article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
rational algebraic curves and surfaces can be described in several different ways , the most common being parametric and implicit representations .
parametric representations describe the geometric object as the image of a rational map , whereas implicit representations describe it as the set of points verifying a certain algebraic condition , e.g. as the zeros of a polynomial equation .
both representations have a wide range of applications in computer aided geometric design ( cagd ) , and depending on the problem one needs to solve , one or the other might be better suited . to give a simple example , the parametric description is better for drawing a surface , as it allows to rapidly generate points on the surface , which can then be interpolated , whereas an implicit representation is better adapted for testing if a given point lies on the surface , since one only needs to check whether the point verifies the algebraic condition that defines the surface .
it is thus interesting to be able to pass from the parametric representation to the implicit equation .
this is a classical problem and there are numerous approaches to its solution , see @xcite and @xcite for good historical overviews .
however , it turns out that the implicitization problem is computationally difficult . a promising alternative suggested in @xcite is to compute a so - called _ matrix representation _ instead , which is easier to compute but still shares some of the advantages of the implicit equation .
here is the definition .
let @xmath4 be a hypersurface .
a matrix @xmath5 with entries in the polynomial ring @xmath6 $ ] is called a _ representation matrix _ of @xmath7 if it is generically of full rank and if the rank of @xmath5 evaluated in a point of @xmath8 drops if and only if the point lies on @xmath7 .
it follows immediately that a matrix @xmath5 represents @xmath7 if and only if the greatest common divisor of all its minors of maximal size is a power of the homogeneous implicit equation @xmath9 $ ] of @xmath7
. one major ingredient in the construction of such matrices are syzygies .
the theory of syzygies has been developed in the theoretical context of commutative algebra at the beginning of the 20th century by mathematicians such as david hilbert .
however , it was only in the 1990s that the cagd and geometric modeling community discovered that the concept of syzygies is useful in their field .
initially unaware of the connections to commutative algebra , @xcite , @xcite , @xcite , and numerous other authors labeled this approach the method of `` moving curves '' ( or `` moving surfaces '' ) and showed how it can be used to express the implicit equation as a determinant .
in the case of a planar rational curve @xmath10 given by a parametrization of the form @xmath11 , @xmath12 , where @xmath13 $ ] are coprime polynomials of degree @xmath14 and @xmath15 is a field , a linear syzygy ( or moving line ) is a linear relation on the polynomials @xmath16 , i.e. a linear form @xmath17 in the variables @xmath18 and with polynomial coefficients @xmath19 $ ] such that @xmath20 .
we denote by @xmath21 the set of all those linear syzygies forms and for any integer @xmath22 the graded part @xmath23 of syzygies of degree at most @xmath22 .
actually , to be precise , one should homogenize the @xmath24 with respect to a new variable and consider @xmath21 as a graded module here .
it is obvious that @xmath23 is a finite - dimensional @xmath15-vector space and one can easily obtain a basis @xmath25 by solving a linear system .
we define the matrix @xmath26 of coefficients of the @xmath27 with respect to a @xmath15-basis of @xmath28_\nu$ ] as @xmath29 that is , the coefficients of the syzygies @xmath27 form the columns of the matrix .
note that the entries of this matrix are linear forms in the variables @xmath30 with coefficients in the field @xmath15 .
let @xmath31 denote the homogeneous implicit equation of the curve and @xmath32 the degree of the parametrization as a rational map .
intuitively , @xmath32 measures how many times the curve is traced .
it is known that for @xmath33 , the matrix @xmath26 is a representation matrix ; more precisely : if @xmath34 , then @xmath26 is a square matrix , such that @xmath35 .
also , if @xmath36 , then @xmath26 is a non - square matrix with more columns than rows , such that the greatest common divisor of its minors of maximal size equals @xmath37 . in other words
, one can always represent the curve as a square matrix of linear syzygies . in principle
, one could now actually calculate the implicit equation .
however , it might be advantageous to avoid the costly determinant computation and work directly with the matrix instead , as it has the advantage of making the well - developed theory and tools of linear algebra applicable to solve geometric problems .
for instance , testing whether a point @xmath38 lies on the curve only requires computing the rank of @xmath26 evaluated in @xmath38 .
other interesting results using square matrix representations directly to solve geometric problems are presented , for example , in @xcite or @xcite , in which intersection problems are treated by means of eigenvalue techniques . it is a natural question whether this kind of matrix representation can be generalized to rational surfaces defined as the image of a map @xmath39 where @xmath40 $ ] are coprime polynomials of degree @xmath14 . in order to put the problem in the context of graded modules ,
one first has to consider an associated projective map @xmath41 where @xmath1 is a 2-dimensional projective toric variety ( for example @xmath2 or @xmath42 ) with coordinate ring @xmath43 and the @xmath44 are homogenized versions of their affine counterparts @xmath24 . in other words
, @xmath1 is a suitable compactification of the affine space @xmath45 @xcite . in this case ,
a linear syzygy ( or moving plane ) of the parametrization @xmath46 is a linear relation on the @xmath47 , i.e. a linear form @xmath48 in the variables @xmath49 with @xmath50 $ ] such that @xmath51 exactly in the same way as for curves , one can set up the matrix @xmath26 of coefficients of the syzygies in a certain degree @xmath22 , but unlike in the curve case , it is in general not possible to choose a degree @xmath22 such that @xmath26 is a square matrix representation of the surface . in recent years
, two main approaches have been proposed to deal with this problem : * one allows the use of quadratic syzygies ( or higher - order syzygies ) in addition to the linear syzygies in order to be able to construct square matrices .
* one only uses linear syzygies as in the curve case and obtains non - square representation matrices . the first approach using linear and quadratic syzygies ( or moving planes and quadrics ) has been treated in @xcite for base - point - free homogeneous parametrizations , i.e. @xmath52 , and @xcite does the same in the presence of base points . in @xcite ,
square matrix representations of bihomogeneous parametrizations , i.e. @xmath3 , are constructed with linear and quadratic syzygies , whereas @xcite gives such a construction for parametrizations over toric varieties of dimension 2 .
the methods using quadratic syzygies usually require additional conditions on the parametrization and the choice of the quadratic syzygies is often not canonical . the second approach ,
even though it does not produce square matrices , has certain advantages , in particular in the sparse setting that we present . in previous publications ,
this approach with linear syzygies , which relies on the use of the so - called approximation complexes has been developed in the case @xmath52 , see for example @xcite , @xcite , and @xcite , and in @xcite for bihomogeneous parametrizations of degree @xmath53 .
however , for a given affine parametrization @xmath54 , these two varieties are not necessarily the best choice of a compactification of affine space , since they do not always reflect well the combinatorial structure of the polynomials @xmath55 . in this paper
we will extend the method to a much larger class of varieties , namely toric varieties of dimension 2 , and we will see that this generalization allows us to choose a `` good '' toric compactification of @xmath45 depending on the polynomials @xmath55 , which makes the method applicable in cases where it failed over @xmath2 or @xmath42 and we will also see that it is significantly more efficient and leads to much smaller representation matrices . the main idea of our method is similar to the one in @xcite .
we use a ( general ) toric embedding to consider our domain as a 2-dimensional toric variety contained in a higher - dimensional projective space , which we present in section [ sec : setting ] .
contrary to the cited paper , this natural domain will not be in general a hypersurface and its coordinate ring will usually not be gorenstein , which means that we have to give new proofs for some of the results in which this property was used . in section [ approxt ]
we proceed to establish the necessary homological tools and in particular to derive bounds on local cohomology in theorem [ locosymalgt ] , our main technical result .
after that , we will see in section [ sec : equation ] that we can deduce the validity of the approach from previous results to produce an efficient representation matrix for the implicit equation ( see corollary [ cor : main ] ) .
the particular case of bihomogeneous parametrizations of any bidegree is illustrated in section [ segret ] .
we then show the advantages of our method through several examples in section [ sec : examples ] .
after some concluding remarks which summarize the scope of the paper in section [ sec : final ] , an implementation in macaulay2 @xcite for the important special case @xmath3 is included as an appendix .
let @xmath15 be a field .
all the varieties considered hereafter are understood to be taken over @xmath15 .
we suppose given a rational map @xmath56 where @xmath40 $ ] are polynomials .
we assume that * @xmath54 is a generically finite map onto its image and hence parametrizes an irreducible surface @xmath57 * @xmath58 , which means that there are only finitely many base points .
we briefly introduce some basic notions from toric geometry .
these constructions are investigated in more detail in ( * ? ? ?
2 ) , @xcite , and ( * ? ? ?
* ch . 5 & 6 ) .
let @xmath59}$ ] .
we define the support @xmath60 to be the set of all the exponents which appear in @xmath61 , i.e. @xmath62 the newton polytope @xmath63 , where @xmath64 , is defined as the convex hull of the union @xmath65 in @xmath66 of the supports of the @xmath24 . in other words , @xmath67 is the smallest convex lattice polygon in @xmath66 containing all the exponents appearing in one of the @xmath24 .
note that our hypothesis that @xmath54 is generically finite implies that @xmath67 is two - dimensional .
furthermore , let @xmath68 be the biggest integer such that @xmath67 equals @xmath69 where @xmath70 is a lattice polygon .
in other words , @xmath70 is the smallest possible homothety of @xmath67 with integer vertices .
then @xmath70 defines a two - dimensional projective toric variety @xmath71 , as explained in @xcite , where @xmath72 is the cardinality of @xmath73 .
it is defined as the closed image of the embedding @xmath74 where @xmath75 .
for example , the triangle between the points @xmath76 , @xmath77 , and @xmath78 corresponds to @xmath2 and @xmath42 has a rectangle as polygon .
the rational map @xmath54 factorizes through @xmath1 in the following way @xmath79 ^ -{f } \ar@{->}[d]^{\rho } & \pp^3 \\ \tc \ar@{-->}[ur]_g } \label{diagram2t}\ ] ] where @xmath46 is given by four polynomials @xmath80 of degree @xmath14 in @xmath81 variables .
thus , we have extended the affine parametrization @xmath54 to a parametrization @xmath46 of @xmath82 over the projective variety @xmath1 @xmath83 for which we will adapt the method of approximation complexes .
this map induces an application between the homogeneous coordinate rings @xmath84 & \xrightarrow{h } & a\\ t_i & \mapsto & g_i(x_0,\ldots , x_m)\end{aligned}\ ] ] where @xmath85 / i(\tc)$ ] is the homogeneous coordinate ring of @xmath1 , which is a domain , as @xmath86 is prime .
note that the variables @xmath87 correspond to monomials @xmath88 and the ideal @xmath86 is the ideal of relations between these monomials .
the implicit equation of @xmath82 is a generator of the principal ideal @xmath89 .
we should remark that the toric ideals @xmath86 are very well understood and there exist highly efficient software systems to compute their grbner bases , for example @xcite . instead of @xmath70
we could actually have chosen any polygon @xmath90 such that a multiple @xmath91 , @xmath68 , contains @xmath67 .
in particular , we could choose @xmath67 itself , in which case the @xmath92 will become linear forms , compare ( * ? ? ?
2 ) . we will see in section [ segret ] that @xmath70 is always a better choice than @xmath67 ; for the moment let us just state that a smaller polygon leads to a less complicated coordinate ring but to a higher degree of the @xmath92 and that the advantages of the former outweigh the inconveniences of the latter .
intuitively , the surface @xmath1 should be understood to be the smallest compactification of @xmath45 through which the map @xmath54 factorizes , so in a way it respects the geometry of the map best and is a natural candidate .
however , we will see in example [ interestingexample ] that in some cases there are better choices than the canonical choice @xmath70 .
we can describe the ring @xmath43 in a more combinatorial way , which will enable us to study its properties in more detail .
let @xmath93 be the cone generated by the polytope @xmath70 .
i.e. the rational cone over @xmath94
. then @xmath95 equals the union of @xmath96 for @xmath97 , where @xmath98 which means that at each height @xmath99 we have a homothety of @xmath70 by a factor of @xmath99 .
in particular , we identify @xmath100 with @xmath101 .
since we are dealing with polygons in dimension two , it holds that @xmath102 this property is called _
normality_. we should note that these considerations are no longer true in higher dimensions .
this is because in dimension @xmath103 there exist non - normal lattice polytopes ( * ? ? ?
in fact , the study of normality of smooth lattice polytopes is a subject of current research @xcite as an illustration , consider the following picture of the cone @xmath93 : now we can associate an affine semigroup ring @xmath104 $ ] to this cone : one takes the @xmath15-vector space freely generated by the elements of @xmath95 and equips it with a natural multiplication , which is induced by the addition of vectors in @xmath105 , see ( * ? ? ?
6 ) for more details .
it is actually a graded @xmath15-algebra , with the grading being induced by the height @xmath99 , i.e. by the decomposition @xmath106 .
if the variable @xmath87 in @xmath43 stands for the monomial @xmath88 , we can identify it with the point @xmath107 .
the multiplication of two monomials in @xmath43 corresponds to the addition of two vectors in @xmath93 .
it is easy to verify that the above correspondence extends to a graded isomorphism of @xmath15-algebras between @xmath43 and @xmath104 $ ] by observing that the relations of @xmath86 correspond to different decompositions of an element of @xmath96 as the sum of elements of smaller degree , so we actually have @xmath108\ ] ] exploiting this combinatorial description of the ring @xmath43 we can deduce some algebraic properties , but let us first recall the definition of the canonical module , a notion we will use in this section .
[ canmoddef ] let @xmath109 $ ] , @xmath110 an ideal of @xmath111 and suppose that @xmath112 is of dimension @xmath14
. then the canonical module of @xmath111 is defined as @xmath113 $ ] and @xmath114)\ ] ] is the canonical module @xmath115 of @xmath5 .
the ring @xmath43 is an affine normal semigroup ring by ( * ? ? ?
6.1.2 and 6.1.4 ) , since @xmath95 is a normal semigroup .
moreover , by ( * ? ?
6.3.5 ) it is cohen - macaulay and its canonical module @xmath116 is the ideal generated by the monomials that correspond to integer points in the interior of @xmath93 .
this shows that @xmath43 is gorenstein if and only if the first @xmath117 with non - empty interior ( either @xmath118 , @xmath119 , or @xmath120 ) contains exactly one point . in this case
, it is actually easy to see the isomorphism between @xmath116 and @xmath43 geometrically : it is nothing else than the translation that moves this point in the interior of @xmath117 to the origin .
note that in the previous works @xcite , @xcite , and @xcite , the ring @xmath43 was always gorenstein and this property was used in some of the proofs . in our context
we have to do without this property , which means that some of the proofs need to be modified .
we will quickly recall the construction of the approximation complex @xmath121 in order to fix notation , compare also @xcite , @xcite , and @xcite .
let us denote by @xmath122 the class of the variable in the homogeneous coordinate ring @xmath123/j$ ] of @xmath1 , where @xmath124 and @xmath125 stands for the sequence @xmath126 .
@xmath43 is a graded ring , each variable having weight 1 .
let @xmath127 be the ideal generated by the @xmath92 , recall that @xmath128 .
we consider the koszul complex @xmath129 associated to @xmath47 over @xmath43 @xmath130 \ar[r]^-{\delta_4 } & a[-3d]^4 \ar[r]^-{\delta_3 } & a[-2d]^6 \ar[r]^-{\delta_2 } & a[-d]^4 \ar[r]^-{\delta_1 } & a } \ ] ] where the differentials are matrices with @xmath131 as non - zero entries .
write @xmath132 , @xmath133 $ ] , @xmath134 $ ] , @xmath135 $ ] , and @xmath136 $ ] .
set @xmath137 , which says that @xmath138 also keeps the degree shift .
note that with this notation the sequence @xmath139 is an exact sequence of graded modules ( with morphisms of degree zero ) .
we set @xmath140 \otimes_a a[\underline{t}]$ ] , which we will consider as bigraded @xmath141$]-modules ( one grading is induced by the grading of @xmath43 , the other one comes from setting @xmath142 for all @xmath143 ) .
now the approximation complex of cycles @xmath144 , or simply @xmath121 , is the complex @xmath145 & \zc_3(-3 ) \ar[r]^-{\epsilon_3 } & \zc_2(-2 ) \ar[r]^-{\epsilon_2 } & \zc_1(-1 ) \ar[r]^-{\epsilon_1 } & \zc_0 } \ ] ] where the differentials @xmath146 are obtained by replacing @xmath92 by @xmath147 for all @xmath143 in the matrices of @xmath148 and where the degree shifts are with respect to the grading by the @xmath147 .
then @xmath149 is generated by the linear syzygies of the @xmath92 and @xmath150/\im(\epsilon_1 ) \simeq \sym_a(i)\ ] ] from now on , when we take the degree @xmath22 part of the approximation complex , denoted @xmath151 , it should always be understood to be taken with respect to the grading of @xmath43 .
hereafter we denote by @xmath152 the maximal ideal @xmath153 .
the geometric intuition behind the @xmath154-complex is quite profound , we only give some hints and refer to ( * ? ? ?
3 ) or @xcite for a more thorough treatment of the subject . the symmetric algebra is closely related to the rees algebra @xmath155 , which can be defined as the quotient of @xmath141 $ ] by all the syzygies ( not only the linear ones ) .
one has thus a canonical surjection from @xmath156 onto @xmath155 , which induces an inclusion @xmath157 now @xmath158 corresponds to the closure of the graph of the map @xmath46 and its image by the projection to @xmath0 equals the surface @xmath82 , while @xmath159 is a priori a bigger object .
however , @xmath156 is in some ways easier to study and under suitable conditions on the base points the inclusion in becomes an isomorphism and one can retrieve the information about @xmath82 contained in the rees algebra from the symmetric algebra .
more precisely , we will see that the implicit equation of @xmath82 can be obtained from the determinant of certain graded parts of the @xmath154-complex .
the next lemma shows that the complex @xmath160 is acyclic if the base points are local complete intersections and finite in number .
this is a standard hypothesis for syzygy - based implicitization methods , see @xcite .
[ zacyclicityt ] let @xmath127 .
suppose that @xmath161 has at most dimension @xmath162 and is locally a complete intersection , then the complex @xmath163 is acyclic .
if there are no base points this follows immediately from ( * ? ? ?
4.7 ) and for finitely many base points from ( * ? ? ?
we only have to check that the hypotheses of these propositions are verified : in our case , we have @xmath164 and we need to check that @xmath165 , which is true because @xmath43 is cohen - macaulay and because @xmath43 is the homogeneous coordinate ring of a ( projective ) surface .
moreover , in the presence of base points , the equality @xmath166 is again a consequence of the cohen - macaulayness of @xmath43 .
it can be shown in a similar way as in ( * ? ? ?
* lemma 1 ) that the @xmath154-complex is still acyclic if the base points are almost local complete intersections , but we will not treat this case here . the following lemma establishes a vanishing criterion on the local cohomology of @xmath156 , which ensures that the implicit equation can be obtained as a generator of the annihilator of the symmetric algebra in a certain degree .
we refer to @xcite for more details on local cohomology , a detailed treatment of which is beyond the scope of this work .
[ anniht ] suppose that @xmath167 has at most dimension @xmath162 and is locally a complete intersection .
if @xmath168 is an integer such that @xmath169 then we have @xmath170 } ( \sym_a(i)_\nu ) = \ann_{\kk[\underline{t } ] } ( \sym_a(i)_\eta ) = \ker(h)\ ] ] for all @xmath171 . the proof of ( *
* lemma 2 ) can be applied verbatim .
as we shall see , the annihilator in the above lemma can be computed as the determinant ( or macrae invariant ) of the complex @xmath172 , so we should give an explicit formula for the integer @xmath168 , but we first need to study the local cohomology of @xmath43 using its combinatorial structure as a semigroup ring .
the following definition is the same as ( * ? ? ?
* def . 11.15 ) .
let @xmath5 be a graded @xmath43-module .
the matlis dual @xmath173 of @xmath5 is the @xmath43-module defined by @xmath174 the multiplication being the transpose .
one has @xmath175 if all the graded parts @xmath176 of @xmath5 are finite - dimensional as @xmath15-vector spaces .
let @xmath5 be a finitely generated graded @xmath43-module of dimension @xmath177 .
then @xmath5 is cohen - macaulay if and only if @xmath178 for all @xmath179 and @xmath180 is the matlis dual to @xmath115 .
this is ( * ? ? ?
* th . 13.37 ) .
so the local cohomology of an @xmath43-module that is cohen - macaulay can be expressed in terms of its canonical module .
let us apply this to the @xmath43-module @xmath43 .
using that @xmath181 and that @xmath43 is cohen - macaulay we immediately deduce [ locoofa ] the local cohomology of @xmath43 is @xmath182 where @xmath183 is the matlis dual to the canonical module @xmath116 .
so the third local cohomology module of @xmath43 is the only one that is non - zero .
actually , we do not need to know this module exactly ; it is sufficient to know in which graded parts it vanishes .
[ canmod - propt ] let @xmath184 and let @xmath185 .
then we have @xmath186 if @xmath187 . by corollary [ locoofa ] and the definition of the matlis dual we have the identities @xmath188 but the module @xmath116
is generated by the elements in the interior of @xmath93 , i.e. by elements of degree at least @xmath189 , so whenever @xmath187 , it follows @xmath190 and the modules in the above equation are all zero .
we can now proceed to investigate the vanishing of the 0th local cohomology of the symmetric algebra .
the proof is similar to the corresponding theorems ( * ? ? ? * and 5.10 ) and ( * ? ? ?
* th . 1 ) .
we give two bounds , an explicit one , which always holds , and a lower but more complicated bound for the case when there are base points .
[ locosymalgt ] suppose that @xmath167 has at most dimension @xmath162 and is locally a complete intersection .
then @xmath191 where @xmath192 as before .
moreover , if there is at least one base point , one even has @xmath193 the proof is virtually the same for the two cases . as the first one has been proven in ( * ? ? ?
* th . 4.11 )
, we only give a proof for the second bound .
consider the two spectral sequences associated to the double complex @xmath194 , both converging to the hypercohomology of @xmath121 . by lemma [ zacyclicityt ] , @xmath121 is acyclic , hence the first spectral sequence stabilizes at step two with @xmath195 the second one has as first screen : @xmath196\otimes_a a[t_1,\ldots , t_4](-q).\ ] ] the comparison of the two spectral sequences shows that @xmath197 vanishes as soon as @xmath198 vanishes for all @xmath61 , in fact we have that @xmath199 where we denote @xmath200 . by corollary [ locoofa ] and
the fact that @xmath201 , @xmath202 .
recall that the sequence @xmath203 shown in is graded exact .
from , applied to @xmath204 ( writing @xmath205 ) we obtain the long exact sequences of local cohomology @xmath206 now @xmath207 , as @xmath110 is an ideal of an integral domain , by corollary [ locoofa ] we have @xmath208 , hence @xmath209 vanishes .
now as @xmath210 , the koszul complex is exact for @xmath211 , i.e. @xmath212 .
it is clear by construction of the koszul complex that @xmath213 and that @xmath214 $ ] . using that @xmath215 for @xmath187 by corollary [ canmod - propt ]
, we can deduce that @xmath216 if @xmath217 .
it follows that @xmath218 it remains to determine @xmath219 . from the short exact sequence @xmath220 we get the exact sequence @xmath221 hence , as @xmath222 by
, there is a surjective graded map @xmath223 .
moreover , setting @xmath224 , by ( * ? ? ?
* lemma 5.8 ) we have the graded isomorphism @xmath225\cong h^0_\mm(\omega_a / i.\omega_a)[4d]$ ] .
hence , we obtain @xmath226 ) \\ & \leq \endd(h^0_\mm(h_1))-2d \\ & = -\indeg(h^0_\mm(\omega_a / i.\omega_a)[4d])-2d \\ & = 2d -\indeg(h^0_\mm(\omega_a / i.\omega_a ) ) .
\end{array}\ ] ] we have shown that @xmath227 , hence @xmath197 vanishes as soon as @xmath228 .
[ naivebound ] clearly , the advantage of the bound @xmath229 is that it does not require the computation of @xmath230 , which can turn out to be difficult even in simple examples .
however , even though it might not be obvious at first sight , the second bound is lower .
for example , take the case studied in @xcite , i.e. @xmath70 is a unit square and @xmath43 is the quotient @xmath231/x_0x_3-x_1x_2 $ ] . by (
2 ) , we can identify @xmath232 $ ] , hence @xmath233)=2d-1-\indeg(i^\sat)$ ] , whereas the naive bound would be @xmath234 .
similarly , in the case @xmath52 , our bound coincides with the known bound @xmath235 from ( * ? ? ?
3.2 ) , as compared to @xmath236 .
also in the general case , one always has @xmath237 due to @xmath116 being generated in degree at least @xmath189 as explained in section [ combinatorialstructure ] , and obviously @xmath238 .
it can now be deduced that the determinant of the @xmath121-complex is a power of the implicit equation of @xmath82 . indeed , using lemma [ zacyclicityt ] , lemma [ anniht ] , and theorem [ locosymalgt ] , a completely analogous proof to ( * ? ? ?
