j. xu , `` connecting leptonic unitarity triangle to neutrino oscillation , '' phys .
d * 89 * , no .
7 , 073002 ( 2014 ) [ arxiv:1311.4496 ] .
z. z. xing and j. y. zhu , `` leptonic unitarity triangles and effective mass triangles of the majorana neutrinos , '' nucl .
b * 908 * , 302 ( 2016 ) [ arxiv:1511.00450 ] .
h. j. he and x. j. xu , `` connecting leptonic unitarity triangle to neutrino oscillation with cp violation in vacuum and in matter , '' arxiv:1606.04054 .
a. cervera , a. donini , m. b. gavela , j. j. gomez cadenas , p. hernandez , o. mena and s. rigolin , `` golden measurements at a neutrino factory , '' nucl .
b * 579 * , 17 ( 2000 ) [ hep - ph/0002108 ] .
m. freund , `` analytic approximations for three neutrino oscillation parameters and probabilities in matter , '' phys .
d * 64 * , 053003 ( 2001 ) [ hep - ph/0103300 ] .
e. k. akhmedov , r. johansson , m. lindner , t. ohlsson and t. schwetz , `` series expansions for three flavor neutrino oscillation probabilities in matter , '' jhep * 0404 * , 078 ( 2004 ) [ hep - ph/0402175 ] .
x. j. xu , `` why is the neutrino oscillation formula expanded in @xmath381 still accurate near the solar resonance in matter ? , ''
jhep * 1510 * , 090 ( 2015 ) [ arxiv:1502.02503 ]
. h. minakata and s. j. parke , `` simple and compact expressions for neutrino oscillation probabilities in matter , '' jhep * 1601 * , 180 ( 2016 ) [ arxiv:1505.01826 ] .
p. b. denton , h. minakata and s. j. parke , `` compact perturbative expressions for neutrino oscillations in matter , '' jhep * 1606 * , 051 ( 2016 ) [ arxiv:1604.08167 ] .
z. z. xing and j. y. zhu , `` analytical approximations for matter effects on cp violation in the accelerator - based neutrino oscillations with e @xmath382 1 gev , '' jhep * 1607 * , 011 ( 2016 ) [ arxiv:1603.02002 ] . c. moler and c. van loan , nineteen dubious ways to compute the exponential of a matrix , siam review * 20 * ( 1978 ) 801 .
s. parke , `` what is @xmath383 ?
, '' phys .
d * 93 * , no . 5 , 053008 ( 2016 ) [ arxiv:1601.07464 ] . z. z. xing , new formulation of matter effects on neutrino mixing and cp violation , phys .
b * 487 * ( 2000 ) 327 [ hep - ph/0002246 ] .
q. y. liu and a. y. smirnov , `` neutrino mass spectrum with muon - neutrino to sterile - neutrino oscillations of atmospheric neutrinos , '' nucl .
b * 524 * , 505 ( 1998 ) [ hep - ph/9712493 ] .
q. y. liu , s. p. mikheyev and a. y. smirnov , `` parametric resonance in oscillations of atmospheric neutrinos ?
, '' phys .
b * 440 * , 319 ( 1998 ) [ hep - ph/9803415 ] . | motivated by tremendous progress in neutrino oscillation experiments , we derive a new set of simple and compact formulas for three - flavor neutrino oscillation probabilities in matter of a constant density .
a useful definition of the @xmath0-gauge neutrino mass - squared difference @xmath1 is introduced , where @xmath2 for @xmath3 are the ordinary neutrino mass - squared differences and @xmath4 is a real and positive parameter . expanding neutrino oscillation probabilities in terms of @xmath5 , we demonstrate that the analytical formulas can be remarkably simplified for @xmath6 , with @xmath7 being the solar mixing angle . as a by - product , the mapping from neutrino oscillation parameters in vacuum to their counterparts in matter
is obtained at the order of @xmath8 .
finally , we show that our approximate formulas are not only valid for an arbitrary neutrino energy and any baseline length , but also still maintaining a high level of accuracy . * looking into analytical approximations for three - flavor neutrino oscillation probabilities in matter * * yu - feng li @xmath9 * , , , + @xmath10institute of high energy physics , chinese academy of sciences , beijing 100049 , china + @xmath11school of physical sciences , university of chinese academy of sciences , beijing 100049 , china + @xmath12center for high energy physics , peking university , beijing 100871 , china pacs number(s ) : 14.60.pq , 25.30.pt |
approximatively half of the elements beyond the iron peak are formed through rapid neutron captures in stars ( r - process ; burbidge et al . 1957 ) .
` rapid ' refers to the timescale of the process relative to the @xmath5-decay rates of the unstable nuclei .
although detailed evaluations of the mechanisms of r - process nucleosynthesis have long since been made ( burbidge et al . 1957 ; seeger , fowler & clayton 1965 ) , the dominant production site of the r - process elements has not yet been unambiguously identified ( see e.g. thielemann et al .
2010 ) .
the challenge identifying the main astrophysical r - process site consists in meeting the various constraints coming from measurements of abundances in galactic stars , from the modeling of possible sources such as supernovae and neutron star mergers , from nuclear physics and , last but not least , from detailed galactic chemical evolution studies .
observations of heavy element abundances in galactic halo stars provide important constraints on the astrophysical site(s ) of r - process nucleosynthesis .
early recognition that the heavy elements abundance patterns in extremely metal - poor stars ( [ fe / h ] @xmath63.0 dex ) involve only r - process products ( truran 1981 ) led to the view that r - process nucleosynthesis must be associated with the environments provided by the evolution of massive stars ( @xmath7 10 m@xmath8 ) .
neutrino - driven winds from proto - neutron stars following the delayed explosions of very massive stars ( @xmath7 20 m@xmath8 ) have been suggested as a promising site to form the solar r - process abundances , though many shortcomings were immediately apparent with this scenario ( takahashi et al .
1994 ; woosley et al .
1994 ; wanajo et al .
proton - rich winds were first found by liebendoerfer et al .
( 2003 ) and discussed in detail also by froelich et al .
( 2006a , b ) as well as pruet et al .
( 2005 , 2006 ) .
recently , hydrodynamical simulations with accurate neutrino transport have shown that the neutrino winds are proton - rich ( arcones et al .
2007 ; fischer et al .
2010 ; hdepohl et al .
2010 ) or slightly neutron - rich ( martnez - pinedo et al . 2012 ; roberts et al .
2012 ) , but never very neutron - rich , as found in older simulations ( e.g. woosley et al . 1994 ; takahashi et al . 1994 ) .
this casts serious doubts on the validity of the neutrino wind scenario .
it seems now established that neutrino - driven winds from proto - neutron stars can not be the main origin of the r - process elements beyond @xmath9110 ( wanajo 2013 ; arcones & thielemann 2013 ) . on the other hand , prompt explosions of massive stars in the 810 m@xmath8 range
may lead to an ejected amount of r - process matter consistent with the observed galactic abundances ( wheeler et al .
yet , it is not clear whether these prompt explosions do occur .
recent 3d magneto - hydrodynamics investigations point to supernova progenitors characterized by high rotation rates and large magnetic fields as an interesting site for the strong r - process observed in the early galaxy ( winteler et al .
2012 , and references therein ) . in this context
, the rarity of progenitors with the required initial conditions would also provide a natural explanation for the scatter in the abundances of r - process elements observed for low - metallicity stars .
this r - process elements production channel certainly deserves further investigation
. another major source of r - process elements might be neutron star mergers ( lattimer & schramm 1974 , 1976 ; lattimer et al .
1977 ; freiburghaus et al .
1999 ; goriely et al .
2011 ; roberts et al .
the resulting abundance patterns are extremely robust with respect to varying the parameters of the merging binary system and the results from a double neutron star and a neutron star black hole merger are practically indistinguishable , see , for example , fig .
4 in korobkin et al .
therefore we refer to both types of systems collectively as compact binary mergers ( cbm ) .
rosswog et al .
( 1999 , 2000 ) and more recently oechslin et al . ( 2007 ) , bauswein et al . (
2013 ) , rosswog ( 2013 ) , hotokezaka et al .
( 2013 ) , kyutoku et al .
( 2013 ) showed that up to 10@xmath10 m@xmath8 of r - process matter may be ejected in a single coalescence event .
though this quantity is orders of magnitude higher than the average r - process ejecta required from sneii , if every snii is expected to produce r - process matter , the rate of cbm in the galaxy is significantly lower than the snii rate , making unclear which one of the sources could potentially be the major r - process elements producer .
neutron star mergers could also provide a natural explanation for the scatter of r - process elements abundances at low metallicity , given their rarity and high r - process element production . in principle ,
chemical evolution studies offer a way to discriminate among different sites for the r - process elements production , through the comparison of model predictions with the observations . unfortunately
, different studies have reached different conclusions ( e.g. travaglio et al .
1999 ; ishimaru & wanajo 1999 ; de donder & vanbeveren 2003 ; argast et al . 2004 ; cescutti et al . 2006 ) . furthermore ,
only rarely all the possible r - process elements sources have been screened in the very same study . in particular ,
argast et al . (
2004 ) analysed the possibilities of r - process production from sneii and cbm , and concluded that it is unlikely that cbm can be responsible for the entire r - process production , because of the delayed eu appearance predicted by their model , which would not allow to reproduce the milky way data at low metallicity .
this was due to the adoption of an inhomogeneous chemical evolution model , dropping the instantaneous mixing approximation ( i.m.a . ) in the early galactic evolutionary phases .
this model was predicting the spread for r - process elements at low metallicity but also for other elements which do not show a large spread . on the other hand , de donder & vanbeveren ( 2003 ) had concluded that neutron star ( ns)/black hole mergers could be responsible for the galactic r - process production , but they did not consider the possible contribution from sneii .
their model assumed i.m.a . in this paper
we intend to study again the problem of r - process production in the galaxy and in particular of europium , in the light of recent and detailed nucleosynthesis calculations of r - process production in cbm ( korobkin et al .
2012 ) , as well as of the existence of new detailed data .
we will explore the possibility of eu production from sneii and cbm with the intent of establishing which one of the two sources is the most likely and what are the shortcomings of both scenarios .
we will adopt a very recent detailed chemical evolution model for the milky way including updated stellar yields and reproducing the majority of the observational constraints .
we will also predict the expected eu gradient along the galactic disc .
the paper is organized as follows . in section 2
we describe the chemical evolution model . in section 3
we show the adopted eu nucleosynthesis prescriptions for the cbm and sneii , as well as the computation of the cbm rate . in section 4 the results are presented and in section 5 the main conclusions are summarized .
we adopted the model 15 of romano et al .
( 2010 ) , which contains updated stellar yields and reproduces the majority of the [ x / fe ] versus [ fe / h ] relations observed in galactic stars in the solar vicinity .
this model is an updated version of the chiappini et al .
( 1997 , 2001 ) two - infall model , where it is assumed that the inner halo and part of the thick disc formed by means of a gas accretion episode independently from the thin disc , which formed by means of another gas accretion episode on a much longer timescale .
this model relaxes the instantaneous recycling approximation ( i.r.a . ) , i.e. the stellar lifetimes are taken into account in detail , but retains i.m.a .
, i.e. the stellar ejecta are assumed to cool and mix instantaneously with the surrounding interstellar medium ( ism ) .
the model does not consider galactic fountains , but see spitoni et al .
( 2009 ) , who showed that the fountains should not be affecting the chemical abundances in the disc , as long as the process is shorter than 100 myr , and does not allow for gas recycling through the hot halo ( but see e.g. brook et al .
2013 ) . in the spitoni et al .
( 2009 ) paper it was shown that the effect of the fountains is also to delay the chemical enrichment , thus breaking the i.m.a .. however , it was found that even a delay of several hundreds million years would not change the chemical results for the evolution of the galactic disc , thus supporting i.m.a .. on the other hand , inhomogeneities in the ism could be important in the early evolutionary phases , during halo formation . in this respect , there are conflicting results : on one hand it is found a very little spread in abundance ratios of many elements down to [ fe / h]=-4.0 dex , on the other hand a large spread is observed in abundance ratios involving s- and r - process elements relative to fe . in fact , the data for europium show a large spread at low metallicities : the interpretation of this spread , however , can be related more to the different stellar producers of these elements rather than to a less efficient mixing ( see cescutti et al .
2006 , 2013 ) , since in this latter case the spread should be seen for all chemical species . in summary
, we think that i.m.a .
is not a bad assumption on the scale of the solar vicinity also because , besides the above mentioned results of spitoni et al .
( 2009 ) , it has been shown ( recchi et al . 2001 ) that mixing in the ism can occur on very short timescales of the order of tens of million years .
the following equation describes the evolution of the surface mass density of the gas in the form of the generic element @xmath11 , @xmath12 : @xmath13 @xmath14 @xmath15 @xmath16dm_{b}\ ] ] @xmath17 @xmath18 @xmath19 where @xmath20 is the abundance by mass of the element @xmath11 at the time @xmath21 and galactic radius @xmath22 ; @xmath23 indicates the fraction of mass restored by a star of mass @xmath24 in the form of the element @xmath11 , the so - called ` production matrix ' as defined by talbot & arnett ( 1973 ) .
the upper mass limit , @xmath25 , is set to 100 m@xmath2 , while @xmath26 , the lightest mass which contributes to the chemical enrichment , is set to 0.8 m@xmath2 .
the parameter @xmath27 represents the fraction of binary systems with the right characteristics to give rise to type ia sne ( sneia ) in the initial mass function ( imf ) ; its value is chosen in order to obtain the best fit to the present time snia rate .
the adopted progenitor model for sneia is the single - degenerate model as suggested by greggio & renzini ( 1983 ) and later reproposed by matteucci & recchi ( 2001 ) .
this formulation , which is different from single - degenerate rates computed by means of population synthesis models ( e.g. mennekens et al .
2010 ) , has proven to be successful in reproducing the chemical evolution of the milky way , as well as of other galaxies , such as ellipticals .
this rate is also very similar to the rate deriving from the double - degenerate model ( see greggio , 2005 ) and the two rates produce the same chemical evolution results , as shown in matteucci et al .
( 2006 , 2009 ) , where the interested reader can find more details .
in other words , from the point of view of chemical evolution , using the single- or double - degenerate model or a combination of the two , does not produce any noticeable effect in the chemical results .
the term @xmath28 represents the gas accretion rate : @xmath29 the quantities @xmath30 are the abundances in the infalling material , which is assumed to be primordial , and are set after romano et al .
( 2006 ) , while @xmath31 gyr is the time for maximum infall onto the thin disc , @xmath32 gyr is the time scale for the formation of the inner halo / thick disc and @xmath33 is the timescale for the formation of the thin disc and is a function of the galactocentric distance ( inside - out formation , matteucci & franois 1989 ; chiappini et al .
2001 ; pilkington et al .
2012 ) . in the framework of our model , for the solar neighbourhood
the best value is @xmath34 gyr ( see chiappini et al .
1997 ; romano et al .
the quantities @xmath35 and @xmath36 are parameters fixed by reproducing the present time total surface mass densities in the halo and disc of the galaxy ( see romano et al .
2000 for details ) .
the adopted imf , @xmath37 , is that of scalo ( 1986 ) and the stellar lifetimes are taken from schaller et al .
the assumed star formation rate ( sfr ) , @xmath38 , is a schmidt - kennicutt law proportional to the surface gas density to the 1.5th power .
the model computes in detail the chemical abundances of 37 species in the ism .
for all elements but europium , the adopted stellar yields , that are used to compute the entries of the @xmath23 matrix , are described in detail in romano et al .
they reproduce very well the abundance patterns of most chemical species observed in the stars of the milky way halo ( see also brusadin et al . 2013 ) and discs ( micali et al . 2013 ) . in the following ,
we review a few basic facts . *
low- and intermediate - mass stars ( @xmath39 0.88 m@xmath8 ) contribute mainly to the chemical enrichment in he , c , n , s - process elements and , perhaps , some @xmath40li and na .
the adopted stellar yields are from karakas ( 2010 ) and rest on detailed stellar evolutionary models . *
massive stars ( @xmath7 8 m@xmath8 ) are responsible for the production of the @xmath41- and iron - peak elements .
the production of ( primary ) nitrogen and s - process elements is boosted at low - metallicity in fast stellar rotators ( meynet & maeder 2002a ; frischknecht , hirschi & thielemann 2012 ) .
here we adopt he , c , n and o yields from pre - supernova models of rotating massive stars from meynet & maeder ( 2002b ) , hirschi et al .
( 2005 ) , hirschi ( 2007 ) and ekstrm et al .
( 2008 ) . for heavier elements ,
yields are from kobayashi et al .
( 2006 ) . *
when in binary systems with the right characteristics to give rise to snia events , white dwarfs ( originating from low- and intermediate - mass stars ) are responsible for the production of the bulk of iron in the galaxy .
the adopted snia yields are those of iwamoto et al .
( 1999 , their model w7 ) . in the next section ,
we discuss how europium production from stars has been implemented in our model .
as discussed in the introduction , two possible sites have been suggested for the production of eu in stars : sneii of either low ( 810 m@xmath2 ) or high mass ( @xmath42 20 m@xmath2 ; cowan et al .
1991 ; woosley et al . 1994 ; wanajo et al . 2001 ) and cbm ( lattimer & schramm 1974 ; eichler et al . 1989 ; freiburghaus et al . 1999 ; rosswog et al .
1999 , 2000 ) .
however , the classical site for the production of r - process elements , namely sneii , has been recently questioned ( e.g. arcones et al . 2007 ) while the r - process production in cbm seems a very robust result ( e.g. korobkin et al . 2012 ) .
we have computed the evolution of the eu abundance in the milky way under several assumptions : i ) eu is produced only in cbm ; ii ) eu is produced only in snii explosions ; iii ) eu is produced both in cbm and snii explosions . to include the production of eu from coalescence of neutron stars galactic chemical evolution model we need to define the following quantities : 1 . the realization probability for cbm , @xmath43 ; 2 . the time delay between the formation of the double neutron star system and the merging event , @xmath44 ; 3 .
the amount of eu produced during the merging event , @xmath45 . in our model , the rate of cbm at the time @xmath21 is computed under the assumption that the rate of formation of double neutron star systems , which will eventually coalesce , is a fraction @xmath43 of the neutron star formation rate at the time @xmath46 : @xmath47 where @xmath48 and @xmath49 m@xmath2 are the canonical lower and upper masses , at birth , which can leave a neutron star as a remnant
( we will come back to the issue of the choice of the upper mass limit in sections [ sec : results ] and [ sec : discuss ] ) .
stars with @xmath7 30 m@xmath2 probably leave black holes as remnants but the situation is quite uncertain and depends on the assumed rate of mass loss in massive stars and its dependence upon stellar metallicity ( e.g. meynet & maeder , 2002a , b ) .
the value of the parameter @xmath43 is chosen by imposing that equation ( 3 ) reproduces the present - time rate of neutron star merging in the galaxy .
several observational estimates of this rate appeared in the literature ( van den heuvel & lorimer 1996 ; kalogera & lorimer 2000 ; belczynsky et al .
2002 ; kalogera et al .
2004 ) . here
, we take that of kalogera et al .
( 2004 ) , @xmath50 83@xmath51 myr@xmath52 , and find @xmath53 0.018 .
based on the energy / angular momentum loss to gravitational waves ( peters and mathews 1964 ) , inspiral times are usually thought to be between 10 and 100 myrs , though some studies ( e.g. belczynski et al .
2002 ) find that a large fraction of systems would merge within less than a 1 myr .
argast et al .
( 2004 ) , in a work similar to ours , considered two different timescales : 1 myr and 100 myr . here
we will consider 1 myr , 10 myr and 100 myr .
it is worth noting that in both this work and argast et al .
( 2004 ) it is assumed that all neutron star binaries have the same coalescence timescale . clearly , a more realistic approach would consider a distribution function of such timescales , in analogy with sneia for which a distribution for the explosion times is defined ( see greggio 2005 ) .
.model parameters . [ cols="<,^,^,^ " , ] in table [ tab : sun ] we list the predicted eu and fe solar abundances by mass from the various models .
they correspond to the abundances in the ism 4.5 gyr ago .
the observed values for the eu and fe solar abundances are @xmath54 3.5 @xmath0 10@xmath55 and @xmath56 1.34 @xmath0 10@xmath57 , respectively ( asplund et al . 2009 ) .
most of the models predict eu solar abundances in agreement with the observed one .
since in all the previous models we fixed the maximum mass giving rise to a neutron star to be 30 m@xmath8 and this value is quite uncertain , we decided to try to change this upper limit . in fig.[fig : ns50 ] we show the effect on model predictions of a different choice of the upper mass limit for neutron star formation in equation ( 3 ) , namely , 50 m@xmath8 , rather than 30 m@xmath8 .
it is shown that in this case cbm alone can , in principle , account for the abundances of eu observed in galactic halo stars , as well as for the solar eu ( see table 2 ) . finally , in figs .
[ fig : grad ] and [ fig : gradb ] we show the predicted gradient of eu along the galactic disc at the present time for a subset of models . in both figures ,
the data are from cepheids by luck et al .
( 2011 ) . in fig .
[ fig : grad ] we show cases with eu production only by cbm . in fig .
[ fig : gradb ] we show cases with eu production from both neutron stars and sneii .
both scenarios can account reasonably well for the observed gradient of europium in the milky way disc .
this is because the gradient depends on the mechanism of formation of the disc , here assumed to be the inside - out one .
this mechanism was suggested by matteucci & franois ( 1989 ) and proven to be valid also in semi - analytical models ( e.g. pilkington et al .
what it changes in figs [ fig : grad ] and [ fig : gradb ] is the absolute value of the eu abundance , [ eu / h ] , as due to the different assumptions made on the eu producers . in this case , the models with both sneii and cbm as eu producers produce the best agreement with the data ( see fig . [
fig : gradb ] ) .
in this paper we have studied the production of eu and assumed that this element can be produced in both compact binary mergers ( cbm ) and sneii . to do that , we have adopted a very detailed chemical evolution model which can predict the evolution of the abundances of many species and that already reproduces the behaviour of several abundances as well as the main features of the solar neighbourhood and the whole disc .
our attention has been focused here on the production of eu by cbm and whether this production alone can explain the solar abundance of eu as well as the [ eu / fe ] versus [ fe / h ] relation .
the main parameters involved in eu production from cbm are : i ) the eu yield , ii ) the time required for the binary neutron star system to coalesce and iii ) the range of progenitors of neutron stars . among these parameters ,
ii ) and iii ) are quite uncertain whereas the yields seem to be more reliable .
our main conclusions can be summarized as follows : * cbm can be entirely responsible for eu production in the galaxy if the ns systems all have a coalescence time scale no longer than 1 myr , a possibility suggested by belczynski et al .
( 2002 ) , each event produces at least 3 @xmath0 10@xmath1 m@xmath2 of eu and all stars with masses in the range 950 m@xmath2 leave a ns as a remnant . in this case
we can well reproduce the average trend of [ eu / fe ] versus [ fe / h ] in the solar vicinity as well as the eu solar abundance . in this case
, there is no need for sneii producing eu .
however , all the uncertainties in these parameters plus the uncertain observed rate of cbm in the galaxy at the present time , induce some caution in drawing firm conclusions . *
perhaps , a more realistic situation would be the one where both cbm and sneii are producing eu . the best model in this case requires that cbm can produce 2 @xmath0 10@xmath1 m@xmath2 of eu and the delay times can be various , spanning between 10 and 100 myr .
sneii should then produce eu in the range 2050 m@xmath2 with yields of the order of 10@xmath310@xmath4 m@xmath2 of eu per supernova .
it is very important to have high stellar masses to produce the eu observed at very low metallicity . * both models with eu produced only by cbm and models with cbm and sne can reproduce the eu abundance gradient observed along the galactic thin disc .
our conclusions are different from those of argast et al .
( 2004 ) who concluded that cbm can not be the only eu producers .
we think that this is due to the fact that argast et al.s model does not assume instantaneous mixing in the early galactic evolutionary phases .
this fact leads to an additional delay in the appearence of eu in the ism , besides the delay for the merging of the two ns , and is probably the reason why their predicted [ eu / fe ] appears at a too high [ fe / h ] values relative to observations . on the other hand
, the assumption that the ism was not well mixed at early times can explain the large spread observed in the r- and s - process abundances relative to fe at low metallicity .
however , such a large spread is not observed for other abundance ratios at the same metallicity .
therefore , either the spread is explained as due to different stellar producers of different elements as suggested in cescutti et al .
( 2006 , 2013 ) or the spread is related to observational errors .
inhomogeneous mixing , in fact , should act on all the elements .
our model assumes instantaneous mixing approximation ( i.m.a . )
and therefore it can not reproduce the observed spread but just the average trends . on the other hand , de donder & vanbeveren ( 2003 ) , using a model similar to this one with i.m.a .
, taken from chiappini et al .
( 1997 ) , and computing population synthesis binary models , explored several cases of mergers : ns / ns and ns / black hole .
they concluded that mergers ns / black hole can produce enough eu by themselves but they did not test the case of eu production from sneii .
more recently , mennekens & vanbeveren ( 2013 ) , adopting again a population synthesis model and a galactic model like in de donder & vanbeveren ( 2003 ) , reached a conclusion similar to that of the present paper : the cbm can account for the entire r - process production except in the first 100 myr , but they did not include the contribution from sneii .
therefore , although firm conclusions on the nature of eu can not be yet drawn , this subject should be pursued by testing the various hypothesis discussed here in a model which takes into account early inhomogeneities as well as a distribution function of the delay times for the merging of neutron stars , and that will be the subject of a forthcoming paper .
f.m . and d.r .
acknowledge financial support from prin miur 20102011 , project `` the chemical and dynamical evolution of the milky way and local group galaxies '' , prot .
2010ly5n2 t .
aknowledges financial support from the helmoltz - university young investigator grant no .
vh - ng-825 .
o.k . and s.r .
have been supported by by dfg grant ro-3399 , aobj-584282 and by the swedish research council ( vr ) under grant 621- 2012 - 4870 .
s.r . has also been supported by compstar .
finally , we thank a highly competent referee for his / her careful reading of the manuscript and useful suggestions .
90 arcones a. , janka h .- th .
, scheck l. , 2007 , a&a , 467 , 1227 arcones , a. , thielemann , f .- k . , 2013 , jphg , 40 , 3201 argast d. , samland m. , thielemann f .- k . ,
qian y .- z .
, 2004 , a&a , 416 , 997 asplund , m. , grevesse , n. , sauval , a. j. , scott , p. , 2009
, ara&a , 47 , 481 bauswein , a. , goriely , s. , janka , h .-
, 2013 , apj , 773 , 78 bensby , t. , feltzing , s. , lundstrm , i. , ilyin , i. , 2005 , a&a , 433,185 belczynski , k. , kalogera , v. , bulik , t. , 2002 , apj , 572 , 407 brook c. b. , stinson g. , gibson b. k. , shen s. , macci a. v. , wadsley j. , quinn t. , 2013 , preprint ( arxiv:1306.5766 ) brusadin g. , matteucci f. , romano d. , 2013 , a&a , 554 , a135 burbidge e. m. , burbidge g. r. , fowler w. a. , hoyle f. , 1957 , rev .
modern phys . , 29 , 547 burris , d.l . , pilachowski , c. a. , armandroff , t. e. , sneden , c. , cowan , j. j. , roe , h. , 2000 , apj , 544 , 302 cescutti g. , franois p. , matteucci f. , cayrel r. , spite m. , 2006 , a&a , 448 , 557 cescutti , g. , chiappini , c. , hirschi , r. , meynet , g. , frischknecht , u. 2013 , a&a , 553 , 51 chiappini c. , matteucci f. , gratton r. , 1997 , apj , 477 , 765 chiappini c. , matteucci f. , romano d. , 2001 , apj , 554 , 1044 cowan j. j. , thielemann f .- k . , truran j. w. , 1991 , phys . rep .
, 208 , 267 de donder , e. & vanbeveren , d. , 2003 , newastr . , 8 , 415 eichler d. , livio m. , piran t. , schramm d.n . , 1989 , nature,340 , 126 ekstrm s. , meynet g. , chiappini c. , hirschi r. , maeder a. , 2008 , a&a , 489 , 685 fischer t. , whitehouse s. c. , mezzacappa a. , thielemann f .- k . , liebendrfer m. , 2010 , a&a , 517 , a80 franois , p. , depagne , e. , hill , v. , spite , m. , spite , f. , plez , b. , beers , t. c. , andersen , j. et al . , 2007 ,
a&a , 476 , 935 frebel , a. , 2010 , astronomische nachrichten , vol.331 , issue 5 , p.474 - 488 freiburghaus c. , rosswog s. , thielemann f .- k . , 1999 , apj , 525 , l121 frischknecht u. , hirschi r. , thielemann f .- k . , 2012 ,
a&a , 538 , l2 froehlich , c. , hix , w. r. , martnez - pinedo , g. , liebendoerfer , m. , thielemann , f .- k . , bravo , e. , langanke , k. , zinner , n. t. , 2006a , newar , 50 , 496 froehlich , c. , martnez - pinedo , g. , liebendoerfer , m. , thielemann , f .- k . ,
bravo , e. , hix , w. r. , langanke , k. , zinner , n. t. , 2006b , phrvl , 96 , 2502 fulbright , j.p . , 2000 , aj , 120 , 1841 goriely , s. , bauswein , a. , janka , h .-
, 2011 , apj , 738 , l32 greggio , l. , 2005 , a&a , 441 , 1055 greggio l. , renzini a. , 1983 , a&a , 118 , 217 hirschi r. , 2007 , a&a , 461 , 571 hirschi r. , meynet g. , maeder a. , 2005 , a&a , 433 , 1013 hotokezaka , k. et al .
, 2013 , phrvd , 88 , 4026 hdepohl l. , mller b. , janka h .-
, marek a. , raffelt g. g. , 2010 , phys . rev . lett .
, 104 , 251101 ishimaru , y. & wanajo , s. , 1999 , apj , 511 , l33 iwamoto k. , brachwitz f. , nomoto k. , kishimoto n. , umeda h. , hix w. r. , thielemann f .- k . , 1999 , apjs , 125 , 439 kalogera v. , kim c. , lorimer d. r. , burgay m. , damico n. , possenti a. , manchester r. n. , lyne a. g. , joshi b. c. , mclaughlin m. a. , kramer m. , sarkissian j. m. , camilo f. , 2004 , apj , 614 , l137 kalogera , v. & lorimer , d.r : 2000 apj , 530 , 890 karakas , a.i . , 2010 , mnras , 403 , 1413 kobayashi c. , umeda h. , nomoto k. , tominaga n. , ohkubo t. , 2006 , apj , 653 , 1145 korobkin o. , rosswog s. , arcones a. , winteler c. , 2012 , mnras , 426 , 1940 kyutoku , k. , ioka , k. , shibata , m. , 2013 , phrvd , 88 , 1503 li , w. , et al .
2011 , mnras , 412 , 1473 luck , r. e. , andrievsky , s. m. , kovtyukh , v. v. , gieren , w. ; graczyk , d. 2011 , aj , 142 , 51 lattimer j. m. , mackie f. , ravenhall d. g. , schramm d. n. , 1977 , apj , 213 , 225 lattimer , j. m. & schramm , d. n. , 1974 , apj , 192 , l145 lattimer , j. m. & schramm , d. n. , 1976 , apj , 210 , 549 liebendoerfer , m. , mezzacappa , a. , messer , o. e. b. , martinez - pinedo , g. , hix , w. r. , thielemann , f .- k . , 2003 , nupha , 719 , 144 martnez - pinedo g. , fischer t. , lohs a. , huther l. , 2012 , phys .
lett . , 109 , 251104 matteucci f. , franois p. , 1989
, mnras , 239 , 885 matteucci f. , recchi s. , 2001 , apj , 558 , 351 matteucci f. , panagia n. , pipino a. , mannucci f. , recchi s. , della valle m. , 2006 , mnras , 372 , 265 matteucci f. , spitoni e. , recchi s. , valiante r. , 2009 , a&a , 501 , 531 mennekens n. , vanbeveren , d. , de greve , j.p . , de donder , e. , 2010 , a&a , 515 , 89 mennekens , n. & vanbeveren 2013 astro - ph/1307.0959 , a&a submitted meynet g. , maeder a. , 2002a , a&a , 381 , l25 meynet g. , maeder a. , 2002b , a&a , 390 , 561 micali a. , matteucci f. , romano d. , 2013 , preprint ( arxiv:1309.1283 ) , mnras in press mishenina , t. v. , gorbaneva , t. i. , bienaym , o. , soubiran , c. , kovtyukh , v. v. , orlova , l. f. , 2007 , astronomy reports , volume 51 , issue 5 , p.382 oechslin , r. , janka , h .-
marek , a. , 2007 , a&a , 467 , 395 peters , p. & mathews , j. , 1964 , phys .
review b , 27 , 173001 pilkington k. , et al . , 2012 , a&a , 540 , a56 portinari l. , chiosi c. , bressan a. , 1998 , a&a , 334 , 505 pruet , j. , woosley , s. e. , buras , r. , janka , h .-
t . , hoffman , r. d. , 2005 , apj , 623 , 325 pruet , j. , hoffman , r. d. , woosley , s. e. , janka , h .-
t . , buras , r. , 2006 , apj , 644 , 1028 ramya , p. , reddy , b. e. , lambert , d. l. 2012 , mnras , 425 , 3188 reddy , b. e. , tomkin , j. , lambert , d. l. , allende prieto , c. , 2003 , mnras , 340 , 304 recchi , s. , matteucci , f. , dercole , a. , 2001 , mnras , 322 , 800 reddy b. e. , lambert d. l. , allende prieto c. , 2006 , mnras , 367 , 1329 roberts , l. f. , kasen , d. , lee , w. h. , ramirez - ruiz , e. , 2011 , apj , 736 , l21 roberts l. f. , reddy s. , shen g. , 2012 , phys .
c , 86 , 065803 romano d. , karakas a. i. , tosi m. , matteucci f. , 2010 , a&a , 522 , a32 romano d. , matteucci f. , salucci p. , chiappini c. , 2000 , apj , 539 , 235 romano d. , tosi m. , chiappini c. , matteucci f. , 2006 , mnras , 369 , 295 rosswog , s. , 2013 , philosophical transactions of the royal society a : mathematical , physical and engineering sciences , vol .
371 , issue 1992 , pp .
20120272 - 20120272 rosswog s. , davies m. b. , thielemann f .- k .
, piran t. , 2000 , a&a , 360 , 171 rosswog s. , liebendrfer m. , thielemann f .- k . , davies m. b. , benz w. , piran t. , 1999 , a&a , 341 , 499 scalo j. m. , 1986 , fundam .
phys . , 11 , 1 schaller g. , schaerer d. , meynet g. , maeder a. , 1992 , a&as , 96 , 269 seeger p. a. , fowler w. a. , clayton d. d. , 1965 , apjs , 11 , 121 spitoni e. , matteucci f. , recchi s. , cescutti g. , pipino a. , 2009 , a&a , 504 , 87 takahashi , k. , witti , j. , janka , h .-
, 1994 , a&a , 286 , 857 talbot r. j. jr .
, arnett w. d. , 1973 , apj , 186 , 51 thielemann f .- k .
, et al . , 2010 , j. phys .
ser . , 202 , 012006 thielemann f
, kppeli r. , winteler c. , perego a. , liebendrfer m. , nishimura n. , vasset n. , arcones a. , 2012 , proceedings of the xii international symposium on nuclei in the cosmos , published online at at http://pos.sissa.it/cgi-bin/reader/conf.cgi?confid=146 , id.61 travaglio , c. , galli , d. , gallino , r. , busso , m. , ferrini , f. , straniero , o. , 1999 , apj , 521 , 691 truran j. w. , 1981 , a&a , 97 , 391 van den heuvel , e. p. j. & lorimer , d. r. , 1996 , mnras , 283 , l37 wanajo s. , 2013 , apj , 770 , l22 wanajo s. , kajino t. , mathews g. j. , otsuki k. , 2001 , apj , 554 , 578 wheeler j. c. , cowan j. j. , hillebrandt w. , 1998 , apj , 493 , l101 winteler , c. , kppeli , r. , perego , a. , arcones , a. , vasset , n. , nishimura , n. , liebendrfer , m. , thielemann , f .- k . , 2012 , apj , 750 , l22 woosley , s.e . & weaver , t.a . , 1995 , apjs , 101 , 181 woosley s. e. , wilson j. r. , mathews g. j. , hoffman r. d. , meyer b. s. , 1994 , apj , 433 , 229 | we have explored the eu production in the milky way by means of a very detailed chemical evolution model . in particular , we have assumed that eu is formed in merging neutron star ( or neutron star black hole ) binaries as well as in type ii supernovae .
we have tested the effects of several important parameters influencing the production of eu during the merging of two neutron stars , such as : i ) the time scale of coalescence , ii ) the eu yields and iii ) the range of initial masses for the progenitors of the neutron stars .
the yields of eu from type ii supernovae are very uncertain , more than those from coalescing neutron stars , so we have explored several possibilities . we have compared our model results with the observed rate of coalescence of neutron stars , the solar eu abundance , the [ eu / fe ] versus [ fe / h ] relation in the solar vicinity and the [ eu / h ] gradient along the galactic disc .
our main results can be summarized as follows : i ) neutron star mergers can be entirely responsible for the production of eu in the galaxy if the coalescence time scale is no longer than 1 myr for the bulk of binary systems , the eu yield is around 3 @xmath0 10@xmath1 m@xmath2 , and the mass range of progenitors of neutron stars is 950 m@xmath2 ; ii ) both type ii supernovae and merging neutron stars can produce the right amount of eu if the neutron star mergers produce 2 @xmath0 10@xmath1 m@xmath2 per system and type ii supernovae , with progenitors in the range 2050 m@xmath2 , produce yields of eu of the order of 10@xmath310@xmath4 m@xmath2 ; iii ) either models with only neutron stars producing eu or mixed ones can reproduce the observed eu abundance gradient along the galactic disc .
= = = = = = = = # 1 # 1 # 1 # 1 @mathgroup@group @mathgroup@normal@groupeurmn @mathgroup@bold@groupeurbn @mathgroup@group @mathgroup@normal@groupmsamn @mathgroup@bold@groupmsamn = `` 019 = ' ' 016 = `` 040 = ' ' 336 = " 33e = = = = = = = = # 1 # 1 # 1 # 1 = = = = = = = = [ firstpage ] nuclear reactions , nucleosynthesis , abundances galaxy : abundances galaxy : evolution . |
two different types of masers are observed in the 1720mhz satellite line transition of hydroxyl ( oh ) ; one is found in star forming regions ( sfrs ) and the other is associated with supernova remnants ( snrs ) .
the radiatively pumped 1720mhz masers in sfrs are accompanied by masers in the other ground - state rotational transitions ( at 1612 , 1665 and 1667mhz ) as a result of the cascade down from higher excitation levels , whereas the collisionally pumped 1720mhz line is the only oh transition observed near snrs @xcite .
the latter masers originate in post - shocked regions where an expanding snr collides with dense molecular clouds in the surrounding interstellar medium ( ism ) , and are observed near snr / ism interaction regions throughout the galaxy ( e.g. , * ? ? ?
* ; * ? ? ?
the galactic center ( gc ) with the sgraeast snr plowing into the ism surrounding the sgracomplex is no exception .
lcccccl date of observation & config . &
@xmath3 & @xmath4-range & channel rms & beam size & p.a .
& remarks + & & ( ^-1 kms@xmath1 ) & ( ^-1 kms@xmath1 ) & ( mjy beam@xmath1 ) & ( @xmath5 at @xmath6 ) & + 1986 jun 30&bna&8.5&@xmath2276 to @xmath0200&@xmath74&@[email protected] , @xmath910.4&@xcite + 1996 nov 22&bna&2.1&@xmath2194 to @xmath0194&19&@[email protected] , @xmath939.8&unpublished@xmath10 part of @xcite + 1998 jul 06&b & 8.5&@xmath2276 to @xmath0200&@xmath73&10.4 @xmath54.4 , @xmath923.3&previously unpublished archival data + 2005 jan 20 & 26&bna&2.1&@xmath2232 to @xmath0232&@xmath73&@xmath83.9 @xmath5 3.6 , @xmath257.9&new observations + + [ datasets ] the line - of - sight toward the sgracomplex consists of the snr sgraeast and a circumnuclear disk ( cnd ) .
sgraeast manifests itself as a radio continuum ridge or shell ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , in part obscured by the torus- or ring - like cnd .
the cnd consists of irregularly distributed clumps of molecular gas ( e.g. , * ? ? ?
* ; * ? ? ?
* ) rotating counter - clockwise in a ring with a mean radius of about 2 pc around the compact radio source sgra * , the dynamical center of the milky way @xcite .
interior to the cnd , most gas is ionized and is distributed in a `` minispiral '' , also known as the region sgrawest .
this line - of - sight also partly overlaps with two giant molecular clouds ( gmcs ) casually called the @xmath020 and @xmath050 ^-1 kms@xmath1clouds ( [email protected]@xmath20.08 and [email protected]@xmath20.07 ; see reviews in * ? ? ?
* ; * ? ? ?
these form the `` molecular belt '' stretching across the sgracomplex , providing the ism that interacts with sgraeast .
it has long been known that a number of bright 1720mhz masers with line - of - sight velocities near 5065 ^-1 kms@xmath1 are observed in the southwest region of sgraeast , toward the snr [email protected] @xcite .
a few other isolated masers are also distributed at other locations along the radio continuum shell with velocities between 40 and 60 ^-1 kms@xmath1 ( ( * ? ? ?
* hereafter yz96 ) ; ( * ? ? ?
* hereafter k03 ) ) . in sect .
[ snrmasers ] we present new , less prominent 1720mhz oh masers that are due to the interaction of sgraeast and the @xmath050 ^-1 kms@xmath1 molecular cloud observed toward the northeast ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in addition , a few masers with very high absolute line - of - sight velocities can be found , at least in projection , near the cnd ( yz96 ; k03 ; this work ) .
their high velocities ( about @xmath0130 and @xmath2130 ^-1 kms@xmath1 ) do not fit the snr / ism interaction model .
instead , their signature resembles the structures of 1667mhz oh and 22ghz h@xmath11o masers found in circumnuclear tori of nearby galaxies @xcite suggesting that these masers arise in the cnd itself . until now this picture for the gc depended on a previous single detection of a @xmath2132 ^-1 kms@xmath1 maser .
section [ cndmasers ] presents support for this model with the detection of a number of new high - negative velocity masers . in sect .
[ observations ] and sect .
[ results ] we outline the data reduction procedures and present the results . in sect .
[ discussion ] the nature of the different groups of 1720mhz masers in the gc is discussed . with this in mind , we present a model of the interaction of sgraeast with its surroundings . unless stated otherwise , all velocities in this paper refer to line - of - sight velocities as measured with respect to the local standard of rest ( lsr ) .
new high - sensitivity and high spatial resolution vla observations were obtained to search for high - velocity masers in the cnd .
the new , 2005 january observations ( table [ datasets ] ) had two if pairs of 1.562mhz bandwidth , each tuned to slightly offset center velocities in order to cover a large total velocity range ( @xmath2232 @xmath12232 ) .
the data were calibrated using vlarun , a pipeline vla data reduction procedure available in nrao s astronomical image processing system ( aips ) . after continuum subtraction in the uv - plane
the data were imaged with natural weighting using standard aips procedures , very similar to the data reduction as described in k03 .
the only difference was that we joined the spectra from the two if pairs into one long spectrum covering the whole velocity range with the higher spectral resolution using ujoin .
self - calibration was performed using the bright @xmath0132 ^-1 kms@xmath1 maser feature , which resulted in positional uncertainty of about 0.20.3 .
the resulting data cube has a synthesized beam full width half maximum ( fwhm ) of 39 @xmath5 36 at a position angle ( p.a . ) of @xmath257.9@xmath6 , and a channel separation of 2.1 ^-1 kms@xmath1 ( i.e. , a channel resolution of 2.5 ^-1 kms@xmath1 ) . with dual polarizations and a total on - source integration time of about three hours , a typical rms noise of 3.2 mjybeam@xmath1 per channel was achieved .
three archival vla data sets ( out of which two are not published ) have comparable velocity coverage , spatial resolution and sensitivity to the new observations and were re - analyzed following the same calibration procedure described above ( see table [ datasets ] ) .
they were included because we were interested in searching for the newly discovered masers in the older data sets ( sect .
[ resmas ] ) , and re - analyzed because such a search would be easier when using the same ( self - calibration ) reference peak .
since the masers typically are as narrow as a couple ^-1 kms@xmath1 , all masers in these archival data sets should have been re - detected in the new data set if the masers would not vary in flux .
all archival data sets are in b1950 coordinates , and the new ( 2005 january ) data set is in j2000 coordinates .
llrlcclcclclrlrll no & & & & & & & & @xmath13 & & v@xmath14 & & flux & & ref.@xmath15 + & & & & & & & & & & ( jy ) & & + 1 & & [email protected] & & 174226.21 & @xmath2290011.2 & & 174536.96 & @xmath2290120.9 & & 4.2 & & @xmath2104 & & 0.11 & & + 2 & & [email protected] & & 174226.53 & @xmath2290012.7 & & 174537.28 & @xmath2290122.4 & & 2.1 & & @xmath2117 & & 0.05 & & + 3 & & [email protected] & & 174227.70 & @xmath2290007.7 & & 174538.45 & @xmath2290117.4 & & 2.1 & & @xmath2141 & & 0.04 & & + 4@xmath16 & & [email protected] & & 174228.02 & @xmath2290005.8 & & 174538.76 & @xmath2290115.5 & & @xmath178.5@xmath16 & & @xmath2132 & & 0.07 & & 1,2 + 5 & & [email protected] & & 174228.10 & @xmath2285833.4 & & 174538.80 & @xmath2285943.0 & & 4.2 & & 42 & & 0.62 & & 1,3 + 6 & & [email protected] & & 174229.02 & @xmath2285737.0 & & 174539.71 & @xmath2285846.6 & & 2.1 & & 100 & & 0.04 & & + 7@xmath16 & & [email protected] & & 174229.67 & @xmath2285832.2 & & 174540.37 & @xmath2285941.7 & & @xmath178.5@xmath16 & & 132 & & 0.29 & & 1,3 + 8 & & [email protected] & & 174229.71 & @xmath2285811.6 & & 174540.40 & @xmath2285921.1 & & 4.2 & & 32 & & 0.15 & & 1 + 9@xmath16 & & [email protected] & & 174229.91 & @xmath2285834.4 & & 174540.62 & @xmath2285944.0 & & @xmath178.5@xmath16 & & 132 & & 0.66 & & 1,3 + 10 & & [email protected] & & 174230.81 & @xmath2290038.5 & & 174541.57 & @xmath2290147.9 & & 6.3 & & 55 & & 0.51 & & 1,3 + 11 & & [email protected] & & 174231.41 & @xmath2285800.1 & & 174542.11 & @xmath2285909.5 & & 2.1 & & 34 & & 0.03 & & + 12@xmath18 & & [email protected] & & 174232.04 & @xmath2290032.1 & & 174542.80 & @xmath2290141.4 & & 4.2 & & 62 & & 0.10 & & 1,3 + 13@xmath18 & & [email protected] & & 174232.05 & @xmath2290033.2 & & 174542.80 & @xmath2290142.6 & & 2.1 & & 49 & & 0.09 & & 1,3 + 14@xmath18 & & [email protected] & & 174232.28 & @xmath2290028.5 & & 174543.04 & @xmath2290137.9 & & 2.1 & & 60 & & 0.10 & & 1,3 + 15@xmath18 & & [email protected] & & 174232.51 & @xmath2290026.0 & & 174543.26 & @xmath2290135.3 & & 4.2 & & 53 & & 0.40 & & 1,3 + 16@xmath18 & & [email protected] & & 174232.53 & @xmath2290023.4 & & 174543.28 & @xmath2290132.7 & & 4.2 & & 62 & & 0.24 & & 1,3 + 17@xmath18 & & [email protected] & & 174232.69 & @xmath2290022.9 & & 174543.45 & @xmath2290132.2 & & 4.2 & & 57 & & 0.60 & & 1,3 + 18 & & [email protected] & & 174233.57 & @xmath2290010.0 & & 174544.32 & @xmath2290119.3 & & 6.3 & & 66 & & 6.07 & & 1,3 + 19@xmath16 & & [email protected] & & 174234.88 & @xmath2285828.0 & & 174545.58 & @xmath2285937.2 & & @xmath178.5@xmath16 & & 39 & & 0.08 & & 1 + 20 & & [email protected] & & 174237.84 & @xmath2285831.4 & & 174548.55 & @xmath2285940.4 & & 2.1 & & 42 & & 0.03 & & + 21 & & [email protected] & & 174237.89 & @xmath2285816.6 & & 174548.59 & @xmath2285925.5 & & 4.2 & & 45 & & 0.10 & & + 22 & & [email protected] & & 174237.99 & @xmath2285916.8 & & 174548.72 & @xmath2290025.7 & & 2.1 & & 60 & & 0.03 & & + 23 & & [email protected] & & 174239.00 & @xmath2285833.6 & & 174549.71 & @xmath2285942.4 & & 4.2 & & 68 & & 0.18 & & + 24 & & [email protected] & & 174239.07 & @xmath2285844.8 & & 174549.78 & @xmath2285953.7 & & 2.1 & & 64 & & 0.08 & & + 25 & & [email protected] & & 174239.31 & @xmath2285810.8 & & 174550.01 & @xmath2285919.6 & & 2.1 & & 53 & & 0.06 & & + 26 & & [email protected] & & 174239.41 & @xmath2285820.8 & & 174550.11 & @xmath2285929.7 & & 2.1 & & 55 & & 0.04 & & 1 + 27 & & [email protected] & & 174239.52 & @xmath2285849.6 & & 174550.24 & @xmath2285958.4 & & 2.1 & & 62 & & 0.06 & & + 28 & & [email protected] & & 174239.78 & @xmath2285821.7 & & 174550.49 & @xmath2285930.5 & & 2.1 & & 60 & & 0.05 & & + + + + + [ masers ]
the observational results can be divided in two parts : point - like 1720mhz ( maser ) emission and extended 1720mhz absorption , presented in turn below . to unambiguously exclude noise spikes as false maser detections , we consider a maser to be detected only if it has a channel flux density larger than 10 times the 1@xmath19 rms noise level in its data set . with this strict constraint ,
we detect 13 new and 13 previously published masers in the 2005 data .
one of the newly discovered masers was also detected in the 1998 data ( maser # 1 in table [ masers ] at 11@xmath19 ) but no other new masers were found above 10@xmath19 in the archival vla data . apart from the general possibility of intrinsic variability ( which exists : sect .
[ cndmasers ] ) , this is most likely due to the much larger channel widths ( 8.5 ^-1 kms@xmath1 ) or the much larger rms noise in the archival data sets . for example , assuming no variability , the brightest new single - channel detection ( maser # 24 at 25@xmath19 ) would be smeared to less than 7@xmath19 rms in a 8.5 ^-1 kms@xmath1channel in the 1986 june data set and thus would remain undetected using our strict 10@xmath19 constraint . actually , knowing the positions and velocities of the newly detected masers , we found most of the masers # 2028 at about 4@xmath196@xmath19 levels back in both the 1986 and 1998 data ; the 1996 data was too noisy
. also maser # 6 was found in the archival data , but at 7@xmath19 in the 1998 data only ; it had not appeared in 1986 and apparently weakened since 1998 .
two previous detections ( masers # 4 and # 19 in table [ masers ] ) were not detected ( down to 4@xmath19 ) in the 2005 data but are included in table [ masers ] for the completeness of the discussion in this paper .
table [ masers ] lists the oh name in galactic coordinates , the position of the maser , the total width in ^-1 kms@xmath1 of the channels in which emission occurs , the velocity of the channel containing the peak flux , the total detected flux integrated over the velocity width , and finally references to the papers where previous detections are reported . for the masers detected multiple times , we have adopted the parameters derived from the most recent 2005 dataset , or from k03 if the maser was confused or undetected in 2005 ( for confusion see * ? ? ?
small positional offsets compared to earlier papers are due to differences in the self - calibration .
velocity offsets and flux differences are due to different observing parameters per data set and possible intrinsic variability of the maser . because the maser emission is generally found only in 1 - 2 channels and may be non - gaussian and narrower than the channel separation , spectral fitting to the feature to obtain a fitted flux , center velocity , fwhm and their associated errors is meaningless .
we therefore assume an uncertainty in the center velocity and velocity range of half the channel separation ( i.e. , 1 ^-1 kms@xmath1 except for the ones adopted from k03 where it is about 4 ^-1 kms@xmath1 ) , and up to 50% uncertainty in the flux .
the flux values given should therefore be regarded as indicative for their relative intensity only .
good fits and associated errors can be obtained with new high spectral resolution observations , e.g. as in @xcite , but this is beyond the scope of this paper .
figure [ maserpos ] shows the masers from table [ masers ] labeled with their velocities on top of a gray scale 1.7ghz radio continuum image made from the frequency averaged 2005 data .
two different populations of masers can be identified : a ) masers with velocities between @xmath030 and @xmath070 ^-1
kms@xmath1 following the outline of the sgraeast snr shell ( `` snr masers '' , marked with squares , crosses and pluses ) , and b ) masers with absolute velocities @xmath20 100 ^-1 kms@xmath1which are consistent with velocities of the cnd at their projected position ( `` cnd masers '' , circles ) .
note that the latter are located in projection , not only on top and near the cnd , but also on top and near the sgraeast snr shell , albeit with a distinct high velocity .
one additional new maser ( at @xmath0100 ^-1 kms@xmath1 ) seems distinct from the two main populations ( diamond , sect .
[ snrnw ] ) .
we find a significant number ( 3 ) of new high - negative velocity masers toward the southwestern part of the cnd ( circles in fig . 1 ) , but
no new high - positive velocity masers .
the cnd masers are only found near the tangent points of the torus - like and inclined cnd , suggesting that the masers are found only there due to the longer path - length ( * ? ? ?
* ; * ? ? ?
* see sect . [ pump ] ) .
furthermore , almost all new snr masers are found toward the northeastern rim of the snr shell , in projection close to the regions a d ( that lie in front of the 50 ^-1 kms@xmath1 cloud ) , and close to where the deepest 1667 mhz oh absorption occurs ( k03 ) .
as these masers are much weaker than the masers toward [email protected] in the southeast ( e.g. , yz96 ) , they have not been detected in previous observations other than a single one in k03 .
it is in this region that @xcite and , more recently , @xcite and @xcite have detected signs of shock interactions using observations of nh@xmath21 , h@xmath11 and cs emission ( sect .
[ snrmasers ] ) .
although individual clumps and streamers do confuse and complicate the overall interpretation , our 1720mhz oh absorption maps are completely consistent with the conclusions drawn by k03 and others ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
that is , _
i ) _ part of the distribution of the @xmath020 and @xmath050 ^-1 kms@xmath1 clouds is located _ in front of _ the sgraeast snr shell , _ ii ) _ the cnd must be located on the near side of sgraeast , and , _ iii ) _ the @xmath020 ^-1 kms@xmath1 and @xmath050 ^-1 kms@xmath1 clouds must be _ mostly behind _ sgra * and the minispiral , thus behind the cnd .
the large scale absorption distribution illustrates conclusion _ i ) _ above .
figure [ absorption ] plots selected velocity maps of the 1720mhz oh absorption superimposed on the radio continuum where each panel displays spectrally smoothed absorption with a width of 21.7 ^-1 kms@xmath1 .
it resembles the on - line material by k03 ( their fig .
4 ) , but is included here as it shows the large extent of the 1720mhz absorption more clearly than in k03 .
figure [ absorption ] demonstrates that the large scale absorption at @xmath058 to @xmath036 ^-1 kms@xmath1occurs in the eastern part of the snr shell , while absorption at lower velocities ( @xmath036 to @xmath014 ^-1 kms@xmath1 ) also covers the western part of the shell , consistent with _
further , the k03 data as well as our absorption data ( not shown here ; see @xcite ) reveal the cnd in absorption at high absolute velocities ( @xmath22 100150 ^-1 kms@xmath1 ) at opposing azimuth angles of the cnd major axis , thus concluding _
ii)_. finally , our data confirm that the minispiral and sgra * are devoid of absorption .
we refer to k03 ( their sect .
3.1.1 ) for discussing the lack of absorption in detail , but note that our results support that the general lack of absorption toward sgra * and the minispiral therefore imply _
the discussion concentrates on the nature of the 1720mhz maser emission , detected at multiple spatial locations in the gc region .
however , the absorption measurements play an important role in untangling the 3d structure of the region and the origin of the different groups of masers observed .
our absorption measurements corroborate the absorption results and conclusions of k03 : parts of the molecular clouds must be _ between _ the cnd in the front , and the snr in the back ( sect .
[ resabs ] , sect .
[ cndmasers ] ) .
previously , the location of the components in the line - of - sight toward the sgra complex have been subject to different interpretations , in particular the line - of - sight location of the cnd , the molecular cloud material and the sgraeast snr .
many authors ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) have tried to picture a three - dimensional ( 3d ) model of the different components and structures .
the latest is presented by @xcite , who prefer a model where the sgraeast snr directly pushes against the cnd .
a slab of compressed molecular ( or atomic ) material might be a part of this interaction , but would not necessarily separate the snr from the cnd physically .
this model is mostly based on larger velocity dispersions and lower velocity centroids in selected h@xmath11 emission slit observations , as compared to what is observed for the nh@xmath21 density clumps ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
however , this does not directly imply that the snr pushes the cnd toward the observer , as indicated by @xcite .
it does indeed mean that the h@xmath11 gas is hotter , but it might as well be part of a different kinematic structure . since @xcite , it is clear that with the many observations and interpretations of the sgraeast complex using different tracers and methods over the past @xmath2240 years , a full review with perhaps more sensitive ( re)observations and/or minute re - interpretations of the available data and facts is needed to explain the structure . that is beyond the scope of this paper , as here
we just want to explain the nature of the 1720mhz oh masers , but we note that the full 3d structure still is not completely clear . a 3d model should also be based on information from absorption lines which yields direct information of the line - of - sight location of the absorbing gas with respect to the background continuum .
we therefore endorse the result of k03 , that the sgraeast snr and the cnd do not necessarily physically interact , since part of the molecular material lies between them .
maser emission is primarily constrained by the narrow range of physical parameters required ( sect .
[ pump ] ) . the amplification and population inversion of the 1720mhz
oh line requires number densities of the order of @xmath23 10@xmath24 @xmath25 , temperatures in the range @xmath26 = 50125 k and oh column densities of @xmath27 10@xmath2810@xmath29 @xmath30 @xcite .
such conditions can be found in sfrs @xcite and in the post - shock regions near snrs @xcite .
modeling of snr / ism interactions show that c - type shocks provide the required inverted oh column densities . however , in this paper we argue that some masers arise in non - standard locations , i.e. , in the cnd ( sect .
[ cndmasers ] ) . an independent check on whether this is reasonable can be made by estimating the oh column density in regions where masers occur , using the absorption data . in local thermal equilibrium ( lte )
@xmath31 where @xmath32 is the excitation temperature in k , @xmath33 is the fwhm line width in ^-1 kms@xmath1 and @xmath34 is the 1720mhz peak optical depth .
the opacity is dependent on the absorbed flux density as well as the flux density of the background continuum . since it is difficult to properly separate the different continuum components ( snr shell , thermal emission from the ionized gas and non - thermal emission from sgra * ) and their location relative to the absorbing gas , the first order estimates given below are apparent opacities given for the assumed , simplest case with all absorbing gas in front of the continuum ( see k03 for @xmath35 maps ) .
another caveat is , of course , that the oh is not likely to obey lte conditions in these regions . to estimate a typical column density toward the snr shell ,
a ( northeastern ) region centered on @xmath36ra = 75 and @xmath36dec = 104 was chosen . here
, the absorption feature has a velocity centroid close to @xmath050 ^-1 kms@xmath1 thus corresponding to gas in which the snr / ism masers arise
. a single - component gaussian fit yields an opacity of [email protected] , @xmath38 15@xmath373 ^-1 kms@xmath1 and an oh column density @xmath39 [email protected]@xmath510@xmath40 @xmath30 , which are typical values in other regions toward the sgraeast shell .
similarly , the oh column density of the cnd is estimated at a position centered at @xmath36ra = @xmath218 and @xmath36dec = 8 . to ensure the cnd opacity
is measured excluding absorption from the molecular belt , a line with a centroid velocity of 98 ^-1 kms@xmath1 was used , resulting in @xmath38 31@xmath373 ^-1 kms@xmath1 , an opacity of [email protected] and an oh column density of @xmath39 [email protected]@xmath510@xmath40 @xmath30 .
compared to the snr / ism interaction sites , the line widths appear broader in the cnd , or alternatively consist of a larger number of overlapping components in velocity .
overall , we deduce that the estimated oh column densities in the cnd and in the snr / ism are similar , providing similar column density conditions .
the expected temperatures of 50 - 125 k can be combined with our estimate of the column density in the cnd , @xmath41 2@xmath510@xmath40 @xmath30 ( sect . [ resabs ] ) .
the resulting observed column density is an order of magnitude larger than the @xmath27 10@xmath2810@xmath29 @xmath30 predicted by 1720mhz maser theories for regions with number densities of 10@xmath24 @xmath25 .
however , this is not necessarily a contradiction , since the masers only occur in regions where the number density is constrained to @xmath23 10@xmath24 @xmath25 . in regions with number densities above this value ( reflected in an increased column density ) maser emission will be quenched , perhaps in favor of 1612mhz emission @xcite , and the gas will be observed in absorption instead . we note that @xcite do comment on 1612mhz emission toward sgraeast , but none is detected toward the cnd .
this subsection discusses the snr masers found in the @xmath030 to @xmath070 ^-1 kms@xmath1 velocity range ; first the new and previously known masers in the northwest ( crosses in fig . [ maserpos ] ) , then the newly found masers in the northeast ( squares ) , and finally the previously known masers in the southeast ( plusses ) .
the masers to the northwest , north of the cnd , appear related to linear filaments traced by h@xmath11 emission ( usually presumed to trace shocked , dense gas with @xmath42 @xmath25 ) .
the positional coincidence of masers and h@xmath11 filaments was first noted by @xcite ( 2001 , hereafter yz01 ) , who used nicmos observations to compare the position of the @xmath042 ^-1 kms@xmath1maser ( # 5 , yz96 ) with the distribution of h@xmath11 line emission .
the maser was found co - located with an h@xmath11 filament labeled the linear filament . in fig .
[ ellipse ] we overplot the same h@xmath11 map with all masers .
we note that the @xmath032 ^-1 kms@xmath1 maser ( # 8 , k03 ) is projected along the same linear filament as maser # 5 , and that the new @xmath0100 ^-1 kms@xmath1 maser ( # 6 ) is located just north of another filament , called the outer filament . the position and velocity of the @xmath0100 ^-1 kms@xmath1maser may indicate that it is related to the @xmath070 ^-1 kms@xmath1 cloud @xcite or to the nw extended dark cloud , but no h@xmath11 velocity information is available for this cloud for comparison ( yz01 ) .
yz01 suggest that the h@xmath11 filaments could be generated by the impact of sgraeast into the backside of the cnd . as a note of caution ,
they also point out that this region is confused by the presence of the @xmath070 ^-1 kms@xmath1 cloud and the nw extended dark cloud , either of which may be falling into the cnd from the foreground ( @xcite ; yz01 ; k03 ) .
yz01 base their interpretation on a spatial alignment of the linear filament with the western boundary of the snr radio continuum ridge .
a similar alignment can , however , also be argued with the nw extended dark cloud .
thus , the suggestion that the h@xmath11 filaments are produced by a direct impact between sgraeast and the cnd seems viable but , at this point , lacks unambiguous evidence . to explain the appearance of the h@xmath11 filaments across the whole eastern part ( including the linear filament , the outer h@xmath11 clumps , and the filament west of the southern lobe ) in connection to the masers in this region
, we postulate an alternative model where the snr triggers a shock in something much larger than the cnd .
this model is depicted in the lower right inset of fig .
[ ellipse ] . such an arc - like interaction model ( outlined by the elliptical band in fig .
[ ellipse ] ) would argue in favor of an impact of sgraeast into more widespread molecular material , located between the snr and the cnd .
this interaction could , but would not necessarily have to shock the cnd . as mentioned before , line emission observations ( whether from h@xmath11 or nh@xmath21 gas , or oh masers ) alone
do not allow a proper line - of - sight determination of the different components we are looking forward to e.g. absorption and/or extinction measurements toward these h@xmath11 filaments and h@xmath11 clumps .
for now , we can state that the h@xmath11 emission filaments are consistent with _ any _ of these interpretations : _ a ) _ sgraeast is interacting with the cnd , _
b ) _ sgraeast is interacting with the @xmath020/50 ^-1 kms@xmath1 cloud in between the snr and the cnd , or _
c ) _ the h@xmath11 filaments are not due to an interaction with sgraeast but due to an interaction with the nw extended dark cloud or the @xmath070 ^-1 kms@xmath1 cloud . nevertheless , with the absorption data available , and a model as shown in fig .
[ ellipse ] ( lower right panel ) , we favor alternative _ b ) _ where these northwestern snr masers are generated in the molecular cloud material at the front ( near ) side of the snr . we can not determine whether the shock does , or does not penetrate the cnd from the back .
one maser excepted , the group of weaker masers in the northeastern part of sgraeast are all new detections .
these masers appear situated near the well known ultra compact regions located farther east of the snr ( a , b , c , & d ; @xcite ) .
therefore , these northeastern masers could be associated with star formation , similar to what is observed in sfrs @xcite .
relative to the snr however , the regions are in the foreground ( k03 ) .
furthermore , the spatial spread and velocity distribution of the masers follow the `` backward c '' shape observed in nh@xmath21 by @xcite with the most red - shifted maser velocities in the center of the distribution .
this combination thus depicts the shell - like interaction of the back velocity signature would have a regular c shape @xcite , this would have been the front side , propagating toward the observer . ]
side of the sgraeast snr plowing into the @xmath050 ^-1 kms@xmath1 cloud and propagating away from the observer .
this interaction and morphology has been described previously by many authors and indeed also implies that part of the @xmath050 ^-1 kms@xmath1 cloud is behind sgraeast ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this part of the @xmath050 ^-1 kms@xmath1 cloud behind the snr has also been referred to as the ( eastern part of the ) sgraeast core @xcite .
the detection of the 1720mhz oh masers independently confirms that there is such an interaction region and that the excitation mechanism for the observed h@xmath11 emission is indeed due to a c - type shocked front @xcite .
unfortunately , h@xmath11 filaments can not be recognized in their slit observations ( sect .
[ snrnw ] ) . in summary
, we conclude that the northeastern snr masers are generated in the molecular cloud material at the back ( far ) side of the snr .
the previously reported bright snr masers were all detected in our observations ( yz96 ; yz01 ; k03 ) .
those are the bright masers near the southeastern edge of sgraeast ( fig .
[ maserpos ] ) .
the near - linear distribution of the masers in the southeast differs from the more spread - out spatial distribution of the masers to the northeast ( sect .
[ snrne ] ) . in the southeast corner of fig .
[ maserpos ] , an @xmath2280-radius semicircular weak continuum feature can be discerned .
this continuum emission outlines another snr shell , [email protected] ( e.g. , * ? ? ?
* ; * ? ? ?
@xcite have detected non - thermal x - ray emission from parts of the [email protected] shell , with a spectrum that suffers large absorption consequently placing the x - ray source in the gc . in the sky
, the [email protected] snr shell overlaps the sgraeast continuum , and is likely responsible for the inward - concave morphology of the southeastern sgraeast shell @xcite .
this is precisely where the bright 1720mhz masers form .
in contrast to the northwestern masers , the nicmos image only shows weak extended h@xmath11 emission , no filaments ( fig .
[ ellipse ] ; yz01 ) .
moreover , this region displays almost no continuum background and oh absorption ( fig . [ absorption ] ; fig .
[ ellipse ] ; k03 ) .
this region likely is a result of two colliding shock fronts and would explain why there is such a sharp line of bright masers .
previous discussions of these masers suggested that they are the result of the interaction of the @xmath050 ^-1 kms@xmath1 cloud and sgraeast .
we here suggest that they probably are due to a snr / snr interaction ( albeit close to the front - side snr / ism interaction in fig .
[ ellipse ] ) , for which 1720mhz maser emission has not been anticipated before .
we here discuss the masers near the tangent points of the cnd with high absolute ( @xmath37104141 ^-1 kms@xmath1 ) velocities ( circles in fig .
[ maserpos ] ) .
a bright maser at @xmath0132 ^-1 kms@xmath1 has been detected in all observations ( i.e. , since 1986 and beyond 2005 ) in the northern part of the disk ( yz96 ; k03 ; this work ) .
previously , only in 1986 a faint , conjugate @xmath43132 ^-1
kms@xmath1 maser was detected in the southern part of the cnd ( k03 and references therein ) which disappeared before the next possible detection in 1996 .
up until now , this single detection has been the only observation suggesting that some 1720mhz masers are distributed symmetrically in position and velocity with respect to sgra * , like the cnd , in sharp contrast to the snr / ism interaction at velocities between 30 and 70 ^-1 kms@xmath1 as discussed above .
the three new detections of high - negative velocity 1720mhz masers toward the southwestern part of the cnd ( table [ masers ] ) now strongly support the existence of conjugate masers with locations and velocities consistent with originating from gas in the cnd . clearly these masers are variable as they may appear and/or disappear on time scales on the order of years ( sect .
[ resmas ] ) .
the estimated oh column density derived above appears similar across the total extent of the cnd , indicating that maser emission potentially could be observed at all azimuthal angles of the cnd .
this in particular would be the case if sgraeast would continuously drive planar shocks from behind into the cnd while the cnd is pushed toward the observer as argued by yz01 and @xcite data do not properly cover the cnd . ] .
in such a case , one would expect to find some masers originating in the @xmath020 ^-1 kms@xmath1 cloud , i.e. , with velocities in the @xmath22 1050 ^-1 kms@xmath1range , in projection _ toward the whole cnd_. though there are three masers in this velocity range north of the cnd ( and associated with h@xmath11 filaments ) , there is no distribution toward the rest of the cnd .
furthermore , a push from behind would result in a relatively constant pumping of the masers , which due to the clear variability of the southwestern ( high - negative velocity ) cnd masers is hard to support .
in contrast , the oh masers in the cnd appear only in two regions : slightly east of north and west of south at positions close to the geometrical tangent points .
extended h@xmath11 emission is found nearby those regions , but not distributed in filaments as for the northwestern masers ( sect .
[ snrnw ] ) .
this h@xmath11 emission is due to both limb brightening effects as well as dissipation of shocks in clump - clump interactions ( yz01 ) . for the case of masers in a disk or ring structure
, this geometry can be understood by the requirement of having long paths of velocity coherent gas in order to build up a large amplification .
closer to the tangent points of a disk , the path - lengths are longer resulting in a larger amplification @xcite .
such a behavior has previously been observed in maser transitions of h@xmath11o and oh in extragalactic sources @xcite .
it is therefore plausible that the cnd masers can be sustained without being a direct result from shocks generated by sgraeast ( sect .
[ pump ] ) . whereas the pumping of the snr masers can be explained by the interaction of the snr shock
plowing into the interstellar material of the @xmath050 ^-1 kms@xmath1 cloud , it is not required that the snr and the cnd directly interact to form the cnd masers . a radiative ( far - infrared ) pump source primarily tends to invert the 1612mhz line , while collisional excitation has been identified as the prime pumping mechanism for 1720mhz oh masers @xcite .
the lack of interstellar oh 1612mhz and mainline emission in the cnd ( k03 ) argues against far - infrared pumping and for a collisional pumping scheme of the 1720mhz cnd masers . in the cnd
we find extended shock excited h@xmath11 emission in the `` lobes '' ( yz01 ) , but not in large - scale shock - front filamentary structure as for the northwestern masers .
the clumpiness and irregular distribution of molecular gas in the cnd imply the possibility of frequent clump - clump collisions which could provide the source of a collisional maser pump .
the dense ( @xmath44 34@xmath510@xmath45 @xmath25 ) molecular clumps defined by hcn and hco@xmath46 observations are found more or less in the same locations where bright h@xmath11 emission has been mapped ( @xcite ; yz01 ) .
yz01 use the h@xmath11 line emission associated with the cnd to argue that dissipation of the random motions of molecular clumps is the most likely cause of the excitation of the h@xmath11 molecules via shocks , implying the presence of c - type shocks within the cnd .
similar to the c - shock chemistry predicted for snr / ism masers , the post - shock regions in the cnd should produce suitable conditions for regions of enhanced oh abundances @xcite . comparing the maser positions with respect to the h@xmath11 ( yz01 ) and hcn emission @xcite and assuming that the cnd masers are excited via shocks , it is not surprising to see a correlation between the maser emission and the h@xmath11 emission .
it is interesting to note that the masers always appear to be offset from , and trailing the brightest hcn and h@xmath11 peaks . at the high densities ( @xmath47 @xmath25 ) traced by hcn
, maser emission will be quenched .
instead , the masers occur in lower density post - shock regions following / trailing the clumps .
the large dispersion ( @xmath22 40 ^-1 kms@xmath1 ) in the southwestern cnd masers may reflect the dispersion of individual clumps in the cnd .
a clumpy medium also appears to be a better maser amplifier than a smooth medium .
extragalactic oh masers have successfully been modeled assuming a clumpy medium @xcite .
modeling of masers in a clumpy medium predicts high maser variability caused by smaller regions of inverted gas moving in and out of the line - of - sight .
however , no clear limits could be set on the variability of individual maser features presented in this paper as each observation so far has very different observing parameters ( table [ datasets ] ) .
a forthcoming paper will properly address the maser variability using homogeneous data sets .
the velocities and locations of most 1720mhz oh masers observed in the gc agree with the commonly adopted model in which the masers arise in regions where the supernova remnant sgraeast is interacting with the @xmath020 and @xmath050 ^-1 kms@xmath1 molecular clouds and the nearby snr [email protected] .
in addition to extending the database of such masers , in particular toward the interaction region toward the northeast rim of sgraeast , we have explored the slightly different possible origins of the maser groups , notably the masers northwest of the cnd .
we showed that they indicate a region of shock interaction of the near side of sgraeast with molecular cloud material extending well over the projected size of the cnd .
the newly found masers in the northeast indicate such an interaction on the far side of sgraeast . furthermore , we have verified the existence of high - negative velocity masers , which together with previously detected high - positive velocity masers bracket sgra * both in position and velocity .
their positions and velocities are consistent with being located near the tangent points of the cnd .
however , we find no need for the snr to interact directly with the cnd to pump and sustain these high - velocity masers .
oh absorption data show that the cnd and sgraeast are separated by molecular cloud material , and therefore the snr shock is unlikely to be responsible for pumping these high - velocity masers .
instead a more likely pumping mechanism is by collisions from dissipation between clumps _ in _ the cnd .
these results imply that 1720mhz oh masers _ not only _ do occur in sfrs and in snr / ism ( and snr / snr ) interactions , but can also arise in clumpy and disturbed circumnuclear disks at small radii .
if sufficiently bright , such 1720mhz masers could serve as an additional molecular line tool to study circumnuclear gas dynamics around galaxy centers .
we would like to thank m. elitzur and v. fish for their most useful comments on an earlier version of the manuscript , and also s. stolovy and m. christopher for kindly sharing their h@xmath11 and hcn maps for reference .
the very large array is operated by the national radio astronomy observatory , which is a facility of the national science foundation operated under cooperative agreement by associated universities , inc . | we present very large array ( vla ) radio interferometry observations of the 1720mhz oh masers in the galactic center ( gc ) .
most 1720mhz oh masers arise in regions where the supernova remnant sgraeast is interacting with the interstellar medium .
the majority of the newly found 1720mhz oh masers are located to the northeast , independently indicating and confirming an area of shock interaction with the @xmath050 ^-1 kms@xmath1 molecular cloud ( [email protected]@xmath20.07 ) on the _ far _ side of sgraeast .
the previously known bright masers in the southeast are suggested to be the result of the interaction between two supernova remnants , instead of between sgraeast and the surrounding molecular clouds as generally found elsewhere in the galaxy .
together with masers north of the circumnuclear disk ( cnd ) they outline an interaction on the _ near _ side of sgraeast .
in contrast to the interaction between the @xmath050 ^-1 kms@xmath1 cloud and sgraeast , oh absorption data do not support a direct interaction between the cnd material and sgraeast .
we also present three new high - negative velocity masers , supporting a previous single detection .
the location and velocities of the high - negative and high - positive velocity masers are consistent with being near the tangent points of , and physically located _ in _ the cnd .
we argue that the high velocity masers in the cnd are pumped by dissipation between density clumps in the cnd instead of a shock generated by the supernova remnant .
that is , the cnd masers are not coupled to the supernova remnant and are sustained independently . |
piston configurations have received considerable interest because they allow for an unambiguous prediction of casimir forces @xcite .
the force depends on the boundary conditions imposed , the cross section of the piston and on properties of additional kaluza - klein dimensions that might be present@xcite . in this contribution
we consider the same geometrical setting as in refs.@xcite , but with fractional boundary conditions imposed .
this is an extension of ref .
@xcite to the case where additional dimensions are considered .
consider a massive scalar field @xmath1 of mass @xmath2 in the background spacetime @xmath3 , where @xmath4 is the @xmath5-dimensional minkowski spacetime and @xmath6 is a compact @xmath7-dimensional internal manifold .
we want to find the casimir force acting between two parallel plates with arbitrary shape in a cylinder of constant simply connected cross section @xmath8 .
to interpolate between dirichlet and neumann boundary conditions , we consider fractional neumann conditions @xmath9,\ ] ] on the plates .
it can be shown that the casimir force only depends on the difference @xmath10 . therefore without loss of generality , we fix dirichlet boundary conditions on the left plate and impose fractional neumann conditions of order @xmath11 on the right plate . for concreteness , on the walls of the cylinder dirichlet boundary conditions are imposed . using the piston approach@xcite
, we first consider the casimir force acting on a piston moving freely inside a closed cylinder @xmath12\times \omega \times \mathcal{n}$ ] , and let the right end of the cylinder go to infinity , i.e. , @xmath13 .
let @xmath14 denotes the position of the piston .
the finite temperature casimir force acting on the piston is given by @xmath15where @xmath16 is the finite temperature casimir energy inside the closed cylinder @xmath17\times \omega \times \mathcal{n}$ ] ; formally @xmath18 the eigenfrequencies of the field @xmath1 which satisfies dirichlet boundary conditions on the plate at @xmath19 and the wall @xmath20\times \partial\omega\times\mathcal{n}$ ] of the cylinder , and fractional neumann conditions of order @xmath11 on the plate at @xmath14 , are given by @xmath21 here , @xmath22 , @xmath23 , are the eigenvalues for the laplace operator with dirichlet boundary conditions on @xmath8 , @xmath24 , @xmath25 , are the eigenvalues of the laplace operator on @xmath26 , with @xmath27 . for generic @xmath28 , @xmath29 . for @xmath30 or 1 , @xmath31 . in the following
we shall assume @xmath28 ; for @xmath32 and @xmath33 the casimir force will have to be divided by two .
it is worth mentioning that if one considers a cylinder of length @xmath34 instead of @xmath35 and if one imposes twisted boundary conditions @xmath36 for a complex massive scalar field @xmath1 , one obtains the same set of eigenfrequencies . using the zeta regularization method to compute the casimir energy , we find that @xmath37where @xmath38 and @xmath39 are terms that are independent of @xmath35 .
@xmath40 does not contribute to the casimir force acting on the piston , and we find by taking the limit @xmath13 in that the casimir force acting between two parallel plates is @xmath41this expression shows that in the high temperature limit , the leading term of the casimir force is linear in @xmath42 given by the sum of the terms with @xmath43 , which is usually called the classical term .
for the low temperature behavior , some computations give ( we use the notation @xmath44 ) @xmath45 the first two terms are the zero temperature casimir force .
the sum of the last two terms is the thermal correction .
it goes to zero exponentially fast when @xmath46 . for a pair of dirichlet - dirichlet plates , i.e.
, @xmath30 , ( divided by two ) shows that the casimir force is always attractive ( negative ) . for dirichlet - neumann plates , i.e. , @xmath47 , ( divided by two )
shows that the casimir force is always repulsive .
in fact , this is true for all @xmath48 since the function @xmath49is positive for @xmath48 and @xmath50 . in all these cases
the force is enhanced by the presence of extra dimensions . for @xmath51
, one can show that when the plate separation is large enough , the casimir force eventually becomes attractive .
when the plate separation @xmath35 is much smaller than the size @xmath52 of the cross section , some computations show that the leading term of the casimir force density acting on the plates is @xmath53where @xmath54 . in the massless case
( @xmath55 ) , the sum of the terms with @xmath56 should be replaced by @xmath57 a form of more suitable to study the low temperature behavior is @xmath58 when @xmath55 , the sum of the terms with @xmath59 in should be replaced by @xmath60one can show that when @xmath61 , the massive correction to the casimir force is of order @xmath62
. therefore , and show that when @xmath61 and @xmath63 , where @xmath64 is the size of the internal manifold , the nature of the casimir force is determined by the sign of @xmath65with @xmath66 if @xmath67 and @xmath68 if @xmath69 . for any @xmath70 ,
the function @xmath71 $ ] is a monotonically decreasing function and it has a unique zero @xmath72 in @xmath73 $ ] .
one can show that @xmath74 .
therefore , we find that when @xmath75 , for @xmath76 , the casimir force can change from attractive to repulsive when one increases the temperature . on the other hand ,
when the plate separation @xmath35 becomes much smaller than the size of the internal manifold , then the internal manifold should be treated in the same way as the large cross section . in this case , we find that the sign of the casimir force is determined by @xmath77 with @xmath78 if @xmath67 and with @xmath79 if @xmath69 .
again , we see that decreasing the distance between the plates such that it is much smaller than the size of the internal manifold can change the casimir force from repulsive to attractive for some values of @xmath11 . 9 r.m .
cavalcanti , _ casimir force on a piston _ , phys .
d * 69 * ( 2004 ) , 065015 .
hertzberg , r.l .
jaffe , m. kardar and a. scardicchio , _ attractive casimir forces in a closed geometry _ , phys .
* 95 * ( 2005 ) , 250402 .
v. marachevsky , _
casimir interaction of two plates inside a cylinder _ , phys .
d * 75 * ( 2007 ) , 085019 .
teo , _ finite temperature casimir effect in kaluza - klein spacetime _ , nucl .
b * 819 * ( 2009 ) , 431452 .
k. kirsten and s.a .
fulling , _ kaluza - klein models as pistons _ , phys .
d * 79 * ( 2009 ) , 065019 .
s.c . lim and l.p .
teo , _ repulsive casimir force from fractional neumann boundary conditions _ ,
b * 679 * ( 2009 ) , 130137 .
e. elizalde , s.d .
odintsov and a.a .
saharian , _ repulsive casimir effect from extra dimensions and robin boundary conditions : from branes to pistons _ , phys .
d * 79 * ( 2009 ) , 065023 . | we consider the finite temperature casimir force acting on two parallel plates in a closed cylinder with the same cross section of arbitrary shape in the presence of extra dimensions .
dirichlet boundary conditions are imposed on one plate and fractional neumann conditions with order between zero ( dirichlet ) and one ( neumann ) are imposed on the other plate .
formulas for the casimir force show that it is always attractive for dirichlet boundary conditions , and is always repulsive when the fractional order is larger than @xmath0 . for some fractional orders
less than @xmath0 , the casimir force can be either attractive or repulsive depending on the size of the internal manifold and temperature . |
semileptonic and radiative beauty decays play an important role in the determination of the ckm matrix elements and in the indirect search for new physics : the former , tree - level dominated , allow for a precise measurement of the ckm elements @xmath1 and @xmath2 ; the latter , loop - induced , are directly sensitive to new physics contributions , and give also information on @xmath3 and @xmath4 .
these decay modes , all characterized by an electroweak current that probes the @xmath0 dynamics , have a lot in common and form a remarkable set of interdependent measurements .
their simplicity , however , is only apparent : if one is interested in precision measurements they display all the rich complexity specific of qcd dynamics .
the main theoretical divide runs between inclusive and exclusive decays : inclusive decays can be studied using an operator product expansion ( ope ) , parameterizing the non - perturbative physics in terms of @xmath0-meson matrix elements of power - suppressed local operators , while exclusive decays require an estimate of the form factors using non - perturbative methods ( lattice qcd , sum rules ) .
having two completely different methods at our disposal represents a huge advantage and makes non - trivial cross - checks possible .
see @xcite for a review .
the exclusive determination of @xmath5 employs the extrapolation of the @xmath6 rate to the kinematic endpoint where the @xmath7 is produced at rest ( zero - recoil ) . in this limit ,
the form factor @xmath8 is known , up to corrections suppressed by at least two powers of @xmath9 that have to be computed , e.g. on the lattice .
since one needs to estimate only the @xmath10 correction to the heavy quark limit , a good accuracy can be reached even with present non - perturbative methods .
in fact , current lattice qcd and sum rule results are both consistent with @xmath11 @xcite .
the overall uncertainty is therefore close to 5% : @xmath12 , but the two most precise experimental results , by babar and cleo , differ by almost 3@xmath13 @xcite .
semileptonic decays to @xmath14 mesons give consistent but less precise results .
progress is expected especially from unquenched lattice calculations .
while the non - perturbative unknowns in the exclusive determination of @xmath5 have to be calculated , those entering the inclusive semileptonic decay , @xmath15 , can be measured in a self - consistent way .
indeed , the inclusive decay rate depends only on the hadronic structure of the decaying @xmath0 meson , but the sensitivity to it is actually suppressed by two powers of @xmath16 , as the highly energetic decay products are ( generally ) unable to probe the long wavelengths characteristic of the @xmath0 meson .
formally , an ope allows us to write the differential @xmath15 rate as a double series in @xmath17 and @xmath16 , whose leading term is nothing but the parton model result .
however , the ope results for the spectra can be compared to experiment only after smearing over a range of energies @xmath18 and away from the endpoints .
the hadronic mass spectrum , for instance , is dominated by resonance peaks that have no counterpart in the ope : the ope results have no _ local _ meaning .
the observables that can be studied in the ope include the total rate and moments ( weighted integrals ) of the lepton energy and hadronic mass spectra , as well as the photon spectrum in radiative decays .
they generally are subject to a lower cut on the charged lepton energy and can be written in a way analogous to that of the integrated rate , @xmath19,\nonumber\end{aligned}\ ] ] where @xmath20 , @xmath21 is the tree - level expression , @xmath22 contains the leading electroweak corrections , the wilson coefficients @xmath23 are series in @xmath17 , and power corrections up to @xmath24 have been kept .
the parameters entering the predictions are @xmath17 , properly defined _ quark _ masses @xmath25 , and the @xmath0 meson matrix elements of the four _ local _ operators that appear up to @xmath26 : @xmath27 and @xmath28 . because they depend on the various parameters in different ways , the moments serve a double purpose : they allow to constrain the non - perturbative parameters and they test the overall consistency of the ope framework .
effects that can not be described by the ope ( and therefore violate _ parton - hadron duality _ ) and higher order power corrections can be severely constrained .
recent experimental results @xcite , analyzed in the light of up - to - date theoretical predictions @xcite , have led to a big step forward , both in completeness and accuracy .
there is a remarkable consistency of a variety of leptonic and hadronic moments , leading to an excellent fit .
the values of the quark masses are in agreement with lattice and spectral sum rule determinations , and the other non - perturbative parameters are determined for the first time at the 10 - 20% level , in agreement with theoretical expectations .
radiative moments from belle , cleo , and babar @xcite can be included as well without deteriorating the quality of the fit , that yields @xmath29 @xcite , in agreement with the exclusive result .
the estimate of the theory error ( missing higher order perturbative and non - perturbative contributions , intrinsic charm etc . ) is particularly delicate ; different recipes in different schemes @xcite have led to compatible results . despite recent progress @xcite , higher order perturbative corrections to the wilson coefficients
are the main source of uncertainty : a 1% determination of @xmath5 is possible but requires new calculations . it should be stressed that the ope parameters describe _ universal _ properties of the @xmath0 meson and of the quarks .
for example , @xmath30 and @xmath31 are determined in the fit within less than 40 and 30 mev , resp .
the reach of the new results therefore extends well beyond the @xmath5 determination , as is well demonstrated by the case of @xmath32 .
the ratio @xmath33 measures the left side of the unitarity triangle , identifying a circle in the @xmath34 plane .
the determination of @xmath32 from @xmath35 semileptonic decays parallels that of @xmath5 , but the exclusive determination ( @xmath36 , etc . )
is penalized by the absence of a heavy quark normalization for the form factors at a certain kinematic point . moreover ,
if theoretical precision is lower , so is statistics , by about two orders of magnitude . in view of the precision reached by @xmath5 , a drastic improvement in the determination of @xmath32 , made possible by the high statistics available ,
has become the top priority .
the relevance of @xmath32 is is illustrated for instance in fig . 1 , where the unitarity triangle is determined for the first time using tree - level processes only .
comparing a high precision determination of this kind , insensitive to new physics , to the standard one based on loop processes would be very instructive @xcite .
in the exclusive case , lattice qcd and light cone sum rules complement each other , but as the first unquenched calculations appear , the error in the high-@xmath37 region accessible to lattice still exceeds 10% , while the @xmath37 extrapolation is well under control , thanks to analyticity and new experimental data @xcite .
sum rules prefer a lower value than lattice , @xmath38 ( first of @xcite ) against @xmath39 @xcite .
the goal of lattice simulation is a 5 - 6% determination within a few years @xcite .
recent proposals include a new @xmath40 form factor normalization based on scet @xcite , and the combination of @xmath41 and @xmath42 data @xcite .
the latter is not yet competitive and the @xmath32 result could depend on new physics in the rare decay .
the inclusive determination of @xmath2 is strongly affected by the kinematic cuts necessary to isolate @xmath35 transitions from the dominant @xmath43 background . in general , cuts placed near the perturbative singularities ( typically , the lepton energy endpoint ) destroy the convergence of the ope and introduce a sensitivity to local @xmath44-quark wave function properties like the fermi motion of the heavy quark in the @xmath0 meson .
these non - perturbative effects are not suppressed by powers of @xmath45 : at leading order in this expansion they are described by a single distribution function , often called _ shape function _ ,
whose first moments are given by expectation values of the same _ local _ operators we have encountered earlier .
the shape function is universal : in principle it can be extracted from the photon spectrum in @xmath46 or studying the differential distributions in @xmath47 , although the two processes are different at subleading order in @xmath45 and @xmath17 .
the shape function gets renormalized by perturbative effects : disentangling the latter from non - perturbative contributions is not trivial and has been done in different ways @xcite .
different cutting strategies have been proposed : cuts on the hadronic invariant mass @xmath48 , on the electron energy , on the @xmath37 of the lepton pair and combinations thereof , on the light - cone variable @xmath49 ; each has peculiar experimental and theoretical systematics @xcite , though the uncertainty on leading and subleading shape functions plays often a central role .
eventually , the variety of complementary approaches that have been developed will be extremely useful .
recent theoretical work has focused on the optimization of the cuts , subleading non - perturbative effects @xcite , the resummation of sudakov logs , the role of the radiative decay spectrum in constraining the shape function , etc . @xcite .
the latest hfag average of inclusive determinations , @xmath50 @xcite agrees with the exclusive determination based on lattice .
the error is close to 8% and dominated by theoretical systematics .
the progress since last year is significant and due to : i ) the implementation of the constraints on @xmath30 and on the moments of the shape function derived from @xmath51 spectral moments ; ii ) larger statistics and better knowledge of the charm background , that has allowed to cut in a milder way , to the relief of theorists .
further improvement can be expected from high statistics data . a more precise determination of the radiative spectrum and better experimental constraints on the weak annihilation ( wa ) contributions @xcite are needed .
particularly promising are new analyses based on fully reconstructed events that allow high discrimination of charmed final states .
they allow the measurement of @xmath52 decay distributions well beyond the kinematic cuts on @xmath43 @xcite .
of course for milder cuts the experimental error tends to increase , while the sensitivity to the shape function decreases : the balance between theoretical and experimental errors can be optimized .
the @xmath52 differential distributions and their truncated moments will help constraining the shape function(s ) , weak annihilation ( wa ) contributions , and the heavy quark parameters @xcite .
while the v - a current involved in the semileptonic @xmath0 decays is conserved , that is not the case of the tensor current that induces radiative decays . as a consequence ,
these decays depend logarithmically on the electroweak scale at which the current is generated by @xmath53 and top quark loops : in addition to the @xmath44 mass and the scale of qcd , a third scale @xmath54 must be taken into account . the large logs @xmath55 are resummed in the context of an effective theory , at leading order in @xmath56 : the resummation of @xmath57 term corresponds to the leading order ( lo ) expression , of @xmath58 terms to the nlo , etc . apart from this complication , and from additional ones due to the presence of charm loops @xcite , the ope for _ inclusive _ radiative decays is analogous to the one for semileptonic ones .
the nlo calculation was practically completed in 1996 @xcite and involves @xmath59 matrix elements .
electroweak effects and power corrections are also known @xcite .
the main theoretical uncertainty in the nlo analysis of the branching ratio ( br ) is related to the mass of the charm quarks circulating in the @xmath59 loops of the matrix element @xmath60 @xcite .
this matrix element vanishes at lo , and the charm dependence is an @xmath59 effect . the _
natural _ scale at which @xmath61 should be normalized is therefore undetermined without and @xmath62 calculation .
the matter is quite relevant : the numerical difference between @xmath63 gev and @xmath64 gev is large enough to shift the br by more than 10% .
using @xmath65 as central value , the br of @xmath46 for @xmath66 gev is @xmath67 , with an error close to 8% @xcite , dominated by the charm scale uncertainty . in practice , a lower cut on the photon energy ( @xmath68 - 1.9 gev )
is always applied to avoid background .
a detailed knowledge of the tail of the spectrum is therefore required for the extrapolation to a region where the ope can be trusted . the same problem also affects the moments of the photon spectrum that are employed in the hqe fit @xcite . both perturbative hard gluon emission and the tail of the shape function concur to form the tail of the photon spectrum . as already mentioned , disentangling them is not straightforward .
the transition from the shape function dominated region to the local ope has been studied by neubert , who found that it can be described by a multi - scale non - local ope @xcite .
he noticed that the cut @xmath69 effectively introduces two new scales that may be relevant besides @xmath30 : @xmath70 and @xmath71 . according to this picture
, the perturbative tail of the spectrum receives contributions at the scale @xmath72 , and may be subject to large higher order perturbative corrections , even for @xmath73 gev , simply because @xmath72 is then close to 1 gev .
this view has been criticized @xcite .
certainly , the results of a new calculation of the dominant @xmath62 effects in the photon spectrum @xcite show only very small deviations from the lo plus blm perturbative spectrum considered in @xcite , excluding large corrections in the standard picture
. moreover , the partial br calculated in the multiscale ope , br(@xmath66 gev)= @xmath74@xmath75 @xcite has a central value very close to that of @xcite .
the values of @xmath30 and @xmath76 extracted from the first and second photonic moments in @xcite following @xcite are also very close to those extracted from the same data using the multiscale ope @xcite .
one would conclude that the local ope is likely to provide a good description of the spectrum in the intermediate region @xmath77
gev of phenomenological interest , and that there is no need to enlarge substantially the theoretical error on the partial br given in @xcite . [ cols="<,<,<,<,<",options="header " , ] in order to compare different experiments ,
the cut rates are usually extrapolated to a conventionally defined _ total _
rate @xcite .
the nlo result @xmath78 @xcite can then be compared with the experimental world average : @xmath79 @xcite .
no deviation from the sm is observed , but in view of the final accuracy expected at the @xmath0 factories the theoretical prediction has to be improved . that is the aim of the nnlo calculation currently under way @xcite . to date , the missing pieces of this challenging enterprise include the four loop anomalous dimension matrix and the finite parts of the three loop matrix elements with charm loop .
the high precision of inclusive radiative @xmath80 decays is not yet matched by the theoretical understanding of _ exclusive _
radiative @xmath0 decays @xcite , despite some recent progress @xcite . because of the difficulty of measuring inclusively @xmath81 , the ratio of @xmath81 over @xmath82 exclusive modes is extremely interesting .
@xmath83 has just been measured for the first time @xcite .
the ratio @xmath84 , in particular , is hardly affected by wa contributions , while the su(3 ) breaking effects can be estimated on the lattice and using light cone sum rules ; the error is 10 - 20% at most @xcite . in this way
, we can extract @xmath85 before @xmath86 is measured .
while the 2004 preliminary result showed an interesting deviation from the global ut fit , the recent belle update has restored consistency with the sm and damped enthusiasm @xcite .
finally , it should also be mentioned that the rare leptonic transitions @xmath87 complement radiative decays in constraining new physics @xcite . the inclusive decay @xmath88 , in particular , has reached experimental and theoretical maturity with theoretical errors comparable to @xmath89 @xcite .
the joint theoretical and experimental effort to study semileptonic and radiative decays to high precision has led to relevant progress in the determination of the ckm matrix elements and in testing the standard model .
table 1 summarizes the present theoretical uncertainty and outlines the main ingredients necessary to improve on it for the various processes .
while no deviation from the standard model has been so far uncovered , more theoretical work is needed to make the most of the wealth of data coming from the @xmath0 factories .
m. battaglia _ et al .
_ , hep - ph/0304132
. b. aubert _ et al . _ [ babar coll .
] , phys .
d69 ( 2004 ) 111103 , phys .
d69 ( 2004 ) 111104 , phys .
* 93 * ( 2004 ) 011803 ; k. abe _ et al . _
[ belle coll .
] , hep - ex/0409015 , hep - ex/0509013 ; s. e. csorna _ et al . _
[ cleo coll .
] , phys .
d * 70 * ( 2004 ) 032002 ; a. h. mahmood _ et al .
_ [ cleo coll .
] , phys .
d * 70 * ( 2004 ) 032003 ; r. miquel [ cdf coll .
] , nucl .
142:205,2005 ; delphi coll . , cern - ph - ep/2005 - 015
; m. battaglia _ et al .
_ , phys .
b * 556 * ( 2003 ) 41 .
p. gambino and n. uraltsev , eur .
j. c * 34 * ( 2004 ) 181 . c. w. bauer _ et al .
_ , phys .
d * 70 * ( 2004 ) 094017 .
o. buchmuller and h. flacher , hep - ph/0507253 .
s. chen _ et al .
_ [ cleo coll .
] , phys .
* 87 * ( 2001 ) 251807 ; p. koppenburg _ et al .
_ [ belle coll .
] , phys .
* 93 * ( 2004 ) 061803 ; b. aubert _ et al . _
[ babar coll .
] , hep - ex/0507001 .
m. trott , phys .
d * 70 * ( 2004 ) 073003 ; n. uraltsev , int .
j. mod .
phys . a * 20 * ( 2005 ) 2099 ; v. aquila _ et al . _
b * 719 * ( 2005 ) 77 .
buras _ et al .
_ , phys .
b * 500 * 161 ( 2001 ) .
p. ball and r. zwicky , phys .
b * 625 * 225 ( 2005 ) ; t. becher and r.j .
hill , hep - ph/0509090 .
j. shigemitsu _ et al .
_ , hep - lat/0408019 ; m. okamoto _ et al .
_ , nucl .
* 140 * ( 2005 ) 461 .
j. shigemitsu , talk at ckm 2005 , san diego , march 2005 ; all presentations can be seen at http://ckm2005.ucsd.edu/. c. n. burrell , m. e. luke and a. r. williamson , phys .
d * 69 * ( 2004 ) 074015 ; s. w. bosch , m. neubert and g. paz , jhep * 0411 * ( 2004 ) 073 .
k. s. m. lee and i. w. stewart , nucl .
b * 721 * ( 2005 ) 325 ; m. beneke _ et al .
_ , jhep * 0506 * ( 2005 ) 071 .
b. o. lange , m. neubert and g. paz , hep - ph/0504071 and hep - ph/0508178 and refs . therein .
see also j. r. andersen and e. gardi , hep - ph/0509360 ; u. aglietti , g. ricciardi and g. ferrera , hep - ph/0507285 and hep - ph/0509095 .
# 1#2#3phys .
* b#1 * ( 19#2 ) # 3 # 1#2#3z .
* c#1 * ( 19#2 ) # 3 # 1#2#3phys .
# 1 * ( 19#2 ) # 3 # 1#2#3rev .
* # 1 * ( 19#2 ) # 3 # 1#2#3phys .
# 1 * ( 19#2 ) # 3 # 1#2#3phys . rev . * d#1 * ( 19#2 ) # 3 # 1#2#3nucl . phys .
* b#1 * ( 19#2 ) # 3 # 1#2#3mod .
# 1 * ( 19#2 ) # 3 k. g. chetyrkin , m. misiak and m. munz , phys.lett.b*400 * ( 1997 ) 206 ; p. gambino , m. gorbahn and u. haisch , nucl . phys .
b * 673 * ( 2003 ) 238 ; a. j. buras and m. misiak , acta phys .
b * 33 * ( 2002 ) 2597 and refs .
therein .
a. czarnecki and w. j. marciano , phys .
* 81 * ( 1998 ) 277 ; p. gambino and u. haisch , jhep 09 ( 2000 ) 001 and 10 ( 2001 ) 020 ; a. f. falk , m. luke , and m. j. savage , phys .
* d49 * ( 1994 ) 3367 ; c. w. bauer , phys .
d * 57 * ( 1998 ) 5611 [ erratum - ibid .
d * 60 * ( 1999 ) 099907 ] .
p. gambino and m. misiak , nucl .
b * 611 * ( 2001 ) 338 .
i. blokland _ et al .
_ , phys .
d * 72 * , 033014 ( 2005 ) ; m. gorbahn , u. haisch and m. misiak , phys .
lett . * 95 * , 102004 ( 2005 ) ; m. misiak and m. steinhauser , nucl .
b * 683 * , 277 ( 2004 ) ; k. bieri , c. greub and m. steinhauser , phys .
d * 67 * , 114019 ( 2003 ) ; h. m. asatrian _ et al .
_ , phys .
b * 619 * , 322 ( 2005 ) .
a. ali , e. lunghi and a. y. parkhomenko , phys .
b * 595 * ( 2004 ) 323 ; s. w. bosch and g. buchalla , jhep * 0501 * ( 2005 ) 035 ; m. beneke , t. feldmann and d. seidel , eur .
j. c * 41 * ( 2005 ) 173 .
k. abe _ et al . _ [ belle coll .
] , hep - ex/0506079 ; b. aubert _ et al . _
[ babar coll .
] , phys .