5.2 ) shows the following .
[ maintht ] suppose that @xmath167 has at most dimension @xmath162 and is locally a complete intersection .
let @xmath184 as before and @xmath239 .
for any integer @xmath240 the determinant @xmath241 of the complex @xmath151 of @xmath242$]-modules defines ( up to multiplication with a constant ) the same non - zero element in @xmath242 $ ] and @xmath243 where @xmath31 is the implicit equation of @xmath82 . by theorem [ locosymalgt ]
, one can replace the bound in this result by the more precise bound @xmath244 if there is at least one base point .
as in @xcite or @xcite , there is a possible generalization of the above theorem to the case of almost local completion intersection base points .
however , the proofs of the corresponding results ( or the one of ( * ? ? ?
4 ) ) do not apply directly here , because they use at some points that @xmath43 is gorenstein , which is not necessarily the case in the toric setting . by (
* appendix a ) , the determinant @xmath241 can be computed either as an alternating sum of subdeterminants of the differentials in @xmath245 or as the greatest common divisor of the maximal - size minors of the matrix @xmath5 associated to the first map @xmath246 .
note that this matrix is nothing else than the matrix @xmath26 of linear syzygies as described in the introduction ; it can be computed with the same algorithm as in @xcite by solving the linear system given by the degree @xmath247 part of . as an immediate corollary
we deduce the following very simple translation of theorem [ maintht ] , which can be considered the main result of this paper .
[ cor : main ]
let @xmath248 be a parametrization of the surface @xmath57 given by @xmath249 with @xmath44 .
let @xmath26 be the matrix of linear syzygies of @xmath47 in degree @xmath250 , i.e. the matrix of coefficients of a @xmath15-basis of @xmath251 with respect to a @xmath15-basis of @xmath252 . if @xmath46 has only finitely many base points , which are local complete intersections , then @xmath26 is a representation matrix for the surface @xmath82 .
we should also remark that by ( * ? ? ?
1 ) ( or ( * ? ? ?
* appendix ) ) the degree of the surface @xmath82 can be expressed in terms of the area of the newton polytope and the hilbert - samuel multiplicities of the base points : @xmath253 where @xmath254 is twice the euclidean area of @xmath67 , i.e. the normalized area of the polygon . for locally complete intersections ,
the multiplicity @xmath255 of the base point @xmath256 is just the vector space dimension of the local quotient ring at @xmath256 .
bihomogeneous parametrizations , i.e. the case @xmath3 , are particularly important in practical applications , so we will now make explicit the most important constructions in that case and make some refinements .
we also include an implementation in macaulay2 @xcite in the appendix .
in this section , we consider a rational parametrization of a surface @xmath82 @xmath258 where the polynomials @xmath55 are bihomogeneous of bidegree @xmath259 with respect to the homogeneous variable pairs @xmath260 and @xmath261 , and @xmath262 are positive integers .
we make the same assumptions as in the general toric case .
let @xmath263 , @xmath264 , and @xmath265 .
so we assume that the newton polytope @xmath67 is a rectangle of length @xmath266 and width @xmath267 and @xmath70 is a rectangle of length @xmath268 and width @xmath269 ( in fact @xmath67 might be smaller , but in this section we homogenize with respect to the whole rectangle ) .
so @xmath270 can be embedded in @xmath271 , @xmath272 through the _ segre - veronese embedding _
@xmath274 we denote by @xmath1 its image , which is an irreducible surface in @xmath271 , whose ideal @xmath275 is generated by quadratic binomials .
we have the following commutative diagram . @xmath276 ^ -{f } \ar@{->}[d]^{\rho } & \pp^3
\\ \tc \ar@{-->}[ur]_g } \label{diagram1t}\ ] ] with @xmath277 , the @xmath92 being polynomials in the variables @xmath126 of degree @xmath14 .
we denote by @xmath85/j$ ] the homogeneous coordinate ring of @xmath1 .
we can give an alternative construction of the coordinate ring ; consider the @xmath278-graded @xmath15-algebra @xmath279_{n e_1 ' } \otimes_\kk \kk[t , v]_{n e_2 ' } \right ) \subset \kk[s , u , t , v]\ ] ] which is finitely generated by @xmath280 as an @xmath281-algebra .
then @xmath270 is the bihomogeneous spectrum @xmath282 of @xmath283 , since @xmath284_{n e_1'})=\proj(\bigoplus_{n\in \nn } \kk[t , v]_{n e_2'})=\pp^1 $ ] .
the segre - veronese embedding @xmath285 induces an isomorphism of @xmath278-graded @xmath15-algebras @xmath286 where @xmath287 for @xmath288 and @xmath289 and the implicit equation of @xmath82 can be obtained by the method of approximation complexes described in the previous sections as the kernel of the map @xmath290 & \rightarrow & a \\
t_i & \mapsto & g_i\end{aligned}\ ] ] the ring @xmath43 is an affine normal semigroup ring and it is cohen - macaulay .
it is gorenstein if and only if @xmath291 ( or equivalently @xmath292 ) , which is the case treated in @xcite . the ideal @xmath275 is easier to describe than in the general toric case ( compare @xcite for the case @xmath293 ) . the generators of @xmath275 can be described explicitly .
let @xmath294 then the ideal @xmath275 is generated by the @xmath295-minors of the @xmath296-matrix below built from the matrices @xmath297 : @xmath298 let us also state the degree formula for this setting , which is a direct corollary of : @xmath299 where as before @xmath255 is the multiplicity of the base point @xmath256 .
we have claimed before that it is better to choose the toric variety defined by @xmath70 instead of @xmath67 .
let us now give some explanations why this is the case .
as we have seen , a bihomogeneous parametrization of bidegree @xmath259 gives rise to the toric variety @xmath3 determined by a rectangle of length @xmath268 and width @xmath269 , where @xmath300 , @xmath263 , and whose coordinate ring can be described as @xmath279_{n e_1 ' } \otimes_\kk \kk[t , v]_{n e_2 ' } \right ) \subset \kk[s , u , t , v]\ ] ] instead of this embedding of @xmath42 we could equally choose the embedding defined by @xmath67 , i.e. a rectangle of length @xmath266 and width @xmath267 , in which case we obtain the following coordinate ring @xmath301_{n e_1 } \otimes_\kk \kk[t , v]_{n e_2 } \right ) \subset \kk[s , u , t , v]\ ] ] it is clear that this ring also defines @xmath42 and we obviously have an isomorphism @xmath302 between the graded parts of the two rings , which means that the grading of @xmath303 is coarser and contains less information .
it is easy to check that the above isomorphism induces an isomorphism between the corresponding graded parts of the approximation complexes @xmath121 corresponding to @xmath283 and @xmath304 corresponding to @xmath305 , namely @xmath306 if the optimal bound in theorem [ maintht ] for the complex @xmath154 is a multiple of @xmath14 , i.e. @xmath307 , then the optimal bound for @xmath308 is @xmath309 and we obtain isomorphic complexes in these degrees and the matrix sizes will be equal in both cases .
if not , the optimal bound @xmath310 is the smallest integer bigger than @xmath311 and in this case , the vector spaces in @xmath312 will be of higher dimension than their counterparts in @xmath313 and the matrices of the maps will be bigger .
an example of this is given in the next section .
we first treat some examples from @xcite .
example 10 in the cited paper , which could not be solved in a satisfactory manner in @xcite , is a surface parametrized by @xmath314 the newton polytope @xmath70 of this parametrization is * * we can compute the new parametrization over the associated variety , which is given by linear forms @xmath47 , i.e. @xmath315 ( since there is no smaller homothety @xmath70 of @xmath67 ) and the coordinate ring is @xmath316/j$ ] where @xmath275 is generated by @xmath317 binomials of degrees 2 and 3 . recall that the 9 variables correspond to the @xmath318 points in the newton polytope . in the optimal degree @xmath319 as in theorem [ locosymalgt ] , the implicit equation of degree @xmath320 of the surface @xmath82 is represented by a @xmath321-matrix , compared to a @xmath322-matrix with the toric resultant method ( from which a @xmath323-minor has to be computed ) and a @xmath324-matrix with the method of moving planes and quadrics . note also that this is a major improvement of the method in @xcite , where a @xmath325-matrix representation was computed for the same example .
example 11 of @xcite is similar to example 10 but an additional term is added , which transforms the point @xmath326 into a non - lci base point .
the parametrization is @xmath327 the newton polytope has not changed , so the embedding as a toric variety and the coordinate ring @xmath43 are the same as in the previous example . again
the new map is given by @xmath47 of degree 1 .
as in @xcite , the method represents ( with @xmath328 ) the implicit equation of degree @xmath320 times a linear extraneous factor caused by the non - lci base point . while the chow form method represents this polynomial as a @xmath329-minor of a @xmath330-matrix , our representation matrix is @xmath331 .
note that in this case , the method of moving lines and quadrics fails .
[ indegfalse ] in this example , we will see that if the ring @xmath43 is not gorenstein , the correction term for @xmath247 is different from @xmath332 , unlike in the homogeneous and the unmixed bihomogeneous cases .
consider the parametrization @xmath333 we will consider this as a bihomogeneous parametrization of bidegree @xmath334 , that is we will choose the embedding @xmath285 corresponding to a rectangle of length 2 and width 3 .
the actual newton polytope @xmath67 is smaller than the @xmath334-rectangle , but does not allow a smaller homothety .
one obtains @xmath335 /j$ ] , where @xmath275 is generated by 43 quadratic binomials and the associated @xmath92 are of degree @xmath336 .
it turns out that @xmath337 is the lowest degree such that the implicit equation of degree 46 is represented as determinant of @xmath313 , the matrix of the first map being of size @xmath338 .
so we can not compute @xmath247 as @xmath339 , as one might have been tempted to conjecture based on the results of the homogeneous case .
this is of course due to @xmath43 not being gorenstein , since the rectangle contains two interior points .
let us make a remark on the computation of the representation matrix .
it turns out that this is highly efficient .
even if we choose the non - optimal bound @xmath340 as given in theorem [ maintht ] , the computation of the @xmath341 representation matrix is computed instantaneously in macaulay2 .
just to give an idea of what happens if we take higher degrees : for @xmath342 a @xmath343-matrix is computed in about 30 seconds , and for @xmath344 we need slightly less than 5 minutes to compute a @xmath345 matrix . in any case
, the computation of the matrix is relatively cheap and the main interest in lowering the bound @xmath247 as much as possible is the reduction of the size of the matrix , not the time of its computation .
this reduction improves the performance of algorithmic applications of our approach , notably to decide whether a given point lies in the parametrized surface .
[ interestingexample ] in the previous example , we did not fully exploit the structure of @xmath67 and chose a bigger polygon for the embedding .
here is an example where this is necessary to represent the implicit equation without extraneous factors .
take @xmath346 .
this is a very sparse parametrization and we have @xmath347 .
the coordinate ring is @xmath348/j$ ] , where @xmath349 and the new base - point - free parametrization @xmath46 is given by @xmath350 .
the newton polytope looks as follows .
* * + @xmath67 * * + * * + for @xmath351 we can compute the matrix of the first map of @xmath352 , which is a @xmath353-matrix .
the greatest common divisor of the @xmath354-minors of this matrix is the homogeneous implicit equation of the surface ; it is of degree 6 in the variables @xmath49 : @xmath355 as in example [ indegfalse ] we could have considered the parametrization as a bihomogeneous map either of bidegree @xmath356 or of bidegree @xmath357 , i.e. we could have chosen the corresponding rectangles instead of @xmath67 .
this leads to more complicated coordinate rings ( @xmath358 resp .
@xmath359 variables and @xmath360 resp .
@xmath361 generators of @xmath275 ) and to bigger matrices ( of size @xmath362 in both cases ) .
even more importantly , the parametrizations will have a non - lci base point and the matrices do not represent the implicit equation but a multiple of it ( of degree @xmath318 ) . instead , if we consider the map as a homogeneous map of degree @xmath363 , the results are even worse : for @xmath364 , the @xmath365-matrix @xmath366 represents a multiple of the implicit equation of degree @xmath317 .
to sum up , in this example the toric version of the method of approximation complexes works well , whereas it fails over @xmath42 and @xmath2 .
this shows that the extension of the method to toric varieties really is a generalization and makes the method applicable to a larger class of parametrizations .
interestingly , we can even do better than with @xmath67 by choosing a smaller polytope .
the philosophy is that the choice of the optimal polytope is a compromise between two criteria : * the polytope should be as simple as possible in order to avoid that the ring @xmath43 becomes too complicated .
* the polytope should respect the sparseness of the parametrization ( i.e. be close to the newton polytope ) so that no base points appear which are not local complete intersections .
so let us repeat the same example with another polytope @xmath90 , which is small enough to reduce the size of the matrix but which only adds well - behaved ( i.e. local complete intersection ) base points : the newton polytope @xmath67 is contained in @xmath367 , so the parametrization will factor through the toric variety associated to @xmath90 , more precisely we obtain a new parametrization defined by @xmath368 over the coordinate ring @xmath369/j$ ] with @xmath370 .
the optimal bound is @xmath371 and in this degree the implicit equation is represented directly without extraneous factors by a @xmath372-matrix , which is smaller than the @xmath353 we had before . as we have seen , the size of the matrix representation depends on the given parametrization and as a preconditioning step it is often advantageous to choose a simpler parametrization of the same surface , if that is possible .
for example , approaches such as @xcite can be used to find a simpler reparametrization of the given surface and optimize the presented methods .
another important factor to consider is that all the methods we have seen represent the implicit equation to the power of the degree of the parametrization . on one hand ,
it can be seen as an advantage that this piece of geometric information is encoded in the matrix representation , but on the other hand , for certain applications one might be willing to sacrifice the information about the parametric degree in order to obtain smaller matrices .
if this is the case , there exist ( for certain surface parametrizations ) algorithms to compute a proper reparametrization of the surface , e.g. @xcite , and in these cases it is highly advisable to do so before computing the matrix representation , because this will allow us to represent the implicit equation directly instead of one of its powers , and the matrices will be significantly smaller .
let us illustrate this with example 2 from @xcite , which treats a parametrization @xmath54 defined by @xmath373 this is a parametrization of bidegree @xmath374 and its newton polytope is the whole rectangle of length 8 and width 4 , so we can apply the method of approximation complexes for @xmath42 .
we obtain a matrix of size @xmath375 representing @xmath376 , where @xmath377 is the implicit equation and @xmath378 . using the algorithm presented in @xcite
one can compute the following proper reparametrization of the surface @xmath82 : @xmath379 this parametrization of bidegree @xmath380 represents @xmath381 directly by a @xmath382-matrix .
representation matrices can be efficiently constructed by solving a linear system of relatively small size ( in our case @xmath383 equations in @xmath384 variables ) .
this means that their computation is much faster than the computation of the implicit equation and they are thus an interesting alternative as an implicit representation of the surface . in this paper , we have extended the method of matrix representations by linear syzygies to the case of rational surfaces parametrized over toric varieties ( and in particular to bihomogeneous parametrizations ) .
this generalization provides a better understanding of the method through the use of combinatorial commutative algebra . from a practical point of view
, it is also a major improvement , as it makes the method applicable for a much wider range of parametrizations ( for example , by avoiding unnecessary base points with bad properties ) and leads to significantly smaller representation matrices .
let us sum up the advantages and disadvantages compared to other techniques to compute matrix representations ( e.g. the ones introduced in @xcite ) .
the most important advantages are : * the method works in a very general setting and makes only minimal assumptions on the parametrization .
in particular , it works well in the presence of base points .
* unlike the method of toric resultants , we do not have to extract a maximal minor of unknown size , since the matrices are generically of full rank . *
the structure of the newton polytope of the parametrization is exploited , so one obtains much better results for sparse parametrizations , both in terms of computation time and in terms of the size of the representation matrix .
moreover , it subsumes the known method of approximation complexes in the case of dense homogeneous parametrizations , in which case the methods coincide . disadvantages of the method are the following . * unlike with the toric resultant or the method of moving planes and surfaces , the matrix representations are not square . *
the matrices involved are generally bigger than with the method of moving planes and surfaces .
it is important to remark that those disadvantages are inherent to the choice of the method : a square matrix built from linear syzygies does not exist in general and it is an automatic consequence that if one only uses linear syzygies to construct the matrix , it has to be bigger than a matrix which also uses entries of higher degree .
the choice of the method to use depends very much on the given parametrization and on what one needs to do with the matrix representation .
in this appendix we show how to compute a matrix representation with the method developed in this paper , using the computer algebra system macaulay2 @xcite . as it is probably the most interesting case from a practical point of view , we restrict our computations to bi - homogeneous parametrizations of a certain bi - degree @xmath259
. however , the method is easily adaptable to the toric case , or more precisely to a given fixed newton polytope @xmath67 and , where it is appropriate , we will give hints on what to change in the code .
moreover , we are not claiming that our implementation is optimized for efficiency ; anyone trying to implement the method to solve computationally involved examples is well - advised to give more ample consideration to this issue .
for example , in the toric case there are better suited software systems to compute the generators of the toric ideal @xmath275 , see @xcite .
let us start by defining the parametrization @xmath54 given by @xmath385 .
.... s = qq[s , u , t , v ] ; e1=4 ; e2=2 ; f1=s^4*t^2 + 2*s*u^3*v^2 f2=s^2*u^2*t*v-3*u^4*t*v f3=s*u^3*t*v+5*s^4*t^2 f4=2*s*u^3*v^2+s^2*u^2*t*v f = matrix{{f1,f2,f3,f4 } } .... the reader can experiment with the implementation simply by changing the definition of the polynomials and their degrees , the rest of the code being identical .
we first set up the list @xmath386 of monomials @xmath387 of bidegree @xmath388 . in the toric case
, this list should only contain the monomials corresponding to points in the newton polytope @xmath70 . .... st= { } ;
l=-1 ; d = gcd(e1,e2 ) ee1=numerator(e1/d ) ; ee2=numerator(e2/d ) ; for i from 0 to ee1 do ( for j from 0 to ee2 do ( st = append(st , s^i*u^(ee1-i)*t^j*v^(ee2-j ) ) ;
l = l+1 ) ) .... we compute the ideal @xmath275 and the quotient ring @xmath43 .
this is done by a grbner basis computation which works well for examples of small degree , but which should be replaced by the matrix formula in for more complicated examples . in the toric case ,
there exist specialized software systems such as @xcite to compute the ideal @xmath275 .
.... sx = qq[s , u , t , v , w , x_0 .. x_l , monomialorder=>eliminate 5 ]
x= { } ; st = matrix { st } ; f = sub(f , sx ) st = sub(st , sx ) te=1 ; for i from 0 to l do ( te = te*x_i ) j = ideal(1-w*te ) for i from 0 to l do ( j = j+ideal ( x_i - st_(0,i ) ) ) j= selectinsubring(1,gens gb j ) r = qq[x_0 .. x_l ] j = sub(j , r ) a = r / ideal(j ) .... next , we set up the list @xmath389 of monomials @xmath387 of bidegree @xmath259 and the list @xmath390 of the corresponding elements of the quotient ring @xmath43 . in the toric case
, this list should only contain the monomials corresponding to points in the newton polytope @xmath67 . .... use sx st= { } ; for i from 0 to e1 do ( for j from 0 to e2 do ( st = append(st , s^i*u^(e1-i)*t^j*v^(e2-j ) ) ; ) ) x= { } ; for z from 0 to length(st)-1 do ( f = st_z ; xx=1 ; is = degree substitute(f,{u=>1,v=>1,t=>1 } ) ; is = is_0 ; it = degree substitute(f,{u=>1,v=>1,s=>1 } ) ; it = it_0 ; iu = degree substitute(f,{t=>1,v=>1,s=>1 } ) ; iu = iu_0 ; iv = degree substitute(f,{u=>1,t=>1,s=>1 } ) ; iv = iv_0 ; ded=0 ; while ded < k do ( for mm from 0 to l do ( js = degree substitute(st_(0,mm),{u=>1,v=>1,t=>1 } ) ; js = js_0 ; jt = degree substitute(st_(0,mm),{u=>1,v=>1,s=>1 } ) ; jt = jt_0 ; ju = degree substitute(st_(0,mm),{t=>1,v=>1,s=>1 } ) ; ju = ju_0 ; jv = degree substitute(st_(0,mm),{u=>1,t=>1,s=>1 } ) ; jv = jv_0 ; if is>=js and it>=jt and iu>=ju and iv>=jv then ( xx = xx*x_mm ; ded = ded+1 ; is = is - js ; it = it - jt ; iv = iv - jv ; iu = iu - ju ; ) ) ) ; x = append(x , xx ) ; ) .... we can now define the new parametrization @xmath46 by the polynomials @xmath47 . .... x = matrix { x } ; x = sub(x , sx ) ( m , c)=coefficients(f , variables= > { s_sx , u_sx , t_sx ,
v_sx},monomials=>st ) g = x*c g = matrix{{g_(0,0),g_(0,1),g_(0,2),g_(0,3 ) } } g = sub(g , a ) .... in the following , we construct the matrix representation @xmath5 . for simplicity
, we compute the whole module @xmath391 , which is not necessary as we only need the graded part @xmath392 . in complicated examples
, one should compute only this graded part by directly solving the linear system given by in degree @xmath247 .
remark that the best bound @xmath393 depends on the parametrization .
.... use a z1=kernel koszul(1,g ) ; nu=2*d-1 s = a[t1,t2,t3,t4 ] g = sub(g , s ) ; z1nu = super basis(nu+d , z1 ) ; tnu = matrix{{t1,t2,t3,t4}}*substitute(z1nu , s ) ; lll = matrix { { x_0 .. x_l } } lll = sub(lll , s ) ll= { } for i from 0 to l do { ll = append(ll , lll_(0,i ) ) } ( m , m)=coefficients(tnu , variables= > ll , monomials=>substitute(basis(nu , a),s ) ) ; m ; .... the matrix @xmath5 is the desired matrix representation of the surface @xmath82 .
we thank laurent bus and marc chardin for useful discussions .
d. a. aruliah and robert m. corless and laureano gonzalez - vega and azar shakoori , geometric applications of the bezout matrix in the lagrange basis , snc 07 : proceedings of the 2007 international workshop on symbolic - numeric computation , london , ontario , canada .
2007 , pp . 5564 .
chardin , marc .
applications of some properties of the canonical module in computational projective algebraic geometry .
symbolic computation in algebra , analysis , and geometry .
j. symbolic comput .
29 , 2000 , 4 - 5 , pp . 527544
. d. a. cox .
equations of parametric curves and surfaces via syzygies . in _ symbolic computation : solving equations in algebra , geometry , and engineering ( south hadley , ma , 2000 ) _ , volume 286 of _ contemp .
_ , pages 120 .
soc . , providence , ri , 2001 .
m. dohm , implicitization of rational algebraic surfaces with syzygy - based methods , ph.d .
thesis , universit de nice - sophia antipolis , nice , france , 2008 .
electronic version available at http://tel.archives-ouvertes.fr/tel-00294484/en .
j. herzog , a. simis , and w. v. vasconcelos .
koszul homology and blowing - up rings . in _ commutative algebra ( trento , 1981 )
_ , volume 84 of _ lecture notes in pure and appl .
_ , pages 79169 .
dekker , new york , 1983 . | in this paper we show that a surface in @xmath0 parametrized over a 2-dimensional toric variety @xmath1 can be represented by a matrix of linear syzygies if the base points are finite in number and form locally a complete intersection .
this constitutes a direct generalization of the corresponding result over @xmath2 established in @xcite and @xcite . exploiting the sparse structure of the parametrization
, we obtain significantly smaller matrices than in the homogeneous case and the method becomes applicable to parametrizations for which it previously failed .
we also treat the important case @xmath3 in detail and give numerous examples . matrix representation , rational surface , syzygy , approximation complex , implicitization , toric variety |
a change - point model for a sequence of independent random variables @xmath0 is a model in which there exist unknown change points @xmath1 , @xmath2 , such that , for each @xmath3 , @xmath4 are identically distributed with a distribution that depends on @xmath5 . here , we consider parametric change - point models in which the distribution of @xmath6 is parametric ; however , the form of the distribution can be different for each @xmath5 .
change - point models are used in many fields .
for example , broemeling and tsurumi ( @xcite ) uses a multiple change - point model for the us demand for money ; lombard ( @xcite ) uses a multiple change - point model to model the effect of sudden changes in wind direction on the flight of a projectile ; reed ( @xcite ) uses a multiple change - point model in the analysis of forest fire data .
a number of authors have used multiple change - point models in the analysis of dna sequences ; see , for example , braun and muller ( @xcite ) , fu and curnow ( @xcite ) and halpern ( @xcite ) .
many further examples are provided in the monographs chen and gupta ( @xcite ) and csrg and horvth ( @xcite ) .
the goal of this paper is to establish the asymptotic properties of maximum likelihood estimators of the parameters of a multiple change - point model , under easily verifiable conditions .
these results are based on the following model .
assume that the vectors in the data set @xmath7 are independently drawn from the parametric model @xmath8 where @xmath9 is a probability density function of a continuous distribution with unknown common parameter @xmath10 for all @xmath11 and unknown within - segment parameters @xmath12 for each @xmath11 ; @xmath9 may have the same functional form for some or all of @xmath11 ; @xmath10 may be a vector ; @xmath12 may be a different vector parameter of different dimensions for each @xmath11 .
in this model , there are @xmath13 unknown change points @xmath14 , where the number of change points @xmath13 is assumed to be known .
the parameter @xmath15 is common to all segments .
there are a number of results available on the asymptotic properties of parameter estimators in change - point models .
see , for example , hinkley ( @xcite ) , hinkley and hinkley ( @xcite ) , battacharya ( @xcite ) , fu and curnow ( @xcite ) , jandhyala and fotopoulos ( @xcite ) and hawkins ( @xcite ) ; the two monographs chen and gupta ( @xcite ) and csrgo and horvth ( @xcite ) have detailed bibliographies on this topic . in particular
, hinkley ( @xcite ) considers likelihood - based inference for a single change - point model , obtaining the asymptotic distribution of the maximum likelihood estimator of the change point under the assumption that the other parameters in the model are known .
hinkley ( @xcite ) and hinkley ( @xcite ) argue that this asymptotic distribution is also valid when the parameters are unknown .
unfortunately , there are problems in extending the approach used in hinkley ( @xcite ) to the setting considered here .
the method used in hinkley ( @xcite ) is based on considering the relative locations of a candidate change point and the true change point .
when there is only a single change point , there are only three possibilities : the candidate change point is either greater than , less than or equal to the true change point . however , in models with @xmath13 change points , the relative positions of the candidate change points and the true change points can become quite complicated and the simplicity and elegance of the single change point argument is lost .
a second problem arises when extending the argument for the case in which the change points are the only parameters in the model to the case in which there are unknown within - segment parameters .
the consistency argument used in the former case is extended to the latter case using a `` consistency assumption '' ( hinkley ( @xcite ) , section 4.1 ) ; this condition is discussed in appendix and examples are given which show that this assumption is a strong one that is not generally satisfied in the class of models considered here .
there are relatively few results available on the asymptotic properties of maximum likelihood estimators in multiple change - point models .
thus , the present paper has done several things . in the general model described above ,
in which there is a fixed , but arbitrary , number of change points , we show that the maximum likelihood estimators of the change points are consistent and converge to the true change points at the rate @xmath16 , under relatively weak regularity conditions . as noted above , a simple extension of the approach used in single change - point models is not available ; thus , the second thing achieved by this paper is the introduction of the tools necessary for analyzing the likelihood function in a multiple change - point model . finally , the asymptotic distribution of the maximum likelihood estimators of the parameters of the within - segment distributions is derived for the general case described above , in which the form of the distribution can change from segment to segment and in which , possibly , there are parameters that are common to all segments .
the paper is organized as follows . the asymptotic theory of maximum likelihood estimators of a multiple change - point model
is described in section [ sec2 ] .
section [ sec3 ] contains a numerical example illustrating these results and section [ sec4 ] contains some discussion of future research which builds on the results given in this paper .
appendix discusses the `` consistency assumption '' used in hinkley ( @xcite ) ; all technical proofs are given in appendix .
consider estimation of the multiple change - point model introduced in section [ sec1 ] .
for any change point configuration @xmath17 , the log - likelihood function is given by @xmath18 estimators of all change points , all within - segment parameters and the common parameter are given by @xmath19 where @xmath20 , and @xmath21 are the parameter spaces of @xmath22 , @xmath23 , and @xmath24 , respectively .
let @xmath25 note that @xmath26 is taken to be a constant vector as @xmath27 goes to infinity .
define @xmath28 the expected information matrix is given by @xmath29 = \pmatrix { e[-\ell^0_{\psi\psi}(\psi,\theta);\phi ] & e[-\ell^0_{\psi\theta}(\psi , \theta);\phi]\vspace*{3pt}\cr e[-\ell^0_{\psi\theta}(\psi,\theta);\phi]^t&e[-\ell^0_{\theta\theta } ( \psi,\theta);\phi]},\ ] ] @xmath30\\ & & \quad=\bigl(e\bigl[-\ell^{(1)}_{\psi\theta_1}(\psi,\theta_1);\phi\bigr],e\bigl[-\ell ^{(2)}_{\psi\theta_2}(\psi,\theta_2);\phi\bigr],\ldots , e\bigl[-\ell^{(k+1)}_{\psi \theta_{k+1}}(\psi,\theta_{k+1});\phi\bigr]\bigr ) , \\ & & e[-\ell^0_{\theta\theta}(\psi,\theta);\phi]\\ & & \quad = \operatorname{diag}\bigl ( e\bigl[-\ell^{(1)}_{\theta_1\theta_1}(\psi,\theta_1);\phi\bigr ] , e\bigl[-\ell^{(2)}_{\theta_2\theta_2}(\psi,\theta_2);\phi\bigr ] , \ldots , e\bigl[-\ell^{(k+1)}_{\theta_{k+1}\theta_{k+1}}(\psi,\theta_{k+1});\phi\bigr]\bigr),\end{aligned}\ ] ] where @xmath31 denotes a diagonal block matrix whose diagonal blocks are in the bracket , other elements are zeros and the average expected information matrix is given by @xmath32 the asymptotic properties of these estimators are based on the following regularity conditions .
other than the parts concerning change points , these conditions are typically similar to those required for the consistency and asymptotic normality of maximum likelihood estimators of parameters in models without change points ; see , for example , wald ( @xcite ) .
particularly , compactness of parameter spaces is a common assumption in the classical likelihood literature .
these conditions are different from those required by ferger ( @xcite ) and dring ( @xcite ) , who consider estimation of change points in a nonparametric setting in which nothing is assumed about the within - segment distributions , using a type of nonparametric m - estimator based on empirical processes .
thus , these authors do not require conditions on the within - segment likelihood functions ; on the other hand , their method does not provide estimators of within - segment parameters .
[ a20 ] it is assumed that for @xmath33 , @xmath34 on a set of non - zero measure .
this assumption guarantees that the distributions in two neighboring segments are different ; clearly , this is required for the change points to be well defined .
[ a21 ] it is assumed that : for @xmath11 , @xmath22 and @xmath12 are contained in @xmath35 , where @xmath35 is a compact subset of @xmath36 ; @xmath24 and @xmath10 are contained in @xmath21 where @xmath21 is a compact subset of @xmath37 ; here , @xmath38 are non - negative integers ; @xmath39 is third - order continuously differentiable with respect to @xmath40 ; the expectations of the first and second order derivatives of @xmath41 with respect to @xmath42 exist for @xmath42 in its parameter space .
compactness of the parameter space is used to establish the consistency of the maximum likelihood estimators of @xmath43 ; see , for example , bahadur ( @xcite ) for further discussion of this condition and its necessity in general models .
if we assume further conditions on models , the compactness of the parameter space may be avoided . but this appears to be a substantial task for future work .
differentiability of the log - likelihood function is used to justify certain taylor series expansions .
both parts of assumption [ a21 ] are relatively weak and are essentially the same as conditions used in parametric models without change points ; see , for example , schervish ( @xcite ) , section 7.3 .
part 3 is very weak and is used in the proof of theorem [ thm3 ] .
[ a22]it is assumed that : for any @xmath11 and any integers @xmath44 satisfying @xmath45 , @xmath46\}\biggr)^2\biggr\}\leq c(t - s)^r , \ ] ] where @xmath47 and @xmath48 is a constant ; for any @xmath11 and any integers @xmath44 satisfying @xmath49 , @xmath50 -v(\psi,\theta_j;\psi^0,\theta^0_j)\}\biggr)^2\biggr\}\\ & & \quad\leq d(t - s)^r , \end{aligned}\ ] ] where @xmath51 is introduced in equation , @xmath47 and @xmath52 is a constant .
parts 1 and 2 of assumption [ a22 ] are technical requirements on the behavior of the log - likelihood function between and within segments , respectively .
this condition is used to ensure that the information regarding the within- and between - segment parameters grows quickly enough to establish consistency and asymptotic normality of the parameter estimators .
these conditions are relatively weak ; it is easy to check that they are satisfied by at least all distributions in the exponential family .
consider a probability density function of exponential family form : @xmath53 it is then straightforward that the schwarz inequality gives @xmath54\}\biggr)^2 \\ & & \quad \leq\biggl[1+\sum_{q=1}^{m}w_{q}(\eta)^2\biggr]\\ & & \qquad{}\times\biggl\ { \biggl[\sum_{i = s+1}^{t}\bigl(\log h(x_i)-e(\log h(x_i))\bigr)\biggr]^2+\sum_{q=1}^{m}\biggl[\sum _ { i = s+1}^{t}\bigl(t_{q}(x_i)-e(t_{q}(x_i))\bigr)\biggr]^2\biggr\}.\end{aligned}\ ] ] therefore , part 1 of assumption [ a22 ] is satisfied with @xmath55 because the function @xmath56 assumed to be continuous can achieve its maximum on the compact parameter space . similarly , part 2 of assumption [ a22 ] is also satisfied with @xmath55 .
the main results of this paper are given in the following three theorems .
[ thm1 ] under assumption , part 1 of assumption and part 1 of assumption , @xmath57 and @xmath58 as @xmath59 , that is , @xmath60 and @xmath61 , where @xmath62 for @xmath63 and @xmath11 .
note that @xmath64 , are not consistent ( hinkley ( @xcite ) ) ; it is the estimators of the change - point fractions @xmath65 , that are consistent .
the consistency of @xmath66 , and @xmath67 is the same as the corresponding result in classical likelihood theory for independent , identically distributed data .
[ thm2 ] under assumptions , we have @xmath68 where @xmath69 @xmath70 that is , @xmath71 for @xmath72 . we now consider the asymptotic distribution of @xmath73 , where @xmath74 . [ thm3 ] under assumptions , @xmath75 where @xmath76 is the @xmath77-dimensional multivariate normal distribution with mean vector zero and covariance matrix @xmath78 .
the proofs of theorems [ thm1][thm3 ] are based on the following approach .
define a function @xmath79 by @xmath80f_i(\psi^0,\theta_i^0;x)\,dx\biggr\}\nonumber\\ & & { } + \frac{1}{n}\sum_{j=1}^{k+1}\sum_{i = n_{j-1}+1}^{n_j}\ { \log f_j(\psi,\theta_j;x_i)-e[\log f_j(\psi,\theta_j;x_i)]\}\\ & & { } -\frac{1}{n}\sum_{j=1}^{k+1}\sum_{i = n_{j-1}^0 + 1}^{n_j^0}\ { \log f_j(\psi^0,\theta_j^0;x_i)-e[\log f_j(\psi^0,\theta_j^0;x_i)]\ } , \nonumber\end{aligned}\ ] ] where @xmath81 is the number of observations in the set @xmath82\cap[n_{i-1}^0 + 1,n_i^0]$ ] for @xmath83 .
we obviously have that @xmath84 thus , the maximum likelihood estimators may be defined as the maximizers of @xmath79 rather than as the maximizers of @xmath85 .
let @xmath86 be defined by @xmath87f_i(\psi^0,\theta _
i^0,x)\,dx \nonumber \\[-8pt ] \\[-8pt ] \nonumber & & \quad\mbox{for } i , j=1,2,\ldots , k+1.\end{aligned}\ ] ] note that @xmath79 may be written @xmath88 , where @xmath89 and @xmath90\ } \nonumber \\[-8pt ] \\[-8pt ] \nonumber & & { } -\frac{1}{n}\sum_{j=1}^{k+1}\sum _ { i = n_{j-1}^0 + 1}^{n_j^0}\{\log f_j(\psi^0,\theta_j^0;x_i)-e[\log f_j(\psi^0,\theta_j^0;x_i)]\}.\end{aligned}\ ] ] alternatively , we may write @xmath91 \nonumber \\[-8pt ] \\[-8pt ] \nonumber & & { } \hspace*{44pt}-\sum_{t\in\tilde{n}_{ji}}[\log f_i(\psi^0,\theta
_ i^0;x_t)-e(\log f_i(\psi^0,\theta_i^0,x_t))]\biggr\},\end{aligned}\ ] ] where @xmath92\cap[n_{i-1}^0 + 1,n_i^0]$ ] .
note that @xmath93 is a weighted sum of the negative kullback
leibler distances ; it will be shown that @xmath94 approaches @xmath95 as @xmath96 .
also , @xmath86 @xmath97 with equality if and only if @xmath98 almost everywhere ( kullback and leibler ( @xcite ) ) .
lemma [ lem1 ] gives a bound for @xmath93 .
[ lem1 ] under assumption and part 1 of assumption , there exist two positive constants @xmath99 and @xmath100 such that , for any @xmath101 and @xmath42 , @xmath102 where @xmath103 and @xmath104 .
lemma [ lem2 ] describes between - segment properties and within - segment properties of this model .
[ lem2 ] under part 1 of assumption [ a21 ] , the following two results follow from parts 1 and 2 of assumption [ a22 ] respectively : for any @xmath11 , any @xmath105 and any positive number @xmath106 , there exist a constant @xmath107 , independent of @xmath108 , and a constant @xmath47 , such that @xmath109\}\biggr|>\varepsilon\biggr ) \nonumber \\[-8pt ] \\[-8pt ] \nonumber & & \quad \leq a_j\frac { ( m_2-m_1)^r}{\varepsilon^2}.\end{aligned}\ ] ] for any @xmath11 and any positive number @xmath106 , there exist a constant @xmath110 , independent of @xmath108 , and a constant @xmath47 , such that @xmath111 \nonumber \\[-8pt ] \\[-8pt ] \nonumber & & { } \hspace*{134pt}-v(\psi,\theta_j;\psi^0,\theta^0_j)\}>\varepsilon\biggr)\leq b_j\frac { ( n^0_j - n^0_{j-1})^r}{\varepsilon^2}.\end{aligned}\ ] ] in practical applications , it is useful to have an estimator of @xmath112
. let @xmath113 & \hat{e}[-\hat{\ell}_{\psi\theta}(\hat{\psi},\hat{\theta});\hat{\phi}]\vspace*{3pt}\cr \hat{e}[-\hat{\ell}_{\psi\theta}(\hat{\psi},\hat{\theta});\hat{\phi } ] ^t&\hat{e}[-\hat{\ell}_{\theta\theta}(\hat{\psi},\hat{\theta});\hat { \phi}]},\\ \hat{e}[-\hat{\ell}_{\psi\psi}(\hat{\psi},\hat{\theta});\hat{\phi } ] & = & \sum_{j=1}^{k+1}\sum_{i=\hat{n}_{j-1}+1}^{\hat{n}_j}\frac { 1}{f_j^2(\hat{\psi},\hat{\theta}_j;x_i)}{f_j}_{\psi}(\hat{\psi},\hat { \theta}_j;x_i){f_j}_{\psi}^t(\hat{\psi},\hat{\theta}_j;x_i),\\ \hat{e}[-\hat{\ell}_{\psi\theta_j}(\hat{\psi},\hat{\theta});\hat{\phi } ] & = & \sum_{i=\hat{n}_{j-1}+1}^{\hat{n}_j}\frac{1}{f_j^2(\hat{\psi},\hat { \theta}_j;x_i)}{f_j}_{\psi}(\hat{\psi},\hat{\theta}_j;x_i){f_j}_{\theta _ j}^t(\hat{\psi},\hat{\theta}_j;x_i),\\
\hat{e}[-\hat{\ell}_{\theta_j\theta_j}(\hat{\psi},\hat{\theta});\hat { \phi}]&=&\sum_{i=\hat{n}_{j-1}+1}^{\hat{n}_j}\frac{1}{f_j^2(\hat{\psi } , \hat{\theta}_j;x_i)}{f_j}_{\theta_j}(\hat{\psi},\hat{\theta } _ j;x_i){f_j}_{\theta_j}^t(\hat{\psi},\hat{\theta}_j;x_i)\end{aligned}\ ] ] for @xmath11 .
then @xmath114 is a consistent estimator of @xmath115 .
consider the problem of analyzing the mineral content of a core sample , which is extensively studied in chen and gupta ( @xcite ) , chernoff ( @xcite ) and srivastava and worsley ( @xcite ) .
in particular , we consider the data in chernoff ( @xcite ) on the mineral content of @xmath116 minerals in a core sample measured at @xmath117 equally spaced points . since some of the minerals have a very low assay , we follow chen and gupta ( @xcite ) and srivastava and worsley ( @xcite ) in analyzing only the @xmath118 variables @xmath119 and @xmath120 with the highest assays .
thus , we assume that @xmath121 has a @xmath122-variate normal distribution with a within - segment mean parameter vector and a variance - covariance matrix that is common to all segments .
the analyses of chen and gupta ( @xcite ) , chernoff ( @xcite ) and srivastava and worsley ( @xcite ) suggest that there are @xmath122 change points of the mean vector and , hence , we make that assumption here .
the estimates of @xmath122 change points , within - segment parameters of mean vectors and common parameter of variance - covariance matrix were computed using maximum likelihood .
the estimated change points are @xmath123 and @xmath124 , which are different from those estimated change points by chen and gupta ( @xcite ) , chernoff ( @xcite ) and srivastava and worsley ( @xcite ) , and are more reasonable .
this is because chen and gupta ( @xcite ) , chernoff ( @xcite ) and srivastava and worsley ( @xcite ) use the binary segmentation procedures which detect multiple change points one by one , not simultaneously , whereas the method in this paper simultaneously estimates multiple change points .
the estimated six within - segment mean vectors are in the following .
they are arranged according to the order of from left to right .
for example , the two vectors on the first line are , respectively , the first and second within - segment mean vectors . @xmath125
the estimated common variance - covariance matrix is @xmath126
this paper establishes the consistency of maximum likelihood estimators of the parameters of a general class of multiple change - point models and gives the asymptotic distribution of the parameters of the within - segment distributions .
the required regularity conditions are relatively weak and are generally satisfied by exponential family distributions .
some important problems in the analysis of multiple change - point models were not considered here .
one is that the asymptotic distribution of the maximum likelihood estimator of the vector of change points was not considered .
the reason for this is that the methods used to determine this asymptotic distribution are quite different from the methods used to establish the consistency of the maximum likelihood estimator ; see , for example , hinkley ( @xcite ) for a treatment of this problem in a single change - point model .
thus , this is essentially a separate research topic .
however , the asymptotic properties obtained in this paper are necessary for the establishment of the asymptotic distribution of the maximum likelihood estimator of the vector of change points in this model
. this will be a subject of future work .
another important problem is to extend the results of this paper to the case in which the number of change points is not known and must be determined from the data .
clearly , a likelihood - based approach to this problem will require an understanding of the properties of maximum likelihood estimators in the model in which the number of change points is known .
thus , the results of the present paper can be considered as a first step toward the development of a likelihood - based methodology that can be used to determine simultaneously the number and location of the change points .
this is also a topic of future research .
consider a change - point model with a single change point , @xmath127 , and suppose that there are no common parameters in the model . in hinkley ( @xcite )
, it is shown that @xmath128 , the maximum likelihood estimator of @xmath127 , satisfies @xmath129 under the condition @xmath130 with probability @xmath131 as @xmath132 , which was described as a `` consistency assumption '' .
note that the random variables in the sum @xmath133 are drawn from the distribution with density @xmath134 .
suppose that @xmath135 converges to @xmath136 as @xmath137 , uniformly in @xmath138 , where @xmath139 is distributed according to the distribution with density @xmath140 .
equation ( [ eqa1 ] ) then holds , provided that @xmath141 note that , by properties of the kullback leibler distance and assumption [ a20 ] , @xmath142 for each @xmath138 .
thus , condition ( [ eqa1 ] ) fails whenever the distribution corresponding to the density @xmath140 is in the closure of the set of distributions corresponding to densities of the form @xmath143 , in a certain sense .
one such case occurs if @xmath144 and @xmath134 have the same parametric form with parameters @xmath145 , respectively , satisfying @xmath146 .
for instance , suppose that the random variables in the first segment are normally distributed with mean @xmath147 and standard deviation @xmath131 and the random variables in the second segment are normally distributed with mean @xmath148 and standard deviation @xmath131 .
then @xmath149 where @xmath150 is normally distributed with mean @xmath148 and variance @xmath151 .
clearly , ( [ eqa1 ] ) does not hold in this case .
a similar situation occurs when the distribution with density @xmath152 can be viewed as a limit of the distributions with densities @xmath143 .
for instance , suppose that @xmath144 is the density of a weibull distribution with rate parameter @xmath153 and shape parameter @xmath154 , @xmath155 , @xmath156 , and @xmath134 is the density of an exponential distribution with rate parameter @xmath157 . in this appendix
, we show that this is a strong assumption that is not generally satisfied by otherwise well - behaved models .
for instance , suppose that @xmath144 and @xmath134 have the same functional form and that the difference between the two distributions is due to the fact that @xmath158 .
again , ( [ eqa1 ] ) will not hold .
thus , the consistency condition used in hinkley ( @xcite ) is too strong for the general model considered here .
proof of lemma [ lem1 ] we first need to prepare some results which are to be used in this proof . for @xmath63 ,
let us define @xmath159,\ ] ] where @xmath160 .
we then have that @xmath161 for @xmath63 .
it is straightforward to show that @xmath162 is a convex function with respect to @xmath154 for any @xmath63 .
let @xmath163 . because @xmath164 for @xmath165 , convexity of @xmath166 gives that @xmath167 noting that @xmath168,\ ] ] it follows from assumption [ a20 ] that @xmath169 .
if we let @xmath170 , then @xmath171 .
let @xmath172 .
consider a change - point fraction configuration @xmath101 such that @xmath173 .
for any @xmath33 , there are two cases : a candidate change - point fraction @xmath174 may be on the left or the right of the true change - point fraction @xmath175 . for any @xmath5 with @xmath174 on the right of @xmath175 , we have that @xmath176 .
then @xmath177 if we define @xmath178 , then the case @xmath173 gives that @xmath179 and @xmath180\\ & \leq&\frac{n_{j , j+1}}{n}g_j(\phi^0)\leq(\lambda_j-\lambda_j^0)\bar { g}(\phi^0).\end{aligned}\ ] ] for any @xmath5 with @xmath174 on the left of @xmath175 , we have that @xmath181 .
similarly , we define @xmath182 . using the fact that @xmath183 , it similarly gives that @xmath184 .
therefore , if @xmath173 , then we obtain that @xmath185 .
on the other hand , @xmath186 we have @xmath187 for any @xmath5 , so @xmath188 now , consider the other case of a change - point fraction configuration @xmath101 , where @xmath189 .
it is clear that there exists a pair of integers @xmath190 such that @xmath191 , @xmath192 and @xmath193 .
let @xmath194 .
for any @xmath42 , we have that @xmath195\\ & \leq&\frac{n_{i , j+1}+n_{ij}}{n}\min(\alpha_{i , j+1},1-\alpha _ { i , j+1})\bar{g}(\phi^0)\\ & \leq&\frac{\delta_{\lambda}^0}{2}\min\biggl(\frac { n_{i , j+1}}{n},\frac{n_{ij}}{n}\biggr)\bar{g}(\phi^0)\\ & \leq&\frac{1}{2}\biggl(\frac{{\delta_{\lambda}^0}}{2}\biggr)^2\bar { g}(\phi^0).\end{aligned}\ ] ] combining the results from the two cases of @xmath196 and @xmath197 , it follows that @xmath198 and @xmath199 \leq-\frac{\delta_{\lambda}^0}{2}\min\biggl[\rho(\phi,\phi^0),-\frac{\delta _ { \lambda}^0}{4}\bar{g}(\phi^0)\\\biggr ] .\ ] ] note that ( [ eqb1 ] ) can be simplified .
if we define @xmath200 then we have that @xmath201 .
it follows from inequality ( [ eqb1 ] ) that @xmath202.\ ] ] if @xmath203 , then we have that @xmath204 if @xmath205 , then @xmath206 .
letting @xmath207 inequality ( [ eqb1 ] ) gives that @xmath208 .
setting @xmath209 , we finally have that @xmath210 which concludes the proof .
proof of lemma [ lem2 ] with part 1 of assumption [ a22 ] in mind , equation ( [ eq6 ] ) can be achieved by induction with respect to @xmath211 .
the induction method is similar to the one used in mricz , serfling and stout ( @xcite ) , so its proof is omitted here . using part 2 of assumption [ a22 ] , equation ( [ eq7 ] )
can be proven similarly by the same induction method .