* 94 * ( 2005 ) 011801 .
p.ball and r.zwicky , phys.rev.d * 71 * , 014029 ( 2005 ) ; s. bosch , talk at ckm 2005 @xcite . | i briefly review the theoretical status of semileptonic and radiative @xmath0 decays in 2005 . |
three jet production in @xmath3 annihilation was investigated in details at petra and later at tristan energies @xcite in connection with the critical tests of perturbative qcd . in particular
the difference between energy spectra of quark and gluon jets and quark and gluon jet multiplicities provided important information for quantitative tests of quantum chromodynamics . besides the studies of hadronic final state as a result of direct @xmath4 annihilation , the measurement of the two photon processes @xmath5 at petra and tristan shows high physics potential .
new tests are possible in lep experiments , especially the precise meauserement of the photon structure function and hadron production at high @xmath0 @xcite at present time new experimental possibilities connected with the generation of @xmath6 beams by compton backscattering of the laser beam on a high energy electron beam @xcite are discussed extensively @xcite .
investigation of the jet phenomena in @xmath7 collisions at the energy of several hundred gev is one of the most interesting questions for @xmath7 colliders phenomenology . in this paper
we discuss three jet production in the region of large @xmath0 originated from t - channel quark exchange between the colliding photons . at next linear colliders energy
there are two other possible mechanisms for the jet production in high @xmath0 region with the cross sections of the same order @xcite : ( 1 ) gluon exchange between two pairs of quark jets ( 2 ) @xmath8 boson production .
however , they have different final state topologies and can easily be distinguished from the mechanism under consideration . at smaller @xmath0
the photon looks more like the hadron - like object and special treatment beyond the simple qcd tree approximation is necessary .
the aims of this paper are the calculation of high @xmath0 quark and gluon jets energy spectra in @xmath7 collisions and comparison of the results with the @xmath3 case .
the first order qcd calculation for three jet production in @xmath10 ( @xmath11 ) gives the following well - known result for the spectra @xcite : @xmath12 where @xmath13 are the energy fractions of quark , antiquark and gluon @xmath14 satisfying the equality @xmath15 distribution ( 1 ) defines the topology of the final state ( or topology of the primary process ) in @xmath3 collision . since quarks and gluons are unobservable , in the following various fragmentation models are used to generate real event samples .
however , event generation will be beyond our analysis at present stage .
the calculation in @xmath4 case for two feynman amplitudes only ( gluon is radiated from quark or antiquark leg ) is not technically difficult .
in @xmath7 case we have 6 amplitudes represented in fig.1 .
symbolic calculation for the corresponding 21 squared amplitudes could be more complicated . however , it is much simplified because the compact form of the result for the sum of 21 squared diagrams can be obtained @xcite : @xmath16 } { \prod_{i=1}^{3}\,(p_ip_4)\,(p_ip_5)}\ ] ] we reproduced this formula using comphep package @xcite for generation of symbolic result and reduce system @xcite in the following algebraic transformations . for the process
@xmath17 we are using the kinematical variables @xmath18 where @xmath19 is the angle between particle 1 and particle 4 in the c.m.s . of ( 1,2 ) , @xmath20 is the helicity angle between the planes ( 1,4 ) and ( 3,4 ) in the c.m.s of ( 3,5 ) .
the phase space in these variables takes the form @xmath21 in the case of @xmath22 process the matrix element has collinear singuliarities only . in the case of @xmath23 process additional t - channel singuliarity appears .
the denominator of the squared amplitude contains four momenta products ( see more details in the appendix ) @xmath24 where @xmath25 integration over @xmath20 leads to the structures @xmath26 and @xmath27 in the denominator ( see more details in appendix ) . for this reason besides the poles of matrix element for the scattering in forward - backward directions ( no gluon emission from final quark ) there are parametrically dependent poles at @xmath28 corresponding to the t - channel singuliarity in the diagrams with gluon emission from the quark leg .
( in this case the antiquark is emitted in the opposite hemisphere at zero angle ) .
while integrating over @xmath29 we introduced kinematical cuts @xmath30 and @xmath31 for quark near the forward pole and the poles defined by ( 13 ) .
after two integrations over the angular variales @xmath32 we obtain the following result for three jet spectra in @xmath7 collisions : @xmath33 \nonumber \\ & & \hspace{15 mm } \times [ { \rm ln } \frac{1}{2\epsilon_1\epsilon_2 ^ 2}+2 \,{\rm ln } \frac{\omega_1\omega_2 } { z_1 z_2 } - \frac{2(z_1+z_2 - 1)-z_1z_2}{z_1z_2}{\rm ln } \frac{\omega_1}{\omega_2 } ] \nonumber \\ & & \hspace{15 mm } + z_1 ^ 2 ( ( 1-z_1)^2(1-z_2)^2+z_2 ^ 4 ) { \rm ln } \frac{2}{\epsilon_1 } \nonumber \\ & & \hspace{15 mm } + z_2 ^ 2 ( ( 1-z_1)^2(1-z_2)^2+z_1 ^ 4 ) { \rm ln } \frac{\omega_1 \omega_2}{\epsilon_2 ^ 2 } \nonumber \\ & & \hspace{15 mm } + 2z_1 ^ 2z_2 ^ 2(z_1 ^ 2+z_2 ^ 2 ) { \rm ln } \frac{\omega_1}{2
z_1 z_2 } \nonumber \\ & & \hspace{15 mm } + z_1 ^ 4z_2+z_1z_2 ^ 4 -5z_1 ^ 3z_2 - 5z_1z_2 ^ 3 + 4z_1 ^ 2z_2 + 4z_1z_2 ^ 2 + z_1 ^ 2z_2 ^ 3+z_1 ^ 3z_2 ^ 2 \nonumber \\ & & \hspace{15 mm } + 4z_1 ^ 3 + 4z_2 ^ 3 - 4z_1 ^ 2z_2 ^ 2 -2z_1 ^ 2 - 2z_2 ^ 2 - 2z_1
^ 4 - 2z_2 ^ 4 \}\nonumber \\ & & \end{aligned}\ ] ] where @xmath34 besides the collinear singuliarities in @xmath7 case the spectrum ( 14 ) has the factor @xmath35 in denominator and logarithmic terms originating from t - channel quark exchange .
kinematical cut for @xmath30 in the case of nonzero fermion masses is @xmath36 , it is easy to show that the cut for @xmath31 is of the same order .
we show double differential spectra of quark and gluon jets in @xmath3 and @xmath7 collisions ( ( 1),(14 ) ) at the energy scale @xmath37 in fig.2 . in the @xmath7 case
the cross section is an order of magnitude larger than for @xmath4 case .
the shape of the spectra are similar . in the case when there are no additional methods of quark and gluon jet identification the spectra are not measurable separately and the usual way of analysis is jet ordering in energy . in fig.3
we show single differential spectra of quark and gluon jets in the case of @xmath4 and @xmath7 collisions . in the @xmath4 case quark and gluon
jet spectra show large difference giving the possibility of good jet discrimination .
for instance , about 70% of the jets at @xmath38 less than @xmath39 are gluon jets and about 50% of the jets at @xmath38 between @xmath40 and @xmath41 are quark jets . in the @xmath7 case
the difference between the spectra is also well pronounced and identification possibilities of the quark and qluon jets by energy ordering seem not worse than for @xmath4 case .
improvement of jet identification in some experimental situations could be provided by flavor tagging @xcite or comparison of @xmath42 and @xmath43 event topologies [ 1 ] .
probably some combination of different methods can give optimal results .
* acknowledgements * the author is grateful to m.fontannaz , j.fujimoto , j.p.guillet , k.kato , t.munehiza and y.shimizu for useful comments and discussions .
the work was partially supported by isf ( grant m9b000 ) and intas ( grant 93 - 1180 , contract 1010-ct93 - 0024 ) .
in this section we shall show some details of the calculation taking squared amplitudes ( 1 ) and ( 5 ) ( see fig.1 ) as an example .
four momenta products in the variables ( 5 ) have the form @xmath44 where @xmath45 it is worth noticing that this set is not invariant under the transposition @xmath46 ( as long as we are using the @xmath47 c.m.s . as the reference frame ) .
the matrix element has the symmetry @xmath48 , so this symmetry must be restored in the final result for the spectra ( at least in the nondivergent terms , artificial cuts near the poles can break the symmetry ) .
squared diagrams ( 1 ) and ( 5 ) sum @xmath49 after the integration over the helicity angle @xmath20 takes the form @xmath50 where @xmath51 integration over @xmath29 must take into account two cases for the sign of the absolute value .
the physical region of the reaction @xmath9 in @xmath52 plane is the triangle with vertices ( 0,1);(1,0);(1,1 ) .
hyperbola reflecting the relation between @xmath53 for t - channel pole ( 13 ) crosses the physical region from ( 0,1 ) to ( 1,0 ) ( see berends et.al in [ 7 ] ) .
integration of the rational function @xmath54 gives the result for the spectrum @xmath55\end{aligned}\ ] ] calculation for the sum of 21 squared diagrams gives the same symbolic structures .
99 collected physics papers ( i ) of tristan experiments , kek , tsukuba , 1994 p.aurenche , a.douiri , r.baier , m.fontannaz , d.schiff , in:_physics at lep _ , ed.by j.ellis,r.peccei , cern report 86 - 02 , 1986 i.ginzburg , g.kotkin , v.serbo , v.telnov , nucl.instr.meth . 205
( 1983 ) 147 e.boos , m.dubinin , v.ilyin , a.pukhov , in:_@xmath4 collisions at 500 gev : the physics potential _ , ed.by p.zerwas , desy report 93 - 123c , 1993 , p.561 i.ginzburg , d.ivanov , v.serbo , in:_proc.of workshop on physics and experiments with linear @xmath4 colliders _ , ed.by f.harris , s.olsen , s.pakvasa , x.tata , world scientific , singapore , 1993 , p.600 j.ellis , m.k.gaillard , g.g.ross , nucl.phys.b111 ( 1976 ) 253 + t.a.degrand , y.j.ng , s.h.h.tye , phys.rev.d16 ( 1977 ) 3251 + a.derujula , j.ellis , e.g.floratos , m.k.gaillard , nucl.phys.b138 ( 1978 ) 387 + g.kramer , g.schierholz , phys.lett.82b ( 1979 ) 102 + p.hoyer , p.osland , h.g.sander , t.f.walsh , p.m.zerwas , nucl
b161 ( 1979 ) 349 + s.nandi , w.wada , phys.rev.d21 ( 1980 ) 76 + h.p.nilles , k.h.streng , phys.rev.d23 ( 1981 ) 1944 f.berends , z.kunszt , g.gastmans , phys.lett.92b ( 1980 ) 186 , nucl.phys.b182 ( 1981 ) 397 + p.aurenche , a.douiri , r.baier , m.fontannaz , d.schiff , z.phys.c24 ( 1984 ) 309 e.boos et.al.,in:_91 electroweak interactions and unified theories ( proc . of the xxvith recontre de moriond ) _ , ed.by j.tran than van , editions frontieres , 1991 , p.501 + e.boos et.al.,in : _ new computing techniques in physics research ii ( proc .
of the second int.workshop on software engineering , artificial intelligence and expert systems in high energy and nuclear physics ) _ , ed.by d.perret-gallix , world scientific , 1992 , p.665 + e.boos , m.dubinin , v.ilyin , a.pukhov , v.savrin , preprint inp msu 94 - 36/358 , 1994 ( * hep - ph/9503280 * ) reduce by a.c.hearn , rand corp .
, cp78 ( rev.7.91 ) , 1991 h.borner , p.grosse-wiesmann , in : _
@xmath3 collisions at 500 gev : the physics potential _ , ed.by p. zerwas , desy report 92 - 123a , 1992 , p.63
* feynman diagrams for @xmath9 .
* double differential spectra @xmath56 at fixed @xmath57 of parent quark and gluon in the reactions @xmath22 and @xmath9 ( @xmath58 ) . @xmath59 * differential spectra @xmath60 ( integrated over @xmath61 ) of parent quark and gluon in the reactions @xmath22 and @xmath9 ( @xmath58 ) . | we consider three jet production at high @xmath0 in the @xmath1 process and calculate quark and gluon jet spectra .
the possibility of quark - gluon jet separation is discussed and compared with the @xmath2 case . |
in this paper we are interested in hamilton - jacobi ( hj ) equations posed on a one dimensional domain containing one single singularity . this is a special case of a more general setting where hj equations are posed in domains that are unions of submanifolds whose dimensions are different @xcite .
an intermediate setting is the study of hj equations on networks , see in particular @xcite .
we will restrict ourselves to a very simple network : the union of a finite numbers of half - lines of the plane with a single common point .
such a domain is referred to as a _ junction _ and the common point is called the _ junction point_. we point out that getting a comparison principle is the most difficult part in such a study ; it is obtained in @xcite for similar special networks ( bounded star - shaped ones ) .
our motivation comes from traffic flows . for this reason ,
it is natural to impose different dynamics on each _ branch _ of the junction .
consequently , the resulting hamiltonian is by nature discontinuous at the junction point . together with the singularity of the domain
, this is the major technical difficulty to overcome .
the analysis relies on the complete study of some _ minimal action _ ( or _ metric _ ) related to the optimal control interpretation of the equation @xcite .
we prove in particular that this minimal action is semi - concave by computing it .
we first present the problem and the main results in details .
then we recall existing results and compare them with ours . in this subsection , the analytical problem is introduced in details .
we first define the junction , then the space of functions on the junction and finally the hamilton - jacobi equation .
[ [ the - junction . ] ] the junction .
+ + + + + + + + + + + + + let us consider @xmath0 different unit vectors @xmath1 for @xmath2 .
we define the branches @xmath3 and the _ junction _ ( see figure [ f6 ] ) @xmath4 the origin @xmath5 is called the _ junction point_. it is also useful to write @xmath6 . for a time @xmath7
, we also define @xmath8 the reader can remark that we chose to embed the junction in a two - dimensional euclidean space .
but we could also have considered an abstract junction , or we could have embedded it for instance in a higher dimensional euclidean space .
we made such a choice for the sake of clarity .
[ [ space - of - functions . ] ] space of functions .
+ + + + + + + + + + + + + + + + + + + for a function @xmath9 , we denote by @xmath10 the restriction of @xmath11 to @xmath12 .
then we define the natural space of functions on the junction @xmath13 in particular for @xmath14 and @xmath15 with @xmath16 , we define @xmath17 then we set @xmath18 [ [ hj - equation - on - the - junction . ] ] hj equation on the junction .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + we are interested in continuous functions @xmath19 which are _ viscosity solutions _
( see definition [ defi : main ] ) on @xmath20 of @xmath21 submitted to an initial condition @xmath22 because of the optimal control problem we have in mind ( see subsection [ ss13 ] below ) , we restrict ourselves to the simplest case of discontinuous hamiltonians ; precisely , we consider @xmath23 where @xmath24 are convex functions whose legendre - fenchel transform is denoted @xmath25 .
we recall that @xmath26 and @xmath27 we also consider @xmath28 therefore equation ( [ eq : main ] ) can be rewritten as follows @xmath29 [ [ the - optimal - control - framework . ] ] the optimal control framework .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in this paragraph , we give an optimal control interpretation @xcite of the hamilton - jacobi equation .
we define the set of admissible controls at a point @xmath30 by @xmath31 for @xmath32\times j$ ] with @xmath33 ( the case @xmath34 being trivial and forcing @xmath35 ) , we define the set of admissible trajectories from @xmath36 to @xmath37 by @xmath38;{\mathbb{r}}^2)\colon \left|\begin{array}{ll } x(\tau)\in j & \mbox{for all}\quad \tau \in ( s , t),\\
\dot{x}(\tau ) \in a(x(\tau ) ) & \mbox{for a.e.}\quad \tau \in ( s , t)\\ x(s)=y & \mbox{and}\quad x(t)=x \end{array}\right . \right\}.\ ] ] for @xmath39 with @xmath40 , we define the lagrangian on the junction @xmath41 with @xmath42 the reader can be surprised by the definition of @xmath43 for @xmath5 .
in fact , if one considers only trajectories that do not stay at the junction point , then the ones staying at @xmath44 are approximated by those staying very close to @xmath44 on a branch @xmath45 and moving `` slowly '' ( @xmath46 ) .
we make the following assumptions : [ [ a0 ] ] * ( a0 ) * + + + + + + the initial data @xmath47 is lipschitz continuous .
[ [ a1 ] ] * ( a1 ) * + + + + + + there exists a constant @xmath48 , and for all @xmath2 , there exists @xmath49 functions @xmath25 satisfying @xmath50 , such that ( [ eq::r14 ] ) and ( [ eq::r13 ] ) hold . [ th::r2 ] assume ( a0)-(a1 ) and let @xmath7 .
then there exists a unique viscosity solution @xmath11 of ( [ eq : main])-([eq : ci ] ) on @xmath20 in the sense of definition [ defi : main ] , satisfying for some constant @xmath51 @xmath52 moreover the function @xmath11 is lipschitz continuous with respect to @xmath37 on @xmath20 . on one hand
, we will see below that the existence of a solution can be obtained with perron s method under weaker assumptions than ( a1 ) ( see theorem [ thm : perron ] ) . on the other hand ,
we are able to get uniqueness of the solution only under assumption ( a1 ) and this is a consequence of the following result : [ th::r1 ] assume ( a0)-(a1 ) .
let @xmath7 and let @xmath11 ( resp .
@xmath53 ) be a sub - solution ( resp .
a super - solution ) of ( [ eq : main])-([eq : ci ] ) on @xmath20 in the sense of definition [ defi : main ] .
we also assume that there exists a constant @xmath51 such that for all @xmath54 @xmath55 then we have @xmath56 on @xmath20 . in order to prove this strong uniqueness result
, we will use in an essential way the _ value function _ associated to the optimal control problem described in subsection [ ss13 ] : for @xmath57 , @xmath58 where @xmath59 is defined in ( [ eq::r15 ] ) and @xmath60 is defined in ( [ eq::r16 ] ) .
[ th::r3 ] assume ( a0)-(a1 ) and let @xmath7 .
the unique solution given by theorem [ th::r2 ] is @xmath61 with @xmath62 given in ( [ eq::r17 ] ) .
moreover , we have the following hopf - lax representation formula @xmath63 with @xmath64 the comparison principle is obtained by combining * a super - optimality principle for counterrevolutions @xmath53 , which implies @xmath65 ; * a direct comparison result with sub solutions @xmath11 , which gives @xmath66 .
we finally have the following result which shed light on the role of the junction condition ( see the second line of ( [ eq::r12 ] ) ) . [ th::r4 ]
assume ( a0)-(a1 ) and let @xmath7 .
let @xmath67 be such that @xmath68 and for each @xmath69 , the restriction @xmath10 of @xmath11 to @xmath12 is a classical viscosity solution of @xmath70 then @xmath11 is a sub - solution of ( [ eq : main])-([eq : ci ] ) on @xmath20 in the sense of definition [ defi : main ] , and @xmath71 .
an immediate consequence of theorem [ th::r4 ] is the fact that the optimal control solution @xmath62 is the maximal continuous function which is a viscosity solution on each open branch .
we apply in section [ s5 ] our hj approach to describe traffic flows on a junction . in particular , we recover the well - known junction conditions of lebacque ( see @xcite ) or equivalently those for the riemann solver at the junction as in the book of garavello and piccoli @xcite ; see also @xcite .
we already mentioned that the main difficulties we have to overcome in order to get our main results are on one hand the singular geometry of the domain and on the other hand the discontinuity of the hamiltonian .
[ [ discontinuity . ] ] * discontinuity . * + + + + + + + + + + + + + + + + several papers in the literature deal with hjb equations with discontinuous coefficients ; see for instance @xcite .
note that in these works the optimal trajectories do cross the regions of discontinuities ( i.e. the junction in the present paper ) only on a set of time of measure zero . in the present paper
, the optimal trajectories can remain on the junction during some time intervals , and the results cited above do not apply then to the problem studied here .
on the other hand , the analysis of scalar conservation laws with discontinuous flux functions has been extensively studied , we refer to @xcite and references therein .
we also point out that a uniqueness result is proved in @xmath72 in the framework corresponding a junction with two branches @xcite . to the best of our knowledge ,
in the case of junctions with more than two branches , there are no uniqueness result .
moreover , the link between hjb equations and conservation laws with discontinuous has been seldom investigated @xcite .
the main differences between the study in @xcite and the one carried out in the present paper lie in the fact that in @xcite the lagrangian can depend on @xmath73 and is continuous with respect to this variable , while we consider a lagrangian which is constant in @xmath73 on each branch but can be discontinuous ( with respect to @xmath73 ) at the junction .
we point out that we can not extend directly our approach to lagrangians depending on @xmath73 since we use extensively the representation formula `` la hopf - lax '' . in order to generalise results in this direction ,
the semi - concavity of the `` fundamental solution '' @xmath74 should be proved without relying on explicit computations .
this question is very interesting but is out of the scope of the present paper .
[ [ networks . ] ] * networks .
* + + + + + + + + + + + it is by now well known that the study of traffic flows on networks is an important motivation that give rise to several difficulties related to scalar conservation laws with discontinuous coefficients .
this topic has been widely studied by many authors , see for instance @xcite and the references therein .
however , the study of hjb equations on networks has been considered very recently ; the reader is referred to @xcite where eikonal equations are considered .
a more general framework was also studied in @xcite where a definition of viscosity solutions on networks , similar to definition [ defi : main ] , has been introduced .
it would be interesting to extend the results of the present paper to more general networks but the obstacle is the same than the one to be overcome if one wants to deal with lagrangians depending on @xmath73 : for a general network , the complete study of the fundamental solution is probably out of reach .
this is the reason why we only consider the very specific case of a junction in order to be able to overcome the difficulty of the discontinuity of the lagrangian . [
[ the - optimal - control - interpretation . ] ] * the optimal control interpretation . * + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + as explained above , the comparison principle is proved by using in an essential way the optimal control interpretation of the hamilton - jacobi equation .
the use of representation formulas and/or optimality principles is classical in the study of hamilton - jacobi equations @xcite .
more specifically , it is also known that a `` metric '' interpretation of the hamilton - jacobi equation is fruitful @xcite .
such an interpretation plays a central role in the weak kam theory @xcite .
as far as our problem is concerned , we are not able to prove uniqueness of viscosity solutions by using the classical techniques of doubling variable .
the idea used here is based on the equivalence between the viscosity super - solution and the super - optimality principle ( also known as weak - invariance principle ) , and by using representation formulas for the viscosity sub solutions .
this representation seems to be new for hjb equations with discontinuous coefficients , see for instance @xcite .
we would like next to be a bit more precise .
the technical core of the paper lies in theorem [ thm : minimal - action ] .
this result implies that the function @xmath75 is semi - concave with respect to @xmath37 and @xmath36 and , if there are at least two branches ( @xmath76 ) , that @xmath74 satisfies @xmath77 ( in a weak sense made precise in the statement of theorem [ thm : minimal - action ] ) . in the case where the lagrangians coincide at the junction point ( @xmath78 )
, it turns out that the restriction @xmath79 of @xmath80 to @xmath81 belong to @xmath82 and is convex .
a more general case is considered in this paper : lagrangians can differ at the junction point and in this case , the functions @xmath83 are not convex nor @xmath84 anymore for some @xmath85 .
let us point out here that the assumptions on the hamiltonian @xmath24 , and in particular the fact that it does not depend on the space variable @xmath73 , plays a crucial role to establish the properties satisfied by the minimal action function @xmath86 .
[ [ generalization - and - open - problems . ] ] * generalization and open problems . *
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + eventually , we briefly mention natural generalizations of our results and some important open problems .
first of all , it would be natural to extend the results of this paper to more general setting where the hamiltonians @xmath24 depend on the space variable @xmath73 .
moreover , it would be interesting to consider general networks with several junction points .
dealing with non - convex and non - coercive hamiltonians is quite challenging and would require first to have a direct proof of the comparison principle which does not need to go through the interpretation of the viscosity solution as the value function of an optimal control problem .
section [ s5 ] is devoted to the application of our results to some traffic flow problems .
in particular , the hj equation is derived and the junction condition is interpreted . in section [ sec : visc ] , the definition of ( viscosity ) solutions is made precise . in section [ sec : oc ] , the first important properties of optimal trajectories are given .
section [ sec : op ] is devoted to the proof of the main results of the paper .
in particular , the comparison principle is proved by proving a super - optimality principle and by comparing sub solutions with the solution given by the optimal control interpretation of the equation . section [ sec : complete ] is devoted to the proof of the technical core of the paper , namely the existence of test functions for the minimal action associated with the optimal control interpretation .
[ [ distance - and - coordinates - in - the - junction . ] ] distance and coordinates in the junction .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we denote by @xmath87 the geodesic distance defined on @xmath88 by @xmath89 for @xmath90 , @xmath91 denotes the ( open ) ball centered at @xmath73 of radius @xmath92 .
we also consider balls @xmath93 centered at @xmath94 of radius @xmath95 . for @xmath30 ,
let us define the index @xmath96 of the branch where @xmath73 lies .
precisely we set : @xmath97 up to reordering the indices , we assume that there exists an index @xmath98 such that @xmath99 we also set @xmath100 [ [ functions - defined - in - mathbfj2 . ] ] functions defined in @xmath101 . + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + for a function @xmath102 defined on @xmath103 , we call @xmath104 its restriction to @xmath105
. then we define the space @xmath106 we also call for @xmath15 with @xmath16 and @xmath107 with @xmath108 @xmath109 and @xmath110 and similarly @xmath111 we also set @xmath112
in this section we present the derivation of the hamilton - jacobi formulation of traffic on a junction .
we also discuss the meaning of our junction condition in this framework and relate it to known results .
we consider a junction ( represented on figure [ f5 ] ) with @xmath113 incoming roads ( labeled by the index @xmath114 ) and @xmath115 outgoing roads ( labeled by @xmath116 ) .
this means that the cars move on the incoming roads in the direction of the junction and then have to choose to go on one of the @xmath117 outgoing roads .
we assume that the proportion of cars coming from the branch @xmath118 is a fixed number @xmath119 ( which may be not realistic for @xmath120 ) , and that the proportion of cars going on each branch @xmath121 is also a fixed number @xmath122 .
we also assume the obvious relations ( for conservation of cars ) @xmath123 we denote by @xmath124 the car density at time @xmath125 and at the position @xmath126 on the branch @xmath127 .
in particular , we assume that the traffic is described on each branch @xmath127 by a flux function @xmath128 .
we assume @xmath129 the typical example of such flux function is given by the lwr model ( lighthill , whitham @xcite and richards @xcite ) , with @xmath130 where @xmath131 and @xmath132 are respectively the maximal velocity and the maximal car density in the model ) . in this model the critical car density @xmath133 where @xmath134 is maximal , is equal to @xmath135 .
we assume that the car densities are solution of non linear transport equations : @xmath136 where we assume that the junction point is located at the origin @xmath137 .
we do not precise yet the junction condition at @xmath137 , and we now proceed formally to deduce the hamilton - jacobi model of such a junction . for a function @xmath138 to determine ,
let us consider the functions @xmath139 then we can compute formally for @xmath140 @xmath141 this shows that for @xmath140 @xmath142 where @xmath143 remark that we can show similarly that ( [ eq::r3 ] ) is still true for the index @xmath144 replaced by @xmath118 with @xmath145 in particular , this shows ( at least when the quantities in ( [ eq::r3 ] ) are well defined ) that we can choose @xmath138 such that the total flux @xmath146 is given by @xmath147 and then we have @xmath148 if and only if @xmath149 which is exactly the expected condition which says that the proportion of incoming cars going in the junction from the branch @xmath150 is @xmath151 and the proportion of cars getting out of the junction which choose to go on the branch @xmath144 is @xmath152 .
let us notice that if we choose the initial condition @xmath153 , then we deduce from ( [ eq::r4 ] ) that we have with @xmath154 @xmath155 which shows that _ @xmath156 can be interpreted as the total quantity of cars passing through the junction point @xmath137 during the time interval @xmath157_. as a consequence , the quantity @xmath158 can also be interpreted as the instantaneous flux of cars passing through the junction point .
we now give a further interpretation of the problem in the special case @xmath159 . in the special case @xmath159 ,
imagine for a moment , that we come back to a discrete description of the traffic , where each car of label @xmath127 has a position @xmath160 with the ordering @xmath161 .
we can be interested in the label @xmath127 of the car @xmath162 which is the closest to the junction point @xmath137 .
let us call it @xmath163 .
we can normalize the initial data such that @xmath164 .
then the quantity of cars that have passed through the junction point @xmath137 during the time interval @xmath157 is equal to @xmath165 , which is the exact discrete analogue of the continuous quantity @xmath166 . on the other hand
the number of cars between the positions @xmath167 and @xmath168 is obviously equal to @xmath169 , and its continuous analogue on the branch @xmath170 with @xmath171 , is @xmath172 .
this shows that @xmath173 can be interpreted as the exact continuous analogue of the discrete labeling of the cars moving in the traffic .
this interpretation is also meaningful on the `` exit '' branches , i.e. for @xmath121 . indeed , for such @xmath144 s
, @xmath174 is the continuous analogue of the discrete label of the car that have decided to choose the branch @xmath144 and which is at time @xmath125 close to the position @xmath175 .
we now set @xmath176 and we define the convex hamiltonians @xmath177 then we deduce from ( [ eq::r3 ] ) that we have @xmath178 with equality of the functions at the origin , i.e. @xmath179 notice that for the choice @xmath180 in ( [ eq::r155 ] ) , we get with @xmath181 for all @xmath182 , that @xmath183 in particular this shows that the @xmath184 are not all the same , even in the simplest case .
a junction condition is still needed so that the solution of ( [ eq::r8 ] ) be uniquely defined . indeed , at first glance , one may think that @xmath185 is equal to @xmath146 which is given by ( [ eq::r4 ] ) ( where we have assumed ( [ eq::r11 ] ) ) .
the point is that this condition can not be satisfied for every time .
one way to be convinced oneself of that fact is to consider the case @xmath186 with @xmath187 .
then , we look at solutions @xmath11 of the hamilton - jacobi equation in @xmath72 with the artificial junction . we can simply associate with it the classical conservation law on the whole real line .
we can then consider a single shock moving with constant velocity for the conservation law .
when this shock will pass through the junction point
( let us say at time @xmath188 ) , this will mean that @xmath189 is discontinuous in space at the junction point .
in particular the formal computations of subsection [ s51 ] are no longer valid at that time @xmath188 , even if they are valid for @xmath190 . for a general problem
, one may expect that our formal computations are only valid for almost every time ( even if it is not clear for us ) . in view of theorem [ th::r4 ] ,
if we restrict our attention to continuous solutions @xmath11 , then we will have @xmath71 where @xmath62 is the solution associated to the optimal control problem .
this shows in particular that we have @xmath191 which means ( in view of ( [ eq::r6 ] ) and the interpretation of @xmath192 given in subsection [ s51 ] ) that we have a universal bound on the total amount of cars passing through the junction point during the time interval @xmath157 .
if we assume moreover that this amount of cars is maximal , then we can choose ( and indeed have to choose ) @xmath61 and the natural junction condition is then @xmath193 with @xmath194 using our assumption ( [ eq::r0 ] ) on the functions @xmath195 , let us define for @xmath196 the demand functions @xmath197 and the supply functions @xmath198 from assumption ( [ eq::r0 ] ) on the functions @xmath195 , we deduce that @xmath199 condition ( [ eq::r9 ] ) means that @xmath200 notice that from ( [ eq::r11 ] ) , it is natural to compare @xmath201 then condition ( [ eq::r10 ] ) is nothing else that the demand and supply condition of lebacque , which claims that the passing flux is equal to the minimum between the demand and the supply , as it is defined in @xcite ( at least in the case @xmath159 ) . in the special case @xmath159 , it is explained in @xcite that this condition ( [ eq::r10 ] ) is also equivalent to the condition defining the riemann solver at the junction point in the book of garavello and piccoli @xcite .
let us notice that this condition is also related to the bardos , le roux , ndelec @xcite boundary condition .
in this section , we consider a weaker assumption than ( a1 ) . we introduce the following assumption : [ [ a1 - 1 ] ] ( a1 ) + + + + + for each @xmath202 ,
* the function @xmath203 is continuous and @xmath204 * there exists @xmath205 such that @xmath24 is non - increasing on @xmath206 $ ] and non - decreasing on @xmath207 ; when ( a1 ) holds true , the function @xmath208 is defined by @xmath209 .
we now make the following useful remark whose proof is left to the reader .
[ lem::r18 ] assumption ( a1 ) implies assumption ( a1 ) .
next we give equivalent definitions of viscosity solutions for .
we give a first definition where the junction condition is satisfied in `` the classical sense '' ; we then prove that it is equivalent to impose it in `` the generalized sense '' .
it is essential if one expects solutions to be stable .
we give a first definition of viscosity solutions for in terms of test functions by imposing the junction condition in the classical sense .
we recall the definition of the upper and lower semi - continuous envelopes @xmath210 and @xmath211 of a function @xmath212 : @xmath213 [ defi : main ] a function @xmath214 is a _ sub - solution ( resp .
super - solution ) of on @xmath20 _ if it is upper semi - continuous ( resp .
lower semi - continuous ) and if for any @xmath215 such that @xmath216 in @xmath217 for some @xmath218 , @xmath95 and such that @xmath219 at @xmath220 , we have @xmath221 that is to say * if @xmath222 , then @xmath223 * if @xmath224 , then @xmath225 a function @xmath214 is a _ sub - solution ( resp .
super - solution ) of - on @xmath20 _ if it is a sub - solution ( resp .
super - solution ) of on @xmath20 and moreover satisfies @xmath226 ( resp .
@xmath227 ) . a function @xmath228 is a _ ( viscosity ) solution of ( resp .
- ) on @xmath20 _ if @xmath210 is a sub - solution and @xmath211 is a super - solution of ( resp .
- ) on @xmath20 . as mentioned above , the following proposition is important in order to get discontinuous stability results for the viscosity solutions of definition [ defi : main ] .
[ prop : equiv - def ] assume ( a1 ) .
a function @xmath229 is a sub - solution ( resp .
super - solution ) of on @xmath20 if and only if for any function @xmath230 such that @xmath216 in @xmath20 and @xmath219 at @xmath231 , * if @xmath222 , then @xmath232 * if @xmath224 , then either there exists @xmath233 such that @xmath234 or holds true .
we classically reduce to the case where the ball @xmath217 is replaced with @xmath20 .
the `` if '' part is clear . let us prove the `` only if '' one .
we distinguish the sub - solution case and the super - solution one .
we start with super - solutions since it is slightly easier . * case 1 : super - solution case .
* we consider a test function @xmath215 such that @xmath235 in @xmath20 and @xmath236 at @xmath237 .
there is nothing to prove
if @xmath238 so we assume @xmath239 .
we have to prove that @xmath240 .
we argue by contradiction and we assume that @xmath241 then it is easy to see that there exists a function @xmath242 such that @xmath243 with equality at the point @xmath244 and such that @xmath245 notice that @xmath246 the first inequality is straightforward . to check the second inequality , we have to distinguish two cases .
either we have @xmath247 , and then @xmath248 and we use the fact that the minimum of @xmath208 is @xmath249 . or @xmath250 and then this common value belongs to the interval @xmath206 $ ] on which we have @xmath251 . since @xmath252 in @xmath20 and @xmath253 at @xmath244 , we conclude that either @xmath254 or there exists @xmath233 such that @xmath255 in view of and , we obtain a contradiction with . * case 2 : sub - solution case . * we consider a function @xmath256 such that @xmath216 in @xmath20 and @xmath236 at @xmath237
. there is nothing to prove if @xmath257 and we thus assume @xmath239 .
we have to prove that @xmath258 .
we argue by contradiction and we assume that @xmath259 in order to construct a test function @xmath260 , we first consider @xmath261 the set of @xmath144 s such that @xmath262 since @xmath263 is coercive , there exists @xmath264 such that @xmath265 .
we next consider a test function @xmath266 such that @xmath267 with equality at @xmath244 and such that @xmath268 notice that for all @xmath269 , @xmath270 where for the inequality , we have in particular used the fact that @xmath263 is non - decreasing on @xmath271 . since @xmath272 in @xmath20 and @xmath253 at @xmath244
, we conclude that either @xmath273 or there exists @xmath269 such that @xmath274 in view of and , we obtain a contradiction with .
this ends the proof of the proposition .
we now prove theorem [ th::r4 ] .
let us consider a function @xmath215 such that @xmath275 with equality at @xmath244 with @xmath276 .
modifying @xmath277 if necessary , we can always assume that the supremum of @xmath278 is strict ( and reached at @xmath244 ) .
for @xmath279 , we set @xmath280 because @xmath11 is continuous at @xmath244 , we get for @xmath281 that @xmath282 where @xmath283 is a point where the supremum in @xmath284 is reached .
moreover given the components @xmath285 for @xmath286 , we can use ( [ eq::r21 ] ) in order to find @xmath287 small enough to ensure that @xmath288 .
then we have in particular the following sub - solution viscosity inequality at that point @xmath289 : @xmath290 therefore passing to the limit @xmath291 , we get @xmath292 because this is true for any @xmath293 , we finally get the sub - solution viscosity inequality at the junction : @xmath294 now the fact that @xmath295 follows from the comparison principle .
this ends the proof of the theorem .
we already mentioned that the optimal control solution of the hamilton - jacobi equation defined by plays a central role in our analysis .
we remark that for @xmath90 and @xmath296 , @xmath297 where @xmath298 more generally , keeping in mind the weak kam theory , we define the so - called _ minimal action
_ @xmath299 by @xmath300 it is convenient to extend @xmath301 to @xmath302 .
we do so by setting @xmath303 under assumptions ( a0)-(a1 ) , it is possible ( and easy ) to prove the following dynamic programming principle : for all @xmath30 and @xmath304 $ ] , @xmath305 notice that a super - optimality principle will be proved in proposition [ prop : superopt ] . before stating the main result of this section , we the following result can be considered as the core of our analysis .
the most important part of the following theorem is the fact that the minimal action is semi - concave with respect to @xmath37 ( resp .
@xmath36 ) .
[ thm : minimal - action ] @xmath301 is finite , continuous in @xmath306 and lower semi - continuous in @xmath307 .
moreover , for all @xmath308 and @xmath309 , @xmath310 , there exist two functions @xmath311 and @xmath95 such that * @xmath312 on a ball @xmath313 with equality at @xmath314 and @xmath315 * @xmath316 on a ball @xmath317 with equality at @xmath318 and @xmath319 moreover , for all @xmath320 , there exists a constant @xmath321 such that we have @xmath322 as we shall see when proving this result , we can even require equalities instead of inequalities in and .
since the proof of theorem [ thm : minimal - action ] is lengthy and technical , we postpone it until section [ sec : complete ] . when proving the main results of our paper in the next section , we also need the following lower bound on @xmath301 .
we remark that this bound ensures in particular that it is finite .
[ lem : am ] assume ( a1 ) .
then @xmath323 where @xmath324 , @xmath325 appears in ( a1 ) , @xmath326 and @xmath327 is chosen as in ( [ eq::r26 ] ) .
moreover , @xmath328 in particular , if @xmath329 , then @xmath330 as @xmath331 .
we only prove the first inequality since the remaining of the statement is elementary .
we have @xmath332 this shows that @xmath333 thus we can write for @xmath334 , @xmath335 then jensen s inequality allows us to conclude .
in this section , we investigate the uniqueness of the solution of - . in particular , we will show that the solution constructed by perron s method coincide with the function @xmath62 coming from the associated optimal control problem . in this subsection
, we will show that a super - solution satisfies a super - optimality principle .
for the sake of clarity , we first give a formal argument to understand this claim .
we consider the auxiliary function , for @xmath336 , @xmath337 and we are going to explain formally that it is non - decreasing with respect to @xmath338 as soon as @xmath11 is a super - solution of .
we call this property a super - optimality principle .
notice that this is strongly related to the fact that the quantity @xmath339 is constant in @xmath338 if @xmath11 is equal to the optimal control solution @xmath62 .
assume that the infimum defining @xmath340 is attained for some @xmath341 .
then we write @xmath342 moreover assuming @xmath301 to be smooth ( which is not the case ) , we formally get from ( [ eq::r61 ] ) the fact that @xmath343 ( at least in the case @xmath76 ) . hence @xmath344
we thus conclude that @xmath345 is non - decreasing if @xmath11 is a super - solution of .
we now give a precise statement and a rigorous proof .
[ prop : superopt ] assume ( a1 ) .
let @xmath19 be a super - solution of on @xmath20 such that there exists @xmath346 such that for all @xmath347 , @xmath348 then for all @xmath347 and @xmath349 $ ] , @xmath350 assume moreover ( a0 ) and that @xmath11 is a super - solution of ( [ eq : main])-([eq : ci ] ) on @xmath20 .
then we have @xmath351 on @xmath352 .
the proof proceeds in several steps
. * step 1 : preliminary . * notice first that from ( [ eq::r41 ] )
, we get @xmath353 using the lower semi - continuity of @xmath301 , we see that the infimum in @xmath354 of this function is then reached for bounded @xmath354 s .
moreover by lower semi - continuity of the map @xmath355 , we deduce in particular that the map @xmath356 ( and then also @xmath357 ) is lower semi - continuous . *
step 2 : the map @xmath358 is non - decreasing .
* we are going to prove that for @xmath359 , @xmath360 in the viscosity sense .
we consider a test function @xmath102 touching @xmath345 from below at @xmath361 .
there exists @xmath362 such that @xmath363 we deduce from the definition of @xmath345 that @xmath364 \le u(s , y ) - u ( \bar s , \bar y).\ ] ] by theorem [ thm : minimal - action ] , there exists a test function @xmath365 such that @xmath366 on a ball @xmath367 with equality at @xmath368 .
hence , we can rewrite the previous inequality by replacing @xmath74 with @xmath365 .
we then obtain that @xmath369 is a test function touching @xmath11 at @xmath370 from below .
since @xmath11 is a super - solution of , we have in the cases @xmath371 or @xmath372 and @xmath373 @xmath374 and in the case @xmath372 and @xmath375 @xmath376 where we used the properties of the function @xmath365 given by theorem [ thm : minimal - action ]
. * step 3 : conclusion .
* let us define for @xmath54 the following kind of lower semi - continuous envelope ( for the past in time ) @xmath377 let us notice that we have @xmath378 given a point @xmath54 , let us consider a sequence @xmath379 such that @xmath380 using lemma [ lem : am ] , we have for any @xmath381 @xmath382 therefore from the lower semi - continuity of @xmath340 , we get @xmath383 again from the lower semi - continuity of the map @xmath384 , we get passing to the lower semi - continuous envelopes in @xmath37 : @xmath385 where we have used ( [ eq::r71 ] ) .
this shows ( [ eq : superopt ] ) for @xmath386 .
this is still true for @xmath34 by definition of @xmath301 .
the proof is now complete .
[ prop : subopt ] let @xmath387 be a sub - solution of -([eq : ci ] ) on @xmath20 , such that there exists @xmath346 such that for all @xmath347 , @xmath388 then we have @xmath295 on @xmath20 . in order to prove proposition [ prop : subopt ] , we first state and prove two lemmas .
[ lem::r46 ] assume ( a0)-(a1 ) .
then the function @xmath62 defined in ( [ eq::r17 ] ) satisfies @xmath389 we first get a bound from below . using ( [ eq::r41 ] ) , we deduce ( denoting by @xmath390 the lipschitz constant for @xmath47 ) : @xmath391 with @xmath392 this implies that @xmath393 we next get a bound from above .
we have @xmath394 with @xmath395 this ends the proof of the lemma .
[ lem::r72 ] assume ( a0)-(a1 ) .
let @xmath396 be a sub - solution of -([eq : ci ] ) on @xmath20 , satisfying .
then there exists a constant @xmath397 such that @xmath398 using the lipschitz regularity of @xmath47 , we can easily consider a smooth approximation @xmath399 of @xmath47 such that @xmath400 and @xmath401 as @xmath402 .
then consider the following supremum for @xmath403 @xmath404 we claim that @xmath405 for some @xmath406 large enough independent on @xmath407 small enough .
the lemma will be obtained by letting @xmath408 and @xmath409 go to @xmath44 .
we argue by contradiction and assume that @xmath410 .
thanks to , the supremum @xmath411 is attained for some @xmath412 .
if @xmath413 , we have @xmath414 .
therefore @xmath296 and we can use the fact that @xmath11 is a sub - solution to obtain for @xmath15 @xmath415 where we have used assumption ( a1 ) to estimate @xmath416 from below . notice that we have also made use of a slight abuse of notation in the case @xmath224 .
choosing @xmath417 allows us to conclude to a contradiction .
this ends the proof of lemma [ lem::r72 ] .
we now turn to the proof of proposition [ prop : subopt ] .
the proof proceeds in several steps .
* step 1 : preliminaries .
* let us consider @xmath418 from lemmas [ lem::r46 ] and [ lem::r72 ] , we deduce that we have @xmath419 .
we want to prove that @xmath420 . to this end
, we perform the usual corrections considering the following supremum for @xmath407 @xmath421 as it is proved classically , we also have that @xmath422 as @xmath423 where @xmath424 we argue by contradiction by assuming that @xmath425 and then @xmath426 for @xmath427 small enough and fixed for the rest of the proof
. * step 2 : reduction to @xmath428 . * notice that the supremum @xmath429 is achieved for points @xmath430 . using again lemmas [ lem::r46 ] and [ lem::r72 ]
, we also deduce that @xmath431 and hence @xmath432 for @xmath408 small enough .
* step 3 : a priori bounds . * using the argument of step 1 of the proof of proposition [ prop : superopt ] , we see that there exists @xmath433 such that @xmath434 therefore we can rewrite @xmath429 as @xmath435 and the supremum is achieved for @xmath436 .
notice that this supremum looks like the classical one for proving the comparison principle for viscosity solutions , with the usual penalization term @xmath437 replaced here by the function @xmath438 . in view of the bound ( [ eq::r41
] ) from below on @xmath301 and , we derive from @xmath439 that @xmath440 where @xmath390 denotes the lipschitz constant of @xmath47 .
we conclude that there exists @xmath441 such that @xmath442 where @xmath441 depends on @xmath443 , @xmath444,@xmath445 , @xmath390 and @xmath325 .
* step 4 : getting the viscosity inequality . * since @xmath446 , we have in particular that @xmath447 by theorem [ thm : minimal - action ] , there exists a test function @xmath277 such that @xmath448 on a ball @xmath449 with equality at @xmath450 .
hence , we can rewrite the previous inequality by replacing @xmath74 with @xmath277 .
we then obtain that @xmath451 touches @xmath11 from above at @xmath452 with @xmath453 .
we use next that @xmath11 is a sub - solution of and get for @xmath454 @xmath455 where we have made use of a slight abuse of notation in the case @xmath456 . on the other hand , we have @xmath457 therefore @xmath458 on the one hand , from ( [ eq : penal ] ) , we have @xmath459 . on the other hand , we can use ( [ eq : penal ] ) and ( [ eq::r73 ] ) in order to conclude that @xmath460 for some constant @xmath461 which does not depend on @xmath408 .
we can now use the fact that the hamiltonians are locally lipschitz continuous in order to get the desired contradiction for @xmath408 small enough .
this ends the proof of the proposition . in this subsection
, we prove the main results announced in the introduction .
we simply apply propositions [ prop : superopt ] and [ prop : subopt ] and get @xmath462 which implies the result . in order to prove theorem [ th::r2 ]
, we should first prove that solutions are lipschitz continuous .
[ pro::r74 ] assume ( a0)-(a1 ) .
let @xmath11 be a solution of - on @xmath20 .
then @xmath11 is lipschitz continuous with respect to @xmath37 on @xmath20 .
we first recall ( see lemma [ lem::r18 ] ) that ( a1 ) implies ( a1 ) . we know that the solution @xmath463 given by theorem [ thm : perron ]
satisfies for some constant @xmath397 and all @xmath347 , @xmath464 from the comparison principle ( theorem [ th::r1 ] ) , we deduce that @xmath465 and then the solution @xmath466 is continuous . for @xmath467 small ( with @xmath468 ) , we now consider @xmath469 and @xmath470 this new function satisfies in particular @xmath471 . therefore @xmath53 is a sub - solution of - on @xmath472 .
we thus conclude from the comparison principle that @xmath473 , which implies @xmath474 arguing similarly , we can prove that @xmath475 . because @xmath476 can be chosen arbitrarily small ,
we conclude that @xmath11 is lipschitz continuous with respect to time on the whole @xmath20 .
since @xmath11 is a viscosity solution of , it satisfies in particular ( in the viscosity sense ) for each @xmath202 : @xmath477 this implies that there exists a constant @xmath478 such that ( in the viscosity sense ) @xmath479 this implies that @xmath11 is lipschitz continuous with respect to the space variable .
this ends the proof of the lemma .
we now turn to the proof of theorem [ th::r2 ] .
the uniqueness of the solution follows from theorem [ th::r1 ] .
the existence is obtained thanks to the optimal control interpretation ( @xmath62 is a solution ) .
the lipschitz regularity was proved in lemma [ pro::r74 ] above .
the proof of theorem [ th::r2 ] is now complete .
the fact that the solution is equal to @xmath62 follows from propositions [ prop : superopt ] and [ prop : subopt ] .
the representation formula ( [ eq::r77 ] ) follows from ( [ eq::r76 ] ) .
we start this section with the following remark : the analysis can be reduced to the case @xmath480 . precisely , using the fact that the hamiltonian does not depend on time and is positively homogeneous with respect to the state
, the reader can check that a change of variables in time yields the following [ lem : reduced ] for all @xmath481 and @xmath482 , we have @xmath483 this is the reason why we consider the reduced minimal action @xmath484 defined by @xmath485 thanks to the previous observation , it is enough to prove the following theorem in order to get theorem [ thm : minimal - action ] . [
thm : minimal - action-0 ] let us assume ( a1 ) .
then for all @xmath486 , the @xmath487 is finite .
it is continuous in @xmath488 and for all @xmath489 , there exists a function @xmath490 such that @xmath491 on @xmath488 , @xmath492 and we have @xmath493 and if @xmath494 , @xmath495 and if @xmath496 , @xmath497 moreover , for all @xmath498 , there exists @xmath321 such that for all @xmath481 , @xmath499 if @xmath500 , then we have @xmath501 .
this good case corresponds to the case where all the @xmath502 s are equal .
we can interpret lemma [ lem : am ] as follows .
assume ( a1 ) .
then @xmath503 where constants are made precise in lemma [ lem : am ] .
we are going to see that the infimum defining the minimal action can be computed among piecewise linear trajectories , and more precisely among trajectories that are linear as long as they do not reach the junction point .
this is a consequence of the fact that the hamiltonians do not depend on @xmath73 and are convex ( through jensen s inequality ) . in order to state a precise statement
, we first introduce that optimal curves are of two types : either they reach the junction point , or they stay in a branch and are straight lines .
this is the reason why we introduce first the action associated with straight line trajectories @xmath504 and the action associated with piecewise linear trajectories passing through the junction point @xmath505 where @xmath506 and @xmath507 by defining the @xmath508 s in such a way , we treat the degenerate cases : @xmath5 or @xmath496 .
indeed , @xmath509 ( resp .
@xmath510 ) measures how long it takes to the trajectory to reach the junction point ( resp . the final point @xmath73 ) from the starting point @xmath354 ( resp .
the junction point ) .
the following facts will be used several times .
[ lem : cont - ei ] the function @xmath511 ( resp .
@xmath512 ) is continuous in @xmath513 \times j^*$ ] ( resp . in @xmath514 ) .
[ lem : lsc - ei ] the function @xmath508 , @xmath515 are lower semi - continuous in @xmath516 \times j$ ] .
consider the function defined for @xmath517\times j$ ] by @xmath518 from the inequality for @xmath519 ( consequence of ( [ eq::r50 ] ) ) : @xmath520 we deduce that @xmath138 is lower semi - continuous .
consequently , the map @xmath511 is lower semi - continuous .
we proceed similarly for @xmath512 .
we first show the main lemma of this subsection .
[ lem : optim - curv ] the infimum defining the reduced minimal action @xmath80 can be computed among piecewise linear trajectories ; more precisely , for all @xmath481 , @xmath521 we write with obvious notation @xmath522 . in order to prove the lemma ,
it is enough to consider a curve @xmath523 and prove that @xmath524 to do so , we first remark that the uniform convexity of @xmath25 implies that for all @xmath525 , we have @xmath526 we now consider an admissible trajectory @xmath527 \to j$ ] and we treat different cases .
case a : @xmath528 .
we assume that a curve @xmath126 stays in one of the branch @xmath529 on the time interval @xmath530 with @xmath531 .
in such a case , we consider the curve @xmath532 with same end points @xmath533 and @xmath534 in @xmath535 but linear .
if @xmath525 is such that @xmath536 for @xmath537 and @xmath538 , we deduce from that @xmath539 case b : @xmath540 ) \subset j_i$ ] with @xmath541 . in that case , let us set @xmath542 for @xmath543 $ ] . using ( [ eq::r42 ] ) with @xmath544 and the definition of @xmath545 as a minimum of the @xmath546 s ( see ( [ eq::r26 ] ) )
, we get that @xmath547 from what we deduce that ( [ eq::r27 ] ) still holds true .
case c : the general case . by assumption , we have @xmath548)$ ] .
we then distinguish two cases .
either @xmath549)$ ] , and then we define @xmath532 as in case a. in this case , implies that @xmath550 or @xmath551)$ ] , and then we call @xmath552\subset [ s , t]$ ] the largest interval such that @xmath553 , and define @xmath532 as follows : it is linear between @xmath44 and @xmath509 , and reaches @xmath44 at @xmath509 ; it stays at @xmath44 in @xmath554 ; then it is linear in @xmath555 and reaches @xmath73 at @xmath556 .
using again the continuity of @xmath126 , we can find a decomposition of @xmath552 $ ] as a disjoint union of intervals @xmath557 ( with an at most countable union ) @xmath558=\bigcup_{k } { \mathcal i}_k\ ] ] such that for each @xmath127 , @xmath559 for some @xmath560 and @xmath137 on @xmath561 .
using case a or case b on each segment @xmath562 , we deduce that @xmath563 [ lem : dzun - cont ] the function @xmath564 is continuous in @xmath488 .
we first prove that @xmath564 is lower semi - continuous .
we know from lemma [ lem : lsc - ei ] that the function @xmath565 is lower semi - continuous for @xmath486 and @xmath566 .
therefore the function @xmath567 is also lower semi - continuous ( since the infimum is taken over a compact set ) . besides , the infimum is in fact a minimum .
we now prove that @xmath564 is upper semi - continuous at any point @xmath568 .
consider first @xmath569 ^ 2 $ ] such that @xmath570 given any sequence @xmath571 , we want to show that @xmath572 we use @xmath573 with an appropriate choice of @xmath574 .
[ [ case-1-yin - j_j - xin - j_i . ] ] case 1 : @xmath575 , @xmath576 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in this case , we choose @xmath577 and we use lemma [ lem : cont - ei ] in order to get @xmath578 and @xmath579 hence we conclude that ( [ eq::r85 ] ) holds true . [ [ case-2-y0-xin - j_i . ] ] case 2 : @xmath496 , @xmath576 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we choose @xmath580 we still have @xmath581 and we can use lemma [ lem : cont - ei ] in order to get @xmath582 we also have ( if @xmath583 ) @xmath584 hence we conclude that ( [ eq::r85 ] ) holds true .
[ [ case-3-yin - j_j - x0 . ] ] case 3 : @xmath575 , @xmath5 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we choose @xmath585 ^ 2 $ ] we still have @xmath586 and then @xmath578 ( since @xmath511 is continuous in @xmath513 \times j^*$ ] )
. we also have ( if @xmath587 ) @xmath588 hence we conclude that ( [ eq::r85 ] ) holds true .
[ [ case-4-y0-x0 . ] ] case 4 : @xmath496 , @xmath5 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we choose @xmath589 $ ] .
we deduce from and .
the function @xmath80 is continuous in @xmath488 .
since @xmath590 is lower semi - continuous , we can use lemmas [ lem : optim - curv ] and [ lem : dzun - cont ] in order to conclude that @xmath80 is lower semi - continuous . consider @xmath591 .
then either @xmath5 or @xmath496 .
moreover for @xmath592 and @xmath15 , @xmath593 therefore for each @xmath202 , we have for @xmath594 , @xmath595 therefore we have with @xmath592 , @xmath15 @xmath596 this implies that @xmath80 is continuous in @xmath488 . in view of
, we see that the study of @xmath80 can now be reduced to the study of @xmath564 . the function @xmath564 is defined as a minimum over a triangle @xmath597 ^ 2 : \tau_1 \le \tau_2 \}$ ] .
we will see below that @xmath564 is defined implicitly when the constraint @xmath598 is active ( @xmath599 ) or defined explicitly if not ( @xmath600 ) . in other words , it will be linear `` as long as '' trajectories stay some time ( @xmath601 ) at the junction point .
we first define for @xmath602 , @xmath603 the continuity of @xmath599 will be used later on .
[ lem::r101 ] the restrictions @xmath604 of @xmath599 are continuous in @xmath605 and continuous at @xmath606 if @xmath607 or @xmath608 . from lemma
[ lem : lsc - ei ] , we deduce that @xmath604 is lower semi - continuous on @xmath81 .
we now show that @xmath604 is upper semi - continuous at any point @xmath609 and also at @xmath606 if @xmath607 or @xmath608 .
we first consider @xmath610 $ ] such that @xmath611 for any sequence @xmath571 with @xmath612 , we want to show that @xmath613 arguing as in lemma [ lem : dzun - cont ] , we use @xmath614 and we choose @xmath615 as follows @xmath616 , \\ \text{if } x=0 , j\in i_0 : & \tau^k=1-|x^k| \in ( 0,1 ] , \\
\text{if } y=0 , x=0 , i\in i_0 : & \tau^k=|y^k| \in [ 0,1 ) .
\end{array}\ ] ] this ends the proof of the lemma .
we next define for @xmath617 @xmath618 where @xmath619 are defined thanks to the following function ( for @xmath620 ) @xmath621 precisely , @xmath622 when @xmath623 ( see lemma [ lem : k ] below ) .
we will see that @xmath624 plays an important role in the analysis of @xmath564 .
in particular , it allows us to define , when @xmath625 and @xmath626 , the following convex subset ( triangle ) of @xmath627 : @xmath628 it is convenient to set @xmath629 if @xmath45 or @xmath630 .
we next state a series of lemmas before proving them .
[ lemma : junc-1 ] @xmath631 [ lemma : junc-2 ] the functions @xmath632 , @xmath633 and @xmath604 are convex and @xmath84 in @xmath634 and , if @xmath635 is one of them , it satisfies for @xmath636 @xmath637 [ lemma : junc-3 ] for @xmath638 , there exists a unique @xmath639 such that @xmath640 moreover , @xmath641 [ lem::r111 ] for @xmath642 , there is a unique @xmath643 $ ] such that @xmath644 moreover @xmath645 and @xmath646 [ lem::r91 ] then we have @xmath647 with @xmath648 where @xmath649 satisfy @xmath650 moreover we have @xmath651 and @xmath652 before proving these lemmas , the reader can check the following useful properties of the function @xmath624 that will be used in their proofs .
[ lem : k ] assume ( a1 ) .
then for any @xmath653 , we have @xmath654 we define @xmath655 as the inverse of the function @xmath624 restricted to @xmath656 $ ] , and @xmath657 as the inverse of the function @xmath624 restricted to @xmath658 .
we set @xmath659 then we have @xmath660 moreover we have @xmath661 the proof proceeds in several steps .
* step 1 : first study of @xmath632 . *
let us define @xmath662 for @xmath663 , and setting @xmath664 and @xmath665 and @xmath666 , we compute @xmath667 therefore @xmath668 is in particular convex on @xmath669 . because @xmath668 is in particular lower semi - continuous on @xmath516\times [ 0,1]\times j_j^*\times j_i^*$ ] , and @xmath670 we deduce that @xmath671 this implies that @xmath632 is also convex in @xmath634 . notice that in particular @xmath672 and @xmath673 the map @xmath674 is then strictly convex on the convex set @xmath675 therefore using again ( [ eq::r96 ] ) and the lower semi - continuity of @xmath668 , we deduce that it has a unique minimum that we denote by @xmath554 satisfying @xmath676 . * step 2 : study of @xmath604 .
* let us consider the following function @xmath677 for @xmath678 , setting @xmath679 and proceeding similarly as in step 1 , we can deduce that @xmath680 and that @xmath604 is also convex on @xmath634 .
we can also deduce that the map @xmath681 is strictly convex on @xmath682 for @xmath638 and that it has a unique minimum that we denote by @xmath678 such that @xmath683 using the derivative with respect to @xmath684 , we see that @xmath684 is characterized by the equation @xmath685 moreover @xmath686 using the regularity @xmath687 of @xmath688 given in assumption ( a1 ) , we see that the unique solution @xmath689 of @xmath690 is continuously differentiable with respect to @xmath568 . therefore we deduce that @xmath691 .
we have @xmath692 writing @xmath684 for @xmath693 , and using the optimality condition , we get @xmath694 where we have used ( [ eq::r98 ] ) in the second and in the fourth line .
hence @xmath604 satisfies ( [ eq::r99 ] ) on @xmath695
. * step 3 : further study of @xmath632 .
* we concluded at the end of step 1 that for @xmath638 we have @xmath696 with @xmath676 .
then we can distinguish two cases . * case 1 : @xmath697 .
* in that case this implies that @xmath698 which can be written as @xmath699 with @xmath649 defined in ( [ eq::r97 ] ) . using lemma [ lem : k ] , we conclude that holds true if and only if @xmath700 and @xmath701 ; i.e. @xmath702 . in this case
we have @xmath703 and @xmath704 and then @xmath705 moreover , we have in this case @xmath706 . using legendre - fenchel s equality together with @xmath707 and @xmath708
, we have @xmath709 and @xmath710 hence @xmath633 satisfies ( [ eq::r99 ] ) on @xmath634 .
finally we deduce from that the condition : @xmath711 is equivalent to @xmath712 and then by continuity of @xmath632 and @xmath633 , we get @xmath713 * case 2 : @xmath714 . *
if for @xmath636 we have @xmath715 with @xmath714 , then we have seen that @xmath716 and @xmath717 . from lemma [ lem::r101 ]
, we also have that @xmath718 if @xmath607 or @xmath608 and in that case @xmath629 . on the other hand
, we have @xmath719 if @xmath720 with @xmath721 in that case .
therefore in all cases we have @xmath722 now from the continuity of @xmath564 , we deduce that @xmath723 * step 4 : on the boundary @xmath724 .
* we already know that @xmath564 is continuous , therefore if @xmath725 : @xmath726 on the other hand , recall that @xmath638 , the real @xmath678 is characterized by ( [ eq : cno ] ) , i.e. @xmath727 notice that if we choose @xmath728 we deduce from @xmath729 that @xmath730 which are obvious solutions of ( [ eq::r103 ] ) .
therefore we conclude that this is the solution . using ( [ eq::r32])-([eq : dzo - dir ] ) and
the expression of @xmath633 , ( [ eq::r104 ] ) implies the equality of the gradients of @xmath633 and @xmath604 on the boundary @xmath731 .
finally this shows that @xmath732 .
this ends the proof of the lemmas .
the proof proceeds in several steps .
* continuity of @xmath443 .
* we set for @xmath733\times j_j\times j_i$ ] @xmath734 from proposition [ lemma : junc-1 ] , we already know that there exists a unique @xmath735 $ ] such that @xmath736 on the other hand , we have @xmath737 notice that in cases 1 and 2 , there is a unique @xmath735 $ ] such that @xmath738 and @xmath739 in case 1 , @xmath740 $ ] in case 2
. then the continuity of @xmath741 in @xmath742 follows from the lower semi - continuity of @xmath743 on @xmath516\times j_j\times j_i$ ] and the uniqueness of @xmath684 such that ( [ eq::r112 ] ) holds .
* computation of @xmath443 .
* we distinguish cases .
case 1 : @xmath744 .
notice that we have @xmath745 if @xmath746 , then @xmath747 and @xmath748 . if @xmath749 , then @xmath750 is a solution of @xmath751 and @xmath752 .
case 2 : @xmath753 .
notice that we have @xmath754 if @xmath755 , then @xmath756 and @xmath757 . if @xmath758 , then @xmath759 is a solution of @xmath760 and @xmath761 .
this ends the proof of the lemma .
the proof proceeds in several steps . * step 1 : continuity . * from proposition
[ lemma : junc-1 ] , we already know that @xmath762 and ( [ eq::r106 ] ) holds true with @xmath763 where @xmath741 in the first line . therefore , in order to prove that @xmath764 , it is sufficient to prove that if @xmath765 , and if @xmath766 is a sequence of points such that @xmath571 , then we have with @xmath767 @xmath768 where @xmath769 is given by ( [ eq::r109 ] ) .
let us recall that @xmath615 is characterized by the equation @xmath770 we will assume ( up to extract a subsequence ) that @xmath771 for some limit @xmath772 $ ] . because we have @xmath773 , it is easy to deduce from ( [ eq::r107 ] ) , that there exists a constant @xmath774 such that @xmath775 this can be proved by contradiction , distinguishing the cases @xmath776 , @xmath777 and @xmath778 . up to extract a subsequence , we can then pass to the limit in ( [ eq::r107 ] ) and get @xmath779 in the following cases , we now identify one of the two quantities @xmath780 or @xmath781 , the other one being determined by ( [ eq::r113 ] ) .
* case 1 : @xmath782 . * from lemma [ lem::r111 ] , we know that @xmath783 , and then @xmath784 and from ( [ eq::r116 ] ) , we get @xmath785 * case 2 : @xmath786 . * from lemma [ lem::r111 ] , we know that @xmath787 , and then @xmath788 and from ( [ eq::r116 ] ) , we get @xmath789 * case 3 : @xmath790 . * this case only occurs if @xmath607 or @xmath608 .
moreover at least one of the two quantities @xmath791 and @xmath792 tends to zero .
if @xmath793 , then @xmath794 and hence @xmath795 if @xmath796 , then @xmath797 and hence @xmath798 this implies that @xmath799 by the uniqueness of the limit , this finally shows that @xmath800 and ( [ eq::r109 ] ) holds .
* step 2 : checking ( [ eq::r116 ] ) and ( [ eq::r115 ] ) . * from ( [ eq::r118 ] ) and ( [ eq::r119 ] ) , we deduce ( [ eq::r116 ] ) on @xmath801 . from @xmath802 on @xmath803 , we deduce that ( [ eq::r116 ] ) is also true on @xmath804 .
then ( [ eq::r115 ] ) follows from a simple computation for @xmath805 .
this is still true for @xmath806 , because @xmath632 is @xmath84 .
this ends the proof of the lemma .
the following lemma will be used below . since it is elementary ,
its proof is omitted .
[ lem::r123 ] for @xmath807 , we have for @xmath808 with @xmath805 if @xmath809 : @xmath810 and @xmath811 we are now in position to prove theorem [ thm : minimal - action-0 ] .
we prove several lemmas successively .
[ lem : dzo-1 ] for @xmath808 , we have @xmath812 in particular @xmath813 in the first two cases .
[ lem : dzo-1bis ] in the case @xmath814 , we have @xmath815 moreover , in this case there exists a curve @xmath816 such that @xmath817 .
this curve connects @xmath818 and @xmath819 and is contained in @xmath820 the results of these two lemmas are illustrated in figures [ fig2 ] and [ fig4 ] .
we only have to treat the case @xmath821 .
the convexity of @xmath25 implies that for @xmath678 : @xmath822 therefore for @xmath638 with @xmath823 , we have @xmath824 when @xmath825 , we have @xmath826 , and then @xmath827 for @xmath638 and then also for @xmath828 , by continuity of the functions . we first prove and then describe the curve @xmath829 . [ [ proof - of - eqr120 . ] ]
proof of ( [ eq::r120 ] ) . + + + + + + + + + + + + + + + + + + + + + + combining and , we obtain @xmath830 on the other hand , we have ( a strictly convex function being above its tangent ) for @xmath831 and @xmath832 @xmath833 this shows that @xmath834 we see that ( [ eq::r121 ] ) and ( [ eq::r122 ] ) imply ( [ eq::r120 ] ) . [ [ description - of - leftmathcald_mathrmlinearjimathcald_mathrmstraightjirightcap - deltaji . ] ] description of @xmath835 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + notice that @xmath836 this means that the functions @xmath633 and @xmath837 coincide at the two points @xmath819 and @xmath838 .
therefore we have @xmath839x_i , y_j[\ ] ] because @xmath633 is linear and @xmath837 is strictly convex as a function of @xmath840 .
+ the function @xmath841 being convex because @xmath842 is linear , we can consider the convex set @xmath843 then for @xmath814 , the set @xmath844 is contained in the boundary of the convex set @xmath845 .
more precisely , we have @xmath846 which shows that @xmath816 is a curve and @xmath847 [ lem : dzo-2 ] for all @xmath848 , and @xmath849 where @xmath83 is @xmath84 : @xmath850 moreover for all @xmath851 ( with @xmath852 if @xmath809 ) @xmath853 and for all @xmath854 ( with @xmath855 if @xmath809 ) @xmath856 we also have @xmath857 for all @xmath858 except for @xmath859 for which we exclude points @xmath860 .
moreover for @xmath861 , we have @xmath862 and @xmath809 , we have @xmath863 using proposition [ lemma : junc-1 ] for @xmath632 , lemma [ lem::r123 ] for @xmath837 , and ( [ eq::r130 ] ) for @xmath633 and the property ( [ eq::r120 ] ) , we get @xmath864 which implies ( [ eq::r131 ] ) .
similarly we get @xmath865 which implies ( [ eq::r132 ] )
. relations ( [ eq::r133 ] ) and ( [ eq::r134 ] ) follow both from proposition [ lemma : junc-1 ] and lemma [ lem::r123 ] .
finally ( [ eq::r140 ] ) and ( [ eq::r141 ] ) follows from the previous results .
this ends the proof of the lemma .
we now can check the equations satisfied by @xmath80 at the boundary .
[ lem : dzo - bdry-1 ] at any point @xmath866 with @xmath867 if @xmath868 , we have for any @xmath869 @xmath870 [ lem : dzo - bdry-2 ] at any point @xmath871 with @xmath832 if @xmath872 , we have for any @xmath873 @xmath874 we first remark the general fact that @xmath875 on the one hand , from lemma [ lem : dzo-1 ] , we have for points @xmath866 where @xmath876 is @xmath84 @xmath877 and this common quantity is independent on @xmath127 . therefore to conclude to ( [ eq::r136 ] ) in the case @xmath76 ,
it is enough to show that there exists at least an index @xmath127 such that @xmath878 [ [ case-1 ] ] case 1 : + + + + + + + @xmath76 and @xmath879 .
then we have @xmath880 and then @xmath881 therefore holds true for @xmath882 .
[ [ case-2 ] ] case 2 : + + + + + + + @xmath372 and @xmath883
. then we have @xmath884 and by lemma [ lem::r123 ] , we have for @xmath885 which is in particular true for @xmath496 .
this shows ( [ eq::r136 ] ) in the case @xmath372 . from lemma [ lem : dzo-1 ]
, we have for points @xmath871 where @xmath886 is @xmath84 @xmath887 and this common quantity is independent on @xmath127 . therefore to conclude to ( [ eq::r137 ] ) , it is enough to show that there exists at least an index @xmath127 such that @xmath888 * case 1 : * @xmath607 .
then from lemma [ lem : dzo-1 ] , we have with @xmath889 @xmath890 therefore ( [ eq::r137 ] ) holds true for @xmath889 .
* case 2 : * @xmath872 .
we distinguish sub - cases .
subcase 2.1 : @xmath891 . from lemma [ lem :
dzo-1 ] , we still have ( [ eq::r142 ] ) with @xmath889 , which again implies ( [ eq::r137 ] ) for @xmath889 .
subcase 2.2 : @xmath758 .
then we choose an index @xmath892 , and lemma [ lem : dzo-1 ] implies that @xmath893 which again implies ( [ eq::r137 ] ) for such @xmath894 .
this ends the proof of the lemma
. we can now prove theorem [ thm : minimal - action-0 ] . from lemma [ lem : dzo-1 ] ,
we know that @xmath80 has the regularity @xmath895 except on certain curves @xmath896 for @xmath809 .
so if @xmath568 is a point of local @xmath895 regularity of @xmath80 , then we simply set @xmath897 if @xmath568 is a point where @xmath80 is not @xmath895 , then we have @xmath898 , and we can simply set @xmath899 the required equalities follow from lemmas [ lem : dzo-1 ] , [ lem : dzo - bdry-1 ] , [ lem : dzo - bdry-2 ] .
estimate ( [ eq : grad - control ] ) follows from the fact that @xmath80 is the minimum of @xmath900 and of functions in @xmath901 for some @xmath150 .
this ends the proof of the theorem .
this section contains classical results from viscosity solutions , whose statements are adapted to the equation studied in the present paper . * consider a family of sub solutions ( resp .
super - solutions ) @xmath902 of on @xmath20 such that the u.s.c .
( resp . l.s.c . )
envelope @xmath11 of @xmath903 is finite everywhere .
then @xmath11 is a sub - solution ( resp .
super - solution ) of on @xmath20 . * consider a family of sub solutions ( resp .
super - solutions ) @xmath904 of on @xmath20 such that the upper ( resp .
lower ) relaxed semi - limit @xmath11 is finite everywhere .
then @xmath11 is a sub - solution ( resp .
super - solution ) of on @xmath20 . the first author is partially supported by the kamfaible project ( anr-07-blan-0361 ) .
the authors thank a. briani for useful discussions and important indications on the recent literature on the subject .
the authors are very grateful to j .- p .
lebacque for fruitful and enlightening discussions on the modeling of traffic flows and indications on the related literature .
the second author thanks m. khoshyaran for discussions on the topic .
the authors also thank y. achdou , g. barles , p. cardaliaguet , e. chasseigne , h. frankowska and m. garavello for discussions about this work .
, _ optimal control and viscosity solutions of hamilton - jacobi - bellman equations _ , systems & control : foundations & applications , birkhuser boston inc . , boston , ma , 1997 . with appendices by maurizio falcone and pierpaolo soravia . height 2pt depth -1.6pt width 23pt , _ solutions de viscosit des quations de hamilton - jacobi _ , vol .
17 of mathmatiques & applications ( berlin ) [ mathematics & applications ] , springer - verlag , paris , 1994 .
height 2pt depth -1.6pt width 23pt , _ first order macroscopic traffic flow models : intersection modeling , network modeling _ , in transportation and traffic theory .
flow , dynamics and human interaction , elsevier , 2005 , pp .
365386 . height 2pt depth -1.6pt width 23pt , _ optimality principles and representation formulas for viscosity solutions of hamilton - jacobi equations
. i. equations of unbounded and degenerate control problems without uniqueness _ , adv .
differential equations , 4 ( 1999 ) , pp . 275296 . height 2pt depth -1.6pt width 23pt , _ optimality principles and representation formulas for viscosity solutions of hamilton - jacobi equations .
equations of control problems with state constraints _ , differential integral equations , 12 ( 1999 ) , pp . | this paper is concerned with the study of a model case of first order hamilton - jacobi equations posed on a `` junction '' , that is to say the union of a finite number of half - lines with a unique common point .
the main result is a comparison principle .
we also prove existence and stability of solutions .
the two challenging difficulties are the singular geometry of the domain and the discontinuity of the hamiltonian . as far as discontinuous hamiltonians are concerned , these results seem to be new .
they are applied to the study of some models arising in traffic flows .
the techniques developed in the present article provide new powerful tools for the analysis of such problems . _ dedicated to j .- b .
hiriart - urruty _ * keywords : * hamilton - jacobi equations , discontinuous hamiltonians , viscosity solutions , optimal control , traffic problems , junctions * msc : * 35f21 , 35d40 , 35q93 , 35r05 , 35b51 |
the @xmath2-ray lightcurve of grb 051109b shows a single soft peak , without a strong emission above 100 kev ( see fig .
[ fig : lc ] , left panel ) .
the estimated duration in the 15350 kev energy band is t@xmath3=15@xmath41 s @xcite .
a faint afterglow was detected by the x - ray telescope ( xrt ) @xmath086 s after the burst and monitored for the following 6 days .
although the burst location was promptly observed by ground - based telescopes ( e.g. at t+52 s by the rotse - iiib telescope ) @xcite and by the uvot ( at t+84 s ) @xcite , no optical counterpart was detected .
a nearby ( z=0.08 ) barred spiral galaxy has been proposed by @xcite as the putative host galaxy .
the field of grb 051109b was then re - observed by swift on aug 2006 but no late - time x - ray emission was seen in the 17 ks exposure pc image with a 3@xmath5 upper limit of 9.8@xmath610@xmath7ctss@xmath8 .
[ fig : lc ]
the time averaged spectrum of the prompt emission is well fit by a simple power law with a photon index of [email protected] . a cut off powerlaw or a band model
do not improve significantly the fit , however fixing the first index @xmath9 of the band function @xmath10 , we put an upper limit to the e@xmath11 energy of 70 kev at the 90% confidence level .
the fluence in the 15150 kev energy band is @xmath02.1@xmath610@xmath12erg@xmath13 and , using the redshift value of 0.08 , we put a lower limit of @xmath03@xmath610@xmath14erg to the isotropic energy . extrapolating the power law model in the whole 1 - 10@xmath15 kev ( rest frame ) band we computed an upper limit of [email protected]@xmath610@xmath17erg .
the average xrt spectrum can be modeled with an absorbed power law of @xmath18 @xmath13 and a photon index @xmath19=@xmath20 .
the fluence emitted by the afterglow in the x - ray band is @xmath03.5@xmath610@xmath12erg@xmath13 , comparable to that of the prompt emission .
the combined bat and xrt lightcurve , shown in fig.[fig : lc ] , displays many of the features seen by swift for other grbs .
the x - ray hardness ratio ( fig.[fig : lc ] ) shows a spectral evolution between the three standard phases of the afterglow emission , in particular a hardening related to the small flare peaking at @xmath0t+1500 s. from spectral and timing analysis we derived a spectral index @xmath21 of the radiating electrons of @xmath02.1 for @xmath22@xmath23@xmath24 .
due to the narrow energy band of the swift bat , it is hard to costrain low peak energy values , however the soft spectral index of the prompt emission suggest grb 051109b can be identified as a xrr burst .
the x - ray lightcurve has a well known shape : an initial fast decay in the first 200 s , interpreted as off - axis prompt emission ; a plateau , attributed to energy injection into the afterglow ; a final decay of index @xmath01 in agreement with standard afterglow models .
using our best fit model and the data reported in @xcite , corrected for extinction , we found f@[email protected]@xmath610@xmath26erg@xmath13s@xmath8 and a f@xmath27@xmath011 @xmath28jy at t+11 hr . we verified that grb 051109b lies well above the region of dark bursts defined in @xcite . | we present swift observations of grb 051109b , a soft long burst triggered by the burst alert telescope ( bat ) .
the soft photon index of the prompt emission suggest it is a x - ray flash ( xrf ) or , at least , a x - ray rich ( xrr ) burst .
the x - ray lightcurve displays the canonical shape of many other grbs , a double broken power law with a small flare superimposed at @xmath0t@xmath1 + 1500 s , and its extrapolation to early times smoothly joins with the bat lightcurve . on the basis of the derived optical to x - ray flux ratio
, it can not be classified as a dark burst . |
gamma - ray bursts ( grbs ) are the most energetic transient events in the universe . the most accepted model to explain
this phenomenon is the fireball " model ( e.g. * ? ? ?
* ; * ? ? ?
* ) and the progenitor of long grbs are believed to be massive stars on the order of 30 solar masses or more .
the isotropic energies inferred from the observed fluences are often of order @xmath5 ergs . on the other hand ,
typical supernova explosions are of the order of only @xmath6 ergs .
one way to solve this energy problem " is by assuming that the radiation is collimated into a jet .
one prediction of the fireball " model is that the afterglow decay rate increases when the relativistic beaming angle equals or exceeds the physical jet opening angle as the jet decelerates in the surrounding medium ( e.g. * ? ? ?
* ; * ? ? ?
this can be seen as an achromatic jet break in the light curve , with a typical decay slope after the jet break of order @xmath1 = 2 ( e.g. zhang et al .
2006 ; nousek et al .
the time of this jet break can be used to infer the jet opening angle by e.g. the relation given by @xcite .
the measurement of the jet break time is therefore most critical for understanding the energetics of grbs .
before the launch of the _ swift _ gamma - ray burst explorer mission @xcite , putative jet breaks were found at optical or radio wavelengths , typically a few days after the burst ( e.g. * ? ?
since its launch , _ swift _ has detected roughly 400 bursts ( end of 2008 ) and typically observes them for up to one or two weeks after the trigger .
for the majority of these bursts , jet breaks have not been detected ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
however , one reason could be that jet breaks occur in x - rays at much later times than previously thought and the follow - up observations by _ swift _ are not late enough in time to detect a jet break , as studies by @xcite and @xcite suggest . as an alternative , @xcite suggested that the jet breaks are hidden and the light curves are mis - interpreted as a single power law decay although a jet break is there .
only a handful of afterglows have been followed for more than a month by _ swift _ , because usually the x - ray afterglow fades below the _ swift _ x - ray telescope ( xrt ; * ? ? ?
* ) detection limit @xmath7 ergs s@xmath8 @xmath9 roughly a week after the trigger .
one of these exceptions is the bright x - ray afterglow of grb 060729 which was detected by _ swift _
xrt even 125 days after the burst @xcite .
the _ swift _ burst alert telescope ( bat ; * ? ? ? * ) triggered on grb 060729 on 2006 july 29 , 19:12:29 ut @xcite and a redshift of @xmath10 was measured @xcite .
the xrt and the uv / optical telescope ( uvot ; * ? ? ?
* ) , started observing the burst about 2 minutes after the trigger .
the _ swift _ uvot was able to follow this afterglow in the uvw1 filter up to 31 days after the bat trigger . in x - rays
_ swift _ s xrt was still detecting the x - ray afterglow of grb 060729 at the end of 2006 november , 125 days after the burst @xcite .
however , by 2006 december the _
swift_-xrt detection limit was reached and only a 3@xmath11 upper limit could be given for the 63.5 ks exposure time obtained in december 2006 . by that time
the x - ray afterglow of grb 060729 did not show any clear evidence for a jet break , giving a lower limit on the jet opening angle of @xmath12 @xcite based on the assumption of a constant circumburst medium . in order to extend the light curve of this exceptional x - ray afterglow , we observed it five times with _
chandra _
acis in 2007 and 2008 .
we report on the detections of the x - ray afterglow of grb 060729 @xcite with _
chandra _ up to 642 days after the burst - the latest detection ever of an x - ray afterglow of a grb at cosmological distance .
previously the burst with the latest detection of an x - ray afterglow was grb 030329 @xcite , which had a detection 258 days after the burst by _ xmm - newton_. our paper is organized as follows : in [ observe ] the observations and data reduction are explained , in [ results ] the measurements of the x - ray light curve are shown , and in [ discuss ] we discuss the implications of this light curve . throughout this paper
the x - ray flux dependence on time and frequency is defined as @xmath13 .
luminosities are calculated assuming a @xmath14cdm cosmology with @xmath15 = 0.30 , @xmath16 = 0.70 and a hubble constant of @xmath17 = 71 km s@xmath8 mpc@xmath8 corresponding to a luminosity distance @xmath18 = 3064 mpc for grb 060729 .
_ chandra _ observed grb 060729 three times between 2007 march 16 and 2007 june 30 .
two very late - time _
chandra _
observations were performed in 2007 december/2008 january for 72.7 ks and in 2008 april / may for 117.3 ks . due to pitch angle constraints some of these had to be split into several visits .
all observations , with start and end times and exposure times , are listed in table[xray_log ] .
all of these observations were performed with the standard 3.2s readout time in very faint mode on the on - axis position on the back - illuminated acis - s3 ccd .
data reduction was performed with the _ chandra _ analysis software ciao version 4.0 and the calibration database caldb version 3.4.3 . in order to reduce the acis particle background all _ chandra _
stage 1 event data were reprocessed using ciao acis_process_events with the vf mode cleaning . only acis grades 0 , 2 , 3 , 4 , and 6 were selected for further analysis .
the background was further reduced by using only photons in the 0.5 - 8.0 kev energy range . before further analysis ,
the observations were combined into one event file each using the ciao task _ merge_all_. source photons were selected in a circle with a radius @xmath19 = 1@xmath20 and background photons in a close - by source - free region with a radius @xmath19 = 10@xmath20 .
count rates were converted into fluxes by using pimms version 3.9b using the parameters from the spectral fits to the _ swift _ data after the break at 1 ms after the burst ( see below ) with an absorption column density @xmath21 @xmath9 and an x - ray spectral index @xmath22= 0.89 .
a description of the reduction and analysis of the _ swift _ data can be found in @xcite . for display purposes and fitting the late - time light curve we rebinned the _ swift _
xrt photon counting data with 250 counts per bin for the times up to 2 ms after the burst and 100 counts per bin for the times thereafter .
spectral analyses were performed for the times 300 - 800 ks and @xmath23 1 ms after the burst .
source photons were collected in a circle with @xmath24 and background photons with @xmath25 with grade selection 0 - 12 . the response matrix _
swxpc0to12s0_20010101v010.rmf _ was used .
the spectra were rebinned with 20 counts per bin for the 300 - 800 ks after the burst spectrum and 15 counts per bin for the spectrum with @xmath26 1 ms .
the spectra were analyzed with xspec version 12.4.0x @xcite . to search for changes in the x - ray spectrum we applied a hardness ratio studythe hardness ratio
is defined as hr=(h - s)/(h+s ) , where s and h are the observed counts in the 0.3 - 2.0 and 2.0 - 10.0 kev energy bands , respectively . ]
segment by segment .
because of the low - number statistics in some of the later segments of the _ swift _ xrt and all _ chandra _ observations , we applied bayesian statistics to determine the hardness ratios as described by @xcite .
the late - time x - ray light curve including the five _ chandra _ pointings is shown in figure[xray_lc ] .
we ignore the first day of the _ swift _ observation because it is not relevant for the study of the late - time light curve .
the early light curve and a detailed discussion can be found in @xcite . the late - time light curve ( figure[xray_lc ] )
was fitted by several power law and multiply - broken power law models as listed in table[lc_fits ] .
fitting the light curve with the decay slope @xmath27 is the steep decay slope after the plateau phase . ] fixed to 1.32 , as reported by @xcite from the _ swift _ data , gives a very poor result ( @xmath28 = 897/46 ) .
the fit can be improved by leaving the decay slope as a free parameter ( table[lc_fits ] , model 2 ) .
this results in a single decay slope of @xmath27 = [email protected] , but the light curve still deviates significantly at later times from this slope , resulting in an unacceptable @xmath28 = 400/45
. a broken power law fit to the entire late - time light curve ( model 3 ) reveals a break at about 2 ms after the burst ; in contrast to the result of @xcite , in which we could fit the late - time _ swift _
data with just one decay slope , the addition of the 2007 _ chandra _ data requires a break in the _ swift _ data .
the late decay slope , @xmath30 = 1.85 , is driven by the last two _ chandra _ observations , while the @xmath31 is also strongly affected by two very high data points at @xmath32 ms and @xmath33 ms . making the assumption that these two high points are late - time x - ray flares unrelated to the afterglow of the external shock , we removed them from further fits ( see models 4 and 5 ) .
we then fit the data between 100 ks and 30 ms with a broken power law ( model 6 ) , obtaining a break time of @xmath34 ms and slopes of @xmath35 and @xmath36 .
this fit is plotted as the dashed line in figure[xray_lc ] .
the last two _ chandra _ observations deviate from this fit , suggesting a break at about a year after the burst . because these last two points have very few counts ( and consequently large uncertainties )
, a broken power law fit to the late - time light curve can not constrain either the break time or the late - time decay slope @xmath37 unless at least one parameter is fixed .
we therefore approached the question of a final break in steps .
a single power law fit to the light curve for @xmath38 ms ( table[lc_fits ] , model 7 ) gives @xmath30 = [email protected] and @xmath28 = 12/15 .
although this is already an acceptable fit , we investigated the possibility of a late - time break which is expected from grb theory ( c.f .
e.g. * ? ? ?
* ; * ? ? ?
at first we fitted the light curve for @xmath39 ms @xmath40 35 ms with a single power law ( model 8) which results in @xmath30 = @xmath41 and @xmath28 = 6/13 .
we then fitted a broken power law model to the entire light curve with @xmath23 1.2 ms with @xmath30 fixed at 1.61 ( the best - fit result when the last two _ chandra _ observations are excluded ; model 8 in table[lc_fits ] ) to determine whether the data require a very late break in the light curve slope .
this fit ( table[lc_fits ] , model 9 ) gives @xmath42 = 41.3@xmath43 ms and @xmath44 , with @xmath45 .
figure[contour ] displays the contour plot between the final break time and the final slope .
it shows that they are still not well - constrained .
although the best - fit break time is 41 ms ( 2007 november ) , a break as early as @xmath46 ms , with a late - time decay slope of @xmath37 = 2.5 , is consistent with the data at the @xmath47 level .
in addition to the broken power law fits with a sharp break , the late - time light curve was also fitted by the smoothed double - broken power law model defined by @xcite .
here we found decay slopes @[email protected] , @[email protected] , and @[email protected] with break times at [email protected] ms and 20 ms ( fixed ) .
the smooth parameter is fixed to 3.0 and 2.0 for the breaks at about 1 ms and 20 ms , respectively .
this results in an acceptable fit with @xmath28 = 42/32 .
possible interpretations of these temporal breaks are discussed in
[ discuss ] .
temporal breaks are often associated with spectral breaks ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
figure[lc_cr_hr ] displays the _ swift _
xrt count rate and hardness ratio light curves for the interval between 100 ks and 5 ms after the burst .
the hardness ratios are plotted segment by segment .
while the hardness ratios before the break at 1 ms after the burst are of order hr @xmath48 , after the break the spectrum hardens to hr @xmath49 with even harder values at later times .
the spectrum before the 1 ms break can be fitted with a single absorbed power law with @xmath50 @xmath9 and an energy spectral slope @xmath51 = [email protected] ( @xmath28 = 81/82 ) .
the spectrum after the 1 ms break was also fitted by an absorbed single power law model . leaving the absorption column density as a free parameter , however , results in an increase of the column density , which does not seem plausible .
therefore we fixed the absorption column density to @xmath52 @xmath9 , the value obtained before the break .
this fit results in a slightly flatter energy spectral slope @xmath51 = [email protected] .
these values were used in pimms to convert the _
chandra _
acis - s count rates into the fluxes given in table[xray_log ] and plotted in figure[xray_lc ] .
the _ chandra _ data must be analyzed in the poisson limit , complicating proper analysis of possible spectral variations at very late times . using the bayesian approach described by @xcite
, we estimated the hardness ratios in the _ chandra _ data and their uncertainties , both before and after the break at 38 ms .
we obtain mean values of hr=@xmath53 for the 2007 march - june data ( before the final break ; 38 counts total ) and hr=@xmath54 for the very late data ( after the final break ; 8 counts total ) , with 85% confidence limits of hr@xmath55 to @xmath56 and hr@xmath57 to @xmath58 , respectively .
although this is a suggestion of spectral softening across the final break , we can not exclude ( at the 85% confidence level ) the possibility that the hardness ratio is constant .
note that due to the different energy bands and detector response matrices it is not possible to compare the _ swift _ and _ chandra _ hardness ratios directly . even though grb 060729 was one of the brightest bursts detected in x - rays , it is not the brightest one so far seen during the _ swift _ mission .
grb 061121 was about 2 - 3 times brighter when the _
swift_-xrt started observing it @xcite , but by a day after the burst it was an order of magnitude fainter than grb 060729 and was not detected after 2 ms post - burst .
the second brightest x - ray afterglow so far seen by _ swift _ was grb 080319b ( racusin et al . , 2008 ) .
figure [ grbs060729_061121 ] displays the observed count rate light curves of grbs 060729 , 061121 , and 080319b .
even though grbs 061121 and 080319b appear to be much brighter than grb 060729 until about 20 ks after the trigger , the long plateau phase in grb 060729 makes it the brightest x - ray afterglow about half a day after the trigger .
grb 061121 already displays a break from the plateau to the second steep decay phase at 2.2 ks after the burst followed by an even steeper decay at 30 ks with a decay slope of @xmath1 = 1.5 ( page et al . , 2007 ) .
this earlier break compared to grb 060729 , which broke at about 60 ks after the burst , and the steeper decay slope of 1.5 compared to 1.3 in grb 060729 made grb 061121 disappear much faster than grb 060729 .
the naked - eye " burst 080319b on the other hand does not even show a noticeable plateau phase and decays rather quickly with late - time decay slopes of @xmath59= 1.17 and 2.61 before and after the late - time jet break at @xmath60 s. we can ask how the intrinsic rest - frame 2 - 10 kev luminosity light curve of grb 060729 compares to the bursts shown in figure 1 in @xcite .
figure[grbs_lc_l_t ] displays grbs 060729 , 061121 , and 080319b with most of the bursts shown in figure1 in @xcite in the rest - frame .
the plot shows that during the plateau phase grb 060729 was not that luminous . as a matter of fact
it was a factor of about 10 to 100 less luminous than bursts such as grbs 061121 or 080319b .
nevertheless after a few days in the rest - frame , grb 060729 becomes the most luminous x - ray afterglow in the 2 - 10 kev band . compared with other grbs shown in figure[grbs_lc_l_t ] ,
the total energy output in the apparent rest - frame 2 - 10 kev band of @xmath61 ergs makes grb 060729 one of the most energetic x - ray afterglows ever detected . in the 2 - 10 kev band , only grbs 061121 and 080319b appear to be more energetic with @xmath62 ergs and @xmath63 ergs , respectively . however , if we attempt to correct for the jet opening angle , the picture changes . with jet opening angles of 4@xmath64 and 0.4@xmath64 as inferred for grbs 061121 and 080319b , respectively @xcite , the beaming - corrected energies are @xmath65 and @xmath66 ergs , respectively . for grb 060729 , however , assuming a jet half opening angle of 14@xmath64 ( see next section ) the beaming corrected energy is still @xmath67 ergs in the rest frame 2 - 10 kev band .
while our original _ swift _ xrt light curve
after the plateau phase seems to be consistent with a single power law decay @xcite , the _ chandra _ data make it apparent that there had to be a break at about 1 ms after the burst .
we showed above that this break coincides with a significant change in the x - ray spectral slope from @xmath22= 1.2 before and 0.9 after that break .
unfortunately , as mentioned in @xcite , _ swift _ could only follow the afterglow in the uvot with the w1 filter up to a month after the burst due to bright stars in the uvot field which caused some scatter in the uvot and an enhanced background at the position of grb 060729 . according to model 9 ( excluding the two x - ray flare points at 2 ms and 5 ms ) in table 2 ,
the late time x - ray afterglow is described by @xmath68 , where ( 1 ) @xmath69 , @xmath70 for 105 ks @xmath71 ms ; ( 2 ) @xmath72 , @xmath73 for @xmath74 s ; ( 3 ) @xmath44 for @xmath75 .
the closure relation for 0.1 ms @xmath76 ms results in @xmath77 , consistent with both the ism and wind models with theoretical expectation of @xmath78 if @xmath79 max@xmath80 , so @xmath81 and @xmath82 .
the power law index of the electron energy distribution is @xmath83 ( [email protected] ) , derived from @xmath27(@xmath84 ) .
the closure relation for @xmath85 ms @xmath86 ms results in @xmath87 , moderately consistent with the theoretical expectation of @xmath88 if the environment is a free wind or 0.0 , which can be excluded at the 5@xmath11 and 2@xmath11 confidence level , respectively . even if we adopt @xmath89 , the derived @xmath90 from the temporal / spectral index is inconsistent with the @xmath91 value derived with @xmath27 and @xmath84 at the earlier stage . ] and the spectral regime is @xmath92 , so @xmath93 and @xmath94 .
in other words , the break at @xmath95 ms can be interpreted as a cooling break in the wind medium scenario in which the cooling frequency @xmath96 crosses the x - ray band .
the power law index of the electron energy distribution derived from the value of @xmath30 ( the spectral index of this epoch has a relatively large uncertainty ) is @xmath97 , quite consistent with the value derived during the previous epoch .
therefore , a wind model is preferred from the observations before @xmath98 ms , breaking the degeneracy of the wind and ism models which are both consistent with the earlier data @xcite .
the x - ray afterglow light curve for @xmath99 s excluding the two flares can be also well fitted by a single smoothed broken power law : ( 1 ) @xmath100 , @xmath70 for 275 ks @xmath101 ms ; ( 2 ) @xmath102 , @xmath73 for @xmath103 ( @xmath104 , smoothness parameter @xmath105 ) .
the steepening of the decay at @xmath106 ms , if not due to the cooling frequency passing through the observing band ( which results @xmath107 , inconsistent with the observation ) , should originate from the post - jet - break evolution .
the change of the temporal decay index , @xmath108 , is consistent with a non - spreading jet break in a wind model with @xmath109 ( if the jet has significant sideways expansion , then the value of @xmath30 should be equal to @xmath110 ) .
however , the spectral hardening around the break time can not be well interpreted in such a model .
furthermore , the transition of a jet break in the wind medium usually takes two orders of magnitude in time ( kumar & panaitescu 2000 ) , which is inconsistent with the observation of grb 060729 . in conclusion , the temporal break at @xmath111 ms is probably not a jet break .
there are two possible interpretations for the last tentative light curve break at @xmath98 ms , as follows : a jet break , or a spectral break in a spherical model .
we discuss these next , after which we consider the implications of the long plateau phase .
the jet + wind model predicts @xmath112 for a sideways expanding jet or @xmath113 for a non - sideways expanding jet . the value of the model - predicted temporal index after the jet break thus can not be excluded at even the @xmath114 confidence level ( see fig.[contour ] ) .
recently zhang & macfadyen ( 2009 ) have performed two dimensional simulations and calculations of grb afterglow hydrodynamics and emission .
they showed that the sideways expansion of grb jets can be neglected during the relativistic phase ( for the sideways expansion of grb jets see also kumar & granot 2003 , granot & kumar 2003 ) and that the change in decay slope was larger than predicted analytically .
their results may further alleviate the above problem of the relatively shallow theoretical slope compared with the observed steep slope .
however , the spectral softening revealed by the hardness ratio evolution around this break time somewhat disfavors the jet break interpretation .
if we interpret this break as a jet break , then a half - opening angle of the jet can be inferred ( e.g. , * ? ? ?
* ) . under the thin - shell approximation for the post - shocked fluid of a relativistic blast wave ,
the conservation of energy ( neglecting the initial baryon loading in the fireball ) reads @xmath115 where @xmath116 is the bulk lorentz factor of the downstream fluid just behind the shock front and @xmath117 is the lorentz factor of the shock .
the swept - up circum - burst mass by the blast wave is @xmath118 where the environmental density @xmath119 , and @xmath120 for the ism case ( @xmath121 ) and @xmath122 @xmath123 for the stellar wind case ( @xmath124 ) .
the evolution of the shock radius @xmath125 measured in the observer s frame is @xmath126 , therefore , @xmath127^{1/(4-k)},\ ] ] and @xmath128^{1/2}.\ ] ] inserting the inferred values of the physical parameters , we have @xmath129 @xmath130 for the ism case , and @xmath131 @xmath132 @xmath133 for the wind case .
the values of @xmath134 and @xmath135 adopted here can be found below .
we adopt the convention of @xmath136 in cgs units .
unless @xmath137 @xmath138 , the jet has already decelerated to be non - relativistic ( @xmath139 ) while the fact that there was no jet break before @xmath140 s argues against the jet in an ism medium . in other words ,
the outflow of grb 060729 is likely spherical if the circum - burst medium is ism .
however , in this way the temporal break at @xmath42 can not be explained with the hydrodynamic / geometry effect .
at such a late time , the jet may also enter the non - relativistic phase in a stellar wind medium and the hydrodynamics is described by the self - similar sedov - von neumann - taylor solution .
the non - relativistic transition time , where @xmath141 years is the time when a grb jet with negligible sideways expansion becomes non - relativistic . ] is @xcite @xmath142 where @xmath143 is the beaming - corrected kinetic energy of the outflow .
the non - relativistic transition predicts a flattening in the light curve @xcite .
the theoretical temporal slope is @xmath144 for @xmath110 , which is not inconsistent with the observations if the large error bars of the observed slope are considered .
however , the trend of steepening after the break contradicts with the trend of flattening when the jet becomes non - relativistic .
the jet is still inside the free wind bubble @xmath145 days after the burst in the observer s frame , indicating that the termination shock radius of the wind bubble is larger than @xmath146 pc .
the size of a grb progenitor star wind bubble depends on the density and pressure of external interstellar medium , and the mass loss history of the progenitor @xcite .
the large size of the wind bubble surrounding the progenitor of grb 060729 suggests ( 1 ) a fast and tenuous wind during the rsg phase prior to the wolf - rayet phase , ( 2 ) the life time of the wolf - rayet stage is relatively long , and ( 3 ) the density and pressure of the external interstellar medium should be low .
therefore , from modeling the afterglow , the progenitor of grb 060729 is not likely to be located in a giant molecular cloud or a star - burst environment .
further deep optical observation of the grb site and its host galaxy may help to test this prediction .
we assume that the initial shock - accelerated electrons at such a late time have a broken power law distribution , @xmath147 where @xmath148 is the low energy power law index and @xmath149 is the high energy power law index , @xmath150 is the break lorentz factor of electrons .
this assumption is much more realistic than the single power law assumption , especially for the late time when the shock lorentz factor decreases to the order of unity / the shock is no longer ultra - relativistic ( e.g. , hededal et al .
2004 ; niemiec & ostrowski 2006 ; spitkovsky 2008 ) . for simplicity , we assume @xmath151 remains a constant in time ( li & chevalier 2001 ) . in this scenario ,
@xmath152 ( @xmath153 ) is equal to @xmath154 s , @xmath155 ( @xmath156 , @xmath157 is the typical synchrotron frequency of @xmath150 electrons ) is equal to @xmath158 s. the last steep decay segment can be described with @xmath159 and @xmath160 .
the high energy power index is therefore derived to be @xmath161 , and the inferred spectral index is @xmath162 . from @xmath163
s and @xmath164 s , we constrain the parameter @xmath165 .
the above @xmath166 , @xmath167 and @xmath168 are quite similar to those derived for grbs 991208 and 000301c in li & chevalier ( 2001 )
. the synchrotron emission from a spherical relativistic blast wave can be described by ( e.g. , chevalier & li 2000 ; wu et al .
2005 ) @xmath169 @xmath170 @xmath171
where @xmath172 , @xmath173 are the energy equipartition fractions of electrons and magnetic field respectively , @xmath135 is the stellar wind parameter .
we have considered the cooling effect on @xmath96 by synchrotron - self - compton scattering processes .
the crossing time @xmath174 in the optical band ( @xmath175 hz ) by @xmath176 no later than @xmath177 s gives @xmath178 which is easily satisfied , while the crossing time @xmath152 in the x - ray band ( @xmath179 kev ) by @xmath96 gives @xmath180 the flux density at frequency @xmath96 , @xmath181 , should be equal to the observed 2 kev x - ray flux density of @xmath182 mjy at @xmath183 s , which reads @xmath184 the above afterglow model parameters can not be tightly constrained due to the lack of more conditions / equations .
however , a large late - time isotropic kinetic energy of @xmath185 ergs and a small wind parameter of @xmath186 are favored if we assume @xmath187 and @xmath188 as inferred from fittings to other afterglows ( e.g. , panaitescu & kumar 2001 ; 2002 ) . within this parameter set , @xmath189 is much higher than those derived from grbs 991208 and 000301c by li & chevalier ( 2001 ) .
a large @xmath134 ( @xmath190 ergs ) at very late time ( after the plateau phase ends ) can be understood if the early long plateau phase detected both in x - ray and optical bands from a few hundred seconds to @xmath191 s since the trigger is interpreted as the period of energy injection , as already discussed in @xcite .
possible scenarios for this enhanced energy injection are refreshed shocks as suggested by @xcite or the continuous energy injection by the strong magnetic field of a newborn fast - rotating magnetar or black hole as suggested by @xcite and @xcite . in the latter model ,
the temporal index @xmath192 of the injected luminosity ( @xmath193 , @xmath194 ) can be derived from the observed values of the temporal and spectral indices during the plateau phase ( e.g. , table 2 of zhang et al .
2006 ) . from table 5 of @xcite
, we can obtain @xmath195 with @xmath196 and @xmath197 for the x - ray afterglow plateau assuming @xmath198 .
such a @xmath192 value is consistent with the theoretical expectation ( @xmath199 ) of the poynting - flux dominated injection model .
therefore the isotropic kinetic energy of the shock increases linearly with time , @xmath200 .
the prompt isotropic gamma - ray energy release in the @xmath201 kev - @xmath202 mev band is @xmath203 ergs ( @xcite ) . adopting a typical grb efficiency of @xmath204 , the initial isotropic kinetic energy remaining in the afterglow shock is about the same order of @xmath205 , i.e. @xmath206 ergs .
after the energy injection is finished , the shock energy @xmath134 is increased by a factor of @xmath207 compared to its initial value , so @xmath134 is the order of @xmath208 ergs .
the lack of a jet break up to @xmath209 days results a lower limit of the half - opening angle of the grb jet , i.e. , @xmath210 which corresponds to a beaming - corrected jet ( double - sided ) energy of @xmath211 this makes it one of the most energetic jets ever seen and is the reason why we were able to detect the x - ray afterglow still 642 days after the burst .
note that the energy of a fastest - rotating magnetar is @xmath212 ergs . a massive and fast - rotating black hole as the central engine is thus more likely to provide the required energy .
what makes the x - ray afterglow of grb 060729 so remarkable is the fact that it was still detected even almost two years after the burst .
this exceptional late - time detectability is related to three things : a ) with an initial 0.3 - 10.0 kev flux of almost @xmath213 ergs s@xmath8 @xmath9 it was one of the brightest afterglows ever detected by _
swift _ , b ) its flat decay phase @xcite extended out to about 60 ks after the burst , and c ) the decay slope after that break is about @xmath1 = 1.3 . despite breaks at @xmath214 ms and @xmath214 year the afterglow was still detected by _
chandra _ nearly two years after the burst .
bursts like grbs 060614 , 061121 , or even 080319b
* ; * ? ? ?