proof of theorem [ thm1 ] let @xmath212 then , for any @xmath213 , it follows from lemma [ lem1 ] that @xmath214 therefore , we obtain that @xmath215\}\biggr|>\frac{c_1}{2}\delta\biggr ) \\ & & \qquad{}+p_r\biggl(\sum_{j=1}^{k+1}\frac{1}{n}\biggl|\sum _ { i = n_{j-1}^0 + 1}^{n_j^0}\{\log f_j(\psi^0,\theta_j^0;x_i)-e[\log f_j(\psi^0,\theta_j^0;x_i)]\biggr|>\frac{c_1}{2}\delta\biggr)\\ & & \quad\leq\sum_{j=1}^{k+1}p_r\biggl(\max_{0\leq n_{j-1}<n_j\leq n,\theta_j\in \theta_j,\psi\in\psi}\frac{1}{n}\biggl|\sum_{i = n_{j-1}+1}^{n_j}\{\log f_j(\psi , \theta_j;x_i)-e[\log f_j(\psi,\theta_j;x_i)]\}\biggr|\\ & & \quad\hspace*{43pt}>\frac{c_1\delta } { 2(k+1)}\biggr)\\ & & \qquad{}+\sum_{j=1}^{k+1}p_r\biggl(\frac{1}{n}\biggl|\sum _ { i = n_{j-1}^0 + 1}^{n_j^0}\{\log f_j(\psi^0,\theta_j^0;x_i)-e[\log f_j(\psi^0,\theta_j^0;x_i)]\}\biggr|>\frac{c_1\delta}{2(k+1)}\biggr ) .\end{aligned}\ ] ] it follows from lemma [ lem2 ] that @xmath216 ^ 2\biggl(\sum_{j=1}^{k+1}a_j\biggr)n^{r-2}\rightarrow0\qquad \mbox { as } n\rightarrow+\infty,\ ] ] noting that @xmath47 . for @xmath73 , we similarly obtain that @xmath217\}\biggr|\\ & & { } \hspace*{51pt}>\frac{c_2\delta } { 2(k+1)}\biggr)\\ & & \qquad{}+\sum_{j=1}^{k+1}p_r\biggl(\frac{1}{n}\biggl|\sum _ { i = n_{j-1}^0 + 1}^{n_j^0}\{\log f_j(\psi^0,\theta_j^0;x_i)-e[\log f_j(\psi^0,\theta_j^0;x_i)]\}\biggr|>\frac{c_2\delta}{2(k+1)}\biggr).\end{aligned}\ ] ] similarly , lemma [ lem2 ] shows that @xmath218 .
noting the fact that @xmath219 if and only if @xmath220 and @xmath221 , it follows that @xmath58 and @xmath222 for @xmath11 , which completes the proof .
proof of theorem [ thm2 ] let us first define @xmath223 for any @xmath213 .
because of the consistency of @xmath224 , we need to consider only those terms whose observations are in @xmath225 , @xmath226 and @xmath227 for all @xmath5 in equation ( [ eqn2.5 ] ) .
therefore , we have @xmath228\\ & & \qquad{}\hspace*{78pt}-\frac{1}{n}\sum_{t\in\tilde{n}_{jj}}[\log f_j(\psi ^0,\theta^0_j;x_t)-e(\log f_j(\psi^0,\theta^0_j;x_t))]\\ & & \qquad{}\hspace*{78pt}+\frac{1}{3(k+1)}j_1\biggr\}>0\biggr)\\ & & \qquad{}+\sum_{j=2}^{k+1}p_r\biggl(\max_{\lambda\in\lambda_{\delta , n},\phi\in\phi}\biggl\{\frac{1}{n}\sum_{t\in\tilde{n}_{j , j-1}}[\log f_j(\psi , \theta_j;x_t)-e(\log f_j(\psi,\theta_j;x_t))]\\ & & \qquad{}\hspace*{92pt}-\frac{1}{n}\sum_{t\in\tilde{n}_{j , j-1}}[\log f_{j-1}(\psi ^0,\theta^0_{j-1};x_t)-e(\log f_{j-1}(\psi^0,\theta^0_{j-1};x_t))]\\ & & \qquad{}\hspace*{92pt}+\frac { 1}{3k}j_1\biggr\}>0\biggr)\\ & & \qquad{}+\sum_{j=1}^{k}p_r\biggl(\max_{\lambda\in\lambda_{\delta , n},\phi \in\phi}\biggl\{\frac{1}{n}\sum_{t\in\tilde{n}_{j , j+1}}[\log f_j(\psi,\theta _ j;x_t)-e(\log f_j(\psi,\theta_j;x_t))]\\ & & \qquad{}\hspace*{91pt}-\frac{1}{n}\sum_{t\in\tilde{n}_{j , j+1}}[\log f_{j+1}(\psi ^0,\theta^0_{j+1};x_t)-e(\log f_{j+1}(\psi^0,\theta^0_{j+1};x_t))]\\ & & \qquad{}\hspace*{92pt}+\frac { 1}{3k}j_1\biggr\}>0\biggr)\\ & & \quad \equiv\sum_{j=1}^{k+1}i_{1j}+\sum_{j=2}^{k+1}i_{2j}+\sum _ { j=1}^{k}i_{3j}.\end{aligned}\ ] ] first , consider the probability formulas @xmath229 in the above equation for any @xmath11 .
the consistency of @xmath224 allows us to restrict our attention to the case @xmath230 . for this case
, we have that @xmath231 therefore , we obtain that @xmath232\\ & & \qquad{}\hspace*{16pt}-v(\psi^*,\theta^*_j;\psi ^0,\theta^0_j)\}>\frac{n^0_j - n^0_{j-1}}{6(k+1 ) } |v(\psi^*,\theta _ j^*;\psi^0,\theta^0_j)|\biggr)\\ & \leq & p_r\biggl(\max_{n_{j-1}^0\leq s <
t\leq n_j^0,\psi\in\psi , \theta_j\in\theta_j}\sum_{i = s+1}^t\{[\log f_j(\psi,\theta _ j;x_t)-\log f_j(\psi^0,\theta^0_j;x_t)]\\ & & \qquad{}\hspace*{112pt}-v(\psi,\theta_j;\psi^0,\theta^0_j)\}>\frac { e}{6(k+1)}(n^0_j - n^0_{j-1})\biggr),\end{aligned}\ ] ] where @xmath233 , @xmath234 , @xmath235 and @xmath236 are , respectively , the maximizing values of @xmath237 , @xmath24 , @xmath238 and @xmath101 obtained through the maximization .
equation ( [ eq7 ] ) of lemma [ lem2 ] can then be applied to show that @xmath239 as @xmath240 .
next , consider the probability formula @xmath241 for any @xmath242 . in this case , @xmath243 .
we have that @xmath244+\frac{1}{6k}j_1\biggr\}>0\biggr)\\ & & { } \hspace*{18pt}+p_r\biggl(\max_{\lambda\in\lambda_{\delta , n},\phi\in\phi}\biggl\ { -\frac{1}{n}\sum_{t\in\tilde{n}_{j , j-1}}[\log f_{j-1}(\psi,\theta _ { j-1};x_t)-e(\log f_{j-1}(\psi,\theta_{j-1};x_t))]\\ & & \qquad{}\hspace*{74pt}+\frac{1}{6k}j_1\biggr\}>0\biggr)\\ & \equiv & i_{2j}^{(1)}+i_{2j}^{(2)}.\end{aligned}\ ] ] @xmath245 and @xmath246 can be handled in the same way , so we just show how to handle @xmath245 .
only two cases have to be considered . if @xmath247 , then @xmath248\biggr|\\ & & { } \quad\hspace*{32pt}>\frac{c_1\delta}{6k}\biggr).\end{aligned}\ ] ] equation ( [ eq6 ] ) of lemma [ lem2 ] gives that @xmath249 as @xmath250 .
if @xmath251 for the other case , then @xmath252 . therefore , we obtain that @xmath253\\ & & \qquad{}\hspace*{113pt}-\frac{c_1}{6k}\biggr)>0\biggr)\\ & \leq & p_r\biggl(\max_{n_{j-1}\leq s < t\leq n^0_{j-1},\theta _ j\in\theta_j,\psi\in\psi}\biggl|\sum_{i = s+1}^{t}[\log f_j(\psi,\theta _
j;x_i)-e(\log f_j(\psi,\theta_j;x_i))]\biggr|\\ & & \quad{}\hspace*{2pt}>\frac { c_1}{6k}(n^0_{j-1}-n_{j-1})\biggr),\end{aligned}\ ] ] which converges to zero as @xmath254 , by equation ( [ eq6 ] ) of lemma [ lem2 ] .
@xmath255 can be handled in the same way as @xmath241 .
therefore , theorem [ thm2 ] is proved .
proof of theorem [ thm3 ] we first have the expansion @xmath256(\hat{\phi } -\phi^0).\ ] ] the fact that @xmath257 then gives that @xmath258^{-1}\frac{\hat{\ell}_{\phi}(\psi^0,\theta^0)}{\sqrt{n}}.\ ] ] now , consider the limit of @xmath259 .
we have that @xmath260 + \frac{1}{\sqrt{n}}\ell^0_{\phi}(\psi^0,\theta^0).\nonumber\end{aligned}\ ] ] because of the consistency of @xmath224 , we can assume that @xmath261 for @xmath33 .
it is then straightforward to obtain that @xmath262\\ & & \quad=\frac{1}{\sqrt{n}}\sum_{j=1}^{k+1}\bigl[\hat{\ell}_{\phi } ^{(j)}(\psi^0,\theta_j^0)-\ell_{\phi}^{(j)}(\psi^0,\theta _ j^0)\bigr]\nonumber\\ & & \quad = \frac{1}{\sqrt{n}}\sum_{j=1}^{k+1}\biggl\ { i(\hat{n}_j\geq n^0_j,\hat{n}_{j-1}\geq n^0_{j-1})\\ & & { } \hspace*{56pt}\times\biggl[\sum_{i = n^0_j+1}^{\hat{n}_j}\frac{\partial}{\partial\phi } \log f_j(\psi^0,\theta^0_j;x_i ) -\sum_{i = n^0_{j-1}+1}^{\hat{n}_{j-1}}\frac{\partial}{\partial\phi}\log f_j(\psi^0,\theta^0_j;x_i)\biggr]\\ & & { } \hspace*{58pt}+i(\hat{n}_j\geq n^0_j,\hat{n}_{j-1 } < n^0_{j-1})\\ & & { } \hspace*{67pt}\times\biggl[\sum_{i = n^0_j+1}^{\hat{n}_j}\frac{\partial}{\partial\phi } \log f_j(\psi^0,\theta^0_j;x_i)+\sum_{i=\hat { n}_{j-1}+1}^{n^0_{j-1}}\frac{\partial}{\partial\phi}\log f_j(\psi ^0,\theta^0_j;x_i)\biggr]\\ & & { } \hspace*{58pt}+i(\hat{n}_j < n^0_j,\hat{n}_{j-1}\geq n^0_{j-1})\\ & & { } \hspace*{66pt}\times\biggl[-\sum_{i=\hat{n}_j+1}^{n^0_j}\frac{\partial}{\partial\phi } \log f_j(\psi^0,\theta^0_j;x_i)-\sum_{i = n^0_{j-1}+1}^{\hat { n}_{j-1}}\frac{\partial}{\partial\phi}\log f_j(\psi^0,\theta ^0_j;x_i)\biggr]\\ & & { } \hspace*{58pt}+i(\hat{n}_j < n^0_j,\hat{n}_{j-1 } < n^0_{j-1})\\ & & { } \hspace*{68pt}\times\biggl[-\sum_{i=\hat{n}_j+1}^{n^0_j}\frac{\partial}{\partial\phi } \log f_j(\psi^0,\theta^0_j;x_i)+\sum_{i=\hat { n}_{j-1}+1}^{n^0_{j-1}}\frac{\partial}{\partial\phi}\log f_j(\psi ^0,\theta^0_j;x_i)\biggr]\biggr\}.\end{aligned}\ ] ] it follows from theorem [ thm2 ] that @xmath263=\frac{1}{\sqrt{n}}\mathrm{o}_p(1),\ ] ] which converges to zero in probability as @xmath264 .
since @xmath265 it follows that @xmath266 in a similar way , we easily obtain that @xmath267 therefore , we have that @xmath268 proving the result .
h. he thanks professor peter hall for his support of this research .
the research of h. he was financially supported by a mascos grant from the australian research council .
the research of t.a .
severini was supported by the u.s
. national science foundation . | models with multiple change points are used in many fields ; however , the theoretical properties of maximum likelihood estimators of such models have received relatively little attention .
the goal of this paper is to establish the asymptotic properties of maximum likelihood estimators of the parameters of a multiple change - point model for a general class of models in which the form of the distribution can change from segment to segment and in which , possibly , there are parameters that are common to all segments .
consistency of the maximum likelihood estimators of the change points is established and the rate of convergence is determined ; the asymptotic distribution of the maximum likelihood estimators of the parameters of the within - segment distributions is also derived .
since the approach used in single change - point models is not easily extended to multiple change - point models , these results require the introduction of those tools for analyzing the likelihood function in a multiple change - point model . |
the israeli electrostatic accelerator fel ( ea - fel ) is based on a 6 mev en - tandem van de graaff accelerator , which was originally used as an ion accelerator for nuclear physics experiments @xcite .
the scheme employs straight geometry for the electron beam transport , where the electron gun and the collector are installed outside of the accelerator region .
lasing was reported in a previous configuration , where radiation was transmitted through the accelerator tubes with 40db attenuation @xcite . in the present version of the fel , which was relocated to ariel
, the millimeter - wave radiation generated in the resonator is separated from the electron beam by means of a perforated talbot effect reflector @xcite .
a quasi - optic delivery system transmits the out - coupled power through a window in the pressurized gas accelerator tank .
the basic parameters of the fel are summarized in table 1 .
the acceleration voltage is set to be @xmath0 in order to tune the frequency of the fel radiation to the w - band near 100 ghz . in the following sections , we present an analysis and the results of spontaneous and stimulated emissions measurements carried out recently .
ll & + + @xmath1electron beam energy : & @xmath2 mev + @xmath1beam current : & @xmath3 a + & + + @xmath1type : & + @xmath1magnetic induction : & @xmath4=2 kg + @xmath1period length : & @xmath5=4.444 cm + @xmath1number of periods : & @xmath6=20 + & + + @xmath1waveguide : & + @xmath1transverse mode : & @xmath7 + @xmath1round - trip length : & @xmath8=2.62 m + @xmath1out - coupling coefficient : & @xmath9=7% + @xmath1total round - trip reflectivity : & @xmath10=65% +
random electron distribution in the e - beam causes fluctuations in current density , identified as _ shot noise _ in the beam current .
electrons passing through a magnetic undulator emit a partially coherent radiation , which is called _ undulator synchrotron radiation_. the electromagnetic fields excited by each electron add incoherently , resulting in a _ spontaneous emission _ with generated power spectral density @xcite : @xmath11 where @xmath12 is the _ expected value _ of the spontaneous emission power , @xmath13 is the slippage time and @xmath14 is the detuning parameter ( @xmath15 is the axial velocity of the accelerated electrons and @xmath16 is the group velocity of the generated radiation ) .
the spontaneous emission null - to - null bandwidth is approximately @xmath17 . in a fel , utilizing a magneto - static planar wiggler , the total power of the spontaneous emission is given by : @xmath18 where @xmath19 is the mode impedance , and @xmath20 is the dc beam current .
the expected value of the total spontaneous emission power generated inside the cavity is about @xmath21 .
the calculated spectrum of the spontaneous emission power of the israeli ea - fel , has a null - to - null bandwidth of 18 ghz . at the resonator output , the spontaneous emission spectrum generated inside the resonator
is modified by a fabry - perot spectral transfer - function @xcite : @xmath22 where @xmath8 is the resonator ( round - trip ) length , @xmath10 is the total power reflectivity of the cavity , @xmath9 is the power transmission of the out - coupler and @xmath23 is the axial wavenumber of the waveguide mode .
the maxima of the resonator transfer function factor occur when @xmath24 ( where @xmath25 is an integer ) , which defines the resonant frequencies @xmath26 of the longitudinal modes .
the _ free - spectral range _
( fsr ) ( the inter - mode frequency separation ) is given by @xmath27 mhz .
the transmission peak is @xmath28 with _ full - width half - maximum _ ( fwhm ) bandwidth of @xmath29 mhz , where @xmath30{r}/(1 - \sqrt{r})=14.56 $ ] is the _ finesse _ of the resonator .
the spectral line - shape of the spontaneous emission power obtained at the resonator output of the ea - fel is shown in fig . 1 .
the _ noise equivalent bandwidth _ is defined as the bandwidth of an ideal band - pass filter producing the same noise power at its output .
the noise equivalent bandwidth of a single resonant longitudinal mode is @xmath31 mhz .
consequently , the spontaneous emission power of mode @xmath25 is given by : @xmath32 the typical bandwidth of the generated spontaneous emission power spectrum ( 1 ) is @xmath33 ghz .
the number of longitudinal modes within the spontaneous emission bandwidth is then @xmath34 .
thus the total spontaneous emission power measured at the output of the resonator is given as follows : @xmath35 using equation ( 2 ) , we expect up to @xmath36 spontaneous emission power to be generated inside the resonator . from ( 5 ) , the power emitted from the resonator out - coupler is reduced to @xmath37 .
the attenuation of the wave - guiding system , delivering the power from the resonator , located inside the high - voltage terminal , to the measurement apparatus is 10db .
consequently , the spontaneous emission power expected at the detector sight is 2.4 @xmath38w .
the traces shown in fig . 2 , describe the electron beam current pulse and the signal obtained at the detector video output , corresponding to the measured spontaneous emission rf power .
in the present operation regime of the fel , the efficiency of energy extraction from the electron beam is given in terms of the number of wiggler s periods @xmath39 by the approximate formula @xmath40 % .
the stimulated radiation power generated inside the resonator at steady state is given as follows : @xmath41 where @xmath42 kw for a beam current of @xmath43 a. the resulted power obtained from the out - coupler is given as follows : @xmath44 and evaluated to be @xmath45 kw . considering the attenuation of the transmission system , 700 w is expected at the detector .
3 shows recent measurement of 150 w radiation power at the end of the optical transmission line in the measurement room .
we note that in the present preliminary experiments , only a fraction of the cathode current was transported through the wiggler , and no beam circulation ( transport up to the collector ) was achieved .
the charging of the terminal caused voltage drop of the terminal of 125 kv during the pulse duration .
evidently , the fel had not yet reached saturation because the radiation mode built inside the resonator went out of synchronism with the beam before reaching saturation .
+ 40 a. gover , e. jerby , h. kleinman , i. ben - zvi , b.v .
elkonin , a. fruchtman , j.s .
sokolowski , b. mandelbaum , a. rosenberg , j. shiloh , g. hazak , o. shahal , nucl .
instr . and meth . a 296 ( 1990 ) 720 .
a. abramovich , m. canter , a. gover , j.s .
sokolovski , y.m .
yakover , y. pinhasi , i. schnitzer , j. shiloh , phys .
bf 82 ( 1999 ) 6774 .
a. abramovich et al . , appl .
71 ( 1997 ) 3776 . | the israeli tandem electrostatic accelerator fel ( ea - fel ) , which is based on an electrostatic van der graaff accelerator was relocated to ariel 3 years ago , and has now returned to operation under a new configuration . in the present fel
, the millimeter - wave radiation generated in the resonator is separated from the electron beam by means of a perforated talbot effect reflector .
a quasi - optic delivery system transmits the out - coupled power through a window in the pressurized gas accelerator tank into the measurement room ( in the previous configuration , radiation was transmitted through the accelerator tubes with 40 db attenuation ) .
this makes it possible to transmit useful power out of the accelerator and into the user laboratories . after re - configuring the fel electron gun and the e - beam transport optics and installing a two stage depressed collector ,
the e - beam current was raised to 2 a. this recently enabled us to measure both spontaneous and stimulated emissions of radiation in the newly configured fel for the first time .
the radiation at the w - band was measured and characterized .
the results match the predictions of our earlier theoretical modeling and calculations . |
in early numerical simulations , the merging of a compact galaxy group was found to lead to the formation of a single elliptical galaxy ( hence a `` fossil group '' ) in a few billion years ( barnes 1989 ) .
the final product retains an x - ray emitting halo of hot gas ( ponman & bertram 1993 ) .
the whole scenario was proved to be plausible by the discovery of the fossil group rxj1340.6@xmath144018 ( ponman et al .
1994 ) .
the standard observational definition of a fossil system is based on the existence of extended x - ray emission with a bolometric luminosity @xmath15 and of an r - band magnitude gap @xmath16 between the brightest and second brightest members within @xmath7 is the radius within which the total mass density of a group / cluster is equal to @xmath17 times the critical density of the universe ( @xmath18 ) .
correspondingly , @xmath19 is the total mass enclosed by @xmath20 . ]
( jones , ponman & forbes 2000 ) . in theoretical studies ,
the analogous definition of a fossil system is less straightforward ( see e.g. , donghia et al .
2005 for a discussion ) . on the basis of the existing observations , mostly limited to the local universe
, fossil groups are expected to be a considerable population among the gravitationally bound systems of galaxies ( vikhlinin et al .
1999 ; romer et al .
2000 ; jones et al . 2003 ; la barbera et al . 2009
; voevodkin et al . 2010 ) . for instance , their estimated spatial density is equal to @xmath21 for an x - ray bolometric luminosity above @xmath22 ( la barbera et al .
x - ray properties and dark - matter ( dm ) content of fossil groups are comparable to those of bright groups and poor clusters of galaxies , with total masses of @xmath23@xmath24 , whereas their brightest members are typically as luminous as a cluster cd galaxy ( @xmath25 ) . in recent high - resolution hydrodynamical cosmological simulations in a @xmath26 cold dark matter ( @xmath26cdm ) universe ,
fossil groups have already assembled half of their final mass at redshifts @xmath27 , and subsequently they typically grow by minor mergers only , whereas non - fossil systems of similar total mass on average form later ( donghia et al .
2005 ; dariush et al . 2007 ) .
the early assembly leaves sufficient time for galaxies with a luminosity @xmath28 ( assuming a schechter 1976 luminosity function ) to merge into the central galaxy by dynamical friction , which yields the magnitude gap defining fossil groups .
this `` early formation '' scenario is alternative to the compact group merging scenario ( barnes 1989 ) .
however , no difference seems to exist in the distribution of neighboring faint galaxy density excess , distance from the red sequence break . among coeval galaxies , the old , passively evolving ones exhibit the reddest colours , if dust attenuation is neglected ( see pierini et al .
2004 for a discussion ) . ] , structural parameters , or stellar population properties between the brightest central galaxies ( bcgs ) in 25 fossil groups at @xmath29 and a sample of bright field ellipticals with a similar distribution in optical luminosity and x - ray properties ( la barbera et al .
these authors argue that this evidence undermines the idea that fossil systems represent the `` end product '' of a physical mechanism that occurred at high redshift ; it rather suggests that fossil groups are the final stage of mass assembly in the universe . whether either interpretation is true or fossil groups correspond to a transient phase in the life of a galaxy group ( von benda - beckmann et al .
2008 ; dariush et al .
2010 ; cui et al .
2011 ) , whether they form a distinct family of groups ( donghia & lake 2004 ; khosroshahi , ponman & jones 2007 ) or not ( sales et al .
2007 ; zibetti , pierini & pratt 2009 ; sun et al .
2009 ; dmocls et al .
2010 ; voevodkin et al .
2010 ; cui et al .
2011 ) , for several reasons it is important to extend the present knowledge of fossil systems beyond the local universe and towards lower total masses .
if undisturbed for a very long time , fossil groups may represent the ultimate example of hydrostatic equilibrium in viriliazed systems ( vikhlinin et al . 1999 ) .
this conclusion is supported by a recent detailed study of two such systems based on high - quality xmm-_newton _ spectro - imaging data ( dmocls et al .
2010 ) .
in addition , fossil groups could represent the least massive , baryonically closed systems ( mathews et al .
2005 ) , their baryon mass fraction being equal to the cosmic mean value . ] .
if so , they could form a reference sample with a recent , quiescent merging activity with which one can investigate the origin of the average baryon deficiency within @xmath30 ( @xmath31 ) found in x - ray emitting groups at @xmath32 ( giodini et al .
2009 , hereafter referred to as g09 ; see also teyssier et al .
in particular , two pair - matched samples of fossil groups with a range in total mass and redshift , one with radio - loud bcgs and the other with radio - quiet bcgs , could be used to estimate the `` instantaneous '' boost of the entropy in the central regions of the icm that is due to radio - galaxy feedback ( cf .
giodini et al .
2010 ) . from simple considerations on the hierachical nature of structure growth via merging ( milosavljevi et al .
2006 ) , fossil systems are expected to be more frequent ( by a factor of ten ) at the group mass - scale than at the cluster mass - scale ( @xmath33@xmath34 ) .
this is confirmed by the most recent study of fossil groups based on hydrodynamical resimulations of five regions in the millennium simulation and on galaxy formation models ( cui et al .