* respectively ) were even brighter in x - rays at about 100 s after the burst than grb 060729 , but their plateau phases are significantly shorter than that of grb 060729 .
they therefore faded more rapidly than grb 060729 at late times .
analysis and modeling of the x - ray afterglow of grb 060729 show that this burst happened in a tenuous wind .
this is consistent with the collapsar picture of long grbs . during the early plateau phase ,
the energy in the external shock increased by two orders of magnitude .
a reanalysis of the _ swift _ xrt light curve together with the three detections by _ chandra _ in 2007 reveal a temporal break at @xmath215 ms after the burst .
the decay slope steepened from @xmath216 to @xmath217 around this break and the x - ray spectrum hardened in the meanwhile , indicating this break is a cooling break ( the cooling frequency of synchrotron radiation crosses the x - ray band ) .
there is another light curve break at @xmath218 year after the burst tentatively indicated by the last two _ chandra _ detections .
this break coincides with a possible spectral softening , suggesting that the break may be of spectral origin , though a hydrodynamic origin ( jet break ) is also possible .
if due to a jet break , then the implied half - opening angle is @xmath219 . if due to a spectral break , such a spectral softening could be the result of a very steep power - law distribution of shock - accelerated electrons responsible for the synchrotron radiation . in this case , with no evidence for a jet break up to 642 days after the burst by _
chandra _ the jet half - opening angle must be @xmath220 and the jet energy @xmath221 erg .
such a large jet energy implies that the central engine must be a fast - rotating massive black hole , not a magnetar .
our _ chandra _ observations presented here have shown again how important _ chandra _ is for late - time observations of grb x - ray afterglows .
_ chandra _ has already been essential for the detection and non - detection of jet breaks in the x - ray afterglows of the short - duration grbs 050724 and 051221a ( * ? ? ?
* ; * ? ? ?
* respectively ) .
we want to thank sandy patel and chryssa kouveliotou for providing the data for the luminosities of _ swift _ and pre-_swift _ bursts , hala eid for fitting the late - time light curve with the beuermann et al .
model , and eric feigelson for useful discussion about statistics .
we thank the referee for his / her suggestions / comments which have helped to improve our paper significantly .
we are extremely thankful to the whole _ chandra_team for successfully planning and performing the observations of grb 060729 .
xfw thanks peter mszros , kenji toma , derek fox , antonino cucchiara , and yizhong fan for their helpful discussion .
this work was also partially supported by nasa nnx 08al40 g ( xfw ) , national natural science foundation of china ( grants 10221001 , 10403002 , 10503012 , 10621303 , 10633040 , and 10873002 ) , and national basic research program of china ( 973 program 2009cb824800 ) ( xyw , xfw , and ewl ) .
swift is supported at penn state by nasa contract nas5 - 00136 .
this research has been supported by sao grants sv4 - 74018 , a12 ( d.g . and g.g . ) and g08 - 9056 x ( d.g . )
arnaud , k. a. , 1996 , asp conf .
ser . 101 : astronomical data analysis software and systems v , 101 , 17 barthelmy , s. d. , 2005 , space science reviews , 120 , 143 beuermann , k. , et al . , 1999 , , 352 , l26 burrows , d. n. , et al .
, 2005 , space science reviews , 120 , 165 burrows , d. n. , et al . , 2006 , , 653 , 468 burrows , d. n. , & racusin , j. , 2007 , il nuovo cimento b , vol 121 , 1273 chevalier , r. a. , & li , z .- y . , 2000 , , 539 , 195 chevalier , r. a. , li , z .- y .
, & fransson , c. , 2004 , , 606 , 369 curran , p. a. , van der horst , a. , & wijers , r. a. m. j. , 2008 , , 386 , 859 dai , z. g. , & lu , t. , 1998 , , 333 , l87 dai , z. g. , & wu , x. f. , 2003 , , 591 , l21 evans , p. a. , et al . , 2009 , , 397 , 1177 frail , d. a. , et al . , 2001 , , 562 , l55 garcia - segura , g. , langer , n. , & mac low , m - m , 1996 , , 316 , 133 gehrels , n. , et al . , 2004 , apj , 611 , 1005 granot , j. , & kumar , p. , 2003
, , 591 , 1086 grupe , d. , burrows , d. n. , patel , s. k. , kouveliotou , c. , zhang , b. , mszros , p. , wijers , r. a. m. , & gehrels , n. , 2006 , , 653 , 462 grupe , d. , et al . , 2007 , , 662 , 443 hededal , c. b. , haugblle , t. , frederiksen , j. t. , & nordlund , , 2004 , , 617 , l107 huang , y. f. , & cheng , k. s. , 2003 , mnras , 341 , 263 kumar , p. , & panaitescu , a. , 2000 , , 541 , l9 kumar , p. , & granot , j. , 2003 , , 591 , 1075 li , z .- y . , & chevalier , r. a. , 2001 , , 551 , 940 liang , e .- w . ,
racusin , j. l. , zhang , b. , zhang , b .- b . , & burrows , d. n. , 2008 , , 675 , 528 mangano , v. , et al .
, 2007 , , 470 , 105 mszros , p. , 2006
phys . , 69 , 2259 mszros , p. , rees , m. j. , & wijers , r. a. m. j. , 1998 , , 499 , 301 niemiec , j. , & ostroswski , m. , 2006 , , 641 , 984 nousek , j. , kouveliotou , c. , grupe , d. , page , k. l. , et al . , 2006 , , 642 , 389 page , k. l. , et al .
, 2007 , , 663 , 1125 panaitescu , a. , & kumar , p. , 2001
, , 560 , l49 panaitescu , a. , & kumar , p. , 2002
, , 571 , 779 park , t. , kashyap , v. l. , siemiginowska , a. , van dyk , d. a. , zezas , a. , heinke , c. , & wargelin , b. j. , 2006 , , 652 , 610 racusin , j. l. , et al . , 2008 , nature , 455 , 183 racusin , j. l. , et al . , 2009 , , 698 , 43 rees , m. j. , & mszros , p. , 1998
, , 496 , l1 rhoads , j. e. , 1999 , , 525 , 737 roming , p. w. a. , et al . , 2005 , space science reviews , 120 , 95 sari , r. , piran , t. , & narayan , r. , 1998 , , 497 , l17 sari , r. , piran , t. , & halpern , j. p. , 1999
, , 524 , l43 sato , g. , et al . , 2007 , , 657 , 359 spitkovsky , a. , 2008 , , 682 , l5 thoene , c. c. , et al . 2006 ,
5373 , http://gcn.gsfc.nasa.gov/gcn3/ 5373.gcn3 tiengo , a. , mereghetti , s. , ghisellini , g. , rossi , e. , ghirlanda , g. , & schartel , n. , 2003 , , 409 , 983 tiengo , a. , mereghetti , s. , ghisellini , g. , tavecchio , f. , & ghirlanda , g. , 2004 , , 423 , 861 troja , e. , et al . , 2007 , , 665 , 599 waxman , e. , 2004 , , 602 , 886 willingale , r. , et al .
, 2007 , , 662 , 1093 wu , x. f. , dai , z. g. , huang , y. f. , & lu , t. , 2005 , apj , 619 , 968 zhang , b. , & mszros , p , 2001 , , 552 , l35 zhang , b. , & mszros , p. , 2004
phys . a , 19 , 2385 zhang , b. , et al . , 2006 , , 642 , 354 zhang , w. q. , & macfadyen , a. , 2009 , , 698 , 1261 lccrcc 7567 & 2007 - 03 - 16 11:39 & 2007 - 03 - 16 19:57 & 27690 & + 8541 & 2007 - 03 - 17 13:02 & 2007 - 03 - 17 14:54 & 4701 & [ -1.5ex]3.93@xmath222 & [ -1.5ex]4.26@xmath223
+ 7568 & 2007 - 05 - 16 09:30 & 2007 - 05 - 16 20:56 & 40268 & 2.67@xmath224 & 2.90@xmath225 + 7569 & 2007 - 06 - 30 06:19 & 2007 - 06 - 30 23:48 & 60400 & 2.07@xmath226 & 2.25@xmath227 + 9086 & 2007 - 12 - 26 01:02 & 2007 - 12 - 26 09:04 & 27395 & + 9801 & 2007 - 12 - 28 19:23 & 2007 - 12 - 28 23:59 & 15068 & + 9802 & 2007 - 12 - 29 18:57 & 2007 - 12 - 30 01:13 & 20239 & [ -1.5ex]0.69@xmath228 & [ -1.5ex]0.75@xmath229 + 9803 & 2008 - 01 - 05 01:04 & 2008 - 01 - 05 04:23 & 9986 & + 9087 & 2008 - 04 - 30 23:09 & 2008 - 05 - 01 08:34 & 32330 & + 9811 & 2008 - 05 - 01 22:45 & 2008 - 05 - 02 07:50 & 30775 & + 9812 & 2008 - 05 - 03 15:51 & 2008 - 05 - 03
23:35 & 26382 & [ -1.5ex]0.26@xmath230 & [ -1.5ex]0.28@xmath231 + 9813 & 2008 - 05 - 04 15:07 & 2008 - 05 - 04
23:22 & 27810 & \1 ) powl fixed & 1.32 ( fixed ) & & & & & 897/46 + 2 ) powl free & [email protected] & & & & & 400/45 + 3 ) bknpowl & 1.32@xmath232 & 2.08@xmath233 & 1.85@xmath234 & & & 168/43 + 4 ) powl & [email protected] & & & & & 160/41 + 5 ) bknpowl & 1.32@xmath235 & 1.23@xmath236 & 1.70@xmath237 & & & 65/39 + 6 ) bknpowl @xmath238 ms & 1.32@xmath235 & 1.01@xmath239 & 1.61@xmath234 & & & 60/37 + 7 ) powl @xmath240 ms & & & [email protected] & & & 12/15 + 8) powl 1.2 ms @xmath241 35 ms & & & 1.61@xmath242 & & & 6/13 + 9 ) bknpowl @xmath240 ms@xmath243 & & & 1.61 ( fixed ) & @xmath244 & @xmath245 & 6/14 + | we report on 5 _ chandra _ observations of the x - ray afterglow of the gamma - ray burst grb 060729 performed between 2007 march and 2008 may . in all five observations
the afterglow is clearly detected .
the last _ chandra _
pointing was performed on 2008-may-04 , 642 days after the burst - the latest detection of a grb x - ray afterglow ever .
a reanalysis of the _ swift _ xrt light curve together with the three detections by _ chandra _ in 2007 reveals a break at @xmath01.0 ms after the burst with a slight steepening of the decay slope from @xmath1 = 1.32 to 1.61 .
this break coincides with a significant hardening of the x - ray spectrum , consistent with a cooling break in the wind medium scenario , in which the cooling frequency of the afterglow crosses the x - ray band .
the last two _ chandra _ observations in 2007 december and 2008 may provide evidence for another break at about one year after the burst . if interpreted as a jet break , this late - time break implies a jet half opening angle of @xmath2 for a wind medium .
alternatively , this final break may have a spectral origin , in which case no jet break has been observed and the half - opening angle of the jet of grb 060729 must be larger than @xmath3 for a wind medium .
we compare the x - ray afterglow of grb 060729 in a wind environment with other bright x - ray afterglows , in particular grbs 061121 and 080319b , and discuss why the x - ray afterglow of grb 060729 is such an exceptionally long - lasting event . |
the edge - on spiral galaxy ngc4945 contains the first h@xmath0o maser discovered in a galactic nucleus ( @xcite ; @xcite ) . within the nucleus
there is a heavily absorbed hard x - ray component ( @xmath13 kev ) , with a measured column density of @xmath14 @xmath15 ( @xcite ) , and true x - ray luminosity ( 220 kev ) of @xmath16 erg s@xmath17 .
the hard component varies on timescales of hours , implying a source size of @xmath18 cm .
the nucleus is very prominent at infrared wavelengths , with a far infrared luminosity @xmath19 erg s@xmath17 from the inner 200 pc ( @xcite ) , at an assumed distance of 3.7 mpc ( @xcite ) .
the nucleus also shows a @xmath20 kpc ionization cone along the galactic minor axis ( @xcite ) and emission knots with liner - like spectra ( @xcite ) .
optical line splittings of up to @xmath21 ( @xcite , @xcite ) indicate that the cone is evacuated by a nuclear wind .
while x - ray characteristics suggest that the active galactic nucleus ( agn ) is driven by an obscured seyfert nucleus , the high ir luminosity and wind indicate the presence of a nuclear starburst .
spectra of the maser consistently show emission ( @xmath22 jy ) within about @xmath23 of the systemic velocity and stronger emission ( @xmath24 jy ) red - shifted by @xmath25 ( e.g. , @xcite , figure 1 ) .
previously , no corresponding emission on the blue side of the systemic velocity has been detected ( e.g. , @xcite ) .
we adopt a heliocentric systemic velocity of @xmath26 , based on co observations ( @xcite ) .
( velocities are defined with respect to the standard radio astronomical definition of doppler shift throughout . )
the importance of nuclear h@xmath0o masers to the study of agn has been widely appreciated after recent observations of the maser in the nucleus of ngc 4258 ( @xcite ) , where the masers were found to trace a thin molecular keplerian disk of 0.13 pc inner radius around a central mass of @xmath27 m@xmath11 . in this letter ,
we report the first sub - arcsecond estimate of the maser position in ngc4945 , the detection of blue - shifted maser features , and the first interferometric map of the distribution of maser emission .
we observed the maser with the very long baseline array of the nrao@xmath28 for 4 hours @xmath28 the national radio astronomy observatory is operated by associated universities , inc , under cooperative agreement with the national science foundation . on 1995 january 22 .
we used the southernmost stations , at mauna kea , hi ( mk ) , saint croix , vi ( sc ) , and kitt peak , az ( kp ) . at kp
the source reached a maximum elevation of @xmath29 , while it was @xmath30 and rising at mk .
we obtained useful data for only 2.5 hours on the mk kp baseline . observing conditions at saint croix were relatively poor and no useful data were obtained from it .
tipping scans showed that the opacity at kp , mk , and sc were 0.07 , 0.05 , and 0.3 at zenith , respectively .
the baseline length was 4470 km and the angular resolutions were @xmath31 and @xmath32 milliarcseconds ( mas ) in right ascension and declination , respectively .
the system temperatures for @xmath33 elevation at kp and mk were 170 and 130 k , respectively .
we recorded four contiguous bandpasses covering the range 405 to 821 , so as to include the systemic emission , the red - shifted emission , and the likely velocity range of counterpart blue - shifted emission .
the spectrum is shown in figure 1 .
we calibrated the data with the nrao aips package .
the interferometer delays and fringe rates were estimated from observations of the continuum sources 4c39.25 , 3c273 , and @xmath34 .
the latter source was observed at elevations @xmath35 and was included so that we could estimate the magnitude of elevation dependent errors in the correlator model and post - correlation analysis .
the delay calibration is accurate to about 3 ns overall , after implementing the u. s. naval observatory station positions and earth orientation parameters ( eubanks , private communication ) .
we analyzed the fringe rate of the maser emission as a function of time to estimate the position of the maser , taking into account the effect of the atmosphere .
we fit a sinusoid to the measured fringe rates for emission between 664 and 719 ( with uniform weighting ) to obtain position offsets of @xmath36 and @xmath37 from the _ a priori _ position @xmath38 , @xmath39 , with respect to a reference frame consisting of the sources 4c39.25 , @xmath34 , and 3c273b .
we conservatively estimate the uncertainties in position based on the residual fringe rates of the maser and calibrator sources . the new maser position is @xmath40 @xmath41 to map the brightness distribution of the maser
, we phase referenced the data to the strong spectral feature at @xmath42 . the phases of this feature were subtracted from the phases of other spectral features , which corrected for the effects of the atmosphere and calibration uncertainties .
we assumed a point source model and fit a sinusoid ( of 24 hour period ) to the difference phase , estimating relative feature positions and formal errors ( e.g. , @xcite ) .
sample phase plots are shown in figure 2 .
the data were averaged in frequency across the line profile for each feature and averaged in time for between 300 s and 640 s so that phase ambiguities could be resolved prior to fitting . for a 1 jy feature that is 1wide ( half power width )
the signal - to - noise ratio in 300 s is about 4 .
the structure of the maser is remarkably linear , with emission distributed over @xmath7 mas between 414 and 719(figure 3 ) .
deviations from linearity are antisymmetric with respect to the systemic emission , which gives the apprearance of a shallow `` s '' on the sky .
the centroids of the blue and red emission are separated by @xmath43 mas or 0.55 pc , at a position angle of @xmath44 , and roughly symmetrically bracket the systemic emission . the wide separations among the systemic , red , and blue emission correspond to many interferometer fringes and are well determined .
maser features that are detected only marginally ( i.e. , with a rms of residuals about the fitted model of @xmath45 rad ) are not shown in the map , although they do not appreciably alter the source structure .
these marginal features include one at @xmath46 , which is more red - shifted than any previously reported maser emission in ngc4945 . in nonetheless , not all of the detected maser features could be mapped , particularly near the systemic and blue - shifted velocities , because a significant fraction are not well approximated by a point source model ( i.e. , the phases could not be fit by a diurnal sinusoid ) .
the interferometer calibration was sufficiently good that uncertainty in the astrometric position of the maser is the largest source of quantifiable systematic error in estimating maser feature positions .
the error is @xmath47 @xmath6as , which is computed for an astrometric uncertainty of 100 mas and for a feature 200 from the reference , and is proportional to both quantities .
( residual atmospheric delay contributes a moderate systematic error because of the low elevations at which we observed the maser ; it is smaller than the error from the astrometry . )
a second source of systematic error arises from the possibility that emission , for each feature , may not be well described by a point source , which is difficult to quantify .
however , the data for each mapped maser feature is well fit by a point source model over the limited range of hour angle available . the systematic error in position for each maser feature is less than the statistical error , which is dominated by the limited range of hour angle over which the phases were measured .
this limitation leads to 1 ) a large correlation between right ascension and declination parameters and 2 ) an uncertainty as to in which lobe of the interferometer fringe pattern the maser lies .
uncertainties in the estimated positions take into account possible lobe ambiguities .
the new astrometric position of the maser clearly shows that it is associated with the center of the galactic nucleus ( figure 5 ) . the previous position of batchelor(1982 ) , obtained with the parkes antenna of the csiro , was substantially more uncertain because of the beamwidth of the antenna .
the maser lies within @xmath48 of the infrared peak at 1.6 @xmath6 m and overlaps the radio continuum source , which whiteoak & wilson ( 1990 ) estimate is @xmath49 in size at 6 ghz ( see figure 5 ) .
we cautiously interpret the data for the maser in ngc4945 in terms of the paradigm established for ngc 4258 .
the extensive data on ngc 4258 , with its clear keplerian signature , show that the masers define a nearly edge on thin annular disk . the data for ngc4945 are more sketchy but they share the characteristic of systemic emission bracketted roughly symmetrically in velocity _ and _ position by high - velocity emission .
moreover , the angular distributions of blue - shifted emission is roughly linear and the velocities decline with angular distance from the systemic emission .
however , the position errors are significant , the red - shifted emission does not appear to mirror this structure , the decline in velocity for the blue - shifted emission is faster than what is characteristic for a keplerian rotation curve , and the non - point source emission that seems to exist at some systemic and blue - shifted velocities could not be mapped ( figure 4 ) .
nonetheless , there is a suggestion that the structures of ngc4945 and ngc4258 may be similar , and we proceed on that assumption . in the context of an edge - on disk ,
the observations suggest a binding mass ( for circular orbits ) of @xmath50 , where @xmath51 is the enclosed mass in units of m@xmath11 , @xmath52 is the orbital velocity in units of , @xmath53 is the distance in mpc , @xmath54 is the radius in mas , @xmath55 is inclination , and @xmath56 is the azimuthal angle of the red - shifted masers in the disk plane ( @xmath57 is the plane of the sky ) .
the binding mass is @xmath58 m@xmath11 for @xmath59 , @xmath60 mas or 0.3 pc , and @xmath61 ( figure 4 ) .
the implied mass density is @xmath62 m@xmath11 pc@xmath63 .
the position angle of the disk , @xmath8 , is comparable to the position angle of the larger galactic disk , @xmath64 , and the senses of rotation of the two disks are the same .
if the observed @xmath16 erg s@xmath17 x - ray luminosity ( @xcite ) implies a bolometric luminosity on the order of @xmath65 erg s@xmath17 , then the central engine radiates 10% of its eddington luminosity .
we note that the shallow `` s '' the mappable maser emission traces may be consistent with a warp in the disk , which could be caused by radiation - driven torques that result from oblique illumination by the central engine ( @xcite ) , though this is not a definitive explanation .
the interpretation of the observed maser structure is not unique and radically different interpretations might be viable . the emission could arise from a complex of star forming regions superposed on or within the nucleus , or from jet - induced shocks in the agn .
the linear extent of the maser is comparable to galactic star formation complexes ( e.g. , w49n ) , although the velocity difference between the systemic and red - shifted features is large for adjacent star forming regions .
maser emission can arise in warm gas behind dissociative shocks ( @xcite ) and nondissociative shocks ( @xcite ) , although the maser spectrum mandates a complex line - of - sight velocity structure for the shocks .
interferometric synthesis observations will undoubtedly clarify the dynamical characteristics of the maser region , providing a comprehensive map of the maser emission within 200 of the systemic velocity .
the dual nature of the nucleus , as both a starburst and a very heavily obscured seyfert object , makes this water maser source particularly important since it may be used to probe the dynamics and structure of the nucleus on sub - parsec scales .
1 spectrum of the water maser emission from the nucleus of ngc4945 made on 1995 october 13 at the parkes radiotelescope of the csiro .
the maser is variable on time scales of months but the spectrum shown is typical .
the integration time is about 120 minutes ( on and off source ) .
the adopted heliocentric systemic velocity of the galaxy , 561 , is indicated by the vertical bar .
the dashed lines indicate the extremes of velocity that we observed with the vlba .
the solid bars above the velocity axis indicate the range of velocity within which maser emission is detected .
the absolute gain calibration is accurate to @xmath66 % .
velocities with respect to the local standard of rest may be obtained from the relation @xmath67 .
2 sample plots of phase vs time for ( _ top _ ) the systemic maser feature at 557 and ( _ bottom _ ) the red - shifted feature at 665 .
the phases sample a 24 hour sinusoid and are relative to the phase of the reference feature at 699 .
the right ascension and declination offsets of the maser features , with respect to the reference , were determined from these curves .
the hour angle of the source with respect to the baseline is @xmath68 h at 12 ut .
3 map of the maser brightness distribution .
the gray - scale shows velocity ( 414719 ) with lighter shading indicative of more blue - shifted emission .
the error bars reflect measurement noise at the @xmath69 level .
the position angle of the galactic disk is @xmath64 ( @xcite ) .
marginal detections , for which the rms of the phase residuals from the model are @xmath70 rad , are also not shown .
the arrow indicated the maser feature closest to the systemic velocity of the galaxy .
4 position - velocity diagram for the maser emission .
the impact parameter is measured with respect to the maser feature closest to the systemic velocity of the galaxy .
the dashed line indicated the systemic velocity .
the error bars represent position uncertainties at the @xmath69 level . fig .
5 comparison of various position determinations for the nucleus of ngc4945 .
these include the new maser position reported here and the _ a priori _ position obtained by batchelor(1982 ) with the parkes antenna , the interferometric position of h i absorption ( @xcite ) , the synthesis positions at 408 mhz ( @xcite ) , 1.4 ghz ( @xcite ) , and 6.0 ghz ( @xcite ) , and the position of the @xmath71 m peak ( whiteoak & gardner , unpublished ) .
the sizes of the symbols correspond to the @xmath48 uncertainties in position .
the deconvolved angular size of the continuum source at 6 ghz is about @xmath49 ( @xcite ) at a position angle of @xmath72 and is shown by the dashed ellipse .
the position angle of the galactic disk is @xmath64 ( @xcite ) . | we present the first interferometer map of the h@xmath0o maser emission in the active nucleus of ngc4945 , which exhibits both starburst and seyfert qualities .
although the declination of the galaxy is @xmath1 , we were able to make the observations with the southernmost antennas of the very long baseline array .
strong maser emission is present in three velocity ranges , one near the systemic velocity and two shifted roughly symmetrically by @xmath2 .
this is the first detection of highly blue - shifted water emission in ngc4945 .
we also report a marginal detection of previously unreported red - shifted emission at a velocity of @xmath3 with respect to systemic . from fringe rate analysis we determined the position of the maser to be @xmath4
the uncertainties in earlier estimates are at least several arcseconds .
the maser lies within @xmath5 ( 36 pc at a distance of 3.7 mpc ) of the peaks in 1.4 ghz continuum and 1.6 @xmath6 m emission from the nucleus .
the mappable maser emission is distributed roughly linearly over @xmath7 milliarcseconds ( 0.7 pc ) at a position angle of @xmath8 , which is close to the @xmath9 position angle of the galactic disk .
the red and blue - shifted emission symmetrically stradle the systemic emission on the sky , which suggests material in edge - on circular motion around a central object .
the position - velocity structure indicates a binding mass of @xmath10 m@xmath11 , within a volume of radius @xmath12 pc , which in combination with estimates of the agn luminosity , implies that the central engine radiates on the order of 10% of its eddington luminosity . |
factorization theory has its genesis in studying the decomposition of natural numbers into its prime ( irreducible ) divisors . in this multiplicative monoid , the fundamental theorem of arithmetic guarantees that such a decomposition is unique .
many other monoids , however , have elements that fail to admit a unique factorization into irreducibles ; in such cases , algebraists are frequently interested in measuring , in various ways , how different the plural factorizations of an element can be .
one does not have to look far for examples .
in fact , certain submonoids of the natural numbers have a rich non - unique factorization theory . _ numerical monoids _
( co - finite , additive submonoids of the natural numbers ) have garnered much recent interest , as their factorization theories are complicated enough to be interesting while controlled enough to be tractable .
the goal of this paper is to introduce various perspectives and tools for studying factorizations in numerical monoids , and to present open problems that are of primary interest in this field .
several natural examples of numerical monoids are found in the literature .
historically , frobenius coin - exchange problem asked what monetary values one could make with relatively prime coin denominations . a more endearing manifestation of numerical monoids is the mcnugget monoid , which enumerates the number of chicken mcnuggets one could buy by using the standard package sizes of 6 , 9 , and 20 .
as with the coin - problem and the mcnugget monoid , all numerical monoids can be described using a finite set of generators .
we provide the formal definition below . in what follows ,
we let @xmath0 denote the set of non - negative integers .
[ d : numericalmonoid ] let @xmath2 be a collection of relatively prime positive integers .
the _ numerical monoid _ @xmath3 generated by these integers is the set of all non - negative , integral linear combinations of these numbers . in other words ,
@xmath4 [ r : mingenset ] this set is an additive monoid with identity @xmath5 , and the condition that @xmath6 guarantees that there exist only finitely many natural numbers that are not elements of @xmath3 .
in fact , every co - finite additive submonoid of @xmath0 can be generated by a finite set of relatively prime natural numbers ; thus , numerical monoids are precisely the co - finite , additive submonoids of @xmath0 .
while generating sets of a fixed numerical monoid are non - unique , there always exists a unique minimal generating set ( with respect to set - theoretic containment ) . as these minimal generators can not be represented in terms of the other generators , they are precisely the _
irreducible _ elements of the numerical monoid . for the remainder of the paper
, we will assume that @xmath7 constitutes the unique minimal generating set for a numerical monoid @xmath3 and that @xmath8 . both frobenius coin problem and the mcnugget problem ask which natural numbers arise as non - negative , integral linear combinations of their respective generators . as such , understanding the largest natural number that is not an element of @xmath3 ( named for the aforementioned coin problem ) is often useful .
[ d : frobenius ] for a numerical monoid @xmath3 , the values in @xmath9 are called the _ gaps _ of @xmath3 , and @xmath10 is called the _ frobenius number _ of @xmath3 . [
e : numericalmonoid ] for two relatively prime positive integers @xmath11 , the numerical monoid @xmath12 has @xmath13 as its frobenius number @xcite .
furthermore , all but finitely many elements of @xmath3 can be written in multiple ways as non - negative integral combinations of @xmath14 and @xmath15 .
the simplest example is the element @xmath16 , which can be seen as @xmath14 copies of @xmath15 or as @xmath15 copies of @xmath14 .
these two representations of @xmath17 as an element of @xmath3 are called _ factorizations _ ; see section [ s : factorization ] for precise definitions and notation .
the mcnugget monoid @xmath18 has frobenius number 43 ; thus , for every @xmath19 , one can purchase exactly @xmath20 mcnuggets using the standard package sizes of @xmath21 , and @xmath22 . as with the numerical monoid @xmath23 ,
almost every element in @xmath24 has multiple distinct factorizations .
for example , @xmath25 can be expressed as three copies of @xmath22 , ten copies of @xmath26 , or four copies of @xmath26 plus four copies of @xmath27 .
each such expression corresponds to a way to purchase exactly 60 mcnuggets .
the existence of a frobenius number for a numerical monoid @xmath3 guarantees that past some value , every natural number lies in @xmath3 .
therefore , as a set , a numerical monoid @xmath3 is a scattering of small " elements ( those less than @xmath28 ) and an infinite ray after @xmath28 .
what is less understood is the factorization theory of these elements , which is hinted at in example [ e : numericalmonoid ] . in general , the factorization theory of numerical monoids with @xmath29 minimal generators is well - understood , while those numerical monoids with 3 or more irreducible elements have a much more complicated factorization structure with numerous open questions .
factorization structures are often classified and quantified using a variety of important _ factorization invariants _ , which generally measure how far from unique a monoid element s factorizations are .
the coming sections introduce several standard invariants , together with results and open questions in the context of numerical monoids .
section [ s : factorization ] describes the strongest ( but most cumbersome ) invariant , the set of factorizations .
then , section [ s : lengthsets ] describes the length set and elasticity invariants , which measure , in different ways , how many irreducible elements appear in the factorizations of an element .
section [ s : deltasets ] gives several important results for delta sets , which measure the gaps in lengths that are possible between distinct factorizations of a single element . while length sets , elasticity , and delta sets record only information about the length of factorizations of an element , the catenary degree , which is described in section [ s : catenary ] , measures how many irreducible elements different factorizations of the same element do not share in common .
lastly , as non - unique factorizations coincide with the presence of irreducible elements that are not prime , section [ s : omega ] describes @xmath1-primality , which measures how far from prime " a monoid element is .
[ d : factorizations ] fix a numerical monoid @xmath30 and an element @xmath31 . a _
factorization _ of @xmath20 is an expression @xmath32 of @xmath20 as a sum of irreducible elements @xmath33 of @xmath3 . collecting like terms
, we often write factorizations of @xmath20 in the form @xmath34 ( see remark [ r : mingenset ] )
. write @xmath35 for the _ set of factorizations _ of @xmath31 .
when there is no ambiguity , we often omit the subscript and simply write @xmath36 .
[ r : ehrhart ] the notation for factorizations in definition [ d : factorizations ] is motivated in part by connections to discrete geometry .
in particular , each factorization of @xmath20 corresponds to an integer point in @xmath37 , and the set @xmath38 forms a codimension-1 linear cross - section of @xmath37 that coincides with the set of integer points of a convex integral polytope .
this viewpoint allows the use of techniques from lattice point enumeration and other tools from combinatorial commutative algebra when examining the factorization structure of numerical monoids @xcite .
the set of factorizations ( definition [ d : setoffactorizations ] ) is an example of a perfect factorization invariant in that it encapsulates enough information to uniquely determine the underlying monoid ( theorem [ t : setoffactorizations ] ) .
however , such complete information comes at a cost : extracting information from it ( or even simply writing it down ) is a nontrivial task .
invariants derived from the set of factorizations , such as the length set and elasticity invariants ( section [ s : lengthsets ] ) , are more manageable and easier to work with , but necessitate a loss of information .
below , @xmath39 denotes the power set of @xmath37 .
[ d : setoffactorizations ] given a numerical monoid @xmath30 , let @xmath40 denote the _ set of factorizations _ of @xmath3 .
theorem [ t : setoffactorizations ] follows from the fact that a minimal presentation of a numerical monoid @xmath3 ( as well as its defining congruence ) can be recovered from @xmath41 .
example [ e : setoffactorizations ] exhibits one possible method of doing so ; see @xcite for more background on monoid congruences and minimal presentations .
in fact , for a fixed @xmath42 , the set @xmath38 alone contains enough information to recover the monoid structure of @xmath3 . as such , it is perhaps not surprising that @xmath41 is often hard to describe in complete detail .
[ t : setoffactorizations ] for two numerical monoids @xmath3 and @xmath43 , the following are equivalent .
a. @xmath44 , b. @xmath45 , and c. @xmath46 .
* theorem 1.1 ) .
[ e : setoffactorizations ] fix a numerical monoid @xmath3 , and suppose that the two - element sets @xmath47 and @xmath48 are each a subset of some element of @xmath41 . since each is a subset of @xmath49 , @xmath3 must have three minimal generators , say @xmath50 . moreover , since @xmath51 and @xmath52 are factorizations of the same element , we must have @xmath53 , and similarly that @xmath54 . since @xmath55 , we conclude that @xmath56 , the mcnugget monoid from example [ e : numericalmonoid ]
. larger elements in a numerical monoid @xmath3 are more likely to have larger factorization sets .
more precisely , if @xmath3 has @xmath57 minimal generators , the size of @xmath36 grows on the order of @xmath58 .
theorem [ t : factorasymp ] , which specializes the asymptotic result in @xcite to numerical monoids , can be found in @xcite .
[ t : factorasymp ] given @xmath30 numerical , there exist periodic functions @xmath59 whose periods divide @xmath60 such that @xmath61 for all @xmath31 .
[ r : computefactor ] much of the recent literature on numerical monoids has focused on algorithms for explicit computation .
it is worth noting that although theorem [ t : factorasymp ] implies @xmath62 is polynomial in @xmath20 , is it @xmath63-hard to compute this set in general , as doing so requires solving a system of diophantine equations .
see @xcite for a survey of recent computational advances in this area .
several of the factorization invariants introduced in the coming sections , including those derived directly from the set of factorizations , benefit from faster computation methods .
in particular , several of these invariants admit algorithms that do not require computing large sets of factorizations ; see , for instance , remarks [ r : deltadynamic ] and [ r : omegadynamic ] .
additional computational traction can also be obtained by specializing to certain families of monoids , such as those with a fixed number of generators or generating sets of a certain form ; see , for instance , theorems [ t : elastarith ] , [ t : deltaarithmetic ] and [ t : catnaryarithmetic ] .
another computation technique in computing invariant values is to first compute the apry set of @xmath3 ( a set determined by the complement of @xmath3 ; see @xcite ) .
although computing the apry set from a list of generators is also @xmath63-hard in general , many invariants can then be computed directly from it .
for instance , computing the frobenius number of a numerical monoid is @xmath63-hard in general @xcite , but if the apry set is known , its computation is trivial @xcite . in this way , the complexity of computing an invariant ( or several invariants ) can be `` transferred '' to the computation of the apery set @xcite .
additionally , for some special families of numerical monoids , the apery set is known to admit a closed form ; see @xcite .
while the set of factorizations for a numerical monoid is a perfect invariant , it is often cumbersome to compute and encode .
the next two sections focus on invariants obtained by passing from factorizations to their lengths . in this section ,
we introduce length sets and elasticity as useful invariants for numerical monoids .
we begin with the definition of a length set .
[ d : lengthset ] fix a numerical monoid @xmath30 , and fix @xmath31 . given a factorization @xmath64 , we denote by @xmath65 the number of irreducibles in the factorization @xmath66 ; that is , @xmath67 .
the set of factorization lengths of @xmath20 , denoted @xmath68 , is called the _ length set _ of @xmath20 .
when there is no ambiguity , we often omit the subscript and simply write @xmath69 .
[ e : lengthset ] the length set of an element is easily computed from its set of factorizations . in the numerical monoid @xmath70
, the element @xmath22 has factorization set @xmath71 and length set @xmath72 .
this length set , however , is shared with other elements ( e.g. , @xmath73 ) .
this example highlights one manner in which information is frequently lost when passing from @xmath36 to @xmath69 .
although distinct elements in a numerical monoid can have identical length sets , one may ask whether distinct numerical monoids can have different sets of length sets . in
what follows , @xmath74 denotes the power set of @xmath0 . [
d : setsoflengthsets ] for a numerical monoid @xmath3 , its _ set of length sets _ is given by @xmath75 investigations into whether sets of length sets determine finitely generated , cancellative , commutative monoids up to isomorphism are widespread in factorization theory .
the most famous conjecture is stated below for certain block monoids ( that is , monoids of zero - sum sequences of abelian group elements under concatenation ) .
since a detailed introduction to block monoids would take us too far afield , we refer the interested reader to @xcite for background on block monoids and @xcite for recent progress on conjecture [ c : blocklensets ] . in what follows , let @xmath76 denote the block monoid over a group @xmath77 ; @xmath78 is defined analogously for this class of monoids . note that for two abelian groups @xmath77 and @xmath79 , @xmath80 if and only if @xmath81 . [ c : blocklensets ] given two finite abelian groups @xmath77 and @xmath79 with @xmath82 , we have @xmath83 implies @xmath84 .
theorem [ t : setsoflength ] answers an analogous question for numerical monoids in the negative .
* corollary 3.4)[t : setsoflength ] there exist numerical monoids @xmath85 with @xmath86 . in particular , sets of length sets do not characterize numerical monoids .
[ r : arithmetical ] the distinct numerical monoids with equal sets of length sets described in @xcite are called _ arithmetical numerical monoids _ , which are numerical monoids generated by an arithmetic sequence @xmath87 with @xmath88 and @xmath89 .
theorem [ t : setsoflength ] is proven by computing the set of length sets of arithmetical numerical monoids in terms of the parameters @xmath90 and @xmath57 and showing that different values of these parameters yield the same set . the description of the set of length sets for arithmetical numerical monoids upon which theorem [ t : setsoflength ] is based is rare and , in general , it is difficult to obtain such a description for other families of numerical monoids .
part of the difficulty comes in computing every factorization length that an element in a numerical monoid can obtain .
many times , the most interesting factorization lengths for an element are the largest and smallest ones .
the quotient of these two values gives a weaker factorization invariant called _ elasticity _ , which is defined below .
[ d : elasticity ] let @xmath3 be a numerical monoid , @xmath31 nonzero , and @xmath69 its length set .
let @xmath91 and @xmath92 the _ elasticity of @xmath20 _ is given by @xmath93 the _ set of elasticities of @xmath3 _ is given by @xmath94 .
the _ elasticity of @xmath3 _ is given by @xmath95 .
in what follows , we see that @xmath96 for numerical monoids and that , in fact , @xmath97 .
moreover , elasticity is one of the few factorization invariants known to admit a closed form for any numerical monoid @xmath3 .
[ t : holdenmoore ] let @xmath98 be minimally generated by @xmath99 with @xmath14 and @xmath100 the smallest and largest generators , respectively .
then , @xmath101 and the set of elasticities @xmath102 has @xmath103 as its unique accumulation point .
[ e : elastsets ] figure [ fig : elastsets ] depicts the elasticity of elements in the numerical monoids @xmath104 and @xmath105 .
the maximal values attained are @xmath106 and @xmath107 , respectively , as ensured by theorem [ t : holdenmoore ] . moreover , the monoid elements with maximal elasticity are precisely the multiples of the least common multiple of the first and last generators .
( left ) and the arithmetical numerical monoid @xmath108 ( right ) , produced using ` sage ` @xcite . here
, the point at @xmath109 indicates that @xmath110.,title="fig:",width=240 ] ( left ) and the arithmetical numerical monoid @xmath108 ( right ) , produced using ` sage ` @xcite . here
, the point at @xmath109 indicates that @xmath110.,title="fig:",width=240 ] more can be said about the structure of the set @xmath102 upon examining the asymptotic behavior of the maximum- and mininum - length factorization functions @xmath111 and @xmath112 for large monoid elements @xmath31
. see ( * ? ? ?
* section 4 ) for more detail .
* corollary 4.5)[t : elasticityset ] fix a numerical monoid @xmath98 .
a. for @xmath113 , we have @xmath114 b. the set @xmath102 is the union of a finite set and a collection of @xmath115 monotone increasing sequences , each converging to @xmath103 .
[ r : elasticityset ] both plots in figure [ fig : elastsets ] demonstrate the claims in theorem [ t : elasticityset ] .
roughly speaking , each plot can be viewed as a union of `` wedges '' of points , where each wedge contains one point from each of the monotone increasing sequences described in theorem [ t : elasticityset](b ) . moreover
, the right - hand plot depicts the elasticies of an arithmetical numerical monoid , where elasticity values are often repeated in quick succession .
it is this repetition of elasticitiy values that gives rise to theorem [ t : elastarith ] .
certainly , two numerical monoids with identical sets of length sets have equal sets of elasticities .
conversely , there exist pairs of numerical monoids with equal elasticity sets and unequal sets of length sets .
for example , for @xmath116 and @xmath117 , a computation shows that @xmath118 , while @xmath119 but @xmath120 ; see ( * ? ? ?
* example 3.11 ) .
thus , in general , sets of elasticities do not determine sets of length sets .
theorem [ t : elastarith ] demonstrates that , within the class of arithmetical numerical monoids , equality of sets of length sets is equivalent to equality of sets of elasticities .
* theorem 1.2)[t : elastarith ] for two distinct arithmetical numerical monoids @xmath121 and @xmath122 , the following are equivalent : a. @xmath118 b. @xmath86 theorem [ t : elasticityset ] demonstrates that computing the set of elasticities is fairly tractable , while the set of length sets is , in general , much more difficult to describe .
therefore , finding classes of numerical monoids ( beyond the arithmetical case ) for which the set of elasticities is as strong an invariant as the set of length sets is of much interest .
[ pr : elastlencomp ] characterize which distinct numerical monoids @xmath3 and @xmath43 satisfy @xmath118 and @xmath123 .
in this section , we consider the _ delta set _ invariant ( definition [ d : deltaset ] ) .
the delta set of a monoid element @xmath20 , like its elasticity ( definition [ d : elasticity ] ) , is derived from the set of factorization lengths @xmath69 . [
d : deltaset ] fix a numerical monoid @xmath30 , and fix @xmath31 . writing @xmath124 ,
the _ delta set _ of @xmath20 is the set @xmath125 of successive differences of factorization lengths of @xmath20 , and @xmath126 .
when there is no ambiguity , we often omit the subscript and simply write @xmath127 .
[ e : delta2gen ] consider the numerical monoid @xmath12 with two relatively prime generators .
if @xmath31 , then there exist @xmath128 such that @xmath129 .
notice that for any @xmath130 , @xmath131 is a factorization so long as @xmath132 and @xmath133 are non - negative .
in fact , it can be shown that all factorizations of @xmath20 in @xmath3 are of this form , corresponding to successive values of @xmath57 . since the length of such a factorization is @xmath134 , all successive length differences in @xmath69 yield @xmath135 .
thus , for any @xmath31 , either @xmath127 is empty ( when its length set is a singleton ) , or @xmath136 in particular , @xmath137 the above example is a special case of arithmetical numerical monoids , whose minimal generators form an arithmetic sequence ( remark [ r : arithmetical ] ) .
such monoids have singleton delta sets consisting of the step size between these generators , a fact which plays a crutial role in characterizing their length sets ( theorems [ t : setsoflength ] ) .
* corollary 2.3)[t : deltaarithmetic ] fix positive integers @xmath138 , @xmath139 , and @xmath57 with @xmath88 and @xmath89 . for @xmath140 and @xmath31 , either @xmath127 is empty , or @xmath141 . in particular , @xmath142 .
even slight generalizations of arithmetical numerical monoids have elusive delta sets ; this leads us to the following open problem .
[ pr : genarithmeticdelta ] let @xmath138 , @xmath143 , @xmath139 , and @xmath57 be positive integers such that @xmath88 and @xmath144 , and consider @xmath145 .
compute @xmath146 .
[ r : genarithmetic ] the generators of the numerical monoid in problem [ pr : genarithmeticdelta ] form what is known as a _
generalized arithmetic sequence_. numerical monoids generated by generalized arithmetic sequences and arithmetic numerical monoids are closely connected : when @xmath147 , a generalized arithmetic sequence reduces to an arithmetic sequence ; also , the last @xmath57 generators of a generalized arithmetic sequence form an arithmetic sequence .
these relationships , together with a precise membership criterion ( * ? ? ?
* theorem 3.1 ) , make working with this family of numerical monoids more tractable .
[ r : deltadifficulty ] aside from the arithmetical setting , few natural families of numerical monoids are known to admit closed forms for their delta sets . in general , given a numerical monoid @xmath3 , it can be hard to prove that @xmath148 for a given integer @xmath139 , especially if @xmath149 , since it must be shown that @xmath139 does not lie in the delta set of even a single element of @xmath3 .
example [ e : deltanumerical ] demonstrates the sublety of this question .
[ e : deltanumerical ] the delta set of elements of the numerical monoid @xmath150 are depicted in figure [ fig : deltanumerical ] .
notice that @xmath151 , but @xmath26 only appears in @xmath152 , @xmath153 , and @xmath154 . indeed , for @xmath155 , @xmath156
; see theorem [ t : deltaperiodic ] .
from example [ e : deltanumerical ] , produced using ` sage ` @xcite . here , each point @xmath157 indicates that @xmath158.,width=576 ] [ r : deltablock ] in contrast to numerical monoids , the delta set of any block monoid @xmath159 over a finite abelian group @xmath77 is known to be an interval , and recent work has focused on determining its maximal element ( * ? ? ? * section 6.7 ) .
in particular , any positive integer @xmath160 must also lie in @xmath161 .
roughly speaking , block monoids have abundant irreducible elements , so if @xmath162 for some @xmath163 , one can produce an element in @xmath24 with @xmath164 in its delta set by simply adding the appropriate irreducible elements to @xmath138 .
this process effectively `` cuts down '' a length gap @xmath139 by carefully filling in a particular factorization length .
the difficulty in controlling the phenomenon discussed in remark [ r : deltablock ] when considering numerical monoids motivates problem [ pr : deltarealization ] , which is known as the _ delta set realization problem for numerical monoids_. [ pr : deltarealization ] let @xmath165 be a finite set of positive integers . determine if there exists a numerical monoid @xmath3 such that @xmath166 .
currently , the only known restrictions on @xmath146 for a numerical monoid @xmath3 are given in theorem [ t : mindelta ] , each of which also holds for a much larger class of monoids .
[ t : mindelta ] if @xmath24 is an atomic , cancellative , commutative monoid with @xmath167 , then , @xmath168 . if , additionally , @xmath24 is finitely generated , then @xmath161 is finite .
[ r : deltarealization ] progress on problem [ pr : deltarealization ] has come from finding specific subsets that can be realized as delta sets .
for example , every set of the form @xmath169 is known to occur as the delta set of a numerical monoid @xcite .
additionally , a family of numerical monoids whose delta sets are @xmath170 for any @xmath171 is given @xcite .
recent investigation into problem [ pr : deltarealization ] has greatly benefited from evidence produced by computer algebra systems .
that said , until very recently , there was no known algorithm to compute the delta set of an entire numerical monoid @xmath3 . although the delta set @xmath127 of an element @xmath31 can be computed by computing its set of factorizations @xmath36 , it is less straight forward to compute the entire set @xmath146 , as this requires taking the union of the delta sets of all ( infintely many ) elements of @xmath3 . before discussing how this issue can be resolved ( see corollary [ c : deltacompute ] ) , we state theorem [ t : maxdelta ] , which demonstrates that the maximal element @xmath172 in the delta set of a numerical monoid @xmath3 can be obtain by computing the delta sets of a special ( computable ) class of monoid elements , called _ betti elements _
( definition [ d : betti ] ) .
[ d : betti ] let @xmath98 be a numerical monoid and @xmath31 .
consider the graph @xmath173 with vertex set given by its set of factorizations @xmath36 and an edge connecting @xmath174 if @xmath66 and @xmath175 have disjoint support as vectors .
that is , @xmath176 and @xmath177 are adjacent in @xmath173 if for all @xmath178 , @xmath179 and @xmath180 are never both non - zero .
an element @xmath31 is called a _ betti element _ if @xmath173 is disconnected .
the _ set of betti elements of @xmath3 _ is denoted by @xmath181 .
[ r : betti ] every numerical monoid contains only finitely many betti elements . when @xmath182 is a numerical monoid with two irreducible elements ,
@xmath17 is the unique betti element . in the case
when @xmath50 has three irreducible elements , there are at most three betti elements , each of which is an integer multiple of one of the generators of @xmath3 ( * ? ? ?
* section 9.3 ) .
[ t : maxdelta ] let @xmath3 be a numerical monoid .
the largest element of @xmath146 is achieved at a betti element of @xmath3 . in other words , @xmath183
betti elements are discussed in more depth in section [ s : catenary ] , as they are useful in finding maximal and minimal catenary degrees achieved in a numerical monoid ( see definitions [ d : catenarydegree ] and [ d : catenaryset ] ) .
[ e : bettidelta ] as stated in remark [ r : betti ] , for the numerical monoid @xmath12 with two irreducible elements , @xmath17 is the unique betti element .
in fact , it has precisely two factorizations , namely @xmath184 and @xmath185 .
since their difference in length equals @xmath135 , we have @xmath186 . of course ,
by theorem [ t : deltaarithmetic ] , @xmath187 theorem [ t : deltaperiodic ] states that delta sets for numerical monoids are eventually periodic with period dividing @xmath188 .
the value of @xmath189 given in @xcite is explicitly computed , albeit cumbersome to state .
most impressive about theorem [ t : deltaperiodic ] is that it provides a concrete upper bound for the beginning of the periodic behavior , and thus a finite set of monoid elements the union of whose delta sets equals the delta set of the entire numerical monoid . as a consequence ,
theorem [ t : deltaperiodic ] also yields the first known algorithm to compute the delta set of any numerical monoid ; see corollary [ c : deltacompute ] .
[ t : deltaperiodic ] for @xmath30 numerical , there exists a number @xmath189 such that for any @xmath31 with @xmath190 , @xmath191 .
[ c : deltacompute ] given a numerical monoid @xmath98 , we have @xmath192 } \!\!\ !
\delta(n),\ ] ] where @xmath193 .
[ e : deltaperiodic ] let @xmath150 as in example [ e : deltanumerical ] .
readily visible in figure [ fig : deltanumerical ] is the eventual periodicity of @xmath194 described by theorem [ t : deltaperiodic ] .
in particular , theorem [ t : deltaperiodic ] ensures that @xmath195 for all @xmath196 .
computing @xmath127 for smaller values of @xmath20 demonstrates that , in fact , @xmath197 holds for all @xmath198 .
[ r : deltacompute ] while corollary [ c : deltacompute ] provides the delta set of the entire numerical monoid by taking the union of only finite many element - wise delta sets , the number of elements @xmath20 for which @xmath127 needs to be computed becomes prohibitively large , especially when larger generators are used .
aside from recent improvements in the special case when @xmath3 has 3 minimal generators @xcite , few refinements are known for the set of elements whose delta sets must be computed .
alternatively , we can utilize a dynamic algorithm for computing length sets to dynamically compute @xmath127 and @xmath146 ; see remark [ r : deltadynamic ] . [
r : deltadynamic ] in the numerical monoid @xmath56 , it is possible to compute @xmath199 directly from @xmath200 , @xmath201 , and @xmath202 , without computing @xmath203 .
the key observation is that one can produce a factorization of @xmath25 from a factorization of @xmath204 , @xmath205 or @xmath206 by adding a single irreducible element , and this opereration always increases factorization length by exactly one .
consider the following computation : @xmath207 this provides an inductive ( dynamic ) method to compute all the length sets necessary to employ corollary [ c : deltacompute ] without having to compute any factorization sets , and the resulting algorithm is significantly faster than one dependent on factorizations . indeed , the set @xmath36 of factorizations of @xmath20 grow large very quickly @xcite , whereas @xmath208 grows linearly in @xmath20 ( * ? ? ?
* theorem 4.3 ) .
for a more detailed analysis , see ( * ? ? ?
* section 3 ) .
many of the factorization invariants discussed above ( e.g. , length sets , delta sets , and elasticity ) measure only information about the lengths of factorizations of an element .
while helpful in measuring the non - uniqueness of factorizations of an element , these invariants lose certain information when passing from the set of factorizations to the set of lengths ( theorem [ t : setsoflength ] ) . the _ catenary degree _ , which we describe in this section ,
is derived directly from factorizations and is not uniquely determined by sets of factorization lengths .
[ d : catenarydistance ] fix a numerical monoid @xmath30 , and fix an element @xmath31 . for @xmath174 , the _ greatest common divisor of @xmath66 and @xmath175 _ is given by @xmath209 and the _ distance between @xmath66 and @xmath175 _ ( or the _ weight of @xmath210 _ ) is given by @xmath211 the function @xmath212 measures how many irreducible elements two factorizations do not share in common . as implied by the notation
, @xmath213 is a metric on the set @xmath36 ( * ? ? ? * proposition 1.2.5.3 ) .
in fact , it is common to represent @xmath36 as the vertices of a complete graph , where the edge between factorizations @xmath66 and @xmath175 is decorated with @xmath214 .
below , we use the metric @xmath139 to define the _ catenary degree _ , which measures the `` overall '' distance between distinct factorizations of a given monoid element .
[ d : catenarydegree ] given @xmath174 and @xmath215 , an _
@xmath216-chain from @xmath66 to @xmath175 _ is a sequence @xmath217 of factorizations of @xmath20 such that ( i ) @xmath218 , ( ii ) @xmath219 , and ( iii ) @xmath220 for all @xmath221 .
the _ catenary degree of @xmath20 _ , denoted @xmath222 , is the smallest @xmath223 such that there exists an @xmath216-chain between any two factorizations of @xmath20 . with respect to the decorated complete graph representing @xmath36 ,
the catenary degree @xmath222 can be computed by repeatedly removing edges of maximal distance until the graph becomes disconnected .
the distance decorating the last removed edge is equal to @xmath222 .
example [ e : catenarydegree ] describes such a computation .
[ e : catenarydegree ] consider the numerical monoid @xmath224 .
the left - hand picture in figure [ fig : catenarydegree ] depicts the factorizations of @xmath225 along with all pairwise distances .
there exists a @xmath226-chain between any two factorizations of @xmath227 ; one such @xmath226-chain between @xmath228 and @xmath229 is depicted with bold red edges . since every @xmath226-chain between these factorizations contains the edge labeled 16 , we have @xmath230 .
this can also be computed in a different way . in the right - hand picture of figure
[ fig : catenarydegree ] , only distances of at most @xmath226 are depicted , and the resulting graph is connected . removing the edge labeled @xmath226 yields a disconnected graph ,
so we again conclude @xmath230 .
, as in example [ e : catenarydegree ] .
each vertex is labeled with an element of @xmath231 , and each edge is labeled with the distance between the factorizations at either end . the diagram on the left depicts all edges , and the diagram on the right includes only those edges labeled at most @xmath230 .
both graphics were created using the computer algebra system ` sage ` @xcite.,title="fig:",width=240 ] , as in example [ e : catenarydegree ] .
each vertex is labeled with an element of @xmath231 , and each edge is labeled with the distance between the factorizations at either end . the diagram on the left depicts all edges , and the diagram on the right includes only those edges labeled at most @xmath230 .
both graphics were created using the computer algebra system ` sage ` @xcite.,title="fig:",width=240 ] while the catenary degree of an individual element contains information about its different factorizations , the range of catenary degrees achieved by all elements in a monoid has also been studied .
definition [ d : catenaryset ] gives two ways to measure how the catenary degree varies throughout a numerical monoid .
[ d : catenaryset ] the _ set of catenary degrees of @xmath3 _ is the set @xmath232 , and the _ catenary degree of @xmath3 _ is the maximal catenary degree : @xmath233 . by far , the more
well - understood of these two invariants is @xmath234 , which has been computed for several classes of numerical monoids .
theorem [ t : catenarycomputation ] gives a summary of some known results ( additionally , see theorem [ t : maxcatdegree ] ) .
[ t : catenarycomputation ] fix a numerical monoid @xmath3 .
a. if @xmath12 , then @xmath235 . b. if @xmath236 for positive integers @xmath237 with @xmath88 and @xmath144 , then @xmath238 . c. if @xmath239 , where @xmath240 and @xmath241 for all @xmath242 , then @xmath243 .
while a closed form for @xmath222 for every element @xmath20 is a numerical monoid is rare , such a formula has been found for numerical monoids generated by an arithmetic sequence .
[ t : catnaryarithmetic ] suppose @xmath244 for positive integers @xmath245 with @xmath88 and @xmath144
. then @xmath246 for each nonzero @xmath31 , where @xmath247 .
in particular , @xmath248 . while it remains a lofty goal to give a closed form for the catenary degree of individual elements in the general setting , the set of catenary degrees may prove to be more approachable .
problem [ pr : catenarysets ] is , in many ways , an analog of problem [ pr : deltarealization ] for the set of catenary degrees .
[ pr : catenarysets ] determine which finite sets equal @xmath249 for some numerical monoid @xmath3 .
many of the arguments that compute the maximal catenary degree @xmath234 for a finitely generated monoid @xmath3 focus on the betti elements of @xmath3 ( definition [ d : betti ] ) .
the factorizations of betti elements contain enough information about @xmath250 to give sharp bounds on the catenary degrees occuring in @xmath249 ( theorem [ t : maxcatdegree ] ) , and can be readily computed using the ` gap ` package ` numericalsgps ` @xcite .
[ t : maxcatdegree ] for any finitely generated monoid @xmath3 , we have @xmath251 [ e : bettielements ] let @xmath252 .
the catenary degrees of @xmath3 are plotted in figure [ fig : bettielements ] .
the only betti elements of @xmath3 are 58 , 150 , and 154 , which have catenary degrees 4 , 12 , and 14 , respectively .
the betti elements can be seen in the plot as the first occurence of each of these three values .
notice that @xmath253 is distinct from each of these values . by theorem [ t : maxcatdegree ]
, every element of @xmath3 with at least two distinct factorizations has catenary degree at least 4 and at most 14 .
discussed in example [ e : bettielements].,width=480 ] as with the delta set , understanding the long - term behavior of the catenary degree ( when viewed as a function @xmath254 ) is of particular interest .
theorem [ t : catenaryperiodic ] shows that the catenary degree in numerical monoids is also eventually periodic , a phenomenon that can be seen in figure [ fig : bettielements ] .
[ t : catenaryperiodic ] fix a numerical monoid @xmath30 .
the catenary degree function @xmath255 is eventually periodic , and moreover , its period divides @xmath60 . [
r : catenaryperiodic ] among the eventual periodicity results in this paper ( theorems [ t : elasticityset ] , [ t : deltaperiodic ] , [ t : catenaryperiodic ] , and [ t : omegaquasi ] ) , theorem [ t : catenaryperiodic ] provides the most room for improvement
. there is currently no known bound on the start of this periodic behavior , as its proof in @xcite relies on the fact that a nonincreasing sequence of positive integers is eventually constant . additionally , the period is only known to divide the least common multiple of the monoid generators , but in practice this bound is far from sharp .
it remains an interesting problem to refine understanding of these properties .
examination of the delta set and the @xmath1-primality invariant ( intoduced in section [ s : omega ] ) both benefit from the use of dynamic algorithms ; see remarks [ r : deltadynamic ] and [ r : omegadynamic ] .
an answer to problem [ pr : catenarydynamic ] is likely within reach , and would aide in refining theorem [ t : catenaryperiodic ] .
[ pr : catenarydynamic ] find a dynamic algorithm to compute the catenary degree of numerical monoid elements .
in cancellative , commutative monoids , non - unique factorizations arise from irreducible elements that are not prime .
the rich factorization theory in numerical monoids stems in part from the fact that no element is prime . in this section ,
we develop the notion of @xmath1-primality , which measures how `` far from prime '' an element is in a natural way ( see @xcite for a more thorough introduction ) .
we begin with a definition .
[ d : omega ] let @xmath98 be a numerical monoid . for @xmath31
, define the
_ @xmath1-primality of @xmath20 _ to be @xmath256 if @xmath257 is the smallest positive integer with the property that whenever @xmath258 satisfies @xmath259 with @xmath260 , there exists a @xmath261 with @xmath262 for each @xmath178 such that @xmath263 and @xmath264 . in a multiplicative monoid @xmath24 ,
an element @xmath265 is prime if whenever @xmath266 for @xmath267 , then either @xmath268 or @xmath269 .
definition [ d : omega ] is a generalization of primeness , written additively in the context of a numerical monoid .
in fact , in a general cancellative , commutative monoid , @xmath270 if and only if @xmath20 is prime . in practice ,
the @xmath1-primality of an element @xmath20 of a numerical monoid @xmath30 is computed by finding those @xmath258 that are maximal among those with the property that @xmath271 lies in @xmath3 ( definition [ d : bullet ] ) .
proposition [ p : bullet ] states their precise relationship to the @xmath1-value of @xmath20 .
[ d : bullet ] we say that @xmath258 is a _ bullet _ for @xmath20 if ( i ) @xmath272 and ( ii ) @xmath273 whenever @xmath274 .
a bullet @xmath66 for @xmath20 is _ maximal _ if @xmath275 . the _ set of bullets _ for an element @xmath20
is denoted by @xmath276 .
in general , an element @xmath20 has several bullets with different lengths . as stated in proposition
[ p : bullet ] , the largest such length coincides with the @xmath277 , justifying the above definition for a _
maximal bullet_. [ e : omega ] let @xmath56 denote the numerical monoid from example [ e : numericalmonoid ] .
consider the following sets : @xmath279 we see that @xmath280 since @xmath281 , but @xmath282 .
similarly , @xmath283 since @xmath284 , but omitting a single @xmath26 or @xmath27 produces a difference lying outside of @xmath3 .
once the above sets have been determined , proposition [ p : bullet ] implies that @xmath285 and @xmath286 .
[ r : omegadynamic ] several recent works examine methods of computing @xmath1-primality in numerical monoids @xcite . at the time of writing ,
the fastest known algorithm ( which is implemented in the ` gap ` package ` numericalsgps ` @xcite ) utilizes dynamic programming to compute bullets @xcite , similar to the dynamic algorithm used to compute the delta set ( remark [ r : deltadynamic ] ) .
example [ e : omegadynamic ] demonstrates the idea behind this algorithm .
[ e : omegadynamic ] resuming notation from example [ e : omega ] , for each bullet @xmath287 , either @xmath288 or @xmath289 .
notice that it is impossible for both of these to lie in @xmath290 by definition [ d : bullet](ii ) .
similarly , bullets in @xmath291 and @xmath292 give rise to bullets in @xmath290 .
moreover , each bullet for @xmath25 is the `` image '' of a bullet for @xmath206 , @xmath205 or @xmath204 in this way .
much like the recurrence relation for sets of factorizations ( remark [ r : deltadynamic ] ) , the set of bullets of an element @xmath31 is determined by the `` images '' of the bullets of @xmath293 , @xmath294 and @xmath295 ; see @xcite for more detail .
in general , the existence of non - unique factorizations coincides with the existence of non - prime irreducible elements . as such
, the @xmath1-primality invariant was originally constructed to measure how far irreducible elements were from being prime ( see , for example , @xcite ) .
although some progress has been made in the setting of numerical monoids @xcite , problem [ pr : omegagens ] remains largely open .
[ r : tamedegree ] both the catenary degree ( definition [ d : catenarydegree ] ) and a variation called the tame degree share a surprising connection to @xmath1-primality . given @xmath31 , the _ tame degree _ @xmath296 equals the minimum value @xmath257 such that every factorization of @xmath20 with at least one zero entry is distance at most @xmath257 ( in the sense of definition [ d : catenarydistance ] ) from a fully supported factorization of @xmath20 , or @xmath297 if @xmath20 has no fully supported factorizations @xcite . as with the catenary degree ,
the tame degree of @xmath3 is defined as @xmath298 and is known to be finite .
it is not hard to show that @xmath299 , but what is perhaps more surprising is that @xmath300 , where @xmath301 is the maximum @xmath1-value obtained at an irreducible @xcite .
this connection has been explored in some of the recent literature , in part as a method of bounding @xmath302 @xcite .
recent investigations focusing on the @xmath1-values of all non - unit elements in monoids have uncovered interesting asymptotic behavior . in the setting of numerical monoids
, @xmath1-primality grows in a periodic , linear fashion when viewed as a function @xmath303 .
[ t : omegaquasi ] let @xmath98 be a numerical monoid .
the @xmath1-primality function is eventually quasilinear .
more specifically , for @xmath304 , we have @xmath305 for every @xmath306 , where @xmath307 is @xmath14-periodic .
-function for the numerical monoids @xmath310 ( left ) and @xmath311 ( right ) , produced using ` sage ` @xcite.,title="fig:",width=240 ] -function for the numerical monoids @xmath310 ( left ) and @xmath311 ( right ) , produced using ` sage ` @xcite.,title="fig:",width=240 ] [ r : omegaquasi ] we resume notation from theorem [ t : omegaquasi ] .
in contrast to theorem [ t : deltaperiodic ] , the period of the function @xmath312 is known to be exactly @xmath14 @xcite . additionally , the value of @xmath313 , though not tight , is in general a relatively good bound on the start of the quasilinear behavior of @xmath314 ( * ? ? ?
* remark 6.8 ) .
each of these improvements on the original statement of theorem [ t : omegaquasi ] in @xcite are heavily motivated by data produced by the dynamic algorithm described in remark [ r : omegadynamic ] .
v. blanco , p. garca - snchez , and a. geroldinger , _ semigroup - theoretical characterizations of arithmetical invariants with applications to numerical monoids and krull monoids _ , illinois journal of mathematics , 55 * 4 * ( 2011 ) , 13851414 s. chapman , p. garca - snchez , d. llena , v. ponomarenko , j. rosales , _ the catenary and tame degree in finitely generated commutative cancellative monoids _
, manuscripta math .
120 ( 2006 ) , no .
3 , 253264 . j. garca - garca , m. moreno - fras , a. vigneron - tenorio , _ computation of the @xmath1-primality and asymptotic @xmath1-primality with applications to numerical semigroups _ , israel j. math .
206 ( 2015 ) , no .
1 , 395411 .
c. oneill and r. pelayo , _ how do you measure primality ?
_ , american mathematical monthly * 122 * ( 2014 ) , no . 2 , 121137
c. oneill , v. ponomarenko , r. tate , and g. webb , _ on the set of catenary degrees of finitely generated cancellative commutative monoids _
, preprint .
available at .
ramrez alfonsn , _ the diophantine frobenius problem _ , oxford lecture series in mathematics and its applications , 30 .
oxford university press , oxford , 2005 .
xvi+243 pp .
isbn : 978 - 0 - 19 - 856820 - 9 ; 0 - 19 - 856820 - 7 | nonunique factorization in commutative monoids is often studied using factorization invariants , which assign to each monoid element a quantity determined by the factorization structure . for numerical monoids ( co - finite , additive submonoids of @xmath0 ) ,
several factorization invariants have received much attention in the recent literature . in this survey article
, we give an overview of the length set , elasticity , delta set , @xmath1-primality , and catenary degree invariants in the setting of numerical monoids . for each invariant , we present current major results in the literature and identify the primary open questions that remain . |
recently , a new flow on riemannian manifolds is introduced by kong and liu(@xcite , @xcite , @xcite ) : + @xmath2 + this is the hyperbolic version of the famous ricci flow .
they studied its short - time existence in compact case , derived evolutions of curvatures which have wave character .
a dissipative flow was also considered in @xcite .
one remarkable result is @xcite , in which the authors proved that on compact surfaces with metric depending on one space variable , the flow has a global solution if the initial velocity is large enough .
so given an initial metric , the geometric surgery can be replaced by choosing suitable velocity to allow the long - time existence . however , the evolutions of riemann curvature , ricci curvature and the scalar curvature are very complicated which contain the first - order derivative of time that is hard to control . in this paper ,
i consider its complex version which is also introduced by kong and liu in @xcite : + @xmath3 + as we will see on calabi - yau manifolds , it can be reduced to a scalar ( complex ) hyperbolic monge - amp@xmath1re equation which is much easier to handle .
the flow considered here can also be regarded as the hyperbolic version of k@xmath0hler - ricci flow : + @xmath4 + in the past several decades , nonlinear partial differential equations played important role in differential geometry .
for example , the resolutions of calabi s conjecture @xcite and poincar@xmath5 s conjecture @xcite @xcite @xcite @xcite @xcite are due to the methods of geometric analysis and the elliptic and parabolic type of equations are intensively studied . however , the hyperbolic equations have been ignored for a long time . for the elliptic and parabolic equations
, we have the powerful tool the maximum principle which can not be applied to hyperbolic ones .
this lack of practical tool may explain the reason why the research of hyperbolic equations on manifolds are not as active as the other two types .
however , perelman @xcite @xcite @xcite introduced several powerful tools into the study of flow like the energy functional , the monotonicity formula , and the space - time geometry . in the euclidean space ,
these methods are available for hyperbolic equations .
so i believe we can refer to perelman s methods to study nonlinear hyperbolic equations on manifolds .
hyperbolic equations are important in physics especially in general relativity .
the famous einstein s equation has been studied for a long time .
in fact , as illustrated in @xcite , the hyperbolic version of ricci flow is closely related to einstein s equation .
so i hope , by studying the wave character of manifolds through hyperbolic equations , we can better understand the basic structure of the universe .
there is already some work on wave equations on lorentzian manifolds which catches more interest from physicists .
although wave equations on riemannian manifolds are also considered , the results are few and have a more analytic style that its influence on curvature are not considered . in my opinion , analyzing wave equations on manifolds from a more geometric viewpoint so that it may give more help to understand the geometry and topology of manifolds .
+ let @xmath6 be a complete k@xmath0hler manifold , where @xmath7 is the metric depending on time and @xmath8 is a fixed complex structure .
the hyperbolic k@xmath0hler - ricci flow is the following evolution equation + @xmath9 + for a family of k@xmath0hler metrics @xmath10 on @xmath11 .
a natural and fundamental problem is the short - time existence and uniqueness theorem of ( 1.1 ) .
we will see as in the case of k@xmath0hler - ricci flow , the evolution equation of metrics can be simplified to a scalar evolution equation . in the present paper ,
i also derive the corresponding wave equations for various geometric quantities . + * theorem 1.1 .
* _ let @xmath12 be a compact calabi - yau manifold with @xmath13 a k@xmath0hler metric .
then there exists a constant @xmath14 such that the initial value problem + @xmath15 + has a unique smooth solution @xmath16 on @xmath17 $ ] and @xmath18 remains k@xmath0hler for any @xmath19 as long as the solution exists and the initial velocity must satisfy @xmath20=0 $ ] where @xmath21 is the k@xmath0hler form of @xmath22 . _
+ similar to k@xmath0hler - ricci flow , i will derive the corresponding wave equations for the curvatures . + * theorem 1.2 . * _ under the hyperbolic k@xmath0hler - ricci flow ( 1.1 ) , the riemannian curvature tensor , ricci curvature and scalar curvature satisfy the evolution equations under a unitary coordinate + @xmath23 + @xmath24 + @xmath25 + here @xmath26 .
_ + particularly , on calabi - yau manifolds , the flow can be reduced to a hyperbolic monge - amp@xmath1re equation : + @xmath27 + where + @xmath28 + the paper is organized as follows . in section 2
, i will review briefly some basics in k@xmath0hler geometry . in section 3 , i will discuss the derivation and basic facts about hyperbolic k@xmath0hler - ricci flow in detail . in section 4 , evolutions of various geometric quantities including the curvatures will be derived . + * acknowledgement . * _ the author thanks professor de - xing kong and professor ke - feng liu for their helpful suggestion without whose help , the come out of this paper is impossible .
the author also thanks wen - rong dai and xiao - feng sun for their useful discussion . _ +
i state some basic facts about k@xmath29hler geometry that will be used in this paper . for a thorough study of k@xmath29hler geometry , @xcite @xcite @xcite @xcite are good choices .
let @xmath11 be an n - dimensional compact k@xmath0hler manifold .
a hermitian metric is given by + @xmath30 + its associated k@xmath0hler form is + @xmath31 + the k@xmath0hler condition requires that @xmath32 is a closed positive global @xmath33-form .
the riemann curvature is locally given by + @xmath34 + and has the symmetries + @xmath35 + the ricci curvature is defined by + @xmath36 + and locally we have + @xmath37 + the associated ricci form is + @xmath38 + it is a real closed @xmath33-form and represents the first chern class . the scalar curvature is + @xmath39 + given a k@xmath0hler metric , there are some other important quantities .
the nonzero christoffel symbols are given by + @xmath40 + the volume form is + @xmath41 + the average scalar curvature is + @xmath42 + finally i give an important lemma which is a consequence of the hodge decomposition theorem . + * lemma 2.1(@xmath43lemma ) . *
_ let @xmath11 be a compact k@xmath0hler manifold .
if @xmath44 is an exact real ( 1,1)-form , then there exists a real - valued function @xmath45 on @xmath11 such that @xmath46 .
that is , + @xmath47 + where @xmath48 and @xmath49 . _
+ * proof .
* this is a standard result in the theory of k@xmath0hler manifolds , see @xcite for example
. @xmath50
in this section i will discuss the hyperbolic k@xmath0hler - ricci flow and its equivalent formulation as a single hyperbolic monge - amp@xmath1re equation .
for the k@xmath0hler - ricci flow theory , please see @xcite .
i also suggest reader to read hamilton s papers for more details about ricci flow @xcite @xcite .
the book @xcite is also a good choice .
given a compact complex manifold @xmath51 , consider the hyperbolic k@xmath0hler - ricci flow equation + @xmath52 + for a 1-parameter family of k@xmath0hler metrics , which is obtained from the ricci flow by dropping the factor of 2 with respect to the convention . on a calabi - yau manifold , the first chern class vanishes , so @xmath53=0 $ ] . under this condition , from the @xmath54 ,
the ricci tensor has a potential , i.e. + @xmath55 + where @xmath56 is a function defined on the manifold .
assume + @xmath57 + where @xmath58 is a function on the manifold , since we have + @xmath59 + hence + @xmath60 + so on calabi - yau manifolds , the hyperbolic k@xmath0hler - ricci flow equation is equivalent to the following hyperbolic ( scalar ) complex monge - amp@xmath1re equation due to the maximum principle for compact manifolds : + @xmath61 + for some function of time @xmath62 satisfying the compatibility condition + @xmath63 + here the volume @xmath64 is under the metric @xmath65 .
further , by a time - dependent translation of @xmath58 , we can drop the factor @xmath62 .
further we have + @xmath66 + so we can let @xmath67=0 $ ] . from the hyperbolicity and short time existence and uniqueness of our hyperbolic monge - amp@xmath5re equation ,
we have proved the theorem 1.1 .
next let us consider the normalized flow which is also studied in @xcite in the real case .
choose the normalization factor @xmath68 ( note : this has nothing to do with the @xmath69 above ) , + @xmath70 + such that + @xmath71 + and choose a new time parameter + @xmath72 + noting that for the normalized metric @xmath73 , we have + @xmath74 + thus + @xmath75 + @xmath76 + where @xmath44 and @xmath77 are certain functions of @xmath78 . + next
let us consider the following hyperbolic system + @xmath79 + where + @xmath28 + let + @xmath80 + we get + @xmath81 + so + @xmath82 + thus we have + @xmath83 + note that the laplacian operator here is time - dependent so the equation is genuinely nonlinear .
the hyperbolic k@xmath0hler - ricci flow is a hyperbolic evolution equation on the metrics .
the evolution of the metrics implies nonlinear wave equations for the riemannian curvature tensor @xmath84 , the ricci curvature @xmath85 and the scalar curvature @xmath86 which i will derive .
i also derive the evolutions of some other quantities in this section . for the evolutions which are similar to those in this section associated to k@xmath0hler - ricci flow
, please cite @xcite .
let @xmath11 be an n - dimentional compact k@xmath0hler manifold .
let us consider the hyperbolic k@xmath0hler - ricci flow on @xmath11 , + @xmath87 + * proposition 4.1(riemannian curvature tensor ) .
* _ in a unitary frame , + @xmath88 + the above formula also holds in arbitrary holomorphic coordinates if repeated indices are contracted via the metric .
i.e. , if @xmath89 is replaced by @xmath90 , etc . _
+ in the k@xmath0hler case , it is sometimes convenient to compute locally in holomorphic coordinates in terms of ordinary derivatives .
the following lemma translates these ordinary derivatives to the covariant derivatives , @xcite . +
* lemma 4.2(relation between ordinary and covariant derivatives ) . *
_ let @xmath91 be a closed ( 1,1)-form . locally it is represented by @xmath92 , which is hermitian symmetric .
let @xmath93 denote the covariant derivatives and @xmath94 denote @xmath95 .
then at the center @xmath96 of normal holomorphic coordinates + @xmath97 @xmath98 + _ * proof .
+ in the above calculation , we use the fact that @xmath91 is closed so @xmath100 .
the second formula is the conjugate of the first .
@xmath50 * proof of proposition 4.1 . *
recall that + @xmath101 + this implies that , in a normal holomorphic coordinate system centered at any given point + @xmath102 + the proposition follows from the following two identities : + @xmath103 + and + @xmath104 + recall that @xmath26 .
@xmath50 * lemma 4.3 . *
_ we will need the following two basic formulas under hyperbolic k@xmath0hler - ricci flow : + @xmath105 + + @xmath106 + _ * corollary 4.4(ricci curvature ) . * _ the ricci curvature satisfies the following in a unitary frame : + @xmath107 _ * proof . * + @xmath108 + put the evolutions of @xmath84 and @xmath109 in , we get the result .
@xmath50 * proposition 4.5(evolution of r ) . * _ the scalar curvature @xmath86 evolves by + @xmath110 + _ * proof .
* first we have + @xmath111 + + @xmath112 + thus + @xmath113 + so finally + @xmath114 + @xmath50 next i present the evolution equations of the christoffel symbols and the volume form .
+ * proposition 4.6 .
* _ the christoffel symbols and the volume form evolve respectively by + @xmath115 + + @xmath116d\mu.\ ] ] + _ * proof .
* recall that + @xmath117 + thus + @xmath118 + so finally in normal coordinate + @xmath119 + the evolution of @xmath120 follows from the quantities + @xmath121 + and + @xmath122 @xmath50 * proposition 4.7(potential of ricci curvature ) .
* _ on compact k@xmath0hler manifolds , the potential @xmath123 of ricci curvature @xmath124 satisfies : + @xmath125 + _ * proof . *
+ @xmath126 + so from the maximum principle of compact manifolds , + @xmath125 + @xmath50 by a suitable time - dependent translation of @xmath123 , we can drop the factor @xmath127 .
* proposition 4.8 .
* _ on compact k@xmath0hler manifolds , the evolutions of @xmath128 and @xmath129 are : + @xmath130d\mu + 2\int_{\mathcal { m}^n}\frac{\partial}{\partial t}r\frac{\partial}{\partial t}d\mu\\ & & -2\int_{\mathcal { m}^n}\biggl\langle\frac{\partial}{\partial t}g_{\alpha\bar{\beta } } , \frac{\partial}{\partial t}r_{\alpha\bar{\beta}}\biggr\rangle d\mu\\ & & + 2\int_{\mathcal { m}^n}r_{k\bar{l}}\biggl(\frac{\partial}{\partial t}g_{n\bar{m}}\biggr)\biggl(\frac{\partial}{\partial t}g_{r\bar{s}}\biggr)g^{r\bar{m}}g^{n\bar{l}}g^{k\bar{s } } d\mu,\end{aligned}\ ] ] + @xmath131 +
_ * proof . * since + @xmath132 + put the evolutions of @xmath86 and @xmath133 in , we get the result for @xmath128 .
+ since @xmath134 + and @xmath135 + put the evolutions of @xmath86 and @xmath133 in , we get the result for @xmath129 .
@xmath50 * remark . * in k@xmath0hler - ricci flow ,
the evolutions of @xmath128 and @xmath129 are + @xmath136 + and + @xmath137 + while in our case , all the extra terms are of first - order derivatives .
the flow considered here is the complex version of the hyperbolic geometric flow introduced by kong and liu @xcite .
kong et al @xcite proved that on riemann surfaces , the long time existence of hyperbolic geometric flow depends on the choice of the initial velocity .
we can expect that its complex version has similar property .
on calabi - yau manifolds , the flow can be simplified to a single complex hyperbolic monge - amp@xmath1re equation .
note that yau @xcite used the elliptic monge - amp@xmath1re equation to prove the famous calabi s conjecture and cao @xcite used its parabolic version and techniques from k@xmath0hler - ricci flow to reprove calabi s conjecture .
i hope the hyperbolic monge - amp@xmath1re equation is also powerful to understand k@xmath0hler manifolds .
we can expect this new flow is helpful to study wave phenomena in the nature especially the einstein equation . in the future , we will study several fundamental problems on the hyperbolic k@xmath0hler - ricci flow , for example , long - time existence , formation of singularities as well as physical applications . bennett chow , sun - chin chu , david glickenstein , christine guenther , jim isenberg , tom ivey , dan knopf , peng lu , feng luo , and lei ni . _
the ricci flow : techniques and application , part i : geometric aspects .
_ mathematical surveys and monographs , 135 , ams , providence , ri , 2007 . | in this paper , the author has considered the hyperbolic k@xmath0hler - ricci flow introduced by kong and liu @xcite , that is , the hyperbolic version of the famous k@xmath0hler - ricci flow .
the author has explained the derivation of the equation and calculated the evolutions of various quantities associated to the equation including the curvatures .
particularly on calabi - yau manifolds , the equation can be simplified to a scalar hyperbolic monge - amp@xmath1re equation which is just the hyperbolic version of the corresponding one in k@xmath0hler - ricci flow . |
galaxy interactions and mergers play important roles in triggering star formation and/or fueling the nuclear activity in the merging host galaxies @xcite .
recent high resolution simulations of major mergers show that large scale tidal forces as well as small scale turbulence and stellar feedback can significantly influence the distribution of gas , forming massive clumps of dense gas with @xmath13 = 10@xmath14
10@xmath15 m@xmath10 ( e.g. , * ? ? ?
* ; * ? ? ?
these simulations also predict that the star formation not only increases as the galaxies first collide , but it also persists at a higher rate throughout the merger process , peaking at the final coalescence .
( ultra-)luminous infrared galaxies ( u / lirgs ; * ? ? ?
* ) at low redshifts are almost exclusively strongly interacting and merging systems @xcite , often found at the mid to final stages of the merger .
the elevated level of infrared luminosity originates from the reprocessed emission from the dust particles surrounding the starburst or the active galactic nuclei ( agns ) , both of which are likely triggered by the tidal interaction . the highest gas surface densities ( @xmath16 = 5.4 @xmath8 10@xmath17 1.4 @xmath8 10@xmath18 m@xmath10 pc@xmath19 ) and consequently the highest star formation activities ( @xmath20 = @xmath21 1000 m@xmath10 yr@xmath3 kpc@xmath19 ) are usually found near the compact nuclear region .
dense molecular gas ( @xmath22 @xmath21 10@xmath18 10@xmath9
@xmath23 ) in u / lirgs directly shows nuclear gas distribution and kinematics ( e.g. , * ? ? ?
* ; * ? ? ?
they are often surrounded by diffuse gas ( @xmath22 @xmath21 10@xmath24
10@xmath25 @xmath23 ) that may or may not be directly associated with star formation activities . it has been demonstrated that the hcn ( 43 ) and hco@xmath4 ( 43 ) emission lines , whose critical densities are 8.5 @xmath8 10@xmath14 and 1.8 @xmath8 10@xmath14 @xmath23 , respectively , can be used as tracers of the dense gas ( e.g. , * ? ? ?
* ; * ? ? ?
on the other hand , co ( 10 ) and @xmath2co ( 10 ) line emission , whose critical densities are 4.1 @xmath8 10@xmath24 and 1.5 @xmath8 10@xmath25 @xmath23 , respectively , have been used extensively for tracing the global gas distribution and kinematics in merging u / lirgs ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in addition , the ratio of these lines ( e.g. @xmath1co/@xmath2co and hcn / hco@xmath4 ) have been used to investigate the properties of the ism @xcite or to search for buried agns ( e.g. , * ? ? ? * ; * ? ? ? * ) .
limitations in sensitivity and angular resolution have been the major obstacles in understanding the detailed distribution and kinematics of both dense and diffuse gas , and investigating the spatial variation of the line ratios and the physical condition of gas . ) . ] in this paper , we present _ atacama large millimeter / submillimeter array _ ( alma ) cycle 0 observations of the ir - bright merging galaxy vv 114 .
vv 114 is one of the best samples for studying the gas response during the critical stage when the two gas disks merge @xcite .
the target molecular lines include @xmath1co ( 10 ) , @xmath2co ( 10 ) , @xmath1co ( 32 ) , hcn ( 43 ) and hco@xmath4 ( 43 ) , and we also present the maps of ch@xmath0oh ( 2@xmath261@xmath26 ) , cs ( 21 ) , cn ( 1@xmath270@xmath27 ) , cn ( 1@xmath280@xmath27 ) , and cs ( 76 ) lines which were observed simultaneously within the same band .
the main aim of this study is to investigate the distribution and kinematics of the diffuse and dense molecular gas and to quantify the spatial variation of the excitation conditions across the two merging disks .
vv 114 is a gas - rich ( @xmath29 = 5.1 @xmath8 10@xmath30 m@xmath10 ; * ? ? ?
* ) nearby ( d = 82 mpc ; 10 = 400 pc ) interacting system ( figure [ fig_hst ] ) with high - infrared luminosity ( @xmath31 = 4.7 @xmath8 10@xmath32 @xmath33 ; * ? ? ?
the projected nuclear separation between the two optical galaxies ( vv 114e and vv 114w ) is about 6 kpc .
@xcite found a large amount of dust ( @xmath34 = 1.2 @xmath8 10@xmath15 m@xmath10 ) distributed across the system with a dust temperature of 20 25 k. about half of the warmer dust traced in the mid - ir ( mir ) is associated with the eastern galaxy , where both compact ( nuclear region ) and extended emission is found @xcite .
@xcite found a bimodal distribution of velocity dispersions of several atomic forbidden lines and emission line ratios indicative of composite activity explained by a combination of wide - spread shocks and star formation .
the wide - spread star formation is also revealed by pa@xmath35 observation using anir camera mounted on minitao ( * ? ? ?
* see also appendix [ a1 ] ) .
@xcite ( hereafter ) identified a highly obscured agn and compact starburst clumps using sub - arcsecond resolution alma cycle 0 observations of hcn ( 43 ) and hco@xmath4 ( 43 ) emission .
this paper is organized as follows .
we describe our observations and data reduction in [ obs ] , and results in [ result ] . in
[ ratio ] and [ der ] , we provide molecular line ratios and physical parameters , such as the gas / dust mass , the gas temperature , and the gas density . in [ dis ] ,
we present the properties of dense " gas ( [ dense ] ) , the comparison between molecular gas and star formation ( [ sfr ] ) , the discussions of the co isotope enhancement ( [ isotope ] ) , the gas - to - dust mass ratio ( [ g / d ] ) , the fractional abundances of cs , ch@xmath0oh , and cn relative to h@xmath7 ( [ chemical ] ) , and a potential tidal dwarf galaxy formations at the tip of the tidal arms of vv 114 ( [ tdg ] ) . we summarize and conclude this paper in [ conclusion ] . throughout this paper , we adopt h@xmath36 = 73 km s@xmath3 mpc@xmath3 , @xmath37 = 0.27 , and @xmath38 = 0.73 .
observations toward vv 114 were carried out as an alma cycle 0 program ( i d = 2011.0.00467.s ; pi = d. iono ) using fourteen twenty 12 m antennas .
the band 3 and band 7 receivers were tuned to the @xmath1co ( 10 ) , @xmath2 co ( 10 ) , @xmath1co ( 32 ) , hcn ( 43 ) , and hco@xmath4 ( 43 ) line emissions in the upper side band ( see table [ table_obs ] ) .
the @xmath1co ( 10 ) data were obtained on november 6 , 2011 and may 4 , 2012 in the compact and extended configurations , respectively .
the @xmath2co ( 10 ) data were obtained on may 27 and july 2 , 2012 in the compact configuration .
the @xmath1co ( 32 ) emission was observed on november 5 , 2011 in the compact configuration ( 7-point mosaic ) .
the hcn ( 43 ) and hco@xmath4 ( 43 ) data were obtained on july 1 , 2 , and 3 , 2012 in the extended configuration ( 3-point mosaic ) , simultaneously .
each spectral window had a bandwidth of 1.875 ghz with 3840 channels , and two spectral windows were set to each sideband to achieve a total frequency coverage of @xmath21 7.5 ghz in these observations .
the spectral resolution was 0.488 mhz per channel .
j1924 - 292 , j0132 - 169 , uranus ( neptune for band 3 observations ) were used for bandpass , phase , and flux calibrations .
detailed observational parameters are shown in table [ table_obs ] .
we used the delivered calibrated data and mapping was accomplished using the clean task in casa @xcite .
we made the data cubes with a velocity width of 5 km s@xmath3 for the @xmath1co line and 30 km s@xmath3 for the other lines .
all maps in this paper , except for @xmath1co ( 32 ) , are reconstructed with a briggs weighting and analyzed with miriad and aips .
the @xmath1co ( 32 ) images are created with uniform weighting ( see [ co32 ] ) .
the synthesized beam size of the @xmath1co ( 10 ) , @xmath2co ( 10 ) , @xmath1co ( 32 ) , and hcn ( 43 ) were 197 @xmath8 135 ( p.a .
= 82.3 deg . ) , 177 @xmath8 120 ( p.a .
= 85.8 deg . ) , 164 @xmath8 117 ( p.a .
= 112.6 deg . ) , and 046 @xmath8 038 ( p.a .
= 51.5 deg . ) , respectively .
we also detected cn ( 1@xmath280@xmath27 ) , cn ( 1@xmath270@xmath27 ) , cs ( 21 ) , ch@xmath0oh ( 2@xmath26 - 1@xmath26 ) , and cs ( 76 ) line emission for the first time in vv 114 .
the properties of these molecular lines are summarized in table [ table_data ] . all images which we constructed are corrected for primary beam attenuation .
the on - source times of band 3 and band 7 were about 40 minutes and 80 minutes , and the rms noise levels of the channel maps with 30 km s@xmath3 resolution are 1.0 mjy beam@xmath3 and 0.8 mjy beam@xmath3 , respectively . furthermore , we made continuum maps at each observing frequency by adding the line - free channels .
the rms level of the continuum images were 0.05 mjy beam@xmath3 , 0.11 mjy beam@xmath3 , and 0.07 mjy beam@xmath3 for band 3 , band 7 in the compact configuration , and band 7 in the extended configuration , respectively .
the continuum emission was subtracted in the @xmath39-plane before making the line images . throughout this paper ,
the pixel scales of the band 3 and the band 7 images are set to 03/pixel and 008/pixel , respectively , and only the statistical error is considered unless mentioned otherwise .
the systematic error on the absolute flux is estimated to be @xmath21 5% and @xmath21 10% for both sidebands in band 3 and band 7 , respectively . in the following sections , we estimate the missing flux of each molecular line for which the single dish data are available in literature .
although the effect of missing flux becomes critical when we evaluate the global gas properties and the corresponding line ratios , the effect is negligible when we discuss structures that are smaller than the maximum recoverable scale " ( mrs ) of each configuration of alma .
this is estimated from the minimum baseline lengths of the assigned antenna configurations and the observed frequencies .
the mrs of our observations are @xmath21 8 and @xmath21 7 in band 3 and band 7 , respectively ( table [ table_obs ] ) .
therefore the missing flux effect in this paper is negligible , since we derive physical parameters ( e.g. , molecular gas mass ) only for structures smaller than @xmath21 2 .
molecular line and continuum images are shown in figures [ fig_mom1 ] , [ fig_mom2 ] , [ fig_mom3 ] , [ fig_vv114e ] , and [ fig_contin ] .
the channel maps and the spectra of all line emissions are shown in appendix [ a2 ] and [ a3 ] .
the integrated intensity , velocity field , and velocity dispersion maps of vv 114 are shown in figures [ fig_mom1]a , [ fig_mom1]b , and [ fig_mom1]c , respectively . the total @xmath1co ( 10 ) integrated intensity of vv 114 is 594.6 @xmath6 1.6 jy km s@xmath3 , which is 1.3 times larger than that detected using the nrao 12 m telescope ( 461 jy km s@xmath3 ; * ? ? ?
this is because the pointing center for the nrao 12 m observation was 250 southwest of the co centroid identified from the alma map ( nrao 12 m : 01h07m45.7s , -17d30m36.5s ; co centroid : 01h07m47.2s , -17d30m25.8s ) . at the adopted distance of vv 114 ( 86 mpc ) , the 197 @xmath8 135 beam of the @xmath1co ( 10 ) observation
gives us a resolution of 790 pc @xmath8 540 pc .
the two crosses shown in all images represent the peaks obtained from the minitao / anir @xmath40s - band observation , and we regard them as the progenitor s nuclei .
the integrated @xmath1co ( 10 ) intensity map of vv 114 ( figure [ fig_mom1]a ) shows that the diffuse / cold gas forms two arm - like structures and a filamentary structure located at the center of the image .
the global gas distribution is consistent with the previous @xmath1co ( 10 ) observations @xcite .
the southeastern ( se ) arm clearly follows the tidal arm seen in the hst / acs image ( figure [ fig_hst ] ; * ? ? ?
* ) , while the northwestern ( nw ) arm has no counterpart in any other wavelengths .
the region from the center of vv 114 to the eastern nucleus shows a strong concentration of molecular gas ( @xmath41 55 west of the eastern nucleus ) , and we refer to this region as the overlap " region with a molecular filament " ( see figure [ fig_mom1 ] ) .
the @xmath1co ( 10 ) velocity field map of vv 114 ( figure [ fig_mom1]b ) shows a significantly broad velocity range across the galaxy disks ( @xmath41 600 km s@xmath3 ) .
the se arm has a blue - shifted velocity from 5650 km s@xmath3 to 5920 km s@xmath3 , while the nw arm has a red - shifted velocity from 5950 km s@xmath3 to 6160 km s@xmath3 .
one possibility for the larger velocity width in the se arm may be a highly inclined tidal arm .
two other arm - like features are also detected in the @xmath1co ( 10 ) observations .
one arm is located @xmath41 40 northeast of the eastern nucleus and shows an arc around the eastern nucleus in the velocity range of 5810 km s@xmath3 to 6180 km s@xmath3 .
the other arm is located @xmath41 100 west of the se arm and has a strong peak ( @xmath41 262.5 @xmath6 1.0 jy km
s@xmath3 ) in the velocity range of 5610 km s@xmath3 to 5900 km s@xmath3 . the overlap region has the highest velocity dispersion ( @xmath41 110 km s@xmath3 ) ( figure [ fig_mom1]c ) .
the nw arm has an average velocity dispersion of @xmath41 30 km s@xmath3 , while the se arm has @xmath41 40 km s@xmath3 .
these values are significantly higher than the dispersions seen in giant molecular clouds ( gmcs ) in the lmc ( 2 14 km s@xmath3 ; * ? ? ?
* ; * ? ? ?
* ) and slightly higher than that in giant molecular associations ( gmas ) in the antennae galaxy ( 6 36 km s@xmath3 ; * ? ? ?
we suggest that the main contribution to the @xmath1co ( 10 ) velocity dispersion is inter cloud turbulent medium along the tidal arm , and/or shocked region induced by the tidal interaction , rather than the velocity dispersion of the gmcs / gmas . the integrated intensity , velocity field , and velocity dispersion maps of @xmath2co ( 10 ) are shown in figures [ fig_mom1]d , [ fig_mom1]e , and [ fig_mom1]f , respectively .
the integrated @xmath2co ( 10 ) intensity map of vv 114 ( figure [ fig_mom1]d ) shows a filamentary structure across the galaxy disks , which is consistent with the region where the @xmath1co ( 10 ) filament is detected .
the total @xmath2co ( 10 ) integrated intensity is 5.9 @xmath42 jy km s@xmath3 .
the strongest peak is located @xmath41 42 southwest of the eastern nucleus .
the @xmath2co ( 10 ) velocity field map of vv 114 ( figure [ fig_mom1]e ) shows a narrower velocity range ( 5670 6000 km s@xmath3 ) than that of the @xmath1co ( 10 ) emission ( 5600 6200 km s@xmath3 ) .
this suggests that the @xmath2co ( 10 ) emission mainly comes from two components , the eastern galaxy and the blue - shifted component of the overlap region .
the @xmath2co ( 10 ) velocity dispersion map of vv 114 ( figure [ fig_mom1]f ) shows a high velocity dispersion component ( @xmath21 100 km s@xmath3 ) between the eastern nucleus and the overlap region .
this significant velocity dispersion may be caused by a superposition of clouds ( see the double - peak spectrum at r39 shown in appendix [ a3 ] ) .
the cs ( 21 ) and ch@xmath0oh ( 2@xmath261@xmath26 ) lines are only detected at the overlap region ( figures [ fig_mom1]g , [ fig_mom1]h , [ fig_mom1]i , [ fig_mom2]a , [ fig_mom2]b , and [ fig_mom2]c ) .
this is the first detection of the ch@xmath0oh ( 2@xmath261@xmath26 ) emission in a merger - induced overlap region .
we observed the blended set of 2@xmath43 1@xmath43 ( @xmath44 = 96.756 ghz , @xmath45 = 28.0 k ) , 2@xmath36
1@xmath36 _ e _
( @xmath44 = 96.745 ghz , @xmath45 = 20.1 k ) , 2@xmath36 1@xmath36 _ _ a__@xmath4 ( @xmath44 = 96.741 ghz , @xmath45 = 7.0 k ) , and 2@xmath46 1@xmath46 _ e _ ( @xmath44 = 96.739 ghz , @xmath45 = 12.5 k ) , thermal transitions of ch@xmath0oh ( hereafter designated the 2@xmath26 1@xmath26 transition ) .
the distribution of these molecular lines is clearly different from the other dense gas tracers detected in the current program .
the peaks of cs ( 21 ) and ch@xmath0oh ( 2@xmath261@xmath26 ) are coincident with one of the peaks of @xmath2co ( 10 ) to within 05 .
the total cs ( 21 ) and ch@xmath0oh ( 2@xmath261@xmath26 ) integrated intensities are 0.4 @xmath6 0.1 jy km s@xmath3 and 0.5 @xmath6 0.1 jy km s@xmath3 , respectively .
the signal to noise is too low to resolve the velocity structure . to 3.4 mjy beam@xmath3 .
( b ) cs ( 76 ) integrated intensity image of vv 114e . the flux in color scale ranges from 0 jy km s@xmath3 to 0.45 jy km s@xmath3 .
( c ) hcn ( 43 ) integrated intensity image of vv 114e . the flux in color scale ranges from 0 jy km s@xmath3 to 2.0 jy km s@xmath3 .
( d ) hco@xmath4 ( 43 ) integrated intensity image of vv 114e . the flux in color scale ranges from 0 jy km s@xmath3 to 2.8 jy km s@xmath3 .
the beam size of each line and continuum is shown in the bottom - left of the images ( table [ table_data ] ) .
the red cross shows the position of the eastern nucleus defined by the peak position of the ks - band observation @xcite . ] two radical cn rotational transitions _
= 1 0 ( _ j _ = 3/2 1/2 and 1/2 1/2 ) are detected at the eastern nucleus .
the @xmath47 = 3/2 1/2 transition is extended toward the overlap region ( figures [ fig_mom2]d , [ fig_mom2]e , [ fig_mom2]f , [ fig_mom2]g , [ fig_mom2]h , and [ fig_mom2]i ) .
we can not resolve their multiplet because of the coarse frequency resolution ( 11.5 mhz @xmath41 30 km s@xmath3 ) . because the critical density of cn is high ( @xmath21 10@xmath14 @xmath23 )
, the cn emission mainly comes from denser gas regions than regions traced by @xmath1co ( 10 ) .
= 3/2 1/2 transition shows a similar distribution to the @xmath2co ( 10 ) emission , but it is less extended over the overlap region .
the total cn ( 1@xmath270@xmath27 ) and cn ( 1@xmath280@xmath27 ) integrated intensities are 2.0 @xmath6 0.1 jy km s@xmath3 and 5.4 @xmath6 0.3 jy km s@xmath3 , respectively .
the highest velocity dispersion in the cn ( 1@xmath280@xmath27 ) image is also detected between the eastern nucleus and the overlap region , and this is likely caused by a superposition of clouds similar to the case of the @xmath2co ( 10 ) image ( see appendix [ a3 ] ) .
the @xmath1co ( 32 ) emission maps are presented in figure [ fig_mom3 ] . the estimated missing flux in our alma observation is 21 @xmath6 1 % ( _ james clerk maxwell telescope _ ( jcmt ) : 2956 @xmath6 133 @xmath48 and alma : 2343.7 @xmath6 4.7 @xmath48 ; * ? ? ?
although our @xmath1co ( 32 ) observation recovers more flux than the _ submillimeter array _ ( sma ) observation ( 1530 @xmath6 16 @xmath48 ; the missing flux = 48 @xmath6 15 % ; * ? ? ?