2011 and references therein ) .
therefore fossil groups may also help in understanding the properties of more massive bound systems at @xmath35 , especially if it turns out that they foster the formation of the most massive progenitor of the bcg of a cluster , as suggested by j0454@xmath360309 at @xmath37 ( schirmer et al .
2010 ; cf .
barnes 1989 and references therein ) . during our ongoing studies of the x - ray
selected groups in the @xmath38 area of the cosmic evolution survey ( _ cosmos _ ; scoville et al .
2007 ) , we have discovered two fossil groups at @xmath39 and @xmath0 . as demonstrated by g09 and giodini et al .
( 2010 ) , the multi - wavelength data set of the _ cosmos _ offers a thus far unique opportunity to characterize x - ray selected groups of galaxies up to @xmath40 .
this enables us not only to fully describe the galaxy populations of the two fossil groups , but also to characterize their surrounding large - scale structures ( lsss ) , and to speculate on their progenitors . in sect .
2 we describe selection and properties of these two groups .
their galaxy stellar mass functions and colour magnitude diagrams are determined in sect .
discussion of results and inference on evolutionary scenarios appear in sect . 4 ; conclusions are drawn in sect .
5 . as in g09
, we adopt a @xmath26cdm cosmological model ( @xmath41 , @xmath42 ) with @xmath43 , consistently with results from _ wmap5 _
( dunkley et al . 2009 ; komatsu et al . 2009 ) .
thus , a redshift of 0.372 ( 0.425 ) corresponds to a lookback time of @xmath44 ( @xmath45 ) , a luminosity distance of @xmath46 ( @xmath47 ) , an angular diameter distance of @xmath48 ( @xmath49 ) , and a scale of @xmath50 ( @xmath51 ) .
hereafter magnitudes are expressed in the ab system unless otherwise noted .
-0.35truecm -1.25truecm -0.35truecm -1.25truecm from the composite mosaic of the xmm-_newton _ and _ chandra _ x - ray data mapping the _ cosmos _ @xmath38 area , extended sources ( i.e. , groups and clusters ) were detected down to a flux limit of @xmath52 ( finoguenov et al . in preparation
; see also finoguenov et al .
g09 selected 91 groups / poor clusters at @xmath53 , which have an x - ray flux estimate at a significance better than @xmath54 , after point - like source subtraction ( see finoguenov et al .
2007 ) . for these groups ,
the membership of a galaxy ( through a precise photometric redshift at the redshift of the two fossil groups under study ( i.e. , @xmath55 ) , the @xmath56 error on the photometric redshift is equal to @xmath57 . ] ) and the total stellar mass in galaxies together with its uncertainty were robustly determined as described in g09 .
the total mass of each group was determined from the @xmath58@xmath2 relation established through weak lensing analysis by leauthaud et al .
uncertainties on @xmath2 are statistical only ( i.e. , they arise from the @xmath58@xmath2 relation ) .
the effects of systematics are estimated to be lower than the statistical uncertainty@xmath2 relation . ] .
we searched for fossil groups in the g09 sample .
as in giodini et al .
( in preparation ) , we defined as member candidates of a group those galaxies within a projected group - centric distance of @xmath59 , and with a photometric redshift consistent with the photometric redshift of the group within two times the uncertainty on the photometric redshift of the group .
then we inspected the brightest member - candidate galaxies within a projected group - centric distance equal to @xmath60 to minimize the potential bias produced by uncertainties on @xmath59 .
as for ranking in luminosity , the available @xmath5-band photometry was considered , since the reference _ cosmos _ photometric catalogue is extracted from imaging in this band ( ilbert et al . 2009 ) .
this choice avoids the use of template- and ( photometric ) redshift - dependent @xmath61-corrections as for the choice of considering rest - frame r - band magnitudes .
furthermore , up to @xmath62 , the @xmath5-filter ( observed frame ) maps emission at rest - frame wavelengths redder than the 4000 - break , which limits the impact of young stellar populations and , potentially , attenuation by a dusty interstellar medium . a number of groups were initially selected as potential fossil systems , the magnitude difference between the bcg and any other bright member - candidate galaxy within @xmath60 being conservatively chosen to be larger than @xmath63-mag .
we set a larger search radius and a smaller magnitude gap than those required by the definition of fossil groups ( cf .
1 ) in order to maximize the number of candidates in this first round of the search for fossil groups .
two groups were successively established as fossil thanks to the existing spectroscopic information ( lilly et al .
2007 ) , limited to sources brighter than @xmath64 , which enabled a finer screening against projected foreground / background bright galaxies .
on one hand , we flagged out galaxies classified as member candidates if they were located within @xmath59 from the x - ray centroid of either fossil group but exhibited a line - of - sight velocity in excess of @xmath65 plus its @xmath66 uncertainty . ] with respect to the spectroscopically determined recession velocity of either group .
the redshift of each group was computed using the median spectroscopic redshifts of the galaxies within @xmath59 through iterative removal of outliers with a recession velocity beyond @xmath67 with respect to the recession velocity of the group .
the starting redshift was defined by the photometric redshifts of galaxies on the red sequence .
contamination was checked by running a red - sequence - finding algorithm between redshifts 0 and 2 and by using spectroscopic group catalogs ( knobel et al .
2009 ) . on the other hand
, the available spectroscopy enabled us to check , in particular , that bright galaxies within @xmath7 from the x - ray centroid of either fossil group were actually not dropped as member candidates because of an inaccurate photometric redshift .
additional considerations on size and luminosity were used to check the goodness of the photometric redshifts of relatively bright galaxies without spectroscopy ( see [email protected] in fig . 2 ) .
figures 1 and 2 show cut - outs of the central regions of the two _ cosmos _ fossil groups under study , which encompass the area within @xmath7 for each system . they reproduce x - ray surface brightness contours overlaid on the @xmath5-band hst / acs images ( koekemoer et al .
2007 ) of the two galaxy systems .
in particular , member candidates ( i.e. , photometric members ) are indicated together with all galaxies brighter than @xmath64 having a spectroscopic redshift .
the bcg of either group can be easily identified at the center of the x - ray emission region .
one system , [email protected] , has @xmath68 ( i.e. , @xmath69 as obtained from the difference between the magnitudes in the rest - frame @xmath70-band ) , @xmath71 , @xmath72 , @xmath39 , and six spectroscopic members .
it is part of a large - scale structure ( see fig . 1 in g09 and fig . 3 in scoville et al .
2007 ) populated by 28 x - ray emitting groups distributed across the entire @xmath38 area of the _ cosmos _ ( corresponding to a cross size of about @xmath73 at @xmath74 ) . furthermore
, its dominant galaxy ( with a rest - frame absolute magnitude @xmath75 ) hosts a point - like radio source ( see giodini et al . 2010 ) .
the other fossil group , [email protected] , has @xmath76 ( i.e. , @xmath77 in the rest - frame @xmath70-band ) , @xmath78 , @xmath79 , @xmath0 , and eight spectroscopic members .
it is isolated and its bcg has @xmath80 .
basic properties of the two fossil groups under study are listed in table 1 .
in addition , we note that they populate the upper half of the distribution of the _ cosmos _ x - ray selected groups in the ( galaxy ) stellar mass fraction group total - mass diagram ( g09 , their fig . 5 ) , where quantities are estimated at @xmath60 .
this is particularly true for [email protected] . as from g09 ,
the stellar mass fraction is equal to 0.039 for the more massive fossil group ( [email protected] ) and 0.143 for the less massive one ( [email protected] ) .
thus , the fraction of the stellar mass in galaxies decreases with increasing total mass for the two fossil groups as for all other coeval groups ( g09 ) .
furthermore , the more ( less ) massive fossil group exhibits a value of the total @xmath5-band - to - x - ray luminosity ratio at @xmath59 ( @xmath81 ) equal to 138 ( 305 ) , after background subtraction including spectroscopic information ( from giodini et al . in preparation ) . therefore [email protected] ( [email protected] ) exhibits a low ( normal ) value for its total mass with respect to the distribution of @xmath81 as a function of @xmath2 for the 51 groups at @xmath820.5 considered by giodini et al .. with a total mass of @xmath83 , [email protected] is among the least massive fossil systems selected in x - rays , whereas [email protected] is at the boundary between the group and cluster mass regimes ( cf .
vikhlinin et al .
1999 ; sun et al .
2009 ; zibetti et al . 2009 ; dmocls et al . 2010 ; voevodkin et al .
2010 ) . as a reference , rxj105453.3@xmath14552102 at @xmath84 , with a total mass of @xmath85 ( aguerri et al .
2011 ) , is among the most massive fossil systems552102 is not considered as a fossil system by aguerri et al .
( 2011 ) , since its magnitude gap is equal to @xmath86 in the observed - frame @xmath70-band .
the bright end of the galaxy luminosity function of this cluster is populated by old , passively evolving objects ( aguerri et al . 2011 ) , so that the introduction of a @xmath61-correction is expected not to significantly modify the magnitude gap in the rest - frame @xmath70-band . however , the fossil criteria fixed by jones et al . ( 2003 ) are somewhat arbitrary and depend on the way magnitudes are measured ( see aguerri et al .
2011 for a discussion ) .
in addition , the likely contribution from diffuse light to the bcg emission is hard to be fully accounted for ( see zibetti et al .
2009 for a discussion ) .
thus , we consider rxj105453.3@xmath14552102 as a fossil system as we did for rxcj2315.7@xmath360222 at @xmath87 with @xmath88 in the rest - frame @xmath89-band ( dmocls et al .
2010 ) . ] .
[ cols="<,^,^,^,^,^,>,>,^ " , ] denomination , spectroscopic redshift ( @xmath90 ) , total mass ( @xmath2 ) , size ( @xmath59 ) , observed @xmath5-band and rest - frame @xmath70-band magnitude gaps between the first two brightest member galaxies ( @xmath91 and @xmath92 , respectively ) , total number of galaxies brighter than the completeness magnitude ( @xmath93 and @xmath94 for [email protected] and [email protected] , respectively ) within @xmath59 ( @xmath95 ) , numbers of expected ( i.e. , the photometric members after statistical subtraction of field galaxies ) and spectroscopic member galaxies within @xmath59 ( @xmath96 and @xmath97 , respectively ) are listed for each fossil group .
we compute the galaxy stellar mass function ( gsmf ) of the two fossil groups in bins of 0.5 dex in stellar mass ( @xmath98 ) . only member - candidate galaxies with a stellar mass above the completeness mass at the redshift of each group
are considered .
the completeness mass is equal to @xmath99 or @xmath100 for , respectively , all member - candidate galaxies of [email protected] or those among them that are classified as passively evolving . ] .
it is equal to @xmath101 or @xmath102 for , respectively , all member - candidate galaxies of [email protected] or the passively evolving ones among them .
we applied a correction for the expected number of interlopers , based on the statistics of the galaxies outside x - ray emitting bound structures in the same photometric - redshift slice , which define the coeval field of either group .
this holds whether or not galaxies are additionally selected according to their star formation activity . in each case , the field gsmf was normalized by the group - to - survey volume ratio and then subtracted to the gsmf of the member - candidate galaxies to obtain the gsmf of the ( bona fide ) member galaxies of each fossil group .
counts per stellar mass bin are normalized to the total mass @xmath2 of each fossil group , to enable comparison with the composite gsmf of the 40 ( 11 ) _ cosmos _ x - ray selected groups at @xmath9 with @xmath103 ( @xmath104 ) from giodini et al .
( in preparation ) .
these two composite gsmfs are computed in an analogous way , but include the number counts from all selected groups with similar masses to that of either fossil group ; they are normalized to the sum of the total masses of the individual groups .
figure 3 shows the normalized gsmfs of the two fossil groups and the normalized composite gsmfs of the two reference samples of x - ray selected groups at @xmath9 for all ( bona fide ) member galaxies and for those classified as passively evolving .
comparison of these normalized gsmfs reveals that about 80% of the galaxies with @xmath105 are passively evolving in the two fossil groups at @xmath55 .
conversely , passive galaxies start dominate the gsmf of the coeval , mass - complete sample of x - ray selected groups at much higher stellar masses , with a fraction of 80% passive galaxies being reached only at @xmath106 .
this holds although the fraction of passively evolving galaxies increases towards larger stellar masses across the entire range in stellar mass .
interestingly , half of the galaxies with @xmath107 do not show evidence of ongoing star - formation activity for similarly massive , optically selected groups at @xmath108 ( wilman et al .
2008 ) .
if the fraction of star - forming ( `` blue '' ) galaxies is a proxy for the dynamical youth of a galaxy system at fixed total mass ( e.g. , tully & shaya 1984 ; balogh , navarro , & morris 2000 ; ellingson et al .
2001 ; zapata et al . 2009 ; braglia et al .
2009 ) , the previous difference suggests that the two fossil groups assembled their dm haloes earlier than the average system among the similarly massive groups at @xmath9 .
the dominance of passively evolving or quiescent galaxies at @xmath105 is confirmed , in a more direct way , by the @xmath12 vs @xmath5 colour magnitude diagram of either fossil group ( see fig .
this diagram is obtained using subaru photometry ( see ilbert et al .
2009 ) . for @xmath109 , the @xmath12 colour nicely brackets the redshifted spectral feature at @xmath110 ( rest frame ) , whose amplitude increases together with the time since the last main episode of star formation but also with increasing metallicity ( e.g. worthey 1994 ) .
both fossil groups exhibit a well defined red - sequence , and a lack of galaxies within @xmath7 that are bluer than this locus by @xmath111 .
most of these blue galaxies are at least @xmath112 magnitudes fainter than the bcg of either fossil group in the @xmath5-band . moving out to @xmath59 , the bona fide member galaxies with a @xmath12 colour at least 0.5 mag bluer than the bcg one are still a few .
this confirms that most of the luminous galaxies within the bound region of either fossil group are old and passively evolving or quiescent , modulo dust attenuation ( cf .
wilman et al .
2008 ) .
the dominance of galaxies with a very low star - formation activity among those that are about two to four magnitudes fainter than the bcg is confirmed for the fossil groups rxcj2315.7@xmath360222 and rxcj0216.7@xmath364749 ( dmocls et al . 2010 ) , awm4 ( zibetti et al . 2009 ; see fig . 2 in koranyi & geller 2002 ) , rxj1416.4@xmath142315 ( see fig . 2 in cypriano , mendes de oliveira , & sodr 2006 ) , and the fossil group candidate rxj1520.9@xmath144840 ( eigenthaler & zeilinger 2009 )
. however , in the literature there is no comparative study of the galaxy populations of fossil and non - fossil systems of similar masses down to such fainter magnitudes , based on multi - band photometry or spectroscopy .
finally , we note that two member - candidate galaxies with @xmath113 and @xmath114 exhibit colors that are much redder than average color of the red - sequence of [email protected] ( @xmath115 and @xmath116 , respectively , vs @xmath117 ) . these two very faint and red galaxies ( about 6.3@xmath118-mag fainter and 0.4@xmath119 redder in @xmath12 than the bcg ) are located at projected distances of 110 and @xmath120 from the bcg ( or the cluster center ) ,
member - candidate galaxies with analogous photometric properties exist in the fossil group candidate rxj1548.9@xmath140851 ( eigenthaler & zeilinger 2009 ) .
in order to understand why the bulk of the galaxies with @xmath105 are dominated by old , passively evolving stellar populations in the two fossil groups at @xmath55 , we need to estimate if there was enough time for the last formed stars to age once their host galaxy had reached the bound region of either group .
hence , we need to estimate the following timescales : the time elapsed since half the dm halo of each fossil group was in place ( i.e. , the so - called `` assembly time '' ) , and the time elapsed since the last significant infall of blue galaxies with @xmath105 occurred .
these two timescales need to be compared with the minimum life - time of the stellar populations characterizing old , passively evolving ( i.e. , red - sequence ) galaxies , which is about 1@xmath121 .
we now make the conservative assumption that the two fossil groups behave as other similarly massive groups observed at @xmath55 .
furthermore , we assume that groups do not significantly increase their total masses between @xmath122 and @xmath55 . ] , in a statistical sense . by doing so
, we can use results on the median formation redshifts of dm haloes obtained from the simulations of gao et al .
( 2004 ) or different analytical models that are reproduced in fig . 5 of giocoli et al .
groups of similar total masses to [email protected] and [email protected] have assembled half of their dm halo at typical redshifts of @xmath123 and @xmath124 , respectively , which correspond to similar lookback times of @xmath125 at @xmath0 and @xmath39 . according to the computations of dynamical friction and galaxy - merging timescales of boylan - kolchin , ma & quataert ( 2008 , their fig .
1 ) , galaxies with @xmath126 or @xmath127 need about 6 or 3 gyr , respectively , to reach an orbit with maximum group - centric distance of @xmath7 from @xmath59 for the first time . since the simulations of boylan - kolchin et al . mostly neglect the effect of the central galaxy in a given dm halo on the merging time - scale , and do not take into account the drag on ambient gas ( e.g. ostriker 1999 ) , we infer that the epoch of last infall of blue galaxies with @xmath105 can date back to less than 36 gyr with respect to the lookback time of either fossil group .
the time of last infall of blue , intermediate - mass galaxies is comparable to the median time elapsed since the assembly of a group as massive as either fossil group , if not longer than that .
thus , we can assume that most of the galaxies with @xmath105 have resided within the bound region of either fossil group since its assembly ( cf .
von benda - beckmann et al . 2008 ; dariush et al . 2010 ) .
this `` residence '' time of @xmath125 is longer than the typical life - times of young stars , which are responsible for the bulk luminosity shortward of the 4000 - break , when they are present .
hence , there is enough time to observe a dominating population of old , passively evolving galaxies with @xmath105 in the two fossil groups at @xmath55 .
this is even more the case if the assembly time of a fossil group is shorter than the median assembly time for all coeval , similarly massive groups ( donghia et al .
2005 ; dariush et al .
environmental effects could be responsible for the raise of the population of old , passively evolving galaxies in the two fossil groups ( see boselli & gavazzi 2006 for a recent review ) .
member - candidate galaxies of either fossil group which are satellites at @xmath55 could reach stellar masses of 1@xmath128 before their star formation turned to its end .
we note that , in spite of the generally held , negative feedback exerted by the group environment on the star formation activity of member galaxies , such an activity seems to be enabled in 20%50% of the galaxies with @xmath107 in groups at similar redshifts and with similar masses ( fig . 2 ; giodini et al .
in preparation ; see also wilman et al .
2008 and references therein ) .
it is easy to understand how tidal interactions ( e.g. , spitzer & baade 1951 ; richstone 1976 ; farouki & shapiro 1981 ; merritt 1983 , 1984 ; icke 1985 ; miller 1986 ; byrd & valtonen 1990 ; valluri & jog 1990 ) or harassment ( moore et al . 1996 ) could have inhibited the growth of satellite galaxies up to @xmath129 .
it is less certain how the same interaction processes could have fostered the growth of more massive satellite galaxies ( e.g. , keel et al .
1985 ; kennicutt et al .
1987 ; byrd & valtonen 1990 ; mihos et al . 1992 ; henriksen & byrd 1996 ; lake , katz & moore 1998 ; iono , yun & mihos 2004 ; braglia , pierini & bhringer 2007 ) .
an earlier assembly time of fossil groups with respect to non - fossil groups of similar masses ( donghia et al .
2005 ; dariush et al .
2007 ) can justify the lower fraction of star - forming galaxies with @xmath130 ( see sect .
4.1 ) . as discussed by donghia et al .
( 2005 ) , this is also consistent with the larger x - ray - to - optical luminosity ratios of fossil groups with respect to similarly massive non - fossil groups in the local universe claimed by jones et al .
in fact , in groups with earlier assembly times the hot gas was collected and started cooling earlier and at higher central densities .
however , no difference in the scaling relation between x - ray and optical luminosities is reported for a sample of seven fossil groups at @xmath131 and total masses @xmath2 of a few times @xmath24 ( voevodkin et al .
2010 ) , at variance with previous results ( jones et al . 2003 ; khosroshahi et al .
2007 ) . as mentioned in sect . 2
, the x - ray - to - optical luminosity ratio is rather high for [email protected] but normal for [email protected] , when compared to those of similarly massive groups at @xmath1320.5 .
interestingly , there is growing evidence that an environment characterized by high galaxy number density and low galaxy velocity dispersion , such as the one found in compact groups , plays a key role in accelerating galaxy evolution by enhancing star - formation processes in galaxies and favoring a fast transition to quiescence ( tzanavaris et al .
2010 ; walker et al .
2010 and references therein ) .
such an environment could represent the immediate progenitor of a fossil system .
this is straightforward if such a progenitor is a compact group , and compact groups are interpreted as transient phases in the dynamical evolution of looser groups ( barnes 1989 ) .
alternatively , an environment with the previous characteristics can be produced if the existence of fossil systems is primarily driven by the relatively early infall of massive satellites ( donghia et al .
2005 ; von benda - beckmann et al .
tentatively this may be achieved through preferential infall on orbits with typically lower angular momentum .
consistently , there is an indication of the presence of galaxies in radial orbits in the external region of the fossil cluster rxj105453.3@xmath14552102 ( aguerri et al .
2011 ) . in a future study
, we will explore if an efficient confinement of the hot , x - ray emitting gas ( mathews et al . 2005 ) could foster the rather large stellar mass fractions of the two fossil groups under study ( cf .
g09 ) .
an earlier assembly time for fossil groups with respect to as massive non - fossil groups is consistent with the observational results discussed in sect .
3 ; it is also predicted by simulations in a statistical sense ( donghia et al .
2005 ; dariush et al . 2007 ) .
the evidence is a gsmf dominated by old , passively evolving systems at @xmath105 ( sect .
3.1 ) plus the corresponding existence of a tight red - sequence ( sect . 3.2 ) .
since the two fossil groups at @xmath55 sit in very different density environments of the lss ( sect .
2 ) , the conditions leading to the formation of a fossil system , in particular to the shaping of its galaxy population , do not appear to be intimately connected to the surrounding lss .
this is also consistent with simulations ( von benda - beckmann et al .
2008 ; cui et al .
2011 ) . an early formation time was advocated by aguerri et al .
( 2011 ) to explain the dynamics and luminosity function of the member galaxies as well as the photometric properties of the bcg for the relaxed , massive ( @xmath133 ) fossil system rxj105453.3@xmath14552102 at @xmath84 . on the other hand , a deep optical image of the bcg of the low x - ray luminosity fossil - group candidate rxj1548.9@xmath140851
( santos , mendes de oliveira & sodr 2007 ) clearly reveals the existence of many shells ( eigenthaler & zeilinger 2009 ) . according to hernquist & spergel ( 1992 ) ,
many shell systems among elliptical and spheroidal galaxies may have originated through major mergers of disc galaxies with comparable masses rather than via minor mergers - between a massive elliptical galaxy and a small disc companion ( quinn 1984 ; hernquist & quinn 1988 , 1989 ) or a small elliptical / spiral companion ( dupraz & combes 1986 ) - or accretions ( thomson & wright 1990 ; thomson 1991 ) .
we speculate that the relatively early infall of massive satellites can better justify the formation of a rather massive fossil group in a denser region of the lss , like [email protected] , whereas compact groups can more likely be the progenitors of observed low - mass and isolated fossil groups , like [email protected] . on one hand
, the lss surrounding the progenitor of a fossil group could be responsible more for the growth of the total mass of the system than for the growth of the stellar mass of its bcg ( see khosroshahi , ponman & jones 2006 ) .
this is consistent with both the shallow dependence of the median luminosity of bcgs on the total mass of the parent halo observed in the local universe ( yang , mo & van den bosch 2008 ) and the result from the hydrodynamic simulations of cui et al .
( 2011 ) that x - ray bright groups are disproportionally rare in low - density regions .
on the other hand , it is reasonable to assume that isolated fossil groups can be identified as such for longer times than non isolated ones of similar mass , because it is more difficult either that their magnitude gap is replenished by infalling galaxies , or that they fall into larger bound structures . in a low density environment ,
a loose group has more chances to keep its identity and dynamically evolve into a fossil group after a compact group phase ( cf .
barnes 1989 ) . either in the case of early assembly
( donghia et al . 2005 ; dariush et al .
2007 ) or in the case of final evolution of a loose group ( barnes 1989 ) , a rapid build - up of the stellar mass and the following quick end of the star - formation activity in most of the brightest member galaxies need to be invoked . interestingly , this is what compact groups in the local universe seem to have experienced ( tzanavaris et al .
2010 ; walker et al .
2010 and references therein ) .
if two channels lead to the formation of a fossil group and there is a sort of density environment bias on identification and inference on the progenitor from observations , the stellar populations of the bcg of a fossil group do not necessarily have to be remarkably different from those of field , massive early - type galaxies of similar optical and x - ray luminosities at @xmath134 ( cf .
la barbera et al .
furthermore , the size of environmental effects on typical red - sequence galaxies is still a controversial issue , and might be quite small at large stellar masses ( cf .
e.g. , thomas et al . 2005 , 2010 ; pannella et al .