* ) , there are significant negative sidelobes at the north and south of the image which is likely the cause of missing flux .
we made the cleaned image with a uniform _ uv _ weighting to minimized the sidelobe level @xcite . the @xmath1co ( 32 ) integrated intensity map of vv 114 ( figure [ fig_mom3]a ) shows two arm - like structures and a filamentary structure similar to the @xmath1co ( 10 ) image , and the strongest peak is at @xmath41 55 west of the eastern nucleus .
the global gas distribution is consistent with the previous @xmath1co ( 32 ) observations @xcite .
the @xmath1co ( 32 ) velocity field map of vv 114 ( figure [ fig_mom3]b ) also shows significant broad velocity range across the galaxy disks ( @xmath41 600 km s@xmath3 ) , similar to the @xmath1co ( 10 ) velocity field map .
the se arm has a blue - shifted velocity from 5650 km s@xmath3 to 5920 km s@xmath3 , while the nw arm has a red - shifted velocity from 5950 km s@xmath3 to 6160 km s@xmath3 .
other two arm - like features are also detected .
one located @xmath41 40 northeast of the eastern nucleus shows an arc around the eastern nucleus and has red - shifted velocities from 5810 km s@xmath3 to 6180 km s@xmath3 .
this arm coincides with the ne arm detected in the @xmath1co ( 10 ) .
the other one located at @xmath41 100 west of the se arm has a strong peak ( @xmath41 262.5 @xmath6 0.9 jy km s@xmath3 ) and blue - shifted velocities from 5610 km s@xmath3 to 5900 km s@xmath3 .
this arm also coincide with the sw arm detected in the @xmath1co ( 10 ) . from the @xmath1co ( 32 ) velocity dispersion map of vv 114 ( figure [ fig_mom3]c )
, we find that the overlap region has the highest velocity dispersion ( @xmath41 110 km s@xmath3 ) .
the velocity dispersion of the nw arm is @xmath41 30 km s@xmath3 , while the se arm is @xmath41 60 km s@xmath3 .
the hcn ( 43 ) and hco@xmath4 ( 43 ) images are shown in figures [ fig_mom3]d , [ fig_mom3]e , [ fig_mom3]f , [ fig_mom3]g , [ fig_mom3]h , and [ fig_mom3]i . while the hcn ( 43 ) emission is only seen near the eastern nucleus of vv 114 and is resolved into four peaks , the hco@xmath4 ( 43 ) emission is more extended and has at least 10 peaks in the integrated intensity map .
the total integrated intensities of hco@xmath4 ( 43 ) and hcn ( 43 ) are 15.3 @xmath6 0.4 jy km s@xmath3 and 4.4 @xmath6 0.2 jy km s@xmath3 , respectively .
the higher hco@xmath4 ( 43 ) flux observed with the sma ( 17 @xmath6 2 mjy , * ? ? ? * ) using a 28 @xmath8 20 beam is likely attributed to missing flux by the alma observation .
a compact component in the eastern nucleus is unresolved with the current resolution , and the upper limit to the size is 200 pc .
the hcn ( 43 ) emission is not detected in the overlap region , where both the high @xmath1co ( 10 ) velocity dispersion and the significant ch@xmath0oh ( 2@xmath261@xmath26 ) and hco@xmath4 ( 43 ) detection suggest the presence of shocked gas @xcite .
we concluded in from their source size , line widths , and the relative strengths of hcn ( 43 ) and hco@xmath4 ( 43 ) that the unresolved eastern nucleus harbors an obscured agn , and the dense clumps in the western galaxy are related to extended starbursts .
the cs ( 76 ) emission has the highest critical density ( @xmath49 @xmath41 10@xmath9 @xmath23 ) of all of the lines detected in our observations .
the cs ( 76 ) emission is marginally ( s / n @xmath21 4 ) detected at the eastern nucleus ( figure [ fig_vv114e ] ) , and the total flux is 0.5 @xmath6 0.1 jy km s@xmath3 . the continuum image at 110 ghz shows a filamentary structure similar to the molecular line image ( figure [ fig_contin]a ) .
we construct low resolution ( 133 @xmath8 112 ) and high resolution ( 045 @xmath8 038 ) images of the 340 ghz continuum ( figures [ fig_contin]b , and [ fig_contin]c ) using the combined data ( compact + extended ) and the extended configuration data , respectively .
we find that the filamentary structure and the unresolved eastern nucleus are both present in dust continuum .
the total flux of the 110 ghz and the low resolution 340 ghz continuum emission are 10.3 @xmath6 0.2 mjy and 38.6 @xmath6 0.3 mjy , respectively .
the estimated missing flux relative to the jcmt 340 ghz observation @xcite is 75 @xmath6 4 % ( sma : 79 @xmath6 7 % ) .
the difference in the recovered flux between @xmath1co ( 32 ) and 340 ghz continuum emission may be caused by the difference in the distribution .
the 110 ghz and 340 ghz continuum emission is detected at the eastern nucleus ( s / n @xmath21 50 and 70 ) and the filamentary structure ( s / n @xmath21 8 and 24 ) identified in the @xmath2co ( 10 ) image , both with high significance .
we assign 39 r " boxes ( 20 @xmath8 20 ; r1 r39 ; see figure [ fig_ratio ] ) for the band 3 and @xmath1co ( 32 ) data and 15 smaller s " boxes ( 12 @xmath8 12 ; s0 s14 ; see figure [ figratio_hcn ] ) for the rest of the data to estimate the physical parameters , such as the molecular gas mass ( @xmath13 ) , dense gas mass ( @xmath50 ) , dust mass ( @xmath34 ) , star formation rate ( sfr ) , kinetic temperature ( @xmath51 ) , gas density ( @xmath52 ) , gas column density ( @xmath53 ) , and molecular abundance relative to h@xmath7 ( [ @xmath54/[h@xmath7 ] ) .
the positions of the boxes are chosen to cover the co ( 32 ) emission ( r1 r39 ) and the hco@xmath4 ( 43 ) emission ( s0 s14 ) .
the sizes of the boxes are chosen such that they are comparable to the beam size . before deriving the parameters and line ratios at each box
, we first matched the @xmath39 range between our data set and reconstructed the integrated intensity image of each line .
the shortest baseline lengths are set to 13.5 k@xmath56 and 40.0 k@xmath56 for the molecular lines in the band 3 and the band 7 , respectively , and the images are convolved into the same resolution ( 20 @xmath8 15 with a p.a . of 83 deg , 12 @xmath8 10 with a p.a
. of 119 deg ) . for each ratio ,
the two integrated intensity images were expressed in the units of k km s@xmath3 before calculating the ratio at locations where both lines are detected above 3 @xmath57 .
the derived box - summed spectra are listed in appendix [ a3 ] .
we carried out a multi gaussian fit ( one - three components ) to reproduce the box - summed spectra , and labeled the components as a " , b " , and c " from the bluest peak ( e.g. , the bluest peak at r21 is labeled as r21a ) . the @xmath1co ( 32)/@xmath1co ( 10 ) ratio , @xmath58 , can be used as an indicator of the dense / warm gas content relative to the total molecular gas .
the @xmath58 of vv 114 varies from 0.2 to 0.8 , as shown in figure [ fig_ratio ] ( left ) and table [ table_t_r ] .
this range is larger than the same ratios derived for normal spirals , which is typically 0.15 0.5 when observed with a similar linear resolution @xcite . at the edge and the center of the filament
, @xmath58 is higher ( 0.53 0.69 ) than the highest peaks of each arm ( @xmath21 0.4 ) .
this suggests that the co emitting gas at the filament have higher excitation conditions than normal spirals , while the conditions of each arm of vv 114 are consistent with arms and nuclei of normal spirals .
the @xmath58 at the eastern nucleus is 0.76 @xmath6 0.01 .
it is suggested that the @xmath58 is much higher ( 3.12 @xmath6 0.03 in ngc 1068 ; * ? ? ?
* ) for gas surrounding an agn , and the low @xmath58 in vv 114 may be due to the difference in filling factor ( 160 @xmath8 140 pc beam averaging for ngc 1068 , while 800 pc box averaging for vv 114 ) .
it is possible , however , that the nuclear excitation conditions are different from source to source .
in general , the @xmath1co lines has higher optical depths than the @xmath2co ( 10 ) line .
therefore , the measured @xmath1co ( 10)/@xmath2co ( 10 ) line intensity ratio , @xmath59 , gives a lower limit to the co/@xmath2co abundance ratio ( hereafter [ co]/[@xmath2co ] ) .
we present the @xmath59 image of vv 114 in figure [ fig_ratio ] ( center ) .
the @xmath59 increases from the arms ( @xmath5 17 ) to the filament ( 15 32 ) .
observationally , @xmath59 increases towards the central region of galaxies @xcite , where the gas is generally warmer and denser .
@xcite suggest that the moderate optical depth of @xmath1co ( 10 ) emission and/or the high [ co]/[@xmath2co ] environment can increase the @xmath59 in nuclei of u / lirgs . in order to understand which of the two ( optical depths or abundances ) is dominant , we calculated and mapped the optical depth of the @xmath1co ( 10 ) and the @xmath2co ( 10 ) as shown in table [ table_lte ] and figure [ fig_radex ] .
we provide an interpretation of these results in [ chi ] . in , the hcn ( 43 ) and hco@xmath4 ( 43 ) maps of vv 114 allowed us to investigate the central region at 200 pc resolution for the first time , and we find that both the hcn ( 43 ) and hco@xmath4 ( 43 ) in the eastern nucleus are compact ( @xmath5 200 pc ) , and broad [ 290 km s@xmath3 for hcn ( 43 ) ] .
we present the hcn ( 43)/hco@xmath4 ( 43 ) , @xmath60 , image of vv 114 in figure [ figratio_hcn ] . from the higher @xmath60 along with the past x - ray and nir observations
, we suggest the presence of an obscured agn in the eastern nucleus .
we also detect a 3 4 kpc long filament of dense gas , which is likely to be tracing the active star formation triggered by the ongoing merger , and this is consistent with the results from the numerical model by @xcite who predict that the fragmentation and turbulent motion of dense gas across the merging disk is responsible for forming dense gas clumps with masses of 10@xmath14 10@xmath15 m@xmath10 .
we present the @xmath60 image in figure [ figratio_hcn ]
. the overlap region does not show significant hcn ( 43 ) emission , and we provide the 3 @xmath57 upper limit in table [ table_t_s ] .
three out of the four boxes ( i.e. , s1
s3 ) in the eastern nucleus have low @xmath60 ( @xmath5 0.5 ) whereas s0 has a high @xmath60 ( 1.34 @xmath6 0.09 ) .
it is suggested that such a high value is only produced around agn environments ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
in this section , we derive the molecular gas mass ( [ x ] ) , and the physical parameters using the radiative transfer code radex ( [ chi ] ) for each box defined in [ ratio ] .
the column density is derived using the optically thin @xmath2co line under the lte assumption .
we estimate the beam filling factor @xmath61 and the relative molecular abundance of molecule @xmath62 ( hereafter expressed as [ @xmath54/[h@xmath7 ] ) ( [ lte ] ) .
finally , we calculate the dust mass using the 340 ghz continuum emission ( [ dust ] ) .
the molecular gas mass m@xmath63 is derived by ; @xmath64,\ ] ] where @xmath65 is the molecular line luminosity - to - h@xmath7 mass conversion factor and @xmath66 is the velocity integrated flux .
we use the conversion factor known to be appropriate for u / lirgs .
this is consistent with the value derived by @xcite in vv 114 ( @xmath11 = @xmath67 m@xmath68 ) .
the molecular gas mass derived at the boxes defined in [ ratio ] ranges between 0.2 @xmath8 10@xmath15 @xmath69 and 4.8 @xmath8 10@xmath15 @xmath69 m@xmath10 ( table [ table_sf_co ] ) .
we also calculate the dense gas mass @xmath50 using @xmath70 = 10 m@xmath68 and the hcn ( 43 ) luminosity which is converted to the hcn ( 10 ) luminosity using hcn ( 43)/hcn ( 10 ) = 0.63 (; * ? ? ?
the dense gas mass ranges between 1.8 @xmath8 10@xmath14 @xmath71 and 3.8 @xmath8 10@xmath9 @xmath71 m@xmath10 ( table [ table_sf_hcn ] ) .
we note that the co luminosity - to - h@xmath7 mass conversion factor , @xmath11 , is very uncertain , and varies significantly from source to source
. it may be possible that @xmath11 varies from region to region within a galaxy . while one would ideally adopt a spatially varying @xmath11 for a better quantification of the h@xmath7 mass , such a study is beyond the scope of this present paper . for simplicity
, here we adopt a constant @xmath11 across all regions in vv 114 , bearing in mind that the uncertainties could be as large as a factor of two .
the same applies to @xmath70 .
we used the non - lte radiative transfer code radex @xcite and varied the parameters until the residuals between the observed line fluxes and the modeled line fluxes are minimized in a @xmath73 sense .
we assumed a uniform spherical geometry ( @xmath74 = 1.0 @xmath75 ) , and derived the physical conditions of molecular gas ( @xmath51 , @xmath52 , and @xmath76 ) .
radex uses an escape probability approximation to solve the non - lte excitation assuming that all lines are from the same region . since the molecular lines in the band 7 have significantly higher critical densities than that in the band 3 , we used two sets of molecular lines ; ( case 1 ) 20 box - summed @xmath1co ( 10 ) , @xmath2co ( 10 ) , and @xmath1co ( 32 ) , and ( case 2 ) 12 box - summed hcn ( 43 ) , hco@xmath4 ( 43 ) , @xmath1co ( 32 ) , and @xmath1co ( 10 ) , to solve for the degeneracy of the physical parameters . in case 2
, we made the _
uv _ and beam - matched hcn ( 43 ) , hco@xmath4 ( 43 ) , and @xmath1co ( 32 ) images ( 12 @xmath8 10 resolution with the p.a .
= 119 deg . ) , and we defined three hco@xmath4 ( 43 ) peaks as e0 , e1 , and e2 ( figure [ fig_spec_radex ] ) .
we also use the @xmath39 and beam - matched @xmath1co ( 10 ) data to constrain the @xmath76 , allowing us to vary the [ hcn]/[hco@xmath4 ] in case 2 .
all line parameters , such as the upper state energies and the einstein coefficients , were taken from the _ leiden atomic and molecular database _ ( lamda ; * ? ? ?
in order to find the set of physical parameters that can reproduce the observed line intensities , we run radex by varying @xmath51 , @xmath52 , and @xmath53 for case 1 , and @xmath51 , @xmath52 , and [ hcn]/[hco@xmath4 ] for case 2 . the adopted @xmath53 are 10@xmath78 , 10@xmath79 , and 10@xmath80 cm@xmath19 , at e0 , e1 , and e2 , respectively .
we varied the gas kinetic temperature within a range of @xmath51 = 5 300 k using steps of d@xmath51 = 5 k , and a gas density of @xmath52 = @xmath82 @xmath83 using steps of d@xmath52 = @xmath84 . for case 1 , we fixed [ @xmath2co]/[h@xmath7 ] = 1.4 @xmath8 10@xmath85 @xcite and [ co]/[@xmath86 = 70 , which are the galactic values . in case 2 , we changed the parameters , @xmath51 = 5 400 k using steps of d@xmath51 = 5 k , @xmath52 = @xmath87 @xmath88 using steps of d@xmath52 = @xmath84 , and fixed [ co]/[h@xmath7 ] = 1.0 @xmath8 10@xmath89 and [ hco@xmath4]/[h@xmath7 ] = 1.0 @xmath8 10@xmath90 , which are the standard values observed in galactic molecular clouds @xcite .
we varied [ hcn]/[hco@xmath4 ] from 1 10 , in steps of one .
the parameters we used are summarized in table [ table_radex_parm ] .
we list the results that are within the 95 % confidence level with 3-degree of freedom ( @xmath91 ) ( tables [ table_radex ] and [ table_radex_hcn ] ) . finally , we created velocity - averaged channel maps of @xmath52 and the optical depth of the transitions ( figure [ fig_radex ] ) .
we note that the uncertainty of the @xmath53 for case 2 did not strongly affect the results , while that of the [ co]/[@xmath86 for case 1 changed .
the effect of varying the [ co]/[@xmath86 will be discussed in [ 1 ] .
future multi - transition hcn / hco@xmath4/co/@xmath2co imaging will help us to derive these parameters directly .
the ( box - averaged ) kinetic temperature near the eastern nucleus ( r21a ) is constrained to within 25 90 k ( the best fit is 50 k ) , as shown in table [ table_radex ] . the @xmath92 ( 58.8 @xmath6 2.9 k ) obtained from the lte assumption at r21a ( see [ lte ] ) is higher than the best - fitted @xmath51 .
in fact , we also find five regions ( r10b , r11b , r14 , r16 , and r25a ) that show similarly high excitation temperatures .
four out of five regions are in the central filament . in general ,
spontaneous emission dominates over collisional excitation in sub - thermally excited conditions , and hence @xmath92 should be lower than @xmath51 .
one reason for this discrepancy could be attributed to the incorrect assumption of [ co]/[@xmath2co ] . by varying this abundance ratio ,
we find that the temperature reversal ( i.e. @xmath93 ) occurs only when [ co]/[@xmath2co ] @xmath94 150 .
this is consistent with the results obtained by @xcite who used radex along with their multi co and @xmath2co line data to find evidence of a cold / dense molecular gas component with extremely high [ co]/[@xmath2co ] of 229 , which is 3 times higher than that of the galactic value .
the derived @xmath51 at the other regions are generally higher than 100 k. the derived @xmath92 in each region are typically 10 40 k , which may suggest sub - thermal conditions .
the kinetic temperatures derived at the se and nw arms are estimated to be @xmath5 90 k , with higher temperature at the nw arm .
the nw arm is also associated with relatively strong pa@xmath35 emission and @xmath40s - band emission , which is consistent with the higher relative temperature due to star - forming activities @xcite . however ,
this is inconsistent with the general understanding that strong tidal shear in tidal arms prevents active star formation to occur @xcite .
the derived @xmath52 in most of the boxes are less than 10@xmath95 @xmath23 , which is consistent with the critical densities of the low-@xmath47 co lines observed here .
the highest density of 10@xmath96
10@xmath97 @xmath23 is estimated at r21a , and this is consistent with the location of the eastern nucleus . since we also observed the strongest hcn ( 43 ) and hco@xmath4 ( 43 ) emission at r21a at the same line - of - sight velocity (
* see also appendix [ a3 ] ) , it is possible that the main contribution to the co emission at r21a arises from dense gas ( 10@xmath96 10@xmath97 @xmath23 ) near the eastern nucleus , with a minor contribution from the diffuse gas clouds along the same line of sight observed within the same beam .
in contrast to the eastern nucleus , the boxes that cover the western galaxy ( r1 r11 and r26 r29 ) show moderately dense condition of 10@xmath98 10@xmath99 @xmath23 .
this extended and moderately dense gas is associated with the disk - like structure seen in optical images @xcite , and the star formation traced in pa@xmath35 emission and uv / x - ray emission @xcite .
we note that the strongest off - nuclear pa@xmath35 peak ( r27 in table [ table_sf_co ] ; sfr = 3.15 @xmath6 0.05 m@xmath10 yr@xmath3 ) coincides with relatively low gas density ( @xmath21 10@xmath95 @xmath23 ) .
the density of the surrounding region labeled r25a is similar ( 10@xmath100 10@xmath97 @xmath23 ) and this is comparable to the nucleus of the eastern galaxy .
the secondary pa@xmath35 peak ( r29 ; sfr = 0.92 @xmath6 0.05 m@xmath10 yr@xmath3 ) is not associated with any molecular line emission . it is usually believed that the @xmath1co ( 10 ) emission is optically thick ( @xmath101 ) , while the @xmath2co ( 10 ) emission is optically thin ( @xmath102 ) even in luminous mergers @xcite . in most regions , we find that the optical depth of the @xmath1co ( 10 ) line is @xmath103 1 ( figure [ fig_radex ] ) .
in contrast , the @xmath1co ( 10 ) opacity at the eastern nucleus and the filament is moderately optically thick ( @xmath104 @xmath21 1 ) .
however , the elevated @xmath59 at the eastern nucleus ( see [ ratio ] ) can not be explained by the relatively low @xmath1co ( 10 ) opacity alone ( the opacity has to be @xmath104 @xmath105 0.1 ; see also @xcite ) .
finally , we find that indeed the @xmath2co ( 10 ) emission is optically thin ( @xmath102 ) averaged over the whole galaxy , except for the southern dust lane ( @xmath106 = 0.3 1.5 ) . from these results
, we suggest that the peak of the molecular gas in the central 800 pc of the eastern galaxy is cold ( @xmath51 = 25 90
k ) , dense ( @xmath52 = 10@xmath96
10@xmath97 @xmath23 ) , and moderately optically thick ( @xmath107 @xmath21 3 ) , while peaks in the overlap region are warm ( @xmath51 @xmath94 50 k , best - fitted @xmath51 is 95 and 175 k at r39a and r39b , respectively ) , moderately dense ( @xmath52 = 10@xmath108
10@xmath109 @xmath23 ) , and moderately optically thick ( @xmath107 @xmath21 1 ) . the derived density of the eastern galaxy is slightly higher than the range of values found in u / lirgs using low-@xmath47 co emission with@xmath21 kpc resolution .
in addition , the low opacities predicted from these analyses are consistent with earlier results that investigate the opacities in m82 ( @xmath110 = 0.5 4.5 ; * ? ? ? * ) and u / lirgs , and the central region of ngc 6240 ( @xmath110 = 0.2 2 ; * ? ? ? * ) .
however , the derived temperature of the eastern galaxy is inconsistent with the high values found in nearby starburst galaxies m82 , ngc 253 , and ngc 6240 .
the disagreement is possibly due to the uncertainties in the [ co]/[@xmath2co ] , or the difference in the observed molecular gas tracers .
the values for @xmath51 , @xmath52 , and the optical depth of hcn ( 43 ) and hco@xmath4 ( 43 ) are shown in table [ table_e ] .
the derived parameters for the unresolved component e0 , are @xmath51 @xmath94 100 k , @xmath52 = 10@xmath97
10@xmath111 @xmath23 , and [ hcn]/[hco@xmath4 ] @xmath94 5 .
the lower limit to the kinetic temperature is higher than those of e1 and e2 , mainly due to the unusually high @xmath60 and @xmath112 .
in contrast to e0 , the derived parameters near e1 show high h@xmath7 densities ( @xmath52 = 10@xmath113 10@xmath114 @xmath23 ) .
the overlap region ( e2 ) , where the star - formation rate ( 1.70 @xmath6 0.05 m@xmath10 yr@xmath3 ) is lower than the eastern nucleus , has densities in the range of @xmath52 = 10@xmath97
10@xmath113 @xmath23 .
finally , the optical depths for the hco@xmath4 ( 43 ) and hcn ( 43 ) lines are calculated for each gas clump , yielding @xmath115 @xmath41 0.7 and @xmath116 @xmath41 0.2 for e0 , @xmath115 @xmath41 0.2 and @xmath116 @xmath41 0.6 for e1 , and @xmath115 @xmath41 0.4 and @xmath116 @xmath41 0.4 for e2 . the higher linear resolution observations of hcn ( 43 ) and hco@xmath4 ( 43 ) toward ngc 1097 @xcite revealed that the gas in the central region of ngc 1097 has @xmath51 = 70 550 k and @xmath52 = 10@xmath117
10@xmath118 @xmath23 .
moreover , by comparing to lvg models , @xcite found that hcn and hco@xmath4 emission in agn - dominated sources appears to emerge from regions with lower h@xmath7 densities , higher temperatures , and higher hcn abundance relative to starburst - dominated ( sb - dominated ) galaxies .
our results obtained toward vv 114 are consistent with these previous results . in order to determine the bulk properties of the co emitting gas
, we used an excitation temperature analysis @xcite .
the excitation temperature at each box can be calculated from @xmath119 \right)^{-1}\ ] ] where @xmath120 [ = 5.53 k for @xmath1co ( 10 ) emission ] , @xmath121 is the frequency of the transition , @xmath122 is the planck s constant , @xmath123 is the boltzmann s constant , @xmath124 is the brightness temperature of @xmath1co ( 10 ) emission in kelvin , @xmath104 is the optical depth of the @xmath1co ( 10 ) emission , and @xmath125 is the cosmic microwave background temperature ( 2.73 k ) . using @xmath51
estimated from the radex calculation ( [ chi ] ) , we estimate the beam filling factor @xmath61 , @xmath126 the optical depth of the @xmath1co ( 10 ) emission is also estimated from the radex calculation in [ chi ] . assuming that the @xmath2co and co arise from the same molecular cloud , and that the @xmath1co ( 10 ) is optically thick
, we estimate the optical depth of a given molecule using , @xmath127\ ] ] where @xmath128 is the optical depth of a given transition , and @xmath129 is the observed brightness temperature for transition x. using @xmath92 and @xmath128 , we estimate the column density for a given molecule from , @xmath130 @xmath131 @xmath132 where @xmath133 is the dipole moment , @xmath134 is the rotational constant , @xmath47 is the lower energy level , and @xmath135 is the integrated intensity @xcite .
the derived column densities are listed in tables [ table_lte ] and [ table_chem ] .
we calculated the dust mass from the 340 ghz ( 880 @xmath133 m ) continuum emission ( table [ table_contin ] ) using @xcite , @xmath136 where @xmath137 is the 340 ghz flux in jy and @xmath138 is the luminosity distance in mpc .
we assumed a dust emissivity , @xmath139 , and the dust temperature @xmath140 of 39.4 k @xcite .
the box - summed dust masses ranges between 2.0 @xmath8 10@xmath17 @xmath141 and 2.8 @xmath8 10@xmath14 @xmath141 m@xmath10 ( table [ table_sf_co ] ) .
we note that we used the dust model for @xmath142 , because the @xmath142 derived from observations has a large error @xcite .
@xcite suggested that the submillimeter continuum emission traces the total ism mass ( @xmath143 ) , since the long wavelength rayleigh - jeans ( rj ) tail of thermal dust emission is often optically thin . in order to compare the @xmath143 with the @xmath13 ( see $ [ x ] ) using spatially - resolved data , we calculated the total ism mass from the 340 ghz continuum emission @xcite . for @xmath44
@xmath144 1199 ghz , @xmath145 @xmath146 where @xmath147 is the observed flux , @xmath143 is the ism mass , @xmath148 is the observed frequency , and @xmath149 and @xmath150 are given by @xmath151 @xmath152 the derived box - summed ism masses of vv 114 range between 5.2 @xmath8 10@xmath9 and 7.2 @xmath8 10@xmath15 m@xmath10 ( table [ table_sf_co ] ) .
this is comparable to the box - summed h@xmath7 masses ( @xmath13 = ( 0.2 4.7 ) @xmath8 10@xmath15 @xmath69 m@xmath10 ) .
we find that the @xmath143/@xmath13 ratio is close to unity ( 0.5 2.0 , the average @xmath143/@xmath13 = 0.9 @xmath6 0.1 ) , while the total @xmath143/@xmath13 ratio is 0.6 @xmath6 0.1 .
this means that the spatially - resolved @xmath143 is a good tracer of the resolved " h@xmath7 mass .
however , the total @xmath143 underestimates the h@xmath7 mass ( even using the @xmath11 for ulirgs to derive the @xmath13 ) because the global distribution of the 340 ghz continuum emission is significantly different from that of the co ( 10 ) emission ( figures [ fig_mom1 ] and [ fig_contin ] ) .
this difference between the 340 ghz continuum and the co ( 10 ) is also seen in recent observations of nearby lirgs ( e.g. , * ? ? ?
our radex modeling yields lower molecular gas density near the agn ( @xmath52 = 10@xmath97
10@xmath111 @xmath23 ) compared to the surrounding clumps ( 10@xmath113 10@xmath114 @xmath23 ) .
similarly high values are obtained near agns in other galaxies @xcite .
@xcite suggest that the gas densities in agn host galaxies ( @xmath5 10@xmath117 @xmath23 ) are lower than starburst host galaxies ( 10@xmath97 10@xmath153 @xmath23 ) , and a common interpretation relies on a clumpy ism near star - forming regions ( which reduces the filling factor ) and a continuous ism near the agn . since our current observations ( @xmath21 200 pc resolution ) cover a significantly large area and the beam filling factor may be small ( @xmath61 at e0 , e1 , and e2 are @xmath144 0.03 , 0.04 0.06 and 0.01 0.04 , respectively ) , higher resolution observations ( @xmath5 05 ) are required to confirm this scenario .
in addition , our modeling shows higher [ hcn]/[hco@xmath4 ] near the eastern nucleus ( @xmath94 5 ) than that in the surrounding clumps ( @xmath5 4 ) and the overlap region ( 1 - 9 ) .
the elevated [ hcn]/[hco@xmath4 ] is explained by two mechanisms @xcite .
one is far - uv radiation from ob stars in young starbursts , and the other is strong x - ray radiation from an agn @xcite . because of
different penetrating lengths between far - uv and x - ray emission , photon dominated regions ( pdrs ) are created at the surface of gas clouds and x - rays penetrate deeply into the circumnuclear disk ( cnd ) , forming large x - ray dominated regions . as a consequence of this volume versus surface effect , the x - ray radiation from an agn may produce higher hcn abundances than the uv radiation of starburst activities @xcite . to some degree
, ionization effects from cosmic rays @xcite such as supernovae or strong shocks are suspected to significantly increase the hco@xmath4 abundance while potentially decreasing the hcn abundance , thus yielding lower @xmath60 in evolved starbursts than in agns .
the high [ hcn]/[hco@xmath4 ] near the eastern nucleus and low [ hcn]/[hco@xmath4 ] and strong / extended 8 ghz continuum detection at the surrounding clumps @xcite are all consistent with a presence of an agn in the eastern nucleus , surrounded by star - forming dense clumps .
observational studies of galaxies at global scales have shown that the surface density of sfr and that of cold gas traced in co ( 10 ) obey a power law relation ( ks law ; * ? ? ?
* ; * ? ? ?
ulirgs are systemically shifted from the normal galaxy population in the @xmath20
@xmath16 phase @xcite .
it is suggested that systems lower in ir luminosity ( e.g. , lirgs ) occupy the region between the starburst " sequence and the normal disk " sequence in the ks law .
galaxies in the starburst " sequence have shorter gas depletion time ( @xmath154 = @xmath16/@xmath20 @xmath21 0.1 gyr ) relative to galaxies in the normal disk " sequence ( @xmath154 @xmath21 1 gyr ; * ? ? ?
* ; * ? ? ?
the spatially resolved surface densities of the sfr and the molecular gas mass of vv 114 are shown in table [ table_sf_co ] and figure [ fig_ks ] .
the star - forming regions of vv 114 fill the gap between the normal disk " and starburst " sequences ( figure [ fig_ks]a ) .
we also show the spatial distribution of @xmath154 in figure [ fig_ks]b .
the data points close to the starburst " sequence are located along the eastern nucleus ( @xmath5 0.2 @xmath69 gyr ) and the overlap region (= 0.2 0.4 @xmath69 gyr ) , while those near the
normal disk " sequence are located in the nw and se arms ( @xmath94 0.8 @xmath69 gyr ) .
the spatial distribution of @xmath20 and @xmath16 are consistent with the distributions of previous optical , uv , and x - ray studies @xcite . regions with higher @xmath20 and @xmath16 clearly show higher @xmath58 and @xmath59 ( figures [ fig_ks]c and [ fig_ks]d ) .
in summary , transition from the normal disk " to starburst " sequence may occur when the molecular clouds become excited and dense at the nuclei and the overlap region . moreover , gas clouds with high @xmath58 have high @xmath20
@xmath16 , and this is consistent with past studies which suggest that the @xmath58 correlates with the local h@xmath35 flux @xcite .
the @xmath59 also shows a similar trend , and this is also consistent with the past studies ( @xmath94 20 in central kpc regions of u / lirgs , 10 15 in normal starburst galaxies , and @xmath21 5 in galactic gmcs ; * ? ? ?
* ) : the reason for the elevated @xmath59 in starburst regions of vv 114 will be discussed in detail in $ [ isotope ] .
we suggest from our radex modelings that the eastern nucleus and the overlap region have extremely high [ co]/[@xmath2co ] ( @xmath94 200 ) , which is at least two times higher than the galactic value .
the pa@xmath35 peaks roughly coincide with the regions where high [ co]/[@xmath2co ] are expected , suggesting that the increased [ co]/[@xmath2co ] is related to the star formation activity .
similarly high values are seen in the overlap region of ngc4038/9 @xcite and the taffy @xcite .
@xcite suggested that the extreme [ co]/[@xmath2co ] value in the bridge is explained by three scenarios , 1 ) selective isotope photodissociation in the diffuse clouds and shocked region , 2 ) co enrichment around starburst activities , and/or 3 ) the destruction and recombination of molecules after shock .
we briefly explain each scenario below , but our current data is insufficient for us to identify the exact cause of the high [ co]/[@xmath2co ] in vv 114 .
the first possibility of [ co]/[@xmath2co ] enhancement is the deficiency in @xmath2co .
@xcite suggest that selective isotope photodissociation can reduce the @xmath2co abundance in diffuse clouds , because co is self - shielded to a greater extent .
thus , the ism surrounding young starbursts and/or shocked regions show elevated [ co]/[@xmath2co ] @xcite .
the ism in the nuclei and the overlap region of vv 114 show extremely high [ co]/[@xmath2co ] , presumably due to intense starburst activities and/or large - scale shocks .
the second possibility is that massive stars end their life as supernovae and expel a large amount of @xmath1c in the interstellar medium . while the elemental abundances ( e.g. c and s ) are not directly related to the molecular abundances ( e.g. , cs ; * ? ? ?
* ) , once the synthesized elements are dispersed in the interstellar medium , molecules ( e.g. , co , cs , and cn ; * ? ? ?
* ) form as soon as the temperature and density conditions are favorable .
this occurs with a timescale of a few 10@xmath18 yr . for the overlap region ,
the destruction and recombination of molecules after shocks ( see [ chemical ] ) are possible mechanisms to enhance the [ co]/[@xmath2co ] ( the third possibility ) .
the recombination timescale of h@xmath7 and co molecules after shock destruction are shorter than that of @xmath2co , since ionized photons from shocked regions lead to selective isotope photodissociation @xcite . shielded regions from the radiation field are needed to form rare @xmath2co ( abundant co can form self - shielded regions ) . moreover
, the rare isotope molecules generally need a longer time to form , because collisions between molecules and dust grains are less frequent @xcite .
the gas - to - dust ratio , @xmath13/@xmath34 , provides an important measure of the relative abundance between gas and metallicity .
the average @xmath13/@xmath34 over the entire galaxy is often derived in single - dish work , and typical @xmath13/@xmath34 is 200 300 for local u / lirgs , and 15 231 in high - z sources .
@xcite found @xmath13/@xmath34 = 357 @xmath6 95 from a sample of 13 u / lirgs , including vv 114 , observed at kpc resolution .
we use the gas and dust masses derived in [ dust ] to investigate the distribution of @xmath13/@xmath34 ( figure [ plot_g / d ] ) . the smallest value of ( 128 @xmath6 16 )
@xmath156 occurs in the eastern nucleus , which is similar to the galactic value ( 100 ; * ? ? ?
* ) , while higher values of ( 371 @xmath6 118 ) @xmath156 and ( 339 @xmath6 60 ) @xmath156 occur in the western nucleus and the overlap region , respectively .
the clear differences between the two nuclei may suggest a local gradient in the metallicity .
for the overlap region , cold dust associated with diffuse gas clouds can not avoid the collision .
this tends to increase the @xmath13/@xmath34 , because shocks destruct dust particles preferentially @xcite . on the other hand
, the low @xmath13/@xmath34 in the eastern nucleus may be due to intense starbursts producing dust - rich environments .
table [ table_chem ] shows the properties of the detected molecular lines which are not used in the radex calculations .
either the dense gas component of vv 114 has extreme variations in excitation among the molecular clumps in the filament ( see [ dense ] ) , or there is widespread chemical differentiation across the filament .
the fractional abundances [ @xmath157 of the different astrochemical species provide evidence of varying chemical influences due to star formation , physical conditions , and dynamics across the galaxy disks .
we use the h@xmath7 column densities , derived from the radex calculations , which are 10@xmath158 , 10@xmath159 , and 10@xmath159 cm@xmath19 at r18 ( agn ) , r21a ( starburst ) , and r39a ( overlap region ) , respectively .
column densities of each molecules are determined by equation ( 7 ) assuming an optically thin emission under lte .
the @xmath92 values determined from equation ( 3 ) are 38.7 @xmath6 1.9 k , 58.8 @xmath6 2.9 k , and 52.6 @xmath6 2.6 k at r18 , r21a , and r39a , respectively . the derived [ @xmath157 are listed in table [ table_chem ] .
( [ @xmath54/[h@xmath7 ] ) of selected extragalactic sources compared to those of specific regions of vv 114 , as presented in table [ table_chem ] .
the red , green , and blue columns show r18 , r21a , and r39a , respectively .
the pink , light blue , and yellow columns show ngc 253 , m 82 , and ic 342 , respectively .
arrows represent upper limits . ] in figure [ plot_molabu ] , we show the fractional abundances for cs , ch@xmath0oh , and cn in vv 114 , and the same ratios for a sample of nearly galaxies , ngc 253 , m82 , and ic 342 , taken from line surveys available in the literature @xcite .
m82 has a relatively old starburst at its core , with an average stellar population age of @xmath41 10 myr @xcite .
this creates strong uv fields , therefore the pdr dominates its chemistry @xcite .
figure [ plot_molabu ] shows that r39a has higher ch@xmath0oh abundance than m82 , and small cs and cn abundances .
a pure pdr similar to m82 may explain the molecular abundances we observe in r18 .
the molecular abundances for the overlap region and ngc 253 share similar characteristics .
ngc 253 is thought to be in an early stage of starburst evolution , and has young stellar populations in its nucleus ( @xmath41 6 myr ; * ? ? ?
the chemistry in the nucleus of ngc 253 is dominated by large - scale shocks @xcite , and we suggest that the overlap region of vv 114 is also dominated by shocks .
the low @xmath60 at the overlap region are further evidences for a shock dominated region @xcite .
tidal dwarf galaxies ( tdgs ) are gas - rich irregular galaxies made out of stellar and gaseous material pulled out by tidal forces from the disks of the colliding parent galaxies into the intergalactic medium .
they are found at the ends of long tails and host active star - forming regions @xcite .
@xcite and @xcite found the hi gas mass of 4.1 @xmath8 10@xmath15 m@xmath10 and the molecular gas mass of 4 @xmath8 10@xmath14 m@xmath10 at the edge of the southern tail of ngc 4038/9 .
we found an elevated @xmath58 ( 0.36 @xmath6 0.01 ) , sfr ( 0.10 @xmath6 0.05 m@xmath10 yr@xmath3 ) , and @xmath13 ( @xmath21 3.8 @xmath8 10@xmath9 @xmath69 m@xmath10 ) at the edge of the southern tidal arm ( r38 ) .
the derived sfr and @xmath13 of r38 are comparable to those of tdg candidates in other galaxies @xcite .
the gas depletion time of ( 0.40 @xmath6 0.22 ) @xmath69 gyr is shorter than the rest of the gas in the tidal arm ( @xmath94 0.5 @xmath69 gyr ) . according to the radex modeling ,
while the ranges of @xmath51 and @xmath52 are not confined well , the best fitting values ( 35 k , 10@xmath160 @xmath23 ) are slightly higher than those in the middle of the tidal arm , r36a and r37a ( 25 30 k , 10@xmath98 10@xmath161 @xmath23 ) .
we suggest that r38 is a forming tidal dwarf galaxy at the edge of the tidal arm of vv 114 .
future high sensitivity optical and high resolution hi observations will allow us to constrain the star formation and the atomic gas properties of r38 .
we investigate the physical conditions of the molecular gas in the mid - stage merger vv 114 .
we present high - resolution observations of molecular gas and dust continuum emission in this galaxy using alma band 3 and band 7 .
this study includes the first detection of extranuclear ch@xmath0oh ( 21 ) emission in interacting galaxies .
the results can be summarized as follows : 1 .
we find that the co ( 10 ) and co ( 32 ) lines show significantly extended structures ( i.e. , the northern and southern tidal arms ) , the central filament across the galaxy disks , and double - peaks in the overlap region , while the @xmath2co ( 10 ) line is only detected at the central filament .
the filament is also identified by the strong cn ( 1@xmath28 0@xmath27 ) , hco@xmath4 ( 43 ) , 110 ghz , and 340 ghz continuum emission .
2 . higher @xmath162 ( 0.5 0.8 ) and @xmath163 ( 20 50 ) are detected at the central filament .
these higher ratios indicate that the central filament has highly excited ( but not thermalized ) molecular ism , and the eastern nucleus is nearly thermalized when it is observed with a 800 pc beam .
the unresolved eastern nucleus has the highest @xmath60 ( 1.34 @xmath6 0.09 ) , while the dense gas clumps near the eastern nucleus have significantly lower values ( @xmath21 0.5 ) . the broad hcn ( 43 ) and hco@xmath4 ( 43 ) ( @xmath21 290 km
s@xmath3 ) emission lines seen in the unresolved eastern nucleus suggests an obscured agn ( see also ) .
radiative transfer analysis of the co ( 10 ) , co ( 32 ) , and @xmath2co ( 10 ) emission enables us to map physical parameters of the diffuse " gas of an interacting lirg with 800 pc scale for the first time .
the analysis suggests that diffuse " gas clouds in the filament have warmer / denser conditions than those in the galaxy disks .
this is consistent with predictions from merger simulations .
our analysis also suggest that the [ co]/[@xmath2co ] is enhanced in the central filament .
the extremely high [ co]/[@xmath2co ] values are more important than the moderately optically thick @xmath1co ( 10 ) emission to explain the high @xmath59 in vv 114 . 5 .
radiative transfer analysis of the hcn ( 43 ) , hco@xmath4 ( 43 ) , and @xmath1co ( 32 ) allow us to compare the dense gas clouds around agn , starburst activities , and the overlap region .
these results show that dense gas clouds around agn have @xmath52 = 10@xmath97
10@xmath111 @xmath23 and @xmath51 @xmath94 100 k with [ hcn]/[hco@xmath4 ] @xmath94 5 , while gas clumps around starburst activities show @xmath52 = 10@xmath113 10@xmath114 @xmath23 and @xmath51 = 40 100 k with [ hcn]/[hco@xmath4 ] @xmath5 4 .
in addition , the analysis shows that the overlap region has @xmath52 = 10@xmath97
10@xmath113 @xmath23 and @xmath51 = 5 90 k with [ hcn]/[hco@xmath4 ] = 1 9
the spatially resolved kennicutt - schmidt law in vv 114 clearly connects the starburst " sequence with the normal disk " sequence .
most of the data points near the starburst " sequence are found in the nuclei and the overlap region , whereas the data points near the normal disk " sequence are found in the tidal arms .
we also find the @xmath162 and @xmath163 are well correlated with the @xmath20 . 7 .
the @xmath13/@xmath34 of ( 128 @xmath6 16 ) @xmath156 in the eastern nucleus of vv 114 is comparable to the galactic value , but it is a factor of two higher than that in the overlap region of ( 339 @xmath6 60 ) @xmath156 .
since the 340 ghz emission is spatially correlated with dense gas tracers , the cold dust in vv 114 appears to be closely related to the dense molecular component in the filament .
the lowest @xmath13/@xmath34 in the eastern nucleus may be due to the dusty starburst .
8 . comparing the cs , cn , and ch@xmath0oh emission with other galaxies
, we suggest that the overlap region is dominated by large - scale shocks similar to the nucleus of ngc 253 . from the abundance analysis and distribution of the line ratios
, we postulate that the hcn - rich agn , the hco@xmath4-rich starbursts , and the ch@xmath0oh - rich overlap region are important drivers of the molecular chemistry of vv 114 .
we find a region with relatively high excitation ( @xmath41 35 k , @xmath41 10@xmath160 @xmath23 ) and star formation ( sfr = 0.10 @xmath6 0.05 m@xmath10 yr@xmath3 ) at the edge of the southern tail .
this region has a shorter @xmath154 of ( 0.40 @xmath6 0.22 ) @xmath69 gyr than the rest of the southern tail ( @xmath94 1.35 @xmath69 gyr ) , and we suggest that it is a forming tidal dwarf galaxy . the authors thanks the anonymous referee for comments that improved the contents of this paper .
ts thanks for yoichi tamura , takuma izumi , and akio taniguchi s help on the radex calculation .
we used a script developed by y. tamura for this calculation ( http://www.ioa.s.u-tokyo.ac.jp/`~`ytamura/wiki/?science%2fusingradex ) .
ts and other authors thank alma staff for their kind support .
ts , j. ueda , and k. tateuchi are financially supported by a research fellowship from the japan society for the promotion of science for young scientists .
d. iono was supported by the alma japan research grant of naoj chile observaory , naoj - alma-0011 and jsps kakenhi grant number 2580016 .
this paper makes use of the following alma data : ads / jao.alma#2011.0.00467.s .
alma is a partnership of eso ( representing its member states ) , nsf ( usa ) and nins ( japan ) , together with nrc ( canada ) and ncs and asiaa ( taiwan ) , in cooperation with the republic of chile .
the joint alma observatory is operated by eso , aui / nrao , and naoj .
99 aalto , s. , booth , r. s. , black , j. h. , & johansson , l. e. b. 1995 , , 300 , 369 aalto , s. , radford , s. j. e. , scoville , n. z. , & sargent , a. i. 1997 , , 475 , l107 aalto , s. , beswick , r. , jutte , e. 2010 , , 522 , a59 aladro , r. , martn , s. , martn - pintado , j. , et al .
2011 , , 535 , a84 alonso - herrero , a. , rieke , g. h. , rieke , m. j. , & scoville , n. z. 2002 , , 124 , 166 armus , l. , mazzarella , j. m. , evans , a. s. , et al .
2009 , , 121 , 559 blake , g. a. , sutton , e. c. , masson , c. r. , & phillips , t. g. 1987 , , 315 , 621 bolatto , a. d. , wolfire , m. , & leroy , a. k. 2013 , , 51 , 207 bournaud , f. , powell , l. c. , chapon , d. , & teyssier , r. 2011 , iau symposium , 271 , 160 braine , j. , lisenfeld , u. , due , p .- a . , & leon , s. 2000 , , 403 , 867 braine , j. , duc , p .- a .
, lisenfeld , u. , et al .
2001 , , 378 , 51 briggs , d. s. , & cornwell , t. j. 1992 , astronomical data analysis software and systems i , 25 , 170 casoli , f. , dupraz , c. , & combes , f. 1992 , , 264 , 55 condon , j. j. , huang , z .- p .
, yin , q. f. , & thuan , t. x. 1991 , , 378 , 65 contini , m. , & contini , t. 2003 , , 342 , 299 daddi , e. , elbaz , d. , walter , f. , et al .
2010 , , 714 , l118 davis , t. a. , heiderman , a. , evans , n. j. , & iono , d. 2013 , , 436 , 570 downes , d. , & solomon , p. m. 1998 , , 507 , 615 draine , b. t. , & lee , h. m. 1984 , , 285 , 89 engel , h. , davies , r. i. , genzel , r. , et al .
2010 , , 524 , a56 evans , a. s. 2008 , frontiers of astrophysics : a celebration of nrao s 50th anniversary , 395 , 113 fernndez - ontiveros , j. a. , prieto , m. a. , & acosta - pulido , j. a. 2009 , , 392 , l16 frayer , d. t. , ivison , r. j. , smail , i. , yun , m. s. , & armus , l. 1999 , , 118 , 139 fujii , k. , minamidani , t. , mizuno , n. , et al .
2014 , , 796 , 123 gao , y. , lo , k. y. , lee , s .- w . , & lee , t .- h .
2001 , , 548 , 172 gao , y. , & solomon , p. m. 2004 , , 606 , 271 garcia - burillo , s. , combes , f. , usero , a. , et al . 2014 , arxiv:1405.7706 genzel , r. , tacconi , l. j. , gracia - carpio , j. , et al .
2010 , , 407 , 2091 grimes , j. p. , heckman , t. , hoopes , c. , et al .
2006 , , 648 , 310 harada , n. , thompson , t. a. , & herbst , e. 2013 , , 765 , 108 henkel , c. , schilke , p. , & mauersberger , r. 1988 , , 201 , l23 henkel , c. , asiri , h. , ao , y. , et al .
2014 , , 565 , a3 henning , t. , michel , b. , & stognienko , r. 1995 , , 43 , 1333 hibbard , j. e. , van der hulst , j. m. , barnes , j. e. , & rich , r. m. 2001 , , 122 , 2969 hildebrand , r. h. 1983 , , 24 , 267 hgbom , j. a. 1974 , , 15 , 417 hopkins , p. f. , hernquist , l. , cox , t. j. , et al .
2006 , , 163 , 1 hopkins , p. f. , cox , t. j. , hernquist , l. , et al .
2013 , , 430 , 1901 huettemeister , s. , mauersberger , r. , & henkel , c. 1997 , , 326 , 59 imanishi , m. , nakanishi , k. , tamura , y. , oi , n. , & kohno , k. 2007 , , 134 , 2366 imanishi , m. , & nakanishi , k. 2013 , , 146 , 91 imanishi , m. , & nakanishi , k. 2014 , , 148 , 9 iono , d. , ho , p. t. p. , yun , m. s. , et al .
2004 , , 616 , l63 iono , d. , yun , m. s. , & ho , p. t. p. 2005
, , 158 , 1 iono , d. , wilson , c. d. , takakuwa , s. , et al .
2007 , , 659 , 283 iono , d. , saito , t. , yun , m. s. , et al .
2013 , , 65 , l7 izumi , t. , kohno , k. , martn , s. , et al .
2013 , , 65 , 100 jackson , j. m. , paglione , t. a. d. , carlstrom , j. e. , & rieu , n .- q .
1995 , , 438 , 695 kartaltepe , j. s. , sanders , d. b. , le floch , e. , et al .
2010 , apj , 721 , 98 kennicutt , r. c. , jr .
1998 , , 498 , 541 kohno , k. , matsushita , s. , vila - vilar , b. , et al .
2001 , the central kiloparsec of starbursts and agn : the la palma connection , 249 , 672 komugi , s. , sofue , y. , nakanishi , h. , onodera , s. , & egusa , f. 2005 , , 57 , 733 konstantopoulos , i. s. , bastian , n. , smith , l. j. , et al .
2009 , , 701 , 1015 krips , m. , neri , r. , garca - burillo , s. , et al .
2008 , , 677 , 262 langer , w. d. , & graedel , t. e. 1989 , , 69 , 241 le floch , e. , charmandaris , v. , laurent , o. , et al .
2002 , , 391 , 417 leroy , a. k. , walter , f. , sandstrom , k. , et al .
2013 , , 146 , 19 maloney , p. r. , hollenbach , d. j. , & tielens , a. g. g. m. 1996 , , 466 , 561 mao , r. q. , henkel , c. , schulz , a. , et al .
2000 , , 358 , 433 martn , s. , mauersberger , r. , martn - pintado , j. , henkel , c. , & garca - burillo , s. 2006 , , 164 , 450 mauersberger , r. , henkel , c. , wilson , t. l. , & harju , j. 1989 , , 226 , l5 mcmullin , j. p. , waters , b. , schiebel , d. , young , w. , & golap , k. 2007 , astronomical data analysis software and systems xvi , 376 , 127 minamidani , t. , mizuno , n. , mizuno , y. , et al .
2008 , , 175 , 485 scoville , n. , aussel , h. , sheth , k. , et al .
2014 , , 783 , 84 papadopoulos , p. p. , van der werf , p. p. , xilouris , e. m. , et al .
2012 , , 426 , 2601 rich , j. a. , kewley , l. j. , & dopita , m. a. 2011 , , 734 , 87 saito , t. , iono , d. , yun , m. , et al .
2013 , astronomical society of the pacific conference series , 476 , 287 sakamoto , k. , aalto , s. , combes , f. , evans , a. , & peck , a. 2014 , arxiv:1403.7117 sanders , d. b. , scoville , n. z. , & soifer , b. t. 1991 , , 370 , 158 schmidt , m. 1959 , , 129 , 243 schier , f. l. , van der tak , f. f. s. , van dishoeck , e. f. , & black , j. h. 2005 , , 432 , 369 seaquist , e. r. , & frayer , d. t. 2000 , , 540 , 765 seaquist , e. , yao , l. , dunne , l. , & cameron , h. 2004 , , 349 , 1428 sheffer , y. , federman , s. r. , lambert , d. l. , & cardelli , j. a. 1992 , , 397 , 482 sliwa , k. , wilson , c. d. , krips , m. , et al .
2013 , , 777 , 126 soifer , b. t. , sanders , d. b. , madore , b. f. , et al .
1987 , , 320 , 238 solomon , p. m. , & vanden bout , p. a. 2005 , , 43 , 677 sternberg , a. , & dalgarno , a. 1995 , , 99 , 565 tamura , y. , saito , t. , tsuru , t. g. , et al .
2014 , , 781 , l39 tateuchi , k. , motohara , k. , konishi , m. , et al .
2012 , publication of korean astronomical society , 27 , 297 teyssier , r. , chapon , d. , & bournaud , f. 2010 , , 720 , l149 thompson , a. r. , moran , j. m. , & swenson , g. w. , jr .
2001 , interferometry and synthesis in radio astronomy by a. richard thompson , james m. moran , and george w. swenson , jr .
2nd ed .
new york : wiley , c2001.xxiii , 692 p. : ill . ; 25 cm . a wiley - interscience publication . includes bibliographical references and indexes .
isbn : 0471254924 tsai , m. , hwang , c .- y . , matsushita , s. , baker , a. j. , & espada , d. 2012 , , 746 , 129 ueda , j. , iono , d. , petitpas , g. , et al .
2012 , , 745 , 65 ueda , j. , iono , d. , yun , m. s. , et al .
2014 , , 214 , 1 van dishoeck , e. f. , & black , j. h. 1988 , , 334 , 771 van der tak , f. f. s. , black , j. h. , schier , f. l. , jansen , d. j. , & van dishoeck , e. f. 2007 , , 468 , 627 warren , b. e. , wilson , c. d. , israel , f. p. , et al .
2010 , , 714 , 571 wild , w. , harris , a. i. , eckart , a. , et al .
1992 , , 265 , 447 wilson , c. d. , scoville , n. , madden , s. c. , & charmandaris , v. 2003 , , 599 , 1049 wilson , c. d. , petitpas , g. r. , iono , d. , et al .
2008 , , 178 , 189 wilson , c. d. , rangwala , n. , glenn , j. , et al .
2014 , , 789 , l36 wilson , t. l. , rohlfs , k. , huttemeister , s. 2009 , tools of radio astronomy , by thomas l. wilson ; kristen rohlfs and susanne httemeister .
isbn 978 - 3 - 540 - 85121 - 9
. published by springer - verlag , berlin , germany , 2009 . ,
wilson , t. l. , & rood , r. 1994 , , 32 , 191 yao , l. , seaquist , e. r. , kuno , n. , & dunne , l. 2003 , , 588 , 771 yun , m. s. , scoville , n. z. , & knop , r. a. 1994 , , 430 , l109 zhu , m. , seaquist , e. r. , & kuno , n. 2003 , , 588 , 243 zhu , m. , gao , y. , seaquist , e. r. , & dunne , l. 2007 , , 134 , 118 lrrrrrrrrrrcrl 2011 nov 6 & 101.5 , 103.5 & 114.0 , 115.1 & & 16 & cmp & 18 - 196&65 - 89 & 18 & uranus & 41 + 2012 may 4 & 101.5 , 103.5 & 114.0 , 115.1 & & 15 & ext & 39 - 402&48 - 62 & 8 & neptune & 40 + 2012 mar 27 & 97.5 , 99.5 & 110.2 , 111.5 & & 17 & ext & 18 - 401&54 - 73 & 19 & neptune & 22 + 2012 jul 2 & 97.5 , 99.5 & 110.2 , 111.5 & & 20 & ext & 16 - 402&71 - 117 & 21 & neptune & 39 + 2012 nov 5 & 331.1 , 333.0 & 343.5 , 345.3 & & 14 & cmp & 12 - 135&125 - 172 & 9 & uranus & 66 + 2012 nov 5 & 331.1 , 333.0 & 343.5 , 345.3 & & 14 & cmp & 12 - 135&108 - 155 & 9 & uranus & 67 + 2012 nov 5 & 331.1 , 333.0 & 343.5 , 345.3 & & 14 & cmp & 12 - 135&124 - 175 & 9 & callisto & 67 + 2012 jun 1 & 342.0 , 344.0 & 354.5 , 356.0 & & 18 & ext & 15 - 402&150 - 213 & 7 & uranus & 78 + 2012 jun 2 & 342.0 , 344.0 & 354.5 , 356.0 & & 20 & ext & 15 - 402&108 - 160 & 7 & uranus & 80 + 2012 jun 3 & 342.0 , 344.0 & 354.5 , 356.0 & & 20 & ext & 15 - 402&103 - 130 & 7 & uranus & 45 + lrrrrrrrrrcrl ch@xmath0oh ( 2@xmath261@xmath26 ) & 3 & 96.74 & 2.03 @xmath8 1.34 & 85.7 & 30 & 1.0 & 46 + cs ( 21 ) & 3 & 97.98 & 2.01 @xmath8 1.37 & 83.6 & 30 & 0.9 & 40 + @xmath2co ( 10 ) & 3 & 110.20 & 1.77 @xmath8 1.20 & 85.8 & 30 & 1.0 & 46 + cn ( 1@xmath270@xmath27 ) & 3 & 113.14 & 1.97 @xmath8 1.27 & @xmath16485.8 & 30 & 1.0 & 37 + cn ( 1@xmath280@xmath27 ) & 3 & 113.49 & 1.98 @xmath8 1.29 & @xmath16484.7 & 30 & 1.1 & 39 + co ( 10 ) & 3 & 115.27 & 1.97 @xmath8 1.35 & 82.3 & 10 & 2.3 & 76 + cs ( 76 ) & 7 & 342.88 & 0.47 @xmath8 0.39 & 54.2 & 30 & 0.7 & 38 + co ( 32 ) & 7 & 345.80 & 1.64 @xmath8 1.17 & 112.6 & 10 & 2.1 & 11 + hcn ( 43 ) & 7 & 354.51 & 0.46 @xmath8 0.38 & 51.5 & 30 & 0.8 & 42 + hco@xmath4 ( 43 ) & 7 & 356.73 & 0.45 @xmath8 0.37 & 53.4 & 30 & 0.9 & 50 + continuum & 3 & 110 & 1.89 @xmath8 1.28 & 81.8 & & 0.05 & 2.1 + continuum & 7 & 340 & 1.33 @xmath8 1.12 & 119.6 & & 0.11 & 0.8 + continuum & 7 & 340 & 0.45 @xmath8 0.38 & 56.2 & & 0.07 & 4.3 + lrrrrrrrrrcrl r7&0.62 @xmath6 0.08&@xmath5 0.18&0.58 @xmath6 0.14 + r8&0.70 @xmath6 0.08&@xmath5 0.18&@xmath5 0.42 + r9&0.82 @xmath6 0.08&0.21 @xmath6 0.06&@xmath5 0.42 + r10&0.81 @xmath6 0.08&0.20 @xmath6 0.06&@xmath5 0.42 + r11&0.92 @xmath6 0.08&@xmath5 0.18&@xmath5 0.42 + r17&0.42 @xmath6 0.08&@xmath5 0.18&@xmath5 0.42 + r18@xmath165&3.37 @xmath6 0.08&1.48 @xmath6 0.06&5.17 @xmath6 0.14 + r19&2.40 @xmath6 0.08&1.03 @xmath6 0.06&3.03 @xmath6 0.14 + r20&1.14 @xmath6 0.08&0.54 @xmath6 0.06&2.41 @xmath6 0.14 + r21@xmath166&5.00 @xmath6 0.08&1.86 @xmath6 0.06&8.20 @xmath6 0.14 + r22&1.85 @xmath6 0.08&0.81 @xmath6 0.06&4.37 @xmath6 0.14 + r23&1.34 @xmath6 0.08&0.50 @xmath6 0.06&3.55 @xmath6 0.14 + r24&0.67 @xmath6 0.08&@xmath5 0.18&1.19 @xmath6 0.14 + r25&1.07 @xmath6 0.08&0.39 @xmath6 0.06&3.01 @xmath6 0.14 + r26&0.81 @xmath6 0.08&0.33 @xmath6 0.06&1.73 @xmath6 0.14 + r27&1.29 @xmath6 0.08&0.41 @xmath6 0.06&1.40 @xmath6 0.14 + r28&0.81 @xmath6 0.08&0.21 @xmath6 0.06&0.76 @xmath6 0.14 + r29&0.35 @xmath6 0.08&@xmath5 0.18&0.71 @xmath6 0.14 + r30&@xmath5 0.24&@xmath5 0.18&0.70 @xmath6 0.14 + r39@xmath167&1.18 @xmath6 0.08&0.21 @xmath6 0.06&3.35 @xmath6 0.14 lrrrrrrrrrcrl r1&2.12 @xmath6 0.06&5.92 @xmath6 0.08&&0.31 @xmath6 0.01 & + r2&1.41 @xmath6 0.05&7.47 @xmath6 0.06&0.29 @xmath6 0.02&0.59 @xmath6 0.02&4 @xmath6 1 + r3&3.62 @xmath6 0.09&5.45 @xmath6 0.09&&0.17 @xmath6 0.01 & + r4&3.54 @xmath6 0.08&8.69 @xmath6 0.10&0.25 @xmath6 0.04&0.27 @xmath6 0.01&13 @xmath6 2 + r5&3.46 @xmath6 0.08&9.16 @xmath6 0.09&0.40 @xmath6 0.03&0.29 @xmath6 0.01&7 @xmath6 1 + r6&3.66 @xmath6 0.10&10.71 @xmath6 0.11&&0.33 @xmath6 0.01 & + r7&5.26 @xmath6 0.10&19.64 @xmath6 0.12&0.40 @xmath6 0.03&0.41 @xmath6 0.01&12 @xmath6 1 + r8&3.71 @xmath6 0.09&13.44 @xmath6 0.12&0.38 @xmath6 0.03&0.40 @xmath6 0.01&8 @xmath6 1 + r9&9.59 @xmath6 0.11&33.88 @xmath6 0.12&0.43 @xmath6 0.04&0.39 @xmath6 0.01&20 @xmath6 2 + r10&12.69 @xmath6 0.11&47.74 @xmath6 0.12&0.37 @xmath6 0.05&0.42 @xmath6 0.01&32 @xmath6 4 + r11&16.26 @xmath6 0.13&58.93 @xmath6 0.14&0.49 @xmath6 0.04&0.40 @xmath6 0.01&30 @xmath6 3 + r12&4.61 @xmath6 0.10&13.53 @xmath6 0.10&&0.33 @xmath6 0.01 & + r13&5.22 @xmath6 0.10&16.02 @xmath6 0.11&0.38 @xmath6 0.05&0.34 @xmath6 0.01&12 @xmath6 2 + r14&4.70 @xmath6 0.09&12.22 @xmath6 0.09&0.35 @xmath6 0.03&0.29 @xmath6 0.01&12 @xmath6 1 + r15&5.03 @xmath6 0.09&10.62 @xmath6 0.09&0.35 @xmath6 0.03&0.23 @xmath6 0.01&13 @xmath6 1 + r16&5.59 @xmath6 0.09&11.23 @xmath6 0.08&0.38 @xmath6 0.02&0.22 @xmath6 0.01&14 @xmath6 1 + r17&6.44 @xmath6 0.10&14.33 @xmath6 0.11&0.38 @xmath6 0.05&0.25 @xmath6 0.01&15 @xmath6 2 + r18@xmath165&16.54 @xmath6 0.13&106.39 @xmath6 0.15&0.60 @xmath6 0.05&0.71 @xmath6 0.01&25 @xmath6 2 + r19&11.09 @xmath6 0.12&54.98 @xmath6 0.14&0.60 @xmath6 0.05&0.55 @xmath6 0.01&17 @xmath6 2 + r20&18.33 @xmath6 0.13&93.14 @xmath6 0.14&0.64 @xmath6 0.06&0.56 @xmath6 0.01&26 @xmath6 2 + r21@xmath166&33.04 @xmath6 0.15&225.47 @xmath6 0.16&1.09 @xmath6 0.07&0.76 @xmath6 0.01&28 @xmath6 2 + r22&28.67 @xmath6 0.15&162.41 @xmath6 0.16&0.90 @xmath6 0.07&0.63 @xmath6 0.01&29 @xmath6 2 + r23&33.65 @xmath6 0.16&177.02 @xmath6 0.17&0.95 @xmath6 0.07&0.58 @xmath6 0.01&32 @xmath6 2 + r24&16.12 @xmath6 0.14&68.28 @xmath6 0.15&0.48 @xmath6 0.07&0.47 @xmath6 0.01&31 @xmath6 4 + r25&25.23 @xmath6 0.16&138.20 @xmath6 0.16&0.86 @xmath6 0.06&0.61 @xmath6 0.01&27 @xmath6 2 + r26&20.44 @xmath6 0.14&97.56 @xmath6 0.15&0.69 @xmath6 0.06&0.53 @xmath6 0.01&27 @xmath6 3 + r27&12.45 @xmath6 0.12&67.10 @xmath6 0.13&0.54 @xmath6 0.06&0.60 @xmath6 0.01&21 @xmath6 2 + r28&5.78 @xmath6 0.10&36.14 @xmath6 0.14&0.35 @xmath6 0.03&0.69 @xmath6 0.01&15 @xmath6 1 + r29 & & & & & + r30&6.39 @xmath6 0.12&30.50 @xmath6 0.12&0.28 @xmath6 0.03&0.53 @xmath6 0.01&21 @xmath6 3 + r31&4.93 @xmath6 0.11&19.03 @xmath6 0.13&0.30 @xmath6 0.04&0.43 @xmath6 0.01&15 @xmath6 2 + r32&2.34 @xmath6 0.05&7.34 @xmath6 0.08&0.21 @xmath6 0.02&0.35 @xmath6 0.01&10 @xmath6 1 + r33&1.44 @xmath6 0.05&2.03 @xmath6 0.05&&0.16 @xmath6 0.01 & + r34&2.81 @xmath6 0.08&11.65 @xmath6 0.10&0.32 @xmath6 0.03&0.46 @xmath6 0.01&8 @xmath6 1 + r35&5.36 @xmath6 0.09&9.40 @xmath6 0.09&0.30 @xmath6 0.03&0.19 @xmath6 0.01&16 @xmath6 2 + r36&5.55 @xmath6 0.12&17.95 @xmath6 0.14&0.31 @xmath6 0.02&0.36 @xmath6 0.01&17 @xmath6 1 + r37&4.84 @xmath6 0.11&8.65 @xmath6 0.11&0.31 @xmath6 0.03&0.20 @xmath6 0.01&14 @xmath6 2 + r38&2.66 @xmath6 0.07&8.51 @xmath6 0.09&0.33 @xmath6 0.02&0.36 @xmath6 0.01&7 @xmath6 1 + r39@xmath167&27.54 @xmath6 0.15&155.31 @xmath6 0.15&0.97 @xmath6 0.06&0.63 @xmath6 0.01&26 @xmath6 2 + lrrrrrrrrrcrl s0@xmath165&1.34 @xmath6 0.09 & 0.88@xmath6 0.10 & 1.52 @xmath6 0.20 + s1&0.11 @xmath6 0.03 & 0.34@xmath6 0.06 & 0.34 @xmath6 0.10 + s2&@xmath5 0.05 & 0.12@xmath6 0.06 & @xmath5 0.43 + s3@xmath166&0.49 @xmath6 0.07 & 1.05@xmath6 0.08 & 0.46 @xmath6 0.07 + s4&0.13 @xmath6 0.06 & 0.63@xmath6 0.08 & 0.20 @xmath6 0.09 + s5@xmath166&0.81 @xmath6 0.07 & 2.36@xmath6 0.09 & 0.34 @xmath6 0.03 + s6&0.40 @xmath6 0.07 & 0.92@xmath6 0.08 & 0.43 @xmath6 0.09 + s7&0.19 @xmath6 0.07 & 0.42@xmath6 0.08 & 0.45 @xmath6 0.18 + s8&0.06 @xmath6 0.04 & 0.14@xmath6 0.06 & 0.44 @xmath6 0.31 + s9&0.07 @xmath6 0.05 & 0.13@xmath6 0.06 & 0.54 @xmath6 0.46 + s10&@xmath5 0.04 & 0.28@xmath6 0.07 & @xmath5 0.14 + s11@xmath167&@xmath5 0.05 & 0.15@xmath6 0.06 & @xmath5 0.33 + s12@xmath167&@xmath5 0.04 & 0.32@xmath6 0.07 & @xmath5 0.12 + s13&@xmath5 0.04 & 0.17@xmath6 0.06 & @xmath5 0.24 + s14&@xmath5 0.05 & 0.19@xmath6 0.07 & @xmath5 0.26 lrrrrrrrrcrl r1&0.40 @xmath6 0.07&0.13 @xmath6 0.01&@xmath5 0.09&0.32 @xmath6 0.07&@xmath94 4 + r2&0.69 @xmath6 0.07&0.25 @xmath6 0.01&@xmath5 0.09&0.37 @xmath6 0.04&@xmath94 8 + r3&0.61 @xmath6 0.07&0.07 @xmath6 0.01&@xmath5 0.09&0.12 @xmath6 0.02&@xmath94 7 + r4&0.98 @xmath6 0.07&0.30 @xmath6 0.01&@xmath5 0.09&0.30 @xmath6 0.02&@xmath94 11 + r5&1.19 @xmath6 0.07&0.42 @xmath6 0.01&0.06 @xmath6 0.03&0.35 @xmath6 0.02&20 @xmath6 2 + r6&0.87 @xmath6 0.07&0.21 @xmath6 0.01&@xmath5 0.09&0.24 @xmath6 0.02&@xmath94 10 + r7a&1.34 @xmath6
0.07&0.32 @xmath6 0.01&@xmath5 0.09&0.24 @xmath6 0.01&@xmath94 15 + r7b&0.45 @xmath6 0.07&0.36 @xmath6 0.01&0.06 @xmath6 0.03&0.79 @xmath6 0.12&7 @xmath6 1 + r8a&0.21 @xmath6 0.07&@xmath5 0.03&@xmath5 0.09&@xmath94 0.14&@xmath94 2 + r8b&0.96 @xmath6 0.07&0.24 @xmath6 0.01&0.08 @xmath6 0.03&0.25 @xmath6 0.02&11 @xmath6 1 + r9a&0.40 @xmath6 0.07&0.12 @xmath6 0.01&@xmath5 0.09&0.30 @xmath6 0.06&@xmath94 4 + r9b&2.60 @xmath6 0.07&0.99 @xmath6 0.01&0.08 @xmath6 0.03&0.38 @xmath6 0.01&33 @xmath6 2 + r10a&0.80 @xmath6 0.07&0.24 @xmath6 0.01&0.05 @xmath6 0.03&0.30 @xmath6 0.03&17 @xmath6 2 + r10b&2.81 @xmath6 0.07&1.27 @xmath6 0.01&0.09 @xmath6 0.03&0.45 @xmath6 0.01&30 @xmath6 1 + r11a&2.11 @xmath6 0.07&0.82 @xmath6 0.01&0.12 @xmath6 0.03&0.39 @xmath6 0.01&18 @xmath6 1 + r11b&3.34 @xmath6 0.07&1.36 @xmath6 0.01&0.04 @xmath6 0.03&0.41 @xmath6 0.01&89 @xmath6 7 + r12a&0.55 @xmath6 0.07&0.23 @xmath6 0.01&@xmath5 0.09&0.41 @xmath6 0.06&@xmath94 6 + r12b&1.30 @xmath6 0.07&0.41 @xmath6 0.01&@xmath5 0.09&0.32 @xmath6 0.02&@xmath94 14 + r13a&1.12 @xmath6 0.07&0.34 @xmath6 0.01&0.06 @xmath6 0.03&0.30 @xmath6 0.02&18 @xmath6 1 + r13b&0.67 @xmath6 0.07&0.30 @xmath6 0.01&@xmath5 0.09&0.45 @xmath6 0.05&@xmath94 7 + r14&1.61 @xmath6 0.07&0.39 @xmath6 0.01&0.08 @xmath6 0.03&0.24 @xmath6 0.01&20 @xmath6 1 + r15&1.77 @xmath6 0.07&0.42 @xmath6 0.01&0.08 @xmath6 0.03&0.24 @xmath6 0.01&22 @xmath6 1 + r16&1.38 @xmath6 0.07&0.33 @xmath6 0.01&0.12 @xmath6 0.03&0.24 @xmath6 0.01&12 @xmath6 1 + r17&1.64 @xmath6 0.07&0.35 @xmath6 0.01&0.05 @xmath6 0.03&0.21 @xmath6 0.01&36 @xmath6 3 + r18@xmath165&3.52 @xmath6 0.07&1.79 @xmath6 0.01&0.14 @xmath6 0.03&0.51 @xmath6 0.01&25 @xmath6 1 + r19a&1.01 @xmath6 0.07&0.54 @xmath6 0.01&0.09 @xmath6 0.03&0.54 @xmath6 0.04&11 @xmath6 1 + r19b&1.99 @xmath6 0.07&0.98 @xmath6 0.01&0.09 @xmath6 0.03&0.49 @xmath6 0.02&23 @xmath6 1 + r20a&3.33 @xmath6 0.07&1.99 @xmath6 0.01&0.11 @xmath6 0.03&0.60 @xmath6 0.01&30 @xmath6 1 + r20b&1.13 @xmath6 0.07&0.67 @xmath6 0.01&@xmath5 0.09&0.59 @xmath6 0.04&@xmath94 13 + r21a@xmath166&5.53 @xmath6 0.07&4.49 @xmath6 0.01&0.23 @xmath6 0.03&0.81 @xmath6 0.01&24 @xmath6 1 + r21b@xmath166&1.22 @xmath6 0.07&0.95 @xmath6 0.01&@xmath5 0.09&0.78 @xmath6 0.05&@xmath94 14 + r21c@xmath166&0.40 @xmath6 0.07&0.72 @xmath6 0.01&@xmath5 0.09&1.80 @xmath6 0.05&@xmath94 4 + r22a&3.60 @xmath6 0.07&1.93 @xmath6 0.01&0.12 @xmath6 0.03&0.54 @xmath6 0.01&29 @xmath6 1 + r22b&3.83 @xmath6 0.07&2.61 @xmath6 0.01&0.16 @xmath6 0.03&0.68 @xmath6 0.01&24 @xmath6 1 + r23a&4.17 @xmath6 0.07&2.56 @xmath6 0.01&0.15 @xmath6 0.03&0.61 @xmath6 0.01&28 @xmath6 1 + r23b&3.39 @xmath6 0.07&1.87 @xmath6 0.01&0.12 @xmath6 0.03&0.55 @xmath6 0.01&28 @xmath6 1 + r24a&1.23 @xmath6 0.07&0.46 @xmath6 0.01&0.05 @xmath6 0.03&0.37 @xmath6 0.02&23 @xmath6 2 + r24b&1.60 @xmath6 0.07&0.96 @xmath6 0.01&0.06 @xmath6 0.03&0.60 @xmath6 0.03&28 @xmath6 2 + r24c&1.78 @xmath6 0.07&0.58 @xmath6 0.01&0.06 @xmath6 0.03&0.32 @xmath6 0.01&32 @xmath6 2 + r25a&4.89 @xmath6 0.07&3.18 @xmath6 0.01&0.23 @xmath6 0.03&0.65 @xmath6 0.01&21 @xmath6 1 + r25b&1.77 @xmath6 0.07&0.59 @xmath6 0.01&0.06 @xmath6 0.03&0.33 @xmath6 0.01&31 @xmath6 2 + r26a&4.49 @xmath6 0.07&2.57 @xmath6 0.01&0.14 @xmath6 0.03&0.57 @xmath6 0.01&33 @xmath6 1 + r26b&1.07 @xmath6 0.07&0.31 @xmath6 0.01&0.07 @xmath6 0.03&0.29 @xmath6 0.02&15 @xmath6 1 + r27a&1.96 @xmath6 0.07&1.31@xmath6 0.01&0.04 @xmath6 0.03&0.67 @xmath6 0.02&47 @xmath6 4 + r27b&1.53 @xmath6 0.07&0.75 @xmath6 0.01&0.13 @xmath6 0.03&0.49 @xmath6 0.02&12 @xmath6 1 + r28a&0.76 @xmath6 0.07&0.37 @xmath6 0.01&0.05 @xmath6 0.03&0.48 @xmath6 0.05&16 @xmath6 2 + r28b&0.86 @xmath6 0.07&0.54 @xmath6 0.01&@xmath5 0.09&0.62 @xmath6 0.05&@xmath94 10 + r29&@xmath5 0.21&@xmath5 0.03&@xmath5 0.09 & & + r30a&0.77 @xmath6 0.07&0.39 @xmath6 0.01&@xmath5 0.09&0.51 @xmath6 0.05&@xmath94 9 + r30b&1.39 @xmath6 0.07&0.69 @xmath6 0.01&@xmath5 0.09&0.50 @xmath6 0.03&@xmath94 15 + r31a&0.72 @xmath6 0.07&0.33 @xmath6 0.01&0.06 @xmath6 0.03&0.46 @xmath6 0.03&13 @xmath6 1 + r31b&0.55 @xmath6 0.07&0.24 @xmath6 0.01&@xmath5 0.09&0.44 @xmath6 0.06&@xmath94 6 + r32&0.55 @xmath6 0.07&0.29 @xmath6 0.01&@xmath5 0.09&0.52 @xmath6 0.07&@xmath94 6 + r33&0.35 @xmath6 0.07&0.14 @xmath6 0.01&@xmath5 0.09&0.41 @xmath6 0.09&@xmath94 4 + r34&0.83 @xmath6 0.07&0.37 @xmath6 0.01&0.05 @xmath6 0.03&0.45 @xmath6 0.04&18 @xmath6 2 + r35&1.42 @xmath6 0.07&0.30 @xmath6 0.01&@xmath5 0.09&0.21 @xmath6 0.01&@xmath94 16 + r36a&0.65 @xmath6 0.07&0.24 @xmath6 0.01&0.07 @xmath6 0.03&0.37 @xmath6 0.04&9 @xmath6 1 + r36b&0.47 @xmath6 0.07&0.22 @xmath6 0.01&@xmath5 0.09&0.47 @xmath6 0.07&@xmath94 5 + r37a&0.79 @xmath6 0.07&0.15 @xmath6 0.01&0.06 @xmath6 0.03&0.19 @xmath6 0.02&14 @xmath6 1 + r37b&0.71 @xmath6 0.07&0.18 @xmath6 0.01&@xmath5 0.09&0.26 @xmath6 0.03&@xmath94 8 + r38&0.38 @xmath6 0.07&0.16 @xmath6 0.01&0.04 @xmath6 0.03&0.42 @xmath6 0.08&9 @xmath6 2 + r39a@xmath167&4.91 @xmath6 0.07&3.15 @xmath6 0.01&0.24 @xmath6 0.03&0.64 @xmath6 0.01&21 @xmath6 1 + r39b@xmath167&1.96 @xmath6 0.07&0.81 @xmath6 0.01&0.06 @xmath6 0.03&0.41 @xmath6 0.02&31 @xmath6 2 + lrrrrrrrrrcrl s0@xmath165&0.32 @xmath6 0.05&0.27 @xmath6 0.05&1.18 @xmath6 0.29 + s1&@xmath5 0.15&0.20 @xmath6 0.05&@xmath5 0.75 + s2&@xmath5 0.15&0.16 @xmath6 0.05&@xmath5 0.94 + s3@xmath166&0.20 @xmath6 0.05&0.65 @xmath6 0.05&0.31 @xmath6 0.08 + s4&0.10 @xmath6 0.05&0.51 @xmath6 0.05&0.19 @xmath6 0.10 + s5@xmath166&0.32 @xmath6 0.05&0.91 @xmath6 0.05&0.35 @xmath6 0.06 + s6&0.14 @xmath6 0.05&0.55 @xmath6 0.05&0.26 @xmath6 0.09 + s7&@xmath5 0.15&0.24 @xmath6 0.05&@xmath5 0.63 + s8&@xmath5 0.15&0.14 @xmath6 0.05&@xmath5 1.07 + s9&@xmath5 0.15&0.17 @xmath6
0.05&@xmath5 0.88 + s10&0.10 @xmath6 0.05&0.17 @xmath6 0.05&0.60 @xmath6 0.34 + s11@xmath167&@xmath5 0.15&0.18 @xmath6 0.05&@xmath5 0.83 + s12@xmath167&@xmath5 0.15&0.24 @xmath6 0.05&@xmath5 0.63 + s13&@xmath5 0.15&0.20 @xmath6 0.05&@xmath5 0.75 + s14&@xmath5 0.15&0.23 @xmath6 0.05&@xmath5 0.65 lrrrrrrrrrcrl e0@xmath165&2.94 @xmath6 0.04&0.06 @xmath6 0.01&0.04 @xmath6 0.01&0.014 @xmath6 0.004&1.55 @xmath6 0.49 + e1@xmath166&5.82 @xmath6 0.04&0.11 @xmath6 0.01&0.32 @xmath6 0.01&0.056 @xmath6 0.002&0.36 @xmath6 0.03 + e2@xmath167&4.71 @xmath6 0.04&0.04 @xmath6 0.01&0.08 @xmath6 0.01&0.017 @xmath6 0.002&0.52 @xmath6 0.14 lrrrrrrrrrcrl 1 & 5 - 300 ( 5 ) & 2 - 5 ( 0.1 ) & 18 - 22 ( 0.1 ) & 70 & 800 + 2 & 5 - 400 ( 5 ) & 3 - 7 ( 0.1 ) & 21.2 , 21.6 , 21.5 & 1 - 10 ( 1 ) & 320 + lrrrrrrrrrcrl r1&3.0 @xmath6 0.6&@xmath5 24.6&@xmath5 0.15&@xmath94 0.20&@xmath5 0.51&@xmath5 17&@xmath94 180&@xmath5 3.6 + r2&2.0 @xmath6 0.4&46.0 @xmath6 8.2&0.29 @xmath6 0.05&0.07 @xmath6 0.02&@xmath5 0.51&@xmath5 17&@xmath94 119&@xmath5 3.6 + r3&5.1 @xmath6 0.8&11.0 @xmath6 8.2&0.07 @xmath6 0.05&0.75 @xmath6 0.57&@xmath5 0.51&@xmath5 17&@xmath94 306&@xmath5 3.6 + r4&5.0 @xmath6 0.7&9.3 @xmath6 8.2&0.06 @xmath6 0.05&0.88 @xmath6 0.78&@xmath5 0.51&@xmath5 17&@xmath94 299&@xmath5 3.6 + r5&4.9 @xmath6 0.7&41.8 @xmath6 8.2&0.26 @xmath6 0.05&0.19 @xmath6 0.05&@xmath5 0.51&@xmath5 17&@xmath94 293&@xmath5 3.6 + r6&5.2 @xmath6 0.9&60.7 @xmath6 8.2&0.38 @xmath6 0.05&0.14 @xmath6 0.03&@xmath5 0.51&@xmath5 17&@xmath94 310&@xmath5 3.6 + r7&7.5 @xmath6 0.9&172.1 @xmath6 8.2&1.07 @xmath6 0.05&0.07 @xmath6 0.01&0.61 @xmath6 0.18&20 @xmath6 6&371 @xmath6 118&5.2 @xmath6 1.4 + r8&5.3 @xmath6 0.8&69.8 @xmath6 8.2&0.43 @xmath6 0.05&0.12 @xmath6 0.02&@xmath5 0.51&@xmath5 17&@xmath94 314&@xmath5 3.6 + r9&13.6 @xmath6 1.2&125.1 @xmath6 8.2&0.78 @xmath6 0.05&0.18 @xmath6 0.02&@xmath5 0.51&@xmath5 17&@xmath94 811&@xmath5 3.6 + r10&18.0 @xmath6 1.3&49.0 @xmath6 8.2&0.30 @xmath6 0.05&0.59 @xmath6 0.11&@xmath5 0.51&@xmath5 17&@xmath94 1074&@xmath5 3.6 + r11&23.1 @xmath6 1.6&106.9 @xmath6 8.2&0.66 @xmath6 0.05&0.35 @xmath6 0.04&@xmath5 0.51&@xmath5 17&@xmath94 1376&@xmath5 3.6 + r12&6.5 @xmath6 0.9&25.5 @xmath6 8.2&0.16 @xmath6 0.05&0.41 @xmath6 0.14&@xmath5 0.51&@xmath5 17&@xmath94 390&@xmath5 3.6 + r13&7.4 @xmath6 1.0&25.9 @xmath6 8.2&0.16 @xmath6 0.05&0.46 @xmath6 0.16&@xmath5 0.51&@xmath5 17&@xmath94 441&@xmath5 3.6 + r14&6.7 @xmath6 0.8&25.5 @xmath6 8.2&0.16 @xmath6 0.05&0.42 @xmath6 0.14&@xmath5 0.51&@xmath5 17&@xmath94 398&@xmath5 3.6 + r15&7.1 @xmath6 0.9&14.7 @xmath6 8.2&0.09 @xmath6 0.05&0.78 @xmath6 0.44&@xmath5 0.51&@xmath5 17&@xmath94 425&@xmath5 3.6 + r16&7.9 @xmath6 0.9&15.9 @xmath6 8.2&0.10 @xmath6 0.05&0.80 @xmath6 0.42&@xmath5 0.51&@xmath5 17&@xmath94 473&@xmath5 3.6 + r17&9.1 @xmath6 1.0&35.2 @xmath6 8.2&0.22 @xmath6 0.05&0.42 @xmath6 0.11&@xmath5 0.51&@xmath5 17&@xmath94 545&@xmath5 3.6 + r18@xmath165&23.5 @xmath6 1.6&277.4 @xmath6 8.2&1.72 @xmath6 0.05&0.14 @xmath6 0.01&5.57 @xmath6 0.58&183 @xmath6 19&128 @xmath6 16&47.1 @xmath6 4.9 + r19&15.7 @xmath6 1.3&150.8 @xmath6 8.2&0.94 @xmath6 0.05&0.17 @xmath6 0.02&2.70 @xmath6 0.32&89 @xmath6 11&177 @xmath6 25&22.9 @xmath6 2.6 + r20&26.0 @xmath6 1.7&78.1 @xmath6 8.2&0.48 @xmath6 0.05&0.54 @xmath6 0.07&2.40 @xmath6 0.29&79 @xmath6 10&329 @xmath6 46&20.3 @xmath6 2.4 + r21@xmath166&46.9 @xmath6 2.7&303.2 @xmath6 8.2&1.88 @xmath6 0.05&0.25 @xmath6 0.02&8.55 @xmath6 0.87&281 @xmath6 29&167 @xmath6 19&72.3 @xmath6 7.3 + r22&40.7 @xmath6 2.4&197.1 @xmath6 8.2&1.22 @xmath6 0.05&0.33 @xmath6 0.02&4.08 @xmath6 0.44&134 @xmath6 15&303 @xmath6 37&34.5 @xmath6 3.7 + r23&47.8 @xmath6 2.7&217.5 @xmath6 8.2&1.35 @xmath6 0.05&0.35 @xmath6 0.02&3.60 @xmath6 0.40&119 @xmath6 13&403 @xmath6 50&30.5 @xmath6 3.3 + r24&22.9 @xmath6 1.7&145.5 @xmath6 8.2&0.90 @xmath6 0.05&0.25 @xmath6 0.02&1.28 @xmath6 0.21&42 @xmath6 7&543 @xmath6 99&10.8 @xmath6 1.7 + r25&35.8 @xmath6 2.2&281.2 @xmath6 8.2&1.74 @xmath6 0.05&0.21 @xmath6 0.01&3.10 @xmath6 0.35&102 @xmath6 12&351 @xmath6 46&26.2 @xmath6 2.9 + r26&29.0 @xmath6 1.9&250.1 @xmath6 8.2&1.55 @xmath6 0.05&0.19 @xmath6 0.01&1.66 @xmath6 0.24&54 @xmath6 8&533 @xmath6 84&14.0 @xmath6 1.9 + r27&17.7 @xmath6 1.4&507.8 @xmath6 8.2&3.15 @xmath6 0.05&0.06 @xmath6 0.01&1.42 @xmath6 0.22&47 @xmath6 7&379 @xmath6 66&12.0 @xmath6 1.7 + r28&8.2 @xmath6 1.0&246.3 @xmath6 8.2&1.53 @xmath6 0.05&0.05 @xmath6 0.01&0.73 @xmath6 0.18&24 @xmath6 6&344 @xmath6 97&6.1 @xmath6 1.4 + r29&@xmath5 1.4&147.8 @xmath6 8.2&0.92 @xmath6 0.05&@xmath5 0.02&0.76 @xmath6 0.19&25 @xmath6 6&@xmath5 55&6.4 @xmath6 1.4 + r30&9.1 @xmath6 1.1&40.2 @xmath6 8.2&0.25 @xmath6 0.05&0.36 @xmath6 0.09&0.79 @xmath6 0.19&26 @xmath6 6&348 @xmath6 92&6.7 @xmath6 1.4 + r31&7.0 @xmath6 1.0&19.0 @xmath6 8.2&0.12 @xmath6 0.05&0.59 @xmath6 0.27&@xmath5 0.51&@xmath5 17&@xmath94 418&@xmath5 3.6 + r32&3.3 @xmath6 0.5&60.3 @xmath6 8.2&0.37 @xmath6 0.05&0.09 @xmath6 0.02&@xmath5 0.51&@xmath5 17&@xmath94 198&@xmath5 3.6 + r33&2.0 @xmath6 0.4&55.0 @xmath6 8.2&0.34 @xmath6 0.05&0.06 @xmath6 0.02&@xmath5 0.51&@xmath5 17&@xmath94 122&@xmath5 3.6 + r34&4.0 @xmath6 0.7&10.9 @xmath6 8.2&0.07 @xmath6 0.05&0.59 @xmath6 0.45&@xmath5 0.51&@xmath5 17&@xmath94 238&@xmath5 3.6 + r35&7.6 @xmath6 0.9&@xmath5 24.6&@xmath5 0.15&@xmath94 0.51&@xmath5 0.51&@xmath5 17&@xmath94 453&@xmath5 3.6 + r36&7.9 @xmath6 1.1&@xmath5 24.6&@xmath5 0.15&@xmath94 0.52&@xmath5 0.51&@xmath5 17&@xmath94 469&@xmath5 3.6 + r37&6.9 @xmath6 1.0&@xmath5 24.6&@xmath5 0.15&@xmath94 0.46&@xmath5 0.51&@xmath5 17&@xmath94 410&@xmath5 3.6 + r38&3.8 @xmath6 0.7&15.2 @xmath6 8.2&0.10 @xmath6 0.05&0.40 @xmath6 0.22&@xmath5 0.51&@xmath5 17&@xmath94 225&@xmath5 3.6 + r39@xmath167&39.1 @xmath6 2.3&273.6 @xmath6 8.2&1.70 @xmath6 0.05&0.23 @xmath6 0.02&3.50 @xmath6 0.58&115 @xmath6 19&339 @xmath6 60&29.6 @xmath6 3.2 + lrrrrrrrrrcrl s0@xmath165&38.3 @xmath6 4.7 & 322 @xmath6 12 & 2.00 @xmath6 0.07 & 19.2 @xmath6 2.5 & 1.71 @xmath6 0.18 & 56.1 @xmath6 6.0&68 @xmath6 11 + s1&3.2 @xmath6 1.8 & 240 @xmath6 12 & 1.49 @xmath6 0.07 & 2.2 @xmath6 1.2 & 0.25 @xmath6 0.07 & 8.1 @xmath6 2.3&40 @xmath6 25 + s2&@xmath5 4.3 & @xmath5 36 & @xmath5 0.21 & & 0.10 @xmath6 0.07 & 3.1 @xmath6 2.2&@xmath5 137 + s3@xmath166&13.9 @xmath6 2.6 & 663 @xmath6 12 & 4.11 @xmath6 0.07 & 3.4 @xmath6 0.6 & 1.28 @xmath6 0.14 & 42.1 @xmath6 4.7&33 @xmath6 7 + s4&3.6 @xmath6 2.2 & 446 @xmath6 12 & 2.76 @xmath6 0.07 & 1.3 @xmath6 0.8 & 0.63 @xmath6 0.09 & 20.6 @xmath6 3.0&17 @xmath6 11 + s5@xmath166&23.2
@xmath6 3.6 & 606 @xmath6 12 & 3.75 @xmath6 0.07 & 6.2 @xmath6 0.7 & 2.56 @xmath6 0.26 & 84.4 @xmath6 8.7&27 @xmath6 5 + s6&11.5 @xmath6 2.6 & 268 @xmath6 12 & 1.66 @xmath6 0.07 & 6.9 @xmath6 1.6 & 0.90 @xmath6 0.11 & 29.7 @xmath6 3.7&39 @xmath6 10 + s7&5.4 @xmath6 2.3 & 138 @xmath6 12 & 0.85 @xmath6 0.07 & 6.4 @xmath6 2.7 & 0.40 @xmath6 0.08 & 13.0 @xmath6 2.5&42 @xmath6 19 + s8&1.8 @xmath6 1.6 & 158 @xmath6 12 & 0.98 @xmath6 0.07 & 1.8 @xmath6 1.6 & 0.32 @xmath6 0.07 & 10.5 @xmath6 2.4&17 @xmath6 16 + s9&2.0 @xmath6 1.8 & 106 @xmath6 12 & 0.66 @xmath6 0.07 & 3.1 @xmath6 2.8 & 0.28 @xmath6 0.07 & 9.2 @xmath6 2.3&22 @xmath6 21 + s10&@xmath5 4.3 & 203 @xmath6 12 & 1.26 @xmath6 0.07 & @xmath5 3.4 & 0.34 @xmath6 0.07 & 11.3 @xmath6 2.4&@xmath5 38 + s11@xmath167&@xmath5 4.3 & 242 @xmath6 12 & 1.50 @xmath6 0.07 & @xmath5 2.9 & 0.21 @xmath6 0.07 & 7.0 @xmath6 2.3&@xmath5 61 + s12@xmath167&@xmath5 1.4 & 323 @xmath6 12 & 2.00 @xmath6 0.07 & @xmath5 2.1 & 0.46 @xmath6 0.08 & 15.3 @xmath6 2.6&@xmath5 28 + s13&@xmath5 4.3 & 291 @xmath6 12 & 1.81 @xmath6 0.07 & @xmath5 2.4 & 0.46 @xmath6 0.08 & 15.2 @xmath6 2.6&@xmath5 28 + s14&@xmath5 1.4 & 219 @xmath6 12 & 1.36 @xmath6 0.07 & @xmath5 3.2 & 0.27 @xmath6 0.07 & 9.0 @xmath6 2.3&@xmath5 48 + lrrrrrrrrrrrl r5&@xmath94 5 & 90 & & @xmath5 3.7 & 2.4 & & 20.4 - 21.2 & 20.7 & 0.0001508 + r7b&5 - 120 & 25 & & @xmath94 2.8 & 4.1 & & @xmath94 20.8 & 21.4 & 0.0008348 + r8b&5 - 30 & 25 & & @xmath5 2.4 & 2.0 & & 20.9 - 21.8 & 21.3 & 0.04646 + r9b&@xmath94 5 & 50 & & 2.2 - 4.0 & 3.0 & & 20.1 - 20.7 & 20.4 & 0.004822 + r10a&5 - 120 & 65 & & @xmath5 3.3 & 2.1 & & 20.4 - 21.8 & 20.9 & 0.001529 + r10b&5 - 20 & 10 & & 3.6 - 4.6 & 4.4 & & 19.9 - 20.4 & 20.1 & 0.00948 + r11a&@xmath94 20 & 75 & & @xmath5 3.2 & 2.5 & & 20.5 - 21.0 & 20.8 & 0.000315 + r11b&&15 & & 2.6 - 4.6 & 3.9 & & @xmath5 20.4 & 18.1 & 0.001707 + r13a&5 - 200 & 85 & & @xmath5 3.5 & 2.2 & & 20.3 - 21.4 & 20.7 & 0.002488 + r14&5 - 200 & 15 & & @xmath5 3.4 & 3.2 & & 20.2 - 20.6 & 20.4 & 0.004464 + r15&5 - 140 & 35 & & @xmath5 3.4 & 2.6 & & 20.2 - 20.8 & 20.5 & 0.001343 + r16&5 - 30 & 15 & & 2.4 - 3.2 & 2.8 & & 20.6 - 21.0 & 20.8 & 0.03139 + r17&5 - 180 & 135 & & @xmath5 3.6 & 2.0 & & 19.6 - 21.2 & 20.2 & 0.003854 + r18@xmath165&@xmath94 35 & 180 & & 2.4 - 3.3 & 2.7 & & 20.6 - 21.0 & 20.8 & 0.000186 + r19a&@xmath94 25 & 85 & & @xmath5 3.2 & 2.5 & & 20.9 - 21.5 & 21.2 & 0.0001408 + r19b&@xmath94 20 & 100 & & 2.2 - 3.5 & 2.8 & & 20.5 - 21.0 & 20.8 & 0.002962 + r20a&@xmath94 30 & 100 & & 2.7 - 3.6 & 3.1 & & 20.5 - 21.1 & 20.8 & 0.001267 + r21a@xmath166&25 - 90 & 50 & & 3.4 - 5.0 & 3.7 & & 20.8 - 21.5 & 21.1 & 0.01197 + r22a&@xmath94 25 & 70 & & 2.6 - 3.6 & 3.1 & & 20.5 - 20.9 & 20.7 & 0.01047 + r22b&@xmath94 90 & 265 & & 2.8 - 3.3 & 3.0 & & 21.0 - 21.6 & 21.3 & 0.00147 + r23a&@xmath94 40 & 140 & & 2.7 - 3.5 & 3.1 & & 20.6 - 21.1 & 20.9 & 0.001661 + r23b&@xmath94 35 & 140 & & 2.5 - 3.5 & 2.9 & & 20.5 - 21.0 & 20.8 & 0.002198 + r24a&@xmath94 5 & 160 & & @xmath5 4.9 & 2.4 & & 20.2 - 21.3 & 20.6 & 0.000268 + r24b&@xmath94 15 & 275 & & 2.5 - 3.6 & 2.8 & & 20.6 - 21.4 & 21.0 & 0.001229 + r24c&@xmath94 5 & 115 & & @xmath5 3.9 & 2.5 & & 20.0 - 20.8 & 20.4 & 0.0001756 + r25a&5 - 40 & 20 & & 3.5 - 5.0 & 3.9 & & 20.5 - 20.8 & 20.7 & 0.002611 + r25b&@xmath94 5 & 285 & & @xmath5 3.9 & 2.2 & & 20.2 - 21.0 & 20.4 & 0.0001641 + r26a&@xmath94 40 & 265 & & 2.6 - 3.4 & 2.8 & & 20.5 - 21.1 & 20.8 & 0.0002885 + r26b&10 - 100 & 50 & & @xmath5 3.3 & 2.2 & & 20.6 - 21.4 & 20.9 & 0.00003488 + r27a&@xmath94 5 & 255 & & @xmath94 2.9 & 3.0 & & @xmath5 21.6 & 20.6 & 0.0004384 + r27b&@xmath94 50 & 160 & & @xmath5 2.6 & 2.0 & & 21.0 - 21.7 & 21.3 & 0.001546 + r28a&@xmath94 5 & 110 & & @xmath5 3.9 & 2.5 & & 20.6 21.6 & 21.0 & 0.00007702 + r31a&@xmath94 20 & 115 & & @xmath5 3.0 & 2.0 & & 20.8 - 22.0 & 21.3 & 0.001847 + r34&@xmath94 5 & 130 & & @xmath5 3.4 & 2.4 & & 20.5 - 21.6 & 20.9 & 0.0006237 + r36a&10 - 40 & 30 & & @xmath5 2.5 & 2.0 & & @xmath94 21.1 & 21.6 & 0.00002201 + r37a&5 - 30 & 25 & & @xmath5 2.7 & 2.0 & & 20.7 - 21.8 & 21.1 & 0.03561 + r38&@xmath94 5 & 35 & & @xmath5 3.7 & 2.5 & & @xmath94 20.8 & 21.2 & 0.0005419 + r39a@xmath167&@xmath94 50 & 95 & & 2.7 - 3.4 & 3.1 & & 20.9 - 21.3 & 21.1 & 0.006462 + r39b@xmath167&@xmath94 5 & 175 & & 2.3 - 4.1 & 2.6 & & 20.0 - 20.9 & 20.5 & 0.005283 + lrrrrrrrrrrrl e0@xmath165&21.2 & @xmath94 100 & 270 & & 5.0 - 5.4 & 5.3 & & @xmath94 5 & 8 & 0.00563 + e1@xmath166&21.6 & 40 - 100 & 70 & & 5.6 - 5.9 & 5.8 & & @xmath5 4 & 2 & 0.01073 + e2@xmath167&21.5 & 5 - 90 & 40 & & 5.0 - 5.6 & 5.1 & & 1 - 9 & 2 & 0.0003303 + lrrrrrrrrrcrl r5&@xmath5 0.24 & 0.10 & 15.3 @xmath6 0.9&3.59 & 0.14 + r7b&@xmath5 0.09 & 0.09 & 8.3 @xmath6 0.8&3.97 & 0.02 + r8b&0.03 - 0.19 & 0.10 & 13.0 @xmath6 0.9&25.28 & 1.57 + r9b&@xmath5 0.52&0.10 & 29.5 @xmath6 1.5&1.27 & 0.02 + r10a&0.01 - 0.16 & 0.10 & 11.4 @xmath6 0.8&6.72 & 0.35 + r10b&0.14 - 0.56 & 0.14 & 23.5 @xmath6 1.1&1.24 & 0.03 + r11a&@xmath5 0.11 & 0.10 & 24.6 @xmath6 1.3&3.88 & 0.19 + r11b&&0.10 & 36.9 @xmath6 1.8&0.01 & 0.01 + r13a&0.01 - 0.22 & 0.10 & 14.6 @xmath6 0.9&6.87 & 0.37 + r14&0.01 - 0.32 & 0.10 & 19.6 @xmath6 1.1&5.89 & 0.35 + r15&0.01 - 0.35 & 0.10 & 21.2 @xmath6 1.1&6.44 & 0.27 + r16&0.05 - 0.28 & 0.10 & 17.2 @xmath6 1.0&9.00 & 0.31 + r17&0.01 - 0.33 & 0.10 & 19.9 @xmath6 1.1&2.59 & 0.06 + r18@xmath165&@xmath5 0.10 & 0.10 & 38.7 @xmath6 1.9&1.34 & 0.03 + r19a&@xmath5 0.04 & 0.04 & 28.7 @xmath6 2.2&6.99 & 0.76 + r19b&@xmath5 0.10 & 0.10 & 23.4 @xmath6 1.2&1.68 & 0.05 + r20a&@xmath5 0.11 & 0.11 & 36.8 @xmath6 1.8&0.57 & 0.01 + r21a@xmath166&0.06 - 0.22 & 0.10 & 58.8 @xmath6 2.9&2.75 & 0.03 + r22a&@xmath5 0.14 & 0.10 & 39.5 @xmath6 1.9&0.90 & 0.01 + r22b&@xmath5 0.04 & 0.04 & 99.3 @xmath6 5.1&0.42 & 0.01 + r23a&@xmath5 0.10 & 0.10 & 45.2 @xmath6 2.2&0.61 & 0.01 + r23b&@xmath5 0.10 & 0.10 & 37.4 @xmath6 1.8&0.87 & 0.01 + r24a&@xmath5 0.25 & 0.10 & 15.7 @xmath6 0.9&4.00 & 0.21 + r24b&@xmath5 0.11 & 0.10 & 19.5 @xmath6 1.1&1.02 & 0.01 + r24c&@xmath5 0.36 & 0.10 & 21.3 @xmath6 1.1&1.83 & 0.05 + r25a&0.12 - 0.98 & 0.12 & 44.3 @xmath6 2.1&2.85 & 0.03 + r25b&@xmath5 0.35 & 0.10 & 21.2 @xmath6 1.1&2.16 & 0.07 + r26a&@xmath5 0.11 & 0.10 & 48.4 @xmath6 2.4&1.56 & 0.02 + r26b&0.01 - 0.11 & 0.10 & 14.1 @xmath6 0.9&6.33 & 0.30 + r27a&@xmath5 0.39 & 0.10 & 23.1 @xmath6 1.2&0.09 & 0.01 + r27b&@xmath5 0.03 & 0.03 & 54.5 @xmath6 3.5&9.17 & 0.92 + r28a&@xmath5 0.15 & 0.10 & 11.0 @xmath6 0.8&4.21 & 0.31 + r31a&@xmath5 0.04 & 0.04 & 21.5 @xmath6 2.0&4.61 & 0.09 + r34&@xmath5 0.17 & 0.10 & 11.7 @xmath6 0.8&4.23 & 0.29 + r36a&0.02 - 0.07 & 0.07 & 12.7 @xmath6
1.1&20.17 & 1.53 + r37a&0.03 - 0.16 & 0.10 & 11.3 @xmath6 0.8&19.92 & 1.04 + r38&@xmath5 0.08 & 0.08 & 8.0 @xmath6 0.9&10.07 & 0.67 + r39a@xmath167&@xmath5 0.10 & 0.10 & 52.6 @xmath6 2.6&1.42 & 0.01 + r39b@xmath167&@xmath5 0.39 & 0.10 & 23.1 @xmath6 1.2&1.29 & 0.02 + llrrrrrrrrcrl cs & r18@xmath165 & @xmath5 0.04 & & @xmath5 4.0 @xmath8 10@xmath32 & @xmath5 6.3 @xmath8 10@xmath168 + & r21a@xmath166 & @xmath5 0.04 & & @xmath5 5.6 @xmath8 10@xmath32 & @xmath5 4.5 @xmath8 10@xmath168 + & r39a@xmath167 & 0.13 @xmath6 0.04 & 44 @xmath6 11 & 1.6 @xmath8 10@xmath1 & 1.3 @xmath8 10@xmath90 + ch@xmath0oh & r18@xmath165 & @xmath5 0.04 & & @xmath5 1.9 @xmath8 10@xmath1 & @xmath5 2.9 @xmath8 10@xmath90 + & r21a@xmath166 & @xmath5 0.04 & & @xmath5 2.6 @xmath8 10@xmath1 & @xmath5 2.1 @xmath8 10@xmath90 + & r39a@xmath167 & 0.17 @xmath6 0.04 & 43 @xmath6 5 & 1.0 @xmath8 10@xmath2 & 8.1 @xmath8 10@xmath90 + cn & r18@xmath165 & 0.23 @xmath6 0.04 & 63 @xmath6 5 & 2.4 @xmath8 10@xmath1 & 3.8 @xmath8 10@xmath90 + & r21a@xmath166 & 0.15 @xmath6 0.04 & 66 @xmath6 3 & 2.2 @xmath8 10@xmath1 & 1.8 @xmath8 10@xmath90 + & r39a@xmath167 & 0.14 @xmath6 0.04 & 49 @xmath6 9 & 1.4 @xmath8 10@xmath1 & 1.1 @xmath8 10@xmath90 | we present alma cycle 0 observations of the molecular gas and dust in the ir - bright mid - stage merger vv 114 obtained at 160 800 pc resolution .
the main aim of this study is to investigate the distribution and kinematics of the cold / warm gas and to quantify the spatial variation of the excitation conditions across the two merging disks .
the data contain 10 molecular lines , including the first detection of extranuclear ch@xmath0oh emission in interacting galaxies , as well as continuum emission .
we map the @xmath1co ( 32)/@xmath1co ( 10 ) and the @xmath1co ( 10)/@xmath2co ( 10 ) line ratio at 800 pc resolution ( in the units of k km s@xmath3 ) , and find that these ratios vary from 0.2 0.8 and 5 50 , respectively .
conversely , the 200 pc resolution hcn ( 43)/hco@xmath4 ( 43 ) line ratio shows low values ( @xmath5 0.5 ) at a filament across the disks except for the unresolved eastern nucleus which is three times higher ( 1.34 @xmath6 0.09 ) .
we conclude from our observations and a radiative transfer analysis that the molecular gas in the vv 114 system consists of five components with different physical and chemical conditions ; i.e. , 1 ) dust - enshrouded nuclear starbursts and/or agn , 2 ) wide - spread star forming dense gas , 3 ) merger - induced shocked gas , 4 ) quiescent tenuous gas arms without star formation , 5 ) h@xmath7 gas mass of ( 3.8 @xmath6 0.7 ) @xmath8 10@xmath9 m@xmath10 ( assuming a conversion factor of @xmath11 = @xmath12 ) at the tip of the southern tidal arm , as a potential site of tidal dwarf galaxy formation . |
leptonic decay @xmath4 is described by the annihilation of the initial quark - antiquark pair into a virtual @xmath5 that materializes as a pair ( fig .
[ diagram ] ) . .
, title="fig:",scaledwidth=40.0% ] [ diagram ] the decay rate is give by @xmath6 where @xmath7 is decay constant @xcite , related to the overlap of the heavy and light quark wave - function at zero spatial separation , @xmath8 is the fermi constant , @xmath9 is the @xmath10 mass , @xmath11 is the final state charged - lepton mass , and @xmath12 is a ckm matrix element , taken as @xmath13 and @xmath14 .
thus , measurement of purely leptonic decays allow a determination of the decay constant @xmath7 .
meson decay constants in the @xmath15 system are used to translate measurements of @xmath16 mixing to ckm matrix elements .
currently , it is not possible to determine @xmath17 accurately from leptonic @xmath15 decays , so theoretical calculations of @xmath17 are used . since the @xmath18 meson does not have @xmath19 decays
, it will never be possible to determine @xmath20 experimentally , thus theory must be relied on . if calculations disagree on @xmath21 mesons , they may be questionable on @xmath15 mesons . if , on the other hand new physics is present , we need understand how it effects sm based predictions of the @xmath15 decay constants .
decay constants can be calculated using lattice quantum - chromodynamics ( lqcd ) .
recent calculation from follana _ et al .
_ using an unquenched lqcd predicts @xmath22 mev and @xmath23 mev @xcite .
we use the reactions @xmath24 , and @xmath25 or @xmath26 .
the @xmath27 is studied at 3770 mev using 818 pb@xmath28 of data @xcite . and the @xmath3
is studied at 4170 mev , using 424 pb@xmath28 for @xmath29 and @xmath30 , @xmath31 , and 300 pb@xmath28 for @xmath30 , @xmath32 .
( eventually cleo will present results using 600 pb@xmath28 . ) [ [ section ] ] we fully reconstruct @xmath33 @xcite as a tag and examine the properties of the other @xmath2 , which can be found even if there is a missing neutrino in the final state .
this method is called as double tag " technique . to reconstruct @xmath33 tags we require that the tag candidates have a measured energy consistent with the beam energy , and have a
beam constrained mass " , @xmath34 , consistent with the @xmath33 nominal mass @xcite , where @xmath35 , @xmath36 is the beam energy and @xmath37 runs over all the final state particles three - momenta .
[ mbc_all ] shows the @xmath38 distribution summed over all the decay modes we use for tagging . selecting events in the mass peak we count 460,055@xmath39787 signal events over a background of 89,472 events .
decay candidates in the final states : @xmath40 , @xmath41 , @xmath42 , @xmath43 , @xmath44 and @xmath45.,scaledwidth=45.0% ] we then search for signal events with one and only one additional track with opposite sign of charge to the tag , not identified as kaon .
the track must make an angle @xmath46 with respect to the beam line ( 90% of the solid angle ) , and in addition we require that there not be any photon detected in the calorimeter with energy greater than 250 mev .
the latter selection can highly suppress @xmath47 background .
we separate these events into two cases , where case ( i ) refers to muon candidates depositing @xmath48 mev , characteristic of 98.8% of muons and case ( ii ) is for candidates depositing @xmath49 mev , characteristic of 45% of the pions , those that happen to interact in the calorimeter and deposit significant energy .
we look for @xmath50 by computing the square of the missing mass @xmath51 where @xmath52 is the three - momentum of the fully reconstructed @xmath33 , and @xmath53 is the energy ( momentum ) of the @xmath54 candidate .
the signal peaks at zero for @xmath55 and is smeared toward more positive values for @xmath56 , @xmath57 .
the fit to the case ( i ) mm@xmath58 distribution shown in fig .
[ case1-taunufix ] contains separate shapes for signal , @xmath59 , @xmath60 , @xmath56 ( @xmath61 , and a background shape describing three - body decays .
here we assume the sm ratio of 2.65 for the ratio of the @xmath62 component and constrain the area ratio of these components to the product of 2.65 with @xmath63)=([email protected])% @xcite and the 55% probability that the pion deposits @xmath64300 mev in the calorimeter .
the @xmath59 background are fixed at 9.2 events obtained from monte carlo ( mc ) simulation .
the normalizations of the signal , @xmath60 , and 3-body background are allowed to float .
the @xmath60 shape is obtained from double tag events of @xmath65 where we ignore one kaon to calculate the mm@xmath58 .
all other shapes are obtained from the mc simulation . for case ( i ) .
the points with error bars show the data .
the black ( dashed ) curve centered at zero shows the signal @xmath55 events . the dot - dashed ( red ) curve that peaks around 0.05
gev@xmath58 shows the @xmath66 , @xmath67 component .
the solid ( blue ) gaussian shaped curve centered on the pion - mass squared shows the residual @xmath59 component .
the dashed ( purple ) curve that falls to zero around 0.03 gev@xmath58 is the sum of all the other background components , except the @xmath60 tail which is shown by the long - dashed ( green ) curve that peaks up at 0.25 gev@xmath58 .
the solid ( black ) curve is the sum of all the components.,scaledwidth=50.0% ] the fit yields [email protected] @xmath55 signal events and 25.8 @xmath56 , @xmath67 events ( for the entire mm@xmath58 range ) .
we also perform the fit allowing the @xmath56 , @xmath67 component to float .
then we find [email protected] @xmath55 events and [email protected] @xmath56 , @xmath67 events , compared with the 25.8 we expect in the sm . performing the fit in this manner
gives a result that is independent of the sm expectation of the @xmath66 rate .
to extract a branching fraction , in either case , we subtract off [email protected] events found from simulations and other studies to be additional backgrounds , not taken into account by the fit .
we find @xmath68 the decay constant @xmath69 is then obtained from eq .
( [ eq : equ_rate ] ) using 1040@xmath397 fs as the @xmath27 lifetime @xcite and 0.2256 as @xmath70 .
our final result , including radiative corrections is @xmath71 [ [ section-1 ] ] to reconstruct the tag , the difference here than the @xmath27 case is that we need to include an additional photon from @xmath72 in the tag .
the @xmath73 tags we reconstructed can either from directly produced @xmath73 mesons or those that result from the decay of @xmath72 mesons .
we calculate the missing mass squared mm@xmath74 recoiling against the photon and the @xmath73 tag , @xmath75 here @xmath76 ( @xmath77 ) is the center - of - mass energy ( momentum ) , @xmath78 ( @xmath79 ) is the energy ( momentum ) of the fully reconstructed @xmath73 tag , and @xmath80 ( @xmath81 ) is the energy ( momentum ) of the additional photon .
we determine number of tags by simultaneously fit to the invariant mass ( @xmath82 ) and mm@xmath74 , shown in fig .
the signal is fit to a sum of two gaussians for the @xmath82 and a crystal ball function for the mm@xmath74 .
the tail parameters of the crystal ball function are obtained from fully reconstructed @xmath83 events .
the background has two components : either comes from the background under the invariant mass peak ( fake @xmath84 ) , or is due to random photon combinations .
the former background is modeled by linear function for the mass and 5th order chebyshev polynomial for the mm@xmath74 .
since the latter background is from a true @xmath84 , its mass distribution has the same shape as that from the signal , while its mm@xmath74 is modeled by another 5th order chebyshev polynomial function .
the total number of single tags is @xmath85 in the invariant mass signal region ( @xmath86 mev from the nominal @xmath84 mass ) and mm@xmath87[3.782 , 4.0 ] gev@xmath58 .
[ 2d ] $ ] gev@xmath58 and mm@xmath74 ( right ) in the invariant mass @xmath88 mev@xmath89 mev in the final states : @xmath90 , @xmath91 , @xmath92 ; @xmath93 , @xmath94 ; @xmath95 , @xmath93 , @xmath96 ; @xmath97 , @xmath98 , @xmath99 , @xmath100 ; @xmath101 , @xmath102 , @xmath103 ; @xmath93 , @xmath98 , and @xmath94 ; @xmath104 .
data ( point ) and fit function ( red curve ) are shown .
the dashed ( blue ) curve corresponds to the fake @xmath84 background , and the dashed ( green ) curve to the random photon background.,title="fig:",scaledwidth=40.0%]$ ] gev@xmath58 and mm@xmath74 ( right ) in the invariant mass @xmath88 mev@xmath89 mev in the final states : @xmath90 , @xmath91 , @xmath92 ; @xmath93 , @xmath94 ; @xmath95 , @xmath93 , @xmath96 ; @xmath97 , @xmath98 , @xmath99 , @xmath100 ; @xmath101 , @xmath102 , @xmath103 ; @xmath93 , @xmath98 , and @xmath94 ; @xmath104 .
data ( point ) and fit function ( red curve ) are shown .
the dashed ( blue ) curve corresponds to the fake @xmath84 background , and the dashed ( green ) curve to the random photon background.,title="fig:",scaledwidth=40.0% ] similarly as the @xmath27 , we reconstruct the signal side .
we veto events with an extra neutral energy cluster @xmath49 mev ( it is 250 mev in the @xmath27 case ) .
it is highly effective in reducing backgrounds , especially for @xmath105 , @xmath106 and the processes @xmath107 .
the missing mass squared , mm@xmath58 , evaluated by taking into account the observed @xmath54 , @xmath73 , and photon should peak at zero ; @xmath108 we also make use of a set of kinematical constraints and fit each event to two hypotheses one of which is that the @xmath73 tag is the daughter of a @xmath109 and the other that the @xmath110 decays into @xmath111 , with the @xmath3 subsequently decaying into @xmath55 .
in addition , we constrain the invariant mass of the @xmath73 tag to the know @xmath84 mass .
this gives us a total of 7 constraints .
the missing neutrino four - vector needs to be determined , so we are left with a three - constraint fit . we perform a standard iterative fit minimizing @xmath112 .
we choose the fitted mm@xmath58 from the hypothesis giving the smaller @xmath112 .
the mm@xmath113 distribution from data is show in fig . [ mms - fitall_38 - 47 ] .
after fixing the ratio of @xmath62 to the sm value we find @xmath114 mev .
the dashed ( grey ) gaussian shaped curve peaked near zero is the @xmath55 component , while the dashed ( purple ) curve that rises sharply from zero and then flattens out shows the @xmath56 component .
the two lines are background components . the solid curve shows the sum . , scaledwidth=45.0% ] we can also use the decay mode @xmath115 .
this result has already been published .
@xcite the technique here is to use only three tagging modes : @xmath116 , @xmath117 and @xmath118 , to ensure that the tags are extremely clean
. then events with an identified @xmath119 and no other charged tracks are selected .
any energy not associated with the tag decay products is tabulated .
those events with small extra energy below 400 mev are mostly pure @xmath120 events . after correcting for efficiencies and residual backgrounds
we find @xmath121 mev .
the preliminary cleo average is @xmath122 mev ( radiatively corrected ) . averaging in the belle result @xcite @xmath123 mev , which differs from the follana _ et al .
_ calculation @xcite by 3.2 standard deviations , while the result for @xmath124 mev is in good agreement .
this discrepancy could be due to physics beyond the standard model @xcite , or systematic uncertainties that are not understood in the lqcd calculation or the experimental measurements , or unlikely statistical fluctuations in the experimental measurements or the lqcd calculation .
fits to the ckm matrix parameters use theoretical predictions of @xmath20/@xmath125 .
as similar calculations are used for @xmath20/@xmath125 , we need to be concerned with them .
i thank the u. s. national science foundation for support .
excellent conversations were had with p. chang , a. kronfeld and s. stone . | using cleo data collected at 3370 mev and 4170 mev , we determine @xmath0 mev and an interim preliminary value of @xmath1 mev , where both results are radiatively corrected .
they agree with the recent most precise unquenched lattice - qcd calculation for the @xmath2 , but do not for the @xmath3 .
several consequences are discussed . |
coherent electron transport through nanostructures in cryogenic two - dimensional electron gas systems is inherently a quantum mechanical phenomenon .
several experiments have demonstrated certain aspects of this quantum behavior , whether it relies directly on the wave nature of the electron and can be probed through a current or conductance measurement ( _ e.g. _ , quantized resistance in the quantum hall effect @xcite , conduction modes of a quantum point contact @xcite ) , or on the particle nature and quantum statistics of the electrons and can be probed through a noise measurement ( _ e.g. _ , hanbury brown and twiss - type experiment @xcite , electron collision @xcite , observation of the fractional charge in the fractional quantum hall effect @xcite ) .
in addition , 2d electron gases could be used to study the fundamental non - local features of quantum mechanics through electron entanglement @xcite .
all these experiments can be successfully explained within the coherent scattering formalism @xcite , a theoretical tool describing coherent and non - interacting electron transport .
this formalism relies on spin - independent stationary states in the leads of the device and , therefore , describes spin - independent transport .
although it is possible to add a local spin - dependent effect directly in the scattering matrix , it is not possible in general to take into account spin effects occurring over the whole system .
one potentially important spin effect occurring in the leads of the conductor is spin - orbit coupling .
any electric field in the reference frame of the laboratory generates a magnetic field in the moving electron reference frame , coupling the electron s orbital degrees of freedom with its spin .
one can find several sources of electric fields in semiconductors .
in three dimensional crystals , the periodic crystal potential generates the dresselhaus effect @xcite , which induces a spin - splitting of the conduction band that is proportional to @xmath0 . in two - dimensional systems , the dominant term results from the asymmetry of the in - plane confining potential .
first introduced by rashba @xcite , this effect causes a spin - splitting @xcite proportional to @xmath1 , and it depends on the strength of the applied electric field .
recently , there has been a growing interest in electronic devices which rely on the spin properties of the electrons .
these spin - dependent devices may be influenced by spin - orbit coupling effects , or may even rely on it as , for example , in a coherent version of a spin - polarized field effect transistor @xcite .
therefore , it may be useful to include spin - orbit coupling in the coherent scattering formalism .
the spin - orbit coupling is governed by the following spin - orbit hamiltonian @xcite , which is obtained using an expansion in @xmath2 of the dirac equation @xmath3 where @xmath4 is the free electron mass , @xmath5 is the momentum operator , @xmath6 are the pauli spin matrices , @xmath7 is the electrostatic potential , and @xmath8 is the gradient operator so that @xmath9 is the electric field .
electron transport in the presence of an electric field results in a spin - orbit effect which couples the electron spin and orbital degrees of freedom through the @xmath10 term . here
, we will neglect the bulk effect arising from the periodic crystal potential ( dresselhaus effect ) and consider only the effect caused by the asymmetry of the confining quantum well ( rashba effect ) , as it is stronger than the dresselhaus effect in most two - dimensional heterostructures @xcite .
let z be the direction of confinement , perpendicular to the plane of motion .
the asymmetry of the confining potential along the z - direction results in a non - zero electric field along the z - axis ( @xmath11 throughout the article , where @xmath12 represents a unit vector in the z - direction ) .
the spin - orbit coupling term of the hamiltonian can be written @xmath13 where @xmath14 is the spin - orbit coupling constant and depends on the strength of the electric field .
it takes values in the range @xmath15 to @xmath16 ev cm for a large variety of systems ( inas / gasb @xcite , inas / alsb @xcite , in@xmath17ga@xmath18as / in@xmath17al@xmath18as @xcite and gaas / al@xmath17ga@xmath18as @xcite ) depending on the shape of the confining well . using the standard effective mass approximation
, we can deduce the system hamiltonian as the free hamiltonian plus the spin - orbit coupling hamiltonian @xmath19 where m is always taken to be the effective mass .
since the operators @xmath20 and @xmath21 commute with @xmath22 , we can search for eigenstates of the form @xmath23 , \ ] ] where @xmath24 and @xmath25 label the up and down states of the z - component of the spin .
we can now diagonalize the hamiltonian @xmath26 in this spin subspace .
the eigenvalues are @xmath27 , and the associated eigenfunctions are @xmath28 \label{eq1 } \\ \left| \psi _ { e_{- } } \right > & = & \frac{e^{i ( k_{x}x + k_{y}y)}}{\sqrt{2 } } [ e^{i \frac{\theta}{2 } } \left| \uparrow \right > - \;e^{-i \frac{\theta}{2 } } \left| \downarrow \right > ] \label{eq2}\end{aligned}\ ] ] where @xmath29 is the angle between @xmath30 and the y - axis ( see fig .
[ fig1 ] ) .
the electrons feel a virtual magnetic field in the 2d plane in a direction perpendicular to @xmath31 . the spin is aligned or antialigned to this field ( see fig .
[ fig1 ] ) , so that @xmath32 , and @xmath33 , where @xmath34 is the angle between @xmath31 and the spin @xmath35 in the state @xmath36 .
the amplitude of the magnetic field depends on the velocity of the electron and vanishes for @xmath37 , preventing a possible spin polarization in the system .
the spin - orbit splitting is usually small compared to the kinetic energy of the electrons ( @xmath38 mev @xmath39 mev for @xmath40 mev ) .
following ref . , we will introduce a transverse confinement in the leads of the conductor , allowing us to address the longitudinal transport modes for each given transverse mode . focusing on one lead , @xmath41
, we will make two simplifying assumptions .
first , we consider only a single independent transverse mode .
second , we neglect the 1d so coupling effect that this transverse confining potential could create , since , to our knowledge , there is no experimental verification of this effect , and it is estimated to be much smaller than the rashba effect @xcite . within these approximations , we can use our previous analysis to deduce the eigenstates and the associated energy dispersion diagram ( see fig . [ fig2 ] ) , with @xmath1 lying in the direction @xmath42 of the lead ( making the angle @xmath43 with the y axis ) .
we will now introduce three labels for the eigenstates that will prove useful in writing the creation and annihilation operators of these states .
@xmath44 or @xmath45 labels the direction of propagation from the sign of the group velocity @xmath46 ( @xmath47 if @xmath48 , b otherwise ) .
the parameter @xmath1 labels the longitudinal mode wavevector .
the so coupling label @xmath49 designates the two different branches of the energy dispersion diagram for a given k ( @xmath50 branch and @xmath51 branch in fig .
[ fig2 ] ) . using these labels , we find the following eigenstates and eigenvalues of the system from eq .
( [ eq1 ] ) and ( [ eq2 ] ) @xmath52 \label{eq3 } \nonumber \\ * e & = & \frac{\hbar k^{2}}{2 m } + \eta_{\sigma } \alpha k \;\;\;\;\ ; -\eta_{\sigma } k_{0 } \leq k < \infty\end{aligned}\ ] ] where @xmath53 when @xmath54 , and @xmath55 when @xmath56 .
@xmath57 is the normalized transverse wavefunction for the transverse mode under consideration . by convention , @xmath1
is only taken in the range @xmath58 for both the @xmath47 and @xmath45 states .
we use @xmath59 to parameterize explicitly the appropriate wavevector range for the @xmath54 propagation direction in the eigenstates phase factor @xmath60 .
the eigenvalues in fig .
[ fig2 ] can be deduced by mirror symmetry .
this convention allows us to track the propagation direction throughout the calculation .
we also notice in fig .
[ fig2 ] that , for a given energy , the corresponding @xmath50 and @xmath51 states with same label @xmath61 do not have the same wavevector : @xmath62 the states with @xmath63 have their spin perpendicular to the direction of propagation @xmath64 and in opposite directions , so that @xmath65 and @xmath66 ( see fig . [ fig3 ] ) .
it is important to notice that the @xmath67 states do not coincide with the @xmath68 states , as it is @xmath64 , and not @xmath31 , which determines the direction of the spin ( _ e.g. _ , in fig .
[ fig3 ] , the spin of the @xmath50 state is perpendicular to and on the right side of @xmath64 ) .
for example , in fig .
[ fig2 ] , the two definitions are consistent for states with positive ( negative ) @xmath69 and group velocity @xmath46 .
however , they are inconsistent for positive ( negative ) @xmath69 and negative ( positive ) @xmath46 .
we choose here to link the direction of the spin to @xmath64 ( rather than @xmath70 ) through the label @xmath71 ( rather than @xmath72 ) , because it is the group velocity which determines the direction of the current .
note that in the case without so coupling , no differentiation between @xmath46 and @xmath69 is necessary since they always share the same sign .
the landauer - bttiker formalism relies on a second quantization formulation of quantum mechanics .
the current is expressed as a function of the field operator , which is expanded in the basis of the scattering states using the operators that create and destroy electrons in the leads of the conductor .
a scattering state is a coherent sum of an incident wave in one lead and the outgoing waves it generates in all the other leads .
the amplitude of each outgoing wave is given by the scattering matrix @xmath73 , whose elements depend on the properties of the scatterer . as our final goal is to study the fluctuations of the current , it is important to find a simple way to express the time dependence of the current operator .
this leads us to choose the stationary states in each lead for the purpose of expanding the field operator . in our rederivation of the coherent scattering formalism
, we will follow the same analysis as in ref . , but
introducing two important changes : we will rederive the expression of the current operator as a function of the field operator , adding a new contribution coming from spin - orbit coupling , and we will expand the field operator in the new stationary basis of the so coupling hamiltonian . one way to find
the current operator for non - interacting electrons is to use the single - particle schrdinger equation @xmath74 where @xmath75 is a two - components spinor .
the current operator @xmath76 can be extracted from the conservation of charge equation @xmath77 + \bm{\nabla}.\mathbf{j } = 0 $ ] ; using eq .
( [ eq5 ] ) and its adjoint , @xmath78 - \frac{e \alpha}{\hbar}\ ; \psi^{\dag } \sigma_{y } \psi \\ j_{y } & = & \frac{e \hbar}{2 m i } \big[\ ; \psi^{\dag } \frac{\partial \psi}{\partial y } - \frac{\partial \psi ^{\dag}}{\partial y } \psi \;\big ] + \frac{e \alpha}{\hbar } \ ; \psi^{\dag } \sigma_{x } \psi\end{aligned}\ ] ] we can identify the usual kinetic term of the current density @xmath79 . \label{eq6}\ ] ] note that so coupling adds a new contribution proportional to @xmath14 that we call the so coupling current density @xmath80 , @xmath81 in the framework of second quantization , @xmath75 becomes a field operator , and we have to expand it using a convenient basis : the stationary states .
the first step is to define the creation ( annihilation ) operators for the so coupling stationary states .
we first introduce the creation operators in the old spin basis : @xmath82 ( @xmath83 ) creates an incoming ( outgoing ) electron with spin up ( down ) in the first mode of lead @xmath41 with momentum @xmath1 , in the state @xmath84 ( @xmath85 ) , where @xmath86 .
these standard operators satisfy the anticommutation relation @xmath87_{+ } = \delta_{k k ' } \delta_{\gamma \beta } \delta_{s s'}$ ] , where @xmath88 labels the spin of the electron . through eq .
( [ eq3 ] ) we introduce @xmath89 , which creates an incoming electron in lead @xmath41 with momentum @xmath1 and spin - orbit coupling label @xmath90 ( not to be confused with the spin @xmath88 ) satisfying the relation @xmath91 .\ ] ] knowing the anticommutation relationships in the spin basis , we can calculate it in our new stationary basis to be @xmath92_{+ } = \delta_{k k ' } \delta_{\alpha \beta } \delta_{\sigma \sigma ' } , \ ] ] where we used the relation @xmath93 .
this result justifies our labeling of the so coupling eigenstates and the use of the spin - orbit coupling label @xmath90 instead of the spin .
the stationary states form a complete basis , and we can use them to expand the field operator in lead @xmath41 . from now on , we will choose @xmath42 and @xmath94 as the reference axis ( so that @xmath95 ) , specifically tracking all of the angular dependence in the scattering matrix .
@xmath96 where @xmath97 is the annihilation operator in the incoming(outgoing ) states and @xmath98 is a two - component spinor obtained from eq .
( [ eq3 ] ) , setting @xmath95 @xmath99 } _ { \chi^{\epsilon}_{\sigma } } .
\label{eq8}\ ] ] as shown in fig .
[ fig4 ] , its spin depends on the so coupling state ( @xmath90 ) and on the direction of propagation ( @xmath47 or @xmath45 ) . at this point
, it is more convenient to go from a discrete sum in @xmath1-space to a continuous integral in energy by defining the operators @xmath100 , which create electrons in the energy continuum .
as our labeling defines a one - to - one correspondence between ( @xmath101 ) and ( @xmath102 ) , we can define unambiguously @xmath103 , so that @xmath104_{+ } = \delta _ { \gamma \beta } \delta_{\sigma \sigma ' } \delta ( e - e')\ ] ] where @xmath105 is the density of states at energy e .
_ a priori _ , @xmath105 could depend on the spin - orbit coupling label @xmath90 , but one can check in appendix [ apa ] that it is not the case . replacing the discrete sum over @xmath1 by an integral over @xmath106 and introducing the new operators and their time - dependence @xmath107 , we find @xmath108 with @xmath109 , and @xmath1 has been replaced by @xmath110 to remind us that , for a given energy , @xmath1 depends on the spin - orbit coupling label @xmath90 .
the current in lead @xmath41 is given by @xmath111 we will begin by calculating the usual kinetic term @xmath112 . using eq .
( [ eq6 ] ) , we have @xmath113 @xmath114\;e^{i\;(\eta_{\epsilon ' } k(e',\sigma ' ) - \eta_{\epsilon } k(e,\sigma ) ) x } .\ ] ] we notice from eq .
( [ eq9 ] ) that the @xmath115 frequency of the current is given by @xmath116 .
following ref . , we calculate the current and noise in the zero - frequency limit , and make the approximation @xmath117 ( as @xmath118 ) for electrons having the same so coupling label , and , using eq .
( [ eq4 ] ) , @xmath119 for electrons having different so coupling labels . from eq .
( [ eq8 ] ) , we also have , ( with @xmath120 if @xmath121 otherwise ) @xmath122 this result is consistent with the fact that only electrons with the same direction of propagation and the same spin - orbit coupling label , or opposite direction of propagation and opposite spin - orbit coupling label , have the same spin .
+ in the @xmath123 case , implying @xmath124 and @xmath125 , we find @xmath126 in the @xmath127 case , implying @xmath128 and @xmath129 , so that @xmath130 , we find @xmath131 if we only consider the kinetic term of the current , a non - vanishing contribution for @xmath132 leads to terms like @xmath133 and @xmath134 in the expression of the current , corresponding to electrons propagating in opposite directions .
now , let us study the contribution of the spin - orbit coupling current @xmath135 .
using eq .
( [ eq7 ] ) to calculate @xmath136 in eq .
( [ eq9 ] ) with @xmath137 , we find @xmath138 using @xmath139 , we have from eq . ( [ eq8 ] ) @xmath140 and @xmath141 , so that @xmath142 .
\label{eq12}\ ] ] we can now calculate the value of the total current from eq .
( [ eq10 ] ) , ( [ eq11 ] ) and ( [ eq12 ] ) @xmath143 although the spin - orbit coupling mixes electrons having different directions of propagation when we consider only the kinetic term of the current , these contributions cancel when we add the spin - orbit coupling current @xmath144 , and we find the standard formula for the current operator , @xmath145\ ; e^{i \frac{e - e'}{\hbar}t } .\end{aligned}\ ] ] in this formula , the definition of the @xmath146(@xmath147 ) states as states with positive(negative ) group velocity , and not necessarily positive(negative ) wavevector , is consistent with the fact that they carry the current in opposite directions .
more importantly , the final expression of the current is similar to the one found in ref . , but with the spin index replaced by the spin - orbit coupling label @xmath148 .
the spin related to this new index depends on the direction of propagation , that is , the angle of the lead @xmath43 .
therefore , we expect a @xmath43 dependence in the scattering matrix relating the spin - orbit coupling states in leads with different orientations . as an example , we investigate the scattering matrix in the case of a four - port beamsplitter ( two input leads , two output leads ) used in electron collisions .
the scattering matrix relates the outgoing states in the outputs ( @xmath149 states ) to the incident states at the input ( @xmath146 states ) @xmath150 in the spin - independent transport case , as the beamsplitter does not act on the spin degrees of freedom , the scattering matrix does not mix different spins ( the scattering matrix is diagonal in each subspace of spin : spin - up and spin - down ) . in this case
the reflection and transmission coefficients do not depend on the spin .
however , when we include spin - orbit coupling , the situation is more complicated , because the spin associated with the spin - orbit coupling label @xmath90 is different in each lead ( the leads having different orientations ) .
the conservation of the spin at the beamsplitter then implies that we have a mixing of the spin - orbit states ( off - diagonal elements in the scattering matrix ) , and this mixing becomes more important when the angle between the leads increases . a beamsplitter , with all leads oriented at the same angle @xmath29 ( the case illustrated in fig .
[ fig5 ] ) is investigated in appendix [ apb ] , and the following scattering matrix is found after some simplifying assumptions @xmath151 we note that the unitarity of the @xmath73 matrix requires @xmath152 and @xmath153 .
we will now use the expression of the current operator and the scattering matrix to calculate the current and noise in two experiments using the four - port beamsplitter setup .
the first is an unpolarized electron collision , which has already been realized experimentally ( see ref . ) .
the second is a spin - polarized electron collision .
this experiment is a collision at a 50/50 beamsplitter of electrons from leads 1 and 2 which are biased at @xmath154 at zero temperature and without spin - orbit coupling , each electron with a specific spin in lead 1 has an identical partner in lead 2 .
the pauli exclusion principle then forbids us to find these two electrons in the same output .
we expect a total suppression of the partition noise . with spin - orbit coupling
, we do not expect this result to be modified .
it relies only on the full occupation of the energy eigenstates , which is not modified at zero temperature by so coupling .
we will confirm this by studying the collision of four electrons at the same energy .
the initial state is : @xmath155 from now on , we will drop the @xmath106 , @xmath1 and @xmath69 labels for the energy and the momentum of the electrons , associating a @xmath50(@xmath51 ) so coupling state with @xmath1(@xmath69 ) .
the output state is found using the new scattering matrix with so coupling using eq .
( [ eq13 ] ) and eq .
( [ eq14 ] ) @xmath156 + t\;\hat{b}^{\dag}_{+ 4}\right]\;\left[r\ ; [ \;i \sin\theta \ ; \hat{b}^{\dag}_{+ 3 } + \cos\theta \ ; \hat{b}^ { \dag}_{- 3}\ ; ] + t\;\hat{b}^{\dag}_{- 4}\right]\nonumber \\ * & & \times\left[t\;\hat{b}^{\dag}_{+ 3 } + r\;[\;\cos\theta\ ; \hat{b}^{\dag}_{+ 4 } - i\sin\theta \;\hat{b}^{\dag}_{- 4 } \;]\right]\ ; \left[t\;\hat{b}^{\dag}_{- 3 } + r\;[\;- i \sin\theta \ ; \hat{b}^{\dag}_{+ 4 } + \cos\theta \;\hat{b}^{\dag}_{- 4}\;]\right ] \left| 0\right > \nonumber \\ * & = & \hat{b}^{\dag}_{+ 3}\hat{b}^{\dag}_{- 3}\hat{b}^{\dag}_{+ 4}\hat{b}^{\dag}_{- 4 } \left| 0\right>\end{aligned}\ ] ] where we have used the relation @xmath157 .
as expected , a full occupation of the two input states at energy e leads to a full occupation of the two output states as well .
a calculation of the noise for the fully occupied outputs would show a complete suppression of the partition noise . the only difference with the spin - independent transport case is that the colliding electrons at the same energy do not have the same momentum ( @xmath158 ) .
therefore , the non - ideality in the noise suppression observed in ref .
can not be caused by the so coupling . in the previous example , an unpolarized collision can not show any so coupling effect , because , starting with two quiet sources of electrons with fully occupied energy levels , all the transport statistics are governed by the pauli exclusion principle independent of the so coupling . however , this is not the case for a spin - polarized collision . as the standard spin basis is not the stationary basis , and as the so coupling scattering matrix mixes different so coupling states ,
some of the partition noise is recovered , depending on the so coupling constant @xmath14 and on the angle between the colliding leads .
we will confirm this by calculating the non - equilibrium noise ( of the electrons above the fermi sea ) between @xmath159 and @xmath160 , assuming that the process of polarization does not effect the transport in the conductor ( for example , there is no magnetic field in the conductor ) .
the input state is then made of all the electrons with the same spin , for example spin - up , between @xmath161 and @xmath162 .
following ref . , we present the initial state and the current operator using the following notations @xmath163 where @xmath164 is given by @xmath165 .
we then use the current fluctuation @xmath166 to calculate the fluctuation correlation function .
@xmath167 although we derived an expression for the current operator in the so coupling basis , it is mathematically more convenient in this case to express all the operators in the standard spin basis @xmath168}{\sqrt{2 d_{\sigma}(k)}}\ ] ] with @xmath169 .
defining @xmath170 , we have , @xmath171 \ ; \nonumber \\ * \!\!\!\!\!\!\!\!\!\!\!\!\ ! & \!\!\!\!\!\ ! & \!\!\!\!\!\ ! \qquad \qquad ... + \;\eta_{\sigma'}\eta_{\sigma '' } \delta _ { \beta \zeta } \delta_{\gamma \delta } \;\delta ( k_{e,\sigma } - k_{e''',\sigma ' '' } ) \delta ( k_{e',\sigma ' } - k_{e'',\sigma '' } ) \ ; n_{\beta \uparrow}(k_{e , \sigma})\ ; \big]\end{aligned}\ ] ] where @xmath172 for @xmath173 , so that @xmath174 $ ] . after integration over @xmath175 and @xmath176 and
making the approximation @xmath177 , we find @xmath178 \ ; + \;\eta_{\sigma'}\eta_{\sigma''}\ ; \delta_{\gamma \delta } \ ; n_{\beta \uparrow}(k_{e , \sigma})\ ; \big ] . \label{eq16}\end{aligned}\ ] ] we can replace @xmath179 by @xmath180 , because @xmath181 , which follows from @xmath182 as the so coupling is small .
we will now calculate the power spectral density at zero frequency in output lead number 3 , @xmath183 , @xmath184 from eq .
( [ eq15 ] ) and eq .
( [ eq16 ] ) , we find @xmath185 \label{eq17}\end{aligned}\ ] ] where @xmath186 1 or 2 , since we ignored scattering from output to output .
@xmath187 is the number of electrons in lead @xmath188 with spin up and momentum @xmath189 @xmath190 assuming that the so coupling is weaker than the bias voltage , that is , @xmath191 , we have @xmath192 \neq 0 $ ] for @xmath193 and @xmath194 \neq 0 $ ] for @xmath195 .
therefore , eq . ( [ eq17 ] ) becomes @xmath196 .\ ] ] given the scattering matrix calculated in eq .
( [ eq14 ] ) , we find @xmath197 and @xmath198 so that @xmath199 after some calculation , we find the current and fano factor to be @xmath200 this result reveals two typical features of so coupling : first , the noise is proportional to the so coupling constant , and second , it depends on the angle between the input leads 1 and 2 .
one can check that the obtained noise is identical to the partition noise for two independent leads where only a fraction ( @xmath201 of the electrons is colliding , and for which the noise is modified by the factor @xmath202 .
this result can also be explained using fig .
[ fig6 ] . in the inputs ,
the spin - up electron states between @xmath161 and @xmath164 are fully occupied .
however , in terms of the so coupling states , only the states between @xmath203 and @xmath204 are jointly occupied , giving no contribution to the noise as in the unpolarized case . between @xmath205 and @xmath206 ,
only the @xmath51 states are filled , and between @xmath204 and @xmath207 , only the @xmath50 states are filled .
these states do not have the same spin in leads 1 and 2 ( their overlap is @xmath208 ) , so that the pauli exclusion principle does not fully suppress the noise , and the classical partition noise is partially recovered through the factor @xmath209 . in this treatment
, we have assumed that the so coupling is small ( @xmath210 ) , so that @xmath211 .
but , a typical value of the applied bias voltage is @xmath212 mev , which is smaller than the typical values of energy splitting caused by the rashba effect ( @xmath213 mev @xmath214 mev ) . in this case , all the electrons contribute to the partition noise , and we have @xmath215 . depending on the angle between the leads , we can go from full noise suppression to the classical limit of the partition noise .
the modification of the noise caused by so coupling in this collision experiment can in principle be measured , even for small values of the so coupling constant .
we have studied the influence of so coupling in the framework of the landauer - bttiker formalism .
a short review of the effect of so coupling in 2degs ( and more precisely of the rashba effect ) has reminded us of its main features : spin - splitting proportional to @xmath1 , and stationary states of the spin perpendicular to the direction of propagation .
the electron feels a magnetic field perpendicular to the direction of propagation , with an amplitude proportional to the velocity .
we have then included the effects of so coupling in the landauer - bttiker coherent scattering formalism .
the so coupling gave rise to two important modifications .
first , the addition of a so coupling term in the hamiltonian modifies the expression of the current operator , and an extra term directly related to so coupling has to be included .
second , the expansion of the current operator is in the basis of the stationary states of so coupling .
the final formula for the current operator was found to be identical to the one derived in the spin - independent transport case , but with the spin replaced by the so coupling label , indicating the alignment of the spin on the virtual magnetic field caused by the so coupling .
the main differences introduced by the so coupling then arise in the calculation of the scattering matrix relating the stationary states of so coupling in different leads with different orientations .
the direction of the virtual magnetic field ( and , consequently , the direction of the spin ) depends on the direction of propagation which is different for each lead in general .
therefore , the scattering matrix is shown to mix states with different so coupling labels , and the strength of this mixing depends on the angle between the leads .
the effect of so coupling on the current noise was then investigated in two examples of electron collision . in the unpolarized electron collision example
, it is shown that the so coupling does not modify the noise ; this case is entirely determined by the pauli exclusion principle . in contrast , the polarized case exhibits a contribution to the noise caused by so coupling , which is proportional to the so coupling constant @xmath14 and depends on the angle between the leads .
a polarized electron collision experiment provides , in principle , another way to measure the strength of the rashba splitting energy .
finally , this new formulation of the current operator can be applied to other coherent scattering experiments in which one wants to investigate or incorporate the effects of so coupling .
the authors gratefully acknowledge useful discussions with e. waks , x. matre and the support of j. f. roch .
we thank the quantum entanglement project ( icorp , jst ) for financial support .
w. d. o. gratefully acknowledges additional support from muri and the ndseg fellowship program .
in this appendix , we calculate the density of states and show that it does not depend on the so coupling label @xmath90 , as is suggested by the symmetry between the @xmath50 and @xmath51 states in the energy dispersion diagram ( see fig .
[ fig2 ] ) , @xmath216 from eq .
( [ eq4 ] ) we deduce @xmath217 which is independent of @xmath90 .
here , we determine the scattering matrix for the four - port beamsplitter described in fig .
the beamsplitter is very simply approximated by a potential barrier at @xmath218 of length @xmath219 .
the plane is then divided into three areas of different potential ( see fig . [ fig7 ] ) in which the solution of the schrdinger equation is known . starting with an incident wave in lead number one , for example
, we can calculate the reflected , refracted , and transmitted waves in lead 3 and 4 , using the continuity of the wavefunction and its derivatives at the beamsplitter interface ( x=0 and x=@xmath219 ) .
for example , let us start with an incident state at energy @xmath106 with momentum @xmath220 in the so coupling state @xmath221
. by the conservation of energy , the reflected wave in lead 3 is a superposition of the states @xmath222 and @xmath223 with @xmath224 the translational invariance of the beamsplitter along the y - axis leads to the conservation of the y - component of the momentum @xmath225 we deduce that @xmath226 , but @xmath227 .
there is dispersion due to the so coupling , leading to an angular separation between the @xmath50 and @xmath51 states after reflection at the beamsplitter ( see fig .
[ fig8 ] ) .
this angular separation is given by @xmath228 \neq \theta_{r + } .\ ] ] starting with an incident state with so coupling label @xmath51 , the angular separation is @xmath229 \neq \theta_{r - } .\ ] ] in analogy with the the total reflection for incident angles below the critical angle in classical optics , we can even have @xmath230 , leading to a suppression of the reflection in the @xmath50 state for a small enough incident angle .
we note that starting with a mixture of @xmath50 and @xmath51 states in the incident beam of electrons , one could suppress the reflection of the @xmath50 state , thus achieving a polarization of the beam .
although interesting , we will neglect this effect of angular dispersion by considering only non - equilibrium electrons above the fermi energy for which @xmath231 ( as so coupling is small compared to the kinetic energy ) , and important incident angles @xmath232 . in this case
the angular separation is very small ( @xmath233 ) .
the equations of continuity of the wavefunction and its derivative are then much easier to solve , and one can find that the incident , refracted , and reflected waves have the same spin at x=0 , and the refracted and transmitted waves have the same spin at x=@xmath219 . using eq.([eq2 ] ) to find the spin overlap between different leads we find , starting with a @xmath50 incident state , @xmath234\ ] ] where @xmath235 is the transmitted wave ( into lead 4 ) and @xmath236 the reflected wave ( into lead 3 ) .
as the transmitted wave has the same direction of propagation as the incident wave , there is no mixing of the so coupling states , and we have only the usual transmission coefficient @xmath237 . for the reflected wave ,
the direction is changed and we have to mix the different so coupling states to obtain the same spin as the incident wave on the interface with the beamsplitter . if we start now with a @xmath51 incident state , we find @xmath238 .\ ] ] the same analysis can be done for an incident state in lead 2 with @xmath239 replaced by @xmath240 .
we then deduce the whole scattering matrix : @xmath241\ ] ] 99 k. von klitzing , g. dorda , and m. pepper , phys .
* 45 * , 494 ( 1980 ) .
b. j. van wees , h. van houten , c. w. j. beenakker , j. g. williamson , l. p. kouwenhoven , d. van der marel , and c. t. foxon , phys .
* 60 * , 848 ( 1988 ) .
d. a. wharam , t. j. thornton , r. newbury , m. pepper , h. ahmed , j. e. f. frost , d. g. hasko , d. c. peacock , d. a. ritchie , and g. a. c. jones , j. phys .
c * 21 * , l209 ( 1988 ) .
m. henny , s. oberholzer , c. strunk , t. heinzel , k. ensslin , m. holland , and c. schnenberger , science * 284 * , 296 ( 1999 ) .
w. d. oliver , j. kim , r. c. liu , and y. yamamoto , science * 284 * , 299 ( 1999 ) .
r. c. liu , b. odom , y. yamamoto , and s. tarucha , nature * 391 * , 263 ( 1998 ) .
r. de - picciotto , m. reznikov , m. heiblum , v. umansky , g. bunin , and d. mahalu , nature * 389 * , 162 ( 1997 ) .
l. saminadayar , d. c. glattli , y. jin , and b. etienne , phys .
rev . lett . * 79 * , 2526 ( 1997 ) .
g. burkard , d. loss , and e. v. sukhorukov , phys .
b * 61 * , 16 303 ( 2000 ) .
x. matre , w. d. oliver , and y. yamamoto , physica e * 6 * , 301 ( 2000 ) .
w. d. oliver , r. c. liu , j. kim , x. matre , l. di carlo , and y. yamamoto , in _ quantum mesoscopic phenomena and mesoscopic devices in microelectronics _ , edited by i. o. kulik and r. ellialtiolu , nato asi , series c , vol . 559 ( kluwer , dordrecht , 457 - 466 , 2000 ) .
r. ionicioiu , p. zanardi , and f. rossi , phys .
a * 63 * , 050101(r ) ( 2001 ) .
p. recher , e. v. sukhorokov , and d. loss , phys .
b * 63 * , 165314 ( 2001 ) .
w. d. oliver , f. yamaguchi , and y. yamamoto , quant - ph/0107084 .
m. bttiker , phys .
b * 46 * , 12 485 ( 1992 ) .
t. martin , and r. landauer , phys .
b * 45 * , 1742 ( 1992 ) .
g. dresselhaus , phys . rev . * 100 * , 580 ( 1955 ) . e. i. rashba , fiz .
tela ( leningrad ) * 2 * , 1224 ( 1960 ) [ sov .
phys . solid state * 2 * , 1109 ( 1960 ) ] .
a. bychkov , and e. i. rashba , pisma zh .
. fiz . * 39 * , 66 ( 1984 ) [ jetp lett .
* 39 * , 78 ( 1984 ) ] .
s. datta , and b. das , appl .
. lett . * 56 * , 665 ( 1990 ) .
f. mireles , and g. kirczenow , phys .
b * 64 * , 024426 ( 2001 ) .
l. i. schiff , _ quantum mechanics _ ( mcgraw - hill , new - york , 1968 ) .
b. das , s. datta , and r. reifenberger , phys .
b * 41 * , 8278 ( 1990 )
. g. l. chen , j. han , t. t. huang , s. datta , and d. b. janes , phys .
b * 47 * , 4084 ( 1993 ) .
j. luo , h. munekata , f. f. fang , and p. j. stiles , phys .
b * 41 * , 7685 ( 1990 ) .
j. p. heida , b. j. van wees , j. j. kuipers , t. m. klapwijk , and g. borghs , phys .
b * 57 * , 11 911 ( 1998 ) . j. nitta , t. akazaki , and h. takayanagi , phys . rev . lett . * 78 * , 1335 ( 1997 ) .
t. hassenkam , s. pedersen , k. baklanov , a. kristensen , c. b. sorensen , p. e. lindelof , f. g. pikus , and g. e. pikus , phys . rev .
b * 55 * , 9298 ( 1997 )
. a. v. moroz , and c. h. w. barnes , phys .
b * 60 * , 14 272 ( 1999 ) . | the influence of spin - orbit coupling in two - dimensional systems is investigated within the framework of the landauer - bttiker coherent scattering formalism .
this formalism usually deals with spin - independent stationary states and , therefore , it does not include a spin - orbit contribution to the current . in this article , we will rederive the coherent scattering formalism , accounting for the changes brought about by the spin - orbit coupling . after a short review of the features of spin - orbit coupling in two - dimensional electron gases ,
we define the creation / annihilaton operators in the stationary states of the spin - orbit coupling hamiltonian and use them to calculate the current operator within the landauer - bttiker formalism .
the current is expressed as it is in the standard spin - independent case , but with the spin label replaced by a new label which we call the spin - orbit coupling label .
the spin - orbit coupling effects can then be represented in a scattering matrix which relates the spin - orbit coupling stationary states in different leads .
this scattering matrix is calculated in the case of a four - port beamsplitter , and it is shown to mix states with different spin - orbit coupling labels in a manner that depends on the angle between the leads . the formalism is then used to calculate the effect of spin - orbit coupling on the current and noise in two examples of electron collision . |
the existence of an alfvenic instability in weakly magnetized rotating plasmas was discovered in 1959 by e. p. velikhov @xcite and implemented on an astrophysical setting by s. chandrasekar in 1960 @xcite and it is well - known as magneto - rotational instability ( mri ) . the intrinsic nature of such unstable mode profile consists of the coupling between alfven waves and the plasma inhomogeneity , due to differential rotation ( indeed mri is suppressed in uniformly rotating systems ) .
the most intriguing feature of mri is however that it vanishes when the magnetic field has sufficiently high amplitude , because of the effect of the restoring force due to magnetic tension , which is able to stabilize the plasma shift dynamics . despite an instability driven by magnetic tension
does not transport matter , nonetheless mri is very relevant for astrophysical systems and , in particular , for accretion structures .
the shakura idea for accretion @xcite relies on the existence of an effective viscosity , emerging from the plasma turbulence and the differential rotation of the accretion profile .
actually mri turns out to be the most appropriate ( if not the only reliable ) mechanism to rise such a turbulent regime in astrophysical accreting systems @xcite ( see also @xcite , where mri was re - analyzed , after the original derivation ) .
the great impact of this issue can be easily understood when realizing that most of the space plasma are in the condition of weakly magnetized systems , especially stellar disk configurations @xcite .
this scenario has led many authors , over the years , to investigate mri in some different contexts and deepening its different morphologies , see for instance @xcite .
a rather general result has been achieved in @xcite , where a stratified and differentially rotating disk has been addressed in view of tracing the feature of mri as a general property of axially symmetric profiles .
the very surprising feature derived in this work consists of the independence of the stability conditions on the intensity and direction of the background magnetic field .
the aim of the present analysis is to demonstrate how the direction of the magnetic field is actually involved in the stability of the stratified disk , as soon as the so - called co - rotation theorem is taken into account @xcite .
such a theorem states that , for a steady , axially symmetric and purely rotating background ( exactly the adopted hypotheses in @xcite ) , the disk angular velocity depends on the magnetic flux function only : in what follows , this will be equivalently formulated as the orthogonality character of the angular velocity gradient with respect to the background magnetic field . here , we perform the linear perturbation analysis by accounting for the co - rotation profile of the background and then we demonstrate that the obtained dispersion relation takes the same form as the one in @xcite , but after the orthogonality constraint between the background magnetic field and the angular velocity gradient is imposed
. then we investigate how such a revised dispersion relation affects the stability properties of the stratified disk .
the main merit of the present study consists of re - introducing a dependence of the stable profile on the angle that the background magnetic field forms with the entropy gradient , _
i.e. _ the stability region depends on the direction of the stationary magnetic configuration .
a specifical example is considered of a stratified disk around a compact massive astrophysical object , showing how a relevant stability region exists in the inner thick part of the disk .
the basic theoretical framework of the present discussion is the ideal magneto - hydrodynamics ( mhd ) theory , as viewed in axial symmetry , _ i.e. _ reducing the dynamical problem to a two - dimensional one . in particular , in what follows , we will implement the momentum conservation equation @xmath0 and the induction equation @xmath1 where @xmath2 and @xmath3 denote the velocity and magnetic fields , respectively , while @xmath4 is the plasma mass density and @xmath5 its total pressure .
the ideal ( hydrogen - like ) plasma thermodynamics is regulated by the mass and entropy conservation equations , which read respectively @xmath6 according to the axial symmetry , in what follows we adopt standard cylindrical coordinates ( @xmath7 ) and we take both the background ( denoted by a suffix @xmath8 ) and the perturbed ( characterized by a corresponding suffix @xmath9 ) quantities as independent of the angle @xmath10 , while the former are also stationary ( independent of the time variable @xmath11 ) . more specifically , the background equilibrium is guaranteed by the simultaneous validity of the gravitostatic equilibrium and the lorentz force - free condition , _
@xmath12 where @xmath13 is the background disk angular velocity , @xmath14 the keplerian angular velocity and we adopt a compact notation for the poloidal vector radius ( _ i.e. _ @xmath15 ) .
we consider separately the force - free condition , because we regard the background field @xmath16 as the radiation one , due to the central star .
furthermore , the basic co - rotation condition must hold @xmath17 the linear perturbation theory is faced in the local approximation limit , when the wavelength of the disturbances is much less than the typical scale of variation characterizing the background , _
i.e. _ we will deal with the condition @xmath18 ( @xmath19 being the wave vector used below ) . such assumption allows to treat the background variables as parametric quantities and the fourier analysis is available .
finally , we will adopt also the so - called boussinesq approximation @xcite , which reduces the mass conservation equation to the local incompressibility of the perturbations , _ i.e. _ equation ( [ cw ] ) becomes the constraint @xmath20 , where @xmath21 denotes the poloidal perturbed velocity .
this assumption corresponds to require that the sound speed is much greater than the alfven one and therefore the plasma @xmath22-parameter is large .
the study of the perturbation dynamics does not require , as we shall see , the specification of the equation of state , relating @xmath23 and @xmath24 , since the perturbed pressure will be provided by requiring the validity of the local incompressibility during the plasma evolution .
the resulting expression for @xmath23 , for sufficiently large wavenumbers , results to be negligible in equation ( [ dw ] ) , which de facto becomes a link between @xmath25 and @xmath21 .
let us consider the case of axisymmetric perturbations , with a poloidal wave - vector @xmath26 . the basic equations from which we start are the perturbed poloidal and azimuthal components of the momentum conservation equation and of the induction equation in the linear approximation , which read @xmath27 and @xmath28 respectively . from and
the components @xmath29 and @xmath30 of the magnetic field can be written in terms of other variables and the resulting expressions can be inserted into .
an expression for the perturbed pressure @xmath23 can be given by multiplying times @xmath19 and using the mass conservation equation in the boussinesq approximation , namely @xmath31 , which must be preserved during the dynamics .
eventually , the perturbed mass density can be given from the entropy conservation equation , which written for adiabatic perturbations , in the limit when @xmath32 ( which is easy to realize accounting for the large value of @xmath33 in the local approximation scheme ) , reduces to the form @xmath34 where @xmath35 .
the dispersion relation can be given by combining the conditions one gets from and after all these manipulations and by considering the following decomposition for the wave vector : since , by virtue of the co - rotation theorem , the two vector @xmath16 and @xmath36 are orthogonal and the wavevector @xmath19 has zero azimuthal component ( the perturbations preserve axial symmetry ) , we can write @xmath37 or more explicitly @xmath38 with @xmath39 being the angle between the background magnetic field and the equatorial plane . the final dispersion relation is @xmath40 with @xmath41\nonumber\\ & \left ( c_{0r}\frac{k_z^2}{k^2 } - c_{0z}\frac{k_rk_z}{k^2 } \right)\label{b } \\
c=&\frac{12}{5}\omega_a^2\omega_0 ^ 2 \bigg[- \frac{k_z^2}{k^2}\left(\frac{\vec{\nabla}p_0\cdot \vec{c}_0}{\rho _ 0 } - \frac{(\vec{\nabla}p_0\cdot \vec{k})(\vec{k}\cdot\vec{c}_0)}{k^2\rho _ 0 } \right)+\nonumber\\ & + \left ( c_{0r}\frac{k_z^2}{k^2 } - c_{0z}\frac{k_rk_z}{k^2 } \right)\left ( \frac{\partial _ rp_0}{\rho _ 0}\frac{k_z^2}{k^2 } - \frac{\partial _
zp_0}{\rho _ 0}\frac{k_rk_z}{k^2}\right ) \bigg]\,.\end{aligned}\ ] ] above , we defined the quantities @xmath42 , @xmath43 and @xmath44 . if as in @xcite one introduces @xmath45 it can be shown that @xmath46 from which it follows that @xmath47 .
moreover , the following relation holds for the second term on the right - had side of @xmath48 similarly , using , and also the expression for @xmath49 can be simplified . collecting all these results
, we end up with the following expressions @xmath50\bigg|_{cr}\\ b=&\left[- 4\omega_a^2\omega _ 0 ^ 2\frac{k_z^2}{k^2}\right]\bigg|_{cr}\\ c=&0\,,\end{aligned}\ ] ] where @xmath51 means the imposition of the co - rotation theorem , _
i.e. _ the decomposition for @xmath19 .
the resulting dispersion relation formally coincides with that given by balbus in @xcite , but for the implementation of the corotation theorem .
similarly , the study of the stability can be performed as in @xcite : the stability condition can be translated into the condition that the dispersion relation does not admit solutions in the limit @xmath52 , which implies the following inequality @xmath53+n_r^2+r\,\partial_r\omega_0 ^ 2>0\ , , \end{aligned}\ ] ] where @xmath54 the inequality above is always verified if @xmath55>0\,.\label{stab2}\end{aligned}\ ] ] the imposition of the co - rotation theorem here means that the derivatives of @xmath56 must be written according to the following relations @xmath57 which come from the decomposition of the radial and vertical directions as in . therefore , the stability condition becomes @xmath58>0\,,\label{sc2}\ ] ] which can be rewritten as follows @xmath59 this relation outlines how for a given matter configuration , _
i.e. _ given @xmath60 and @xmath61 , the stability of the system depends on the direction of the magnetic field .
let us consider as an application of the previous analysis the case of a specific background morphology , describing a stratified ( thick ) accretion disk around a compact astrophysical object . according to equations ,
the background magnetic field is essentially that generated by the central object and it is here modeled by a dipole - like configuration ( a reliable approximation for many concrete astrophysical systems ) , _ i.e. _ it is determined by the function @xmath62 , where the constant @xmath63 fixes the magnetic field amplitude , and it leads to the following magnetic fields components @xmath64 the gradients of the pressure can be inferred from , by assuming @xmath65 , @xmath66 being the mass of the central object .
this assumption comes from implementing the co - rotation theorem and requiring that on the equatorial plane the angular velocity be keplerian @xcite . by investigating the consistency conditions coming from , it turns out that the matter density is the product of @xmath67 times an arbitrary function of @xmath68 , while the pressure equals the primitive of such arbitrary function modulo a constant .
we choose an exponential function , leading to the following particular configuration @xmath69 @xmath70 and @xmath71 being two integration constants .
it is easy to check that these expressions solve , as it comes by direct substitution .
the corresponding profile is characterized by an exponential decay along @xmath72 , with a decreasing decay length along the radial direction ( see figure 1 ) .
[ fig : graficodisco ] the stability condition can be shown to be already contained into , such that the stable region is solely determined by the direction of the magnetic field according to ( it is drawn in figure 1 ) .
this is clarified in figure 2 : in the stable region the angle between the magnetic field and the entropy gradient is acute , while in the unstable region it is obtuse .
this analysis outlines how the relative orientation of the magnetic field with respect to the entropy gradient is crucial for the stability of the whole system . in particular , for the considered configuration mri is suppressed close to the central object in the inner part of the stratified disk ( in correspondence to the lobes in figure 1 ) and therefore no turbulent transport can easily emerge there .
[ fig : vector ]
we developed a perturbation analysis which accounts for the validity of the corotation theorem on the background disk configuration , here translated as the orthogonality of the background magnetic field and the angular velocity gradients .
the effect of such a restriction on the dispersion relation corresponds to restore a dependence on the background magnetic field direction into the mode spectrum ( absent in @xcite ) .
such an effect is relevant only for a stratified plasma disk since , when the @xmath72-dependence is weak or removed ( like in a thin disk morphology ) , the corotation constraint is automatically satisfied : the magnetic field is mainly vertical and the angular velocity gradient is essentially radially directed .
we have seen how , in a specific sample , the restored magnetic field dependence leads to a significant modification of the mri morphology across the disk , since a threshold angle emerges .
in fact , mri can develop only when the angle between the magnetic field and the entropy gradient of the background configuration is bigger than @xmath73 . from an astrophysical point of view , the obtained result could affect the transport properties of thick stratified disks , since in some regions of the plasma profile the turbulent flow due to mri is clearly suppressed ( as elucidated in the previous section ) , together with the associated effective viscosity .
this feature is particularly enhanced far from the equatorial plane ( when also the dipole magnetic field is significantly inclined ) and for stiff radial pressure gradients , _
i.e. _ when the disk rotation deviates from the keplerian behavior @xcite . independently from the astrophysical range of applicability for the corotation theorem ( for a discussion see @xcite ) , the present study removes the surprising and , a bit unnatural , feature discussed in @xcite , about the independence of mri spectrum on the magnetic configuration , _
i.e. _ on its own trigger mechanism : actually the magnetic field direction enters the spectrum and also a forbidden spatial region for mri emerges .
this work has been developed in the framework of the cgw collaboration ( www.cgwcollaboration.it ) . | we re - analyze the physical conditions for magneto - rotational instability ( mri ) within a steady axisymmetric stratified disk of plasma , in order to account for the so - called corotation theory ( the differential angular velocity depends on the magnetic flux surface ) .
we develop the study of linear stability around an astrophysical background configuration , following the original derivation in @xcite , but implementing the corotation condition as the orthogonality between the background magnetic field and the angular velocity gradient .
we demonstrate that a dependence on the background magnetic field direction is restored in the dispersion relation and , hence , the emergence of mri is affected too . |
we acknowledge support from epsrc programme grant ep / j007544/1 , erc advanced grant excipol , 320570 and the leverhulme trust .
our system comprises a planar waveguide with gaas core , algaas bottom cladding and silicon nitride top cladding .
three shallow in@xmath54ga@xmath55as quantum wells are embedded in the core parallel to waveguide plane and at the peak of the electric field density .
details of the structure are the same as those in ref . except for the use of three quantum wells .
the transverse - electric ( te ) polarised guided optical mode couples strongly to the quantum well 1s heavy - hole exciton at wavelength 839.6 nm to form te polarised polaritons .
the vacuum rabi splitting is 9mev .
the linear loss length of 400 @xmath2 m is due to a combination of photon tunneling through the cladding and optical absorption by the gaas core material and tail of the exciton line .
excitation laser pulses were generated by spectrally filtering 100fs pulses from a mode - locked titanium doped sapphire laser with a diffractive pulse shaper employing an adjustable slit .
the laser repetition rate was 80mhz .
the pulses were injected into the waveguide through the grating couplers of period 250 nm by tuning their incidence angle and central energy to correspond to those of the guided mode .
coupling was maximised at around 10 degrees by monitoring the emission from the output grating .
light coming from the sample was collected by an objective lens and spatially filtered so that only light from the output grating was detected .
the temporal profile of the pulses was measured by projecting an image of the output grating onto the entrance slit of a streak camera .
the streak camera has a temporal resolution of 2.0ps , measured using the reflection of the excitation laser pulse from an unpatterned part of the sample surface .
the spectrally resolved emission was measured by projecting the fourier plane of the objective onto the entrance slit of an imaging spectrometer and integrating the light with a ccd detector .
the pulse energy was determined by focussing all light from the output grating onto a commercial photodiode - based power meter and dividing the average beam power by the laser repetition rate .
a ratio of 1:4 for the power scattered towards the power meter and that lost in the substrate was determined by modelling pulse outcoupling through the grating using a two - dimensional finite - difference - time - domain ( fdtd ) method and has been included to obtain the total pulse energy before outcoupling .
the wavevector resolved time - of - flight measurements were performed by projecting the fourier plane of the objective onto the entrance slit of the streak camera .
the final lens was scanned in order to align each wavevector component with the slit . the photoluminescence emission presented in fig [ fig1]b )
was measured using continuous wave laser excitation at 685 nm wavelength .
the conservative soliton solutions are parameterized by the soliton frequencies @xmath59 and velocities @xmath60 and can be sought in the form @xmath61 where @xmath62 , @xmath63 , @xmath64 and @xmath65 are real functions of @xmath66 . substituting the anzats ( [ eq : analitycal_solution ] ) into equation ( [ eq : max_lorentz_temp ] ) we obtain the expressions for amplitudes @xmath67 , @xmath68 of the field .
they read carusotto , i. et al .
fermionized photons in an array of driven dissipative nonlinear cavities _ phys .
lett . _ * 103 , * 033601 ( 2009 ) teras , h. , flayac , h. , solnyshkov , d. d. and malpuech , g. , non - abelian gauge fields in photonic cavities and photonic superfluids _ phys
lett . _ * 112 , * 066402 ( 2014 ) peschel , u. , peschel , t. and lederer f. , temporal solitary waves near linear resonances _ j. opt .
soc . am . b _ * 14 , * 2994 - 3002 ( 1997 ) ; gabitov , i. r. et .
al . , double - resonant optical materials with embedded metal nanostructures _ j. opt
. soc . am .
b _ * 23 , * 535 - 542 ( 2006 ) . | new functionalities in nonlinear optics will require systems with giant optical nonlinearity as well as compatibility with photonic circuit fabrication techniques .
here we introduce a new platform based on strong light - matter coupling between waveguide photons and quantum - well excitons . on a sub - millimeter length scale
we generate sub - picosecond bright temporal solitons at a pulse energy of only 0.5 pico - joules . from this
we deduce an unprecedented nonlinear refractive index 3 orders of magnitude larger than in any other ultrafast system .
we study both temporal and spatio - temporal nonlinear effects and for the first time observe dark - bright spatio - temporal solitons .
theoretical modelling of soliton formation in the strongly coupled system confirms the experimental observations .
these results show the promise of our system as a high speed , low power , integrated platform for physics and devices based on strong interactions between photons .
the introduction of strong nonlinearity into photonic circuits has many applications ranging from optical information processing @xcite and soliton physics @xcite to the study of inter - particle interactions @xcite in photonic analogs of important physical systems such as photonic topological insulators@xcite , optical analogs of quantum hall systems . in this work
we address the problem of introducing strong optical nonlinearity by studying ultra - low - power optical solitons in a novel slab waveguide geometry we recently introduced @xcite where photons are strongly coupled to quantum - well excitons .
temporal optical solitons are one of the most fundamental nonlinear phenomena in optics .
they arise due to a competition between group velocity dispersion ( gvd ) , where different frequency components accumulate different phase during propagation , and nonlinear phase modulation where intensity dependent phase shifts can exactly balance those due to gvd @xcite .
they have been proposed for applications @xcite and observed @xcite in long - haul communications lines and more recently investigated in the micro- and nano - scale waveguides suitable for design of all optical processing chips @xcite .
on - chip applications require that solitons form and interact over small distances , which requires that both the gvd and nonlinear effects develop on very short length scales .
it is also important for information processing applications that the underlying nonlinearity responds on picosecond or faster timescales . achieving short nonlinear length scales while minimising power requirments necessitates very large nonlinearity . to generate significant gvd ,
photonic crystal band engineering and semiconductor photonic wires have been used in a number of schemes@xcite . in this work
we simultaneously achieve giant nonlinearity and gvd by exploiting the unique optical properties of exciton - polaritons @xcite .
these quasi - particles are the quantum eigenmodes formed when photons couple strongly to a quantum - well exciton resonance .
they have characteristics coming from both their light and matter constituents , with the proportion depending on the detuning from resonance .
the photonic component contributes long lifetime and detailed control over system properties via photonic circuit fabrication techniques @xcite .
matter - like interparticle scattering leads to enormous nonlinearity which responds on an ultrafast timescale .
we demonstrate formation of bright sub - picosecond temporal polariton solitons with pulse energies of less than 0.5 pj . from the soliton formation threshold we deduce a nonlinear refractive index
three orders of magnitude larger than in any other ultrafast optical system , which verifies the highly nonlinear nature of our platform .
we show that the frequency dependent transition of polaritons from photon - like to matter - like leads to giant group velocity dispersion on the order of 1000 ps@xmath0 m@xmath1 .
this is comparable to that achieved in fiber bragg gratings and photonic crystal waveguides@xcite . as we achieve the gvd necessary for soliton formation without the need for transverse photonic confinement we are able to study spatio - temporal effects as well as purely temporal ones .
we experimentally demonstrate dark - bright spatio - temporal solitons , which to our knowledge have not previously been seen in any experimental setting with either photons or polaritons .
these are solitons where the intensity is peaked in the propagation direction and extended in the transverse direction apart from a dark notch .
note that bright@xcite and spatial dark@xcite exciton - polariton solitons have previously been reported in strongly coupled bragg microcavities .
compared to these our system has the advantages of an order of magnitude faster soliton velocity , simplicity of design and fabrication , compatibility with existing designs of high performance waveguide photonic circuit elements and , crucially , does not depend on external continuous wave pumping to balance loss for soliton formation to be possible . apart from these ,
the major advantages of our platform for nonlinear optics are the combination of very large nonlinearity and ultra - fast temporal response .
the intrinsic nature of the large dispersion is also advantageous for on - chip soliton formation , pulse compression @xcite and the study of two - dimensional effects . between polariton and exciton energies.,width=8 ]
the structures we use in this work are planar algaas waveguides with quantum wells parallel to the substrate @xcite , see schematic diagram in fig .
[ fig1]a ) .
diffraction in the crystal growth direction is prevented by total internal reflection .
as the photonic losses are moderate and broadening of the exciton line is suppressed ( the temperature is 10k ) the system is in the strong coupling regime and exciton - polaritons are formed .
the sample is excited by pulses with a square spectral envelope of 5.5mev corresponding to an approximately 350 femto - second ( fs ) temporal full width at half maximum ( fwhm ) .
the pulses are coupled in at near - zero incidence angle through the input grating , propagate 600 @xmath2 m and couple out through the output grating .
fig .
[ fig1]b ) shows the polariton dispersion relation determined by angle - resolved photoluminescence spectroscopy@xcite .
as the momentum increases along the lower polariton branch the polaritons become more matter like and their group velocity approaches zero . in order to model the experimental results we describe a light pulse interaction with a narrow exciton resonance using the maxwell - lorentz system .
any group velocity dispersion ( gvd ) related to the waveguide geometry is negligible over the spectral range close to the exciton resonance and in comparison with the gvd induced by the photon exciton coupling .
the resulting set of equations accounting for the dominant nonlinearity originating in the two - body exciton - exciton interaction is [ eq : max_lorentz ] @xmath3 here @xmath4 and @xmath5 are the propagation constant and the group velocity of the photonic waveguide mode at the exciton frequency @xmath6 and @xmath7 . @xmath8 and @xmath9 are the coordinates , respectively , along and across the propagation direction .
the waveguide width along @xmath9 is much larger than the diffractive spread achieved over its length .
@xmath10 is time , @xmath11 is the amplitude of the photonic mode , which is tightly bound in the direction perpendicular to the waveguide plane , and @xmath12 is the excitonic polarization scaled to be in the same units as @xmath11 .
@xmath13 are the photon and the exciton coherence decay rates and @xmath14 is the rate of the light matter coupling . for plane - waves @xmath11 and @xmath12 proportional to @xmath15 , where @xmath16 is a propagation constant offset from @xmath4 and @xmath17 is a frequency offset from @xmath6 , the lossless dispersion law is @xmath18 . here
@xmath19mev is the energy splitting between the two polariton branches at @xmath20 .
the gvd parameter @xmath21 .
thus the gvd is normal ( i.e. @xmath22 ) for pulses with central frequency @xmath23 detuned down from the resonance ( @xmath24 ) towards the lower polariton branch and anomalous ( @xmath25 ) otherwise . in our experiment
the pulse central ( carrier ) frequency @xmath26 varies from @xmath27mev to @xmath28mev , so that we operate on the part of the lower polariton branch where the polaritons are between 74% and 84% photonic and @xmath29 is between 400 and 1000 ps@xmath0 m@xmath1 ( see fig .
[ fig1]b ) . within our experimental and theoretical frameworks
we study both quasi - one dimensional effects when diffraction along @xmath9 is not important and two - dimensional effects , when diffraction is relevant for the physics of the process .
mathematically , the system under consideration is similar to the equations used to study gap solitons @xcite and solitons with dispersion provided by material resonances of different physical origin @xcite .
note that the latter type of solitons has mostly been studied theoretically and the experimental studies have been focused on so - called self - induced transparency solitons in atomic gases @xcite .
spatio - temporal solitons are a separate large area of research , where the most convincing experimental measurements of quasi - solitonic effects have been demonstrated in @xmath30 crystals with very challenging dispersion control @xcite and in arrays of coupled waveguides @xcite .
previous efforts to use nanophotonic systems to observe spatio - temporal effects @xcite have not been very successful .
we proceed by describing our results on observation of one dimensional temporal solitons .
we first present experimental measurements which demonstrate temporal soliton formation and then confirm this by comparing the data with numerical modelling .
dependencies of the output pulse width vs total output pulse energy are shown in fig .
[ fig_temporal_results]a ) for several detunings of the central pulse frequency from the exciton , @xmath23 .
the detector resolution causes the incident 350fs pulse to be measured as 2.0ps . below a threshold pulse energy
the measured fwhm after propagation though the waveguide is between 2.7 to 4.2 ps depending on @xmath23 , which indicates significant temporal speading due to the group velocity dispersion .
the inset of fig .
[ fig_temporal_results]a ) shows the temporal profile of the pulse collected from the output grating for pulse energies above and below the threshold .
one should note that the pulse shape is highly asymmetric below threshold and becomes symmetric above threshold .
fig .
[ fig_temporal_results]b ) shows the corresponding output spectral widths while the inset shows pulse spectra above and below threshold . for low pulse energies
the output spectrum reflects that of the incident pulse slightly modified by the frequency dependent coupling efficiency of the grating coupler ( see inset ) .
as the pulse energy is increased above threshold the pulse becomes significantly shorter and more symmetrical in time until , for the strongest pulses , the measured fwhm reaches a minimum between 2.4ps and 2.7ps , depending on @xmath17 .
at the same time the spectrum shows strong narrowing about the central frequency reaching values in the range 1.8 to 2.3mev , as shown in fig .
[ fig_temporal_results]b ) . during this process
the total power collected from the output grating remains linear with pump power to within a few percent indicating that the spectral and temporal compression are due to a coherent transfer of energy between frequency components rather than a nonlinear absorption or filtering effect .
such coherent spectral and temporal compressions occuring at a critical input energy and levelling off with further energy increase are characteristic of the transition from dispersive propagation to soliton formation .
another key signature of soliton formation is linearisation of the soliton dispersion relation resulting in a single group velocity for all soliton components .
figure [ fig3 ] shows the time of flight of different wavevector components of the pulse above and below threshold .
below threshold the time of flight is simply determined by the polariton dispersion relation with polaritons close to resonance travelling slower , see fig .
[ fig1](b ) . above threshold
, however , there is a very clear departure from this trend . to within the experimental error
the time of flight becomes constant with wavevector across the range where there is significant intensity .
this should be compared with a difference of 4ps over the same wavevector range below threshold .
the dispersion is therefore cancelled out above threshold confirming the formation of a soliton .
the soliton velocity of 39 @xmath2 m ps@xmath1 is faster than that of linearly propagating polaritons at the same energies but significantly slower than the 58 @xmath2 m ps@xmath1 expected for uncoupled photons .
this confirms that the strong light - matter coupling is preserved during the propagation of the high power pulse .
to summarise , the experimental evidence for soliton formation consists of strong coherent temporal and spectral narrowing of the output pulses at a critical energy and a strong reduction in the spread of velocities among different wavenumber components of the pulses .
soliton formation requires the characteristic length scale @xmath31 for the buildup of nonlinear phase to be comparable to the dispersion length . here , @xmath32350fs is the pulse width , @xmath33 is the characteristic pulse energy for the soliton formation and @xmath34 is the nonlinear waveguide parameter @xcite .
the dispersion length is given by @xmath35 m where @xmath36 is the @xmath37 pulse half width .
we note that solitons in any lossy system will lose intensity as they propagate with a characteristic loss length @xmath38400@xmath2 m in our system .
provided that @xmath39 the nonlinearity will adiabatically adjust the pulse width to compensate the loss and solitons may still be expected . from these considerations we now estimate the effective nonlinear refractive index in our system . taking the threshold pulse energy at the waveguide exit as 100fj and @xmath40 400@xmath2
m we estimate @xmath41fj at the input . setting @xmath42
we obtain @xmath43 -18,000/(wm ) .
this may be related to the nonlinear refractive index by @xmath44 where @xmath45 is the vacuum wavenumber and @xmath46 6.6 @xmath2m@xmath0 is the effective mode area @xcite .
we therefore obtain @xmath47 - 1.6x10@xmath48 m@xmath0w@xmath1 . this is more than three orders of magnitude larger than 6x10@xmath49 m@xmath0w@xmath1 in silicon @xciteand ingap @xcite which have recently been used in a suspended membrane photonic crystal geometry in two of the most promising demonstrations of solitons for integrated optics until now . in those systems solitons
are formed at pulse energies of 12pj for ingap @xcite and 9pj for silicon @xcite which are 1.3 - 1.4 orders of magnitude higher than in our system .
we note that in those works the effective mode areas are smaller ( @xmath500.5@xmath2m@xmath0 ) and the dispersion lengths are much longer ( 3.6 mm and 1 mm ) than in our system both of which can be used to reduce the threshold , the latter at the expense of increased circuit size .
figs .
[ fig : theory](a , b ) show numerically computed time domain and spectral pulse profiles vs pulse energy at the end of the waveguide under the assumption that loss is neglected . by tracing the evolution of the pulse along the waveguide
we have confirmed that exact temporal solitons are excited for pulse energies above @xmath51fj .
experimental values of the losses introduced into the modeling modify the pulse dynamics only quantitatively , see figs .
[ fig : theory](c , d ) .
they accelerate pulse broadening and shift the threshold for formation of solitons towards higher energies , @xmath52fj . figs .
[ fig : theory](e , f ) show evolution of the pulse profile along the waveguide with the real losses below ( e ) and above ( f ) the soliton formation threshold . for every profile shown in
( f ) the red curves indicate a matching analytical soliton solution , see methods , thus confirming that we are dealing with adiabatically decaying solitons . for comparison with experimental results ,
the modelled dependencies of the temporal and spectral widths of the output pulses on pulse energy are shown in figs .
[ fig_temporal_results](c , d ) .
the 2ps resolution of the streak camera has been included in the temporal widths . as in the experimental case
there is a temporal compression of up to 1ps depending on detuning which occurs over a few tens of femto - joules pulse energy .
also as in the experiment the spectral widths reduce by @xmath50 2mev over the same pulse energy range . this good semi - quantitative agreement between experiment and theoretical soliton solutions
futher confirms the observation of temporal solitons .
we now shift our attention to the conditions where diffraction in the transverse direction is important .
since the polariton - polariton interaction is repulsive , corresponding to defocussing nonlinearity , a gaussian beam is expected to exhibit enhanced diffraction as the power increases . as well as this defocussing the system
is also expected to support dark spatial solitons @xcite .
ideally , these consist of a high intensity background extending to infinity in the direction transverse to propagation with a dark notch at the centre whose width is defined by the background intensity . in analogy to temporal solitons the fundamental spatial soliton is formed when the nonlinear length matches the diffraction length so that diffraction is balanced by self phase modulation .
the intrinsically large gvd in our system makes the dispersion and diffraction lengths comparable so that both may be compensated at the same low pulse energy and thus form spatio - temporal solitons . before proceeding
it is worth mentioning that if transverse effects are undesirable for a particular application they may be removed simply by employing transverse confinement such as a waveguide ridge .
staying in the planar geometry , the excitation beam transverse profile was modified to excite a dark soliton following the approach used in ref . .
the initially gaussian beam was passed through a thin - film phase plate in order to introduce a @xmath53 phase jump at its centre .
the intensity distribution at the input grating then consisted of two bright lobes with a dip in intensity between them where the phase is discontinuous .
figure [ fig : bright_dark ] shows the width of the full distribution and of the dark notch measured at the output grating as a function of pulse energy .
the upper inset of the figure shows the transverse profile for high and low powers . as the power
is increased the total width of the distribution increases by 6 microns due to nonlinear defocussing as expected .
in contrast , the width of the intensity notch decreases by 5 microns .
a simple defocussing effect would be expected to enhance the diffraction of the notch as well as the sidelobes .
the fact that the notch narrows indicates that we form a dark soliton .
in fact , the combination of sidelobe broadening and notch narrowing represents the distribution gradually approaching that of the ideal dark soliton as the nonlinear interaction becomes important . at the same powers as the spatial soliton forms the temporal duration decreases from 4.7ps to 3.6ps .
thus we have observed clear experimental signatures of spatio - temporal dark - bright solitons .
these solitons have been previously studied only theoretically@xcite .
we have also numerically modeled formation of these solitons and found our results to be in very good agreement with the experiment .
the numerically computed transverse widths are shown in fig .
[ fig : bright_dark]b ) . as in the experimental case
the nonlinear interaction widens the total distribution and narrows the dark notch by several microns with the rate of change of width increasing with power .
the good semi - quantiative agreement of these trends again confirms that the experiment observes spatio - temporal solitons . in conclusion ,
we have demonstrated formation of picosecond light - matter solitons with unprecedented low pulse energy density .
we have studied both temporal and spatio - temporal nonlinear effects and observed the dark - bright spatio - temporal soliton for the first time .
the solitons exist in a novel regime where intrinsically large dispersion is balanced by a nonlinearity several orders of magnitude larger than in any other ultrafast system .
these properties allow nonlinear phases to accumulate and be controlled over short distances consistent with on - chip applications at low power .
the waveguide exciton - polariton system therefore represents an important platform for high speed integrated optics requiring large nonlinearity . |
the 2-matrix hermitian models are matrix models with the probability measure @xmath6 defined on the space of pairs @xmath7 of @xmath8 hermitian matrix . the constant @xmath9 is the normalization constant of the measure , @xmath10 is a coupling constant and @xmath11 , @xmath12 are the standard lebesgue measures on the space of hermitian matrices . in ( [ eq:2 mm ] ) ,
@xmath2 and @xmath13 are called potentials of the matrix model . in this paper
, we shall consider @xmath2 to be a general even polynomial and @xmath13 to be the monomial @xmath14 .
let @xmath15 and @xmath16 be the eigenvalues of the matrices @xmath3 and @xmath17 respectively , then the eigenvalues of the matrix model ( [ eq:2 mm ] ) are distributed according to @xmath18 where @xmath19 is a normalization constant and @xmath20 , @xmath21 .
the two - matrix model was introduced in @xcite , @xcite as a generalization of the one matrix model to study critical phenomena in physical systems .
the 2 matrix model is needed to represent all conformal models in statistical physics @xcite .
it is also a powerful tool in the studies of random surfaces as the large @xmath22 expansion of the partition function @xmath19 is expected to be the generating function of discretized surface @xcite . since its introduction , the 2 matrix model has become a very active research area @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite and one of the major problems is to obtain rigorous asymptotics for the eigenvalue statistics .
a good review of the subject can be found in @xcite , @xcite .
a particular important object in the studies of eigenvalue statistics is the correlation function @xmath23 @xmath24 in @xcite , @xcite , a connection between biorthogonal polynomials and the correlation functions of 2 matrix models ( [ eq : corel ] ) was found .
let @xmath25 and @xmath26 be monic polynomials of degrees @xmath27 and @xmath28 such that @xmath29 for some constants @xmath30 , then these polynomials exist and are unique @xcite , @xcite .
these polynomials are known as biorthogonal polynomials .
let us define some integral transforms of the biorthogonal polynomials by @xmath31 and define the kernels to be @xmath32 then the correlation function @xmath33 ( [ eq : corel ] ) has the following determinantal expression .
@xmath34 upon averaging over the variables @xmath35 , we see that the @xmath36-point correlation function @xmath37 for the eigenvalues of the matrix @xmath3 is given by the kernel @xmath38 , @xmath39 the purpose of this paper is to provide a rigorous asymptotic expression for the kernel @xmath40 as @xmath4 for @xmath14 and @xmath41 being a general even polynomial . due to a generalized christoffel - darboux formula , the kernels @xmath40 can be expressed in terms of a finite sum of the biorthogonal polynomials ( see @xcite , @xcite , @xcite and @xcite ) .
this reduces the problem of finding an asymptotic expression for @xmath40 into finding the asymptotics of the biorthogonal polynomials . until recently , most results in the asymptotics of biorthogonal polynomials have been obtained through heuristic argument ( see @xcite , @xcite ) . for a long time , the only rigorous result was the case when both @xmath42 and @xmath41 are quadratic polynomials @xcite . in the recent work by duits and kuijlaars @xcite ,
( see also @xcite , chapter 5 , which was later made into the publication @xcite ) , the deift - zhou steepest decent method ( @xcite , @xcite , @xcite , see also @xcite ) was successfully applied to obtain the asymptotics of biorthogonal polynomials with @xmath14 and @xmath41 an even polynomial in the case when the limiting eigenvalue density for @xmath3 is supported on a single interval .
the main idea in @xcite is to transform and approximate the riemann - hilbert problem satisfied by the biorthogonal polynomials @xcite , @xcite ( see section [ se : rhp ] for details ) via the use of suitable equilibrium measures and then solve the approximated riemann - hilbert problem explicitly to obtain asymptotic formula for the biorthogonal polynomials .
the results in @xcite was obtained in the case when one of these equilibrium measures is supported on a single interval .
this case is called the one - cut regular case " in @xcite . to be precise , let @xmath43 be the following energy function @xmath44 where the integral is performed on the supports of the measures @xmath45 and @xmath46 .
then the equilibrium measures @xmath47 , @xmath48 and @xmath49 are the measures that minimize the following energy function @xmath50 . [ de : equ ] ( definition 2.2 in @xcite ( see also definition 5.2.1 in @xcite ) ) the equilibrium measure @xmath51 is the triplet of measures that minimizes the following energy function .
@xmath52 amount non - negative borel measures @xmath53 , @xmath54 and @xmath55 that satisfy the following properties . 1 .
all the measures @xmath46 , @xmath56 have finite logarithmic energies ; 2 .
@xmath53 and @xmath55 are measures supported on @xmath57 with @xmath58 and @xmath59 ; 3 .
@xmath54 is a measure supported on @xmath60 with @xmath61 ; 4 .
let @xmath62 be the unbounded measure on @xmath60 given by @xmath63 where @xmath64 is the unit arc length on @xmath60 , then @xmath54 satisfies the constraint @xmath65 . let @xmath66 be the logarithmic potential of the measure @xmath67 .
@xmath68 then it was shown in @xcite that the logarithmic potentials of @xmath47 and @xmath48 satisfy the following properties @xmath69 for some constant @xmath28 , where @xmath70 is the support of @xmath47 .
the properties of the equilibrium measures @xmath47 , @xmath48 and @xmath49 were studied in @xcite and we have the following [ thm : equ ] ( theorem 2.3 in @xcite ( see also theorem 5.2.2 in @xcite ) ) let @xmath2 be an even polynomial and @xmath71 .
then there is a unique minimizer @xmath51 of @xmath72 in ( [ eq : ev ] ) that satisfies the conditions in definition [ de : equ ] .
let us denote the support of the borel measure @xmath67 by @xmath73 , then we have 1 .
@xmath70 consists of finitely many disjoint intervals @xmath74 , \end{split}\ ] ] where @xmath75 and the points are ordered such that @xmath76 if @xmath77 .
moreover , @xmath47 is absolutely continuous with respect to the lebesgue measure , and on each interval @xmath78 $ ] , it has a continuous density of the form @xmath79,\ ] ] where @xmath80 is analytic and non - negative on @xmath78 $ ] .
2 . let @xmath62 be the measure in ( [ eq : sigma ] ) , then @xmath81 and there exists a constant @xmath82 such that the support @xmath83 of @xmath84 is given by @xmath85 moreover , @xmath84 has an analytic density on @xmath83 that vanishes as a square root at @xmath86 .
3 . @xmath87 and @xmath49 has a density which is analytic in @xmath88 .
4 . for @xmath56
, we have @xmath89 for any borel set @xmath90 . [ re : sim ] in particular , by 4 . in the above
, we see that @xmath70 is symmetric under the map @xmath91 , that is , we have @xmath92 .
we then have the following definition of regularity .
( see definition 2.5 in @xcite ) [ de : reg ] the potential @xmath41 is regular if the following conditions are satisfied . 1 .
the inequality in ( [ eq : reg ] ) is strict outside of @xmath70 ; 2 .
the density @xmath93 vanishes like a square - root at the end points of @xmath70 ; 3 .
the density @xmath93 does not vanish in the interior of @xmath70 .
it is known that a generic potential @xmath41 is regular @xcite .
the one - cut regular case " is the case when @xmath70 consists only of a single interval and that @xmath41 is regular . in @xcite , rigorous asymptotics of biorthogonal polynomials
was obtained for this case .
the asymptotics of the biorthogonal polynomials were then used to obtain an asymptotic expression for the kernel @xmath40 in ( [ eq : ker ] ) . in this paper
we will extend these result to the case when @xmath47 is supported on any number of intervals .
( see theorem [ thm : main1 ] and theorem [ thm : main2 ] )
in this paper we obtain universality results for the 2 matrix model with potentials @xmath14 and @xmath41 a general even polynomial .
moreover , we will assume the potential @xmath41 satisfies the regularity condition in definition [ de : reg ] and that @xmath5 is a multiple of @xmath94 .
then we have the following result on the global eigenvalue distribution of the matrix @xmath3 .
[ thm : main1 ] let @xmath51 be the equilibrium measures that minimize the functional ( [ eq : ev ] ) .
then as @xmath4 and @xmath95 , we have @xmath96 uniformly for @xmath97 , where @xmath98 is the density of @xmath47 in definition [ de : reg ] .
as explained in @xcite , the requirement @xmath99 is not essential and is only imposed to minimize the technicality .
the other result is the universality property of the kernel @xmath40 .
[ thm : main2 ] let @xmath51 be the equilibrium measures that minimize the functional ( [ eq : ev ] ) . then as @xmath4 and @xmath95 , we have the followings . 1 .
let @xmath100 be a point in the interior of the support @xmath70 .
then we have @xmath101 uniformly for @xmath102 , @xmath103 in compact subsets of @xmath57 .
2 . let @xmath104 be such that @xmath105 as @xmath106 , @xmath107 inside of @xmath70 , where @xmath108 are defined as in ( [ eq : smu1 ] ) .
then we have the following @xmath109 uniformly for @xmath102 , @xmath103 in compact subsets of @xmath57 , where @xmath110 is the airy function .
recall that the airy function is the unique solution to the differential equation @xmath111 that has the following asymptotic behavior as @xmath112 in the sector @xmath113 , for any @xmath114 .
@xmath115 where the branch cut of @xmath116 in the above is chosen to be the negative real axis .
although the steepest decent analysis in @xcite already covers the general case without the 1-cut assumption , solution to a ` modeled riemann - hilbert problem ' ( see ( [ eq : outer ] ) ) must be obtained to complete the riemann - hilbert analysis and to extend the universality results to the general case .
the main difficulties are to show that a solution of the modeled riemann - hilbert problem exists for all @xmath5 and to find an explicit expression of it .
this involves the study of the theta divisor , which is the set of points in which a theta function vanishes .
( see section [ se : theta ] for a more detailed description of the theta function ) this is a difficult problem with very few results available . in this paper we managed to construct the solution of the modeled riemann - hilbert problem ( [ eq : outer ] ) with the use of theta functions and by using results from @xcite and @xcite
, we were able to show the existence of the solution @xmath117 to ( [ eq : outer ] ) for all @xmath5 .
this allows us to extend the universality results in @xcite to the case when @xmath41 is a general even polynomial and obtain theorem [ thm : main1 ] and theorem [ thm : main2 ] . in many applications of the deift - zhou steepest decent method
, theta function is needed to solve a modeled riemann - hilbert problem and the solvability of these modeled riemann - hilbert problems is important to guarantee the validity of these asymptotic formula for all @xmath5 , as @xmath4 .
we believe the techniques and results in this paper will not only be valuable to the random matrix community studying 2 matrix models , but it will also be important to many other problems in which the deift - zhou steepest decent method is applicable .
the author is indebted to m. duits and a. b. j. kuijlaars for many fruitful discussions and for inviting me to katholieke universiteit leuven in which this work initiated .
i am also grateful to them for providing me with an early copy of @xcite and for encouraging me to complete this paper .
i am also grateful to m. bertola for showing me the relevant results in @xcite which led to the completion of section [ se : nonvan ] .
the funding of this research is provided by the epsrc grant ep / d505534/1 .
in this section we will summarize the results in @xcite that are relevant to our analysis .
we will not repeat the analysis in @xcite but will simply state the results that are applicable to our studies .
the biorthogonal polynomials @xmath118 in ( [ eq : bop ] ) satisfies a riemann - hilbert problem @xcite , @xcite similar to the one that is satisfied by orthogonal polynomials @xcite .
this allows the implementation of the deift - zhou steepest decent method .
( @xcite , @xcite , @xcite , @xcite ) let @xmath119 be the weights defined by @xmath120 assuming @xmath5 is divisible by @xmath94 and consider the following riemann - hilbert problem @xmath121 this riemann - hilbert problem has a unique solution given by the biorthogonal polynomial @xmath118 and some other polynomials , together with their cauchy transforms @xcite .
@xmath122 where @xmath123 , @xmath124 are some polynomials of degree @xmath125 and @xmath126 is the cauchy transform of the function @xmath127 .
@xmath128 in @xcite , the deift - zhou steepest decent method ( @xcite , @xcite , @xcite , @xcite ) was extended to the riemann - hilbert problem ( [ eq : rhsol ] ) . with the help of the equilibrium measures
@xmath47 , @xmath48 and @xmath49 that minimize ( [ eq : ev ] ) , the authors of @xcite were able to transform and approximate the riemann - hilbert problem ( [ eq : rhsol ] ) by an explicitly solvable one . to state their results , let us assume @xmath41 is regular .
let us denote an interval in the support @xmath70 ( [ eq : smu1 ] ) of @xmath47 by @xmath129 and a gap between the intervals by @xmath130 .
@xmath131,\quad j=1,\ldots , g+1,\\ \tilde{\xi}_{j}&=[\lambda_{2j},\lambda_{2j+1}],\quad j=1,\ldots , g,\\ \tilde{\xi}_0&=(-\infty,\lambda_1],\quad \tilde{\xi}_{g+1}=[\lambda_{2g+2},\infty ) .
\end{split}\ ] ] we will define @xmath132 to be the constants @xmath133 note that , since @xmath41 is an even polynomial , by theorem [ thm : equ ] , we have @xmath134 for any borel set @xmath90 . therefore the constants @xmath135 in ( [ eq : apj ] ) satisfy the symmetry @xmath136 let us define the following riemann - hilbert problem for a matrix @xmath117 .
( see section 8 of @xcite and 5.10 of @xcite ) @xmath137 where @xmath83 is oriented upward , the branch of @xmath138 is chosen such that @xmath139 for @xmath140 and the branch cut is chosen to be the negative real axis .
the @xmath141 are given by ( with @xmath142 ) @xmath143 the jump matrices @xmath144 in ( [ eq : outer ] ) are given by the followings @xmath145 the steepest decent analysis in @xcite leads to the ` modeled riemann - hilbert problem ' ( [ eq : outer ] ) .
provided a solution @xmath117 of ( [ eq : outer ] ) exists and is uniformly bounded in @xmath5 away from the singularities , the analysis in @xcite that leads to the asymptotic forms ( [ eq : global ] ) , ( [ eq : bulk ] ) and ( [ eq : edge ] ) can be carried out with the parametrix @xmath117 .
[ thm : dk ] let @xmath146 be a fixed small number independent on @xmath5 .
let @xmath147 and @xmath148 be small discs of radius @xmath149 centered at @xmath108 and @xmath86 respectively .
let @xmath150 be a compact subset in @xmath151 and let @xmath152 be the set @xmath153 suppose the solution @xmath117 of the riemann - hilbert problem ( [ eq : outer ] ) and its inverse @xmath154 exist and satisfy the following conditions . 1
. both @xmath117 and @xmath154 are bounded in @xmath5 uniformly inside @xmath152 for any compact subset @xmath155 ; 2 . for any @xmath156 independent on @xmath5 ,
there exist constants @xmath157 and @xmath158 , @xmath159 , independent on @xmath5 , such that , for @xmath160 , @xmath161 then as @xmath4 and @xmath162 , the asymptotic behavior of the kernel @xmath40 is given by ( [ eq : global ] ) , ( [ eq : bulk ] ) and ( [ eq : edge ] ) . in the following sections we will construct the solution @xmath117 with the help of theta functions and we will show that the solution satisfies the conditions in theorem [ thm : dk ] in section [ se : nonvan ] .
we will now construct a riemann surface from the equilibrium measures and use the theta function on this riemann surface to construct a parametrix for the riemann - hilbert problem ( [ eq : outer ] ) .
the riemann surface is realized as a four - sheeted covering of the riemann sphere .
define four copies of the riemann sphere by @xmath163 , @xmath164 @xmath165 where @xmath166 is the riemann sphere obtained by adding the point @xmath167 to @xmath151 .
the riemann surface @xmath168 is constructed as follows : @xmath169 is connected to @xmath170 via @xmath70 , @xmath170 is connected to @xmath171 via @xmath83 and @xmath171 is connected to @xmath172 via @xmath173 , as shown in figure [ fig : connection ] . .
] let us define the functions @xmath174 by @xmath175 then we have the following result [ le : riemsur ] ( lemma 5.1 in @xcite , ( see also lemma 5.4.1 in @xcite ) ) the function @xmath176 defined by @xmath177 has an extension to a meromorphic function ( also denoted by @xmath178 ) on @xmath168 .
the meromorphic function has a pole of order @xmath179 at infinity on the first sheet , and a simple pole at the other points at infinity .
we shall denote the restriction of @xmath180 to the sheet @xmath181 by @xmath182 . the riemann surface @xmath168 is of genus @xmath183 .
let us define a set of canonical basis of cycle as in figure [ fig : cycle ] .
[ ] [ ] [ 1][0.0]@xmath184 [ ] [ ] [ 1][0.0]@xmath185 [ ] [ ] [ 1][0.0]@xmath186 [ ] [ ] [ 1][0.0]@xmath187 [ ] [ ] [ 1][0.0]@xmath188 [ ] [ ] [ 1][0.0]@xmath189 [ ] [ ] [ 1][0.0]@xmath190 [ ] [ ] [ 1][0.0]@xmath191 [ ] [ ] [ 1][0.0]@xmath192 [ ] [ ] [ 1][0.0]@xmath193 [ ] [ ] [ 1][0.0]@xmath194 [ ] [ ] [ 1][0.0]@xmath195 [ ] [ ] [ 1][0.0]@xmath196 and @xmath197 cycle of riemann surface @xmath168.,title="fig : " ] the figure should be understood as follows .
the top left rectangle denotes the first sheet @xmath169 , the top right rectangle denotes @xmath170 , the lower left one denotes @xmath171 and the lower right one denotes @xmath172 .
a @xmath197-cycle is a loop in @xmath169 around the branch cuts that is symmetric with respect to the real axis , while an @xmath198-cycle @xmath199 consist of a path in the upper half plane in @xmath169 that goes from @xmath200 to @xmath201 ( @xmath129 is defined in ( [ eq : cutgaps ] ) ) , together with a path in the lower half plane in @xmath170 that goes from @xmath201 to @xmath200 .
the loop formed by these 2 paths is an @xmath198-cycle .
we will also choose these 2 paths such that their projection on the complex @xmath202-plane are mapped onto each other under complex conjugation .
we can now define the basis of holomorphic differential that is dual to this basis of cycle .
let @xmath203 be holomorphic differential 1-forms on @xmath168 such that @xmath204 the 1-forms @xmath203 are known as the holomorphic 1-forms that are dual to the basis of cycles @xmath205 .
let the @xmath197-period of these 1-forms be @xmath206 @xmath207 then the @xmath208 matrix @xmath209 with entries @xmath206 is symmetric and @xmath210 . the theta function @xmath211 associated to the riemann surface @xmath168 and this choice of basis is defined by @xmath212 the theta function has the following quasi - periodic property , which will be important to the construction of the parametrix .
[ pro : per ] the theta function is quasi - periodic with the following properties : @xmath213 \theta(\vec{s } ) , \end{split}\ ] ] where @xmath214 denotes the usual inner product in @xmath215 . we will now define the abel map on @xmath168 . the abel map @xmath216 is defined by @xmath217 where @xmath218 is a point on @xmath168 .
we will choose @xmath218 so that @xmath218 is the point on @xmath169 that projects to @xmath188 in @xmath166 .
we will denote this point by @xmath219 .
the composition of the theta function with the abel map is then a multi - valued function from @xmath168 to @xmath151 .
it is either identically zero or it has @xmath183 zeros on @xmath168 .
the following lemma tells us where the zeros are .
[ le : zero ] let @xmath220 be a non special divisor of degree @xmath183 on @xmath168 , then the multi - valued function @xmath221 has precisely @xmath183 zeros located at the points @xmath222 , @xmath223 .
the vector @xmath224 is the riemann constant @xmath225 recall that a divisor of degree @xmath36 is a formal sum of @xmath36 points ( counting multiplicity ) on the riemann surface and that two divisors @xmath226 and @xmath227 are equivalent if and only if there exists a meromorphic function @xmath228 on @xmath168 with poles at the points of @xmath226 and zeros at the points of @xmath227 .
a divisor @xmath229 is special if there exists a non - constant meromorphic function on @xmath168 with @xmath183 poles at the points @xmath230 .
the condition of @xmath231 being non special is equivalent to the condition that @xmath232 does not vanish identically .
[ thm : rr ] let @xmath230 be @xmath183 points on a riemann surface and let the multiplicity of @xmath233 within these @xmath183-tuple of points be @xmath234 .
then the function @xmath235 is identically zero if and only if there exist a function @xmath236 that has poles of order @xmath234 at @xmath233 for @xmath237 and holomorphic elsewhere .
this is a consequence of the riemann - roch theorem . in general , for a given @xmath238 points ( counting multiplicity ) on a riemann surface , there are @xmath28 independent meromorphic functions with poles exactly at these points .
this can be thought of as an extension of the liouville s theorem .
let @xmath239 be the anti - holomorphic involution on @xmath168 defined by @xmath240 where @xmath180 is the function on @xmath168 given by lemma [ le : riemsur ] .
then by the definition of the cycles in figure [ fig : cycle ] , we see that under the involution @xmath241 , we have @xmath242 where the symbol @xmath243 means that @xmath244 is homologic to @xmath199 .
in particular , if we consider the holomorphic 1-forms @xmath245 on @xmath168 , we have @xmath246 hence by the uniqueness of holomorphic 1-forms that is dual to the cycles @xmath205 , we have @xmath247 . by computing the @xmath197-periods of @xmath245 and making use of ( [ eq : involcyc ] )
, we obtain the following . [ le : imag ] the period matrix @xmath209 of @xmath168 is purely imaginary . in particular , by ( [ eq : thetadef ] ) , we see that @xmath248 is real and positive and from lemma [ le : zero ]
, we see that @xmath249 has @xmath183 zeros on @xmath168 .
let us denote these zeros by @xmath250 . then by the following result in @xcite
, we can simplify the expression of the riemann constant @xmath251 .
[ pro : constant ] ( see p.308 - 309 of @xmath252 ) suppose @xmath249 is not identically zero .
then it has @xmath183 zeros in @xmath168 .
let @xmath250 be its zeros , then the riemann constant is given by @xmath253 [ re : iota ] let @xmath254 be the intervals in @xmath255 that projects onto the gaps @xmath130 in ( [ eq : cutgaps ] ) , then as we shall see in corollary [ pro : fay2 ] , there is exactly one point @xmath256 that belongs to @xmath257 for @xmath237 .
we shall label the @xmath256 such that @xmath258 .
we would like to express the function @xmath249 as a meromorphic function on @xmath151 with jump discontinuities . to do so
, we need to define the contour of integration in the abel map ( [ eq : abel ] ) such that the integral can be defined without ambiguity .
we will define the contour of integration as in figure [ fig : contour ] .