2009 ; cooper et al .
hence , there might be less tension between our conclusions and the result of la barbera et al .
( 2009 ) .
future multi - wavelength studies of statistical samples of fossil groups with different masses and redshifts ( e.g. , santos et al . 2007 ) , located in different density regions of the lss , will provide a test of our interpretation and hypotheses . here
we discuss if the contribution from low - surface brightness , diffuse stellar emission not associated with galaxies ( the so - called intracluster light - icl ) could explain the rather low value of @xmath81 found in sect . 2 for the more massive fossil group , [email protected] .
the impact on the total stellar mass budget of the icl has already been pointed out in the literature , and with different scales , from observations ( cf .
e.g. , zibetti et al . 2005 and gonzalez et al .
2007 ) . in simulations ,
most of the icl seems to be associated with the build - up of the bcg ( e.g. , murante et al .
2007 and references therein ) .
the expected amount of icl is uncertain as well as the observed one - 3% to 30% of all stars in groups at @xmath35 ( e.g. , kapferer et al .
2010 and references therein ) - and seems to be independent of the mass - scale of the galaxy system ( e.g. , dolag , murante & borgani 2010 and references therein ) .
groups of galaxies that formed early ( as fossil groups , potentially ) might be particularly effective at producing icl , as interactions in the group environment begin to liberate tidal material .
galaxy systems as massive as the two fossil groups under study have smaller velocity dispersions than do massive clusters , meaning that interactions on the group scales can be slow and damaging as opposite to the impulsive high speed encounters , e.g. by harrassment , on the cluster scale .
these slow tidal interactions can strip material from rotating disc galaxies into loosely bound tidal structures , forming long tails of stars and gas ( donghia et al .
it is then interesting to estimate this additional amount of unbound stars that can escape detection in the present data .
in particular , we use the approximations developed in donghia et al .
( 2010 ) to compute the fraction of icl that a disc galaxy with a dynamical mass equal to @xmath135 ( of which @xmath136 is in the disc ) produces passing on a parabolic orbit within the potential of a typical group .
such a galaxy has a stellar mass of the order of @xmath137 .
figure 5 shows that the smaller the pericentric distance the larger the fraction of light stripped into the intergalactic medium .
the fraction of icl computed for the model galaxy is given by the ratio of the number of unbound stars to the total amount of stars bound to all the galaxies of the group .
this galaxy , passing at 30 - 40 kpc from the group center , can contribute up to 3% of the total light in the group galaxies .
hence , slow tidal interactions likely provide a marginal contribution to the total stellar budget of a low - mass fossil group as [email protected] , but a non - neglible one for a fossil group as massive as [email protected] . given these rough estimates
, we will estimate the icl contribution to the total stellar luminosity of the two fossil groups under study in the future , thanks to the deep b- and r - band photometry recently obtained by us with the wide field imager ( baade et al .
1999 ) , mounted at the cassegrain focus of the mpg / eso @xmath138 telescope in la silla , chile .
for the first time , the galaxy stellar mass function ( gsmf ) was computed , down to a stellar mass @xmath139 and in a robust statistical way , for two fossil groups discovered at @xmath55 : [email protected] and [email protected] .
these groups have also well - determined total masses ( @xmath2 ) equal to @xmath3 and @xmath4 , respectively .
this was possible because of the high - quality multi - wavelength information available for the @xmath38 area of the cosmic evolution survey ( _ cosmos _ ; scoville et al .
2007 ) where the two systems were initially detected in x - rays ( finoguenov et al . in preparation ; g09 ) .
in particular , the identification as fossil and the selection of the member - candidate galaxies of either group rest on robust photometric redshifts , and @xmath8 limited spectroscopy .
interestingly , the large - scale structure ( lss ) in which the two fossil groups are embedded is very different : [email protected] appears isolated , whereas [email protected] belongs to a lss that covers the entire @xmath38 area of the _ cosmos _ ( corresponding to a cross size of about @xmath73 at @xmath55 ) and is traced by a total of 28 x - ray emitting groups .
comparison of the gsmf of these fossil groups and the composite gsmf of the _ cosmos _ x - ray emitting groups at @xmath9 that span a similar range in total mass ( giodini et al . in preparation ) highlights a lack of star - forming galaxies with @xmath105 in the two fossil groups .
consistently , the distribution of their photometric member galaxies in the @xmath12 vs @xmath5 colour magnitude diagram reveals a well - defined red sequence and a lack of significantly bluer , luminous galaxies out to the virial radius @xmath59 of either fossil group .
in addition , the total stellar mass fraction of these groups within @xmath60 is relatively large for their total masses ( g09 ) , in particular for the less massive one ( i.e. , [email protected] ) .
this evidence suggests that the overall conversion of ( cold ) gas into stars was accelerated and/or more efficient in fossil groups with respect to the average x - ray emitting group in the same mass range and at similar redshift .
the star - formation activity in member galaxies rapidly ceased afterwards .
at the same time , no significant infall of galaxies with @xmath13 took place in the last 36 gyr for either fossil group at @xmath55 .
if the processes shaping the galaxy component of a fossil group seem to be independent of the lss , the progenitor of an observed fossil group may depend on it . on the basis of the measured stellar mass fractions ( g09 ) and x - ray - to - optical luminosity ratios ( giodini et al . in preparation ) , we propose that compact groups are the most likely progenitors of observed low mass and isolated fossil groups , like [email protected] .
conversely , the relatively early infall of massive satellites likely originates massive fossil groups in dense regions of the lss , like [email protected] .
we thank the referee , paul eigenthaler , for a careful reading of the original version of the paper and the useful comments that improved the presentation of the paper .
dp acknowledges the kind and fruitful hospitality at the max - planck - institut fr extraterrestrische physik ( mpe )
. 99 aguerri j.a.l .
2011 , a&a , 527 , 143 arimoto n. , yoshii , y. 1987 , a&a , 173 , 23 baade d. et al . 1999 ,
msgnr , 95 , 15 balogh m.l . , navarro j.f . , morris s.l
. , 2000 , apj , 540 , 113 barnes j.e . , 1989 , nature , 338 , 123 boselli a. , gavazzi g. , 2006 , pasp , 118 , 517 boylan - kolchin m. , ma c .- p . , quataert e. , 2008 , mnras , 383 , 93 braglia f.g . , pierini d. , bhringer h. , 2007 , a&a , 470 , 425 braglia f.g . , pierini d. , biviano a. , bhringer h. , 2009 , a&a , 500 , 947 byrd g. , valtonen m. , 1990 , apj , 350 , 89 chabrier g. , 2003 , apj , 586 , l133 conroy c. , gunn j.e . , 2010 , apj , 712 , 833 cooper m.c .
, gallazzi a. , newman j.a . , yan r. , 2010 , mnras , 402 , 1942 cui w. , springel v. , yang x. , de lucia g. , borgani s , 2011 , mnras , in press ( arxiv:1102.4269 ) cypriano e.s . , mendes de oliveira c.l .
, sodr l. jr . , 2006 ,
aj , 132 , 514 dariush a. , khosroshahi h.g . ,
ponman t.j . , pearce f. , raychaudhury s. , hartley w. , 2007 , mnras , 382 , 433 dariush a. , raychaudhury s. , ponman t.j . , khosroshahi h.g . ,
benson a.j . ,
bower r.g . , pearce f. , 2010 , mnras , 405 , 1873 dmocls j. , pratt g.w . , pierini d. , arnaud m. , zibetti s. , donghia e. , 2010 , a&a , 517 , 52 dolag k. , murante g. , & borgani s. , 2010 , mnras , 405 , 1544 donghia e. , lake g. , 2004 , apj , 612 , 628 donghia e. , sommer - larsen j. , romeo a.d .
, burkert a. , pedersen k. , portinari l. , rasmussen j. , 2005 , apj , 630 , l109 donghia e. , besla g. , cox t.j . ,
hernquist l. , 2009 , nature , 460 , 605 donghia e. , vogelsberger m. , figuere - ciguere c. , hernquist l. , 2010 , apj , 725 , 353 dunkley j. et al . , 2009 ,
apjs , 180 , 306 dupraz c. , combes f. , 1986 , a&a , 166 , 53 eigenthaler p. , zeilinger w. w. , 2009 , astronomische nachrichten , 330 , 978 ellingson e. , lin h. , yee h.k.c . , carlberg r.g . , 2001 ,
apj , 547 , 609 elmegreen d.m . ,
elmegreen b.g . ,
chromey f.r . ,
fine m.s . , 2000 ,
aj , 120 , 733 evrard a.e . , metzler c.a . , navarro j.f . , 1996 , apj , 469 , 494 farouki r. , shapiro s.l . , 1981 , apj , 243 , 32 finoguenov a. et al . 2007 , apjs , 172 , 182 gao l. , white s.d.m .
, jenkins a. , stoehr f. , springel v. , 2004 , mnras , 355 , 819 giocoli c. , moreno j. , sheth r.k . , tormen , g. , 2007 , mnras , 376 , 977 giodini s. et al .
2009 , apj , 703 , 982 ( g09 ) giodini s. et al .
2010 , apj , 714 , 218 gonzalez a.h .
, zabludoff a.i . , zaritsky d. , 2007 , apj , 666 , 147 henriksen m. , byrd , g. , 1996 , apj , 459 , 82 hernquist l. , quinn p.j .
, 1988 , apj , 331 , 682 hernquist l. , quinn p.j . , 1989 , apj , 342 , 1 hernquist l. , spergel d.n . , 1992 , apj , 399 , l117 icke v. , 1985 , a&a , 144 , 115 ilbert o. et al . , 2009 , apj , 690 , 1236 iono d. , yun m.s .
, mihos j.c . , 2004 , apj , 616 , 199 jones l.r .
, ponman t.j . ,
forbes d.a . , 2000 ,
mnras , 312 , 139 jones l.r .
, ponman t.j .
, horton a. , babul a. , ebeling h. , burke d.j . , 2003 , mnras , 343 , 627 kapferer w. , schindler s. , knollmann s.r . , van kampen e. , 2010 , a&a , 516 , 41 keel w.c . , kennicutt r.c .
jr . , hummel e. , van der hulst j.m . , 1985 ,
aj , 90 , 708 kennicutt r.c .
, roettiger k.a . ,
keel w.c . , van der hulst j.m . ,
hummel e. , 1987 , aj , 93 , 1011 khosroshahi h.g . ,
ponman , t.j .
, jones l.r . , 2006 , mnras , 372 , 68 khosroshahi h.g . ,
ponman , t.j . ,
jones l.r . , 2007 , mnras , 377 , 595 knobel c. et al . , 2009 , apj , 697 , 1842 koekemoer a.m. et al . , 2007 , apjs , 172 , 196 komatsu e. et al . , 2009 , apjs , 180 , 330 koranyi d.m . , geller m.j . , 2002 ,
aj , 123 , 100 la barbera f. , de carvalho r.r .
, de la rosa i.g . , sorrentino g. , gal r.r . , kohl - moreira j.l . , 2009 , aj , 137 , 3942 lake g. , katz n. , moore b. , 1998 , apj , 495 , 152 leauthaud a. et al . , 2010 , apj , 709 , 97 lilly s.j .
et al . , 2007 ,
apjs , 172 , 70 mathews w.g .
, faltenbacher a. , brighenti f. , buote d.a . , 2005 , apj , 634 , l137 merritt d. , 1983 , apj , 264 , 24 merritt d. , 1984 , apj , 276 , 26 mihos j.c . ,
richstone d.o . ,
bothun g.d . , 1992 ,
apj , 400 , 153 miller r.h . , 1986 , a&a , 167 , 41 milosavljevi m. , miller c.j .
, furlanetto s.r .
, cooray a. , 2006 , apj , 637 , l9 moore b. , katz n. , lake g. , dressler a. , oemler a. , 1996 , nature , 379 , 613 murante g. , giovalli m. , gerhard o. , arnaboldi m. , borgani s. , dolag k. , 2007 , mnras , 377 , 2 ostriker e.c . , 1999 , apj , 513 , 252 pannella m. et al . , 2009 , apj , 701 , 787 pierini d. , maraston c. , bender r. , witt a.n .
, 2004 , mnras , 347 , 1 pierini d. , zibetti s. braglia f. , bhringer h. , finoguenov a. , lynam p.d .
, zhang y .- y . , 2008 , a&a , 483 , 727 ponman t.j . , bertram d. , 1993 , nature , 363 , 51 ponman t.j . , allan d.j .
, jones l.r . , merrifield m. , mchardy i.m .
, lehto h.j . ,
luppino g.a . , 1994 ,
nature , 369 , 462 quinn p.j .
, 1984 , apj , 279 , 596 richstone d.o
. , 1976 , apj , 204 , 642 romer a.k .
2000 , apjs , 126 , 209 sales l.v .
, navarro j.f .
lambas d.g . ,
white s.d.m . , croton d.j . , 2007 , mnras , 382 , 1901 salpeter e.e . , 1955 ,
apj , 121 , 161 santos w.a . , mendes de oliveira c. , sodr l. jr . , 2007 , aj , 134 , 1551 schechter p. , 1976 ,
apj , 203 , 297 schirmer m. , suyu s. , schrabback t. , hildebrandt h. , erben t. , halkola a. , 2010 , a&a , 514 , 60 scoville n. et al . , 2007 , apjs , 172 , 1 spitzer l. jr .
, baade w. , 1951 , apj , 113 , 413 sun m. , voit g.m . , donahue m. , jones c. , forman w. , vikhlinin a. , 2009 , apj , 693 , 1142 teyssier r. , moore b. , martizzi d. , dubois y. , mayer l. , 2011 , mnras , 414 , 195 thomas d. , maraston c. , bender r. , mendes de oliveira c. , 2005 , apj , 621 , 673 thomas d. , maraston c. , schawinsk , k. , sarzi m. , silk j. , 2010 , mnras , 404 , 1775 thomson r.c .
, 1991 , mnras , 253 , 256 thomson r.c . ,
wright a.e . , 1991 , mnras , 247 , 122 tully r.b . , shaya e.j .
, 1984 , apj , 281 , 31 tzanavaris p. et al . , 2010 , apj , 716 , 556 valluri m. , jog c.j . , 1990 , apj , 357 , 367 vikhlinin a. , mcnamara b.r . , hornstrup a. , quintana h. , forman w. , jones c. , way m. , 1999 , apj , 520 , l1 voevodkin a. , borozdin k. , heitmann k. , habib s. , vikhlinin a. , mescheryakov a. , hornstrup a. , burenin r. , 2010 , apj , 708 , 1376 von benda - beckmann a.m. , donghia e. , gottlber s. , hoeft m. , khalatyan a. , klypin a. , mller v. , 2008 , mnras , 386 , 2345 walker l.m . , johnson k.e . , gallagher s.c .
, hibbard j.e . ,
hornschemeier a.e . ,
tzanavaris p. , charlton j.c .
, jarrett t.h .
, 2010 , aj , 140 , 1254 wilman d.j .
et al . , 2008 , apj , 680 , 1009 worthey g. , 1994 , apjs , 95 , 107 yang x. , mo h.j .
, van den bosch f.c . , 2008 , apj , 676 , 248 zapata t. , perez j. , padilla n. , tissera p. , 2009
, mnras , 394 , 2229 zibetti s. , white s.d.m . , schneider d.p . , brinkmann j. 2005 , mnras , 358 , 949 zibetti s. , pierini d. , pratt g.w .
2009 , mnras , 392 , 525 | we report on two fossil groups of galaxies at redshifts @xmath0 and @xmath1 discovered in the cosmic evolution survey ( _ cosmos _ ) area . selected as x - ray extended sources , they have total masses ( @xmath2 ) equal to @xmath3 and @xmath4 , respectively , as obtained from a recent x - ray luminosity mass scaling relation .
the lower mass system appears isolated , whereas the other sits in a well - known large - scale structure ( lss ) populated by 27 other x - ray emitting groups .
the identification as fossil is based on the @xmath5-band photometry of all the galaxies with a photometric redshift consistent with that of the group at the @xmath6 confidence level and within a projected group - centric distance equal to @xmath7 , and @xmath8 limited spectroscopy .
both fossil groups exhibit high stellar - to - total mass ratios compared to all the x - ray selected groups of similar mass at @xmath9 in the _ cosmos_. at variance with the composite galaxy stellar mass functions ( gsmfs ) of similarly massive systems , both fossil group gsmfs are dominated by passively evolving galaxies down to @xmath10 ( according to the galaxy broad - band spectral energy distributions ) .
the relative lack of star - forming galaxies with @xmath11 is confirmed by the galaxy distribution in the @xmath12 vs @xmath5 colour magnitude diagram .
hence , the two fossil groups appear as more mature than the coeval , similarly massive groups .
their overall star formation activity ended rapidly after an accelerated build up of the total stellar mass ; no significant infall of galaxies with @xmath13 took place in the last 3 to 6 gyr .
this similarity holds although the two fossil groups are embedded in two very different density environments of the lss , which suggests that their galaxy populations were shaped by processes that do not depend on the lss .
however , their progenitors may do so .
we discuss why the late merging of a compact group is favoured over the early assembly as a formation scenario for the isolated , low - mass fossil group .
[ firstpage ] x - rays : galaxies : clusters galaxies : groups : general galaxies : evolution galaxies : luminosity function , mass function . |
as the development of optical clocks is currently pursued in many laboratories worldwide @xcite , a broad range of applications is taking shape with optical clocks on ground and in space employed for high precision tests of fundamental physics @xcite , chronometric levelling - based geodesy @xcite , improved rf standards for navigation @xcite and observations of cosmic radio sources with very - long baseline interferometry ( vlbi ) @xcite .
for all these purposes today s complex and bulky optical - clock experimental setups need to be re - engineered into more compact and power efficient systems , ensuring at the same time a high stability and accuracy , but also high operation reliability in critical environments @xcite , such as application in the field or even satellites .
a first step in the engineering challenge leading to space based optical clocks is to demonstrate transportable clocks .
these are interesting because frequency comparisons of today s best optical clocks can not be done through satellite links and tests with optical fiber links require dedicated equipment @xcite .
a frequency transfer standard will allow for comparing clocks at the accuracy level provided by the transportable clock @xcite .
in this paper we present the realization of a transportable strontium optical clock , based on the integration of two subsystems , a compact atomic source including a compact cooling and trapping laser set - up , and a transportable clock laser , providing the laser - cooled sample of strontium atoms and the ultra - stable radiation source for interrogating the clock transition , respectively .
the clock laser , transported by van from ptb in braunschweig to lens in florence was employed to perform high resolution spectroscopy of the clock transition @xmath2s@xmath3-@xmath4p@xmath5 of @xmath6sr in lamb - dicke regime and to characterize systematic frequency shifts of the optical transition . in the following
are presented in detail novel design solutions employed for each subsystem that allowed us to reduce size , weight and power consumption with respect to a traditional laser cooling apparatus .
the compact laser - cooling strontium source mainly consists of the following modules : the cooling laser sources , a frequency distribution breadboard and a vacuum system .
the three modules are connected through optical fibers and mounted directly on a 120@xmath7@xmath890@xmath7 cm breadboard .
the control electronics is hosted under the main breadboard in a 19 " rack ( size 60@xmath7@xmath860@xmath7@xmath8180@xmath7 cm ) .
cooling and trapping of strontium atoms is performed through a two - stage magneto - optical - trap ( mot ) on the dipole allowed @xmath2s@xmath3-@xmath9p@xmath10 transition and on the intercombination line @xmath2s@xmath3-@xmath4p@xmath10 , respectively ( see fig .
[ fig.livellisr ] ) .
all the employed lasers are based on semiconductor diodes @xcite .
the laser set is composed of a frequency doubled 300@xmath7mw 461@xmath7 nm laser , a 50@xmath7mw 689 nm laser , a 420@xmath7mw 813@xmath7 nm ( master + tapered amplifier ) laser and two @xmath1120@xmath7mw 679@xmath7 nm and 707@xmath7 nm repumper lasers .
these sources are used for the two - stage laser - cooling and subsequent optical trapping in a one - dimensional ( 1d ) lattice at the magic wavelength @xcite .
.[fig.livellisr],scaledwidth=50.0% ] all the optical beams for the first laser - cooling stage at 461 nm are produced in a compact module ( size 30@xmath7@xmath840@xmath7@xmath88@xmath7 cm ) .
the module contains ultra - stable mountings for mirrors , polarization cubes , plates , lenses and acousto - optic - modulators ( aoms ) to which the 461 nm radiation is delivered by fibre @xcite . in order to reduce the number of beam shaping optical elements used in the breadboard while maintaining a high diffraction efficiency on aoms
, an input beam @xmath12 diameter has been set to @xmath13 mm . as a result
we obtain high diffraction efficiency on aoms ( @xmath1480 % in single pass configuration ) without the use of any additional telescope , while still keeping the possibility of resolving the diffraction orders at short distances .
moreover with this choice a coupling efficiency into optical fibers of about 65 % is achieved .
then , the overall efficiency of a typical @xmath15 cm long optical path inside the breadboard , including the aom diffraction efficiency ( in single - pass configuration ) and the coupling efficiency into the output fiber , is typically @xmath16 .
the frequency detunings from the @xmath2s@xmath3-@xmath9p@xmath10 resonance for zeeman slower and mot beams , of @xmath17mhz and @xmath18mhz respectively , are obtained by using two aoms in single - pass configuration , driven at 170 mhz ( @xmath19 order ) and 110 mhz ( @xmath20 order ) , respectively .
all output beams can also be completely shut off by fast and compact mechanical shutters .
a beam for frequency stabilization of the first stage cooling light is obtained from a double - pass configuration through an aom driven at 75@xmath7mhz and sent to the atomic beam used to load the mot ( see fig .
[ fig : vacuum_system ] ) . in order to extract an error signal from spectroscopy on the cooling transition ,
the driving rf signal is also frequency modulated at @xmath11 10@xmath7khz with a peak to peak deviation of @xmath1110 mhz . a low power ( @xmath11 1 mw ) resonant beam for absorption imaging is generated in a similar double pass configuration .
for both double - pass aoms a cats - eye configuration has been realized by focusing the optical beam on the retro - reflecting mirror with a lens at distance f = 60 mm from aom and retro - reflection mirror .
the atomic sample is trapped in a vacuum system ( size 1200@xmath7@xmath840@xmath7 cm @xmath21 36@xmath7 cm ) , see fig .