[ ] [ ] [ 1][0.0]@xmath184 [ ] [ ] [ 1][0.0]@xmath185 [ ] [ ] [ 1][0.0]@xmath186 [ ] [ ] [ 1][0.0]@xmath187 [ ] [ ] [ 1][0.0]@xmath188 [ ] [ ] [ 1][0.0]@xmath259 [ ] [ ] [ 1][0.0]@xmath260 [ ] [ ] [ 1][0.0]@xmath261 [ ] [ ] [ 1][0.0]@xmath262 [ ] [ ] [ 1][0.0]@xmath195 [ ] [ ] [ 1][0.0]@xmath196 [ ] [ ]
[ 1][0.0]@xmath263 [ ] [ ] [ 1][0.0]@xmath264 .,title="fig : " ] for a point @xmath202 in @xmath169 ( @xmath170 ) , the contour of integration @xmath259 goes from @xmath188 to @xmath202 in @xmath169 ( @xmath170 ) without intersecting @xmath265 and the branch cuts on the imaginary axis . for a point @xmath202 in the upper ( lower ) half plane in @xmath171 ,
the contour of integration @xmath261 consists of 2 parts .
the first part lies in @xmath170 , goes from @xmath188 to a point @xmath263 ( @xmath264 ) on the branch cut in the imaginary axis without intersecting @xmath265 in @xmath170 and enter the branch cut from the left hand side of this point .
the second part lies in the upper ( lower ) half plane in @xmath171 , goes from the right hand side of @xmath263 ( @xmath264 ) to the point @xmath202 . for a point @xmath202 in the upper ( lower ) half plane in @xmath172 ,
the contour of integration consists of 3 parts .
the first part lies in @xmath170 , goes from @xmath188 to a point @xmath264 ( @xmath263 ) on the branch cut in the imaginary axis without intersecting @xmath265 in @xmath170 and enter the branch cut from the left hand side of this point .
the second part lies in the lower ( upper ) half plane in @xmath171 , goes from the right hand side of @xmath264 ( @xmath263 ) to the origin .
the last part lies in the upper ( lower ) half plane in @xmath172 , goes from the origin to the point @xmath202 .
the choice of the point @xmath266 in the construction is immaterial as long as it lies on the branch cut on the imaginary axis .
let @xmath267 be the point on @xmath163 that projects to @xmath202 in @xmath166 and @xmath268 be a @xmath183-dimensional constant vector .
we can now define four functions @xmath269 on the complex @xmath202-plane by @xmath270 these functions will have jump discontinuities in the complex @xmath202-plane . by using the periodicity of the theta function ( [ eq : period ] ) , we can compute their jump discontinuities . [
pro : thetajump ] let @xmath271 be a @xmath183 dimensional vector .
the functions @xmath269 are analytic in @xmath272 . on @xmath273 , they satisfy the following conditions @xmath274 let us first consider the discontinuities of @xmath275 and @xmath276 across @xmath129 .
let @xmath277 be the projection of @xmath168 onto the riemann sphere .
suppose @xmath202 is a point in @xmath129 .
let @xmath278 and define the points @xmath279 to be @xmath280 we will now choose @xmath114 to be real and positive and let @xmath281 . from the definition of the integration contour in figure [ fig : contour ] and the canonical basis of cycles in figure [ fig : cycle ] , we have the following relation between the points @xmath282 as @xmath283 .
@xmath284 where @xmath285 is a vector with 1 in the @xmath286 entry and zero elsewhere and @xmath287 is the vector @xmath288 from this and the periodicity of the theta function ( [ eq : period ] ) , we obtain @xmath289 let us now consider a point @xmath202 in @xmath130 .
again , from the definition of the integration contour and the canonical basis , we have , as @xmath283 , the following @xmath290 where @xmath291 . from this and the periodicity of the theta function
, we see that @xmath292 from the definition of the integration contour , it is clear that @xmath275 and @xmath276 are analytic across @xmath293 and that @xmath275 is analytic across @xmath83 . let us now consider the discontinuities of @xmath294 and @xmath276 on @xmath83 .
let @xmath202 be a point on @xmath83 , from the definition of the contours , it follows immediately that @xmath295 let us now consider the boundary value of @xmath276 on the minus side of @xmath83 . for small and positive @xmath296 , we have @xmath297 where @xmath298 is the close loop on @xmath168 depicted in figure [ fig : gamma ] . since this loop is contractible , we have @xmath299 finally , the conditions @xmath300 follow directly from the definition of the contour of integration .
[ ] [ ] [ 1][0.0]@xmath184 [ ] [ ] [ 1][0.0]@xmath185 [ ] [ ] [ 1][0.0]@xmath186 [ ] [ ] [ 1][0.0]@xmath187 [ ] [ ] [ 1][0.0]@xmath188 [ ] [ ] [ 1][0.0]@xmath298 [ ] [ ] [ 1][0.0]@xmath202 [ ] [ ] [ 1][0.0]@xmath195 [ ] [ ] [ 1][0.0]@xmath196 [ ] [ ] [ 1][0.0]@xmath263 [ ] [ ] [ 1][0.0]@xmath264 .,title="fig : " ] another key ingredient in the construction of the parametrix is meromorphic differentials on the riemann surface .
most of the results that we will be using can be found in @xcite .
[ pro : mero ] let @xmath301 be @xmath27 distinct points on a riemann surface @xmath168 .
let @xmath302 be complex numbers with @xmath303 .
then there exists a meromorphic 1-form @xmath304 on @xmath168 , holomorphic on @xmath305 such that @xmath306 where @xmath307 is a local coordinate near @xmath233 such that @xmath308 at @xmath233 .
this result can be found for example , in @xcite .
( p.52 , theorem ii.5.3 ) a meromorphic 1-form with simple poles only is called a meromorphic 1-form of the third type .
let @xmath304 be a meromorphic 1-form of the third type . in order to define the periods of @xmath304 unambiguously
, one has to define the periods to be integrals around close loops @xmath309 and @xmath310 that are homologic to the @xmath198 and @xmath197-cycles in figure [ fig : cycle ] in @xmath311 , where @xmath312 is the set of poles of @xmath304 . by adding suitable multiples of holomorphic 1-forms to a given meromorphic 1-form
, we can obtain meromorphic 1-forms with arbitrary @xmath198-periods .
for example , if a meromorphic 1-form given by proposition [ pro : mero ] has the following @xmath198-periods @xmath313 then the meromorphic 1-form @xmath314 will have @xmath198-periods @xmath315 but with the same pole structure and residues .
of course , we can not control both the @xmath198 and the @xmath197-periods of the 1-form .
in fact , meromorphic 1-forms with prescribed @xmath198-period and pole structure is uniquely defined . a meromorphic 1-form with all @xmath198-periods zero
is called a normalized meromorphic 1-form .
let @xmath301 be @xmath27 distinct points on a riemann surface @xmath168 .
let @xmath302 be complex numbers with @xmath303 and let @xmath316 be arbitrary complex numbers .
then there exists a unique meromorphic 1-form of the third type @xmath304 on @xmath168 , holomorphic on @xmath305 such that @xmath317 we have already shown the existence part . to see the uniqueness part ,
let @xmath304 and @xmath318 be 2 meromorphic 1-forms of the third type with the properties ( [ eq : mero ] ) .
let @xmath319 be their difference . then , since both @xmath304 and @xmath318 have the same singular behavior at the points @xmath233 , the 1-form @xmath319 does not have any pole and is therefore holomorphic . moreover , all its @xmath198-periods vanish .
since a holomorphic 1-form with vanishing @xmath198-periods has to be zero itself , ( see , for example , @xcite , p.65 , proposition iii.3.3 ) the proposition is proven .
we will conclude this section with a result that relates the periods of a normalized meromorphic 1-form to the values of the abel map at its poles .
[ thm : rb ] ( see e.g. @xcite , p.65 , iii.3 ) let @xmath320 be a meromorphic differential of the third type with simple poles at the points @xmath321 and @xmath322 be a holomorphic differential .
let @xmath323 and @xmath324 be their periods @xmath325 then the riemann bilinear relation is the following @xmath326 where @xmath218 is an arbitrary point on @xmath168 .
we will now construct the local parametrix with the the theta function and meromorphic 1-forms .
let us now define a local coordinate @xmath327 near @xmath328 , the point on @xmath170 , @xmath171 and @xmath172 that projects onto @xmath329 in the riemann sphere .
@xmath330 where @xmath142 and the branch of @xmath138 is chosen such that @xmath331 .
one can check that @xmath327 is indeed holomorphic in @xmath168 in a neighborhood of @xmath328 .
let us now define four meromorphic 1-forms of the third type @xmath332 , @xmath164 by the following properties .
[ de : deltaj ] the normalized meromorphic 1-forms @xmath332 are holomorphic in @xmath333 where @xmath334 are the points in @xmath170 that project onto @xmath86 and @xmath335 are the zeros of @xmath249 . at these points
they have simple poles with residues @xmath336 provided none of the @xmath337 is equal to @xmath338 for some @xmath27 . if some @xmath337 is equal to @xmath339 for some @xmath27 , then the residue at @xmath337 will be @xmath340 .
these 1-forms are then uniquely defined .
we will denote the @xmath197-period of these 1-forms by @xmath341 .
@xmath342 to avoid ambiguity in the @xmath197-periods , let @xmath343 be the projection of @xmath335 on @xmath344 ( see remark [ re : iota ] ) .
then the @xmath197-periods are computed as integrals on @xmath197-cycles in @xmath169 that intersects @xmath344 at any point @xmath345 if @xmath346 .
if @xmath347 , then the @xmath197-cycle can intersect @xmath344 at any point @xmath348 in @xmath169 .
we will now define four functions in the riemann surface @xmath168 .
first let @xmath349 be the images of @xmath350 under the maps @xmath351 , that is , @xmath352 let @xmath353 be a point in @xmath354 .
the exact choice of @xmath353 is immaterial to the construction as long as @xmath355 or @xmath219 .
we will now define the functions @xmath356 on @xmath168 as follows .
@xmath357 where the function @xmath358 is given by @xmath359 and the path of integration is defined in the same way as the ones for the abel map , except that every path now starts at @xmath353 .
let @xmath360 be the point on @xmath181 that projects to @xmath202 in @xmath166 .
as before , we will now define four functions @xmath361 on the complex @xmath202-plane by @xmath362 then these functions have the following jump discontinuities in the complex @xmath202-plane . [
pro : deltajump ] the functions @xmath363 are analytic in @xmath272 . on @xmath273 , they satisfy the following conditions @xmath364 where @xmath365 is the @xmath286 component of the vector @xmath341 . the proof follows similar argument as the ones used in the proof of proposition [ pro : thetajump ] .
first let us consider the jump discontinuities on @xmath350 .
let @xmath202 be a point in @xmath350 and define the points @xmath366 as in ( [ eq : zepsilon ] ) in the proof of proposition [ pro : thetajump ] .
first consider the boundary values @xmath367 and @xmath368 .
choose integration contours @xmath369 and @xmath370 from @xmath353 to the points @xmath371 and @xmath372 as in figure [ fig : gammadelta ] .
let @xmath373 .
then @xmath298 can be deformed into the sum @xmath374 of the @xmath198-cycles and a loop @xmath375 around the point @xmath219 in @xmath376 , where @xmath377 is the set of poles of @xmath332 .
( see figure [ fig : loop ] ) .
@xmath378 [ ] [ ] [ 1][0.0]@xmath184 [ ] [ ] [ 1][0.0]@xmath185 [ ] [ ] [ 1][0.0]@xmath186 [ ] [ ] [ 1][0.0]@xmath187 [ ] [ ] [ 1][0.0]@xmath188 [ ] [ ] [ 1][0.0]@xmath369 [ ] [ ] [ 1][0.0]@xmath370 [ ] [ ] [ 1][0.0]@xmath195 [ ] [ ] [ 1][0.0]@xmath196 for @xmath367 and @xmath368.,title="fig : " ] [ ] [ ] [ 1][0.0]@xmath184 [ ] [ ] [ 1][0.0]@xmath185 [ ] [ ] [ 1][0.0]@xmath186 [ ] [ ] [ 1][0.0]@xmath187 [ ] [ ] [ 1][0.0]@xmath188 [ ] [ ] [ 1][0.0]@xmath298 [ ] [ ] [ 1][0.0]@xmath375 [ ] [ ] [ 1][0.0]@xmath195 [ ] [ ] [ 1][0.0]@xmath196 for @xmath367 and @xmath368.,title="fig : " ] therefore we have @xmath379 where the last equality follows from the fact that @xmath332 has residue @xmath380 at the point @xmath219 .
let us now consider the boundary values @xmath381 and @xmath382 on @xmath350 . since @xmath383 , the integration contours @xmath370 and @xmath369 can now be chosen to lie in the lower ( upper ) half plane of @xmath169 ( @xmath170 ) . the loop @xmath373 can now be deformed into the sum @xmath384 of the @xmath198-cycles .
however , such a deformation will necessarily go pass the poles @xmath385 and @xmath386 of @xmath332 ( recall that by the remark after proposition [ pro : constant ] , there is exactly one point @xmath335 that belongs to @xmath387 ) .
since the residues of @xmath332 at these points are given by @xmath388 from the @xmath389 and @xmath390 from the @xmath337 when all @xmath337 and @xmath391 are distinct , the total residue at these points is zero .
it is clear from definition [ de : deltaj ] that , when some @xmath337 coincide with the @xmath391 , the total residue at these points remains unchanged .
hence we have @xmath392 let us now consider the boundary values on the gaps @xmath344 .
let @xmath393 .
for the boundary values @xmath394 , we choose @xmath395 to be integration contours that go from @xmath353 to @xmath396 in @xmath169 without intersecting @xmath397 except at @xmath353 and @xmath396 .
let the loop @xmath298 be @xmath373 , then for @xmath398 , @xmath298 can be deformed into the @xmath197-cycle @xmath399 without passing any pole of @xmath332 , except possibly @xmath335 ( recall the definition of the @xmath197-periods of @xmath332 in definition [ de : deltaj ] ) .
when @xmath335 is not equal to @xmath400 or @xmath401 , @xmath332 has integer residue at @xmath335 , and when @xmath335 is equal to either @xmath400 or @xmath401 , the deformation from @xmath298 to @xmath399 will not have to go pass @xmath335 .
this implies @xmath402 for @xmath403 , the loop @xmath298 can be deformed into a small loop around the point @xmath404 .
since the 1-forms @xmath332 have integer residues at @xmath404 , we have @xmath405 if @xmath406 , then the loop @xmath298 will be contractible in @xmath376 .
hence we have @xmath407 on the other hand , for the boundary values @xmath408 on @xmath344 , let us consider @xmath395 to be integration contours that go from @xmath353 to @xmath409 in @xmath170 without intersecting @xmath397 except at @xmath353 and @xmath409 .
let @xmath410 , and let @xmath343 be the projection of @xmath335 onto the @xmath202-plane . then depending on the relative positions of @xmath202 , @xmath343 and @xmath400 , the loop @xmath373 can be deformed into @xmath411 , together with small loops around the poles @xmath412 and @xmath413 in @xmath376 ; or it can be deformed into the sum of @xmath411 and small loops around the poles @xmath414 and @xmath413 in @xmath376 . in either cases , the total residue of @xmath332 at these points will be an integer .
therefore we have @xmath415 similarly , for @xmath403 , the loop @xmath298 can be deformed into a small loop around the points @xmath328 and @xmath334 . since the total residue the 1-form @xmath332 at these points is an integer
, we have @xmath416 if @xmath406 , then the loop @xmath298 will be contractible in @xmath376 .
hence we have @xmath417 we now consider the boundary values @xmath368 and @xmath418 at @xmath83 , let @xmath419 .
let us again denote by @xmath369 and @xmath370 contours of integration from @xmath353 to @xmath420 in @xmath171 and @xmath421 in @xmath170 .
then depending on whether @xmath202 is in the upper or lower half plane , the loop @xmath373 can be deformed into to a small loop around the pole @xmath422 or @xmath423 in @xmath376 ( see figure [ fig : gamma ] .
the loop @xmath298 in this case is the same except that it begins and ends at @xmath353 instead of @xmath219 ) . since the residue of @xmath332 around @xmath422 or @xmath423 is @xmath380 , we have @xmath424 the rest of the jump discontinuities in ( [ eq : deltajump ] ) now follow directly from the definition of the integration contours as in the proof of proposition [ pro : thetajump ] . let us denote by @xmath425 the projection of @xmath356 onto the @xmath286-sheet , that is , @xmath426 , where @xmath182 is the function @xmath180 on @xmath181
. then we have the following . [ thm : outer ] let @xmath427 be the @xmath428 matrix whose elements are given by @xmath429 where @xmath356 are defined in ( [ eq : nj ] ) .
suppose we have @xmath430 let the constants @xmath431 be @xmath432 where @xmath327 is the local coordinate near @xmath328 defined in ( [ eq : w ] ) and the limits in @xmath433 , @xmath434 and @xmath435 are taken as @xmath436 in the first quadrant of @xmath170
. then the matrix @xmath437 satisfies the riemann - hilbert problem ( [ eq : outer ] ) , where @xmath438 is the following in the expansion of @xmath439 at @xmath440 @xmath441 as @xmath436 in the first quadrant of @xmath170 .
the constants @xmath431 , @xmath164 can be represented as @xmath442 where the limits are taken as @xmath436 in the first quadrant of @xmath170 .
first note that , by using proposition [ pro : thetajump ] and proposition [ pro : deltajump ] , one can verify that @xmath427 does satisfy the jump discontinuities in ( [ eq : outer ] ) .
since @xmath427 satisfies the jump discontinuities of ( [ eq : outer ] ) , the matrix @xmath443 does not have any jump discontinuities in @xmath151 .
moreover , this matrix does not grow faster than @xmath444 at @xmath167 and has at worst square root singularities at the points @xmath108 and @xmath334 . since it has no jump discontinuities , all these singularities are removable and therefore we have @xmath445 for some constant matrix @xmath446 .
to determine the constant matrix @xmath446 , we will have to study the behavior of @xmath427 as @xmath112 .
the behavior of @xmath117 is given by the following @xmath447 where the @xmath448 lower right block is given by @xmath449 for @xmath112 in the first quadrant .
from the relation between the local coordinate @xmath327 and @xmath202 in ( [ eq : w ] ) and the jump discontinuities of @xmath427 near @xmath328 , we see that , if we can show that the functions @xmath356 behave as @xmath450 when @xmath436 in the first quadrant of @xmath170 , then the matrix in ( [ eq : sol ] ) will be the unique solution of the riemann - hilbert problem ( [ eq : outer ] ) .
the asymptotic behavior of @xmath451 and @xmath452 follows immediately from the definition of the functions @xmath356 ( [ eq : nj ] ) , the constants @xmath431 ( [ eq : lj ] ) and behavior of the 1-forms @xmath332 ( [ eq : deltaj ] ) .
we will now prove the equations in ( [ eq : nbeha ] ) for @xmath439 and @xmath453 .
let the involution @xmath454 on @xmath168 be @xmath455 . to simplify the notation
, we shall simply denote @xmath456 by @xmath457 .
let us consider the functions @xmath458 for @xmath459 .
the singularity structure of this function is the same as @xmath356 . by proposition [ pro : thetajump ] and
[ pro : deltajump ] and the expression of @xmath356 ( [ eq : nj ] ) , we see that the functions @xmath439 and @xmath453 satisfies the following jump discontinuities on @xmath168 .
@xmath460 where @xmath461 is defined in ( [ eq : xipm ] ) , @xmath462 is the interval on @xmath255 that projects to @xmath344 .
that is @xmath463 the intervals @xmath464 are defined to be @xmath465 on the other hand , from ( [ eq : jump23 ] ) , we see that the function @xmath458 has the following jump discontinuities @xmath466 note that the union of the contours @xmath467 and @xmath468 divides @xmath168 into 2 disjoint sets , which are the first sheet @xmath169 and the union of the other sheets @xmath469 . similarly , the contour @xmath470 divides @xmath168 into the sets @xmath471 and @xmath472 .
let @xmath473 be @xmath474 since the constants @xmath135 satisfy the symmetry @xmath475 ( [ eq : symmetry ] ) , from ( [ eq : jump23 ] ) , ( [ eq :- jump23 ] ) and ( [ eq : hatn ] ) we see that the function @xmath476 is either a meromorphic function on @xmath168 with poles exactly at the @xmath183 zeros of @xmath477 or it is a constant . by the assumption of the theorem ,
this theta function is not identically zero .
hence by theorem [ thm : rr ] , we see that @xmath478 must be a constant @xmath479 . by using the jump discontinuities ( [ eq : jump23 ] ) , ( [ eq :- jump23 ] ) of the @xmath356 near @xmath167 , and the relation between the coordinate @xmath138 and @xmath327
, we have the following behavior of @xmath356 near @xmath328 @xmath480 as @xmath436 in the first quadrant of @xmath170 , where the branch cut of @xmath138 is chosen to be the negative real axis . on the other hand , since @xmath481 in the third quadrant when @xmath436 in the first quadrant , the functions @xmath473 have the following behavior @xmath482 as @xmath436 in the first quadrant . on the other hand , since @xmath478 in ( [ eq : tilden ] ) is a constant @xmath479 , we also have @xmath483 as @xmath436 in the first quadrant . by comparing ( [ eq : nhatinf ] ) and ( [ eq : nhatinf2 ] ) , we obtain ( [ eq : nbeha ] ) .
this concludes the proof of the theorem .
we will now prove that the normalization constants @xmath484 , @xmath164 and @xmath485 does not vanish for any @xmath486 . then the solution @xmath487 of the riemann - hilbert problem ( [ eq : outer ] ) constructed in theorem [ thm : outer ] exists and is well - defined .
we will then show that it satisfies the conditions in theorem [ thm : dk ] .
we will use the results in chapter 6 of @xcite .
first let us define a contour @xmath488 that divides the riemann surface @xmath168 into 2 halves .
let @xmath488 be the set of points that is fixed under the map @xmath241 in ( [ eq : phi ] ) .
that is , @xmath489 then @xmath488 is a disjoint union of @xmath490 closed curves @xmath491 , @xmath492 on @xmath168 , given by the followings .
@xmath493 where @xmath494 is the contour on @xmath181 that projects to @xmath130 .
that is @xmath495 in other words , the contours @xmath491 are the closed loops on @xmath168 that start from the branch point @xmath496 , going through the interval @xmath497 $ ] on @xmath169 , then enters @xmath170 at @xmath498 and go back to @xmath496 through the interval @xmath497 $ ] on @xmath170 .
the contour @xmath499 starts at @xmath500 , goes to @xmath501 on the real axis on @xmath169 , then from @xmath502 to @xmath219 on the real axis on @xmath169 , from which it enters @xmath170 and goes to @xmath502 along the real axis on @xmath170 , then goes back from @xmath501 on @xmath170 to @xmath503 along the real axis .
( see figure [ fig : gamma ] ) .
[ ] [ ] [ 1][0.0]@xmath184 [ ] [ ] [ 1][0.0]@xmath185 [ ] [ ] [ 1][0.0]@xmath186 [ ] [ ] [ 1][0.0]@xmath187 [ ] [ ] [ 1][0.0]@xmath188 [ ] [ ] [ 1][0.0]@xmath499 [ ] [ ] [ 1][0.0]@xmath504 [ ] [ ] [ 1][0.0]@xmath505 [ ] [ ] [ 1][0.0]@xmath506 [ ] [ ] [ 1][0.0]@xmath195 [ ] [ ] [ 1][0.0]@xmath196 .,title="fig : " ] note that the images of the cuts @xmath129 on @xmath169 and @xmath170 do not belong to @xmath488 . for example
, let @xmath507 be a point on @xmath508 , then @xmath509 similarly , the images of the real axis on @xmath171 and @xmath172 do not belong to @xmath488 either .
the curve @xmath488 divides the riemann surface @xmath168 into 2 halves , @xmath510 and @xmath511 , each of which is an open riemann surface with boundary @xmath488 .
the riemann surface @xmath512 consists of the upper ( lower ) half planes of @xmath169 , @xmath170 and @xmath171 and the lower ( upper ) half plane of @xmath172 .
the riemann surface @xmath168 can now be thought of as a union of @xmath510 , @xmath511 and @xmath488 .
moreover , the @xmath198-cycles defined in figure [ fig : cycle ] is homologic to the contours @xmath491 .
that is , we have @xmath513 we can think of @xmath168 as the riemann surface formed by gluing two copies of @xmath510 along the boundary @xmath488 with an anti - holomorphic involution @xmath241 that fixes @xmath488 and maps @xmath510 onto @xmath511 .
a riemann surface formed in this way is called a schottky double .
since @xmath168 is a schottky double , we can apply the results in chapter 6 of @xcite to the theta function of @xmath168 .
[ pro : prefay2 ] ( proposition 6.4 of @xmath520 ) for any point @xmath521 , @xmath522 , the function @xmath523 either vanishes identically or has modulo 2 , @xmath524 zeros on @xmath525 , where @xmath526 is the @xmath286 component of the vector @xmath527 .
let us take @xmath529 , @xmath530 and @xmath531 in proposition [ pro : prefay2 ] , then @xmath532 and by the paragraph after lemma [ le : imag ] , we see that @xmath249 is not identically zero and hence by proposition [ pro : prefay2 ] , it has @xmath533 zero on each of the contour @xmath525 , @xmath410 .
[ pro : fay1 ] ( corollary 6.13 of @xmath520 ) let @xmath534 be the universal covering of @xmath535 , @xmath536 then the theta function @xmath537 is real and positive for @xmath538 .
that is , @xmath537 is real and positive for all @xmath539 .
we can now prove that the periods @xmath341 of @xmath332 in definition [ de : deltaj ] are purely imaginary .
this , together with proposition [ pro : fay1 ] will imply the non - vanishing of the theta functions @xmath484 for @xmath485 and @xmath164 .
first note that , by corollary [ pro : fay2 ] , all the points @xmath337 and @xmath389 are invariant under the involution @xmath241 .
hence the meromorphic 1-form @xmath540 has the same poles and residues as @xmath541 .
let us show that the @xmath198-periods of @xmath542 are zero .
we have @xmath543 [ ] [ ] [ 1][0.0]@xmath184 [ ] [ ] [ 1][0.0]@xmath185 [ ] [ ] [ 1][0.0]@xmath186 [ ] [ ] [ 1][0.0]@xmath187 [ ] [ ] [ 1][0.0]@xmath188 [ ] [ ] [ 1][0.0]@xmath544 [ ] [ ] [ 1][0.0]@xmath195 [ ] [ ] [ 1][0.0]@xmath196 to @xmath192 when @xmath545 .
the deformations for other cycles are similar.,title="fig : " ] from figure [ fig : cycle ] , we see that the curve @xmath546 consists of a path from the lower half plane in @xmath169 that goes from @xmath547 to @xmath548 , and another path in the upper half plane of @xmath170 that goes from @xmath548 to @xmath547 .
there are 3 poles of @xmath332 between the loops @xmath549 and
@xmath546 : @xmath335 , @xmath400 and @xmath401 ( see figure [ fig : phia ] ) . from ( [ eq : deltaj ] ) , we see that @xmath332 has a combined residue of 0 at these points , and hence we can deform @xmath546 onto @xmath549 without affecting the value of ( [ eq : deltaper ] ) . therefore , by ( [ eq : deltaper ] ) , we see that @xmath550 by the uniqueness of normalized 1-form , this implies @xmath551 .
now we use ( [ eq : involcyc ] ) for the @xmath197-periods , since the relations for the @xmath197-cycles in ( [ eq : involcyc ] ) are exact and not up to deformation , we have @xmath552 where @xmath365 is the @xmath286 component of the vector @xmath341 . on the other hand , since @xmath551 , the above is also equal to @xmath365 .
this implies the proposition .
let us consider the normalized 1-form @xmath555 that has simple poles at @xmath219 and @xmath556 with residues -1 and 1 for @xmath485 .
then by similar argument used in the proof of lemma [ pro : purim ] , we see that @xmath557 and hence the @xmath197-periods of @xmath555 are all purely imaginary .
now by the riemann bilinear formula ( [ eq : rb ] ) and the definition of the abel map ( [ eq : abel ] ) , we see that @xmath558 hence @xmath559 , @xmath485 are real . therefore , by proposition [ pro : fay1 ] and lemma [ pro : purim ] , we see that @xmath560 . by the periodicity of the theta function ( [ eq : period ] ) , we see that the theta function is in fact a map from @xmath561 , where @xmath562 is the torus @xmath563 . by proposition
[ pro : fay1 ] , the restriction of the theta function on the compact set @xmath564 is real and positive and hence there exists @xmath553 such that @xmath565 for all @xmath566 . this then implies the theorem .
let us first show that the function @xmath427 in ( [ eq : nmatrix ] ) is bounded in @xmath5 uniformly in @xmath152 , where @xmath152 is defined in ( [ eq : calt ] ) . since the entries of @xmath427 are restrictions of the functions @xmath356 in ( [ eq : nj ] ) on different sheets of the riemann surface , we only need to show that @xmath356 is bounded inside the set @xmath568 in @xmath168 that projects onto @xmath152 . from the periodicity property of the theta function ( [ eq : period ] )
, we see that @xmath356 can be written as @xmath569 where @xmath570 is a finite vector given by @xmath571,\quad l=1,\ldots , g,\ ] ] where @xmath572 $ ] is the biggest integer that is smaller than @xmath573 . from ( [ eq : nmod ] ) and
the fact that @xmath249 is not identically zero ( proposition [ pro : constant ] ) , we see that @xmath356 is bounded in @xmath5 uniformly in @xmath574
. we will now show that the constants @xmath575 in ( [ eq : lexp ] ) are bounded in @xmath5 . from the singularity behavior of the meromorphic 1-forms @xmath332 in ( [ eq : deltaj ] )
, we see that following constants @xmath576 in ( [ eq : lexp ] ) are all bounded and non - zero . since they are all independent on @xmath5 , they are also bounded away from infinity and zero as @xmath4 . by proposition
[ pro : constant ] , we see that @xmath249 is not identically zero and will only vanish at the points @xmath337 that belong to @xmath577 . since neither @xmath404 nor @xmath328 belongs to @xmath577 for @xmath578 ,
the constants @xmath579 , @xmath485 are non - zero . moreover , from the definition of the abel map ( [ eq : abel ] ) , we see that @xmath580 and @xmath581 are both finite and hence @xmath582 and @xmath583 are both bounded and are independent on @xmath5 .
let us now consider the factors @xmath484 for @xmath485 and @xmath164 .
by theorem [ thm : nonvan ] , there exists @xmath553 , independent on @xmath5 such that these constants are greater than @xmath584 . on the other hand , from the periodicity of the theta function ( [ eq : period ] ) and the fact that the period matrix @xmath209 is purely imaginary , ( lemma [ le : imag ] ) while the vector @xmath585 in ( [ eq : apj ] ) is real , we see that @xmath484 is bounded in @xmath5 as @xmath4 . hence the constants @xmath575 in ( [ eq : sol ] ) and ( [ eq : lexp ] ) are bounded away from infinity and zero as @xmath4 . finally , by considering the asymptotic expansion of @xmath356 in the local parameter @xmath327 in ( [ eq : w ] ) at @xmath167 and making use of ( [ eq : nmod ] )
, we see that @xmath438 in ( [ eq : sol ] ) is bounded in @xmath5 as @xmath4 .
since all the constants @xmath431 and @xmath438 are bounded in @xmath5 as @xmath4 and that all the @xmath356 are bounded in @xmath5 uniformly in @xmath574 , we see that @xmath487 is also bounded in @xmath5 uniformly in @xmath152 . to see that this is also the case for the inverse @xmath567 ,
let us consider the determinant of @xmath487 .
since @xmath487 is a solution to the riemann - hilbert problem ( [ eq : outer ] ) , the determinant @xmath586 has no jump discontinuity in @xmath151 and it behaves as @xmath587 as @xmath112 . from the expression of @xmath427 in ( [ eq : nmatrix ] ) , we see that at @xmath588 , only the first and second columns of @xmath427 have fourth - root singularities , while at the points @xmath86 , only the second and the third columns of @xmath427 have fourth - root singularities .
therefore the determinant of @xmath487 can at worst have square - root singularities at these points .
since @xmath586 has no jump discontinuities in @xmath151 , we see that @xmath586 can not have square - root singularities at these points .
hence @xmath586 is holomorphic in the whole complex plane . by liouville
s theorem , this implies that @xmath589 .
since the entries of @xmath567 are degree 3 polynomials in the entries of @xmath487 divided by @xmath589 , we see that the entries of @xmath567 are also bounded in @xmath5 uniformly in @xmath152 . finally , by considering the asymptotic expansion of @xmath356 in the local parameter @xmath327 in ( [ eq : w ] ) at @xmath167 and making use of ( [ eq : nmod ] ) , it is easy to see that condition 2 . in theorem
[ thm : dk ] is satisfied for @xmath487 and its inverse .
m. bertola , b. eynard and j. harnad .
differential systems for biorthogonal polynomials appearing in 2-matrix models and the associated riemann - hilbert problem .
_ * 243 * ( 2003 ) , no .
2 , 193240 .
p. deift , t. kriecherbauer , k. t. r. mclaughlin , s. venakides and x. zhou .
strong asymptotics of orthogonal polynomials with respect to exponential weights .
. pure appl .
_ , * 52 * ( 1999 ) , no . 12 , 14911552 .
p. deift , t. kriecherbauer , k. t. r. mclaughlin , s. venakides and x. zhou .
uniform asymptotics for polynomials orthogonal with respect to varying exponential weights and applications to universality questions in random matrix theory .
pure appl . math .
_ , * 52 * ( 1999 ) , no . 11 , 13351425
. p. di francesco .
2d topological and quantum gravities , matrix models and integrable differential systems . in : `` the painlev property '' ,
( r. conte ed . ) crm ser .
phys . , springer , new york , ( 1999 ) , 229285 . | in this paper we studied the asymptotic eigenvalue statistics of the 2 matrix model with the probability measure @xmath0 in the case where @xmath1 and @xmath2 is a general even polynomial .
we studied the correlation kernel for the eigenvalues of the matrix @xmath3 in the limit as @xmath4 .
we extended the results of duits and kuijlaars in @xcite to the case when the limiting eigenvalue density for @xmath3 is supported on multiple intervals .
the results are achieved by constructing the parametrix to a riemann - hilbert problem obtained in @xcite with theta functions and then showing that this parametrix is well - defined for all @xmath5 by studying the theta divisor . |
the last several decades have seen a crumbling of the barrier between the quantum and macroscopic realms .
high-@xmath0 superconductivity provides an exquisite example of long range quantum coherence at temperatures above 77 k. cooperative quantum tunneling has also emerged as an important class of phenomena , whose manifestations include coherent josephson tunneling @xcite and tunneling of vortices , @xcite which can occur more readily in layered high-@xmath0 superconductors due to their pancake - like structures . @xcite vortices are known to nucleate in superfluid helium above a critical velocity , @xcite underscoring the point that magnetic flux is not crucial to vortex nucleation . in fact , as will be discussed , the magnetostatic energy contained within a superconducting flux vortex actually impedes vortex nucleation for applied currents below a critical value .
both abrikosov and josephson vortices , the latter being especially important in high-@xmath0 superconducting ( hts ) grain boundaries , play important roles in limiting the critical currents of hts films and tapes .
a josephson vortex is a 2@xmath3 flux soliton in a wide josephson junction ( wide compared to the josephson penetration length @xmath4 ) , whose energy includes the extra josephson coupling energy in the region where the phase difference @xmath5 across the junction advances by 2@xmath3 from one minimum to the next .
though spatially extended , josephson vortices are extremely light . using the expression provided by grosfeld and stern , @xcite the effective mass of a josephson vortex spanning a 1-@xmath6 m thick film
is estimated to be @xmath7 .
the mass of a single pancake josephson vortex ( josephson pancake @xcite ) within the @xmath8 superconducting cuprate or pnictide layer , whose phase difference @xmath9 advances by 2@xmath3 , is several orders of magnitude smaller still .
an applied current in zero external magnetic field can induce a vortex - antivortex pair to nucleate in close proximity by quantum tunneling @xcite and subsequantly expand outward , as illustrated in fig .
[ fig1](a ) .
conceptually , this process is analogous to electron - hole pair creation via landau - zener tunneling through a tilted bandgap in a semiconductor or schwinger pair production of electron - positron pairs out of the vacuum .
@xcite in this case , the applied current @xmath10 acts as the driving force . neglecting the magnetostatic energy and using the expression derived by cohen and mcgady @xcite for the pair production rate , one finds the voltage across a wide grain boundary jj to be proportional to @xmath10 exp[-@xmath11 .
although highly nonlinear , this lacks a sharply defined threshold current below which the voltage is identically zero .
however , recent experiments probing hts bi - crystal grain boundaries with femtovolt precision @xcite suggest the existence of a sharp critical current below which no josephson vortices nucleate , either in the middle or at the edges of the grain boundary .
the coulomb blockade effect is well known for single electron tunneling , @xcite and has also been proposed for charge soliton pair creation in the massive schwinger model , @xcite essentially a quantum sine - gordon model that includes electrostatic effects , and in density waves .
@xcite here we propose a _
magnetic _ blockade threshold current for abrikosov or josephson vortex pair creation @xcite in hts films and grain boundaries .
@xcite this threshold , for josephson vortex nucleation in a grain boundary , can be much smaller than the ambegaokar baratoff critical current even when _
d_wave pairing symmetry is taken in to account .
@xcite a magnetic weber " blockade effect has also recently been proposed to interpret magneto - conductance oscillations of thin , narrow superconducting strips .
@xcite figure [ fig1](a ) illustrates nucleation , in zero applied magnetic field , of a pair of josephson vortices in a wide hts grain boundary .
the vortex pair generates a magnetic field ( yellow shading ) : @xmath12 that links the vortex and the anti - vortex . here
@xmath13 or @xmath14 for a josephson or abrikosov vortex pair , respectively , @xmath15 is the flux quantum , @xmath16 is the london penetration length , and @xmath17 is a geometrical factor .
a current density @xmath18 in a film of thickness @xmath19 creates a magnetic field , @xmath20 , which can partially cancel the internal field linking the vortex and antivortex and thus overcome the magnetic barrier ( coulomb barrier analog ) , provided its value is high enough . at low applied currents , the difference in magnetostatic energy densities with and without the vortex pair , @xmath21 , is positive when @xmath22 .
however , when @xmath23 , the formerly lowest energy state becomes a metastable state or `` false vacuum , '' as illustrated in fig .
[ fig1](b ) .
the phases @xmath9 tunnel coherently into the lower energy well , creating a bubble of `` true vacuum '' bounded by flux solitons ( josephson vortices ) when @xmath2 exceeds @xmath3 . setting @xmath24
thus yields the pair creation threshold current per unit width : @xmath25 for the case of josephson vortex nucleation , where @xmath26 , this yields a scaling with classical critical current density @xmath27 of the form , @xmath28 , since @xmath29 for both planar and sandwich - type junctions .
@xcite for currents below the vortex nucleation threshold , @xmath2 can be written in terms of the supercurrent per unit width @xmath30 as @xmath31 .
if the vortex and antivortex are separated by a distance @xmath32 , then the pair creation critical current between vortices can be written as @xmath33 , where @xmath34 is the inductance of the flux toroid coupling the vortices .
this is essentially the dual of the coulomb blockade voltage @xmath35 for a small capacitance tunnel junction .
@xcite nucleation of a vortex near one edge and antivortex near the other can be treated similarly using the image vortex concept .
@xcite when @xmath36 as the phases @xmath37 tunnel coherently into the lower well .
* ( c)*. potential energy vs. @xmath2 for @xmath38 . *
( d)*. modified resistively shunted junction model , in which time - correlated tunneling of josephson vortices and antivortices is represented by analogy to time - correlated set .
@xcite[fig1 ] ] hts coated conductors generally have small - angle grain boundaries @xcite and their critical currents vs. thickness @xmath19 often plateau , so that @xmath39 decreases with @xmath19 .
@xcite this is readily explained by the vortex pair creation current density , @xmath40 when @xmath41 , resulting from the magnetic blockade effect . for a strip of width @xmath42 ,
the pair creation current becomes : @xmath43 , which is independent of @xmath19 until it becomes large enough for vortex rings to nucleate .
when @xmath44 , the size of the vortices increases since the effective 2-d penetration length @xmath45 scales inversely with @xmath19 : @xcite @xmath46 .
since @xmath47 when @xmath48 , we use the approximation : @xmath49 , which yield the vortex pair creation critical current : @xmath50 ^ 2}.\label{eq : eq1}\ ] ] figure [ fig2 ] shows a favorable comparison of eq .
( [ eq : eq1 ] ) with measured critical currents vs. thickness of hts coated conductors .
the introduction of multiple insulating ceo@xmath51 spacer layers into hts coated conductors has been found to increase critical current .
@xcite our model provides a straightforward interpretation
sufficiently thick spacer layers decouple vortex pair nucleation events on adjacent hts layers , enabling @xmath52 to increase with the number of layers .
= 380 nm , @xmath53= 1030 a and 398 a to fit 2008 and 2005 data , respectively.)[fig2 ] ]
the `` vacuum '' or theta angle @xmath2 , discussed extensively in the quantum field theory literature ( e.g. see ref . @xcite and citing papers ) is related in our model to total displacement flux @xmath54 by : @xmath55 , where @xmath56 .
the potential energy of the @xmath8 layer for a cuprate or pnictide superconductor can then be written , similar to the massive schwinger model , @xcite as : @xmath57=u_{_j}\left[1-cos\phi_k(x)\right]+u_{_m}\left(\theta-\phi_k(x)\right)^2~,\label{eq : eq2}\ ] ] where the first term is the josephson coupling energy and the quadratic term is the magnetostatic contribution .
figure [ fig1](c ) shows plots of @xmath58 vs. @xmath2 when the energy is minimized for @xmath59 ( dropping the subscript @xmath60 ) when @xmath61 .
each @xmath62 branch has a parabolic form , @xmath63 , analogous to the charging energy parabolas , @xmath64 , in time - correlated set @xcite and time - correlated soliton tunneling .
@xcite the phases @xmath9 tunnel coherently into the next well as each parabola , or branch , in fig .
[ fig1](c ) crosses the next at @xmath1 and the other instability points @xmath65 .
regardless of the detailed shape of the josephson coupling energy ( which may or may not be sinusoidal and may even include disorder ) , the behavior that emerges is clearly nonsinusoidal in nature , which affects the shapiro steps in the @xmath66 curve in the presence of a microwave signal .
the approach proposed here is based on the hypothesis that the amplitudes @xmath67 and @xmath68 for the system to be on branches @xmath69 and @xmath70 , respectively , are coupled via coherent tunneling of josephson pancakes , represented as microscopic quantum solitons .
@xcite this picture is motivated by feynman s intuitive derivation @xcite of the dc and ac josephson effects .
advancing @xmath37 by @xmath71 in a finite region is equivalent to creating a pair of microscopic solitons essentially pancake josephson vortices @xcite with extremely small effective masses .
@xcite similarly , abrikosov pancake vortices @xcite are pointed out to have small effective masses .
@xcite in this picture , many coupled pancake vortices behave as a quantum fluid , @xcite within which their coherent tunneling is viewed as a secondary josephson effect .
we use a modified version of the resistively shunted junction ( rsj ) model [ fig .
[ fig1](d ) ] , in which the analogy to time - correlated single electron tunneling is used to simulate time - correlated vortex tunneling and obtain voltage - current characteristics of an hts grain boundary junction .
the voltage across the junction is : @xmath72 , yielding a normal current through the shunt resistance : @xmath73 . for a grain boundary junction with total current @xmath10 , we define @xmath74 . in a wide junction
, advancing the phase by @xmath75 ( the average phase for all layers ) near the middle of the junction creates a josephson kink - antikink pair whose circulating currents either reduce or reinforce the existing supercurrent between kinks .
this yields a net supercurrent given by : @xmath76 $ ] . using @xmath77 and introducing the time constant @xmath78 , yields the following equation for the time evolution of @xmath2 : @xmath79.\label{eq : eq3}\ ] ] the expectation value @xmath75 is computed by solving the time - dependent schrdinger equation : @xmath80 viewed as an emergent classical equation following feynman , @xcite to compute the original and emerging macrostate amplitudes @xmath81 and @xmath82 .
these are coupled via a tunneling matrix element @xmath83 with a zener - like force dependence , which transfer the phase @xmath9 from one well to the next [ fig .
[ fig1](b ) ] .
our model represents the amplitudes @xmath84 for the system to be on branches 0 and 1 [ fig .
[ fig1](c ) , more generally @xmath69 and @xmath70 ] by : @xmath86 for a layered superconductor , @xmath87 is the fraction of superconducting layers on the respective branch . in this picture
@xmath88 are @xmath89 this superconducting order parameters on opposite sides of the junction but , rather , represent the amplitudes for the phases @xmath9 to be in either of the two wells shown in fig .
[ fig1](b ) .
the phases @xmath90 are the berry phases for the vortices , which would , for example , lead to an aharonov - casher effect for vortices traveling along two branches of a ring - shaped josephson junction surrounding an island charge .
@xcite advancing @xmath37 by 2@xmath3 within a given region , taking @xmath9 from one branch to the next , is equivalent to creating a pair of microscopic 2@xmath3-solitons ( josephson vortices ) .
the driving force is the energy difference per unit length between potential minima at @xmath59 and @xmath91 . when @xmath92 , this force is : @xmath93 , where @xmath94 .
following bardeen s procedure , @xcite the matrix element for zener tunneling through the soliton energy gap is estimated as : @xmath95~,\label{eq : eq6}\ ] ] where @xmath96 , @xmath97 is a mean free path length , @xmath98 is the microscopic soliton energy for a pancake josephson vortex , @xmath99 is the phase ( swihart ) velocity , @xcite and @xmath100 .
this expression for @xmath101 is similar to the rate of zener tunneling or schwinger pair production in 1-d .
@xcite since any negative energy difference within the `` bubble '' is balanced by the positive flux soliton pair energy , the matrix element couples states of equal energy , @xmath102 .
thus , defining @xmath103 $ ] , the schrdinger equation [ eq . ( [ eq : eq4 ] ) ] reduces to : @xmath104 below the pair creation threshold , we can write : @xmath105 where @xmath106 within a grain boundary region of width @xmath32 , which may be slightly smaller than @xmath42 due to the finite width of the vortices . due to multiple vortex pair nucleation events in a wide junction
, @xmath107 may not decrease monotonically with @xmath108 .
the variables are put into dimensionless form : @xmath109 , @xmath110 , proportional to total current @xmath10 , @xmath111 , where @xmath112 , @xmath113 , and @xmath114 is the normalized supercurrent , and the normalized voltage is : @xmath115 . finally , setting @xmath116 and @xmath117 in eq .
( [ eq : eq7 ] ) , taking @xmath118 and @xmath119 to be real , yields the following coupled differential equations : @xmath120 c_0~,\\ \frac{dc_0}{dt'}&= -[{\gamma q_n}'\exp(-q_0/{q_n } ' ) ] c_1~ , \end{aligned } \label{eq : eq8}\ ] ] for @xmath121 and where @xmath122 .
these are integrated numerically with initial values @xmath123 and @xmath124 .
the phase expectation value is : @xmath125 $ ] , where @xmath126 , and the transition is considered complete when @xmath127 exceeds a cutoff close to one . for each value of @xmath128 , a time average over several complete cycles is performed to compute @xmath129 .
a similar approach has recently been used by the authors to compute current - voltage characteristics due to soliton pair production in density waves .
@xcite as in the density wave case , each transition from one branch to the next takes place over a relatively long time interval , suggesting that many pancake vortices `` flow '' through the barrier like a quantum fluid rather than tunneling abruptly as a single massive flux line .
figure [ fig22 ] shows resulting theoretical normalized voltage vs. current plots for several values of @xmath130 for a fixed value of @xmath131 and @xmath132 .
the _ v - i _ plots exhibit piecewise linear behavior , which fits neither the classical rsj model nor a straightforward thermally activated flux flow model .
similar piecewise quasi - linear behavior is seen quite frequently in high-@xmath0 superconducting ybco grain boundary junctions .
the main effect of increasing @xmath130 is to increase the slope of the _ v - i _ curve , saturating at a slope of one for the normalized plot .
figure [ fig23 ] shows similar plots for a fixed value of @xmath133 and several values of @xmath134 . here , the main effect of increasing @xmath134 , which corresponds to increasing the energy required to nucleate each vortex - antivortex pair , is to increase the degree of rounding in the _ v - i _ plot .
moreover , note that the simulated `` measured '' critical current @xmath52 increases to well above the nominal magnetic blockade critical current @xmath135 as @xmath134 becomes significantly greater than one .
and several values of @xmath130.[fig22 ] ] and several values of @xmath134.[fig23 ] ] figure [ fig3](a ) shows comparisons between theory and the measured _ v - i _ characteristics of a ybco @xcite grain boundary junction at several temperatures . the simulated 86 k plot in fig .
[ fig3](a ) ( top ) was obtained using the classical rsj model in the overdamped limit : @xmath136 , without invoking thermal activation .
the fact that the 86 k data fits the classical rsj model almost perfectly suggests that the effective josephson penetration length is comparable to or longer than the junction width ( short junction limit ) due to the small josephson coupling energy at this temperature . the theoretical plots for the remaining temperatures in fig .
[ fig3](a ) , as well as for figs .
[ fig3](b ) and [ fig3](c ) , were obtained from the schrdinger equation as discussed above , using the parameters shown in table [ table1 ] .
the increase in @xmath134 going from 82.5 k down to 70.0 k is consistent with increasing josephson coupling energy , leading to a larger @xmath137 and smaller @xmath4 , as the temperature decreases .
the increased rounding of the _ v - i _ curves with decreasing @xmath138 is also consistent with the system going from the short- to the long - junction limit as @xmath4 decreases .
moreover , the fact that the _ v - i _ curves are more rounded for low than for high temperatures provides powerful evidence that the rounding of the _ v - i _ curves is _ not _ due to thermal activation .
the extremely light masses of josephson vortices coupled with quantum fluidic properties , discussed above , enable quantum effects to dominate over the entire temperature range .
further evidence that even abrikosov vortices exhibit quantum , rather than thermally activated , behavior over a wide temperature range is provided by magnetic relaxation rates of trapped flux in ybco , which go _ down _ with increasing temperature @xcite from 4 k to about 86 k. ] figure [ fig3](b ) shows excellent agreement between our quantum model and the _ v - i _ curve of an iron pnictide superconducting bicrystal , @xcite consisting of coupled srfe@xmath139co@xmath140as@xmath51 and ba@xmath141k@xmath142fe@xmath51as@xmath51 crystals , each @xmath143300 @xmath6 m wide , and thus in the long junction limit .
similar piecewise linear behavior to that seen in the figure also occurs frequently in cuprate grain boundary junctions [ e.g. fig .
[ fig3](a ) , 77.2 k data and ref . @xcite ] and is analogous to the piecewise linear _
i - v _ curve of an ideal coulomb blockade tunnel junction .
figure [ fig3](c ) shows a plot of our simulation as compared to the measured _ v - i _ characteristic of a thallium - based cuprate , @xcite exhibiting good agreement with the data within experimental error .
table [ table1 ] shows the parameters used for the simulations .
.[table1]parameters used to generate the simulated _ v - i _ curves in fig . [ fig3 ] . [ cols="^,^,^,^,^ " , ] the use of the time - dependent schrdinger equation and generalized tunneling matrix element @xmath101 , coupled with the analogy to time - correlated single electron tunneling suggested in fig .
[ fig1 ] , provides a simple , yet powerful approach to modeling the dynamics of vortex tunneling . the top plot in fig .
[ fig4 ] shows predicted voltage oscillations , exhibiting highly non - sinusoidal behavior , for a uniform junction with the parameters @xmath144 and @xmath145 , for an applied current @xmath146 . and
@xmath145 ; and nonuniform junctions , represented as 100 junctions in parallel using the same average @xmath130 and @xmath134 , but ( see text ) with @xmath147 , @xmath148 , and @xmath149 .
[ fig4 ] ] in the remaining plots of fig .
[ fig4 ] we model non - uniformities by representing the junction as 100 junctions in parallel , for which @xmath130 and @xmath134 are psuedo - randomly varied with a uniform distribution within the ranges @xmath150 and @xmath151 , respectively . the phases between adjacent domains @xmath152 are coupled via an additional term that adds to the net force for each domain : @xmath153 $ ] , applying periodic boundary conditions .
the remaining plots in fig .
[ fig4 ] also assume @xmath154 and the same average values for @xmath130 and @xmath134 as before , but with different values of @xmath155 and @xmath60 .
the middle two plots , for which @xmath156 and @xmath157 ( second from top ) and @xmath158 , show greatly reduced voltage oscillation amplitudes as well an apparent amplitude modulation effect consistent with reports of subharmonic shapiro steps .
@xcite the bottom plot ( @xmath159 and @xmath158 ) , representing the highest degree of disorder for this series , shows oscillation amplitudes that are even further reduced .
this is consistent with the idea that shapiro steps would essentially disappear in a highly nonuniform junction .
a key premise of this paper is that the concept of coherent josephson - like tunneling should be generalized to include other phenomena with quantum fluidic properties . rather than treating the system as a massive object and computing the total euclidean action or bounce , " a generalized tunneling hamiltonian matrix element
is introduced to connect the original and emerging macrostates .
the `` disconnectivity '' between macrostates , introduced by leggett , @xcite is extremely small as in josephson tunneling .
this notion is further supported by examination of schrieffer s real - space description of the bcs ground state : @xcite @xmath160 , where @xmath161 is the relative coordinate wavefunction of a pair , @xmath162 is the corresponding spin function , and @xmath163 is the antisymmetrization operator .
as @xmath164 evolves into a state , @xmath165 , containing a vortex - antivortex pair , each pair state individually evolves into a deformed state @xmath166 topologically connected to @xmath161 . beyond improving our understanding of superconducting flux vortex nucleation and dynamics , the concepts proposed here could have scientific impact in other areas , such as @xmath167 instabilities in spontaneous cp violation , @xcite charge and spin density waves , @xcite topological insulators , @xcite and the quantum hall effect .
@xcite finally , better understanding of the quantum behavior of flux solitons could potentially lead to topologically robust forms of quantum information processing .
the authors acknowledge support by the state of texas through the texas center for superconductivity at the university of houston .
additional support was provided by r21ca133153 from nih ( nci ) and by arra supplement : 3 r21 ca133153 - 03s1 ( nih , nci ) .
references b. d. josephson , physics letters * 1 * , 251 ( 1962 ) .
g. blatter , v. b. geshkenbein , and v. m. vinokur , physical review letters * 66 * , 3297 ( 1991 ) .
p. ao and d. j. thouless , physical review letters * 72 * , 132 ( 1994 ) .
a. wallraff , a. lukashenko , j. lisenfeld , a. kemp , m. v. fistul , y. koval , and a. v. ustinov , nature * 425 * , 155 ( 2003 ) .
e. m. chudnovsky , and j. tejada , _ macroscopic quantum tunneling of the magnetic moment _
( cambridge university press , cambridge ; new york , 1998 ) , p. 173
. g. g. ihas , o. avenel , r. aarts , r. salmelin , and e. varoquaux , physical review letters * 69 * , 327 ( 1992 ) .
e. grosfeld and a. stern , proceedings of the national academy of sciences * 108 * , 11810 ( 2011 ) . r. g. mints and i. b. snapiro , physical review b * 49
* , 6188 ( 1994 ) .
a. widom and y. srivastava , physics letters a * 114 * , 337 ( 1986 ) .
t. d. cohen and d. a. mcgady , physical review d * 78 * , 036008 ( 2008 ) .
b. kalisky , j. r. kirtley , e. a. nowadnick , r. b. dinner , e. zeldov , ariando , s. wenderich , h. hilgenkamp , d. m. feldmann , and k. a. moler , applied physics letters * 94 * , 202504 ( 2009 ) .
d. v. averin and k. k. likharev , journal of low temperature physics * 62 * , 345 ( 1986 ) .
s. coleman , annals of physics * 101 * , 239 ( 1976 ) .
i. v. krive and a. s. rozhavsky , solid state communications * 55 * , 691 ( 1985 ) .
j. h. miller , c. ordez , and e. prodan , physical review letters * 84 * , 1555 ( 2000 ) .
j. h. miller , jr .
, a. i. wijesinghe , z. tang , and a. m. guloy , physical review letters * 108 * , 036404 ( 2012 ) . h. hilgenkamp and j. mannhart , reviews of modern physics * 74 * , 485 ( 2002 ) . j. h. xu , j. l. shen , j. h. miller , jr . , and c. s. ting , physical review letters * 73 * , 2492 ( 1994 ) .
d. pekker , g. refael , and p. m. goldbart , physical review letters * 107 * , 017002 ( 2011 ) .
s. k. tolpygo and m. gurvitch , applied physics letters * 69 * , 3914 ( 1996 ) .
a. a. abrikosov , ( north - holland , elsevier science publishers , amsterdam , 1988 ) , p. 630
. s. r. foltyn , l. civale , j. l. macmanus - driscoll , q. x. jia , b. maiorov , h. wang , and m. maley , nature materials * 6 * , 631 ( 2007 ) . j. pearl , applied physics letters * 5 * , 65 ( 1964 )
. v. selvamanickam et al . , applied superconductivity , ieee transactions on * 19 * , 3225 ( 2009 ) .
a. maiti and j. h. miller , physical review b * 43 * , 12205 ( 1991 ) .
z. hermon , a. shnirman , and e. ben - jacob , physical review letters * 74 * , 4915 ( 1995 ) .
r. p. feynman , r. b. leighton , and m. sands , _ the feynman lectures on physics _
( addison - wesley , reading , massachusetts , 1965 ) , vol .
a. grigorenko , s. bending , t. tamegai , s. ooi , and m. henini , nature * 414 * , 728 ( 2001 ) .
a. e. koshelev , physical review letters * 77 * , 3901 ( 1996 ) .
w. j. elion , j. j. wachters , l. l. sohn , and j. e. mooij , physical review letters * 71 * , 2311 ( 1993 ) .
j. bardeen , physical review letters * 6 * , 57 ( 1961 ) . c. b. duke , _ tunneling in solids _
( academic press , new york , 1969 ) , p. xii + 356 .
j. swihart , j. appl . phys . * 32 * , 461 ( 1961 ) .
r. d. redwing , b. m. hinaus , m. s. rzchowski , n. f. heinig , b. a. davidson , and j. e. nordman , applied physics letters * 75 * , 3171 ( 1999 ) . y. y. xue , z. j. huang , p. h. hor , and c. w. chu , physical review b * 43 * , 13598 ( 1991 ) .
x. zhang , s. r. saha , n. p. butch , k. kirshenbaum , j. paglione , r. l. greene , y. liu , l. yan , y. s. oh , k. h. kim , and i. takeuchi , applied physics letters * 95 * , 062510 ( 2009 ) .
d. dimos , p. chaudhari , and j. mannhart , physical review b * 41 * , 4038 ( 1990 ) .
f. lombardi , z. g. ivanov , p. komissinski , g. m. fischer , p. larsson , and t. claeson , applied superconductivity * 6 * , 437 ( 1998 ) . c. dark , s. speller , h. wu , a. sundaresan , y. tanaka , g. burnell , and c. r. m. grovenor , applied superconductivity , ieee transactions on * 15 * , 2931 ( 2005 ) . d. terpstra , r. p. j. ijsselsteijn , and h. rogalla , applied physics letters * 66 * , 2286 ( 1995 ) .
a. j. leggett , progress of theoretical physics supplement * 69 * , 80 ( 1980 ) .
j. r. schrieffer , _ theory of superconductivity _
( benjamin cummings , reading , massachusetts , 1964 ) , p. 332 .
m. h. g. tytgat , physical review d * 61 * , 114009 ( 2000 ) .
d. boer , and j. k. boomsma , physical review d * 78 * , 054027 ( 2008 ) .
k.t . chen and p.
a. lee , physical review b * 83 * , 125119 ( 2011 ) .
a. m. m. pruisken , r. shankar , and n. surendran , physical review b * 72 * , 035329 ( 2005 ) . | we propose a magnetic dual of the coulomb blockade effect for quantum nucleation of flux vortex pairs in high-@xmath0 superconducting ( hts ) films and grain boundaries in zero applied field .
the magnetic blockade instability occurs at @xmath1 , where @xmath2 is the `` vacuum '' or theta angle .
the @xmath2 term has recently been discussed in the context of several other systems , including charge and spin density waves , topological insulators , the quantum hall effect , and spontaneous cp violation .
our model predicts a sharp pair creation threshold current at @xmath1 , analogous to the coulomb blockade voltage of a tunnel junction , and explains the observed thickness dependence of critical currents in hts coated conductors .
we use the schrdinger equation to compute the evolving macrostate amplitudes , coupled by a generalized tunneling matrix element .
the simulations yield excellent quantitative agreement with measured voltage - current characteristics of bi - crystal and other hts grain boundary junctions .
the model also predicts non - sinusoidal behavior in the voltage oscillations resulting from time - correlated vortex tunneling . |
at present , all solid evidence for dark matter is gravitational . at the same time , the possibility that dark matter has electromagnetic or strong interactions is highly constrained @xcite . a straightforward possibility
, then , is that dark matter is hidden , consisting of particles that have no standard model ( sm ) gauge interactions .
hidden sectors have a long and distinguished history .
for example , the idea that a hidden sector may restore parity on cosmological scales is as old as the idea of parity violation itself @xcite .
such hidden sectors , containing `` mirror matter , '' have been studied by many groups , as have their possible implications for dark matter @xcite .
string theory also motivates hidden sectors with their own gauge interactions @xcite , as in the case of the heterotic string @xcite and intersecting brane models @xcite .
more recently , hidden sectors have been central to several phenomenological developments , including higgs portals @xcite , hidden valleys @xcite , unparticles @xcite , and quirks @xcite , and the possibility that dark matter may reside in a hidden sector has been discussed in several recent works @xcite . in considering hidden dark
matter , one would seemingly be sacrificing critical virtues of more conventional dark matter candidates , namely , the connection of dark matter to the gauge hierarchy problem , and the fact that weakly - interacting massive particles ( wimps ) naturally have the desired thermal relic density .
this is not necessarily true , however . in the recently proposed framework of wimpless dark matter @xcite , hidden sector dark matter
also naturally has the desired thermal relic density .
the essential idea is that , very generally , a dark matter candidate that decoupled from the thermal bath non - relativistically has a thermal relic density @xmath1 where @xmath2 is the thermally - averaged annihilation cross section times velocity , and @xmath3 and @xmath4 are the dark matter particle s gauge coupling and mass . for wimps with weak interaction coupling constant @xmath5 and weak - scale mass @xmath6 , this relic mass density is naturally near the observed @xmath7 . in wimpless models , hidden sector particles @xmath8 have masses and couplings that may be very different from wimps , but which nevertheless satisfy @xmath9 in the wimpless examples discussed in ref .
@xcite , eq .
( [ scaling ] ) follows from the structure of gauge - mediated supersymmetry breaking ( gmsb ) , which generates hidden sector masses that are proportional to couplings squared .
the wimpless framework therefore naturally generalizes the wimp paradigm to other cold dark matter candidates without sacrificing the key thermal relic density virtue of wimps .
if additional connector particles with both sm and hidden gauge quantum numbers are present , they will mediate sm - hidden interactions .
the wimpless framework may therefore also predict qualitatively new signals , such as an explanation of the dama signal in terms of gev dark matter @xcite . in this paper , we study hidden thermal relics in general , and wimpless dark matter in particular . for the most part
, we assume that there are no connector particles , so that the hidden sector interacts with the sm only very weakly . in sec .
[ sec : bbn ] , we carry out a model - independent analysis of constraints on hidden sectors from big bang nucleosynthesis ( bbn ) , the cosmic microwave background ( cmb ) , and other cosmological data .
as is well known , these exclude the possibility that the hidden sector is an exact copy of the sm @xcite .
this statement assumes that the hidden and observable sectors are at the same temperature at late times , however . in the present context
, there is no strong reason to assume that the observable and hidden sectors are reheated to the same temperature @xcite , and , even if they are , for a truly decoupled hidden sector , their temperatures will differ by the time of bbn . following previous works in the context of mirror matter ( see , _ e.g. _ ,
@xcite ) , we consider the cosmological constraints allowing different temperatures and find that even large hidden sectors are easily accommodated .
for example , a hidden version of the sm is allowed even if the hidden sector s reheating temperature is as much as 70% of the observable sector s . alternatively ,
if the hidden sector contains a 1-generation version of the mssm with characteristic mass scale below 1 mev , even identical reheating temperatures are allowed .
we then present a concrete model for the hidden sector in sec .
[ sec : concrete ] and study the freezeout of dark matter in this hidden sector in sec .
[ sec : relic ] .
the relations of eqs .
( [ omega ] ) and ( [ scaling ] ) are , of course , only rough relations capturing the parametric dependences . in these sections , we present a detailed analysis of a specific case that highlights several subtleties and generalizes previous freezeout analyses to cases with sectors at different temperatures .
this analysis confirms the validity of conclusions based on eqs .
( [ omega ] ) and ( [ scaling ] ) and highlights various subtleties of the wimpless framework . in particular
, we find that wimpless dark matter may have the desired thermal relic density for the entire range of @xmath10 , where the lower bound is set by the requirement that dark matter freezeout is non - relativistic , and the upper bound follows from the requirement of perturbative gauge couplings .
wimpless dark matter therefore encompasses as large a range of masses as one could expect of dark matter that has the naturally correct thermal relic density , and it provides a unified framework for addressing many diverse dark matter signals and phenomenology . finally , in sec .
[ sec : adding ] , we discuss modifications from the presence of connector particles with both hidden and observable sector gauge interactions .
we find that , under general assumptions , these fields do not upset the conclusions derived earlier .
we present our conclusions in sec .
[ sec : conclusions ] .
we assume that the observable sector is the minimal supersymmetric standard model ( mssm ) , which is supplemented by a single hidden sector .
hidden sector parameters are denoted by the superscript @xmath11 . for most of this work ,
we assume that there are no connectors , fields with both sm and hidden gauge charges .
the observable and hidden sectors therefore interact extremely weakly with each other , and the observable sector s temperature @xmath12 need not equal the hidden sector s temperature @xmath13 .
we define @xmath14 to parametrize the mismatch .
in addition , it will be convenient to introduce the standard relativistic degrees of freedom parameters .
letting @xmath15 we define @xmath16 where the sum is over observable sector particles , and @xmath17 where the sum is over hidden sector particles .
@xmath18 is the temperature of particle @xmath19 , and @xmath20 denotes its internal degrees of freedom . throughout this work
, we assume that the gravitino has a negligible relic density , as is the case , for example , for low and moderate reheating temperatures , and the gravitino is not included eqs .
( [ obsg ] ) and ( [ hiddeng ] ) .
bbn is sensitive to the expansion rate of the universe at time @xmath21 and temperature @xmath22 .
the expansion rate is determined by the energy density , and so constrains light particles even if they have no sm interactions @xcite .
the constraint is conventionally quoted as a bound on @xmath23 , the effective number of light neutrino species , and may be taken to be @xmath24 ( 95% cl ) , where the baryon density has been fixed to the value determined by the cmb , and both @xmath25he and d data are included @xcite . in our framework , this implies @xmath26 where @xmath27 is the temperature of the hidden sector at time @xmath28 .
if the hidden sector is an exact copy of the mssm and @xmath29 , @xmath30 .
this violates the bound of eq .
( [ bound ] ) , as is well - known @xcite , and may create the impression that large hidden sectors of the size of the mssm are completely excluded .
if the observable and hidden sectors are not in thermal contact , however , the hidden sector may be colder than the observable sector
. this would be the case if , for example , the inflaton couplings to the observable and hidden sectors are not identical , so that they reheat to different temperatures @xcite .
alternatively , the observable and hidden sectors may initially have the same temperature , either because they have the same inflaton couplings or because they are in thermal contact , but may cool independently and have different temperatures at later times .
in fact , it quite reasonable that the observable and hidden sectors lose thermal contact before bbn .
thermal contact would require efficient interactions between the two sectors at low temperatures .
these would require light connector fields to mediate these interactions , but light fields with sm gauge charges generically violate experimental constraints .
we conclude , then , that it is far from guaranteed that the observable and hidden sectors will be at the same temperature at @xmath31 ; generically , they will be at different temperatures . to explore this possibility , we assume that the observable and hidden sectors reheat at time @xmath32 to temperatures @xmath33 and @xmath34 , respectively .
we further assume that the entropy per comoving volume is conserved in each sector separately , so that @xmath35 given mass spectra in the observable and hidden sectors , reheat temperatures @xmath33 and @xmath34 , and eq .
( [ entropy ] ) , the ratio @xmath36 is fixed , and one can determine if eq .
( [ bound ] ) is satisfied . to provide numerical examples ,
we define @xmath37 the ratio of temperatures just after reheating , and assume @xmath33 is above the mass of all mssm particles .
we also define @xmath38 for the observable sector and @xmath39 for the hidden sector , where we have assumed all hidden relativistic degrees of freedom have a common temperature . with these definitions , eq .
( [ entropy ] ) becomes @xmath40 solving for @xmath36 and substituting the mssm values @xmath41 and @xmath42 , we find that the bbn bound of eq .
( [ bound ] ) becomes @xmath43 bounds in the @xmath44 plane for various values of @xmath45 are given in fig .
[ fig : gstar ] . several aspects deserve comment . as evident from eq .
( [ bound2 ] ) , the bound is extremely sensitive to @xmath45 .
the @xmath46 contour is barely contained on the plot , implying that if the hidden sector reheats to half the temperature of the observable sector , very large hidden sectors are allowed .
several example hidden sectors , defined precisely in sec .
[ sec : concrete ] , are also given on the plot .
model a is a 1-generation flavor - free version of the mssm , with all yukawa couplings of order 1 .
we see that it is allowed for @xmath47 , that is , reheat temperatures that are almost identical to the observable sector .
this is our prototype model , which will be used in our detailed freezeout analysis in sec .
[ sec : relic ] .
model b is a full 3-generation flavor - free version of the mssm .
we see that even a hidden sector as `` large '' as the mssm is allowed , provided @xmath48 .
perhaps even more interesting , fig .
[ fig : gstar ] shows that model c , a 1-generation version of the mssm with all degrees of freedom relativistic at bbn , is allowed , even for @xmath49 .
naively , one might expect a hidden sector that starts at the same temperature as the mssm and has over 100 light degrees of freedom to be completely excluded .
in fact , however , such a hidden sector is much colder than the mssm at bbn times , since its cooling is not slowed by the disappearance of heavy degrees of freedom , in contrast to the mssm .
this cooling is critical , given the 4th power of the temperature in eq .
( [ bound ] ) , and makes such a hidden sector allowed . for similar reasons , fig .
[ fig : gstar ] shows that an excluded hidden sector is only made more excluded by the addition of heavy degrees of freedom .
note , however , that fig .
[ fig : gstar ] also shows that in some cases , rather counter - intuitively , an excluded hidden sector may be made allowed by the addition of _ light _ degrees of freedom .
cmb temperature anisotropy measurements provide another constraint on the number of relativistic species .
wmap 5 year data , combined with distance information from baryon acoustic oscillations , supernovae ( snia ) and hubble constant measurements , imply @xmath50 ( 68% cl ) @xcite .
in contrast to the bbn bound , this is a 68% cl bound ; at 95% cl , the bound is very weak .
this bound is therefore numerically much weaker than the bbn bound . in the future , however , cmb data from , for example , the planck satellite @xcite will achieve a much higher sensitivity in constraining the amount of radiation energy @xcite . at the same time , of course , the cmb bound constrains the energy density at much later times with redshift @xmath51 and observable sector temperatures @xmath52 .
it is thus interesting to explore the cmb constraint when the hidden sector includes particles with mass between 1 mev and 1 ev . ( see also ref .
@xcite for a combined constraint . )
the current cmb constraint in our framework is @xmath53 where @xmath54 is the neutrino to photon temperature ratio after the era of @xmath55 annihilation .
the fact that the sm neutrinos were not fully decoupled at @xmath55 annihilation is taken into account by using the modified number of sm neutrinos of 3.046 instead of 3 @xcite . following the bbn analysis , this cmb bound implies @xmath56 where @xmath57 and @xmath58 are defined as in eq .
( [ gh_heavy_light ] ) , but with @xmath27 replaced by @xmath59 .
we show current cmb bounds in the @xmath60 plane in fig .
[ fig : gstarcmb ] , along with four example models .
figure [ fig : gstar ] s model c , with all degrees of freedom light compared to @xmath27 , is now differentiated into many models , depending on how many of the degrees of freedom are light compared to @xmath59 .
these define a line in fig .
[ fig : gstarcmb ] , with endpoints given by models c@xmath61 and c@xmath62 .
we find that if these models are allowed by the bbn constraints , they are also allowed by the current cmb constraint .
similarly , fig .
[ fig : gstar ] s model d is differentiated into models d@xmath61 and d@xmath62 in fig .
[ fig : gstarcmb ] , but again we find that in such models , if the bbn constraint is satisfied , the current cmb constraint is also . in these models , then , the current cmb constraint is never as stringent as the bbn constraint .
this may change with the strengthening of the cmb constraint in the future .
in addition , it is interesting that in the presence of hidden sectors , there may be many particles with mass between @xmath27 and @xmath59 , and bbn and the cmb yield independent information about the hidden sector mass spectrum . this contrasts with the case of the sm , where cmb constraints may be viewed as consistency checks on bbn results .
in sec . [ sec : bbn ] , we found that hidden sectors may contain many degrees of freedom without violating cosmological bounds . with this as motivation , in the following section , we will examine freezeout in hidden sectors whose temperatures may differ from the observable sector s . to prepare the way , in this section we first specify a model and mass spectrum for the hidden sector .
the wimpless models of ref .
@xcite are supersymmetric , with supersymmetry breaking mediated by gauge interactions .
a worthwhile endeavor would be to build a complete gmsb wimpless model . in this study , however , we instead work with a `` gmsb - inspired '' model .
we do this for two reasons .
first , our primary aim here is to clarify cosmological issues .
building a complete gmsb model with a wimpless hidden sector , complete with @xmath63 term solution and other particle physics details , would take us too far afield , especially since , as we will see , the relic density is independent of many of these details . second , as noted in sec .
[ sec : intro ] , the essential virtue of wimpless models , that they have the naturally correct thermal relic density , relies solely on the fact that masses in the hidden sector are proportional to gauge couplings squared .
this is true for gmsb , of course , but is also valid in many other contexts , such as anomaly - mediated supersymmetry breaking models , lessening the motivation for detailed consideration of any one gmsb model .
the model we consider has an mssm - like hidden sector , with @xmath64 gauge interactions .
in contrast to the mssm , however , we assume that all yukawa couplings are @xmath65 .
we refer to this as a flavor - free version of the mssm . for simplicity ,
we will focus on a 1-generation flavor - free version of the mssm , but , as all yukawa couplings are large , we will use 3rd generation nomenclature for this one generation of hidden matter fermions and sfermions .
all hidden superpartners are assumed to have mass @xmath66 , the superpartner mass scale .
in addition , since @xmath67 also sets the scale for the higgs boson parameters , the hidden higgs bosons @xmath68 , @xmath69 , @xmath70 , and @xmath71 also have this mass , as do the hidden weak gauge bosons @xmath72 and @xmath73 .
given @xmath65 yukawa couplings , the hidden fermions @xmath74 , @xmath75 , and @xmath76 are therefore also at the superpartner mass scale .
the light , effectively massless particles are the hidden photon @xmath77 , gluon @xmath78 and neutrino @xmath79 .
as is typical of gmsb , the gravitino is also light , and its relic density from late decays to the gravitino is negligible .
we further assume low or moderate reheating temperatures so that the abundance of gravitinos produced during reheating is also insignificant .
wimpless dark matter requires that one particle at the @xmath67 mass scale be stable .
we assume that the lightest @xmath67 scale particle is the hidden right - handed stau @xmath80 .
this is the lightest superpartner in much of gmsb parameter space , and it is also reasonable to assume that it is lighter than the @xmath74 , given @xmath65 yukawa couplings , just as the stop may be lighter than the top in the mssm .
the stau s stability is then guaranteed by hidden sector electric charge conservation .
one might worry about the viability of a charged dark matter candidate .
efficient compton scattering of hidden staus on hidden photons @xmath81 would lead to dissipative energy loss , making the dark matter particles behave more like baryons .
however , this scattering cross section is @xmath82 , which , given eq .
( [ scaling ] ) , is as weak as the sm weak interactions .
hidden sector charged dark matter is therefore consistent with the known properties of dark matter from structure formation .
given the few light states , there are only a few dark matter annihilation channels : @xmath83 the neutrino channel is mediated by an @xmath84-channel @xmath73 .
the gauge boson final states are produced through processes with @xmath85 in @xmath86- and @xmath87-channels and 4-point @xmath88 and @xmath89 vertices .
the relic density is therefore dependent on only a few of the many supersymmetric parameters , which we will now discuss .
note that we assume no left - right stau mixing .
all three gauge couplings enter the analysis .
the su(2 ) and u(1 ) gauge couplings enter the annihilation cross sections .
the su(3 ) gauge coupling must also be specified , because it determines the number of relativistic degrees of freedom at bbn .
for simplicity , for the hidden gauge couplings , we adopt assumptions motivated by unification .
gauge coupling unification implies that the hidden weak mixing angle is @xmath90 at the gut scale . in principle , this is modified by renormalization group ( rg ) evolution .
the rg equations for the hidden sector mssm gauge couplings are @xmath91 where @xmath92 for the 1-generation case , and @xmath93 for the 3-generation case . as we will see in sec .
[ sec : relic ] , however , wimpless models are valid in the ranges @xmath94 and @xmath95 , depending on the reheating temperature ratio @xmath45 .
for most of this range , then , @xmath96 , and the fractional change in the gauge couplings from rg evolution is @xmath97 ; that is , the gauge couplings evolve very little . given this
, we will assume @xmath98 at @xmath67 .
for @xmath99 , this is not the prediction of unification , and our assumption therefore implicitly assumes some more complicated structure .
the unification assumption also typically implies that the hidden confinement scale @xmath100 is far below @xmath27 .
for example , for the 1-generation case , taking @xmath101 and @xmath102 , we find @xmath103 .
we therefore assume @xmath104 , and so the hidden gluons contribute to @xmath105 at bbn . again
, this does not follow from unification for the largest @xmath67 .
we will identify @xmath106 with the hidden su(2 ) gauge coupling @xmath78 .
as for the hidden mssm mass spectrum , we assume that all massive hidden particles have masses in the range @xmath107 , and that all relativistic degrees of freedom are always at the same temperature
. the effective relativistic degrees of freedom in the hidden sector are then @xmath108 for the 1-generation ( 3-generation ) case , and we linearly interpolate between these two values for @xmath13 between @xmath67 and @xmath109 .
the exact spectrum in this range is largely irrelevant for the relic abundance calculation , since we consider only dark matter that is non - relativistic at freezeout . given these assumptions , the thermal relic density is completely determined by only four parameters : @xmath110 in fig .
[ fig : xi ] we plot the relative evolution of the observable and hidden photon temperatures for the cases @xmath111 , 1 mev , 1 kev and 1 ev . to make this plot
, we must determine @xmath13 as a function of @xmath12 for a given @xmath67 . to get the hidden sector temperature @xmath13 at a given observable sector temperature @xmath12 ,
the value of @xmath112 is required , which is in fact unattainable unless one knows @xmath13 .
our approach is the following : we first solve for @xmath113 from eq .
( [ entropy ] ) with @xmath114 and @xmath27 replaced by @xmath12 and @xmath13 , respectively .
we then determine @xmath13 by dividing this quantity by @xmath112 as determined by eq .
( [ gstar_model ] ) in the three trial intervals @xmath115 , @xmath116 , and @xmath117 , and choose the value of @xmath13 that yields a self - consistent solution .
the general feature of fig . [ fig : xi ] is that as @xmath12 decreases from @xmath33 , @xmath118 drops when observable degrees of freedom become non - relativistic and rises when hidden degrees of freedom become non - relativistic .
rapid drops occur at @xmath119 when the observable superpartners become non - relativistic and @xmath120{g_{\ast s } ( t)}$ ] , and at @xmath121 , where the qcd transition occurs .
the rise occurs when the hidden @xmath67 mass particles become non - relativistic at @xmath122 . at @xmath123 , where the exact value depends on @xmath67 , the hidden sector phase transition concludes , and @xmath124 begins to track @xmath125{g_{\ast s } ( t)}$ ] again
we are now ready to determine thermal relic densities in the case of observable and hidden sectors with unequal temperatures . for the hidden sector
, we assume the 1-generation flavor - free version of the mssm described in sec .
[ sec : concrete ] . the temperature of the hidden sector thermal bath at any instant can be read off from fig .
[ fig : xi ] . the number density of dark matter particles @xmath126
is determined by the competition between the expansion rate of the universe and the dark matter annihilation rate through the boltzmann equation @xmath127 \ .\ ] ] the hubble parameter depends on the effective relativistic degrees of freedom in both the observable and hidden sectors and is @xmath128^{\frac{1}{2 } } \ , \ ] ] where @xmath129 in contrast to @xmath130 , @xmath2 , the thermally - averaged product of the total annihilation cross section and the mller velocity , and @xmath131 , the equilibrium number density , are determined by the hidden temperature alone . to solve the boltzmann equation ,
it is standard to change to dimensionless variables through @xmath132 these new variables are natural from the observable sector viewpoint , as they use the observable sector s temperature @xmath12 as the `` clock , '' and the observable sector s entropy density @xmath133 as the fiducial quantity characterizing the universe s expansion .
note , however , that in the present context there is also another useful change of variables , @xmath134 that is far more natural from the hidden sector point of view .
we will return to this below . in terms of the observable sector variables of eq .
( [ observablevariables ] ) , the boltzmann equation becomes @xmath135\ , , \ ] ] with the expansion and annihilation effects encapsulated in the quantity @xmath136 we will solve this equation in the non - relativistic regime , where @xmath137 .
the equilibrium value of the number of dark matter particles in a comoving volume is @xmath138 where @xmath3 denotes the number of internal degrees of freedom of the dark matter particle , and the last approximation of eq .
( [ yeq ] ) is valid for @xmath137 .
following ref .
@xcite , the thermally - averaged cross section times velocity is determined using @xmath139 in eqs .
( [ yeq ] ) and ( [ sigmav ] ) , @xmath140 are modified bessel functions of order @xmath19 . equation ( [ sigmav ] ) was derived for particles in thermal equilibrium obeying maxwell - boltzmann statistics , but it is applicable to other statistics , provided @xmath141 .
we determine the @xmath142 total annihilation cross section @xmath143 with the help of the calchep package @xcite .
expanding around @xmath144 ( see ref .
@xcite ) , we find @xmath145 \ , , \ ] ] where @xmath146\ , \sin^4 \theta^h_w \\
a_1 & = & \frac{3}{2}\ , \left[- \frac{1}{6 \pi } - \frac{1}{3 \pi}\ , \tan^2 \theta^h_w + \frac{1}{12 \pi}\ , \frac{1}{\cos^4 \theta^h_w\ , \left[\left(-4 + \frac{m^2_{z^h}}{m^2_x } \right)^2 + \frac{m^2_{z^h } \gamma^2_{z^h}}{m^4_x } \right ] } \right]\ , \sin^4 \theta^h_w\ , \end{aligned}\ ] ] and @xmath147 is the @xmath73 boson decay width . in keeping with our non - relativistic approximation ,
we neglect the @xmath148 and higher order terms . for @xmath149 , @xmath150
is greatly enhanced by the @xmath84-channel resonance . off this resonance , however , the neutrino channel is typically sub - dominant , because it is @xmath151-wave suppressed , as can be seen above . in our numerical study , we fix @xmath152 .
the @xmath73 boson remains in equilibrium with the hidden thermal bath through the decay and inverse decay processes @xmath153 .
we now present several results for @xmath154 . because of the `` stiffness '' of the boltzmann equation of eq .
( [ boltzmann_eq_y ] ) , we adopt the implicit trapezoidal method with adaptive step size , developed by the authors of the darksusy package @xcite , to solve for @xmath154 numerically .
we first examine the @xmath45 dependence of the thermal relic density . in fig .
[ fig : freezeout_xi ] , we plot @xmath154 for various values of @xmath45 , @xmath155 , and @xmath156 .
the latter two parameters are naturally accommodated in the wimpless gmsb framework and yield @xmath157 in the required range , as we will see below .
as @xmath45 decreases , we see two effects .
first , for a given @xmath158 , the hidden sector is colder , and so @xmath159 is smaller .
second , since freezeout occurs when @xmath160 , when @xmath45 decreases and @xmath126 drops , freezeout occurs earlier . as can be seen from fig .
[ fig : freezeout_xi ] , these effects act against each other , and the thermal relic density is fairly insensitive to @xmath45 . these results may be understood more precisely using standard analytic approximation formulas for thermal relic freezeout .
the equilibrium number density is @xmath161 where @xmath162 , as defined above .
following the procedure given in ref .
@xcite and assuming @xmath163-wave annihilation , an approximate expression for @xmath158 at freezeout is @xmath164 \nonumber \\ & & -\frac{1}{2 } \xi \ln \left\ { \xi \ln \left [ 0.038 \ , { m_{\text{pl}}}\ , m_x \ , \sigma_0 ( g/\sqrt{g^{\text{tot } } _ * } ) \ , \xi^{3/2 } \ , \delta(\delta+2 ) \right ] \right\ } \ , \label{xf}\end{aligned}\ ] ] where @xmath165 and the parameter @xmath166 is tuned to make these analytical results fit the numerical results . , we find that choosing @xmath167 yields agreement typically better than 3% . for larger @xmath67 or higher @xmath45 ,
the final relic density is less sensitive to @xmath166 . ] for @xmath163-wave annihilation , the final relic abundance is approximately @xmath168 which depends linearly on @xmath124 through @xmath169 .
this linear dependence is verified by the numerical results presented in fig .
[ fig : freezeout_xi ] .
the relative values of @xmath170 obtained from eq .
( [ y0 ] ) are @xmath171 for @xmath172 , 0.5 , 0.3 , respectively , which almost exactly match the results from our numerical computation , which yields @xmath173 . the dependence on @xmath45 may also be understood from the hidden sector viewpoint . in fig .
[ fig : freezeout_xi_hidden ] , we consider the same model parameters as plotted in fig .
[ fig : freezeout_xi ] , but we now present @xmath174 . from a hidden observer s point of view
, the dependence on @xmath45 arises solely because @xmath45 determines how hot the observable sector is , and this impacts the expansion rate @xmath130 . for a fixed @xmath175 ,
lower @xmath45 implies higher @xmath12 , faster expansion , and earlier freezeout .
this is seen in fig .
[ fig : freezeout_xi_hidden ] . at first sight ,
the result that lower @xmath45 implies larger @xmath176 may seem to contradict the results obtained earlier , where lower @xmath45 implies lower relic density .
note , however , that to convert @xmath176 to a relic density , one must multiply @xmath176 by the current @xmath177 , which , unlike @xmath84 , is not independent of @xmath45 . for lower @xmath45 , the current hidden sector temperature is lower , and the current @xmath177 is also lower .
this effect makes the relic density smaller for lower @xmath45 , in accord with our previous results derived from the observable sector viewpoint .
we now examine the dependence of the thermal freezeout process on @xmath67 and @xmath106 . in fig .
[ fig : freezeout_mx ] , we plot @xmath154 for fixed @xmath45 , but for a wide range of @xmath178 with @xmath179 , the scaling relation of wimpless scenarios . for low @xmath67
, freezeout occurs earlier , as expected given eq .
( [ xf ] ) . in our analysis
, we have assumed that freezeout occurs in the non - relativistic regime , where @xmath137 .
converting @xmath175 to @xmath158 using fig .
[ fig : xi ] , we see from fig . [
fig : freezeout_mx ] that our analysis is self - consistent even for @xmath67 as low as 10 kev . in fig .
[ fig : wmapgxmx ] we present contours of constant @xmath157 in the @xmath178 plane .
these contours result from our numerical analysis ; although we have confined our previous plots to @xmath180 , the results of fig .
[ fig : wmapgxmx ] use the relic density as determined at @xmath181 , where it has truly stabilized .
figure [ fig : wmapgxmx ] summarizes all of our work so far , and there are several noteworthy features .
first , we see that these contours essentially follow the scaling relation @xmath182 .
if , for example , @xmath67 is lowered by a factor of 4 and @xmath106 is lowered by a factor of 2 , the weakened interaction implies earlier freezeout and larger @xmath170 , but this compensates the lower @xmath67 to keep the relic density @xmath183 approximately invariant .
alternatively , using the analytic results presented above , this scaling relation keeps @xmath184 invariant , and , given eq .
( [ y0 ] ) , this also keeps @xmath157 invariant . of course
, this is as expected from the general arguments given in sec .
[ sec : intro ] , but fig .
[ fig : wmapgxmx ] shows that these arguments , based essentially on dimensional analysis , are in fact supported by a detailed calculation in the context of a concrete model with dark matter masses varying greatly and far beyond the conventional wimp range .
second , for comparison , fig .
[ fig : wmapgxmx ] also includes two dashed contours to indicate the corresponding observable sector weak mass scale @xmath185 .
the definition of @xmath185 depends on how one chooses to characterize the natural weak scale in supersymmetric theories ( the @xmath63 parameter , the bino mass , etc . ) , and these depend on several parameters , such as the number of messengers and the messenger scale .
we simply choose @xmath186 , where @xmath187 is the u(1 ) gauge coupling of the mssm , and plot contours for @xmath188 and 1 tev .
the fact that there exist values in the @xmath189 plane that give the correct @xmath157 is , of course , not surprising .
however , the fact that the two solid curves lie between the two dashed curves means that the preferred values of @xmath178 are those that correspond to models in which the observable sector s superpartners are at the weak scale .
we also see from fig .
[ fig : wmapgxmx ] that , for a fixed @xmath67 , lower @xmath45 requires lower @xmath106 to obtain the correct relic density .
this is because , at a given observable temperature @xmath12 , colder hidden sectors imply lower equilibrium number densities . to keep the relic density constant
, @xmath106 must be lower so that freezeout happens earlier .
this is consistent with the results shown in fig .
[ fig : freezeout_xi ] .
finally , note that the @xmath190 contour ends at @xmath191 .
this is because the criterion that dark matter freezeout while non - relativistic , that is , @xmath192 , is violated below this mass . in fig .
[ fig : mxin ] , we plot , as a function of @xmath45 , the lower mass limit resulting from requiring that freezeout occurs with @xmath192 and yields @xmath193 .
we can see that the lower mass limit goes up with smaller @xmath45 .
this is because for colder hidden sectors , freezeout occurs earlier , and to satisfy @xmath194 , we need larger @xmath67 .
this exercise establishes that the wimpless dark matter framework may be valid at least down to dark matter masses of @xmath195 . below this mass ,
freezeout is not non - relativistic , and our analysis breaks down .
it would be interesting to perform a more precise analysis without assuming non - relativistic freezeout , but this is beyond the scope of our study .
throughout this work , we have ignored the possible existence of connector particles @xmath196 with both sm and hidden gauge charges .
if these are added , do they significantly affect the conclusions derived above ? there are two significant roles that connector particles can have .
first , they likely establish thermal connectivity between the observable and hidden sectors when the temperatures are above the connector mass scale @xmath197 . at temperatures below @xmath197 , however , the hidden - sm interactions become weak , and the observable and hidden sectors decouple . as shown in figs .
[ fig : gstar ] and [ fig : gstarcmb ] , the bbn and cmb constraints actually allow , for example , a 1-generation flavor - free version of the mssm with @xmath198 , even when @xmath199 . in this case , then , there is no problem if the two sectors are in thermal contact right after reheating . for other hidden sectors
, it may be possible that the existence of connectors enforces @xmath199 , or alternatively , the assumption of @xmath200 implies an upper bound on the reheating temperatures .
one must check on a case by case basis whether connector particles upset the premises of our calculations .
a second affect is that connector particles open new avenues for dark matter annihilation . as an example
, we assume that fermionic connector particles @xmath196 couple through yukawa couplings @xmath201 , where @xmath202 is a sm fermion .
we expect @xmath203 @xcite , and also require @xmath204 to prevent the decay @xmath205 .
such connector interactions induce annihilations @xmath206 through processes with @xmath196 particles in the @xmath86-channel .
there are two cases . if @xmath207 , @xmath208 , and @xmath209 on the other hand , if @xmath210 , @xmath211 , and @xmath212 in either case , the annihilation cross section to sm particles is small relative to the completely hidden annihilation cross sections @xmath213 , provided @xmath214 .
this is a rather weak criterion , and so typically , annihilation to sm particles is sub - dominant , and the analysis given above holds even in the presence of connector particles .
in the paper , we have studied the possibility that dark matter is hidden , that is , has no sm gauge interactions .
hidden sectors appear in many frameworks for physics beyond the sm , and hidden dark matter is perfectly viable , given all observations to date .
we are particularly motivated by the recent proposal of wimpless dark matter . as with wimps , wimpless dark matter has the key virtue that its thermal relic density is in the right range to be cold dark matter .
in contrast to wimps , however , wimpless dark matter may have masses and couplings that differ drastically from wimps .
although hidden sectors interact very weakly with the sm , hidden particles contribute to the energy density of the universe .
this impacts the expansion rate of the universe , and so bounds from bbn and the cmb constrain hidden sectors .
as is well known , current constraints completely exclude a hidden sector that is an exact copy of the mssm .
we find , however , that these constraints are rather brittle and easily avoided if the hidden sector is just slightly colder than the observable sector or if the mass scale of hidden superpartners differs from the mssm .
several viable examples are discussed in sec .
[ sec : bbn ] , and figs .
[ fig : gstar ] and [ fig : gstarcmb ] give model - independent bounds that can be used to determine the viability of other models . in sec .
[ sec : concrete ] , we then defined a concrete model , a 1-generation flavor - free version of the mssm .
this model has features generic to gmsb models and provides a wimpless dark matter candidate , the hidden stau @xmath85 .
the @xmath85 relic density depends sensitively on only a small number of parameters .
this relic density was examined in sec .
[ sec : relic ] by solving the boltzmann equation numerically , and understanding these results through well - known analytic approximations , generalized to hidden sectors with different temperatures .
these results confirm that in wimpless models with @xmath215 , the relic density is naturally in the right range for dark matter masses @xmath10 , greatly extending the conventional wimp mass range .
wimpless dark matter therefore provides a class of dark matter candidates that shares the key relic density virtue of wimps .
this generalization may have interesting applications , given the diverse and interesting phenomenology and observational anomalies already present at the kev , mev , gev , and tev mass scales .
we have determined the allowed range of @xmath67 by requiring that this dark matter freezes out with the right relic density while non - relativistic .
of course , this range may be further constrained by other considerations .
in particular , the lightest wimpless candidates , with @xmath195 , may have interesting implications for structure formation .
after freeze out , they can still couple to the hidden sector thermal bath through elastic scattering processes @xmath216 and @xmath217 .
after their kinetic decoupling at temperature @xmath218 , they then stream freely between over- and under - dense regions .
free - streaming of thermal relics causes damping in density perturbations @xcite at a comoving scale @xmath219 ; in contrast to wimps @xcite , this is the dominant effect for mev or lighter dark matter @xcite .
if this scale falls in the range @xmath220 mpc , the linear matter power spectrum from high - resolution lyman-@xmath221 forest data can be used to place a lower bound on the mass of the dark matter particle .
current lower mass limits on warm dark matter of @xmath222 from lyman-@xmath221 data @xcite therefore imply lower mass bounds in the wimpless scenario , which depend on the kinetic decoupling temperature @xmath218 for different @xmath45 .
in addition , another kind of dark matter lower mass limit follows from the observation that the phase - space density of the dark matter particles in galaxy halos can not exceed the maximum value of the phase - space density when the dark matter particles were in kinetic equilibrium @xcite . with knowledge of the wimpless dark matter s chemical potential
, one can then apply these generalized tremaine - gunn bounds from ref .
@xcite to our scenario .
detailed considerations of these effects are beyond the scope of this study , but they may modify the lower bound on @xmath67 or , alternatively , provide interesting astrophysical signals for the hidden dark matter scenarios discussed here .
we thank manoj kaplinghat and jason kumar for many helpful discussions .
jlf and ht are supported in part by nsf grants phy0239817 , phy0314712 , and phy0653656 , nasa grant nng05gg44 g , and the alfred p. sloan foundation .
hy is supported by nsf grant phy0709742 .
99 s. dimopoulos , d. eichler , r. esmailzadeh and g. d. starkman , phys .
d * 41 * , 2388 ( 1990 ) .
g. d. starkman , a. gould , r. esmailzadeh and s. dimopoulos , phys .
d * 41 * , 3594 ( 1990 ) .
t. d. lee and c. n. yang , phys .
rev . * 104 * , 254 ( 1956 ) .
s. i. blinnikov and m. y. khlopov , sov .
j. nucl .
phys . * 36 * , 472 ( 1982 ) [ yad .
fiz . * 36 * , 809 ( 1982 ) ] ; sov .
astron .
* 27 * , 371 ( 1983 ) [ astron .
60 * , 632 ( 1983 ) ] .
h. m. hodges , phys .
d * 47 * , 456 ( 1993 ) .
z. g. berezhiani , a. d. dolgov and r. n. mohapatra , phys .
b * 375 * , 26 ( 1996 ) [ arxiv : hep - ph/9511221 ] .
r. n. mohapatra and v. l. teplitz , phys .
d * 62 * , 063506 ( 2000 ) [ arxiv : astro - ph/0001362 ] .
z. berezhiani , d. comelli and f. l. villante , phys .
b * 503 * , 362 ( 2001 ) [ arxiv : hep - ph/0008105 ] .
r. n. mohapatra , s. nussinov and v. l. teplitz , phys .
d * 66 * , 063002 ( 2002 ) [ arxiv : hep - ph/0111381 ] .
a. y. ignatiev and r. r. volkas , phys .
d * 68 * , 023518 ( 2003 ) [ arxiv : hep - ph/0304260 ] .
r. foot , phys .
d * 69 * , 036001 ( 2004 ) [ arxiv : hep - ph/0308254 ] .
z. berezhiani , p. ciarcelluti , d. comelli and f. l. villante , int .
j. mod .
d * 14 * , 107 ( 2005 ) [ arxiv : astro - ph/0312605 ] .
r. foot and r. r. volkas , phys .
d * 70 * , 123508 ( 2004 ) [ arxiv : astro - ph/0407522 ] .
m. b. green and j. h. schwarz , phys .
b * 149 * , 117 ( 1984 ) .
d. j. gross , j. a. harvey , e. j. martinec and r. rohm , phys .
* 54 * , 502 ( 1985 ) .
r. blumenhagen , m. cvetic , p. langacker and g. shiu , ann .
sci . * 55 * , 71 ( 2005 ) [ arxiv : hep - th/0502005 ] .
r. schabinger and j. d. wells , phys .
d * 72 * , 093007 ( 2005 ) [ arxiv : hep - ph/0509209 ] .
b. patt and f. wilczek , arxiv : hep - ph/0605188 . m. j. strassler and k. m. zurek , phys .
b * 651 * , 374 ( 2007 ) [ arxiv : hep - ph/0604261 ] .
h. georgi , phys . rev .
lett . *
98 * , 221601 ( 2007 ) [ arxiv : hep - ph/0703260 ] . j. kang and m. a. luty , arxiv:0805.4642 [ hep - ph ] .
x. chen and s. h. tye , jcap * 0606 * , 011 ( 2006 ) [ arxiv : hep - th/0602136 ] .
t. kikuchi and n. okada , phys .
b * 665 * , 186 ( 2008 ) [ arxiv:0711.1506 [ hep - ph ] ] .
j. march - russell , s. m. west , d. cumberbatch and d. hooper , jhep * 0807 * , 058 ( 2008 ) [ arxiv:0801.3440 [ hep - ph ] ] .
d. hooper and k. m. zurek , phys .
d * 77 * , 087302 ( 2008 ) [ arxiv:0801.3686 [ hep - ph ] ] .
j. mcdonald and n. sahu , jcap * 0806 * , 026 ( 2008 ) [ arxiv:0802.3847 [ hep - ph ] ] .
y. g. kim , k. y. lee and s. shin , jhep * 0805 * , 100 ( 2008 ) [ arxiv:0803.2932 [ hep - ph ] ] .
w. krolikowski , arxiv:0803.2977 [ hep - ph ] .
y. gong and x. chen , arxiv:0803.3223 [ astro - ph ] .
j. l. feng and j. kumar , arxiv:0803.4196 [ hep - ph ] .
r. foot , arxiv:0804.4518 [ hep - ph ] .
j. l. feng , j. kumar and l. e. strigari , arxiv:0806.3746 [ hep - ph ] .
e.g. _ , e. w. kolb , d. seckel and m. s. turner , nature * 314 * , 415 ( 1985 ) .
r. foot and r. r. volkas , astropart .
* 7 * , 283 ( 1997 ) [ arxiv : hep - ph/9612245 ] .
g. steigman , d. n. schramm and j. e. gunn , phys .
b * 66 * ( 1977 ) 202 .
for a recent review , see g. steigman , ann .
nucl . part .
* 57 * ( 2007 ) 463 [ arxiv:0712.1100 [ astro - ph ] ] .
r. h. cyburt , b. d. fields , k. a. olive and e. skillman , astropart .
* 23 * ( 2005 ) 313 [ arxiv : astro - ph/0408033 ] .
b. fields and s. sarkar , arxiv : astro - ph/0601514 .
e. komatsu _ et al . _
[ wmap collaboration ] , arxiv:0803.0547 [ astro - ph ] .
[ planck collaboration ] , arxiv : astro - ph/0604069 .
s. hannestad , h. tu and y. y. y. wong , jcap * 0606 * ( 2006 ) 025 [ arxiv : astro - ph/0603019 ] .
v. simha and g. steigman , jcap * 0806 * ( 2008 ) 016 [ arxiv:0803.3465 [ astro - ph ] ]
. g. mangano , g. miele , s. pastor , t. pinto , o. pisanti and p. d. serpico , nucl .
b * 729 * ( 2005 ) 221 [ arxiv : hep - ph/0506164 ] .
t. s. coleman and m. roos , phys .
d * 68 * ( 2003 ) 027702 [ arxiv : astro - ph/0304281 ] .
a. pukhov _ et al .
_ , arxiv : hep - ph/9908288 .
a. pukhov , arxiv : hep - ph/0412191 .
p. gondolo and g. gelmini , nucl .
b * 360 * ( 1991 ) 145 .
p. gondolo , j. edsjo , p. ullio , l. bergstrom , m. schelke and e. a. baltz , jcap * 0407 * ( 2004 ) 008 [ arxiv : astro - ph/0406204 ] .
e. bertschinger , phys .
d * 74 * ( 2006 ) 063509 [ arxiv : astro - ph/0607319 ] .
a. loeb and m. zaldarriaga , phys .
d * 71 * , 103520 ( 2005 ) [ arxiv : astro - ph/0504112 ] .
s. profumo , k. sigurdson and m. kamionkowski , phys .
lett . *
97 * , 031301 ( 2006 ) [ arxiv : astro - ph/0603373 ] .
d. hooper , m. kaplinghat , l. e. strigari and k. m. zurek , phys .
d * 76 * ( 2007 ) 103515 [ arxiv:0704.2558 [ astro - ph ] ] .
v. k. narayanan , d. n. spergel , r. dave and c. p. ma , astrophys . j. * 543 * ( 2000 ) l103 [ arxiv : astro - ph/0005095 ] .
m. viel , j. lesgourgues , m. g. haehnelt , s. matarrese and a. riotto , phys .
d * 71 * ( 2005 ) 063534 [ arxiv : astro - ph/0501562 ] .
k. abazajian , phys .
d * 73 * ( 2006 ) 063513 [ arxiv : astro - ph/0512631 ] .
m. viel , g. d. becker , j. s. bolton , m. g. haehnelt , m. rauch and w. l. w. sargent , phys .
* 100 * ( 2008 ) 041304 [ arxiv:0709.0131 [ astro - ph ] ] .
s. tremaine and j. e. gunn , phys .
lett . * 42 * ( 1979 ) 407 .
j. madsen , phys .
d * 44 * ( 1991 ) 999 . | dark matter may be hidden , with no standard model gauge interactions . at the same time , in wimpless models with hidden matter masses proportional to hidden gauge couplings squared , the hidden dark matter s thermal relic density may naturally be in the right range , preserving the key quantitative virtue of wimps .
we consider this possibility in detail .
we first determine model - independent constraints on hidden sectors from big bang nucleosynthesis and the cosmic microwave background .
contrary to conventional wisdom , large hidden sectors are easily accommodated .
a flavour - free version of the standard model is allowed if the hidden sector is just 30% colder than the observable sector after reheating .
alternatively , if the hidden sector contains a 1-generation version of the standard model with characteristic mass scale below 1 mev , even identical reheating temperatures are allowed .
we then analyze hidden sector freezeout in detail for a concrete model , solving the boltzmann equation numerically and understanding the results from both observable and hidden sector points of view .
we find that wimpless dark matter indeed obtains the correct relic density for masses in the range @xmath0 .
the upper bound results from the requirement of perturbativity , and the lower bound assumes that the observable and hidden sectors reheat to the same temperature and is raised to the mev scale if the hidden sector is 10 times colder .
wimpless dark matter therefore generalizes the wimp paradigm to the largest mass range possible for viable thermal relics and provides a unified framework for exploring dark matter signals across nine orders of magnitude in dark matter mass . |
the standard cosmological model ( for review see @xcite ) gives us an accurate description of the evolution of the universe . in spite of its success ,
the standard cosmological model has a series of problems such as the initial singularity , the cosmological horizon , the flatness problem , the baryon asymmetry , and the nature of dark matter and dark energy . under the dynamical laws of general relativity ,
the standard flrw cosmology becomes singular at the origin of universe .
the matter density and geometrical invariants diverge as the volume of the universe goes to zero .
the big bang singularity seems to be an unavoidable aspect of the currently established cosmological model @xcite which probably only a full quantum theory of gravity could resolve . a bouncing universe with an initial contraction to a non - vanishing minimal radius , then subsequent an expanding phase provides a possible solution to the singularity problem of the standard big - bang cosmology .
bouncing cosmologies , in which the present era of expansion is preceded by a contracting phase , have been studied as potential alternatives to inflation in solving the problems of standard frw cosmology . the first explicit semi - analytic solution for a closed bouncing frw model filled by a massive scalar field
was found by starobinskii @xcite .
later explicit solutions for a bouncing geometry were obtained by novello and salim @xcite , and melnikov and orlov @xcite . for the review of the cosmological bounce
one can see review @xcite . in this paper
we would like to show that a universe bounce can be not only a short time event but it can be a long - term process .
in this section we will investigate a cosmological solution for a closed universe filled with a higgs scalar field .
our goal is to find a regular solution with contracting , expanding and bouncing off phases .
we start with the lagrangian @xmath0\ ] ] where @xmath1 is the 4d scalar curvature , @xmath2 is the scalar higgs field with the potentials @xmath3 .
the corresponding field equations are @xmath4 the potential @xmath3 is the mexican hat potential @xmath5 where @xmath6 and @xmath7 are constants and @xmath7 can be considered as a cosmological constant .
the energy - momentum tensor @xmath8 for the scalar field is @xmath9 for the investigation of bouncing off of universe we consider the cosmological metric @xmath10 .
\label{2 - 60}\ ] ] after the substitution of the metric into field equations we have following equations set @xmath11 , \label{2 - 70}\\ \frac{2 \ddot a}{a } + \frac{\dot a^2}{a^2 } + \frac{1}{a^2 } & = & \varkappa \left [ - \frac{\dot \phi^2}{2 } + \frac{\lambda}{4 } \left ( \phi^2 - m^2 \right)^2 - v_0 \right ] , \label{2 - 80}\\ \ddot \phi + \frac{3 \dot a \dot \phi}{a } & = & \lambda \phi \left ( m^2 - \phi^2 \right ) .
\label{2 - 90}\end{aligned}\ ] ] to begin with we would like to define the notion `` bouncing off of the universe '' .
we will say , that the universe undergo bouncing off if enough a long compression stage is replaced by enough a long stage of the expansion .
a long time interval @xmath12 means that @xmath13 or @xmath14 where @xmath15 is the life time of the universe .
why we need for such definition ?
below we will show that bouncing off can be a complicated process lasting a long enough time .
the statement of the problem :
we will investigate a long term bouncing off solutions with a point of inflection with following conditions @xmath16 here @xmath17 is the point of inflection . in the consequence of the existence of the inflection point the functions
@xmath18 can be presented in the form @xmath19 using the conditions and equations - one can find the following constraints on the initial conditions @xmath20 and the cosmological constant @xmath7 @xmath21 , \label{2 - 110}\\ \frac{2 \dot a^2_0}{a^2_0 } + \frac{2}{a^2_0 } & = & \varkappa \dot \phi^2_0 .
\label{2 - 120}\end{aligned}\ ] ] it means that the cosmological constant @xmath7 is defined uniquely @xmath22 for the numerical investigation we introduce the dimensionless time @xmath23 and dimensionless functions @xmath24 , @xmath25 . then equations - becomes @xmath26 , \label{2 - 140}\\ \frac{2 \ddot a}{a } + \frac{\dot a^2}{a^2 } + \frac{1}{a^2 } & = & \left [ - \frac{\dot \phi^2}{2 } + \frac{\lambda}{4 } \left ( \phi^2 - m^2 \right)^2 - v_0 \right ] , \label{2 - 150}\\ \ddot \phi + \frac{3 \dot a \dot \phi}{a } & = & \lambda \phi \left ( m^2 - \phi^2 \right ) \label{2 - 160}\end{aligned}\ ] ] with the following initial conditions @xmath27 the numerical solutions are presented in fig s [ fg1]-[fg5 ] . in fig .
[ fg1 ] the profiles for different @xmath28 are presented . from this figure
we see that there exist the bounce with different time duration .
every such solution can be enumerated with the number of minima or maxima . on fig s [ fg1]-[fg4 ]
the profiles of @xmath29 , hubble parameter @xmath30 and the state equation @xmath31 for @xmath32 are presented .
it is useful to present the profile of a state equation ( see , fig .
[ fg4 ] ) in the form @xmath33 where @xmath34 is the pressure and @xmath35 is the energy density . we see that in the contracting and expanding stages @xmath36 and additionally in during of bouncing off stage there are points where @xmath36
. how long could be such prolonged bouncing off ?
for the investigation of this question we are addressing to fig .
we see that for @xmath37 the solution has bouncing off but for @xmath38 the solution is singular one .
it means that there exists a special solution with @xmath39 where @xmath40 . from fig .
[ fg1 ] and fig .
[ fg5 ] we see that such solution should have an infinite long bouncing off with a contracting phase at @xmath41 and an expansion phase at @xmath42 .
the most interesting is that by @xmath28 close enough to @xmath43 we may have a bouncing off time with any duration . in during of bouncing off the size of universe is @xmath44 .
it means that the existence time of universe in such state can be somehow long with possible expansion at any time .
here we have shown that the process of universe bouncing off is not a simple process .
implicitly is supposed that bouncing off happens at . here
we have shown that it can be taking place at the planck region .
it allows us to consider a quantum birth of the universe as following process : the universe exists in the state with @xmath39 and in the consequence of quantum fluctuations of @xmath28 the universe pass to the state with @xmath45 .
after that the universe go into an inflation phase . | the cosmological solution with a long - term bouncing off time is presented .
the solution has a preceding contracting and a subsequent expanding phases and between them there exists a bouncing off phase with arbitrary time duration . |
the recent interest in the frustrated heisenberg antiferromagnets is motivated by high @xmath1-superconductivity ; the undoped compounds show long - range antiferromagnetic order , similar to the heisenberg model . upon doping superconductivity
occurs . adding frustration to the heisenberg model can be thought of as to mimic the effect of hole doping .
we consider the frustrated heisenberg model on a square lattice with @xmath2 sites .
it is described by the following hamiltonian for quantum spins @xmath3 on a lattice : @xmath4 where @xmath5 denotes a pair @xmath6 of nearest - neighbor sites and @xmath7 a pair of next - nearest - neighbor sites .
the spin length is fixed ; @xmath8 . both @xmath9 and @xmath10 are taken to be non - negative .
if @xmath11 is small , the antiferromagnetic long - range order is recovered ( nel - like ) . for @xmath11 large , the system decomposes in two nel ordered sublattices which , however , have the same quantization axis . alternating strips of up and down - spins will occur ; the so - called collinear ordering . clearly these couplings frustrate each other . if the spins were classical objects a large number of phases , among which the nel and collinear phases , would become degenerate for @xmath12 . for the quantum case , a quantum phase transition to a spin - liquid phase might occur.@xcite in this paper , our intention is to employ the spin stiffness @xmath13 to measure magnetic order in the system .
we calculate the spin stiffness in the framework of the schwinger boson mean - field approximation ( sbmfa ) using a general formula for @xmath13 .
previous evaluations of the spin stiffness were indirect @xcite .
however , apart from minor adjustments , we confirm their results .
furthermore the scaling behavior of @xmath14 in this approximation is derived .
this is useful to sensibly extrapolate results of more exact approaches , like quantum monte carlo and exact diagonalization .
also , the typical system sizes for which scaling is valid can be estimated in this way.@xcite
the sbmfa improves upon standard mean - field theory by incorporating correlations between neighboring spins @xcite . with this approximation
we derive the energies and wavefunctions of all states of the frustrated heisenberg model for both nel and collinear order .
our notation below will generally follow that of mila et al.@xcite the schwinger boson transformation is a representation of the separate spin operators by pairs of boson operators ; @xmath15 , @xmath16 , @xmath17 $ ] , supplemented by the local constraint @xmath18 . to transform the hamiltonian in a convenient form
an appropriate rotation in spin space is applied .
define @xmath19 and @xmath20 .
the hamiltonian becomes @xmath21 where we have inserted @xmath22 for the pairs @xmath6 of spins parallel in the @xmath23-direction and @xmath24 for the antiparallel pairs .
for the two orderings considered this is depicted in figure [ fig : rho_s ] .
the parameter @xmath25 equals @xmath9 for nearest - neighbors and @xmath10 for next - nearest - neighbors .
the mean - field decoupling is made using the fields @xmath26 and @xmath27 .
the local constraint @xmath28 is replaced by a global one and enforced by means of a lagrange multiplier @xmath29 .
after a fourier transform from @xmath30 to @xmath31 and a bogoliubov transformation from @xmath32 to @xmath33 ; @xmath34 the hamiltonian becomes @xmath35 where we have introduced the quasiparticle energies @xmath36 below the quantities @xmath37 , @xmath38 and @xmath39 will be defined for the nel and collinear orderings separately .
the fields @xmath40 , @xmath41 and the lagrange multiplier @xmath29 are obtained through consistency equations .
we consider two types of order : _ nel order : _ the specific form of @xmath37 , @xmath42 , @xmath39 and the three consistency equations are : @xmath43 , \\
\kappa & = & \frac{1}{n } \sum_{\bf p } \frac{h_{\bf p}+\lambda}{2 \omega_{\bf p } } \cos p_x \cos p_y \label{eq : sumkappa } , \\
\gamma & = & \frac{1}{n } \sum_{\bf p } \frac{\delta_{\bf p}}{4 \omega_{\bf p } } ( \cos p_x + \cos p_y ) \label{eq : sumgamma } , \\ 1 & = & \frac{1}{n } \sum_{\bf p } \frac{h_{\bf p}+\lambda}{2 \omega_{\bf p } } .\label{eq : sum1}\end{aligned}\ ] ] for future applications it is also useful to define the `` condensate '' @xmath44 @xcite by @xmath45 this is the combination of the -equivalent- @xmath46 and @xmath47 terms in ( [ eq : sumkappa ] ) and ( [ eq : sum1 ] ) , which both diverge for @xmath48 . in the same limit , @xmath49 , @xmath44 also equals the corresponding terms in ( [ eq : sumgamma ] ) . _ collinear order : _ for this phase we introduce quantities with a bar where confusion might arise ; @xmath50 the condensate @xmath51 and @xmath52 are defined in a similar manner as before in ( [ eq : condensate ] ) and ( [ eq : omegaq ] ) .
the discussion above has given us the ground states @xmath53 with energy @xmath54 for both orderings .
these ground states are characterized by the absence of quasi - particles ; @xmath55 .
excited states are given by : @xmath56|a \rangle .
\label{eq : excitatione}\end{aligned}\ ] ]
the spin stiffness @xmath14 is non - zero if there exists magnetic order in the system and is , at @xmath57 , associated with an increase in energy upon twisting the order parameter of the system ( @xmath58 with @xmath59 the wave vector of the twist . )
@xcite . in line with einarsson and schulz @xcite
we introduce this twist through a replacement of @xmath22 and @xmath24 in ( [ eq : sb - h ] ) by @xmath60 with @xmath61 .
the resulting hamiltonian @xmath62 is now evaluated within the sbmfa .
defining @xmath63 and @xmath64 , the mean - field hamiltonian becomes @xmath65 \nonumber \\ & & + \sum_{fm } j_{ij } \kappa_{ij}({\bf q})(d_{ij}^\dagger ( { \bf q } ) + d_{ij}({\bf q } ) - 2 \kappa_{ij}({\bf q } ) ] \nonumber \\ & & + \lambda \sum_{i } ( a_i^\dagger a_i + b_i^\dagger b_i - 1)+\hbox{constants}.\end{aligned}\ ] ] since within the sbmfa we know what the excited states are [ see ( [ eq : excitation ] ) and ( [ eq : excitatione ] ) ] , the spin stiffness can be directly evaluated from second order perturbation theory @xmath66 with the quantities @xmath67 and @xmath68 defined by @xmath69 in ( [ eq : rhofinal ] ) , we have also defined the abbreviations @xmath70 and @xmath71 for the two terms in @xmath14 .
from this point on , we set @xmath72 . of the two terms for @xmath14 in ( [ eq : rhofinal ] ) , @xmath70 is evaluated more easily .
we obtain after straightforward manipulations with brillouin zone summations : @xmath73 these simple equations hold for all system sizes @xmath74 . the quantity @xmath71 requires more effort ; the operator @xmath68 has to be expressed in the operators @xmath75 and @xmath76 defined in ( [ eq : bogoliubov ] ) . for the wavefunctions @xmath77 and energies @xmath78 of the excitations we use ( [ eq : excitation ] ) and ( [ eq : excitatione ] ) .
the resulting values for @xmath71 are written as summations over the brillouin zone : @xmath79 \nonumber \\ & & - \sin \phi \sin p_y[j_1 \gamma_1(\bar{h}_{\bf p } + \bar{\lambda } ) + 2 j_2 \gamma_2 ( \bar{h}_{\bf p } + \bar{\lambda})\cos p_x ] ) ^2 . \nonumber\end{aligned}\ ] ] only for the infinitely large lattice these equations can be simplified by replacing summations by integrals and partially integrating .
the expression for the spin stiffness @xmath80 then simplifies considerably and becomes : @xmath81 \label{eq : rhoscoll}.\end{aligned}\ ] ] ivanov and ivanov @xcite apply a different method to derive @xmath14 .
they use a modified spin wave theory which leads to the same consistency equations as the schwinger boson approach .
@xmath14 is then obtained by calculating the correlation length @xmath82 associated with the spin - spin correlation function @xmath83 and comparing this @xmath82 to the expression for @xmath82 obtained for the non - linear sigma model to two - loop order by chakravarty et al .
@xcite ( where @xmath84 with @xmath85 the temperature ) . for nel ordering their expression
is identical to our result ( [ eq : rhosneel ] ) .
it is gratifying to see that the non - linear sigma model also is the effective field theory for the low - energy physics of the _ frustrated _ heisenberg antiferromagnet . for the collinear ordering they obtain the geometrical average of our @xmath86 and @xmath87-terms , whereas we take explicitly the anisotropy of this phase into account , still both expressions for @xmath88 vanish at the same value of @xmath11 .
table [ tab : tj ] and figure [ fig : rho_s ] contain our numerical results .
it is necessary to know the size dependence of observables to obtain a good approximation for their limit values .
neuberger and ziman @xcite derived the scaling behavior for an unfrustrated heisenberg antiferromagnet explicitly .
here we extend this to the case of frustration .
here we only treat the nel ordering .
we want to know the scaling behavior of the condensate @xmath44 and the two terms @xmath71 and @xmath70 that make up @xmath14 ( @xmath80 ) .
the latter two will turn out to have different scaling behavior . as can be seen from ( [ eq : tyneel ] ) only the scaling behavior of @xmath89 and @xmath90 is required for @xmath70 .
these two are part of the set @xmath91 of mutually dependent quantities .
we will now argue what is the exponent of their scaling behavior and therefore of @xmath70 , without trying to obtain the precise prefactor ( which would be quite tedious ) .
name the equations ( [ eq : sumkappa])-([eq : sum1 ] ) @xmath92 , @xmath93 , and @xmath94 , respectively .
they contain poles at @xmath46 and @xmath47 .
with help of ( [ eq : omegaq ] ) , ( [ eq : hq ] ) and ( [ eq : dq ] ) we rearrange them as @xmath95 , @xmath96 and @xmath97 .
we neglect the @xmath98 and @xmath99 terms in the summations .
it is easy to show that this will give rise to errors of the order @xmath100 .
next we expand these equations to first order round their infinite size values @xmath101 .
define the size dependences @xmath102 , @xmath103 and @xmath104 to obtain the equation @xmath105 = \vec{a } + \vec{\vec{b } } \cdot \left [ \begin{array}{c } \delta \kappa_n \\ \delta \gamma_n
\\ \delta \lambda_n \end{array } \right ] , \label{eq : roughly}\ ] ] where @xmath106 and @xmath107 contain summations over the brillouin zone dependent on the infinite size parameters @xmath108 , @xmath109 , and @xmath29 .
the remaining size dependence of @xmath107 can be neglected as it leads to higher order terms . on the other hand
the summations in @xmath106 will be replaced by integrations plus size dependent corrections . using neuberger and ziman @xcite we obtain @xmath110 = \frac{1}{n^{3/2 } } \vec{c}.\ ] ] the parameters @xmath89 , @xmath90 and @xmath111 thus scale with @xmath112 .
a direct consequence of this is that @xmath113 .
if the size dependence of the parameters is neglected [ @xmath114 _ inside _ the summations ( [ eq : sumkappa])-([eq : sum1 ] ) ] , we also find the prefactor : @xmath115 .
\label{eq : tscalesimple}\ ] ] upto @xmath116 this is in good agreement with the numerical solution of ( [ eq : sumkappa])-([eq : sum1 ] ) for various sizes @xmath74 .
next we consider the condensate @xmath117 and @xmath118 , for which we can even derive the prefactors .
replacement of @xmath91 by @xmath101 in the summation of ( [ eq : condensate ] ) leads to errors of the order @xmath119 .
we will neglect these .
neuberger and ziman @xcite state a lemma applicable to this summation .
this leads to : @xmath120 for @xmath118 we find in a similar fashion : @xmath121 these formulas are in excellent agreement with the numerical solution to ( [ eq : sumkappa])-([eq : sum1 ] ) . for @xmath118
this is depicted in figure [ fig : jyscale ] . in conclusion , we have obtained the scaling behavior of @xmath70 and @xmath71 from an analysis of the formulas in the sbmfa for the frustrated heisenberg antiferromagnet . the qualitative scaling behavior ( i.e. the exponents ) is the same as for the _ unfrustrated _ case , which was discussed by neuberger and ziman @xcite .
the scaling behavior that was utilized by einarsson and schulz @xcite does agree with our findings but we confirm the message of feiguin et al.@xcite that the clusters they used are too small for the scaling behavior of @xmath71 to have set in .
numerically we see that the scaling behavior starts around size @xmath122 ( see figure [ fig : jyscale ] ) whereas the largest cluster they used is @xmath123 .
p. chandra , b. doucot , phys . rev .
b * 38 * , 9335 ( 1988 ) .
f. mila , d. poilblanc , c. bruder , phys .
b * 43 * , 7891 ( 1991 ) . n.b . ivanov , p. ch .
ivanov , phys .
b * 46 * , 8206 ( 1992 ) .
dotsenko , o.p .
sushkov , phys .
b * 50 * , 13821 ( 1994 ) .
t. einarsson , h.j .
schulz , phys .
b * 51 * , 6151 ( 1995 )
feiguin , g.j .
gazza , a.e .
trumper , h.a .
ceccatto , phys .
b * 52 * , 15043 ( 1995 ) .
schulz , t.a.l .
ziman , d. poilblanc , j. phys . 1 france * 6 * , 675 ( 1996 ) .
sushkov , cond - mat 9602150 .
m. zhitomirsky , k. ueda , phys .
b * 54 * , 9007 ( 1996 ) d.p .
arovas , a. auerbach , phys .
b * 38 * , 316 ( 1988 ) , a. auerbach , d.p .
arovas , phys .
lett . * 61 * , 617 ( 1988 ) .
a. auerbach , _ interacting electrons and quantum magnetism _ ( springer , new york , 1994 ) .
s. chakravarty , b.i .
halperin , d.r .
nelson , phys .
b * 39 * , 2344 ( 1988 ) . j.e .
hirsch , s. tang , phys .
b * 39 * , 2850 ( 1989 ) .
fisher , m.n .
barber , d. jasnow , phys .
rev . a * 188 * , 1111 ( 1973 ) .
van leeuwen , m.s.l .
du croo de jongh , p.j.h .
denteneer , j. phys . a : math
* 29 * , 41 ( 1996 ) .
h. neuberger , t. ziman , phys .
b * 39 * , 2608 ( 1989 ) .
.[tab : tj ] the limit ( @xmath124 ) values for @xmath70 , @xmath125 and @xmath14 as function of the ratio @xmath11 where @xmath126 .
the two orderings considered are nel ( n ) and collinear ( c ) order . for the collinear ordering there are two directions : along the antiparallel spins , @xmath127 ( listed first ) and along the parallel spins , @xmath128 ( listed secondly ) . [ cols="^,^,^,^,^,^,^,^",options="header " , ] | we calculate the spin stiffness of the @xmath0 frustrated heisenberg antiferromagnet directly from a general formula which is evaluated in the schwinger boson mean - field approximation .
both nel and collinear ordering are considered . for collinear
ordering , we take the anisotropy of this phase into account , unlike previous approaches . for nel
ordering , a detailed study is made of the finite - size scaling behavior of the two terms that make up the spin stiffness .
the exponents of the scaling with the system size of the two terms comprising the spin stiffness turn out to be identical to those of the unfrustrated case .
pacs numbers : 75.10.jm , 75.30.kz , 75.40.cx |
a major impediment to using semi - classical general relativity is calculating the renormalized expectation value of the stress tensor .
properly renormalized values for @xmath2 and @xmath3 provide information on particle production and spontaneous symmetry breaking , and are also required to calculate backreaction . since in general relativity energy density is itself a source of curvature , great care must be taken in deciding what may be dismissed as ` unphysical ' .
fortunately , there are several generally accepted renormalization schemes for curved space - times @xcite . our purpose is to present a compact formula for the renormalization terms that may be applied to @xmath2 and @xmath3 calculations in arbitrary black hole space - times of even dimension .
calculating @xmath3 for a general d - dimensional black hole space - time is difficult . for a scalar field , @xmath4 and its derivatives ,
so we start with the simpler problem of calculating @xmath5 , where @xmath6 is the hartle - hawking vacuum .
note that @xmath2 is the coincidence limit of the two point function @xmath7 , and so may be expressed in terms of green s functions . in particular , the feynman green s function is related to the time ordered propagator , @xmath8 .
a wick rotation allows us to work in euclidean space where @xmath9 .
the euclidean green s function , @xmath10 , now obeys @xmath11 where @xmath12 is the laplace - beltrami operator in @xmath13-dimensional curved euclidean space . to solve for @xmath10 , start with the euclidean metric for a static space - time in @xmath13 dimensions with line element
@xmath14 here @xmath15 is the euclidean time , @xmath16 , @xmath17 is a radial coordinate , and @xmath18 represents a @xmath19-dimensional angular space .
the only restriction for this method is that the metric must be diagonal .
if the scalar field is at temperature @xmath20 , then the green s function is periodic in @xmath21 with period @xmath22 . assuming a separation of variables , standard green s function techniques lead to the formal solution @xmath23 where @xmath24 , and @xmath25 are eigenfunctions of the helmholtz equation obtained from the from the angular part of eq .
( [ eq : green ] ) . for black holes with spherical topology
these are equivalent to the set of hyperspherical harmonics .
the radial function @xmath26 obeys a complicated differential equation obtained by putting the above expression into eq .
( [ eq : green ] ) .
this expression is divergent in the sum over @xmath27 .
the @xmath28 computation has been reduced to computating the coincidence limit of the green s function a divergent quantity . to assign physical meaning to @xmath2 it
must be rendered finite via some renormalization process , and the standard approach is to renormalize the expression for @xmath29 via christensen s point splitting method applied to the dewitt - schwinger expansion of the propagator @xcite . in @xmath13 dimensions ,
the adiabatic dewitt - schwinger expansion of the euclidean propagator is @xcite @xmath30 equation ( [ eq : schwingdewitt ] ) introduces several new variables .
let @xmath31 be the geodesic distance between @xmath32 and @xmath33 , then define @xmath34 and @xmath35 .
the @xmath36 are called dewitt coefficients , and @xmath37 is a hankel function of the second kind .
lastly , @xmath38 is the van vleck
morette determinant , where @xmath39 and @xmath40 . using the derivative properties of bessel functions , noting that @xmath41 is purely imaginary in euclidean space , and defining @xmath42 , eq .
( [ eq : schwingdewitt ] ) can be written as latexmath:[\[\label{eq : sdwintermediate } g_{\rm e}^{\rm ds}({x},{x ' } ) = \frac{-2i \triangle^{1/2}}{(4\pi)^{d/2 } } \sum_{k= 0 } a_k({x},{x'})(2m^2)^{\nu } the dewitt - schwinger expansion is a wkb expansion of the euclidean propagator for a generic space - time when the point separation is small . for a particular space - time
, this procedure does not give the correct results for the green s function with finite point separation because it ignores global space - time properties that determine the green s function such as the effective potential around a black hole
but it should reproduce the same divergent terms in the coincidence limit . therefore ,
if the divergent terms of the dewitt - schwinger expansion can be isolated , then subtracting these terms from @xmath29 will make it finite as @xmath44 .
since we are working in euclideanized space the physical renormalization terms come from the real part of eq .
( [ eq : sdwintermediate ] ) .
the asymptotic behavior of @xmath45 as @xmath46 implies that only terms with @xmath47 contribute divergences in the coincidence limit , so @xmath48 to renormalize @xmath10 , eq .
( [ eq : gdivterms ] ) must be made commensurate with eq .
( [ eq : greenfuncnonzerotemp ] ) .
we have shown @xcite that an integral representation of @xmath49 for small @xmath50 and integer - valued @xmath51 is @xmath52 changing of variables and using the plana sum formula to convert the integral to a sum , the renormalization terms for the @xmath13-dimensional space - time of eq .
( [ eq : metric ] ) are @xcite @xmath53\kappa\sqrt{\pi}}{(-f)^{\nu } \gamma(\nu+\frac{1}{2 } ) } \bigg[\sum_{n=1}^{\infty}\cos(\kappa\varepsilon n ) \left(\kappa^2n^2+m^2f\right)^{\nu-\frac{1}{2 } } - \frac{1}{2}(\kappa^2+m^2f)^{\nu-\frac{1}{2 } } \\ - i\int_0^{\infty } \frac{dt}{e^{2\pi t}-1}\left\{\left[(1+it)^2\kappa^2+m^2f\right]^{\nu-1/2 } - \left[(1-it)^2\kappa^2+m^2f\right]^{\nu-1/2 } \right\ } \\ + ( m^2f)^{\nu-\frac{1}{2}}\phantom{}_2f_1 \left(\frac{1}{2},\frac{1}{2}-\nu,\frac{3}{2},-\frac{\kappa^2}{m^2f}\right ) \bigg]+ [ a_k]e_{\nu } + \sum_{n=1}^{\nu}\sum_{p=1}^{2n}\sum_{j=0}^{p } \frac{2^{2n-1}(-m^2)^{\nu - n}\gamma(n)}{\gamma(\nu - n+1 ) } \frac{a^j_k\delta^{1/2}_{p - j}}{(\sigma^{\rho}\sigma_{\rho})^n } \bigg\}\end{gathered}\ ] ] for a scalar field at nonzero temperature @xmath54 . in this expression the @xmath55
are terms depending on the metric function @xmath56 and have been tabulated elsewhere @xcite , while @xmath57 and @xmath58 represent the @xmath59 term in an expansion in powers of @xmath60 .
this expression generalizes previously known four - dimensional results @xcite .
the corresponding renormalization terms for a scalar field at zero temperature @xmath61 are similarly found @xcite .
semi - classical general relativity requires calculation of @xmath62 in complicated possibly higher dimensional space - times .
the first step in calculating @xmath62 for a scalar field is calculating @xmath63 .
we have presented a compact expression for the renormalization terms for @xmath28 in even dimensional , static , black hole space - times .
this work partially funded by fct , through project no . cern / fp/109276/2009 | a compact expression for the dewitt - schwinger renormalization terms suitable for use in even - dimensional space - times is derived
. this formula should be useful for calculations of @xmath0 and @xmath1 in even dimensions . |
the neutrino oscillation phenomenon has been clearly established by the study of solar , atmospheric , reactor and beam neutrinos .
the pmns mixing matrix relates the three neutrino mass eigenstates to the three neutrino flavor eigenstates .
this can be parametrized by three mixing angles ( @xmath8 , @xmath0 , @xmath9 ) and one cp violating phase @xmath10 ( if neutrinos are dirac particles ) . during the last years
, tremendous progress has been achieved in the experimental field trying to measure the values of @xmath11 and the two squared mass differences @xmath12 which govern the oscillation probabilities .
the two mass differences and the @xmath8 and @xmath9 mixing angles have been measured with good precision @xcite . however ,
the @xmath0 angle , the sign of @xmath13 ( mass hierarchy ) and the @xmath10 phase are still unknown . in particular , only an upper limit on the value of @xmath0 has been established indicating that the angle is very small compared to the other mixing angles .
a three - flavor global analysis of the existing data provides a constraint on @xmath0 being @xmath14 @xmath15 0.028 at 90% c.l @xcite .
this limit is essentially dominated by the result obtained by the chooz reactor experiment @xcite in france .
this experiment measured the fraction of @xmath16 s surviving at a distance of 1.05 km from the reactor cores to be r = 1.01 @xmath17 2.8% ( stat ) @xmath17 2.7% ( syst ) .
this result was mainly limited by the systematic uncertainties induced by the imperfect knowledge of the neutrino production and interaction .
the measurement of this angle is of fundamental interest not only for the final understanding of neutrino oscillations but because it determines the possibilities to observe cp violation in the leptonic sector with the forthcoming neutrino experiments .
the information on the @xmath0 mixing angle can be essentially obtained from accelerator or nuclear reactor neutrino experiments .
the long baseline accelerator experiments measure the appearance of @xmath18 s in a @xmath19 beam generated at long distance from the detector .
the @xmath19 @xmath20 @xmath18 transition depends on several oscillation parameters like the cp phase and the sign of @xmath13 .
moreover , they can also be sensitive to matter effects due to the long baselines .
therefore , the measurement of @xmath0 will be affected by correlations and degeneracies between parameters and the sensitivity of the accelerator experiments to this parameter will be reduced . on the other hand ,
reactor experiments are unique to provide an unambiguous determination of @xmath0 .
nuclear reactors are very intense sources of @xmath16 s coming from the @xmath21-decay of the neutron - rich fission fragments of fuel elements ( u and pu ) .
reactor neutrino experiments will look for the disappearance of @xmath16 s with energies extending up to 10 mev over distances of the order of kilometers ( short baselines ) to maximize the disappearance probability . due to the low energy of the emitted electron antineutrinos
, only disappearance measurements can be performed .
the survival probability of @xmath16 s emitted by a nuclear power station can be written as : @xmath22 where @xmath23 is the neutrino energy and @xmath24 the distance from the source to the detector .
only the first term is relevant for short baselines and therefore the oscillation amplitude is proportional to @xmath14 and independent of the cp phase and the mass hierarchy .
a clean measurement of @xmath0 can be performed with reactor neutrino experiments since they do not suffer , unlike accelerator experiments , from degeneracies and correlations between different oscillation parameters .
since they are short baseline experiments , they are not affected by matter effects .
reactor antineutrinos are detected through the inverse beta decay @xmath16 + p @xmath20 n + @xmath25 .
the energy threshold for this reaction is 1.806 mev .
the signature of the neutrino interaction is the coincidence of the photons from the prompt signal due to the @xmath25 annihilation and the delayed signal ( @xmath4 t @xmath26 30 @xmath27s ) from the neutron capture .
many liquid scintillator @xmath28 experiments use scintillator loaded with gadolinium in their fiducial volume because of its large neutron capture cross section and high total @xmath29 yield of 7 - 8 mev . the neutrino spectrum results from the convolution of the neutrino flux and the cross section .
the spectrum peaks at @xmath26 4 mev .
reactor experiments will look for a suppression of the @xmath16 rate and a distortion of the spectrum shape due to oscillations .
the signature of a neutrino interaction can be mimicked by two kind of background events : accidentals and correlated .
the accidental background corresponds to the coincidence of a positron - like signal coming from natural radioactivity of the surrounding environment or of the detector materials , with a neutron induced by cosmic muon spallation in the surrounding rock and captured in the detector .
the correlated background are events that mimic both parts of the coincidence signal .
they come from fast neutrons produced by cosmic muons , which can produce proton - recoils in the target scintillator ( misidentified as e@xmath30 ) and then are captured after thermalisation or from the @xmath21 - n decay of long - lived cosmogenic radioisotopes ( @xmath31li , @xmath32he ) produced by muon interactions in the scintillator .
some of the largest systematic uncertainties of the chooz reactor experiment are related to the accuracy to which the original neutrino flux and spectrum are known . in order to improve the chooz sensitivity to the @xmath0 mixing angle ,
a relative comparison between two or more identical detectors located at different distances from the power plant is required .
the first one , which is located at few hundred meters from the nuclear cores , monitors the neutrino flux and spectrum before neutrinos oscillate .
the second detector located 1 - 2 km away from the cores searches for a departure from the overall 1/l@xmath5 behavior of the neutrino energy spectrum . at the same time , the statistical error can also be reduced by increasing the exposure and the fiducial volume of the detector .
moreover , backgrounds can be further reduced with a better detector design , using veto detectors and external shields against muons and external radioactivity .
several nuclear reactor neutrino experiments are foreseen to measure the @xmath0 mixing angle : double chooz @xcite in france , daya bay @xcite in china , reno @xcite in south korea and angra @xcite in brazil . in the next section
the first of these experiments is explained in detail .
the double chooz experiment @xcite will improve our knowledge on the @xmath0 mixing angle within a competitive time scale and for a modest cost .
the double chooz collaboration is composed by institutes of brazil , france , germany , japan , russia , spain , uk and usa . the experiment is being installed in the chooz - b nuclear power plant in the northeast of france . the maximum operating thermal power of the two cores is 8.54 gw@xmath33 .
the far detector is located at 1050 meters from the cores in the same laboratory used by the chooz experiment ( see fig . [ chooz ] ) .
it provides a quickly prepared and well - shielded ( 300 meters of water equivalent ( m.w.e . ) ) site with near - maximal oscillation effect . a second identical detector ( near detector )
will be installed at 400 m away from the reactor cores to cancel the lack of knowledge of the neutrino spectrum and reduce the systematic errors related to the detector . for the near site
, a tunnel will be excavated under a small natural hill ( overburden 115 m.w.e . ) and a near lab will be equipped .
location of the two double chooz detectors in the chooz - b nuclear power plant.,width=321 ] in addition to the neutrino oscillation studies , double chooz aims to study the possibility of using this kind of detector to verify the non - proliferation of nuclear weapons in the framework of the international atomic energy agency missions .
a near detector could be able to provide a precise measurement of the reactor power and isotopic composition by looking at the energy spectrum and rate of the emitted antineutrinos , following its evolution with time .
the chooz detector design can also be optimized in order to reduce backgrounds .
the double chooz detectors ( see fig . [ detector ] ) consist of concentric cylinders and an outer plastic scintillator muon veto .
the innermost volume ( `` target '' ) contains about 10 tons of gd - loaded liquid scintillator ( @xmath260.1% gd ) within a transparent acrylic vessel
. this will be the volume for neutrino interactions .
it is surrounded by a 55 cm thick layer of unloaded scintillator ( `` @xmath29-catcher '' ) contained in a second acrylic vessel .
this scintillating volume is necessary to fully contain the energy deposition of gamma rays from the neutron capture on gd as well as the positron annihilation gamma rays inside the central region .
it also improves the rejection of the fast neutron background .
surrounding the @xmath29-catcher , a 105 cm thick region contains non - scintillating oil inside a stainless steel `` buffer '' vessel .
this volume reduces by two orders of magnitude with respect to chooz the level of accidental backgrounds coming mainly from the radioactivity of the photomultiplier tubes ( pmts ) .
390 10 pmts are installed on the inner wall and lids of the tank to collect the light from the central scintillating volumes , providing about 13% photocathode coverage .
the central detector is encapsulated within a `` inner muon veto '' shield , 50 cm thick , filled with scintillating organic liquid and instrumented with 78 8 pmts .
it allows the identification of muons passing near the active detector that can create spallation neutrons and backgrounds coming from outside . because of space constraint , the 70 cm sand shielding of chooz
is replaced by a 15 cm iron layer to protect the detector from rock radioactivity and to increase the target volume .
an `` outer muon veto '' covers the top of the main system and provides additional rejection power for cosmic - induced events .
it can be used for constant mutual efficiency monitoring with the inner veto .
the double chooz detector design.,width=321 ] the near and far detectors will be identical inside the pmt support structure , allowing a relative normalization error of 0.6% or less , to be compared with the 2.7% systematic error of the chooz experiment .
figure [ farlab ] shows a view of the far detector inside the experimental hall .
calibration systems will be deployed periodically into the target and @xmath29-catcher allowing to check the stability of the system .
the deployment of radioactive sources must be performed in a clean environment under a dry nitrogen atmosphere .
a glove box interface with an associated clean room will be installed on the top of the detector .
the electronic racks will be located at the end of the lab .
the double chooz far detector in the experimental hall.,width=302 ] an intense r&d effort has been carried out by the double chooz collaboration to validate the robustness of the detector concept and the feasibility of the sensitivity goals .
some examples of the activities performed are : * the construction and operation of a 1:5 scale prototype allowed us to validate the technical choices for the vessels construction and integration of the components , check the material compatibility with the scintillator , study the liquid scintillation performances and filling system , the calibration techniques , safety and all the instrumentation needed to operate such a detector .
* production and tests of stable high - quality gd - loaded scintillator , critical for the performance of the experiment .
the long - term stability , optical transmission properties , light yield , transparency and radiopurity have been tested for several years . *
magnetic measurements of the performance of the 10 pmts inside a controlled magnetic environment and design and optimization of individual pmt magnetic shields . *
development of 8-bit flash adcs waveform digitizers for the readout of each pmt signal , recording the pulse shape for possible offline particle identification studies .
many systematic uncertainties that affected chooz and all previous single - baseline reactor neutrino experiments are greatly reduced by having both near and far detectors .
table [ sys ] summarizes the systematic uncertainties in the measurement of the antineutrino flux comparing both chooz and double chooz detectors .
.systematic uncertainties in chooz and double chooz reactor experiments . [ cols="<,^,^",options="header " , ] the veto systems are also in production .
the encapsulation process of the inner veto pmts is on going .
a prototype of the outer veto system is in construction .
a complete calibration program has been designed including the deployment of gamma and neutron radioactive sources in the different liquids of the detector with different systems depending on the volume to be accessed ( guide tubes , articulated arm , fish - line , etc . ) and light flashers ( leds and lasers ) to measure the pmts .
all these systems are in construction
. the far detector integration will be finished by summer 2009 .
the near lab site will be available at the beginning of 2010 to accommodate the second detector . in 2011 , both near and far detectors , will be operative and taking data for three more years in order to achieve double chooz expectations .
double chooz is the first reactor neutrino experiment of a new generation using two identical detectors at different distances to measure the still unknown @xmath0 mixing angle .
double chooz has already started its installation and it is expected to start taking data with the far detector in 2009 and with both near and far detectors in 2011 .
double chooz will be able to measure @xmath1 to 3@xmath2 if @xmath1 @xmath3 0.05 .
otherwise , it will exclude the mixing angle down to @xmath1 @xmath3 0.03 at 90% c.l .
after three years of operation with both detectors in case of no oscillation were observed .
this will represent an improvement of about a factor 7 compared to the chooz limit and will open the way for a new level of accuracy in reactor neutrino experiments .
the information gained with double chooz will complement future results with accelerator experiments , affected by degeneracy problems , helping to better constrain the last undetermined mixing parameters . | the double chooz reactor neutrino experiment will be the next detector to search for a non vanishing @xmath0 mixing angle with unprecedented sensitivity , which might open the way to unveiling cp violation in the leptonic sector .
the measurement of this angle will be based on a precise comparison of the antineutrino spectrum at two identical detectors located at different distances from the chooz nuclear reactor cores in france .
double chooz is particularly attractive because of its capability to measure @xmath1 to 3@xmath2 significance if @xmath1 @xmath3 0.05 or to exclude @xmath1 down to 0.03 at 90% c.l . for @xmath4m@xmath5 = 2.5 @xmath6 10@xmath7 ev@xmath5 in three years of data taking with both detectors .
the installation of the far detector started in may 2008 and the first neutrino results are expected in 2009 .
the advantages of reactor neutrino experiments to measure the @xmath0 mixing angle are described in this article and in particular , the design , current status and expected performance of the double chooz detector . |
the most accepted explanation for the energy source of quasars and active galactic nuclei ( agn ) is the accretion of matter onto supermassive s in the center of the host galaxy .
the amount of accreted matter required to explain the the observed energy output is 0.001-few solar masses per year , while the central mass required to maintain steady accretion ( the eddington limit ) is of the order of @xmath0 . comparing these two numbers gives an age of @xmath1 years , which indicates a timescale for the evolution of the mass and the agn properties .
conservation of angular momentum directs the accreted material to form a disk around the central and the properties of such an accretion disk may be calculated under a few basic assumptions ( e.g. sakura and sunyaev , 1973 ) . in particular one can calculate the luminosity and spectral distribution of the radiation from the accretion disk , given the mass and the . as the mass grows due to the accreted material , the luminosity and the emerging spectrum change with time .
since the continuum radiation ionizes the line - emitting material and drives the broad agn s , the change in the properties of the radiation from the central source changes also the properties of the emission lines .
in particular the spectral shape of the ionizing radiation would determine the ( the ratio between the energy flux in the line to the continuum energy flux at the line frequency ) of the s. baldwin ( 1977 ) showed that the of the civ decreases with increasing continuum luminosity , a correlation known as the baldwin effect. a similar relation has been found also for other broad s. the origin of this effect is not understood .
in this work we suggest that at least part of the can be attributed to evolution .
we estimate the relation between the and the continuum spectrum , and calculating the evolution of the continuum emission from the we show that for plausible evolutionary tracks the is anti - correlated with the continuum luminosity .
the basic properties of a compact , accretion powered radiation source must obey several basic relations : in order to maintain steady spherical accretion the luminosity must be less than the luminosity , @xmath2 or @xmath3 where @xmath4 , @xmath5 is the ratio , and @xmath6 .
if a luminosity @xmath7 comes from a region of radius r and temperature @xmath8 , then @xmath9 if a spectral feature at a photon energy of @xmath10 is due to emission from a region of size @xmath11 , then the temperature is given by @xmath12 and reversing the relation above we have an upper limit on the mass ( for a true spectrum this becomes an approximate equality ) :
@xmath13 where @xmath14 cm is the radius and @xmath15 is the spectral energy in rydbergs .
combining eqs .
[ equ : med ] and [ equ : mt ] we can eliminate @xmath16 obtaining ( for emission ) an estimate of the size : @xmath17 the shortest time scale for global variations in the luminosity is the light travel time across the radius , hence if the luminosity is observed to vary significantly on a time scale @xmath18 , the mass has to be @xmath19 where @xmath20 is the effective radius of emission in units of @xmath21
. however , for continuum changes near the feature of spectral energy @xmath10 the light travel time is ( from eq .
[ equ : rle ] ) @xmath22 and the dynamical time is @xmath23 which gives time scales of the order of the observed uv variability in seyferts and quasars .
* d. evolution of the mass : * we may define the time @xmath24 as the time required for an accretion - powered object radiating at the luminosity to double its mass , due to accreted matter , @xmath25 the observed luminosity implies an accretion rate of @xmath26 where @xmath27 is the efficiency .
combining these two expressions gives the mass , accumulated if the accretion rate is maintained during a time @xmath24 : @xmath28 where @xmath29 is the fraction of the active time for intermittently active agn .
many authors tried to fit spectra to the observed agn continuum , thus finding the parameters ( @xmath30 ) and the angular momentum of the ( or kerr ) which best fit the observed continuum in the uv ( wandel and petrosian 1988 ; sun and malkan 1991 ) or in the soft x - rays ( laor 1990 ) . the radiative energy output of the is dominated by the release of gravitational potential energy , the rate of which is @xmath31 . a more accurate calculation yields @xmath32\ ] ] where @xmath33 is the inner disk radius .
in the outer part of the thin disk the opacity is dominated by true absorption and the local spectrum is a spectrum . for this part of the disk , comparing eq .
[ equ : lr ] to the radiation flux gives for the disk surface temperature @xmath34 where @xmath35 and @xmath36 .
when the approaches the rate , the intermediate disk region becomes dominated by electron - scattering , the spectral function will be of a modified and the surface temperature is higher than given by eq .
[ equ : tr ] . at still smaller radii ,
the pressure in the disk is dominated by radiation , rather than by gas pressure , and the thin disk solution becomes thermally unstable . in that inner region
the thin disk solution is probably not valid , and has to be replaced by a hot disk solution ( e.g. wandel and liang 1991 ) .
it turns out that the intermediate modified region is relatively narrow , and close to the - electron scattering boundary the spectrum is nearly a one , so both regimes may be approximated by the solution .
the spectrum of the ( and modified ) disk is given by integrating over the entire disk , @xmath37 dr\ ] ] where @xmath38 is the planck function and @xmath39 is the transition radius from the intermediate to the inner radiation - pressure dominated region ( for @xmath40 the region extends down to the inner edge of the disk and @xmath41 ) .
since the @xmath42 has a sharp peak at @xmath43 and cuts off at higher frequencies , the highest frequency of the part of the disk spectrum comes from the radius @xmath39 , with the highest temperature for which the disk is still optically thick .
[ equ : fn ] gives a spectrum which depends on the radial extent of the part of the disk .
if that part is extended ( @xmath44 ) the spectrum is almost flat ( @xmath45 ) and cuts off beyond @xmath46 ( or @xmath47 for @xmath48 ) . if @xmath49 a few , the spectrum will be merely a somewhat broadened planck spectrum .
while in the uv band the thin multiple spectrum may be a good approximation , the soft x - rays may be produced by a hotter medium due to processes other than emission , such as comptonization , a two - temperature disk ( wandel and liang 1991 ) or hot corona ( haardt , maraschi and ghisellilni 1994 ; czerny , witt and zycki 1996 ) .
we want to find an approximate relation between the mass and and the continuum spectrum .
the is determined straightforwardly by the luminosity , via the relation @xmath51 , where @xmath52 is the bolometric correction for the ionizing continuum . to estimate the mass we may use the multi- -spectrum . as
discussed in the previous subsection the uv bump cutoff frequency is determined by the highest surface temperature in the part of the disk . if the region extends down to a radius @xmath39 , and the euv cutoff energy is @xmath53 , then eq . [ equ : mt ] gives ( we assume the inner region is optically thin and much hotter , so its contribution in the frequencies of interest can be neglected ) @xmath54 in several models the regime extends close to the inner disk edge .
this is the case in the `` @xmath55 '' disk model with viscosity proportional to the gas pressure ( @xmath56 ) . for low s ( @xmath57 )
this is true also for the `` @xmath58 '' model with viscosity proportional to the total pressure ( @xmath59 ) in this case , since the disk emissivity peaks at @xmath60 ( for a non - rotating ) we may use eq . [ equ : mte ] with @xmath61 in order to find the mass : @xmath62 .
it is possible to do a more refined treatment , calculating for each pair of parameters @xmath16 and @xmath50 fitting not the total luminosity and blue - bump temperature , but the actual observables , e.g. the uv luminosity and spectral index .
wandel and petrosian ( 1988 ) have calculated the flux and spectral index at 1450 , for a grid of parameters @xmath63 .
inverting the grid they have obtained contours of constant mass and constant ratio ( @xmath64 ) in the @xmath65 plane ( fig 1 ) . plotting in this plane samples of agn it is possible to read off the contours the corresponding parameters . comparing several groups of agn
a systematic trend appears : higher redshift and more luminous objects tend to have larger masses and luminosities closer to the limit ( see table 1 ) .
similar results are obtained by sun and malkan ( 1991 ) .
lccc & & & + agn group & log@xmath66 & log @xmath67 & @xmath68 + & & & + & & & + seyfert galaxies & 28 - 29.5 & 7.5 - 8.5 & 0.01 - 0.5 + low z quasars & 29 - 30.5 & 8 - 9 & 0.02 - 0.1 + medium z quasars & 30 - 31.5 & 8 - 9.5 & 0.1 - 0.5 + high z quasars & 31 - 32 & 9 - 9.5 & 0.03 - 2 + & & & + 1 wandel1w2fig.ps 1 .
accretion - disk evolution tracks in the @xmath69 plane ( wandel and petrosian 1988 ) .
crosses : seyfert galaxies , triangles , open circles and filled circles : low , medium and high redshift quasars , respectively .
continuous lines : constant mass ; dashed lines : constant @xmath70 ( note that the labeling of the curves in the figure follows the notation @xmath71 ) .
the of an is actually the ratio between the luminosity in the line and the continuum luminosity at the frequency .
the line luminosity is determined by the photoionizing flux , near 1ryd ( 13.6 ev ) . since @xmath72 is just below 13.6 ev ( see below ) ,
this frequency is on the wien part , and the flux is sensitively influenced by the effective disk temperature , @xmath73 . in the parameter range of interest
we may therefore assume that the is strongly correlated with @xmath73 .
quantitatively , the correlation is related to the logarithmic derivative @xmath74 where @xmath75 . for @xmath761 ryd , which is true in a wide range of cases , as we show below , eq .
[ equ : lblt ] gives @xmath77 namely a strong correlation with t. in order to relate the uv ionizing spectrum near 1ryd to the accretion parameters , it is necessary to determine the effective temperature of the emitting region .
we have already derived a quite general temperature - radius relation for emission from an accretion flow ( eq . [ equ : rle ] ) .
we can apply this relation to determine the temperature or radiation peak energy as a function of the observed luminosity and estimated mass . in order to determine a characteristic temperature we need either
indicate the size of the emitting region ( and use eq .
[ equ : lrt ] ) or assume an emission model , as done below for the model .
the size of the emitting region can be bounded by variability analyses . from eqs .
[ equ : lrt ] and [ equ : mdt ] we have @xmath78 if we assume a linear relation between luminosity and variability time , for example , that for objects with a continuum luminosity @xmath79 the uv variability time is one month , eq .
[ equ : elt ] gives @xmath80 in the framework of the thin model , we consider four models for determining the effective temperature .
* a. the inner disk temperature * when the disk is nearly up to the inner edge , we can approximate the spectrum by emission near the maximum - emissivity radius , which gives @xmath81 as discussed in sec . 3 above
, this model applies to the `` @xmath55 '' disk and to low s ( @xmath57 ) in the @xmath58 disk . *
b. the inner disk radiation - pressure temperature * this model applies to high s ( @xmath82 ) in the @xmath58 disk , provided the inner disk maintains the temperature profile given by the thin - disk solution .
the temperature in the inner , radiation dominated region is given by ( sakura and sunyaev 1973 ) @xmath83^{1/4}~k\ ] ] substituting the maximal - emissivity radius @xmath84 gives @xmath85 the gas - radiation pressure boundary - taking as @xmath86 the temperature at the inner boundary between the and the radiation - dominated disk * c. the -radiation boundary temperature * the effective temperature is taken at the boundary between the intermediate and inner disk regions , at @xmath87 .
the energy of the peak in the spectrum is then derived by substituting @xmath88 into eq . [
equ : trad ] , which gives @xmath89 this model is adequate in the case of a high - @xmath58 disk when the inner disk does not follow the thin disk model but establishes a stable hot solution ( e.g. wandel and liang 1991 ) .
* d. the `` photospheric '' temperature * in this model we consider as the disk effective temperature the temperature at the radius where the disk becomes effectively optically thin .
the effective optical depth in the disk is given by @xmath90 where @xmath91 and @xmath92 are the disk vertical scale - height and density , respectively , and @xmath93 and @xmath94 are the electron - scattering and absorption opacities . equating @xmath95 to unity and eliminating @xmath8 and @xmath20 by using the temperature profile @xmath96 gives @xmath97 note that the last two models give very similar results .
the evolution of the mass over cosmological times depends on the , which may of course change over time . from the efficiency and the observed luminosity we can estimate the present . in order to determine the evolution , we need to specify the variation of the over time .
we consider three representative scenarios : * constant * limited accretion ( constant ratio ) * spherical accretion from a homogeneous medium this would correspond to external feeding of the , which is regulated by an external potential , e.g. by stellar encounters , where the stellar motions are governed by the gravitational potential in the bulge of the host galaxy , far from the influence of the . as the mass of the grows , @xmath50 does not change , and hence the accretion parameter , @xmath98 .
eventually will become very sub- , and the will be starved , or at least on a diet . since @xmath99 , in this scenario the ( bolometric ) luminosity also is also constant .
note that the spectral energy distribution does change ( though slowly ) as @xmath16 increases
. this would be the case in a that is over - fed .
the can not become very super- , and the source will regulate the to @xmath100 or @xmath101 . since @xmath102 , @xmath103 constant implies @xmath104 and hence @xmath105 .
in this case we assume the matter supply comes from a homogeneous distribution ( gas or stars ) , due to the gravitational potential of the . on large scales , the accretion will be spherical , and the accretion radius at which the gravitational potential becomes significant is given by @xmath106 where @xmath107km / s is the stellar velocity dispersion .
the is given by the bondi formula @xmath108 where @xmath109 is the stellar number density .
in this case @xmath110 and @xmath111 , that is , the ratio increases with time .
eventually the ratio will approach unity and the accretion will become limited . a similar expression can be derived for a non - homogeneous medium , with a density profile @xmath112 . assuming the velocity dispersion is independent of @xmath11 , eqs .
[ equ : rac ] and [ equ : sar ] give for the in that case @xmath113 if also the velocity dispersion has a radial dependence , @xmath114 then we get a more complicated dependence , @xmath115 in this case , any functional dependence is possible . if for example @xmath116 and @xmath117 , @xmath118 , and if @xmath119 ( point mass ) and @xmath117 , @xmath120const .
in the previous section we have found the relation between the accretion parameters @xmath16 and @xmath50 , and found the functional dependence of the continuum luminosity on the evolving mass for several accretion scenarios .
this enables us to draw evolutionary curves in the @xmath121 plane ( fig .
1 ) . for example , an bound will evolve along the constant @xmath64 curves ( dashed curves in fig .
1 ) . a homogeneously accreting will evolve along curves of @xmath111 , which correspond to nearly horizontal ( from left to right ) lines in fig . 1 , and a constant will yield @xmath122 which are almost vertical ( upward ) lines in fig . 1 .
llll & & & + accretion scenario : & constant @xmath50 & constant @xmath123 & spherical + @xmath124&const&@xmath125&@xmath126 + + @xmath73 model : & & & + & & & + @xmath127&@xmath128 & @xmath129&const + & & & + @xmath130&@xmath129 & @xmath129&@xmath129 + & & & + @xmath131 or @xmath8(bb / rad)&const & @xmath129&@xmath128 + & & & + & & & + @xmath8(bb - var)@xmath132&const & @xmath129&@xmath128 + & & & + combining this with the dependence of the effective radiation - temperature on the accretion parameters , and using the result that the luminosity is positively correlated with the radiation temperature , we can find the relation between the equivalent width and the continuum luminosity for each pair of accretion scenario and temperature model .
the dependence of the effective temperature on the mass is summarized in table 2 .
the header shows the three accretion scenarios and the corresponding dependence of the continuum luminosity on the mass .
the next row gives the dependence of the luminosity on the mass for each scenario .
the left column shows the temperature models . for each pair
the table gives the functional dependence of @xmath73 on m. applying the dependence of l on m it is possible to deduce the dependence of t on l. combining the -radiation boundary and the photospheric models , which have a very similar dependence on the accretion parameters , the table has nine ( @xmath133 ) pairs , for models , and an additional row for the model independent - variability estimate of @xmath73 ( marked `` @xmath8(bb - var ) '' ) .
for all pairs except two ( constant @xmath50 with @xmath134 and homogeneous medium accretion with @xmath127 ) . the effective disk temperature is decreasing with increasing mass .
for all accretion scenarios the optical and uv continuum luminosity is increasing with @xmath16 .
this is certainly the case for bounded accretion and for accretion from a homogeneous medium ; for a constant the bolometric luminosity is constant , but as the mass increases the peak of the spectrum moves to lower frequencies .
since the optical or near uv continuum are on the rayleigh - jeans part of the spectrum , the luminosity in these bands increases as the mass increases even if the total luminosity remains constant .
we note that this is true also for the general temperature - luminosity relation ( eqs .
[ equ : elt ] and [ equ : el ] ) , so the model is not essential . in that case
the anti - correlation between temperature and luminosity is built into the the expression for the temperature , eq .
[ equ : el ] .
* decreasing . * in the cases we looked at the increases with time or remains constant .
what if the accretion rate decreases with time ?
in the two latter -temperature models ( 5.2.c and d ) we find @xmath135 , so @xmath73 increases , and so does the line luminosity . if we take @xmath136 the continuum luminosity decreases , and there is still a negative correlation .
this is the case also in the general l - t relation .
for the first two models ( 5.2.a , b ) the situation is less clear , because @xmath73 in these two models will decrease ( as @xmath137 and @xmath129 , respectively ) and also @xmath99 decreases .
the question whether @xmath73 is correlated or anti correlated with @xmath7 would depend on the details of the behavior of the . how does the vary ?
the is defined as @xmath138 , where @xmath139 is the flux per unit wavelength at the wavelength of the line .
we have argued that @xmath140 depends on the ionizing continuum , which correlates strongly and steeply with @xmath73 .
the luminosity in the line is also roughly linearly dependent on the continuum luminosity , @xmath141 , so that @xmath142 which gives for the @xmath143 the last approximation should not be taken as a linear relation , but rather as a strong correlation .
as shown above , for @xmath144 few ryd , @xmath145 is steeply increasing ( and hence strongly correlated ) with @xmath8 , and hence @xmath7(line ) is strongly correlated with the temperature .
we conclude that for many accretion - scenarios and temperature - models ( accretion - disk models or general effective related to variability ) the line luminosity decreases with increasing mass , except the two for which the effective temperature is constant , there is an evolutionary : as the evolves over cosmological times due to the mass accreted , the continuum luminosity ( optical or near uv ) increases , while the decreases .
even the two pairs with a constant effective temperature will eventually show a . to see this
, note that for the ( @xmath146(spherical ) ) pair , @xmath64 is increasing as @xmath16 , so it will eventually approach the , and move from the @xmath127 model to one of the other @xmath73 models , which ( for @xmath50(spherical ) ) do have a . in the other pair , ( constant @xmath50 with @xmath134 ) , the situation is opposite , but with the same result : since there @xmath122 , eventually the will become enough sun - so that the @xmath127 model will apply , which for the constant scenario does have a .
we suggest that the growth of the central mass due to the accreted matter causes the luminosity and the spectrum to evolve over over the agn life time . using general model - independent relations , or the thin spectrum
, we estimate the evolution of the and the continuum spectrum , and show that for plausible evolutionary tracks as well as for the model - independent temperature estimate and for most variants of the thin model the decreases with increasing continuum luminosity implying an evolutionary origin to the baldwin effect .
baldwin , j.a .
1977 , , 214 , 679 .
czerny , b , witt , h.j . and zycki , p.t . 1996 , astro - ph/9609180 .
haardt , f. , maraschi , l. and ghisellini , g. 1994 , , 432 , l95 laor , a. 1990 , , 246 , 369 .
mathews , w.g . and
ferland , g.j .
1987 , , 323 , 456 .
netzer , h. 1990 in `` active galactic nuclei '' , saas - fee advanced course 20 , eds .
courvoisier and m. major , springer verlag , berlin , p. 57 .
peterson , b.m.an introduction to active galactic nuclei cambridge university press , 1997 .
rees , m. , netzer h. , ferland , g.j .
1989 . , 347 , 640 .
sun , w.h . and
malkan , m.a .
1991 , wandel , a. 1997 , 430 , 131 .
wandel , a. and boller , th .
1998 , , 331 , 884 .
wandel , a. and liang , 1991 , 380 , 84 .
wandel , a. and mushotzky , r.f .
1986 , , 306 , l61 .
wandel , a. and petrosian , v. 1988 , , 329 , l11 .
wandel , a. and yahil , a. 1985 , , 295 , l1 .
zheng , w. , kriss , g.a . , telfer , r.c . , grimes , j.p . , and davidsen , a.f .
1997 , , 475 , 469 . | assuming active galactic nuclei are powered by accretion onto a massive we suggest that the growth of the central mass due to the matter accreted over the agn lifetime causes evolution of the luminosity and spectrum .
we show that the effective temperature of the uv continuum spectrum is likely to be anti - correlated with the mass and with the luminosity .
we estimate the change in the of the emission lines due to the growth of the and show that for plausible evolutionary tracks and effective temperature models the is anti - correlated with continuum luminosity thus implying an evolutionary origin to the baldwin effect .
# 1#2#3#4 = # 1 cm # 2 cm .3 cm |
due to the rich variety of their physical properties , 2d materials are attracting much attention .
huge thermal conductivity values ( @xmath1 ) in graphene @xcite and in boron nitride @xcite , various electronic properties in transition metal dichalcogenides @xcite , high - t@xmath2 superconductivity in cuprates @xcite are among some of the specific properties of interest .
about two - thirds of the mx@xmath0 transition metal dichalcogenides @xcite crystallize in layered structures and they exhibit a wide range of tunable electronic states ( semiconducting , insulating , semi - metallic and superconducting ) , as well as exotic optical and mechanical properties .
their crystals can be easily cleaved as successive x layers are only weakly bonded by van der waals forces .
the well - known sulfides derived from tis@xmath0 have recently been under active investigation in the new context of thermoelectric materials research @xcite .
while the thermoelectric power factor , @xmath3 , ( @xmath4 and @xmath5 are the electrical resistivity and seebeck coefficient , respectively ) of tis@xmath0 single crystals is similar to that of bi@xmath6te@xmath7 , the thermoelectric material most commonly used in applications , the thermal conductivity ( @xmath8 ) exhibits relatively high values@xcite : @xmath9 , compared to @xmath10 for bi@xmath6te@xmath7 @xcite .
this reduces the dimensionless figure of merit , @xmath11 , which measures the theoretical efficiency of energy conversion .
several attempts have been made to reduce @xmath8 in dense ceramics , for example via intercalation in the van der waals gap as in cu@xmath12tis@xmath0 , or substitution at either anion or cation sites as in tis@xmath13se@xmath14 or ta@xmath15ti@xmath16s@xmath0 @xcite .
these investigations have lead to zt values of 0.4 at 650k .
however , for all these ceramic series , it is difficult to distinguish whether changes in @xmath8 are a result of changes in the charge carrier concentration , the concentration of substituted or intercalated ions , or grain boundary effects .
moreover , the chemical formula should more properly be written as non - stoichiometric ti@xmath17s@xmath0 .
it is known to be a degenerate extrinsic semiconductor with extra ti@xmath18 cations in the van der waals gap , the donated electrons of which populate the empty ti d - band of the tis@xmath0 layer @xcite . meanwhile
, the crystallographic texturation of ceramic samples implies that the transport properties always result from a convolution of in - plane and out - of - plane contributions , not to mention scattering by grain boundaries . in this article
, we report on in - plane electronic and thermal transport for several metallic single crystals derived from tis@xmath0 .
the changes in @xmath8 ( in - plane ) are dominated by the lattice part with a peak at @xmath19/5 ( @xmath19=debye temperature ) expected for phonon - dominated heat transport .
we show that reduction in the lattice heat transport ( @xmath20 ) deriving from the substitution or intercalation of foreign cations accounts for the majority of the change in @xmath8 , and we identify the species that are most effective at reducing the thermal conductivity .
the @xmath20(300k ) values obtained for crystals derived from tis@xmath0 can be reduced by a factor of around five , from @xmath21 ( for the most stoichiometric crystals @xcite ) to @xmath10 ( this study ) .
we conclude that the thermoelectric properties of the 2d dichalcogenides could be improved by thermal conductivity engineering , even before nanostructuring is pursued .
large platelet - like crystals were grown by a vapour transport technique using iodine as the transport agent @xcite .
intercalated and chemically substituted single crystals were grown by pre - reacting co and ta with tis@xmath0 precursor at 500@xmath22c prior to the growth phase .
shiny thin crystals were obtained for all the different batches and powder x - ray diffraction confirmed the presence of the tis@xmath0 phase . scanning electron microscopy with
coupled edx chemical analysis on all samples confirmed the nominal compositions within an uncertainty of @xmath23% . in the case of co@xmath24tis@xmath0
, this means that co was detected but with greater error than the nominal composition .
one ceramic sample of tis@xmath0 grown with a slight nominal excess of ti and densified by spark plasma sintering was also used for comparison to the single crystals .
full details of this sample are in ref . .
single crystals and derivatives , as well as the ceramic sample .
the resistivity of the densified ceramic tis@xmath0 is much higher than the single crystal despite very similar charge carrier concentration .
dashed lines show the fits to @xmath25 . *
b ) * seebeck coefficient of the same samples.,scaledwidth=47.0% ] transport measurements were made on single crystal platelets of typical dimension @xmath26 mm .
electrical contacts were made using dupont 4929 silver paste .
electrical resistivity and hall effect was measured using a standard six probe technique .
thermal conductivity and seebeck coefficient were measured on single crystals using a custom built steady - state experiment installed in a physical properties measurement system cryostat .
calibrated heat pipes were used to measure the thermal power entering and leaving the sample @xcite . power loss ( principally from radiation )
did not exceed 30% at 300k .
this contributes to an additional uncertainty in the thermal conductivity measurements of approximately 10% at 300k , which becomes negligible below 150k .
one crystal of tis@xmath0 was also measured using the harman technique @xcite and the results are in good agreement .
electrical resistivity of the samples is shown in fig . [ fig : tep]a .
they show metallic - like behavior with @xmath27 in between 0.6 and 1@xmath28 and resistivity ratio @xmath29 related to residual resistivity values at 5k of 0.2 to 0.3@xmath28 .
the temperature dependence of the curves follows the previously seen @xmath30 with @xmath31 behaviour over a wide temperature range ( 60 - 300k ) that has been ascribed to the approximately equal contributions of inter- and intra - valley scattering to the relaxation time @xcite .
values of @xmath32 and @xmath33 are shown in table [ tab : params ] and the fits are included in fig .
[ fig : tep]a . a key problem with tis@xmath0 and derivatives is controlling the charge carrier concentration , which arises both from deliberate substitution or intercalation , and from the tendency for excess ti@xmath18 to self - intercalate in the van der waals gap . in the following ,
we establish the concentration of substituted and/or intercalated cations , @xmath34 , by reference to well - characterised ceramic series using the c - axis lattice parameters @xcite , and obtain the corresponding charge carrier concentration , @xmath35 , from hall effect measurements . in the case of the tis@xmath0 single crystal , @xmath36 gives@xcite @xmath37 and @xmath38 .
these values are consistent with a stoichiometry close to ti@xmath39s@xmath0 , assuming that all of the excess ti@xmath18 is intercalated and that all four electrons become free carriers .
the ceramic sample of the same nominal stoichiometry presents nearly identical seebeck coefficient , which we consider to be a more reliable measure of the doping level in ceramic materials than @xmath35 , as was shown in the cuprate superconductors @xcite . the value of @xmath40 for the single crystals of ref
. corresponds to between 1 and 1.5% excess titanitum according to the same ceramic series@xcite .
the co@xmath24tis@xmath0 crystal has @xmath41 and a correspondingly larger @xmath42 .
previous studies of co intercalation at high concentrations were consistent with the co@xmath43 oxidation state @xcite .
@xmath44 implies a stoichiometry close to the nominal co@xmath24 composition@xcite , and shows the contraction of the lattice characteristic of the strong covalent nature of bonding between the co@xmath43 ion and the sulphur atoms in adjacent layers@xcite .
by contrast , the presence of intercalated ti@xmath18 tends to enlarge the van der waals gap .
the measured @xmath35 therefore corresponds to @xmath45 and a stoichiometry co@xmath46 ; this is certainly within the margin of error afforded by the lattice parameter measurements .
there is no suggestion of significant ti self - intercalation in this case , and we speculate that the contraction of the lattice leads to a much lower ability of ti@xmath18 to enter van der waals gap .
the highly doped ta@xmath47ti@xmath48s@xmath0 has a much smaller @xmath49 and a correspondingly larger @xmath50 .
ta substitution at the ti site provides one free carrier per ion , as can be seen from the linear dependence of @xmath35 on doping in ref . .
the value of @xmath51 is close to that obtained for a ceramic sample of the same nominal ta@xmath47 composition @xcite , as is @xmath52 . given this composition , only @xmath53 charge carriers are expected .
we conclude that the rest of the carriers arise from excess ti intercalation , such that the composition is approximately ta@xmath47ti@xmath54s@xmath0 , giving @xmath55 and @xmath56 .
the nominal and derived chemical formulae for all of the single crystals are summarised in table [ tab : params ] . .
the red circles and yellow diamonds represent the steady state and harman measurements on nominally stoichimetric tis@xmath0 respectively ( see text ) . the electronic contribution estimated from the wiedemann - franz law ( solid lines )
is subtracted to give the lattice thermal conductivity * b)*. the peak height of @xmath20 at @xmath57k is reduced as impurity scattering increases .
the fits to the debye - callaway model ( black solid lines ) are dominated by the variation in this parameter .
the estimated minimum lattice thermal conductivity is also shown.,scaledwidth=47.0% ] the in - plane thermal conductivity ( fig .
[ fig : kappa]a ) , particularly the lattice contribution ( fig .
[ fig : kappa]b ) , is a powerful probe of @xmath34 , especially in the absence of grain boundaries .
the only previously reported thermal conductivity value for a slightly self - intercalated tis@xmath0 crystal@xcite was @xmath9 , with @xmath58 corresponding to @xmath59 . for the latter ,
the electronic part of @xmath8 was only @xmath60 against @xmath61 for the present tis@xmath0 and co@xmath24tis@xmath0 crystals ( assuming that the wiedemann - franz law is valid in this regime for electronic heat transport ) .
generally speaking , the thermal conductivity curves follow the typical form expected for heat transported mainly by phonons : a large quasi - linear rise at low temperature as phonons are excited , a peak limited by cation and defect concentration , followed by a continuing decrease at higher temperatures as phonon - phonon umklapp scattering processes become dominant .
we extract the lattice thermal conductivity ( fig .
[ fig : kappa]b ) by subtracting the electrical contribution to the total thermal conductivity according to the wiedemann - franz law , @xmath62 , where we use the sommerfeld value of the lorenz number @xmath63@xmath64 .
@xmath65 extracted in this way is usually an upper bound since a ) inelastic scattering processes , maximally active at @xmath66 , scatter the thermal but not the electrical current effectively and b ) if the temperature broadening of the fermi distribution approaches the bandwidth , scattering processes are no longer quasi - elastic .
for the high carrier concentrations in the samples measured here , the fermi temperature is in excess of 1000k and our measurements remain in the quasi - elastic regime @xcite .
we analyse @xmath20 using the debye - callaway model with contributions from boundary , impurity , umklapp and planar defect scattering : @xmath67 where @xmath68 and @xmath69 .
@xmath70@xmath71 is the in - plane sound velocity @xcite and @xmath72 mm is the relevant limit for boundary scattering , corresponding to the average sample dimension .
@xmath73 is the prefactor for scattering by point - like defects ( lattice imperfections or foreign cations ) , @xmath74 and @xmath75 are the parameters describing the umklapp processes , and @xmath76 is related to the density of planar defects .
the inclusion of @xmath76 significantly improves the quality of the fits at low temperature .
the impurity , defect and boundary scattering control the shape of the curve at low temperatures , while umklapp scattering controls the decline at high temperatures .
@xmath77k is used for all samples@xcite . in the case of the ceramic sample ,
we use @xmath78 nm in order to compare to the results of ref . .
[ cols="^,^,^,^,^,^,^,^,^,^,^ " , ] the ceramic sample shows a smaller @xmath73 than the single crystal of tis@xmath0 , but also a much lower @xmath20 .
the primary reason for the difference between these two cases is the parameter @xmath79 , which describes scattering by grain boundaries . if we take the identical parameters as for the single crystal , a value of @xmath80 nm produces a very similar @xmath20 to the one shown in fig .
[ fig : kappa]b , showing that this is indeed the dominant scattering mechanism .
the reduction in thermal conductivity on going from crystal to ceramic is almost perfectly balanced by the increase in resistivity , however , leading to no change in @xmath81 at 300k .
in conclusion , we have shown using well calibrated concentrations of substituted or intercalated ions in single crystal materials that the most effective species at reducing the thermal conductivity in the tis@xmath0 system is ta@xmath18 substitution at the ti site .
ta substitution is particularly suitable as it provides only 1 electron per ion , compared to 2 and 4 for co and ti intercalants , introducing more scattering centers for a given charge carrier concentration .
the thermoelectric performance of ta doped materials so far studied has however been limited by parasitic self - intercalation of ti .
improvements in thermoelectric performance will only be realised once this effect can be controlled
. then studies of ceramic tis@xmath0 and derivatives for potential applications can be pursued , as similar performance can be obtained even in the presence of grain boundaries .
klipstein , a.g .
baguall , w.y .
liang , e.a .
marseglia , r.h .
friend , j. phys .
c : solid state phys . ,
* 14 * 4067 ( 1981 ) .
k. kawano , j. takahashi , r. nakata and m. sumita , j. phys .
soc . jpn . *
58 * 949 ( 1989 ) .
m. beaumale , t. barbier , y. brard , g. guelou , a.v .
powell , p. vaqueiro , e. guilmeau , acta mater .
, * 78 * 86 , ( 2014 ) .
allen , x. du , l. mihaly and l. forro , phys .
b * 49 * 9073 ( 1994 ) .
w. kobayashi , w. tamura , and i. terasaki , j. phys .
jpn * 77 * 074606 ( 2008 ) .
m. inoue , h.p .
hughes and a.d .
yoffe , advances in physics , * 38 * 565 ( 1989 ) .
obertelli , j.r .
cooper and j. l. tallon , phys .
b * 46 * 14928 ( 1992 ) .
m. danot , j. rouxel and o. gorochov , mat .
* 9 * 1383 ( 1974 ) .
t. kawasaki and k. ohshima , j. phys .
jpn * 80 * 044601 ( 2011 ) . c. wan , y. wang , n. wang , w. norimatsu , m. kusunoki and k. koumoto , sci .
technol . adv .
mater . * 11 * 044306 ( 2010 ) .
m. inoue , y. muneta , h. negishi , and m. sasaki , j. low temp .
phys . * 63 * , 235 ( 1986 ) . thermal conductivity ed .
tritt , kluwer academic/ plenum publishers , new york ( 2004 ) . | * the promising thermoelectric material tis@xmath0 can be easily chemically doped and intercalated .
we present here studies of single crystals that are intercalated with excess ti or co , or substituted with ta .
we demonstrate the intrinsic impact of these dopants on the thermal transport in the absence of grain boundary scattering .
we show that ta doping has the greatest impact on the thermal scattering rate per ion added , leading to a five - fold reduction in the lattice thermal conductivity as compared to stoichiometric single crystals . * |
the problem of reconstructing a signal @xmath0 from non - uniformly spaced measurements @xmath1 arises in areas as diverse as geophysics , medical imaging , communication engineering , and astronomy .
a successful reconstruction of @xmath0 from its samples @xmath1 requires a priori information about the signal , otherwise the reconstruction problem is ill - posed .
this a priori information can often be obtained from physical properties of the process generating the signal . in many of the aforementioned applications
the signal can be assumed to be ( essentially ) band - limited .
recall that a signal ( function ) is band - limited with bandwidth @xmath2 if it belongs to the space @xmath3 , given by @xmath4 where @xmath5 is the fourier transform of @xmath0 defined by @xmath6 for convenience and without loss of generality we restrict our attention to the case @xmath7 , since any other bandwidth can be reduced to this case by a simple dilation .
therefore we will henceforth use the symbol @xmath8 for the space of band - limited signals .
it is now more than 50 years ago that shannon published his celebrated sampling theorem @xcite .
his theorem implies that any signal @xmath9 can be reconstructed from its regularly spaced samples @xmath10 by @xmath11 in practice however we seldom enjoy the luxury of equally spaced samples .
the solution of the nonuniform sampling problem poses much more difficulties , the crucial questions being : * under which conditions is a signal @xmath9 uniquely defined by its samples @xmath12 ? * how can @xmath0 be stably reconstructed from its samples @xmath1 ?
these questions have led to a vast literature on nonuniform sampling theory with deep mathematical contributions see @xcite to mention only a few .
there is also no lack of methods claiming to efficiently reconstruct a function from its samples @xcite .
these numerical methods naturally have to operate in a finite - dimensional model , whereas theoretical results are usually derived for the infinite - dimensional space @xmath8 . from a numerical point of view
the `` reconstruction '' of a bandlimited signal @xmath0 from a finite number of samples @xmath13 amounts to computing an approximation to @xmath0 ( or @xmath5 ) at sufficiently dense ( regularly ) spaced grid points in an interval @xmath14 . hence in order to obtain a `` complete '' solution of the sampling problem following questions have to be answered : * does the approximation computed within the finite - dimensional model actually converge to the original signal @xmath0 , when the dimension of the model approaches infinity ? * does the finite - dimensional model give rise to fast and stable numerical algorithms ? these are the questions that we have in mind , when presenting an overview on recent advances and new results in the nonuniform sampling problem from a numerical analysis view point .
in section [ ss : truncated ] it is demonstrated that the celebrated frame approach does only lead to fast and stable numerical methods when the finite - dimensional model is carefully designed .
the approach usually proposed in the literature leads to an ill - posed problem even in very simple situations .
we discuss several methods to stabilize the reconstruction algorithm in this case . in section [ ss : trigpol ]
we derive an alternative finite - dimensional model , based on trigonometric polynomials .
this approach leads to a well - posed problem that preserves important structural properties of the original infinite - dimensional problem and gives rise to efficient numerical algorithms .
section [ s : numeric ] describes how this approach can be modified in order to reconstruct band - limited signals for the in practice very important case when the bandwidth of the signal is not known .
furthermore we present regularization techniques for ill - conditioned sampling problems .
finally section [ s : applications ] contains numerical experiments from spectroscopy and geophysics .
before we proceed we introduce some notation that will be used throughout the paper .
if not otherwise mentioned @xmath15 always denotes the @xmath16-norm ( @xmath17-norm ) of a function ( vector ) . for operators ( matrices )
@xmath18 is the standard operator ( matrix ) norm .
the condition number of an invertible operator @xmath19 is defined by @xmath20 and the spectrum of @xmath19 is @xmath21 .
@xmath22 denotes the identity operator .
the concept of frames is an excellent tool to study nonuniform sampling problems @xcite .
the frame approach has the advantage that it gives rise to deep theoretical results and also to the construction of efficient numerical algorithms _ if _ ( and this point is often ignored in the literature ) the finite - dimensional model is properly designed .
following duffin and schaeffer @xcite , a family @xmath23 in a separable hilbert space @xmath24 is said to be a frame for @xmath24 , if there exist constants ( the _ frame bounds _ ) @xmath25 such that @xmath26 we define the _ analysis operator _ @xmath19 by @xmath27 and the _ synthesis operator _ , which is just the adjoint operator of @xmath19 , by @xmath28 the _ frame operator _
@xmath29 is defined by @xmath30 , hence @xmath31 .
@xmath29 is bounded by @xmath32 and hence invertible on @xmath24 .
we will also make use of the operator @xmath33 in form of its gram matrix representation @xmath34 with entries @xmath35 .
on @xmath36 the matrix @xmath37 is bounded by @xmath38 and invertible .
on @xmath17 this inverse extends to the _ moore - penrose inverse _ or pseudo - inverse @xmath39 ( cf .
@xcite ) . given a frame @xmath23 for @xmath24 ,
any @xmath40 can be expressed as @xmath41 where the elements @xmath42 form the so - called dual frame and the frame operator induced by @xmath43 coincides with @xmath44 .
hence if a set @xmath23 establishes a frame for @xmath24 , we can reconstruct any function @xmath40 from its moments @xmath45 .
one possibility to connect sampling theory to frame theory is by means of the _ sinc_-function @xmath46 its translates give rise to a _ reproducing kernel _ for @xmath8 via @xmath47 combining with formulas and we obtain following well - known result @xcite .
if the set @xmath48 is a frame for @xmath8 , then the function @xmath9 is uniquely defined by the sampling set @xmath12 . in this case
we can recover @xmath0 from its samples by @xmath49 or equivalently by @xmath50 with @xmath37 being the frame gram matrix with entries @xmath51 and @xmath52 .
the challenge is now to find easy - to - verify conditions for the sampling points @xmath53 such that @xmath48 ( or equivalently the exponential system @xmath54 ) is a frame for @xmath8 .
this is a well - traversed area ( at least for one - dimensional signals ) , and the reader should consult @xcite for further details and references .
if not otherwise mentioned from now on we will assume that @xmath48 is a frame for @xmath8 . of course
, neither of the formulas and can be actually implemented on a computer , because both involve the solution of an infinite - dimensional operator equation , whereas in practice we can only compute a finite - dimensional approximation .
although the design of a valid finite - dimensional model poses severe mathematical challenges , this step is often neglected in theoretical but also in numerical treatments of the nonuniform sampling problem .
we will see in the sequel that the way we design our finite - dimensional model is crucial for the stability and efficiency of the resulting numerical reconstruction algorithms . in the next two sections we describe two different approaches for obtaining finite - dimensional approximations to the formulas and .
the first and more traditional approach , discussed in section [ ss : truncated ] , applies a finite section method to equation .
this approach leads to an ill - posed problem involving the solution of a large unstructured linear system of equations .
the second approach , outlined in section [ ss : trigpol ] , constructs a finite model for the operator equation in by means of trigonometric polynomials .
this technique leads to a well - posed problem that is tied to efficient numerical algorithms .
according to equation we can reconstruct @xmath0 from its sampling values @xmath1 via @xmath55 , where @xmath56 with @xmath57 . 0 since @xmath37 is a compact operator it can be diagonalized via its singular system ( or eigensystem since @xmath37 is self - adjoint ) @xmath58 as follows @xcite @xmath59 with a corresponding complete orthogonal set of vectors @xmath60 .
the moore - penrose inverse @xmath61 can be expressed as @xmath62 where , as usual , only the non - zero singular values of @xmath37 are used in the above sum . in order
to compute a finite - dimensional approximation to @xmath63 we use the finite section method @xcite . for @xmath64 and @xmath65
we define the orthogonal projection @xmath66 by @xmath67 and identify the image of @xmath66 with the space @xmath68 . setting @xmath69 and @xmath70 , we obtain the @xmath71-th approximation @xmath72 to @xmath73 by solving @xmath74 it is clear that using the truncated frame @xmath75 in for an approximate reconstruction of @xmath0 leads to the same system of equations . if @xmath48 is an exact frame
( i.e. , a riesz basis ) for @xmath8 then we have following well - known result .
let @xmath48 be an exact frame for @xmath8 with frame bounds @xmath76 and @xmath77 and @xmath78 as defined above .
then @xmath79 converges strongly to @xmath80 and hence @xmath81 for @xmath82 .
since the proof of this result given in @xcite is somewhat lengthy we include a rather short proof here .
note that @xmath37 is invertible on @xmath17 and @xmath83 .
let @xmath84 with @xmath85 , then @xmath86 .
in the same way we get @xmath87 , hence the matrices @xmath88 are invertible and uniformly bounded by @xmath89 and @xmath90 the lemma of kantorovich @xcite yields that @xmath91 strongly .
if @xmath48 is a non - exact frame for @xmath8 the situation is more delicate .
let us consider following situation .
* example 1 : * let @xmath9 and let the sampling points be given by @xmath92 , i.e. , the signal is regularly oversampled at @xmath93 times the nyquist rate
. in this case the reconstruction of @xmath0 is trivial , since the set @xmath94 is a tight frame with frame bounds @xmath95 .
shannon s sampling theorem implies that @xmath0 can be expressed as @xmath96 where @xmath97 and the numerical approximation is obtained by truncating the summation , i.e. , @xmath98 using the truncated frame approach one finds that @xmath37 is a toeplitz matrix with entries @xmath99 in other words , @xmath88 coincides with the prolate matrix @xcite .
the unpleasant numerical properties of the prolate matrix are well - documented .
in particular we know that the singular values @xmath100 of @xmath88 cluster around @xmath101 and @xmath102 with @xmath103 singular values in the transition region .
since the singular values of @xmath88 decay exponentially to zero the finite - dimensional reconstruction problem has become _ severely ill - posed _
@xcite , although the infinite - dimensional problem is `` perfectly posed '' since the frame operator satisfies @xmath104 , where @xmath22 is the identity operator .
of course the situation does not improve when we consider non - uniformly spaced samples . in this case
it follows from standard linear algebra that @xmath105\}$ ] , or expressed in words , the singular values of @xmath37 are bounded away from zero .
however for the truncated matrices @xmath88 we have @xmath106\}\ ] ] and the smallest of the singular values of @xmath88 will go to zero for @xmath82 , see @xcite .
let @xmath107 be the singular value decomposition of a matrix @xmath108 with @xmath109 .
then the moore - penrose inverse of @xmath108 is @xmath110 , where ( e.g. , see @xcite ) @xmath111 for @xmath112 this means that the singular values close to zero will give rise to extremely large coefficients in @xmath113 .
in fact @xmath114 for @xmath82 and consequently @xmath72 does not converge to @xmath73 .
practically @xmath115 is always bounded due to finite precision arithmetics , but it is clear that it will lead to meaningless results for large @xmath71 .
if the sampling values are perturbed due to round - off error or data error , then those error components which correspond to small singular values @xmath116 are amplified by the ( then large ) factors @xmath117 .
although for a given @xmath88 these amplifications are theoretically bounded , they may be practically unacceptable large .
such phenomena are well - known in regularization theory @xcite .
a standard technique to compute a stable solution for an ill - conditioned system is to use a truncated singular value decomposition ( tsvd ) @xcite .
this means in our case we compute a regularized pseudo - inverse @xmath118 where @xmath119 in @xcite it is shown that for each @xmath71 we can choose an appropriate truncation level @xmath120 such that the regularized inverses @xmath121 converge strongly to @xmath122 for @xmath82 and consequently @xmath123 , where @xmath124 with @xmath125 the optimal truncation level @xmath120 depends on the dimension @xmath71 , the sampling geometry , and the noise level .
thus it is not known a priori and has in principle to be determined for each @xmath71 independently .
since @xmath120 is of vital importance for the quality of the reconstruction , but no theoretical explanations for the choice of @xmath120 are given in the sampling literature , we briefly discuss this issue . for this purpose
we need some results from regularization theory . 0 in realistic situations the samples
are usually perturbed by noise , with @xmath126 where @xmath127 denotes a perturbed sample . the noise level
@xmath128 is in practice in general much larger than the truncation level for noise - free data .
thus @xmath120 will mainly depend on @xmath128 . for noise - free data
the accuracy is determined by the roundoff level via the machine precision of the computer .
let @xmath129 be given where @xmath108 is ill - conditioned or singular and @xmath130 is a perturbed right - hand side with @xmath131 . since in our sampling problem
the matrix under consideration is symmetric , we assume for convenience that @xmath108 is symmetric . from a numerical point of view
ill - conditioned systems behave like singular systems and additional information is needed to obtain a satisfactory solution to @xmath132 .
this information is usually stated in terms of `` smoothness '' of the solution @xmath133 .
a standard approach to qualitatively describe smoothness of @xmath133 is to require that @xmath133 can be represented in the form @xmath134 with some vector @xmath135 of reasonable norm , and a `` smoothing '' matrix @xmath29 , cf .
often it is useful to construct @xmath29 directly from @xmath108 by setting @xmath136 usually , @xmath137 is assumed to be fixed , typically at @xmath138 or @xmath139 .
we compute a regularized solution to @xmath140 via a truncated svd and want to determine the optimal regularization parameter ( i.e. , truncation level ) @xmath141 . under the assumption that @xmath142 it follows from theorem 4.1 in @xcite that the optimal regularization parameter @xmath141 for the tsvd is @xmath143 where @xmath144 ( see section 6 in @xcite ) .
however @xmath135 and @xmath145 are in general not known . using @xmath146 and @xmath147
we obtain @xmath148 .
furthermore , setting @xmath149 implies @xmath150 hence combining and we get @xmath151 applying these results to solving @xmath78 via tsvd as described in the previous section , we get @xmath152 where @xmath153 is the upper frame bound .
fortunately estimates for the upper frame bound are much easier to obtain than estimates for the lower frame bound . thus using the standard setting @xmath138 or @xmath139
a good choice for the regularization parameter @xmath141 is @xmath154\ , .
\label{thresh}\ ] ] extensive numerical simulations confirm this choice , see also section [ s : applications ] .
for instance for the reconstruction problem of example 1 with noise - free data and machine precision @xmath155 , formula implies @xmath156 $ ] .
this coincides very well with numerical experiments .
if the noise level @xmath128 is not known , it has to be estimated .
this difficult problem will not be discussed here .
the reader is referred to @xcite for more details .
although we have arrived now at an implementable algorithm for the nonuniform sampling problem , the disadvantages of the approach described in the previous section are obvious . in general
the matrix @xmath88 does not have any particular structure , thus the computational costs for the singular value decomposition are @xmath157 which is prohibitive large in many applications .
it is definitely not a good approach to transform a well - posed infinite - dimensional problem into an ill - posed finite - dimensional problem for which a stable solution can only be computed by using a `` heavy regularization machinery '' .
the methods in @xcite coincide with or are essentially equivalent to the truncated frame approach , therefore they suffer from the same instability problems and the same numerical inefficiency . as mentioned above one way to stabilize
the solution of @xmath78 is a truncated singular value decomposition , where the truncation level serves as regularization parameter . for large @xmath71
the costs of the singular value decomposition become prohibitive for practical purposes .
we propose the conjugate gradient method @xcite to solve @xmath78 .
it is in general much more efficient than a tsvd ( or tikhonov regularization as suggested in @xcite ) , and at the same time it can also be used as a regularization method .
the standard error analysis for cg can not be used in our case , since the matrix is ill - conditioned .
rather we have to resort to the error analysis developed in @xcite .
when solving a linear system @xmath132 by cg for noisy data @xmath130 following happens .
the iterates @xmath158 of cg may diverge for @xmath159 , however the error propagation remains limited in the beginning of the iteration .
the quality of the approximation therefore depends on how many iterative steps can be performed until the iterates turn to diverge .
the idea is now to stop the iteration at about the point where divergence sets in . in other words
the iterations count is the regularization parameter which remains to be controlled by an appropriate stopping rule @xcite .
in our case assume @xmath160 , where @xmath161 denotes a noisy sample .
we terminate the cg iterations when the iterates @xmath162 satisfy for the first time @xcite @xmath163 for some fixed @xmath164 .
it should be noted that one can construct `` academic '' examples where this stopping rule does not prevent cg from diverging , see @xcite , `` most of the time '' however it gives satisfactory results .
we refer the reader to @xcite for a detailed discussion of various stopping criteria .
there is a variety of reasons , besides the ones we have already mentioned , that make the conjugate gradient method and the nonuniform sampling problem a `` perfect couple '' .
see sections [ ss : trigpol ] , [ ss : ml ] , and [ ss : regul ] for more details . by combining the truncated frame approach with the conjugate gradient method ( with appropriate stopping rule ) we finally arrive at a reconstruction method that is of some practical relevance .
however the only existing method at the moment that can handle large scale reconstruction problems seems to be the one proposed in the next section .
in the previous section we have seen that the naive finite - dimensional approach via truncated frames is not satisfactory , it already leads to severe stability problems in the ideal case of regular oversampling . in this section
we propose a different finite - dimensional model , which resembles much better the structural properties of the sampling problem , as can be seen below .
the idea is simple . in practice only a finite number of samples @xmath13
is given , where without loss of generality we assume @xmath165 ( otherwise we can always re - normalize the data ) . since no data of @xmath0 are available from outside this region we focus on a local approximation of @xmath0 on @xmath166 $ ] .
we extend the sampling set periodically across the boundaries , and identify this interval with the ( properly normalized ) torus @xmath167 . to avoid technical problems at the boundaries in the sequel
we will choose the interval somewhat larger and consider either @xmath168 $ ] or @xmath169 $ ] with @xmath170 .
for theoretical considerations the choice @xmath168 $ ] is more convenient .
since the dual group of the torus @xmath167 is @xmath171 , periodic band - limited functions on @xmath167 reduce to trigonometric polynomials ( of course technically @xmath0 does then no longer belong to @xmath8 since it is no longer in @xmath16 ) .
this suggests to use trigonometric polynomials as a realistic finite - dimensional model for a numerical solution of the nonuniform sampling problem .
we consider the space @xmath172 of trigonometric polynomials of degree @xmath173 of the form @xmath174 the norm of @xmath175 is @xmath176 since the distributional fourier transform of @xmath137 is @xmath177 we have @xmath178 $ ] .
hence @xmath172 is indeed a natural finite - dimensional model for @xmath8 . in general the @xmath1
are not the samples of a trigonometric polynomial in @xmath172 , moreover the samples are usually perturbed by noise , hence we may not find a @xmath175 such that @xmath179 .
we therefore consider the least squares problem @xmath180 here the @xmath181 are user - defined weights , which can be chosen for instance to compensate for irregularities in the sampling geometry @xcite . by increasing @xmath173 so that @xmath182 we can certainly find a trigonometric polynomial that interpolates the given data exactly .
however in the presence of noise , such a solution is usually rough and highly oscillating and may poorly resemble the original signal .
we will discuss the question of the optimal choice of @xmath173 if the original bandwidth is not known and in presence of noisy data in section [ ss : regul ] .
the following theorem provides an efficient numerical reconstruction algorithm .
it is also the key for the analysis of the relation between the finite - dimensional approximation in @xmath172 and the solution of the original infinite - dimensional sampling problem in @xmath8 .
@xcite [ th : act ] given the sampling points @xmath183 , samples @xmath184 , positive weights @xmath185 with @xmath186 .
+ step 1 : compute the @xmath187 toeplitz matrix @xmath188 with entries @xmath189 and @xmath190 by @xmath191 step 2 : solve the system @xmath192 step 3 : then the polynomial @xmath193 that solves is given by @xmath194 * numerical implementation of theorem / algorithm [ th : act ] : * + step 1 : the entries of @xmath188 and @xmath195 of equations and can be computed in @xmath196 operations ( where @xmath197 is the required accuracy ) using beylkin s unequally spaced fft algorithm @xcite .
+ step 2 : we solve @xmath198 by the conjugate gradient ( cg ) algorithm @xcite . the matrix - vector multiplication in each iteration of cg can be carried out in @xmath199 operations via fft @xcite .
thus the solution of takes @xmath200 operations , where @xmath201 is the number of iterations .
+ step 3 : usually the signal is reconstructed on regularly space nodes @xmath202 . in this case @xmath203 in can be computed by fft . for non -
uniformly spaced nodes @xmath204 we can again resort to beylkin s usfft algorithm .
there exists a large number of fast algorithms for the solution of toeplitz systems .
probably the most efficient algorithm in our case is cg .
we have already mentioned that the toeplitz system can be solved in @xmath205 via cg .
the number of iterations @xmath201 depends essentially on the clustering of the eigenvalues of @xmath188 , cf .
it follows from equation below and perturbation theory @xcite that , if the sampling points stem from a perturbed regular sampling set , the eigenvalues of @xmath188 will be clustered around @xmath206 , where @xmath206 is the oversampling rate . in such cases
we can expect a very fast rate of convergence .
the simple frame iteration @xcite is not able to take advantage of such a situation . for the analysis of the relation between the solution @xmath207 of theorem [ th : act ] and the solution @xmath0 of the original infinite - dimensional problem we follow grchenig @xcite .
assume that the samples @xmath12 of @xmath9 are given .
for the finite - dimensional approximation we consider only those samples @xmath1 for which @xmath53 is contained in the interval @xmath208 $ ] and compute the least squares approximation @xmath207 with degree @xmath173 and period @xmath209 as in theorem [ th : act ] .
it is shown in @xcite that if @xmath210 $ ] for all @xmath173 with @xmath211 then @xmath212 } |f(t ) - \plsp(t)|^2 \ , dt = 0 , \label{trigconv}\ ] ] and also @xmath213 uniformly on compact sets . under the nyquist condition @xmath214 and using weights @xmath215 grchenig has shown that @xmath216\ , , \label{condest}\ ] ] independently of @xmath173 , see @xcite .
these results validate the usage of trigonometric polynomials as finite - dimensional model for nonuniform sampling .
* example 1 reconsidered : * recall that in example 1 of section [ ss : truncated ] we have considered the reconstruction of a regularly oversampled signal @xmath9 . what does the reconstruction method of theorem [ th : act ] yield in this case ?
let us check the entries of the matrix @xmath188 when we take only those samples in the interval @xmath217 $ ] .
the period of the polynomial becomes @xmath218 with @xmath219 where @xmath220 is the number of given samples .
then @xmath221 for @xmath222 , where @xmath223 is kronecker s symbol with the usual meaning @xmath224 if @xmath225 and @xmath101 else .
hence we get @xmath226 where @xmath22 is the identity matrix on @xmath227 , thus @xmath188 resembles the structure of the infinite - dimensional frame operator @xmath29 in this case ( including exact approximation of the frame bounds ) .
recall that the truncated frame approach leads to an `` artificial '' ill - posed problem even in such a simple situation .
the advantages of the trigonometric polynomial approach compared to the truncated frame approach are manifold . in the one case we have to deal with an ill - posed problem which has no specific structure ,
hence its solution is numerically very expensive . in the other case
we have to solve a problem with rich mathematical structure , whose stability depends only on the sampling density , a situation that resembles the original infinite - dimensional sampling problem . in principle
the coefficients @xmath228 of the polynomial @xmath207 that minimizes could also be computed by directly solving the vandermonde type system @xmath229 where @xmath230 for @xmath231 and @xmath232 is a diagonal matrix with entries @xmath233 , cf .
several algorithms are known for a relatively efficient solution of vandermonde systems @xcite .
however this is one of the rare cases , where , instead of directly solving , it is advisable to explicitly establish the system of normal equations @xmath234 where @xmath235 and @xmath236 .
the advantages of considering the system @xmath198 instead of the vandermonde system are manifold : * the matrix @xmath188 plays a key role in the analysis of the relation of the solution of and the solution of the infinite - dimensional sampling problem , see and above .
* @xmath188 is of size @xmath237 , independently of the number of sampling points . moreover , since @xmath238 , it is of toeplitz type .
these facts give rise to fast and robust reconstruction algorithms . *
the resulting reconstruction algorithms can be easily generalized to higher dimensions , see section [ ss : multi ] .
such a generalization to higher dimensions seems not to be straightforward for fast solvers of vandermonde systems such as the algorithm proposed in @xcite .
0 an interesting finite - dimensional model is proposed in @xcite .
the bernstein - boas formula yields an explicit way to reconstruct a function @xmath9 from its ( sufficiently dense ) nonuniform samples @xmath239 , cf .
this formula involves the numerically intractable computation of infinite products .
however since only a finite number of samples can be used in a numerical reconstruction one may assume that the sequence of sampling points has regular structure outside a finite interval .
this allows to replace the infinite products by finite products which yields following approximation formula for @xmath0 @xmath240 and an estimate for the approximation error .
although their approach is computationally more expensive than the algorithm proposed in section [ s : trigpol ] their approach may be an attractive alternative if only a small number of samples in a short interval @xmath241 $ ] are available and if at the same time the signal to be reconstructed is `` strongly '' non - periodic on @xmath241 $ ] .
we point out that other finite - dimensional approaches are proposed in @xcite .
these approaches may provide interesting alternatives in the few cases where the algorithm outlined in section [ ss : trigpol ] does not lead to good results .
these cases occur when only a few samples of the signal @xmath0 are given in an interval @xmath242 $ ] say , and at the same time we have @xmath243 and @xmath244 , i.e. , if @xmath0 is `` strongly non - periodic '' on @xmath242 $ ] .
however the computational complexity of the methods in @xcite is significantly larger .
the approach presented above can be easily generalized to higher dimensions by a diligent book - keeping of the notation .
we consider the space of @xmath245-dimensional trigonometric polynomials @xmath246 as finite - dimensional model for @xmath247 . for given samples @xmath1 of @xmath248 , where @xmath249
, we compute the least squares approximation @xmath207 similar to theorem [ th : act ] by solving the corresponding system of equations @xmath198 . in 2-d for instance the matrix @xmath188 becomes a block toeplitz matrix with toeplitz blocks @xcite . for a fast computation of the entries of @xmath19
we can again make use of beylkin s usfft algorithm @xcite . and similar to 1-d , multiplication of a vector by @xmath188
can be carried out by 2-d fft .
also the relation between the finite - dimensional approximation in @xmath246 and the infinite - dimensional solution in @xmath250 is similar as in 1-d .
the only mathematical difficulty is to give conditions under which the matrix @xmath188 is invertible . since the fundamental theorem of algebra does not hold in dimensions larger than one , the condition @xmath251 is necessary but no longer sufficient for the invertibility of @xmath188 .
sufficient conditions for the invertibility , depending on the sampling density , are presented in @xcite .
in this section we discuss several numerical aspects of nonuniform sampling that are very important from a practical viewpoint , however only few answers to these problems can be found in the literature . in almost all theoretical results and numerical algorithms for reconstructing a band - limited signal from nonuniform samples
it is assumed that the bandwidth is known a priori .
this information however is often not available in practice .
a good choice of the bandwidth for the reconstruction algorithm becomes crucial in case of noisy data .
it is intuitively clear that choosing a too large bandwidth leads to over - fit of the noise in the data , while a too small bandwidth yields a smooth solution but also to under - fit of the data . and
of course we want to avoid the determination of the `` correct '' @xmath2 by trial - and - error methods .
hence the problem is to design a method that can reconstruct a signal from non - uniformly spaced , noisy samples without requiring a priori information about the bandwidth of the signal .
the multilevel approach derived in @xcite provides an answer to this problem .
the approach applies to an infinite - dimensional as well as to a finite - dimensional setting .
we describe the method directly for the trigonometric polynomial model , where the determination of the bandwidth @xmath2 translates into the determination of the polynomial degree @xmath173 of the reconstruction .
the idea of the multilevel algorithm is as follows .
let the noisy samples @xmath252 of @xmath9 be given with @xmath253 and let @xmath254 denote the orthogonal projection from @xmath8 into @xmath172 .
we start with initial degree @xmath255 and run algorithm [ th : act ] until the iterates @xmath256 satisfy for the first time the _ inner _ stopping criterion @xmath257 for some fixed @xmath164 .
denote this approximation ( at iteration @xmath258 ) by @xmath259 . if @xmath259 satisfies the _
outer _ stopping criterion @xmath260 we take @xmath259 as final approximation .
otherwise we proceed to the next level @xmath261 and run algorithm [ th : act ] again , using @xmath259 as initial approximation by setting @xmath262 . at level @xmath263
the inner level - dependent stopping criterion becomes @xmath264 while the outer stopping criterion does not change since it is level - independent .
stopping rule guarantees that the iterates of cg do not diverge .
it also ensures that cg does not iterate too long at a certain level , since if @xmath173 is too small further iterations at this level will not lead to a significant improvement .
therefore we switch to the next level .
the outer stopping criterion controls over - fit and under - fit of the data , since in presence of noisy data is does not make sense to ask for a solution @xmath265 that satisfies @xmath266 . since the original signal @xmath0 is not known , the expression @xmath267 in can not be computed . in @xcite
the reader can find an approach to estimate @xmath267 recursively .
a variety of conditions on the sampling points @xmath268 are known under which the set @xmath48 is a frame for @xmath8 , which in turn implies ( at least theoretically ) perfect reconstruction of a signal @xmath0 from its samples @xmath1 .
this does however not guarantee a stable reconstruction from a numerical viewpoint , since the ratio of the frame bounds @xmath269 can still be extremely large and therefore the frame operator @xmath29 can be ill - conditioned .
this may happen for instance if @xmath270 in goes to 1 , in which case @xmath271 may become large .
the sampling problem may also become numerically unstable or even ill - posed , if the sampling set has large gaps , which is very common in astronomy and geophysics .
note that in this case the instability of the system @xmath198 does _ not _ result from an inadequate discretization of the infinite - dimensional problem .
there exists a large number of ( circulant ) toeplitz preconditioners that could be applied to the system @xmath198 , however it turns out that they do not improve the stability of the problem in this case .
the reason lies in the distribution of the eigenvalues of @xmath188 , as we will see below .
following @xcite , we call two sequences of real numbers @xmath272 and @xmath273 _ equally distributed _ , if @xmath274 = 0 \label{defdist}\ ] ] for any continuous function @xmath275 with compact support and @xmath276 are required to belong to a common interval . ] .
let @xmath277 be a @xmath278 circulant matrix with first column @xmath279 , we write @xmath280 .
the eigenvalues of @xmath277 are distributed as @xmath281 .
observe that the toeplitz matrix @xmath282 with first column @xmath283 can be embedded in the circulant matrix @xmath284 thms 4.1 and 4.2 in @xcite state that the eigenvalues of @xmath282 and @xmath285 are equally distributed as @xmath286 where @xmath287 the partial sum of the series is @xmath288 to understand the clustering behavior of the eigenvalues of @xmath188 in case of sampling sets with large gaps , we consider a sampling set in @xmath289 , that consists of one large block of samples and one large gap , i.e. , @xmath290 for @xmath291 for @xmath292 .