[ fig : vacuum_system ] ) consisting of the oven region , where a collimated sr atomic beam is produced , a zeeman slower and the science region where the sr atoms are subsequently zeeman slowed , trapped in a two - stage mot and eventually transferred in a 1d vertical optical lattice for clock spectroscopy .
the oven region is pumped by a 40 l / s ion pump ( pressure during operation @xmath22 pa ) , the science region is evacuated by a 55 l / s ion pump and a titanium sublimation pump ( pressure @xmath23 pa ) . a differential pumping tube ( internal diameter 5 mm , length 7.5 cm ) ensures to maintain the pressure difference between the two regions . in the oven region
, atoms are sublimated by a compact and efficient dispenser @xcite based on a heater placed in vacuum , providing an atomic flux intensity of @xmath24 s@xmath25sr@xmath25 ( @xmath6sr atoms ) at the oven temperature of @xmath26c with a total power consumption of 26 w. the atomic beam is collimated by a nozzle of about 120 capillaries ( internal diameter 200@xmath27 m , length @xmath28 mm ) .
the high collimation ( @xmath29 mrad ) and flux allow the laser stabilization to be performed with high signal - to - noise employing a simple transverse fluorescence spectroscopy on the atomic beam in the oven chamber ( see fig . [
fig : vacuum_system ] ) .
the atomic beam propagates along a 23@xmath7 cm long tube externally wrapped with coils for zeeman slowing .
the magnetic field shape for zeeman slowing has been designed to smoothly match the off - axis component of the mot coils field ( see fig . [
fig : slowing_dynamics_final ] ) @xcite .
the first part of the zeeman slower ( before the inversion of the field given by the mot coils ) has a length of @xmath30 cm and it operates with a power consumption of 17@xmath7w .
the decelerated atomic beam is trapped in a mot at the center of a science cell with eight cf40 and sixteen cf16 optical windows .
two custom cf150 flanges host the pair of coils ( in anti - helmholtz configuration as needed for mot operation ) outside the mot chamber 2.6@xmath7 cm away from the atoms . in this configuration a magnetic field gradient of @xmath31
mt / m is obtained with a total power consumption of @xmath32 w. with these levels of power consumption for oven , zeeman slower and mot coils we could avoid the complication of water cooling , with a typical chamber temperature of 318 k. to make the alignment of the mot beams long - term stable , the cooling beams at 461 nm and 689 nm are delivered by three dichroic fiber - coupled beam expanders fastened to the science cell . a system of setting screws and counter screws allows fine alignment and locking of the mot beams .
similarly , the mot beam retroreflectors , repumping , imaging and clock spectroscopy output - couplers are fastened onto the science cell .
the alignment of the mot beams has been maintained for more than one year without any further adjustment .
the main clock laser breadboard is based on two diode lasers in a master - slave setup .
the master is a filter stabilized extended cavity laser with resonator length 10 cm @xcite .
about 0.5 mw of the master - laser power is used to injection lock the slave laser .
the breadboard has a total size of 60 cm @xmath21 45 cm @xmath21 10 cm , its design is similar to frequency distribution breadboard presented in the previous section . a 200 mhz aom in double - pass configuration is used to bridge the frequency gap between the optical resonator with free spectral range of 1.5 ghz and the frequency of the clock transition .
this aom is used to scan the frequency of the laser and also to enable frequency feedback when the laser is locked to the clock transition .
two output ports for clock spectroscopy and for counting of the laser frequency provide a maximum optical power of about 2 mw each .
both ports are provided with aoms for switching which can also be used to stabilize the optical path length of the fibers . the interferometric setup for this stabilization is included on the breadboard @xcite .
the laser was locked to a transportable high finesse cavity whose transportability had already been demonstrated within the soc project @xcite .
in order to develop a more compact and low power consumption experimental system for the production of ultra - cold strontium atoms , several design solutions have been adopted , together with an optimization of the efficiency of each cooling and trapping step . in order to slow down the atoms sublimating from the oven from 430@xmath33 to below 130@xmath33 , an optical beam at @xmath34 nm , circularly polarized , is sent counter propagating to the atomic beam .
the laser beam has an optical power of @xmath35mw , an initial @xmath12 radius of @xmath36 mm , focused after 1@xmath7 m to compensate the absorption of photons in the slowing process .
atoms are kept in resonance during the slowing by compensating the variation of doppler shift with the proper zeeman shift from the magnetic field provided by the zeeman slower .
the sample of cold @xmath6sr atoms is produced through two laser - cooling stages .
the first stage consists of a mot operating on the @xmath2s@xmath3-@xmath9p@xmath10 transition at @xmath34 nm .
this so called blue mot is realized by three pairs of counter - propagating optical beams , circularly polarized , detuned by @xmath37mhz , with a saturation parameter for each beam of @xmath38 and a magnetic field gradient of @xmath39 mt / m .
the blue mot capture velocity is @xmath40 where @xmath41 is the maximum acceleration exerted by cooling light , with @xmath42 the wavevector of the photons at @xmath43 nm , @xmath44mhz the natural linewidth of the @xmath2s@xmath3-@xmath9p@xmath10 transition and @xmath45 the atomic mass of strontium 88 . as expected we find that the zeeman slower increases the loading rate of the mot , resulting in a net increase of the blue mot population by a factor of @xmath46 ( see fig . [
fig : bluemot_loading_cropped ] ) .
the @xmath2s@xmath3-@xmath9p@xmath10 transition used for the blue mot is not perfectly closed due to the decay channel of the 5p@xmath47p@xmath10 state towards the 4d@xmath47d@xmath48 state , which has a lifetime of @xmath49 ms @xcite .
atoms in the latter state may decay to the ground state through the 5p@xmath50p@xmath10 within less than 1@xmath7ms or may decay to the metastable 5p@xmath50p@xmath48 state and be lost . in order to recycle the atoms stored in the metastable 5p@xmath51p@xmath48 state a 10 mw repumper laser at 707@xmath7
nm is used to pump these atoms in the 6s@xmath51s@xmath10 state .
an additional 10 mw laser at 679@xmath7 nm is necessary to deplete the @xmath0p@xmath5 state since it is also coupled to the 6s@xmath51s@xmath10 state .
the repumping laser beams are superimposed , expanded to 10 mm of diameter and sent to the blue mot and retroreflected .
the repumping increases the atom number in the blue mot by a factor @xmath32 .
we studied the loading dynamics of the blue mot population @xmath52 , given by @xmath53 , where @xmath54 is the effective loading rate of atoms in the mot , @xmath55 is the linear loss rate ( mainly due to background collisions when repumpers are operating ) and @xmath56 is the the coefficient for two - body collisional loss @xcite . the previous equation with the initial condition @xmath57 has the standard solution @xmath58 where @xmath59 is the stationary number of trapped atoms , @xmath60 is the effective trap loading time and @xmath61 is the collisional loss fraction @xcite . with repumpers we measured @xmath62 @xmath6sr atoms , corresponding to an atomic density of @xmath63 @xmath64 , @xmath65ms and @xmath66 s@xmath25 , leading to a rate of captured atoms of @xmath67 .
considering that the rate of atoms effused by the oven in the solid angle covered by the zeeman slowing beam is @xmath68 , about @xmath69 % of the atoms are actually trapped into the blue mot .
this efficiency is mostly determined by the optical pumping into the 4d@xmath47d@xmath48 state , where atoms can be considered lost from the slowing dynamics since the state lifetime is comparable to the slowing time scale .
the probability of an atom of decaying into the ground state after the absorption of @xmath70 photons is simulation of the slowing dynamics with the field profile given by the zeeman slower coils and the off - axis quadrupole field of the mot coils . ] @xmath71 where @xmath72 s@xmath25 and @xmath73 s@xmath25 are the decay rates of the state @xmath2p@xmath10 into @xmath2s@xmath5 and @xmath2d@xmath48 states , respectively .
therefore the actual fraction of slowed atoms can be estimated through @xmath74 where @xmath75 is the minimum final velocity of atoms at the end of the slowing dynamics , @xmath76 is the maximum atom velocity that can be slowed down ( according to the results of numerical simulation shown in fig .
[ fig : slowing_dynamics_final ] ) , @xmath77 is the number of photons necessary to reduce the atom velocity from @xmath78 to @xmath79 , @xmath80 is the recoil velocity due to the absorption of a photon at @xmath34 nm and @xmath81 is the velocity distribution in an effusive atomic beam at @xmath82c @xmath83 therefore the estimation of the zeeman slower efficiency @xmath84 given by eq .
[ eq : zeeman_slower_efficiency ] is close to the measured value .
additionally the blue mot population enhancement factor due to the operation of zeeman slower can be estimated from the ratio between the actual fraction of slowed atoms @xmath85 and the fraction of trapped atoms without zeeman slower @xmath86 which is close to the measured value of @xmath46 .
the final temperature of the blue mot is minimized by continuously reducing the optical power of the cooling beams from the initial total saturation parameter of @xmath87 to a final value of @xmath88 in @xmath89ms .
this power - reduced phase lasts for 50@xmath7ms , leading to a final temperature of the blue mot of about @xmath90mk measured with absorption imaging ( see fig .
[ fig : laser_cooling ] ) .
the second laser - cooling stage is performed by using the @xmath2s@xmath3-@xmath4p@xmath10 intercombination transition at 689@xmath7 nm ( red mot ) .
the optical beams needed for this phase are superimposed on the blue mot beams through the three dichroic beam expanders .
the @xmath12 diameter of the beams for the red mot is 10@xmath7 mm and the total intensity incident on the atoms is @xmath91 .
the atomic sample at the end of the first cooling stage ( @xmath92mk ) has a doppler width of @xmath90mhz , too large to be efficiently transferred into the red mot operating on a @xmath93khz natural linewidth transition .
for this reason the spectrum of the 689@xmath7 nm laser is artificially broadened @xcite .
this is realized by modulating the radio frequency driving of the aom setting the frequency detuning of the cooling beams .
the modulation frequency is @xmath94khz , with a span of @xmath95mhz , leading to an optical spectrum of @xmath11 80 sidebands , with a saturation parameter of 400 for each one , with the closest to resonance by @xmath96khz .
this so called red mot broadband cooling phase lasts for 60@xmath7ms and more than 70@xmath97 of the blue mot population is transferred into the red mot . at the beginning of this phase the zeeman slower field is turned off and the mot magnetic field gradient is only 20 mt / m in order to have all the atomic velocity classes resonant with the cooling light and it is linearly increased to 70 mt / m in a time interval of @xmath35ms to compress the atomic sample . with a blue mot loading time of @xmath98
ms the final population of the broadband red mot is @xmath99 atoms , with an atomic density of @xmath100 and a temperature of @xmath101k .
absorption imaging _ in - situ _ of the atomic sample at the relevant phases of the laser cooling sequence lasting a time @xmath102 , with a constant blue mot loading time of 400 ms .
a ) blue mot full power .
b ) blue mot low power .
c ) red mot broadband .
d ) red mot single frequency.,scaledwidth=60.0% ] the second laser - cooling phase is completed by employing a single frequency red mot , with a detuning of @xmath96khz , a constant magnetic field gradient of 20 mt / m and reducing the total intensity of the cooling beams down to @xmath103 in @xmath94ms .
the single frequency red mot phase produces a sample of @xmath104 atoms , with an atomic density of @xmath105 and a temperature of @xmath106k . since the gravity force is comparable to the radiation force of the red mot beams , the atomic sample sags down from the center of the mot quadrupole field , assuming an half - disk - like shape with a vertical and horizontal diameter of @xmath107 m @xcite . in order to have a long doppler - free interrogation time of the clock transition ,
the laser - cooled strontium sample is trapped into a vertical lattice realized by retroreflecting a @xmath108mw laser beam near the magic wavelength for @xmath6sr @xmath109 nm @xcite . at this wavelength
the light shifts of the @xmath2s@xmath5 and @xmath0p@xmath5 level are equal , so that the frequency of the clock transition @xmath2s@xmath3-@xmath4p@xmath5 is not light - shifted by the lattice . the lattice laser output is coupled into a single mode fiber delivering an optical beam with @xmath12 diameter of @xmath110 mm , which is expanded and focused into the red mot by a @xmath111 mm focal length lens .
the latter is mounted on a 3-axis translation stage with micro - metric actuators , so that the alignment of the beam onto the red mot can be finely tuned .
the resulting beam waist radius on the atomic cloud is @xmath112 m . after the focus ,
the divergent lattice beam is then collimated and retroreflected by means of a dichroic mirror .
the latter is employed to couple into the lattice the clock probe beam at @xmath113 nm . the resulting beam diameter of the clock light on the atoms is @xmath114 m .
taking into account the power losses due to the telescope , focusing optics and cell windows the estimated lattice trap depth is @xmath115 ( corresponding to @xmath116k , in temperature units ) , where @xmath117 is the photon recoil energy with the wavevector @xmath118 of the lattice light and @xmath119 the atomic mass of @xmath6sr . at this depth of the potential
the estimated longitudinal trap frequency is @xmath120khz .
the lattice is continuously operating and at the end of the second laser - cooling stage about @xmath121 atoms remain trapped , populating @xmath122 sites , for a total extension of @xmath123 m corresponding to the vertical size of the red mot .
the @xmath12 radius of the atomic cloud trapped in the lattice is measured with absorption imaging to be @xmath124 m .
the longitudinal @xmath12 radius of a single site , of the order of @xmath125 , can not be resolved by absorption imaging and is estimated from the size @xmath126 of the wave function of atoms populating the ground longitudinal vibrational state ( is a good approximation since we have @xmath127 ) .
thus the trapping volume for lattice site is @xmath128 . in order to keep the shift and broadening effects on the clock transition due to atomic collisions @xcite
, we reduced the number of trapped atoms to below @xmath129 ( less than 10 atoms per site ) by reducing the blue total mot loading time to 100 ms , leading to a peak atomic density per site below @xmath130 .
the number of atoms is controlled by varying the duration of the blue mot stage and by finely tuning the position of the gate valve separating the oven from the science region .
the lifetime of atoms trapped into the lattice is measured to be @xmath131s , limited by background gas collisions . both power and frequency of the lattice laser
are not stabilized .
the relative rms power fluctuation is below the @xmath132 % level .
the lattice wavelength is tuned near to the magic wavelength of @xmath6sr @xcite , monitored by a wave - meter and measured to be stable at @xmath133 nm over a time scale of several hours . as a first test of the performance of the transportable clock laser after the two - days ,
1300 km - long transportation , the laser was compared with a stationary clock laser @xcite .
[ fig : beatclocklaser ] shows the beat note recorded just after re - installation of the clock laser in the lab .
the installation process took about one day , mainly do to re - thermalization of the transportable cavity , which was not actively temperature stabilized during transportation .
the beat note shows a linewidth of the order of 1 hz compatible with the laser frequency stability of @xmath134 at 1 s. beatnote of the transportable laser with a stationary clock laser at 698 nm .
the estimated emission linewidth for each laser is 1 hz.,scaledwidth=60.0% ] the comparison with the stationary clock has also been used to estimate the absolute frequency of the transportable clock laser with an uncertainty of less than @xmath135mhz .
spectra of the @xmath6sr clock transition for different values of the mixing magnetic field @xmath136 , clock probe beam intensity @xmath137 and excitation pulse length @xmath138 .
the frequency axes have arbitrary offsets .
the clock resonance is fitted with a lorentzian function and the obtained fwhm is compared to the fourier limit linewidth @xmath139 and to the rabi linewidth @xmath140 , where @xmath141 and @xmath142 are the second order zeeman shift and clock probe light shift , respectively ( see the text ) .
a ) shows a typical search - scan spectrum with the maximum number of atoms loaded into the lattice @xmath143 .
b ) is taken with clock interrogation @xmath144 pulses and a lattice population of @xmath145.,scaledwidth=90.0% ] considering that the single photon @xmath6sr clock transition @xmath2s@xmath3-@xmath4p@xmath5 is forbidden at any order , the magnetic - field - induced spectroscopy method is used to controllably allow the clock transition , by means of an external magnetic field coupling the @xmath0p@xmath5 to the @xmath0p@xmath10 state @xcite .
the search - scan is performed by using a mixing magnetic field of @xmath146mt and a clock probe beam intensity of @xmath147 , leading to a rabi frequency of @xmath148hz , given by @xmath149 where @xmath150 @xcite . in order to have a high - contrast spectrum
the excitation pulse length is @xmath151ms , thus overdriving the clock transition having an estimated @xmath144 pulse duration of @xmath152ms . in the search - mode
the clock laser frequency is changed by @xmath95khz , covering a span of @xmath153khz in about 50@xmath7s ( experiment cycle 1@xmath7s ) .
a typical search - mode scan is shown in fig .
[ fig : clock_spectra ] .
the excitation probability is given by the ratio @xmath154 $ ] , where @xmath155 and @xmath156 is the atomic population of the @xmath157 and @xmath158 state , respectively .
the @xmath158 population is obtained through absorption imaging of the atomic sample after the clock transition interrogation .
atoms in the @xmath158 state are blown away from the lattice by the resonant 461@xmath7 nm imaging beam .
the atoms excited into the @xmath157 level by the clock probe beam are pumped back into the @xmath158 state by means of a 50@xmath7ms pulse at @xmath159 nm and @xmath160 nm .
the @xmath158 population is then measured through an additional absorption imaging sequence . from the spectra in fig .
[ fig : clock_spectra]a , the motional sidebands are measured at @xmath161 khz from the carrier , thus corresponding to a lamb - dicke parameter @xmath162 , where @xmath163 is the atomic recoil frequency shift associated to the absorption of a photon with @xmath164 nm .
the excitation probability in the search - mode scan is only @xmath165 since the actual clock transition is under - sampled because of the need of covering a large scanning span in reduced time .
the excitation pulse duration in this high - resolution mode is chosen to realize an effective @xmath144 pulse for each configuration of magnetic field and probe intensity .
for this purpose the actual atomic rabi frequency is measured through the observation of the rabi oscillations ( see fig . [
fig : rabi_oscillations ] ) . on the carrier
the excitation probability approaches the @xmath166 % level .
we find that the fitted rabi frequency is about @xmath167 % of the one calculated from eq .
[ eq : rabi ] , an effect we attribute , together with the reduced excitation probability of @xmath166 % , to the inhomogeneous distribution of the rabi frequency among the atoms that could be given by a residual spatial inhomogeneity of the clock probe , by a residual misalignment between the probe and lattice axes ( estimated to 1 mrad ) and by the thermal transverse atomic motion in the lattice potential @xcite .
this effect can be significantly reduced by employing a probe beam with larger diameter and a colder atomic sample .
measurement of the rabi oscillations in condition of resonance with the clock transition for different values of the mixing magnetic field and clock probe beam intensity .
the data are fitted with the function @xmath168 , where @xmath169 is the actual atomic rabi frequency and @xmath170 gives the decoherence time - scale .
the measured atomic rabi frequency @xmath169 is @xmath16 of the estimated rabi frequency @xmath171 calculated from the mixing magnetic field and clock laser intensity ( see eq .
[ eq : rabi ] ) .
the corresponding @xmath144 pulse interrogation length @xmath172 was employed for the clock spectra b ) of fig .
[ fig : clock_spectra].,title="fig:",scaledwidth=40.0% ] measurement of the rabi oscillations in condition of resonance with the clock transition for different values of the mixing magnetic field and clock probe beam intensity .
the data are fitted with the function @xmath168 , where @xmath169 is the actual atomic rabi frequency and @xmath170 gives the decoherence time - scale .
the measured atomic rabi frequency @xmath169 is @xmath16 of the estimated rabi frequency @xmath171 calculated from the mixing magnetic field and clock laser intensity ( see eq .
[ eq : rabi ] ) .
the corresponding @xmath144 pulse interrogation length @xmath172 was employed for the clock spectra b ) of fig .
[ fig : clock_spectra].,title="fig:",scaledwidth=40.0% ] by reducing both the mixing magnetic field and the clock probe intensity the low drift rate of the clock laser ( few @xmath173 ) allowed us to reliably measure rabi oscillations lasting for more than @xmath174ms and fourier limited spectra below the @xmath89hz fwhm linewidth ( see fig .
[ fig : rabi_oscillations ] ) .
systematic shifts of the @xmath2s@xmath3-@xmath4p@xmath5 clock transition in @xmath6sr optical lattice clock .
a ) measurement of the second order zeeman shift as a function of the hall probe voltage .
b ) measurement of the probe ac linear stark shift as a function of the photodiode voltage .
the respective curves are used to calibrate the hall probe and the photodiode to estimate the applied magnetic field @xmath175 and the probe light intensity @xmath137 through the knowledge of the shift coefficients @xcite .
the two calibration coefficients are @xmath176 mv / mt for the hall probe and @xmath177 104(1 ) mv/(w @xmath178 ) for the photodiode , respectively.,title="fig:",scaledwidth=49.0% ] systematic shifts of the @xmath2s@xmath3-@xmath4p@xmath5 clock transition in @xmath6sr optical lattice clock .
a ) measurement of the second order zeeman shift as a function of the hall probe voltage .
b ) measurement of the probe ac linear stark shift as a function of the photodiode voltage .
the respective curves are used to calibrate the hall probe and the photodiode to estimate the applied magnetic field @xmath175 and the probe light intensity @xmath137 through the knowledge of the shift coefficients @xcite .
the two calibration coefficients are @xmath176 mv / mt for the hall probe and @xmath177 104(1 ) mv/(w @xmath178 ) for the photodiode , respectively.,title="fig:",scaledwidth=40.0% ] systematic shifts of the @xmath2s@xmath3-@xmath4p@xmath5 clock transition in @xmath6sr optical lattice clock .
a ) measurement of the second order zeeman shift as a function of the hall probe voltage .
b ) measurement of the probe ac linear stark shift as a function of the photodiode voltage .
the respective curves are used to calibrate the hall probe and the photodiode to estimate the applied magnetic field @xmath175 and the probe light intensity @xmath137 through the knowledge of the shift coefficients @xcite .
the two calibration coefficients are @xmath176 mv / mt for the hall probe and @xmath177 104(1 ) mv/(w @xmath178 ) for the photodiode , respectively.,title="fig:",scaledwidth=49.0% ] systematic shifts of the @xmath2s@xmath3-@xmath4p@xmath5 clock transition in @xmath6sr optical lattice clock .
a ) measurement of the second order zeeman shift as a function of the hall probe voltage .
b ) measurement of the probe ac linear stark shift as a function of the photodiode voltage .
the respective curves are used to calibrate the hall probe and the photodiode to estimate the applied magnetic field @xmath175 and the probe light intensity @xmath137 through the knowledge of the shift coefficients @xcite .
the two calibration coefficients are @xmath176 mv / mt for the hall probe and @xmath177 104(1 ) mv/(w @xmath178 ) for the photodiode , respectively.,title="fig:",scaledwidth=45.0% ] a study of the systematics has been carried out and values of estimated shifts on the @xmath2s@xmath3-@xmath4p@xmath5 clock transition are reported in table [ tab : final_budget ] .
the 2@xmath179 order zeeeman and probe stark shifts have been evaluated by scanning around the clock transition and interleaving measurements with different values of the magnetic field and probe power . as frequency reference we rely on the clock cavity resonance whose typical 1
hz / s linear drift was compensated to better than 10 mhz / s with a programmable feed - forward aom driver .
magnetic field and clock probe power are monitored at @xmath180 precision level with a hall probe and a photodiode , respectively .
they are calibrated through the known coefficient @xmath181 and @xmath182 @xcite by measuring the clock frequency shift as a function of the magnetic field ( @xmath183mt range , also inverting the direction of the field ) and probe intensity ( @xmath184 range ) , as shown in fig .
[ fig : shifts ] .
values are calculated for the @xmath185hz linewidth clock transition ( see table [ tab : final_budget ] ) , obtained with a magnetic field of @xmath186mt and probe beam intensity of @xmath187 , corresponding to a second order zeeman shift and probe ac stark shift of @xmath188hz and @xmath189hz respectively , where @xmath190 and @xmath191 @xcite .
the uncertainties on the second order zeeman shift and probe light shift are given by the quadratic sum of the standard error from the fit and the error associated by the voltage reading on the hall probe and the photodetector .
.[tab : final_budget ] systematic frequency shift and uncertainty budget for the @xmath6sr optical lattice clock as extracted from spectra of the clock transition .
the values are reported for the operating conditions of the @xmath185hz clock transition linewidth ( fig .
[ fig : clock_spectra ] b ) ) . [ cols="<,^,^",options="header " , ] the effects of collisions on the @xmath6sr clock transition have been studied in detail in @xcite . in our optimal conditions
the number of atoms trapped into the lattice is kept at @xmath192 ( @xmath193 atoms per site ) , so that the corresponding peak atomic density per lattice site of @xmath194 leads to a collisional shift and broadening of @xmath195hz and @xmath196hz respectively , with the coefficients @xmath197 and @xmath198 @xcite .
the uncertainties on collisional shift and broadening take into account our experimental @xmath199 fluctuation of the number of atoms shot - to - shot and the uncertainties of the @xmath200 and @xmath201 coefficients and density determination . during the clock spectroscopy measurement
the vacuum chamber temperature is monitored with one thermistor directly attached to the main chamber .
no water cooling is employed to cool the mot coils and after a warm up of about three hours , the temperature of the whole cell stabilizes to @xmath202k . while a more detailed study of temperature gradients is mandatory for accuracy level below the 10@xmath203 level , this lies beyond the scope of the present article .
we estimated the blackbody radiation shift to @xmath204hz with an uncertainty of 0.5 hz , including the uncertainty due to temperature gradients up to @xmath205 10 k inside the main cell .
the ac stark shift induced by the lattice light has been evaluated considering the effect of the detuning of 0.40 pm with respect to the magic wavelength for @xmath6sr ( with a scalar coefficient @xmath206 hz / nm/@xmath207 ) .
taking also into account the uncertainty in the knowledge of the magic wavelength and the smaller contribution from hyperpolarizability effect @xcite , the shift amount to 0.3(4 ) hz .
the amplified spontaneous emission ( ase ) from the tapered amplifier used for lattice laser is about 40 nm wide and symmetric around the emission wavelength . with a typical ase intensity 40 db below the carrier , we estimated an additional shift of -1.0(5 ) hz due to this effect .
the total value for the ac stark shift has been evaluated to be -0.7(9)hz . at @xmath28hz clock transition
linewidth the achieved total fractional uncertainty is @xmath208 , with a transition quality factor of @xmath209 . with a current clock cycle time of @xmath210 s ( mainly limited by technical delays in the absorption imaging detection ) and an interrogation time @xmath211 ms ,
the stability of the clock is limited at @xmath212 at 1 s by the dick effect @xcite . with minor changes to the detection system a reduction of the clock cycle time to @xmath213 ms
can be implemented , maintaining the same number of atoms into the lattice , thus reducing the contribution of the dick noise below @xmath214 level .
we presented a compact and transportable optical clock based on strontium atoms .
novel design solutions allowed us also to reduce the volume , weight and power consumption with respect to traditional laser cooling apparatus . as a result the experimental physics package is contained in a volume of less than 2 m@xmath0 and no water cooling is needed to operate the clock .
furthermore , a modular architecture ensured a high degree of operation reliability of the apparatus both in stationary condition and after a transportation of the experimental set up . to ensure high clock frequency stability , cooling and trapping stages
have been optimized to allow high efficiency transfer among different cooling and trapping stages , thus allowing for faster clock cycle time with high duty cycle .
spectroscopy on @xmath2s@xmath5 -@xmath0p@xmath5 clock transition on bosonic @xmath6sr isotopes has been demonstrated with an 8 hz resolution .
eventually , an evaluation of the main systematic frequency shifts on the clock transition has been done and the fractional uncertainty of the clock is @xmath1 .
the authors acknowledge financial support from esa and the european union seventh framework programme ( fp7/2007 - 2013 grant agreement 263500 , project `` space optical clocks '' ) and the european metrology research programme ( emrp ) under ind14 .
the emrp is jointly funded by the emrp participating countries within euramet and the european union .
we also acknowledge support by the dfg rtg 1729 `` fundamentals and applications of ultra - cold matter '' .
we thank d. v. sutyrin for useful discussions .
b. j. bloom , t. l. nicholson , j. r. williams , s. l. campbell , m. bishof , x. zhang , w. zhang , s. l. bromley , and j. ye . an optical lattice clock with accuracy and stability at the 10@xmath215 level .