( recall that we identify the interval with the torus ) .
then the entries @xmath293 of the toeplitz matrix @xmath188 of ( with @xmath294 ) are @xmath295 to investigate the clustering behavior of the eigenvalues of @xmath188 for @xmath296 , we embed @xmath188 in a circulant matrix @xmath297 as in . then becomes @xmath298}\ ] ]
whence @xmath299}$ ] for @xmath296 , where @xmath300}(x ) = 1 $ ] , if @xmath301 and 0 else .
thus the eigenvalues of @xmath188 are asymptotically clustered around zero and one .
for general nonuniform sampling sets with large gaps the clustering at 1 will disappear , but of course the spectral cluster at 0 will remain . in this case
it is known that the preconditioned problem will still have a spectral cluster at the origin @xcite and preconditioning will not be efficient .
fortunately there are other possibilities to obtain a stabilized solution of @xmath198 .
the condition number of @xmath188 essentially depends on the ratio of the maximal gap in the sampling set to the nyquist rate , which in turn depends on the bandwidth of the signal .
we can improve the stability of the system by adapting the degree @xmath173 of the approximation accordingly .
thus the parameter @xmath173 serves as a regularization parameter that balances stability and accuracy of the solution .
this technique can be seen as a specific realization of _ regularization by projection _ , see chapter 3 in @xcite .
in addition , as described in section [ ss : regul ] , we can utilize cg as regularization method for the solution of the toeplitz system in order to balance approximation error and propagated error . the multilevel method introduced in section [ ss : ml ]
combines both features . by optimizing the level ( bandwidth ) and the number of iterations in each level it provides an efficient and robust regularization technique for ill - conditioned sampling problems .
see section [ s : applications ] for numerical examples . 0 in many applications the physical process that generates the signal implies not only that the signal is ( essentially ) band - limited but also that its spectrum of the signal has a certain rate of decay .
for instance geophysical potential fields have exponentially decaying fourier transform .
this a priori knowledge can be used to improve the accuracy of the approximation . instead of the usual regularization methods , such as tikhonov regularization , we propose a different , computationally much more efficient method .
assume that the decay of the fourier transform of @xmath0 can be bounded by @xmath302 .
typical choice in practice are @xmath303 or @xmath304 . for a given system
@xmath305 define the diagonal matrix @xmath306 by @xmath307 .
instead of solving @xmath308 we consider the `` weighted problem '' @xmath309 or @xmath310 in the first case the solution is @xmath311 and in the second case we have @xmath312 of course , if @xmath19 is invertible both solutions coincide with the solution of @xmath313 .
however if @xmath19 is not invertible , then both equations lead to a weighted minimal norm least squares solution .
note that @xmath306 is not chosen to minimize the condition number of the problem , since as outlined above standard preconditioning will not work in this case .
systems and can be solved by conjugate gradient methods .
hence the computational effort of such an approach is of the same order as algorithm [ th : act ] .
a detailed numerical analysis of the convergence properties of this approach has still to be completed .
for a numerical example see section [ ss : geo ] .
we present two numerical examples to demonstrate the performance of the described methods .
the first one concerns a 1-d reconstruction problem arising in spectroscopy . in the second example we approximate the earth s magnetic field from noisy scattered data .
the original spectroscopy signal @xmath0 is known at 1024 regularly spaced points @xmath53 .
this discrete sampling sequence will play the role of the original continuous signal .
to simulate the situation of a typical experiment in spectroscopy we consider only 107 randomly chosen sampling values of the given sampling set .
furthermore we add noise to the samples with noise level ( normalized by division by @xmath314 ) of @xmath315 .
since the samples are contaminated by noise , we can not expect to recover the ( discrete ) signal @xmath0 completely . the bandwidth is approximately @xmath316 which translates into a polynomial degree of @xmath317 .
note that in general @xmath2 and ( hence @xmath173 ) may not be available .
we will also consider this situation , but in the first experiments we assume that we know @xmath2 .
the error between the original signal @xmath0 and an approximation @xmath318 is measured by computing @xmath319 .
first we apply the truncated frame method with regularized svd as described in section [ ss : truncated ] .
we choose the truncation level for the svd via formula .
this is the optimal truncation level in this case , providing an approximation with least squares error @xmath320 .
figure [ fig : spect](a ) shows the reconstructed signal together with the original signal and the noisy samples . without regularization
we get a much worse `` reconstruction '' ( which is not displayed ) .
we apply cg to the truncated frame method , as proposed in section [ ss : cgtrunc ] with stopping criterion ( for @xmath321 ) .
the algorithm terminates already after 3 iterations .
the reconstruction error is with @xmath322 slightly higher than for truncated svd ( see also figure [ fig : spect](b ) ) , but the computational effort is much smaller . also algorithm [ th : act ] ( with @xmath323 ) terminates after 3 iterations .
the reconstruction is shown in figure [ fig : spect](c ) , the least squares error ( @xmath324 ) is slightly smaller than for the truncated frame method , the computational effort is significantly smaller .
we also simulate the situation where the bandwidth is not known a priori and demonstrate the importance of a good estimate of the bandwidth .
we apply algorithm [ th : act ] using a too small degree ( @xmath325 ) and a too high degree ( @xmath326 ) .
( we get qualitatively the same results using the truncated frame method when using a too small or too large bandwidth ) .
the approximations are shown in figs .
[ fig : spect](d ) and ( e ) , the approximation errors are @xmath327 and @xmath328 , respectively .
now we apply the multilevel algorithm of section [ ss : ml ] which does not require any initial choice of the degree @xmath173 .
the algorithm terminates at `` level '' @xmath329 , the approximation is displayed in fig .
[ fig : spect](f ) , the error is @xmath330 , thus within the error bound @xmath128 , as desired .
hence without requiring explicit information about the bandwidth , we are able to obtain the same accuracy as for the methods above .
exploration geophysics relies on surveys of the earth s magnetic field for the detection of anomalies which reveal underlying geological features .
geophysical potential field - data are generally observed at scattered sampling points .
geoscientists , used to looking at their measurements on maps or profiles and aiming at further processing , therefore need a representation of the originally irregularly spaced data at a regular grid .
the reconstruction of a 2-d signal from its scattered data is thus one of the first and crucial steps in geophysical data analysis , and a number of practical constraints such as measurement errors and the huge amount of data make the development of reliable reconstruction methods a difficult task . it is known that the fourier transform of a geophysical potential field @xmath0 has decay @xmath331 .
this rapid decay implies that @xmath0 can be very well approximated by band - limited functions @xcite .
since in general we may not know the ( essential ) bandwidth of @xmath0 , we can use the multilevel algorithm proposed in section [ ss : ml ] to reconstruct @xmath0 .
the multilevel algorithm also takes care of following problem .
geophysical sampling sets are often highly anisotropic and large gaps in the sampling geometry are very common .
the large gaps in the sampling set can make the reconstruction problem ill - conditioned or even ill - posed . as outlined in section [ ss : regul ] the multilevel algorithm iteratively determines the optimal bandwidth that balances the stability and accuracy of the solution .
figure [ fig : geo](a ) shows a synthetic gravitational anomaly @xmath0 .
the spectrum of @xmath0 decays exponentially , thus the anomaly can be well represented by a band - limited function , using a `` cut - off - level '' of @xmath332 for the essential bandwidth of @xmath0 .
we have sampled the signal at 1000 points @xmath333 and added 5% random noise to the sampling values @xmath334 . the sampling geometry shown in figure [ fig : geo ] as black dots exhibits several features one encounters frequently in exploration geophysics @xcite .
the essential bandwidth of @xmath0 would imply to choose a polynomial degree of @xmath335 ( i.e. , @xmath336 spectral coefficients ) . with this choice of @xmath173
the corresponding block toeplitz matrix @xmath188 would become ill - conditioned , making the reconstruction problem unstable . as mentioned above
, in practice we usually do not know the essential bandwidth of @xmath0 .
hence we will not make use of this knowledge in order to approximate @xmath0 .
we apply the multilevel method to reconstruct the signal , using only the sampling points @xmath337 , the samples @xmath338 and the noise level @xmath339 as a priori information .
the algorithm terminates at level @xmath340 .
the reconstruction is displayed in figure [ fig : geo](c ) , the error between the true signal and the approximation is shown in figure [ fig : geo](d ) . the reconstruction error is @xmath341 ( or @xmath342 mgal ) , thus of the same order as the data error , as desired .
k. grchenig .
non - uniform sampling in higher dimensions : from trigonometric polynomials to band - limited functions . in j.j .
benedetto and p.j.s.g ferreira , editors , _ modern sampling theory : mathematics and applications_. birkhuser , boston , to appear . | we give an overview of recent developments in the problem of reconstructing a band - limited signal from non - uniform sampling from a numerical analysis view point .
it is shown that the appropriate design of the finite - dimensional model plays a key role in the numerical solution of the non - uniform sampling problem . in the one approach ( often proposed in the literature ) the finite - dimensional model leads to an ill - posed problem even in very simple situations . the other approach that we consider leads to a well - posed problem that preserves important structural properties of the original infinite - dimensional problem and gives rise to efficient numerical algorithms .
furthermore a fast multilevel algorithm is presented that can reconstruct signals of unknown bandwidth from noisy non - uniformly spaced samples .
we also discuss the design of efficient regularization methods for ill - conditioned reconstruction problems .
numerical examples from spectroscopy and exploration geophysics demonstrate the performance of the proposed methods . 0 subject classification : 65t40 , 65f22 , 42a10 , 94a12 + non - uniform sampling , band - limited functions , frames , regularization , signal reconstruction , multi - level method . |
a supernova remnant ( snr ) reflects the abundance of the progenitor star when the remnant is young and that of the interstellar matter ( ism ) when it becomes old . in this way
, we can study the evolution of the ejecta and the ism .
the cygnus loop is a proto - typical middle - aged shell - like snr .
the angular diameter is about [email protected] and it is very close to us ( 540pc ; blair et al . 2005 ) , implying a diameter of @xmath123pc .
the estimated age is about 10000years , less than half that based on the previous distance estimate of 770pc @xcite .
since the cygnus loop is an evolved snr , the bright shell mainly consists of a shock - heated surrounding material .
its supernova ( sn ) explosion is generally considered to have occurred in a preexisting cavity @xcite .
levenson et al . ( 1997 ) found that the cygnus loop was a result of a cavity explosion that was created by a star no later than b0 .
it is almost circular in shape with a break - out in the south where the hot plasma extends out of the circular shape .
miyata et al . (
1994 ) observed the northeast ( ne ) shell of the loop with asca and revealed the metal deficiency there @xcite . since dopita et al .
( 1977 ) reported the metal deficiency of the ism around the cygnus loop , they concluded that the plasma in the ne - shell is dominated by the ism . due to the constraints of the detector efficiency
, they assumed that the relative abundances of c , n and o are equal to those of the solar value @xcite .
more recently , miyata et al .
( 2007 ) used the suzaku satellite @xcite to observe one pointing position in the ne rim .
they detected emission lines from c and n and determined the relative abundances @xcite .
they concluded that the relative abundances of c , n and o are consistent with those of the solar values whereas the absolute abundances show depletion from the solar values @xcite .
katsuda et al .
( 2007 ) observed four pointings in the ne rim and detected a region where the relative abundances of c and n are a few times higher than that of o. hatsukade & tsunemi ( 1990 ) detected a hot plasma inside the cygnus loop that is not expected in the simple sedov model @xcite .
they reported that the hot plasma was confined inside the loop .
miyata et al . ( 1998 ) detected strong emission lines from si , s and fe - l from inside the loop @xcite .
they found that the metal abundance is at least several times higher than that of the solar value @xcite , indicating that a few tens of higher than that of the shell region .
they concluded that the metal rich plasma was a fossil of the sn explosion .
the abundance ratio of si , s and fe indicated the progenitor star mass to be 25 .
miyata & tsunemi ( 1999 ) measured the radial profile inside the loop and found a discontinuity around 0.9r@xmath3 where r@xmath3 is the shock radius .
they measured the metallicity inside the hot cavity and estimated the progenitor mass to be 15 .
levenson et al . ( 1998 ) estimated the size of the cavity and the progenitor mass to be 15 .
therefore , the progenitor star of the cygnus loop is a massive star in which the triple-@xmath4 reaction should have dominated rather than the cno cycle .
if the surrounding material of the cygnus loop is contaminated by the stellar activity of the progenitor star , it may explain the c abundance inferred for this region with suzaku @xcite . in order to study the plasma condition inside the cygnus loop ,
we observed it from the ne rim to the south - west ( sw ) rim with the xmm - newton satellite .
we report here the result covering a full diameter by seven pointings .
we performed seven pointing observations of the cygnus loop so that we could cover the full diameter from the ne rim to the sw rim ( from pos-1 to pos-7 ) during the ao-1 phase .
we concentrate on the data obtained with the epic mos and pn cameras .
all the data were taken by using medium filters and the prime full window mode . fortunately , all the data other than pos-4 suffered very little from background flares .
obs ids , the observation date , the nominal point , and the effective exposure times after rejecting the high - background periods are summarized in table [ obs ] .
all the raw data were processed with version 6.5.0 of the xmm science analysis software ( xmmsas ) .
we selected x - ray events corresponding to patterns 012 and 0 for mos and pn , respectively .
we further cleaned the data by removing all the events in bad columns listed in the literature @xcite .
after filtering the data , they were vignetting - corrected using the xmmsas task evigweight . for the background subtraction , we employed the data set accumulated from blank sky observations prepared by @xcite . after adjusting its normalization to the source data by using the energy range between 5kev and 12kev , where the emission is free from the contamination @xcite
, we subtracted the background data set from the source .
figure [ image ] displays an exposure - corrected rosat hri image of the entire cygnus loop ( black and white ) overlaid with the xmm - newton color images of the merged mos1/2 , pn data from all the xmm - newton observations . in this figure , we allocated color codes as red ( 0.30.52kev ) , green ( 0.521.07kev ) and blue ( 1.073kev ) .
we see that the outer regions are reddish rather than bluish while the central region is in bluish .
the ne rim is the brightest in our field of view ( fov ) , showing a bright filament at @xmath5 to the radial direction corresponding to ngc6992 .
the sw rim is also bright in our fov where there is a v - shape structure @xcite . in the center of the loop ,
an x - ray bright filament runs through pos-4 and pos-5 forming a circular structure . in the rosat image
, we can see it and find that it forms a large circle within the cygnus loop . in this way
, there are many fine bright filaments in intensity . however , we find that there is a clear intensity variation along our scan path : dim in the center and bright in the rim .
figure [ all_spectra ] shows spectra for seven pointings ; each is the sum of the entire fov .
the ne rim ( pos-1 ) and the sw rim ( pos-7 ) show strong emission lines below 1kev including o ,
fe - l and ne , while the center ( pos-4 ) shows strong emission lines from si and s. we can see that the equivalent width of si and s emission lines are bigger in the center and gradually decrease toward the rim . we show the comparison of spectra between pos-1 and pos-4 in figure [ spec ] .
prominent emission lines are o - he@xmath4 , o - ly@xmath4 , fe - l complex , ne - he@xmath4 , mg - he@xmath4 , si - he@xmath4 , and s - he@xmath4 .
we see that the emission line shapes for o are quite similar to each other while there is a big difference at higher energy band .
since the spectrum in the ne rim can be well represented by a single temperature plasma model @xcite , we need an extra component in the center .
although there are many fine structures , no matter how finely we divide our fov , each region would contain different plasma conditions due to the integration of the emission along the line of sight .
therefore , we concentrate on large scale structure along the scan path .
first of all , we divided our fov into two parts along the diameter : the north path and the south path .
then we divided them into many small annular sectors whose center is located at ( 20@xmath651@[email protected] , 31@xmath900@xmath1000@xmath11 ) , i.e. , the center of the nominal position of pos-4 .
there are 141 and 172 annular sectors for the north path and the south path , respectively .
these small annular sectors , shown in figure [ image ] , are divided such that each has at least 60,000photons ( @xmath120,000 for mos1/2 and @xmath140,000 for pn ) to equalize the statistics .
we extracted the spectrum from each sector using the data set accumulated from blank sky observations as sky background .
we have confirmed that the emission above 3kev is statistically zero . in this way
, we obtained 313 spectra .
these sectors can be identified by the angular distance , `` r '' , from the center ( east is negative and west is positive as shown in figure [ image ] ) .
the width of the sector depends on r. the sector widths range from [email protected] to [email protected] in the north path and from [email protected] to [email protected] in the south path .
the widest sectors are in pos-4 due to its short exposure because of the background flare .
the narrowest sectors are in the ne rim where the surface brightness is the highest .
we fitted the spectrum for each sector with an absorbed non - equilibrium ionization ( nei ) model with a single @xmath0 , using the wabs @xcite and vnei model ( nei version 2.0 ) @xcite in xspec v12.3.1 @xcite ) .
we fixed the column density , @xmath12 to be @xmath13 ( e.g. , @xcite , @xcite ) .
free parameters were @xmath0 ; the ionization time scale , @xmath14 ( a product of the electron density and the elapsed time after the shock heating ) ; the emission measure ( hereafter em ; em = @xmath15 , where @xmath16 and @xmath17 are the number densities of hydrogens and electrons , @xmath18 is the plasma depth ) ; and abundances of c , n , o , ne , mg , si , s , fe , and ni . we set abundances of c and n equal to that of o , that of ni equal to fe , and other elements fixed to the solar values @xcite . in the fitting process , we set 20 as the minimum counts in each spectral bin to perform the @xmath19 test .
we determined the value of the minimum counts such that it did not affect the fitting results .
figure [ chi2_dist ] shows the distribution of the reduced-@xmath19 in black as a function of r along both the north path and the south path .
we found that values of reduced-@xmath19 for all the sectors are between 1.0 and 2.0 .
if we took into account a systematic error @xcite of 5% , the reduced-@xmath19 was around 1.5 or less . in general ,
values of the reduced-@xmath19 are a little higher in the central part of the cygnus loop .
miyata et al . ( 1994 ) observed the ne rim with asca and found that the spectra were well represented with a one - temperature vnei model with a temperature gradient towards the inside .
the suzaku observation in the ne rim @xcite reveals that the x - ray spectrum can be represented by a two - temperature model : one component is 0.20.35kev and the other is 0.090.15kev . in our fitting ,
the value of @xmath0 obtained is 0.20.25kev .
therefore , we detected a hot component that suzaku detected .
there may be an additional component with low temperature that seems difficult to detect with xmm - newton due to the relatively lower sensitivity below 0.5kev compared to suzaku .
the asca observation @xcite also shows that the metal abundance in the ne rim is deficient .
the authors concluded that the plasma in the ne - rim consists of the interstellar matter ( ism ) rather than the ejecta .
this is confirmed with the suzaku observation @xcite that indicates the abundances of c , n and o to be @xmath10.1 , 0.05 and 0.1 solar , respectively .
we also obtained the metal deficiency in the ne rim data ; the best - fit results are given in figure [ ex_spec ] ( left ) and in table [ param1 ] .
leahy ( 2004 ) measured the x - ray spectrum of the southwest region of the cygnus loop and reported that the oxygen abundance there is about 0.22 solar @xcite .
therefore , the x - ray measurements of the cygnus loop show that the metal abundances are depleted .
cartledge et al . ( 2004 ) measured the interstellar oxygen along 36 sight lines and confirmed the homogeneity of the o / h ratio within 800pc of the sun .
we found that they measured it in the direction about @xmath20 away from the cygnus loop .
the oxygen abundance they measured is about 0.4 times the solar value @xcite .
wilms et al . ( 2000 ) employed 0.6 of the total interstellar abundances for the gas - phase ism oxygen abundance , and suggest that this depletion may be due to grains .
although the ism near the cygnus loop may be depleted , the abundances are still much higher than what we obtained at the rim of the cygnus loop .
it is difficult to explain such a low abundance of oxygen in material originating from the ism .
therefore , the origin of the low metal abundance is open to the question .
since the cygnus loop is thought to have exploded in a pre - existing cavity , we can say that the cavity material shows low metal abundance .
the abundance difference between our data and those from suzaku may be due to the difference in the detection efficiency at low energy . taking into account the projection effect , the plasma of the rim regions consists only of the cavity material while that of the inner regions consists both of the cavity material and of an extra component filling the interior of the loop . to further constrain the plasma conditions ,
we applied a two - component nei model with different temperatures . in this model
, we added an extra component to the single temperature model .
the extra component is also an absorbed vnei model with @xmath0 , @xmath14 , and em as free parameters .
the metal abundances of the extra component are fixed to those determined at the ne rim so that the extra component represents the cavity material .
figure [ chi2_dist ] shows reduced-@xmath19 values in red along the path .
applying the @xmath21-test with a significance level of 99% to determine whether or not an extra component is needed , we found that most of the spectra required a two - component model , particularly in the central part of the loop .
sectors that do not require two - component model are mainly clustered in r@xmath22 - 65@xmath10 , + 25@xmath23r@xmath24 , and + 60@xmath23r .
therefore , we considered that the outer sectors ( @xmath25r@xmath26 ) can be safely represented by a one - component model while other sectors can be represented by a two - component model . in this way
, we performed the analysis by applying a two - component vnei model with different temperatures .
we assumed that the low temperature component comes from the surrounding region of the cygnus loop and the high temperature component occupies the interior of the loop .
we found that the values of reduced-@xmath19 are 1.01.8 even employing a two - component model .
this is partly due to the systematic errors .
looking at the image in detail , there are fine structures within the sector .
furthermore , the spectrum from each sector is an integration along the line of sight .
since we only employ two vnei plasma models , the values of reduced-@xmath19 are mainly due to the simplicity of the plasma model employed here .
therefore , we think that the plasma parameters obtained will represent typical values in each sector . figure [ ex_spec ] ( right ) and table [ param2 ] shows an example result that comes from the sector at r=@xmath27 . fixed
parameters in the low @xmath0 component come from the fitting result at the ne rim obtained by suzaku observations @xcite .
metal abundances for the high @xmath0 component show higher values by an order of magnitude than those of the low @xmath0 component , surely confirming that the high @xmath0 component is dominated by fossil ejecta
. figure [ kt_dist ] shows temperatures as a function of position .
the low @xmath0 component is in the temperature range of 0.120.34kev while the high @xmath0 component is above 0.35kev .
there is a clear temperature difference where a two - component model is required rather than a single temperature model .
the low @xmath0 component represents the cavity material surrounding the cygnus loop while the high @xmath0 component represents the fossil ejecta inside the loop .
figure [ flux_dist ] shows the fluxes for the two components as a function of position .
the low @xmath0 component shows clear rim brightening . the east part is stronger than the west part , showing asymmetry of the loop . on the other hand ,
the high @xmath0 component has a relatively flat radial dependence . from the center to the sw
, we see that the flux of the high @xmath0 component is stronger than that of the low @xmath0 component .
as shown in figure [ kt_dist ] , the low @xmath0 component shows relatively constant temperature with radius .
the distribution of the flux shown in figure [ flux_dist ] shows peaks in the rim and relatively low values inside the loop .
there are some differences between the north path and the south path .
the biggest one is a clear difference in peak position in the ne rim that is due to the bright filament at @xmath5 to the radial direction , as seen in figure [ image ] .
however , these two paths show a globally similar behavior in flux
. therefore , we can see that they are quite similar to each other from a large scale point of view .
we notice that there are many aspects showing asymmetry and non - uniformity .
the ne half is stronger in intensity than the sw half .
the flux in the inner part of the loop shows relatively small values in the west half , particularly at + 25@xmath23r@xmath24 .
the ne half is brighter by a factor of 510 than the sw half
. furthermore , the sw half shows stronger intensity variation than the ne half .
this suggests that the thickness of the cavity shell is far from uniform . the cavity shell in the sw
half is much thinner than that in the ne half .
since we assumed the metal abundances of the low @xmath0 component equal to those of the ne rim , we can calculate the em .
furthermore , we assumed the ambient density to be 0.7@xmath28 based on the observation of the ne rim , and we estimate the mass of the low @xmath0 component to be 130 .
however , we should note that there are is evidence that the sn explosion which produced the cygnus loop occurred within a preexisting cavity ( e.g. , hester et al .
1994 ; levenson et al . 1998 ; levenson et al .
the model predicts that the original cavity density , @xmath29 , is related to the wall density @xmath30 by @xmath31 .
assuming that @xmath32 equals the ambient density , @xmath30 , we estimate @xmath29 to be 0.14 @xmath28 .
then , we calculate the total mass in the preexisting cavity to be @xmath125 m@xmath33 .
the flux distribution from the ejecta along the path is shown with filled circles in figure [ flux_dist ] .
it has a relatively flat structure with two troughs around r=@xmath34 and r=+50@xmath10 .
since we left the metal abundances as free parameters , we obtained distributions of em of various metals ( c = n = o , fe = ni , ne , mg , si and s ) in the ejecta .
these are shown in figure [ metal_dist ] , where black crosses trace the north path and red crosses trace along the south path .
if we assume uniform plasma conditions along the line of sight , the em represents the mass of the metal .
most elements show similar structure between the north path and the south path while there is a big discrepancy in fe / ni distribution at @xmath35 r @xmath36 . in this region ,
the south path shows two times more abundant fe / ni than the north path .
a similar discrepancy is seen in o ( @xmath37 r @xmath38 ) and in ne ( at @xmath35 r @xmath39 ) .
therefore , the distribution of metal abundance shows a north - south asymmetry along the path .
the distributions of o and ne show a central bump and increase at the outer sectors . however , those of mg , si , s and fe only show a central bump .
the increase of o and ne at the outer sectors indicates that the outer parts of the ejecta mainly consists of o and ne and they may be well mixed .
similarly heavy elements , mg , si , s and fe / ni forming central bumps may show that they are well mixed . therefore , a significant convection has occurred in the central bumps while an `` onion - skin '' structure remains in the outer sectors .
the cygnus loop appears to be almost circular with a blowout in the south .
the rosat image indicates no clear shell in this blowout region .
levenson et al . ( 1997 ) revealed that there is a thin shell left at the edge of the blowout region .
therefore , there is a small amount of cavity material in this region that surrounds the ejecta .
this also indicates the non - uniformity of the cavity wall . if the cavity wall is thin , the ejecta can produce a blowout structure .
looking at the component of the cavity material along our path shown in figure [ flux_dist ] , the flux is very weak at + 15@xmath23r@xmath24 .
this indicates that the cavity wall is very thin in this region .
when we calculate the flux ratio between the ejecta plasma and the cavity material , we find that the ratio becomes high ( larger than 4 ) at + 15@xmath23r@xmath40 in the north path and + 30@xmath23r@xmath40 in the south path .
therefore , we guess that the thin shell region is larger in the north path than in the south path .
this also shows the asymmetry between the north and the south as well as that between the east and the west . if the thin shell region corresponds to a blowout similar to that in the south blowout , this region must have a blowout structure along the line of sight either in the near side or far side or both
this structure roughly corresponds to pos-5 and will extend further in the northwest direction . looking at the rosat image in figure [ image ] , we see a circular region with low intensity
. it is centered at ( 20@xmath649@xmath711@xmath8 , 31@xmath905@xmath1020@xmath11 ) with a radius of 30@xmath10 .
we guess that this circular region corresponds to a possible blowout in the direction of our line of sight .
ccd observation just north of our path will answer this hypothesis .
we obtained ems of o , ne , mg , si , s , and fe for the ejecta along the north path and the south path . multiplying the ems by the area of each sector
, we obtained emission integral ( hereafter ei , ei@xmath41 , @xmath42 is the x - ray - emitting volume ) along the path .
since we only observed the limited area of the cygnus loop from the ne rim to the sw rim , we have to estimate the eis for the entire remnant in order to obtain the relative abundances as well as the total mass of the ejecta .
therefore , we divide our observation region into four parts : left - north part , right - north part , left - south part , and right - south part .
we assume that each part represents the average eis of the corresponding quadrant of the loop . in this way
, we can calculate the total eis for o , ne , mg , si , s , and fe that are described in table [ emission_integral ] .
the south quadrant , corresponding to the right - south path , contains the largest mass fraction of 31% , while the other quadrants contain 23% each .
then , we calculate the relative abundances of ne , mg , si , s , and fe to o in the entire ejecta .
since we can not measure the abundance of light elements like he , it is quite difficult to estimate the absolute abundances .
however , the relative abundance to o is robust .
since the cygnus loop is believed to be a result from a core - collapse sn , we compared our data with core - collapse sn models .
there are many theoretical results from various authors ( e.g. , woosley & weaver 1995 ; thielemann et al .
1996 ; rauscher et al . 2002 ; tominaga et al .
we also employed a sn type ia model @xcite for comparison .
we calculated the relative abundance for various elements to o and compared them with models .
figure [ rel_abund ] shows comparisons between the model calculations and our results where we picked up woosley s model with one solar abundance @xcite for the core - collapse case @xcite .
the type ia model yields more si , s and fe than our results , but less ne . models with massive stars produce better fits to our results than the type ia model . among them
, we found that the model with 15 showed good fits to our results .
they fit within a factor of two with an exception of fe .
we also noticed that the model with one solar abundance looked better fit than that with depleted abundance .
therefore , we can conclude that the cygnus loop originated from an approximately 15 star with one solar abundance . assuming that the ejecta density is uniform along the line of sight
, we estimate the total mass of the fossil ejecta to be 21 . in this calculation
, we assumed that the electron density is equal to that of hydrogen and that the plasma filling factor is unity , although the fossil ejecta might be deficient in hydrogen .
if it is the case , the total mass of the fossil ejecta reduces to @xmath112 , whereas the relative abundances are not affected .
the most suitable nucleosynthetic model predicts that the total mass ejected is about 6 without h. therefore , there might be a significant amount of contamination from the swept - up matter into the high-@xmath0 component , which we consider the ejecta .
otherwise , the assumption that the density of the ejecta is uniform might be incorrect since rim - brightening for the ems of o , ne , mg , and fe is clearly seen in fig . [ metal_dist ] .
non - uniformity reduces the filling factor and also the mass of the high-@xmath0 component .
there is observational evidence of the asymmetry of supernova explosions both for massive stars @xcite and for type ia @xcite .
we found that the ejecta plasma shows asymmetric structure between ne half and sw half .
ne and fe are evenly divided while two thirds of o and mg are in the ne half . on the contrary ,
two third of si and s are in the sw half .
we calculated the ejecta mass for each quadrant and found that the south quadrant contains the largest ejecta mass .
similar asymmetries are seen in other snr , such as puppis a , which shows asymmetric structure with o - rich , fast - moving knots ( winkler & kirscher 1985 ; winkler et al .
the central compact object in puppis a is on the opposite side of the snr from the o - rich , fast - moving knots @xcite .
if the asymmetry of the ejecta in the cygnus loop is similar to that of puppis a , we may expect a compact object to be in the north direction .
we have observed the cygnus loop along the diameter from the ne rim to the sw rim employing xmm newton .
the fov is divided into two paths : the north path and the south path .
then it is divided into many small annuli so that each annulus contains a similar number of photons to preserve statistics .
the spectra from the rim regions can be expressed by a one-@xmath0 component model while those in the inner region require a two-@xmath0 component model .
the low @xmath0 plasma shows relatively low metal abundance and covers the entire fov .
it forms a shell that originates from the preexisting cavity .
the high @xmath0 plasma shows high metal abundance and occupies a large part of the fov .
the origins of these two components are different : the high @xmath0 plasma with the high metal abundance must come from the ejecta while low @xmath0 plasma with low metal abundance must come from the cavity material .
we find that the thickness of the shell is very thin in the south west part where , we guess , the ejecta plasma is blow out in the direction of our line of sight .
we estimate the mass of the metals . based on the relative metal abundance , we find that the cygnus loop originated from a 15 star .
the distribution of the ejecta is asymmetric , suggesting an asymmetric explosion .
this work is partly supported by a grant - in - aid for scientific research by the ministry of education , culture , sports , science and technology ( 16002004 ) .
this study is also carried out as part of the 21st century coe program , .
s. k. is supported by jsps research fellowship for young scientists .
anders , e. , & grevesse , n. 1989 , geochim .
acta , * 53 * , 197 arnaud , k. a. 1996 , asp conf .
ser . , * 101 * , 17 ashcehnbach , b. & leahy , d. a. 1999 , a&a , * 341 * , 602 burrows , adam ; hayes , john , 1996 , physical review letters , * 76 * , 352 blair , w. p. , sankrit , r. , & raymond , j. c. , 2005 , aj , * 129 * , 2268 borkowski , k. j. , lyerly , w. j. , & reynolds , s. p. 2001 , apj , * 548
* , 820 cartledge , s. i. b. , lauroesch , j. t. , meyer , d. m. , & sofia , u. j. 2004 , apj , 613 , 1037 dopita , m. a. , mathewson , d. s. , and ford , v. l. 1977 , apj , * 214 * , 179 fujita , y. , sarazin , c. l. , reiprich , t. h. , andernach , h. , ehle , m. , murgia , m. , rudnich , l. , & slee , o. b. 2004 , apj , * 616 * , 157 ghavamian , p. , raymond , john . , smith , r. c. , & hartigan , p. 2001 , apj , * 547
* , l995 hatsukade , i. , & tsunemi , h. 1990 , apj , * 362 * , 566 hester , j. j. , raymond , j. c. , & blair , w. p. 1994 , apj , * 420
* , 721 inoue , h. , koyama , k. , matsuoka , m. , ohashi , t. , tanaka , y. , tsunemi , h. 1980 apj , * 238 * 886 iwamoto , brachwitz , f. , nomoto , k. , kishimoto , n. , umeda , h. , hix , w. r. , & thielemann , f. k. , 1999 , apjs , * 125 * , 439 kahn , s. m. , charles , p. a. , bowyer , s. , blissett , r. j. 1980 , apj , * 242 * , l19 katsuda , s. , tsunemi , h. , uchida , h. , miyata , e. , nemes , n. , miller , e. d. , and hughes , j. p. 2007
, pasj in press kirsch , m. 2004 , epic status of calibration and data analysis , xmm - soc - cal - tn-0018 , issue 2.3 kirsch , m. 2006 , xmm - epic status of calibration and data analysis , xmm - soc - cal - tn-0018 , issue 2.5 leahy , d. a. 2004 , mnras , * 351 * , 385 leonard , d. c. , filippenko , a. v. , ganeshalingam , m. , serduke , f. j. d. , li , w. , swift , b. j. , gal - yam , a. , foley , r. j. , fox , d. b. , park , s. , hoffman , j. l. , and wong , d. s. , 2006 , nature , * 440 * , 505 .
levenson , n. a. , et al .
, 1997 , apj , * 484 * , 304 .
levenson , n. a. , graham , j. r. , keller , l. d. , richter , m. j. 1998 , apjs , * 118 * , 541 levenson , n. a. , graham , j. r. , & snowden , s. l. 1999 , apj , * 526 * , 874 .
* 438 * , l115 mccray , r. & snow , t. p. , jr . 1979 , ara&a , * 17 * , 213 .
minkowski , r. , rev .
mod . phys .
1958 , * 30 * , 1048 .
mitsuda , k. et al . , 2007 ,
pasj , * 59s * , 1 .
miyata , e. , tsunemi , h. , pisarki , r. , and kissel , s. e. 1994 , pasj , * 46 * , l101 miyata , e. , tsunemi , h. , kohmura , t. , suzuki , s. , and kumagai , s. 1998 , pasj , * 50 * , 257 miyata , e. , & tsunemi , h. 1999 , apj , * 525 * , 305 miyata , e. , katsuda , s. , tsunemi , h. , hughes , j. p. , kokubun , m. , & porter , f. s. 2007 , pasj , * 59s * , 163 .
morrison , r. & mccammon , d. 1983 , apj , * 270 * , 119 motohara , k. , maeda , keiichi , g. , christopher l. , nomoto , k. , tanaka , m. , tominaga , n. , ohkubo , t. , mazzali , p. a. , fesen , r. a. , hoflich , p. , wheeler , j. c. 2006 , apj .
* 652 * l101 nevalainen , j. , lieu , r. , bonamente , m. , & lumb , d. 2003 , apj , * 584 * , l716 petre , r. , becker , c. m. , & winkler , p. f. 1996 , apj , * 456 * , l43 rauscher , t. , heger , a. , hoffman , r. d. , woosley , s. e. 2002 , apj , * 576 * , 323 read , a. , m. , & ponman , t. j. 2003 , a&a , * 409 * , 395 sato , k. , furusho , t. , yamasaki , y. , ishida m. , matsushida , k. , and ohashi , t. 2005 , pasj , * 57 * , 743 thielemann , f .- k . ,
nomoto , k. , & hashimoto , m. , 1996 , apj , * 460 * , 408 tominaga , n. , umeda , h. , nomoto , k. in prep .
uchida , h. katsuda , s. , miyata , e. , tsunemi , h. , hughes , j. p. , kokubun , m. , & porter , f. s. 2006 , suzaku conference in kyoto wilms , j. , allen , a. , and mccray , r. 2000 , apj , * 542 * , 914 winkler , p. f. & kirshner , r. p. 1985 ,
apj , * 299 * , 981 winkler , p. f. , tuttle , j. h. , kirshner , r. p. , & irwin , m. , j. 1988 , in iau colloq . 101 : supernova remnants and the interstellar medium , ed .
r. s. roger & t. l. landecker , 65-+ woosley , s. e. & weaver , t. a. 1995 , apjs , * 101 * , 181 .
the best - fit curves are shown with solid lines and the lower panels show the residuals .
right : same as left but for the sector at r=@xmath27 .
both the ejecta and cavity components are shown only for mos1 spectrum as dashed lines . ]
component estimated for the entire loop ( black solid line ) .
dash - dotted red lines represent cdd1 and w7 model of type - ia @xcite .
dotted green , blue , light blue , and magenta lines represent core - collapse models whose progenitor masses are 13 , 15 , 20 , and 25 , respectively @xcite . ] | the cygnus loop was observed from the northeast to the southwest with xmm - newton .
we divided the observed region into two parts , the north path and the south path , and studied the x - ray spectra along two paths .
the spectra can be well fitted either by a one - component non - equilibrium ionization ( nei ) model or by a two - component nei model .
the rim regions can be well fitted by a one - component model with relatively low @xmath0 whose metal abundances are sub - solar ( 0.10.2 ) .
the major part of the paths requires a two - component model . due to projection effects
, we concluded that the low @xmath0 ( @xmath10.2kev ) component surrounds the high @xmath0 ( @xmath10.6kev ) component , with the latter having relatively high metal abundances ( @xmath15 times solar ) .
since the cygnus loop is thought to originate in a cavity explosion , the low @xmath0 component originates from the cavity wall while the high @xmath0 component originates from the ejecta .
the flux of the cavity wall component shows a large variation along our path .
we found it to be very thin in the south - west region , suggesting a blowout along our line of sight .
the metal distribution inside the ejecta shows non - uniformity , depending on the element .
o , ne and mg are relatively more abundant in the outer region while si , s and fe are concentrated in the inner region , with all metals showing strong asymmetry .
this observational evidence implies an asymmetric explosion of the progenitor star .
the abundance of the ejecta also indicates the progenitor star to be about 15 . |
a non trivial problem that appears in several applications , e.g. in multivariate regression and bayesian statistics , is the estimation of the mean of a truncated normal distribution .
the problem arises in cases where a random vector @xmath0^t$ ] follows a normal distribution with mean @xmath1 and covariance matrix @xmath2 , denoted by @xmath3 , but @xmath4 is restricted to a closed subset @xmath5 of @xmath6 .
however , the restrictions considered in most of the cases are of the form @xmath7 , with @xmath8 , @xmath9 . if @xmath10 and @xmath11 ( @xmath12 and @xmath13 ) , we have a _
one - sided truncation _ from the left ( right ) , while in the case where both @xmath14 and @xmath15 are reals , we have _ two - sided truncation_. the problem has attracted the attention of several researchers from the sixties . since then , several deterministic approaches have been proposed with some of them trying to estimate not only the mean but also other moments ( such as the variance ) of a multivariate truncated normal distribution .
these approaches can be categorized according to whether they are suitable for one - sided truncated normal distributions ( @xcite , @xcite , @xcite , @xcite , @xcite ) or for two - sided truncated normal distributions ( @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ) , or according to whether they consider the bivariate , @xmath16 , ( @xcite , @xcite , @xcite , @xcite ) or the multivariate case ( @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite ) .
in addition , some of these methods put additional restrictions to the distribution paramaters ( e.g. @xcite , @xcite , @xcite require that @xmath17 ) . most of these methods either perform direct integration ( e.g. @xcite ) or they utilize the moment generating function tool ( e.g. @xcite ) . an alternative strategy to deal with this problem has been followed in @xcite .
more specifically , in @xcite , a markov chain monte carlo ( mcmc ) iterative scheme has been developed . according to this , at each iteration , @xmath18 succesive samplings
take place , one from each one - dimensional conditionals of the truncated normal distribution .
the mean of the @xmath19-th such distribution is the mean of @xmath20 conditioned on the @xmath21 . after performing several iterations , the estimation of the mean of the truncated normal results by performing an averaging over the produced samples .
convergence issues of this scheme to the mean of the truncated normal distribution are a subject of the markov chain theory .
a relative work that accelerates the method in @xcite is exhibited in @xcite .
the work presented in this paper for approximating the mean of a multivariate truncated normal distribution has been inspired from that of @xcite .
specifically , instead of selecting a sample from each one of the above one - dimensional distributions , we select its mean .
thus , the proposed scheme departs from the statistical framework adopted in @xcite and moves to the deterministic one .
this work is an extension of a relative scheme used in @xcite in the framework of spectral unmixing in hyperspectral images .
in addition , a convergence proof of the proposed scheme is given when certain conditions are fulfilled .
the quality of the approximation of the mean offered by the proposed method is assessed via the case where @xmath2 is the exponential correlation matrix .
experimental results show that the new scheme converges significantly faster than the mcmc approach of @xcite .
the paper is organized as follows .
section 2 contains some necessary definitions and a brief description of the work in @xcite . in section 3 ,
the newly proposed method is described and in section 4 conditions are given under which it is proved to converge . in section 5 , the proposed method is applied to the case where @xmath2 is the exponential correlation matrix . in section 6
simulation results are provided and a relevant discussion is presented .
finally , section 7 concludes the paper .
let us consider the @xmath18-dimensional normal distribution @xmath22 where the @xmath18-dimensional vector @xmath1 is its mean and the @xmath23 matrix @xmath2 is its covariance matrix .
let @xmath5 be a subset of @xmath6 with positive lebesgue measure .
we denote by @xmath24 the truncated normal distribution which results from the truncation of @xmath25 in @xmath5 . speaking in mathematical terms @xmath26
note that @xmath24 is _ proportional _ to @xmath27 , where @xmath28 , if @xmath29 and @xmath30 , otherwise . in the scheme discussed in @xcite , a markov chain monte carlo ( mcmc ) method
is proposed , to compute the mean of single or doubly truncated ( per coordinate ) normal distribution @xmath24 , where @xmath31 \times [ a_2,b_2 ] \times \ldots \times [ a_n , b_n]$ ] .
the method relies on the sampling of the @xmath18 one - dimensional conditionals of the truncated normal distribution .
more specifically , letting @xmath32 @xmath33 $ ] and @xmath34 denoting the expectation and the variance of @xmath20 conditioned on the rest coordinates , respectively follows the ( non - truncated ) @xmath19-th conditional of @xmath25 . ] , and @xmath35}(\mu_i^*,$ ] @xmath36 denoting the ( one dimensional ) truncated normal distribution which results from the truncation of a normal distribution with mean @xmath37 and variance @xmath34 in @xmath38 $ ] , the iterative sampling scheme proposed in @xcite can be written as @xmath39}(e[x_1|x_2^{(t-1)},\ldots , x_n^{(t-1)}],\sigma_1^{*2 } ) \\ 2 . & x_2^{(t ) } \sim { \cal n}_{[a_2,b_2]}(e[x_2|x_1^{(t)},x_3^{(t-1)}\ldots , x_n^{(t-1)}],\sigma_2^{*2 } ) \\ \vdots & \\ n. & x_n^{(t ) } \sim { \cal n}_{[a_n , b_n]}(e[x_n|x_1^{(t)},x_2^{(t)}\ldots , x_{n-1}^{(t)}],\sigma_n^{*2 } ) \end{array}\ ] ] where @xmath40 denotes the sampling action and @xmath41 denotes the current iteration . after performing several ,
say @xmath42 , iterations ( and after discarding the first few , say @xmath43 , ones ) the mean of each coordinate is estimated as @xmath44=\frac{1}{k - k ' } \sum_{t = k'}^k x_i^{(t ) } , \ \
i=1,\ldots , n\ ] ] the quantities @xmath37 and @xmath45 of each one of the above one dimensional conditionals are expressed in terms of the parameters @xmath1 and @xmath2 of the non - truncated multidimensional normal distribution as follows @xmath46 @xmath47 with @xmath48 being the @xmath49 matrix that results from @xmath2 after removing its @xmath19-th column and its @xmath19-th row , @xmath50 being the @xmath19-th column of @xmath2 excluding its @xmath19-th element and @xmath51 , @xmath52 being the ( @xmath53-dimensional ) vectors that result from @xmath4 and @xmath1 , respectively , after removing their @xmath19-th coordinates , @xmath20 and @xmath54 , respectively . note that @xmath37 depends on all @xmath55 s _ except _ @xmath20 .
in the sequel , we focus on the case where @xmath5 is a set of the form @xmath56 \times [ a_2,b_2 ] \times \ldots \times [ a_n , b_n]$ ] , where for each interval @xmath38 $ ] it is either , ( i ) @xmath57 and @xmath58 or ( ii ) @xmath59 and @xmath60 .
this means that , along each dimension , the truncation is one - sided and more specifically , case ( i ) corresponds to right truncation while case ( ii ) corresponds to left truncation .
the proposed model for estimating the mean of @xmath61 @xmath62 , is of iterative nature and , at each iteration , it requires the computation of the ( one - dimensional ) @xmath63 function .
this model has a close conceptual affinity with the one ( briefly ) presented in the previous section ( @xcite ) .
more specifically , instead of utilizing the samples produced by the ( one dimensional ) distributions @xmath35}(e[x_i|x_1^{(t)},x_2^{(t ) } \ldots , x_{i-1}^{(t ) } , x_{i+1}^{(t-1)},\ldots,$ ] @xmath64,\sigma_i^{*2})$ ] , we use the corresponding mean values ( here denoted by @xmath65 @xmath66^t$ ] ) . as it is well known ( see e.g. @xcite ) , the mean @xmath67 of a truncated one dimensional normal distribution @xmath68 } ( \mu^*,\sigma^{*2})$ ] , which has resulted from the ( non - truncated ) normal distribution with mean @xmath69 and variance @xmath70 , is expressed as @xmath71 where @xmath72 , @xmath73 and @xmath74 . however , since in the present paper we consider the cases where either ( i ) @xmath59 and @xmath60 or ( ii ) @xmath57 and @xmath58 , let us see now how eq .
( [ truncmean1d ] ) becomes for each of these cases .
\(i ) @xmath75 , @xmath76 . in this case
it is @xmath77 and as a consequence , @xmath78 , @xmath79 .
thus , taking also into account the definitions of @xmath80 and @xmath81 and the fact that @xmath82 , eq .
( [ truncmean1d ] ) gives ,
@xmath83 \(ii ) @xmath84 , @xmath85 . in this case
it is @xmath86 and , as a consequence , @xmath87 , @xmath88 .
working as in case ( i ) , eq .
( [ truncmean1d ] ) gives @xmath89 eqs .
( [ case - i ] ) and ( [ case - ii ] ) can be expressed compactly via the following single equation @xmath90 \sigma^*\ ] ] where @xmath91 ( @xmath92 ) is an indicator function which equals to @xmath93 if @xmath94 ( @xmath95 ) and @xmath30 otherwise and @xmath96 let us now return to the multidimensional case . since from the ( truncated ) conditional one - dimensional normals we no longer perform sampling but , instead , we consider their means , eq . ( [ trunc - mean ] ) is altered to @xmath97 where @xmath98 results from the current estimate of the ( @xmath18 - dimensional ) mean vector @xmath65 of the truncated normal distribution , after removing its @xmath19-th coordinate .
putting now all the previous ingredients together ( that is , utilizing eqs .
( [ cases - i - ii ] ) , ( [ trunc - mean2 ] ) and ( [ trunc - var ] ) ) we obtain the following iterative scheme @xmath99 \sigma_1^ * \\ 2 . & w_2^{(t ) } = \mu_2^{*(t ) } + \sqrt{\frac{2}{\pi } } \left [ f(\frac{\mu_2^{*(t)}-a_2}{\sqrt{2 } \sigma_2^ * } ) i_{a_2 \in { \cal r } } - f(\frac{b_2-\mu_2^{*(t)}}{\sqrt{2 } \sigma_2^ * } ) i_{b_2 \in { \cal r } } \right ] \sigma_2^ * \\ \vdots & \\ n. & w_n^{(t ) } = \mu_n^{*(t ) } + \sqrt{\frac{2}{\pi } } \left [ f(\frac{\mu_n^{*(t)}-a_n}{\sqrt{2 } \sigma_n^ * } ) i_{a_n \in { \cal r } } - f(\frac{b_n-\mu_n^{*(t)}}{\sqrt{2 } \sigma_n^ * } ) i_{b_n \in { \cal r } } \right ] \sigma_n^ * \end{array}\ ] ] with @xmath100 being computed via eq .
( [ trunc - mean2 ] ) , where @xmath101 ( the only parameter in ( [ trunc - mean2 ] ) that varies through iterations ) is defined as @xmath102^t$ ] , that is , the most recent information about @xmath103 s is utilized .
more formally , we can say that the above scheme performs sequential updating and ( following the terminology used in @xcite ) it is a _ gauss seidel _ updating scheme .
it is reminded that , due to the type of truncation considered here ( only left truncation or only right truncation per coordinate ) , the bracketed expression in each equation of ( [ new ] ) contains _ only one _ non identically equal to zero term . in the sequel , we consider separately the cases corresponding to @xmath10 and @xmath13 , i.e. , @xmath104 and @xmath105 where @xmath106 and @xmath107 @xmath9 , respectively , and @xmath108 is defined as in eq .
( [ f ] ) .
in this section we provide sufficient conditions under which the proposed scheme is proved to converge . before we proceed , we give some propositions and remind some concepts that will be proved useful in the sequel . _
proposition 1 : _ assume that @xmath2 is a symmetric positive definite @xmath23 matrix , @xmath50 is the @xmath19-th column of @xmath2 , after removing its @xmath19-th element and @xmath48 results from @xmath2 after removing its @xmath19-th row and its @xmath19-th column .
also , let @xmath109 be the @xmath110 element of @xmath111 and @xmath112 be the @xmath53-dimensional vector resulting from the @xmath19-th row of @xmath111 after ( i ) removing its @xmath19-th element , @xmath109 , and ( ii ) multiplying the remaining elements by @xmath113 .
then , it holds \(i ) @xmath114 and \(ii ) @xmath115 .
the proof of this proposition is straightforward from the inversion lemma for block partitioned matrices ( @xcite , p. 53 ) and the use of permutation matrices , in order to define the schur complement @xmath116 for each row @xmath19 of @xmath2 .
_ proposition 2 : _ it is @xmath117 where @xmath118 denotes the derivative of @xmath119 , which is defined in eq .
( [ f ] ) .
the proof of proposition 2 is given in the appendix .
_ definition 1 : _ a mapping @xmath120 , where @xmath121 , is called _ contraction _ if for some norm @xmath122 there exists some constant @xmath123 ( called _ modulus _ ) such that @xmath124 is called _ contracting iteration_. _ proposition 3 ( @xcite , pp .
182 - 183 ) : _ suppose that @xmath125 is a contraction with modulus @xmath123 and that @xmath126 is a closed subset of @xmath6 .
then \(a ) the mapping @xmath127 has a unique fixed point @xmath128 . is called _ fixed point _ of a mapping @xmath127 if it is @xmath129 . ]
\(b ) for every initial vector @xmath130 , the sequence @xmath131 , generated by @xmath132 converges to @xmath133 geometrically .
in particular , @xmath134 let us define the mappings @xmath135 , @xmath9 as @xmath136 \sigma_i^ * \label{t - i}\end{aligned}\ ] ] where @xmath137 and @xmath138 and the mapping @xmath127 as @xmath139 let us define next the mapping @xmath140 as @xmath141 performing the sequential updating as described by eq .
( [ new ] ) ( one at a time and in increasing order ) is equivalent to applying the mapping @xmath142 , defined as @xmath143 where @xmath144 denotes function composition . following the terminology given in @xcite ,
@xmath145 is called _ the gauss seidel mapping based on the mapping @xmath127 _ and the iteration @xmath146 is called _ the gauss seidel algorithm based on mapping @xmath127_. a direct consequence of ( * ? ? ?
1.4 , pp.186 ) is the following proposition : _ proposition 4 : _ if @xmath125 is a contraction with respect to the @xmath147 norm , then the gauss - seidel mapping @xmath145 is also a contraction ( with respect to the @xmath147 norm ) , with the same modulus as @xmath127 . in particular ,
if @xmath126 is closed , the sequence of the vectors generated by the gauss - seidel algorithm based on the mapping @xmath127 converges to the unique fixed point of @xmath127 geometrically . having given all the necessary definitions and results
, we will proceed by proving that ( a ) for each mapping @xmath148 it holds @xmath149 , @xmath150 , where @xmath151 , @xmath152 are the @xmath153 and @xmath147 norms , respectively , ( b ) if @xmath111 is diagonally dominant then @xmath127 is a contraction and ( c ) provided that @xmath127 is a contraction , the algorithm @xmath146 converges geometrically to the unique fixed point of @xmath127 .
we remind here that the @xmath53-dimensional vector @xmath154 results from the @xmath19-th row of @xmath111 , exluding its @xmath19-th element @xmath109 and dividing each element by the negative of @xmath109 .
_ proposition 5 : _ for the mappings @xmath148 , @xmath9 , it holds @xmath155 _ proof : _ ( a ) we consider first the case where @xmath10 .
let us consider the vectors @xmath156 . since @xmath1 is constant , utilizing eq .
( [ mu - i ] ) it follows that @xmath157 also , it is @xmath158 taking the difference @xmath159 we have @xmath160 since @xmath161 is continuous in @xmath162 , the mean value theorem guarantees that there exists @xmath163 @xmath164 $ ] such that @xmath165 substituting eq .
( [ mvt ] ) to ( [ diff - t ] ) we get @xmath166 substituting ( [ diff - mu - i ] ) and ( [ diff - a - i ] ) into ( [ diff - t1 ] ) and after some manipulations it follows that @xmath167 taking absolute values in eq .
( [ diff - t2 ] ) and applying hlder s inequality @xmath168 , for @xmath169 and @xmath170 , it follows that @xmath171 ( from proposition 2 ) , and the ( trivial ) fact that @xmath172 , it follows that latexmath:[\[\label{norm - t } thus , the claim has been proved .
\(b ) we consider now the case where @xmath13 .
working similarly to the previous case , the difference @xmath174 is @xmath175 while the difference @xmath159 is expressed as @xmath176 utilizing the mean value theorem we have that there exists @xmath177 @xmath178 $ ] such that @xmath179 substituting eq .
( [ mvt1 ] ) to ( [ diff - t - b ] ) we get @xmath180 substituting eqs .
( [ diff - mu - i ] ) and ( [ diff - b - i ] ) into ( [ diff - t1-b ] ) , we obtain @xmath181 from this point on , the proof is exactly the same with that of ( a ) .
_ proposition 6 : _ the mapping @xmath127 is a contraction in @xmath6 , with respect to the @xmath147 norm , provided that @xmath111 is diagonally dominant . _ proof : _ let @xmath182 .
taking into account proposition 5 , it easily follows that @xmath183 @xmath184 now , ( a ) taking into account that the @xmath53-dimensional vector @xmath154 results from the @xmath19-th row of @xmath111 , exluding its @xmath19-th element @xmath109 and dividing each element by the negative of @xmath109 and ( b ) recalling that @xmath185 is the @xmath19-th row of @xmath111 excluding its @xmath19-th element @xmath109 , it is @xmath186 provided that @xmath111 is diagonally dominant , it is @xmath187 or @xmath188 which proves the claim .
_ theorem 1 : _ the algorithm @xmath146 converges geometrically to the unique fixed point of @xmath127 , provided that @xmath111 is diagonally dominant .
_ proof : _ the proof is a direct consequence of the propositions 3 , 4 and 6 exposed before , applied for @xmath189 .
an issue that naturally arises with the proposed method is how accurate the estimate of the mean is . since it is very difficult to give a theoretical analysis of this issue , mainly due to the highly complex nature of the propsoed iterative scheme ( see eq .
( [ new ] ) ) , we will try to gain some insight for this subject via experimentation . to this end
, we set @xmath2 equal to the _ exponential correlation _ matrix , which is frequently met in various fields of applications , e.g. , in signal processing applications .
its general form is @xmath190 , \ \ ( 0 \leq \rho < 1)\ ] ] it is easy to verify that the inverse of @xmath191 is expressed as @xmath192\ ] ] also , it is straightforward to see that @xmath193 is diagonally dominant for all values of @xmath194 . thus , it is a suitable candidate for our case .
in addition , it is `` controlled '' by just a single parameter ( @xmath195 ) , which facilitates the extraction of conclusions .
note that for @xmath196 , @xmath193 becomes the identity matrix , while as @xmath195 increases towards @xmath93 the diagonal dominancy of @xmath193 decreases ( while its condition number increases ) . for @xmath195 close to @xmath93 , @xmath193 is alomost singular . in the sequel
, we consider the case of a zero mean normal distribution with covariance matrix as in ( [ exp - corr-1 ] ) , which is truncated in the region @xmath197^n$ ] , that is the truncation point is the same along all dimensions . without loss of generality , this choice has been deliberately selected , in order to keep our experimental framework controlled by just two parameters , @xmath195 and @xmath198 .
performing simulations for various combinations of the values of @xmath195 and @xmath198 ( and for various dimensions @xmath18 ) we can gain some insight on the accuracy of the approximation of the mean provided by the proposed method . in the sequel , we use as benchmark the estimate of the mean provided by the ( widely accepted as reliable ) mcmc method ( @xcite ) .
figure 1 , shows a three - dimensional graph of the difference ( assessed by its euclidean norm divided by @xmath18 ) between the estimates of the truncated mean obtained by the mcmc and the proposed methods , against @xmath195 and @xmath198 .
it is worth noting that for smaller values of @xmath195 ( less than @xmath199 ) , the difference remains low ( less than @xmath200 ) , independently of the value of the cutting point @xmath198 . on the other hand , for larger values of @xmath195 ( above @xmath199 ) ,
the difference increases .
more specifically , it increases more for values of @xmath198 between ( approximately ) @xmath201 and @xmath202 . in figure 2 , the shaded regions in the @xmath203 space correspond to low difference ( less than @xmath200 per dimension ) , for @xmath204 .
it can be deduced that the behaviour of the proposed method is only slightly affected influenced by the dimensionality of the feature space .
as a general conclusion , one can argue that the `` more diagonally dominant '' the @xmath193 is ( i.e. , the smaller the @xmath195 is ) , the more accurate the estimate of the mean provided by the proposed method is . from another perspective , the more @xmath191 `` approaches diagonality '' ( again , as @xmath195 becomes smaller ) ,
the more accurate the obtained estimates are .
the latter is also supported by the fact that in the extreme case of a diagonal covariance matrix , one has to solve @xmath18 independent one - dimensional problems , for which an analytic formula exists . in this case , it is easy to verify that the proposed method terminates after a single iteration ( see also comments in the next section ) .
after having gained some insight on the capability of the proposed method to approximate the mean of a multivariate truncated normal distribution in the previous section , we proceed in this section with experiments where now the involved covariance matrices have no specific structure . as in the previous section , the mcmc method is used as benchmark . _
1st experiment : _ the purpose of this experiment is to compare the estimates of the mean of a truncated normal distribution obtained by the proposed and the mcmc methods , for dimensions @xmath18 varying from @xmath205 to @xmath206 . to this end
, for each dimension @xmath18 , @xmath207 different truncated normal distributions ( defined by the means and the covariance matrices of the corresponding untruncated normals , as well as their truncation points ) have randomly been generated , such that , the corresponding inverse covariance matrix is diagonally dominant . for the @xmath19-th such distribution , @xmath208 , both the proposed and the mcmc methods have been applied . letting @xmath209 and @xmath210 denote the respective resulting estimates ,
the mean difference per coordinate between the two estimates is computed , i.e. , @xmath211 and , averaging over @xmath19 , the quantity @xmath212 is obtained . in figure 3 , @xmath213 is plotted versus @xmath18 . from this figure , it can be concluded that the proposed scheme gives estimates that are very close to those given by the mcmc .
thus , the fixed point of the proposed scheme ( when the diagonal dominance condition is fulfilled ) can be considered as a reliable estimate of the mean of the truncated normal distribution .
next , in order to show the rapid convergence of the proposed scheme against the mcmc method , we focus on a single example ( however , the resulting conclusions are generally applicable ) . more specifically , table 1 shows the values of the @xmath153 norm of the difference @xmath214 divided by @xmath18 , i.e. the quantity @xmath215 , as the number of iterations evolves , for the @xmath216-dimensional left truncated normal distribution defined by @xmath217 @xmath218^t$ ] , @xmath219=,@xmath219 while its truncation point is @xmath220 @xmath221^t$ ] .
it is clear that the proposed method converges rapidly to its estimate , while mcmc exhibits rather slow converge .
this is the strong point of the proposed method compared to the mcmc ( keeping , however , in mind the requirement for diagonal dominance of the inverse covariance matrix , for the proposed method ) .
.the evolution of @xmath222 for the mcmc and the proposed methods , for the first few iterations . [ cols=">,>,>",options="header " , ] in the sequel we try to get an indication about the performance of the proposed method when the diagonal dominance requirement does not hold .
the following two experiments are in this direction . _
2nd experiment : _ we consider the three - dimensional case where @xmath223^t$ ] and @xmath219=,@xmath219 while the truncation point is @xmath224^t$ ] note that , although in this case the inverse of the covariance matrix , which is @xmath219 ^ -1=,@xmath219 is not diagonally dominant , the diagonal dominance condition is slightly violated ( only in the third row ) .
the estimate of the mean of the truncated normal distribution obtained by the proposed method is @xmath225^t$ ] , which is quite close to the estimate obtained by the mcmc ( the mean absolute difference per coordinate is @xmath226 ) .
it is worth noting that in this case , although eq .
( [ norm - t ] ) is not satisfied ( @xmath227 , since @xmath111 is not diagonally dominant ) , the quantities @xmath228 , @xmath9 , are less than @xmath93 for all pairs of points @xmath229 is through @xmath230 . ] .
thus , the ( tighter ) bound of eq .
( [ norm - t1 ] ) is satisfied for all the pairs @xmath229 , which guarantees the rapid convergence of the method .
however , note that , since @xmath228 depends on the data ( through @xmath230 ) , it can not be used as an upper bound in the proof of the contraction . _
3rd experiment : _ we consider now the case where @xmath231^t$ ] and @xmath219=,@xmath219 while the truncation point is @xmath232^t$ ] . the matrix @xmath219 ^ -1=,@xmath219 is non - diagonally dominant and , in addition , the diagonal dominance condition is strongly violated ( in the last two rows ) .
we run the proposed method for several different initial conditions .
it is noted that , in contrast to the 2nd experiment , the quantity @xmath233 is now greater than @xmath93 for all pairs of @xmath229 .
the algorithm in all cases converges to the vector @xmath234^t$ ] .
however , in this case , the mean absolute difference per coordinate between this estimate and the one resulting from mcmc is @xmath235 , which is significantly greater than that in the previous experiment . _
4th experiment : _ in order to exhibit the scalability properties of the proposed method with respect to the dimensionality , an additional experiment has been conducted for @xmath236 .
the mean , and the covariance matrix of the corresponding normal distribution , as well as the truncation points have been selected as in the 1st experiment ( note that the inverse of the covariance matrix is diagonally dominant ) .
the algorithm gives its response in less than one minute , a time that is substantially smaller than that required from the mcmc method .
analysing the previous results we may draw the following conclusions : * provided that the inverse of the covariance matrix of the untrucated normal distribution is diagonally dominant , the proposed method gives very accurate estimates of the mean of the truncated normal distribution , using as benchmark the estimates of the mcmc method . *
the proposed method converges much faster ( in very few iterations ) compared to the mcmc method . *
even if the diagonal dominance constraint is slightly violated , the algorithm seems to converge to an accurate estimate of the mean of the truncated distribution , as the 2nd experiment indicates .
* when the diagonal dominance condition is strongly violated , the algorithm still ( seems to ) converge to a vector .
however , this vector is a ( much ) less accurate estimation of the mean of the truncated normal ( see the third experiment ) .
* in the special case , where @xmath2 ( and , as a consequence @xmath111 ) is diagonal , the matrix @xmath237 is ( obviously ) equal to zero .
thus , both eqs .
( [ trunc - mean ] ) and ( [ trunc - mean2 ] ) implies that @xmath238 and eq .
( [ trunc - var ] ) imply that @xmath239 . in other words , in this case ,
both the proposed and the mcmc methods solve @xmath18 independent one - dimensional problems , with the deterministic method giving its estimate in a single iteration , since in this case , it reduces to ( [ case - i ] ) or ( [ case - ii ] ) .
an additional observation is that the estimates provided by the two methods in this case are almost identical . letting intuition enter into the scene and generalizing a bit
, one could claim that as @xmath2 approaches `` diagonality '' , the estimates of the mean of both methods are expected to be even closer to each other .
this observation , may be an explanation of the decreasing trend observed in figure 1 , since , as the dimension increases , @xmath2 moves closer to `` diagonality '' , due to the diagonal dominance condition .
* finally , the proposed method scales well with the dimensionality of the problem .
in this paper , a new iterative algorithmic scheme is proposed for the approximation of the mean value of a one - sided truncated multivariate normal distribution .
the algorithm converges in very few iterations and , as a consequence , it is much faster than the mcmc based algorithm proposed in @xcite .
in addition , the algorithm is an extension of the one used in @xcite .
the quality of the approximation of the mean is assessed through the case where the exponential correlation matrix is used as covariance matrix .
the proof of convergence of the proposed scheme is provided for the case where @xmath111 is diagonally dominant . however , experimental results indicate that , even if this condition is softly violated , the method still provides estimates of the mean of the truncated normal , however , less accurate .
finally , the method exhibits good scalablity properties with respect to the dimensinality .
_ proof of proposition 2 : _ in order to prove ( [ q3 ] )
it suffices to show that @xmath263 , or , @xmath264 since in this case the sign of the second degree polynomial will be the opposite of that of the coefficient of @xmath265 ( i.e. , @xmath266 ) .
we proceed again be considering separately the cases @xmath245 and @xmath254 .
\(i ) let @xmath267 .
we set @xmath251 . in this case
( [ prove ] ) becomes @xmath268 where @xmath269 .
taking into account ( [ ineq ] ) , the right hand side inequality of ( [ prove1 ] ) holds . in order to prove the left hand inequlity of ( [ prove1 ] ) ( again taking into account ( [ ineq ] ) ) , it suffices to prove that \(ii ) let @xmath245 .
the left hand side inequality of ( [ prove ] ) holds trivially , since in this case @xmath273 .
we focus now on the right hand side inequality of ( [ prove ] ) . taking into account that ( a ) @xmath274 , for @xmath275 ( see e.g. @xcite ) and ( b ) @xmath82 , it is @xmath276 combining the right hand side inequality of ( [ ineq ] ) with ( [ q4 ] )
, the right hand side inequality of ( [ prove ] ) holds if @xmath277 or @xmath278 or @xmath279 since @xmath280 , for @xmath281 ( from the taylor series expansion ) , the previous inequality holds if @xmath282 or @xmath283 since @xmath284 , the previous inequality holds if @xmath285 the discriminant of the above second degree polynomial of @xmath286 is @xmath287 .
thus , the above second degree polynomial is always positive since the coefficient of the term @xmath288 is positive . in other words , ( [ q5 ] ) holds .
therefore , the right hand side inequality of ( [ prove ] ) also holds .
thus , for @xmath245 , it is also @xmath263 and therefore @xmath272 . as a consequence ( [ q3 ] ) also holds
. q.e.d .
m.chiani , d.dardari , m.k .
simon , `` new exponential bounds and approximations for the computation of error probability in fading channels '' , _ ieee transactions on wireless communications _ , 2(4 ) , 840 - 845 ( 2003 ) .
r. nabben , r.s .
varga , `` a linear algebra proof that the inverse of strictly ultrametric matrix is a strictly diagonally dominant stieltjes matrix '' , _ siam journal of matrix anal .
_ , 15 , 107 - 113 ( 1994 ) .
k. themelis , a. rontogiannis , k. koutroumbas , `` a novel hierarchical bayesian approach for sparse semisupervised hyperspectral unmixing '' , ieee transactions on signal processing , 60(2 ) , 585 - 599 ( 2012 ) . | a non trivial problem that arises in several applications is the estimation of the mean of a truncated normal distribution . in this paper , an iterative deterministic scheme for approximating this mean
is proposed , motivated by an iterative markov chain monte carlo ( mcmc ) scheme that addresses the same problem .
conditions are provided under which it is proved that the scheme converges to a unique fixed point .
the quality of the approximation obtained by the proposed scheme is assessed through the case where the exponential correlation matrix is used as covariance matrix of the initial ( non truncated ) normal distribution . finally , the theoretical results are also supported by computer simulations , which show the rapid convergence of the method to a solution vector that ( under certain conditions ) is very close to the mean of the truncated normal distribution under study . _
keywords : _ truncated normal distribution , contraction mapping , diagonally dominant matrix , mcmc methods , exponential correlation matrix |
the paper deals with a system of coupled oscillators .
this system has a large number of applications . in the framework of classical theory
it can be used e.g. for describing oscillations of electrons and atoms under forced combinative scattering , see @xcite-@xcite . for instance consider isotropic medium and denote by @xmath1 the normal coordinate of oscillations of electrons in atoms and by @xmath2 the normal coordinate of oscillations of atoms in molecules .
the frequency of oscillations @xmath1 corresponds to the ultraviolet scale range while the frequency of oscillations @xmath2 lies in the ultrared scale range . to a linear approximation
the oscillations of electrons and atoms occur independently of each other . in the general case on taking third order terms into account the expression for the potential energy
has the form @xmath3 where @xmath4 and @xmath5 are constraint elasticity coefficients in a molecule , the coefficient @xmath6 determines the nonlinearity of quasielasticity of oscillations @xmath2 , @xmath7 determines the parametric stimulation of oscillations @xmath2 by means of electron oscillations , @xmath8 determines the nonlinearity of electron oscillations which are responsible for the generation of optical harmonics .
the coefficient @xmath9 corresponds to the nonlinear interaction of electrons and nucleus and it determines the process of combinative scattering
. here we will consider the process of forced combinative scattering which is related to the presence of antiharmonic terms in the expression for the potential energy @xmath10 .
we assume in the sequel that the constants @xmath4 , @xmath5 , @xmath6 , @xmath7 , @xmath8 are essentially less than @xmath9 . under the assumptions that both @xmath1 and @xmath2 are small we get the system @xmath11 see @xcite . here @xmath12 and @xmath13 are reduced masses of electronic and atomic oscillators , @xmath14 and @xmath15
are eigen oscillation frequences , @xmath9 is the constant that describes the nonlinear interaction of electrons and nucleus , @xmath16 is the electron charge , @xmath17 stands for the electric field of light waves , and @xmath18 , @xmath19 are dissipation coefficients . on assuming the atomic dissipation to be essentially less that the electronic dissipation
, we will actually neglect @xmath19 in the sequel .
change the dependent variables in system ( [ sysakh ] ) by @xmath20 and @xmath21 .
on taking the radiation damping in an electronic oscillator into account , see @xcite , @xcite , we arrive at the system @xmath22 where @xmath23 is a small positive parameter , @xmath24 , the quantity @xmath25 is periodic in @xmath26 , and @xmath27 , @xmath28 , @xmath29 , @xmath30 are positive constants
. other examples of systems of coupled oscillators concern the description of propagation of surface and interior gravitational waves , see @xcite , @xcite .
the simplest analogue of system ( [ cauposcil ] ) is a nonlinear oscillator with resonance pumping .
the behaviour of such an oscillator is characterised at large times by beginning of permanent periodic nonlinear oscillations of the envelope whose amplitude is large when compared with the pumping .
this process is described by the primary resonance equation , see for instance @xcite , @xcite .
the system ( [ cauposcil ] ) belongs to the other class of resonantly perturbed problems .
the pumping fails here to occur directly , i.e. by including the perturbation of eigen frequency to the right - hand side of the equation , but rather by means of nonlinear interaction of oscillators . in this situation
the mathieu equation proves to be of crucial role . in the paper
an analogue of the nonlinear resonance equation for the envelope of resonant component is derived in the case of special nonlinear coupling in system ( [ cauposcil ] ) . in the initial stage
the solution amplitude growths linearly , see fig .
[ fig1a ] .
this growth corresponds to the linear resonance in the second equation of the system .
the envelope is well approximated by the straight line @xmath31 . for large values of @xmath32
the nonlinear effects of interaction between oscillators become essential .
-component of solution of ( [ cauposcil ] ) in the initial stage .
the solution amplitude is well approximated by the straight line @xmath31.,width=377,height=226 ] in fig .
the results of numerical simulation of the behaviour of system ( [ cauposcil ] ) of weakly coupled oscillators are shown under zero initial data .
the parameter values are @xmath33 , @xmath34 , @xmath35 , @xmath36 and @xmath37 numerical simulations are realized by the adams method . ) with @xmath38 and @xmath37 .
one can see pulsations of the @xmath39-component of solution of ( [ cauposcil ] ) . in the initial stage
the solution amplitude is well approximated by the straight line @xmath31 .
for large values of @xmath32 the nonlinear effects of interaction between oscillators become essential.,width=377,height=226 ] the main result of the paper consists in describing dynamics of the envelope of oscillations for large times .
it turns actually out that the envelope oscillates as well .
we show formulas for the amplitude and period of these oscillations .
the paper consists of 5 sections . in section [ result ]
we formulate the results obtained . in section [ construction ]
an asymptotic solution to the initial system ( [ cauposcil ] ) is constructed . in section [ lambdabig ]
we study in detail the physically interesting case where the frequency quotient @xmath40 is large enough . in this case
explicit formulas for the amplitude and period of oscillations are given .
in the last section [ s.conclusion ] we shortly comment on the results .
the asymptotic solution of system ( [ cauposcil ] ) for the large time @xmath41 has the form @xmath42 for @xmath43 , where @xmath44 and @xmath45 @xmath46 standing for the real part of a complex number @xmath47 .
the function @xmath48 satisfies the inhomogeneous mathieu equation @xmath49 where @xmath50 write @xmath51 , where @xmath52 and @xmath53 are the real and imaginary parts of the function @xmath54 , respectively .
they are solutions of the averaged system @xmath55 where @xmath56 . in the case
@xmath57 which is important for applications the last system simplifies and reduces to the hamiltonian system with hamiltonian of form @xmath58 and @xmath59 @xmath60 the maximal value of the amplitude @xmath32 for @xmath61 and for @xmath41 with @xmath43 is evaluated by @xmath62 where @xmath63 and @xmath64 the period is in turn determined by the formula @xmath65 the integration being over the cycle @xmath66 in the complex plane of the variable @xmath67 given by @xmath68 .
in this section we write down the problems for determining the coefficients of asymptotic solutions to system ( [ cauposcil ] ) . for convenience ,
we introduce a new independent variable @xmath70 .
on designating @xmath71 we get @xmath72 the solution of this system is constructed by the two scales method ( see @xcite ) in the form @xmath73 where @xmath74 is a slow variable .
let us substitute ( [ secondex ] ) into system ( [ cauposcil1 ] ) and group the coefficients of the same powers of the small parameter @xmath75 .
the equation for @xmath76 takes obviously the form @xmath77 the general solution of this equation is of the form @xmath78 with arbitrary function @xmath79 to be chosen later on .
the main term @xmath48 is determined from the equation @xmath80 which can be rewritten in the form @xmath81 with @xmath82 cf .
( [ nonhommathieu ] ) .
let @xmath83 , @xmath84 be a fundamental system of solutions of the homogeneous equation corresponding to ( [ nonhommathieum ] ) .
then any solution of ( [ nonhommathieum ] ) can be represented in the form @xmath85 the denominator being the wronsky determinant of the linearly independent system @xmath86 . the change of dependent variables @xmath87 reduces the homogeneous equation corresponding to ( [ nonhommathieum ] ) @xmath88 to the mathieu equation with slowly varying coefficient @xmath89 .
more precisely , we get @xmath90 where @xmath91 and @xmath92 .
it is known that the general solution of the mathieu equation changes within a period as @xmath93 where @xmath94 are characteristic indices of the ( homogeneous ) mathieu equation , see @xcite . for @xmath95 and @xmath89
such that both the indices are purely imaginary , the solution of the mathieu equation is bounded .
if either of the characteristic indices is real and positive then the solution has exponential growth .
depending on the values @xmath95 and @xmath89 the solution can grow exponentially or remain bounded , see @xcite . in fig .
[ fig2 ] the dependence upon the characteristic index @xmath96 on the parameters @xmath97 and @xmath98 is shown .
of the mathieu equation ( [ maa ] ) upon parameters @xmath97 and @xmath98 .
this figure is obtained by numerical simulations for different values of parameters , where @xmath99 $ ] and @xmath100 $ ] .
the grid step for numerical simulations equals 0.01 for both parameters .
, width=377,height=226 ] .
the dashed line is related to the dissipation in equation ( [ hommathieu ] ) .
the coefficient @xmath101 of dissipation equals 0.1 for numerical simulations.,width=377,height=226 ] a section of the surface @xmath102 for a fixed value @xmath97 and a line corresponding to dissipation are presented in fig .
[ fig21 ] , where @xmath103 , @xmath104 . in the case under consideration the coefficient @xmath105 changes
slowly . in each @xmath106-interval of the variable
@xmath107 the multiplicator of the solution of ( [ hommathieu ] ) has the form @xmath108 consider the properties of solutions of ( [ hommathieu ] ) for the fixed value @xmath109 , see fig .
[ fig21 ] . for @xmath110
the multiplicators are less than @xmath111 .
hence it follows that the modulus of the solution decreases on a sequence of intervals . if @xmath112 then the solution growths exponentially on a sequence of intervals . in fig .
[ fig22 ] the integral index @xmath113 is shown depending on time .
the solution of equation ( [ hommathieu ] ) does not depend in the main on the initial data as long as @xmath114 . upon @xmath115.,width=377,height=226 ] for equation ( [ nonhommathieu ] ) this means that when studying the general solution one can restrict oneself to the particular solution of zero initial data to the inhomogeneous equation and neglect the solution of the homogeneous equation as long as the condition @xmath114 is fulfilled .
the correction @xmath116 is determined from the differential equation @xmath117 we look for a solution @xmath116 of the form @xmath118 @xmath119 satisfying the equation @xmath120 if @xmath95 and @xmath89 are in the domain where the characteristic indices of the mathieu equation are purely imaginary and @xmath121 then @xmath122 for @xmath123 .
the averaging over the fast variable on the right - hand side of ( [ antisec ] ) determines the derivative of @xmath54 as function of slow time .
the evolution of @xmath48 over the fast variable is determined by the inhomogeneous equation ( [ nonhommathieu ] ) . the dependence of @xmath48 upon the slow variable is explained by the slow perturbation of the coefficients of the mathieu equation .
the explicit dependence of @xmath48 upon the slow variable has not so far been defined .
however , one can verify ( [ antisec ] ) by means of numerical simulation of the initial system ( [ cauposcil ] ) .
the numerical solution of system ( [ cauposcil ] ) allows one to evaluate separately the left - hand side and the right - hand side of ( [ antisec ] ) and to determine the relative residual .
we construct a numerical solution for the parameter values @xmath33 , @xmath124 , @xmath35 , @xmath36 and @xmath37 .
the system ( [ cauposcil ] ) is solved by the runge - kutta method of @xmath125th order . as result
we get a numerical solution @xmath126 , @xmath127 .
we then divide the entire integration interval into subintervals of length @xmath128 and on each subinterval we compute the fourier coefficients of @xmath129 and @xmath130 for the function @xmath131 . from the data obtained in this way we evaluate the difference derivative @xmath132 .
the averaging operator on the right - hand side of ( [ antisec ] ) is written in the form convenient for analytical computations . however , in numerical data the dependencies upon fast and slow variables are not separated from each other .
hence , it is not possible to directly apply formula ( [ antisec ] ) .
instead the averaging operator over the fast variable of ( [ antisec ] ) is replaced by the averaging over the interval of length @xmath133 with centre at the point @xmath134 . in this way
we get the values of the right - hand side at the points @xmath135 . in fig .
[ fig8 ] one compares the derivatives in @xmath136 on the left - hand side of equation ( [ antisec ] ) , which are evaluated numerically , and the integral on the right - hand side . of equation ( [ antisec ] ) .
here @xmath137 and @xmath138 are numerical evaluations of the left - hand and right - hand sides of ( [ antisec ] ) , respectively.,width=302,height=226 ] we thus conclude that the substitution of the numerical solution for the genuine solution of ( [ antisec ] ) leads to an inessential residual .
in this section we carry out the analysis of the behaviour of the main terms of representation ( [ secondex ] ) under the assumption that @xmath57 .
this assumption corresponds to the case of forced combinative scattering , where @xmath139 attains the value @xmath140 , see @xcite , @xcite .
equation ( [ nonhommathieu ] ) for the main term @xmath48 can be obviously rewritten in the form @xmath141 the asymptotics of the particular solution for large values @xmath139 is @xmath142 our next objective is to treat equation ( [ antisec ] ) for @xmath63 . to this end
we substitute ( [ asymath ] ) into equation ( [ antisec ] ) and evaluate the integral explicitly . as result we get @xmath143 where @xmath144 .
this is a hamiltonian system with hamiltonian @xmath145 the system ( [ anti ] ) is easily seen to have a stable equilibrium point of type ` centre ' @xmath146 see fig .
[ fig10 ] . ) in a neighborhood of ` centre ' @xmath147 .
this figure is obtained numerically with diverse initial data for system ( [ anti]).,width=302,height=226 ] the amplitude of oscillations can be found from the system @xmath148 in fig .
[ fig11 ] the dependence of the largest @xmath149 ( dashed curve ) and smallest @xmath150 ( solid curve ) values of solutions of ( [ large - small ] ) upon the hamiltonian is shown .
the vertical line corresponds to @xmath151 ( dashed ) and @xmath150 ( solid ) evaluated by ( [ rpm ] ) on @xmath152 .
the vertical line corresponds to the value @xmath153 .
an intersection of these lines allows one to estimate the maximal value of envelope amplitude.,width=302,height=226 ] this allows one to evaluate the maximal value of the amplitude of oscillation envelope through @xmath154 where @xmath155 the period is determined by @xmath156 where the integration is over the cycle @xmath66 in the complex plane of the variable @xmath67 given by @xmath157 .
the component @xmath32 is obtained by numerical simulations for the solution of system ( [ cauposcil ] ) .
the constant envelope function ( dashed curve ) equals @xmath158 which is related to the stationary solution @xmath147 of hamiltonian system ( [ anti]).,width=302,height=226 ] in fig .
[ fig12 ] one sees the results of numerical simulation for the solution of system ( [ cauposcil ] ) with initial data corresponding to the stationary solution @xmath159 the analytical result obtained in the paper agrees well to the numerical simulation . for the values of parameters @xmath160 and @xmath161 , which are used in the numerical simulation
, the hamiltonian value proves to be @xmath162 . in this case the amplitude of pulsations is evaluated by @xmath163 .
it is shown in fig .
the horizontal line corresponds to @xmath163 .
the explicit value for the period of envelope oscillations can be evaluated from ( [ period ] ) .
our numerical simulations with parameters @xmath164 give the value @xmath165 in the variable @xmath136 . in fig .
[ fig13 ] the pulsations of the @xmath39-component are shown as well as the envelope superposed on them .
the envelope is computed numerically as solution of system ( [ anti ] ) with @xmath166 , @xmath167 , @xmath168 .
all these numerical evaluations are realized by the adams method with the precision @xmath169
-component of the solution and the envelope .
the envelope function is evaluated as the solution of ( [ anti]).,width=302,height=226 ] system ( [ anti ] ) gives an approximation for determining the behaviour of functions @xmath160 and @xmath161 . in fig .
[ fig9 ] the solid curve corresponds to the envelope of the @xmath39-component of the solution .
this curve is found from a numerical solution of system ( [ cauposcil ] ) .
the dashed curve @xmath170 gives an approximation of the solution envelope as solution of system ( [ anti ] ) .
a point @xmath171 is an unstable node equilibrium point .
initial data for system ( [ anti ] ) correspond to the linear resonance of @xmath39-component of the solution for ( [ cauposcil ] ) in the initial stage , see fig .
[ fig1a ] . in fig .
[ fig9 ] the vertical line shows the value of the period @xmath165 .
it is determined numerically by expression ( [ period ] ) .
the horizontal line is an estimate of the maximal value @xmath172 of envelope amplitude .
-component of the solution ( solid curve ) and the approximation of the solution envelope as solution of system ( [ anti ] ) ( dashed curve ) .
the vertical line corresponds to the value of the period @xmath165 .
it is determined numerically by ( [ period ] ) .
the horizontal line is an estimate of maximal value @xmath172 of envelope amplitude.,width=302,height=226 ]
it is shown in the paper that for the large time @xmath0 with @xmath75 being a coupling parameter the behaviour in the main of a resonantly perturbed system of coupled oscillators is described by a hamiltonian system .
this system is studied in the physically interesting case when the quotient of oscillator frequencies @xmath173 is large enough .
certain pulsations prove to occur in the solution of the initial system of coupled oscillators .
the hamiltonian system obtained allows one to evaluate the period and amplitude of these pulsations .
* acknowledgements * the research was supported by the rfbr grant 09 - 01 - 92436-ke - a , the dfg grant ta 289/4 - 1 and by grant 2215.2008.1 for russian scientific schools .
the first author wishes to thank the daad and ministry of education and science of the rf for financial support in the framework of program `` mikhail lomonosov . '' | we consider a resonantly perturbed system of coupled nonlinear oscillators with small dissipation and outer periodic perturbation .
we show that for the large time @xmath0 one component of the system is described in the main by the inhomogeneous mathieu equation while the other component represents pulsation of large amplitude .
a hamiltonian system is obtained which describes the behaviour of the envelope in the main .
the analytic results agree to numerical simulations . |
the model of a particle bouncing off a circular constraint on a polygonal path was originally devised by newton to motivate the concept of centripetal force@xcite associated with circular motion of the particle .
it serves the precise purpose of highlighting the notion of a force continuously operating on the particle on a circle as the limiting case of motion on an inscribed polygon with the corners acting as force centers .
however , the intuitive simplicity of the circle case has contributed to the consistent ignoring of newton s original model .
recently the problem has received some attention @xcite mostly from the historical perspective .
no quantum mechanical treatment of the problem is available in the literature .
the quantum mechanics of the polygon model will be of much pedagogical value for introductory courses in quantum theory inasmuch as it is an illustration of the fact that the symmetry ( the n - fold discrete rotational symmetry in the present case ) of a problem is fully reflected in the wavefunction . in this work
we first reduce the lagrangian for the system by using the constraint equation and then use a suitable generalized coordinate to bring the hamiltonian to the free particle form . the resulting eigenvalues are shown to reduce to those of a particle on a circle in the limit as the number of sides @xmath0 and to those of a particle in an infinite potential well in the @xmath1 case .
we begin by constructing the full lagrangian for a particle of unit mass constrained to move on an @xmath2-sided regular polygon .
the @xmath2-gon can be parametrized by @xmath3 where @xmath4 labels the sides of the polygon . as shown in fig .
[ fig1 ] , @xmath5 is the length of the normal from the center of the circumcircle ( of radius @xmath6 ) to the side .
the length of a side is then given by @xmath7 . the lagrangian will be given by @xmath8 where @xmath9 is a lagrange multiplier implementing the constraint @xmath10 to reduce the lagrangian to the only relevant degree of freedom i.e. , @xmath11 , we employ the constraint directly to bring it to the form @xmath12 the momentum conjugate to the coordinate @xmath11 is @xmath13 finally , the classical reduced hamiltonian is given by @xmath14
a closer look at the hamiltonian eq .
[ clhamilt ] reveals that if we introduce a new generalized coordinate @xmath15 it reduces to the simple free particle form @xmath16 following the usual schrdinger prescription for quantization , the quantum canonical momentum is @xmath17 .
the hamiltonian is therefore the usual free particle schrodinger operator @xmath18 a more formal approach to the quantization of such a hamiltonian eq .
[ clhamilt ] would be to construct the laplace - beltrami operator in terms of the properly constructed quantum momentum operator @xcite .
we emphasize that the same quantum hamiltonian is obtained by this procedure also .
the free particle solutions are now given by the plane waves @xmath19 reverting back to the @xmath11 variable we have @xmath20 for the wavefunction in the @xmath21 side . the normalization is obtained from @xmath22 whence @xmath23 .
the boundary condition following from the singlevaluedness of the wavefunction @xmath24 leads to the quantization condition @xmath25 the energy eigenvalues are now given by @xmath26 we note that the above quantization conditions are also obtained by considering symmetric and antisymmetric solutions about the midpoint of a side , @xmath27 namely , @xmath28 and @xmath29 and imposing the periodic boundary conditions @xmath30 in fig .
[ fig2 ] we show the solutions for a hexagon ( @xmath31 ) .
the familiar example of a particle moving on a circle@xcite would correspond to the limit @xmath0 for the polygon model . in this limit , as each side of the polygon reduces to just a point , it is necessary to redefine the conjugate variables as @xmath32 for the angle variable and @xmath33 for the angular momentum in units of @xmath34 .
we see that the eigenvalues now become @xmath35 and the eigenfunctions reduce to the well known solutions @xmath36 for a free particle on a circle .
in the interesting case of @xmath1 , classically the polygon reduces to just a segment of length @xmath37 traversed in both directions as discussed very lucidly by anicin @xcite .
quantum mechanically this is equivalent to a particle confined in an infinite potential well of width @xmath37 . indeed , the eigenvalue equation eq .
[ energy ] reduces to the familiar eigenvalues @xmath38 note that in the above we have effected the replacement of wavevector @xmath39 by @xmath40 to take care of the two - fold reflection symmetry which is inherent in the reduction of the polygon to the segment of length @xmath37 .
we remark that since the parameterization eq .
[ rcoord ] of the polygon does not survive down to the @xmath1 case for obvious reasons , the eigenfunctions can not be expressed in terms of the variable @xmath41 or @xmath11 . the eigenvalue equation eq .
[ energy ] , nevertheless , is robust and gives the true eigenvalues .
we have quantized newton s polygon model and derived the general eigenvalues and eigenfunctions .
we have recovered the quantum mechanics in the circle limit of the polygon . moreover ,
the energy eigenvalues for the @xmath1 case have been identified with those of a particle in an infinite potential well . in the light of the above results it would be interesting exercise for students of introductory courses on quantum mechanics to look into the polygon analogues of the well studied problems of stark effect for rotor@xcite , aharonov - bohm effect for a particle on a circle etc .. we gratefully acknowledge fruitful discussions with v. v. shenoi , and j. kamila .
one of the authors ( rkp ) would like to thank the institute of physics , bhubaneswar , for hospitality where a part of the work was carried out . | we study the quantization of a model proposed by newton to explain centripetal force namely , that of a particle moving on a regular polygon .
the exact eigenvalues and eigenfunctions are obtained .
the quantum mechanics of a particle moving on a circle and in an infinite potential well are derived as limiting cases . 2 |
possible existence of exotic mesons consisting of two quarks and two anti - quarks was first suggested by jaffe in the framework of the mit bag model @xcite . since then , there have been continuous discussions on whether the mesons in the scalar nonet are candidates for such tetraquark mesons .
recently , interest in tetraquark mesons has been extended to include those containing heavy quarks @xcite , as several heavy mesons , that were observed in @xmath2 meson decays , do not seem to fit well within the conventional quark model @xcite .
tetraquark mesons with two heavy anti - quarks ( @xmath3 ) , henceforth called @xmath4 , are particularly interesting as they are explicitly exotic from flavor considerations @xcite . moreover
, simple theoretical consideration based on the color - spin interaction @xcite shows that for such configurations the binding energy increases as the mass of the heavy quark increases .
calculations based on the flavor - spin interaction @xcite or the instanton induced interactions @xcite also show that the mass of @xmath0 is below that of two charmed mesons . for a similar reason , the chance of having a stable heavy pentaquark ( @xmath5 ) increases as the mass of heavy anti - quark becomes larger .
experimental observation of such explicitly exotic hadrons is crucial in refining our understanding of multiquark interactions in low energy qcd .
however , producing the @xmath4 from an elementary process is highly suppressed as it involves creating two @xmath6 pairs from the vacuum .
in contrast , in relativistic heavy ion collisions at lhc , @xmath7 pairs are expected to be abundantly produced @xcite . since the hadronization from the quark - gluon plasma produced in these collisions tends to follow a statistical description , production of exotic hadrons in heavy ion collisions at lhc is thus much more favorable than in elementary reactions @xcite . in this work ,
we first give a qualitative argument why multiquark hadrons consisting of heavy quarks are likely to exist .
using both the quark coalescence model and the statistical hadronization model , we then give estimates of how many @xmath4 and charmed pentaquark baryons , if they exist , will be produced in central heavy ion collisions at both rhic and lhc . furthermore
, possible decay modes of these charmed exotic hadrons are discussed .
sophisticated constituent quark model calculations have been performed to study possible stable multiquark hadrons that consist of heavy quarks .
these results can be roughly understood in terms of simple arguments based on the color - spin interaction . to illustrate the mechanism , we introduce the following simplified form for the color - spin interaction @xcite : @xmath8 here @xmath9 and @xmath10 are the mass and spin of the constituent quarks @xmath11 and @xmath12 .
the strength of the color - spin interaction @xmath13 should depend on the wave function and the exact form of the interaction as well as the color structure of either the quark - quark or quark - antiquark pair .
the color factor would be @xmath14 for diquarks in the color antitriplet channel and @xmath15 for quark and anti - quark pair in the color singlet channel .
this simple form with @xmath16 for a diquark and @xmath17 for a quark - antiquark pair can capture some of the essential physics in hadron masses .
to illustrate this point , we assume the following constituent quark masses : @xmath18 , @xmath19 , @xmath20 , and @xmath21 .
.baryon mass relations .
the first column is fit to experiments . [ cols="^,^,^,^,^ " , ] [ table11 ]
lastly , we comment on the possibility to measure doubly charmed baryons in heavy ion collisions .
the doubly charmed baryon @xmath22 have been observed by the selex collaboration in the @xmath23 and in the @xmath24 decay modes with a mass of ( @xmath25 ) mev @xcite .
the same collaboration has also successfully measured @xmath26 in the @xmath27 decay mode with a mass of 3460 mev @xcite . on the other hand
, attempts by the focus collaboration in the photoproduction experiment , and by the babar in @xmath28 annihilation experiments @xcite have so far failed to establish the existence of the doubly charmed baryons .
hence , it is an interesting problem to search for the doubly charmed baryons in heavy ion collisions . using the coalescence model , we find that the number of @xmath26 produced are @xmath29 at rhic , and @xmath30 at lhc .
therefore , we will be able to realistically measure @xmath26 and @xmath22 through their decay vertices to @xmath23 and @xmath24 , and to @xmath27 , respectively .
based on the consideration of the color - spin interaction between diquarks , which describes reasonably the mass splittings between many hadrons and their spin flipped partners , we have shown that tetraquark mesons and pentaquark baryons that consist of two charmed quarks could be bound . using the quark coalescence model , their yields in heavy ion collisions at both rhic and lhc
are estimated . because of the expected large charm quark number in central pb+pb collisions at lhc , the abundances of the tetraquark meson @xmath0 and pentaquark baryon @xmath1 are about @xmath31 and @xmath32 , respectively .
we have also discussed their decay modes to illustrate how they can be identified in heavy ion collisions . in our studies , we have not taken into account the hadronic effect on the abundance of these charmed exotics , as hadronic reactions that affect their annihilation and production are unknown .
since the yields of @xmath0 and @xmath1 from the coalescence model is significantly smaller than those expected from the statistical hadronization model , including the hadronic effect is expected to increase their yields substantially and reduces the differences from the predictions from the quark coalescence model and the statistical hadronization model . also , charmed hadrons would be more abundantly produced , particularly the @xmath0 , if charm quarks are produced from the qgp formed in these collisions
. we also comment on the possible measurement of doubly charmed baryons in heavy ion collisions , and the estimated numbers are @xmath29 and @xmath30 at rhic and lhc , respectively .
we thus expect that the open and hidden charmed hadron physics will be an interesting subject in the forthcoming heavy ion collision experiments .
the work of shl and sy was supported by the korea research foundation krf-2006-c00011 , while that of cmk and wl was supported by the u.s .
national science foundation under grant no .
phy-0457265 and the welch foundation under grant no .
a-1358 .
r. l. jaffe , phys .
d * 15 * , 267 ( 1977 ) ; phys .
d * 17 * , 1444 ( 1978 ) .
s. zouzou , b. silvestre - brac , c. gignoux and j. m. richard , z. phys .
c * 30 * , 457 ( 1986 ) . c. semay and b. silvestre - brac , z. phys .
c * 61 * ( 1994 ) 271 .
s. l. zhu , arxiv : hep - ph/0703225 .
a. v. manohar and m. b. wise , nucl .
b * 399 * , 17 ( 1993 ) .
a. de rujula , h. georgi and s. l. glashow , phys .
d * 12 * , 147 ( 1975 ) .
l. y. glozman and d. o. riska , phys .
rept . * 268 * , 263 ( 1996 ) [ arxiv : hep - ph/9505422 ] .
s. pepin , f. stancu , m. genovese and j. m. richard , phys .
b * 393 * , 119 ( 1997 ) [ arxiv : hep - ph/9609348 ] .
d. m. brink and f. stancu , phys .
d * 57 * , 6778 ( 1998 ) .
s. chernyshev , m. a. nowak and i. zahed , phys . rev .
d * 53 * , 5176 ( 1996 ) [ arxiv : hep - ph/9510326 ] .
b. w. zhang , c. m. ko , and w. liu , nucl - th/0709.1684 . c. nonaka , b. muller , m. asakawa , s. a. bass and r. j. fries , phys
. rev .
c * 69 * , 031902 ( 2004 ) [ arxiv : nucl - th/0312081 ] .
l. maiani , a. d. polosa , v. riquer and c. a. salgado , phys .
b * 645 * , 138 ( 2007 ) [ arxiv : hep - ph/0606217 ] . b. silvestre - brac and c. semay , z. phys .
c * 59 * , 457 ( 1993 ) .
f. s. navarra , m. nielsen and s. h. lee , arxiv : hep - ph/0703071 .
t. nakano _ et al .
91 ( 2003 ) 012002 .
e. hiyama , m. kamimura , a. hosaka , h. toki and m. yahiro , phys .
b * 633 * , 237 ( 2006 ) .
f. stancu , phys .
b * 595 * , 269 ( 2004 ) [ erratum - ibid .
b * 598 * , 295 ( 2004 ) ] .
f. stancu , aip conf .
proc . * 775 * , 32 ( 2005 ) .
r. l. jaffe , f. wilczek , phys .
* 91 * ( 2003 ) 232003 .
k. maltman , phys .
b * 604 * , 175 ( 2004 ) .
y. sarac , h. kim , s. h. lee , phys .
d * 73 * , 014009 ( 2006 ) .
h. j. lipkin , phys .
b * 195 * , 484 ( 1987 ) .
e. m. aitala _ et al .
_ [ e791 collaboration ] , phys .
lett . * 81 * , 44 ( 1998 ) [ arxiv : hep - ex/9709013 ] .
e. m. aitala _ et al .
_ [ e791 collaboration ] , phys .
b * 448 * , 303 ( 1999 ) .
r. mattiello , a. jahns , h. srge , h. stcker , and w. greiner , phys .
lett . * 74 * , 2180 ( 1995 ) ; r. mattiello , h. srge , h. stcker , and w. greiner , phys .
c * 55 * , 1443 ( 1997 ) . v. greco , c. m. ko , and p. lvai , phys .
90 * , 202302 ( 2003 ) ; phys .
c * 68 * , 034904 ( 2003 ) ; v. greco , c. m. ko , r. rapp , phys . lett .
b595 , 202 ( 2004 ) arxiv : nucl - th/0312100 . l. w. chen and c. m. ko , phys .
c * 73 * , 044903 ( 2005 ) .
m. mattson _
[ selex collaboration ] , phys .
* 89 * , 112001 ( 2002 ) [ arxiv : hep - ex/0208014 ] .
a. ocherashvili _ et al .
_ [ selex collaboration ] , phys .
b * 628 * , 18 ( 2005 ) [ arxiv : hep - ex/0406033 ] .
j. s. russ [ selex collaboration ] , arxiv : hep - ex/0209075 . s. p. ratti , nucl .
suppl . * 115 * , 33 ( 2003 ) .
b. aubert _ et al .
_ [ babar collaboration ] , phys .
d * 74 * , 011103 ( 2006 ) [ arxiv : hep - ex/0605075 ] . | based on the color - spin interaction in diquarks , we argue that charmed multiquark hadrons are likely to exist .
because of the appreciable number of charm quarks produced in central nucleus - nucleus collisions at ultrarelativistic energies , production of charmed multiquark hadrons is expected to be enhanced in these collisions . using both the quark coalescence model and the statistical hadronization model
, we estimate the yield of charmed tetraquark meson @xmath0 and pentaquark baryon @xmath1 in heavy ion collisions at rhic and lhc .
we further discuss the decay modes of these charmed exotic hadrons in order to facilitate their detections in experiments . |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.