, 506:7175 , 2014 .
s. schiller , g. m. tino , p. gill , c. salomon , u. sterr , e. peik , a. nevsky , a. goerlitz , d. svehla , g. ferrari , n. poli , l. lusanna , h. klein , h. margolis , p. lemonde , p. laurent , g. santarelli , a. clairon , w. ertmer , e. rasel , j. mueller , l. iorio , c. laemmerzahl , h. dittus , e. gill , m. rothacher , f. flechner , u. schreiber , v. flambaum , wei - tou ni , liang liu , xuzong chen , jingbiao chen , kelin gao , l. cacciapuoti , r. holzwarth , m. p. hess , and w. schaefer .
einstein gravity explorer - a medium - class fundamental physics mission .
, 23(2):573610 , 2009 . p. wolf , ch .
j. borde , a. clairon , l. duchayne , a. landragin , p. lemonde , g. santarelli , w. ertmer , e. rasel , f. s. cataliotti , m. inguscio , g. m. tino , p. gill , h. klein , s. reynaud , c. salomon , e. peik , o. bertolami , p. gill , j. paramos , c. jentsch , u. johann , a. rathke , p. bouyer , l. cacciapuoti , d. izzo , p. de natale , b. christophe , p. touboul , s. g. turyshev , j. anderson , m. e. tobar , f. schmidt - kaler , j. vigue , a. a. madej , l. marmet , m. c. angonin , p. delva , p. tourrenc , g. metris , h. mueller , r. walsworth , z. h. lu , l. j. wang , k. bongs , a. toncelli , m. tonelli , h. dittus , c. laemmerzahl , g. galzerano , p. laporta , j. laskar , a. fienga , f. roques , and k. sengstock .
quantum physics exploring gravity in the outer solar system : the sagas project .
, 23(2):651687 , 2009 .
t. m. fortier , m. s. kirchner , f. quinlan , j. taylor , j. c. bergquist , t. rosenband , n. lemke , a. ludlow , y. jiang , c. w. oates , and s. a. diddams .
generation of ultrastable microwaves via optical frequency division . 5(7):425429 , 2011 .
a. e. rogers , r. j. cappallo , h. f. hinteregger , j. i. levine , e. f. nesman , j. c. webber , a. r. whitney , t. a. clark , c. ma , j. ryan , b. e. corey , c. c. counselman , t. a. herring , i i shapiro , c. a. knight , d. b. shaffer , n. r. vandenberg , r. lacasse , r. mauzy , b. rayhrer , b. r. schupler , and j. c. pigg .
very - long - baseline radio interferometry : the mark iii system for geodesy , astrometry , and aperture synthesis . , 219(4580):514 , 1983 .
k. predehl , g. grosche , s. m. f. raupach , s. droste , o. terra , j. alnis , th .
legero , t. w. hnsch , th .
udem , r. holzwarth , and h. schnatz . a 920-kilometer optical fiber link for frequency metrology at the 19th decimal place .
, 336(6080):4414 , 2012 .
d. calonico , e. k. bertacco , c. e. calosso , c. clivati , g. a. costanzo , a. godone , m. frittelli , a. mura , n. poli , d. v. sutyrin , g. m. tino , m. zucco , and f. levi .
high accuracy coherent optical frequency transfer over a doubled 642 km fiber link , in publication on doi : 10.1007/s00340 - 014 - 5917 - 8 , 2014 .
s. bize , p. laurent , m. abgrall , h. marion , i. maksimovic , l. cacciapuoti , j. grunert , c. vian , f.p . dos santos , p. rosenbusch , p. lemonde , g. santarelli , p. wolf , a. clairon , a. luiten , m. tobar , and c. salomon .
cold atom clocks and applications . , 38(9 , si):s449s468 , 2005 .
n. poli , r. e. drullinger , m. g. tarallo , g. m. tino , and m. prevedelli .
strontium optical lattice clock with all semiconductor sources . in _ proceedings of the 2007 ieee international frequency control symposium - jointly with the 21st european frequency and time forum , vols 1 - 4 _ , pages 655658 , 2007 .
a. v. taichenachev , v. i. yudin , c. w. oates , c. w. hoyt , z. w. barber , and l. hollberg .
magnetic field - induced spectroscoy of forbidden optical transitions with application to lattice - based optical atomic clocks .
, 96:083001 , 2006 .
s. blatt , j. w. thomsen , g. k. campbell , a. d. ludlow , m. d. swallows , m. j. martin , m. m. boyd , and j. ye .
rabi spectroscopy and excitation inhomogeneity in a one - dimensional optical lattice clock .
, 80:052703 , 2009 . | we report on a transportable optical clock , based on laser - cooled strontium atoms trapped in an optical lattice .
the experimental apparatus is composed of a compact source of ultra - cold strontium atoms including a compact cooling laser set - up and a transportable ultra - stable laser for interrogating the optical clock transition .
the whole setup ( excluding electronics ) fits within a volume of less than 2 m@xmath0 .
the high degree of operation reliability of both systems allowed the spectroscopy of the clock transition to be performed with 10 hz resolution .
we estimate an uncertainty of the clock of @xmath1 . |
intermetallic compounds consisting of rare earth and transition metals , as well as metalloids , have versatile magnetic properties.@xcite compounds with itinerant @xmath5 electrons are of particular interest when they are in the vicinity the stoner transition : such systems , characterized as nearly or weakly ferromagnet , manifest strongly correlated electronic properties.@xcite on the other hand , heavy rare earth ions manifest magnetic versatility associated with the @xmath2 electrons : null magnetism ( y@xmath3 or lu@xmath3 ) , pure spin , local moment magnetism ( gd@xmath3 ) , potentially anisotropic , crystal electric field ( cef ) split , local moment magnetism ( tb@xmath3 - tm@xmath3 ) , and more exotic magnetism : yb ions may hybridize with conduction electrons and manifest so - called heavy fermion behavior .
needless to say , series of materials that combine these interesting versatilities have attracted the attention of physicists .
for example , the binary rco@xmath0 ( r = rare earth ) compounds , with the nearly ferromagnetic ( fm ) end members yco@xmath0 and luco@xmath0 , and the local moment , fm members ( r = pr , nd , gd - tm ) , have been studied for more than 35 years.@xcite discovered by nasch _ et . al @xcite , the rt@xmath0zn@xmath1 ( r = rare earth , t = transition metal in the fe , co and ni groups ) series of compounds crystallize in the cubic cecr@xmath0al@xmath1 structure @xcite .
the r and t ions occupy their own single , unique , crystallographic sites with cubic and trigonal point symmetry respectively , whereas the zn ions have three unique crystallographic sites ( fig .
[ fig1poly ] a ) .
both the r and the t ions are fully surrounded by the shells of zn that consist of the nearest neighbors ( nns ) and the next nearest neighbors ( nnns ) .
this means that there are no r - r , t - t or r - t nearest neighbor and the shortest r - r spacing is @xmath6 6 . for the r ions , a frank - kasper polyhedron with coordinate number ( cn ) 16
is formed from 4 zn nns on the @xmath7 site and 12 zn nnns on the @xmath8 site ( fig .
[ fig1poly ] b ) .
_ zn@xmath1 .
( b ) the cn16 frank - kasper polyhedron of rare earth ions.,title="fig:",scaledwidth=45.0% ] + recently , studies of the thermodynamic and transport properties of these intermetallics have revealed varied magnetic behavior.@xcite yfe@xmath0zn@xmath1 and lufe@xmath0zn@xmath1 are archetypical examples of nearly ferromagnetic fermi liquids ( nffl ) with stoner enhancement factors of @xmath9 ( where @xmath10 ) . by embedding large
, heisenberg type moments associated with gd@xmath3 ions in this highly polarizable medium , gdfe@xmath0zn@xmath1 manifests highly enhanced fm order . on the other hand ,
gdco@xmath0zn@xmath1 manifests ordinary , low temperature , antiferromagnetic ( afm ) order , correspondent to the ` normal metal ' behavior of the conduction electron host , yco@xmath0zn@xmath1.@xcite in addition to these interesting , @xmath11 electron and local moment properties , six related ybt@xmath0zn@xmath1 compounds ( t = fe , co , ru , rh , os and ir ) show heavy fermion ground states , associated with different kondo temperatures ( @xmath12 ) and yb ion degeneracies.@xcite given the similarities and differences between the rfe@xmath0zn@xmath1 and rco@xmath0zn@xmath1 ( r = gd , y , lu ) series , it becomes important to study all of the r = y , gd - lu members in detail .
a comparative study of the rfe@xmath0zn@xmath1 and rco@xmath0zn@xmath1 series will help to further understand the magnetic interaction between the local moments by means of the strongly polarizable medium , particularly with the crystal electronic field ( cef ) effect associated with non - zero orbital angular momentum .
furthermore , given the very similar cn-16 frank - kasper polyhedron for r ions , as well as the less than 2% difference of lattice constants for the whole rt@xmath0zn@xmath1 families , the study of the cef effect on these local moment members will also help in the understanding of the varied heavy fermion states of ybt@xmath0zn@xmath1 , which were thought to be due to the competition between temperature scales associated with the cef splitting and the kondo effect.@xcite in this paper , we present the results of magnetization , heat capacity and resistivity measurements on rfe@xmath0zn@xmath1 and rco@xmath0zn@xmath1 ( r = y , nd , sm , gd - lu ) compounds . compared with the weakly correlated behaviors for yco@xmath0zn@xmath1 and luco@xmath0zn@xmath1 , yfe@xmath0zn@xmath1 and lufe@xmath0zn@xmath1 , manifest clear , nffl behaviors associated with the spin fluctuation of the itinerant electrons . for the rfe@xmath0zn@xmath1 compounds ( r = gd - tm ) , the well - defined , local moment members all manifest enhanced fm ordering with @xmath4 values that roughly scale with the de gennes factor .
their anomalous , temperature dependent susceptibility and resistivity can be explained as the result of local moments embedded in a nffl host . in contrast , for the rco@xmath0zn@xmath1 series , only gd and tb members manifest afm ordering above 2 k , and the magnetic properties for r = dy - tm clearly manifest features associated with single ion cef effects on the r ions in the cubic symmetry coordination . for the r
= tb - tm members in the co series , the cef parameters can be determined from the magnetic anisotropy and the specific heat data , and are roughly consistent with the results of calculations using a point charge model .
for the fe series , the r = tb - tm members show moderate magnetic anisotropy in their ordered states , mainly due to the cef effect on the r ions , which is consistent with the magnetic anisotropy found for the co members .
these results , as well as the analysis of the heat capacity and resistivity data , indicate that the fm state emerges from the fully degenerate hund s rule ground state for rfe@xmath0zn@xmath1 ( r = gd - ho ) , whereas erfe@xmath0zn@xmath1 and tmfe@xmath0zn@xmath1 manifests cef splitting above their curie temperatures .
single crystals of rfe@xmath0zn@xmath1 and rco@xmath0zn@xmath1 ( r = y , gd - lu ) were grown from high temperature , zn rich solutions with the initial concentration of starting elements being r:(fe
/ co):zn = 2:4:94 , as described previously @xcite .
the small amount of residual zn on the resulting crystals was removed by submerging the crystals into an ultrasonic bath of 0.5 vol .
% hcl in h@xmath0o for 0.5 - 1 hour .
attempts were made to grow even lighter rare earth members ( r = eu , nd and ce ) for the fe series , but no ternary compound was found to form . on the other hand , for the co series , single crystals for the r = nd and sm members were grown using the same growth method .
room temperature , powder x - ray diffraction measurements , using cu @xmath13 radiation , were performed on the samples with additional si powder ( @xmath14 ) used as standard .
figure [ fig2lattice ] shows the lattice constants for the two series of compounds , obtained by using the rietica , rietveld refinement program , plotted with respect to the effective radius of r@xmath3 with cn = 9 @xcite , since the data is absent for larger cn .
the variation of the lattice constant manifests the well - known , lanthanide contraction for r = gd - lu with no evident deviation for r = yb .
however , the relatively larger lattice constants for yfe@xmath0zn@xmath1 and yco@xmath0zn@xmath1 indicate that , with this large cn , the effective ionic radii of y@xmath3 is between tb@xmath3 and dy@xmath3 rather than between dy@xmath3 and ho@xmath3 .
this deviation for y@xmath3 ions is not unprecedented in the isostructural compounds rru@xmath0zn@xmath1 @xcite and rmn@xmath0in@xmath15zn@xmath16 @xcite , as well as the similar structure compounds rco@xmath0 @xcite .
additional single crystal x - ray diffraction measurements were preformed on r = gd , tb , er and lu members of the @xmath17 series and demonstrated full occupancy on all crystallographic sites ( within the detection errors ) and the same lattice as the powder x - ray values.@xcite this result is complicated by the difficulty in resolving the difference between fe and zn elements with very similar x - ray scattering strengths . ) for rfe@xmath0zn@xmath1 and rco@xmath0zn@xmath1 versus the radius of the trivalent rare earth ion with cn = 9 .
@xcite the error bars were estimated from the standard variation of four seperate measurement made on one batch of sample.,title="fig:",scaledwidth=45.0% ] + it is worth noting though , that for the local moment bearing members in fe series , single crystals obtained from different ratios of starting element concentrations manifest detectably different magnetic ordering temperatures .
these differences , tentatively associated with very subtle variations of element occupancy on the crystallographic sites@xcite , are related to the extreme sensitivity to the small disorder for compounds with such strongly correlated conduction electron backgrounds .
a detail discussion of this is presented in appendix a. measurements of electrical resistivity were preformed by using a standard ac , four - probe technique .
the samples were cut into bars , with typical lengths of 23 mm , along their crystallographic [ 110 ] direction .
these bars were measured in a quantum design physical properties measurement system , ppms-14 and/or ppms-9 instruments ( @xmath18 @xmath19 k ) with @xmath20 hz , @xmath21
@xmath22 ma .
temperature dependent specific heat measurements were also performed in these ppms units , using the heat capacity option . in order to study the magnetic part of specific heat , the contribution from
the conduction electrons and lattice vibrations was subtracted .
this was performed by subtracting off the specific heat of the non - magnetic members yt@xmath0zn@xmath1 or lut@xmath0zn@xmath1 [ @xmath23 .
the magnetic part of entropy can then be estimated as @xmath24 , in which the specific heat data below 1.8 k ( 0.4 k for @xmath25 ) were approximated by a linear extrapolation to @xmath26 at 0 k. magnetization measurements were carried out in quantum design magnetic properties measurement system ( mpms ) , superconducting quantum interface device ( squid ) magnetometers in varied applied fields ( @xmath27 koe ) and temperatures ( @xmath28
@xmath29 k ) . in the measurements of magnetization for the fm - ordered , rfe@xmath0zn@xmath1 compounds ,
the effects of the demagnetizing field are relatively small , because of the dilute nature of the magnetic moments:@xcite the maximum demagnetizing field is about 3000 oe for a plate - like shaped sample .
even so , in order to obtain accurate magnetization isotherms near @xmath4 , the samples were cut and measured with the applied field along their long axis so as to minimize this already small demagnetization effect .
we start characterizing the compounds with the non - magnetic rare earth ions of the series : y(lu)fe@xmath0zn@xmath1 and y(lu)co@xmath0zn@xmath1 . without any @xmath2 electronic magnetism , these compounds manifest the electronic and magnetic properties associated with the conduction electron background of each series .
next , we will introduce the two series of compounds with well - defined @xmath2 local moments : r = gd - tm .
we will examine the magnetization and specific heat data for the co series at first .
then an overview of the magnetic properties for the fe series will be presented next .
after that , the magnetization , specific heat and resistivity data will be presented for each fe member separately . finally , similar data for the r = yb heavy fermion compounds , ybfe@xmath0zn@xmath1 and ybco@xmath0zn@xmath1 will be presented .
divided by the applied field @xmath30 for yfe@xmath0zn@xmath1 and lufe@xmath0zn@xmath1 as well as their co analogues for @xmath31 koe and 50 koe .
inset : a blow - up plot at low temperature .
( b ) @xmath32 for yfe@xmath0zn@xmath1 and lufe@xmath0zn@xmath1 .
the solid lines present the modified curie - weiss [ @xmath33 fit for the data above 100 k. inset : field dependent magnetization at 1.85 k.,title="fig:",scaledwidth=45.0% ] + temperature dependent magnetization data ( divided by the applied field ) for y(lu)fe@xmath0zn@xmath1 and y(lu)co@xmath0zn@xmath1 are shown in fig .
[ figylu1 m ] a. the fe members manifest similar , strongly enhanced , temperature - dependent paramagnetic signals , whereas the co members manifest essentially temperature - independent , pauli paramagnetic signals .
the low temperature features for the two fe compounds are shown in the inset of fig .
[ figylu1 m ] . in the applied field of 10 koe ,
the magnetization signals of both fe members show a faint maximum below 10 k , whereas the high magnetic field ( 50 koe ) suppresses the lowest temperature @xmath34 values , as well as the maximum . in our experience on the measurements of different batches of samples ,
these low temperature features are moderately sample - dependent ( different samples may show 20% different magnetization signal and 23 k difference in the temperature of the maximum , @xmath35 ) . nevertheless , the maximum of the temperature dependent susceptibility , @xmath36 , is a common feature in the nffls , such as pd @xcite , yco@xmath0 and luco@xmath0 @xcite , as well as tibe@xmath0 @xcite , although quantitative calculation of @xmath36 still presents a challenge even for the simplest case of pd @xcite .
the field suppression of the magnetization ( and @xmath35 ) at low temperature is not attributed to the possible existence of a paramagnetic impurity contribution ( which would contribute more to the value of @xmath34 at lower temperature and lower field , and therefore suppress the maximum of @xmath34 in lower field ) , but , as discussed below , to the intrinsic variation of @xmath37 with respect to @xmath30 at different temperatures .
figure [ figylu1 m ] b shows that above a characteristic temperature ( @xmath38 k ) , the susceptibility of yfe@xmath0zn@xmath1 and lufe@xmath0zn@xmath1 can be approximately fitted by a curie - weiss ( cw ) term [ @xmath39 plus a temperature - independent term ( @xmath40 ) .
the values of effective moment ( @xmath41 ) , @xmath42 and @xmath40 are extracted as 1.0 @xmath43/fe , -16 @xmath44 , @xmath45emu / mol and 1.1 @xmath43/fe , -33 @xmath44 , @xmath46emu / mol for yfe@xmath0zn@xmath1 and lufe@xmath0zn@xmath1 , respectively .
these values of @xmath41 are significantly larger than the estimated induced moment of fe site in the fm ground state of gdfe@xmath0zn@xmath1 , @xmath47/fe .
@xcite such apparent cw - like behavior was also observed in other nffl systems . @xcite in the context of the spin fluctuation model @xcite , itinerant electronic systems can manifest cw - like behavior with a curie constant related to the local amplitude of the spin fluctuation .
the magnetization data at the base temperature ( 1.85 k ) show nearly linear dependent with the applied field ( inset in fig .
[ figylu1 m ] b ) , which is distinct from the brillouin function type of magnetization curves associated with local moments . for lufe@xmath0zn@xmath1 .
from right to left : @xmath48 koe , 5 koe , 10 koe , 20 koe and 30 koe.,title="fig:",scaledwidth=45.0% ] + for lufe@xmath0zn@xmath1 at varied temperature.,title="fig:",scaledwidth=45.0% ] + in order to better understand the variation of the maximum in temperature dependent @xmath34 data for yfe@xmath0zn@xmath1 and lufe@xmath0zn@xmath1 , @xmath49 and @xmath50 measurements were performed on lufe@xmath0zn@xmath1 for varied applied field and temperature respectively .
figure [ figlu1 ] shows that the magnetic field suppresses the values of @xmath34 , as well as the maximum of @xmath34 , which disappears when @xmath51 koe .
the values of the field dependent susceptibility , @xmath52 , were extracted as @xmath53 with @xmath54 oe from the @xmath50 data at varied temperature ( figure [ figlu2 ] ) . for @xmath55
k , the values of @xmath52 monotonically decrease with increase @xmath30 , whereas a local maximum appears around @xmath56 koe in the data sets as @xmath57 k. this critical temperature ( @xmath58 k ) is close to the @xmath35 ; the maximum of @xmath52 ( @xmath59 koe ) is also close to the suppression field determined by fig .
[ figlu1 ] .
this curious , field dependent , susceptibility at varied temperature is reminiscent to the one of tibe@xmath0 , albeit the amplitude of local maximum in @xmath52 is much smaller .
@xcite in the case of tibe@xmath0 , the reason of anomalous field - dependent magnetization is also still not clear .
@xcite zn@xmath1 and lufe@xmath0zn@xmath1 ( plotted as @xmath60 versus @xmath61 ) , as well as the co analogues.,title="fig:",scaledwidth=45.0% ] + figure [ figylu2cp ] presents the low temperature specific heat data for yfe@xmath0zn@xmath1 and lufe@xmath0zn@xmath1 , as well , as for the co analogues , plotted as @xmath60 versus @xmath61 .
all four compounds manifest clear fermi liquid behavior ( @xmath62 ) .
the similar @xmath63 values ( represented as the slopes of the data sets in the plot , @xmath64 ) indicate the similar debye temperatures for these 4 compounds ( @xmath65 k @xcite ) , consistent with their similar molar mass , similar composition and similar lattice parameters . on the other hand , the over 2.5 times larger values of electronic specific heat ( @xmath66 ) of the fe members indicate a larger density of states at fermi level [ @xmath67 , compared to the co analogues ( consistent with the band structure calculation results @xcite ) .
the values of the electronic specific heat and magnetic susceptibility can be employed to estimate the stoner enhancement factor , @xmath68 , in the context of the stoner theory : that is , the static susceptibility is enhanced by @xmath69 , whereas the electronic specific heat is not .
@xcite the estimated @xmath68 values of yfe@xmath0zn@xmath1 and lufe@xmath0zn@xmath1 are 0.88 , 0.89 , respectively , comparable with the estimated values of the canonical nffl systems : pd : 0.83 , and yco@xmath0 : 0.75.@xcite zn@xmath1 and lufe@xmath0zn@xmath1 , as well as their co analogues .
( b ) : estimated spin fluctuation contribution to the resistivity for yfe@xmath0zn@xmath1 and lufe@xmath0zn@xmath1 .
the error bars were estimated as @xmath70 of the values of the resistivity for yco@xmath0zn@xmath1 and luco@xmath0zn@xmath1 respectively.,title="fig:",scaledwidth=45.0% ] + the temperature dependent electrical resistivity data for yfe@xmath0zn@xmath1 and lufe@xmath0zn@xmath1 are larger than that for the co analogues over the whole temperature range ( fig .
[ figylu3rou ] a ) .
this is not unexpected for a nffl since the spin fluctuations will affect the scattering process of the conduction electrons , which leads to an additional contribution to the resistivity . in order to study the spin fluctuation contribution to the resistivity , the total electrical resistivity @xmath71 is assumed to be : @xmath72 where the first , second and third terms represent residual , phonon and spin fluctuation reisitivity , respectively .
assuming the phonon scattering contribution , @xmath73 , is essentially same for the fe and co analogues ( as suggested by the similar @xmath63 terms ) , then , the spin fluctuation scattering contribution , @xmath74 , can be estimated as : @xmath75 shown in fig .
[ figylu3rou]b , @xmath74 for these two compounds increases with temperature and is close to a saturated value ( 10 @xmath76 ) at 300 k , within the accuracy of the measurements .
versus @xmath61 for yfe@xmath0zn@xmath1 and lufe@xmath0zn@xmath1 , as well as their co analogues .
the solid lines present the linear fit of the data sets from 2 k to 9 k.,title="fig:",scaledwidth=45.0% ] + the analysis of the low temperature resistivity data reveals a quadratic , standard fermi liquid , behavior [ @xmath77 for all 4 compounds ( fig .
[ figylu4rout2 ] ) .
the @xmath78 values of the fe compounds are about 7 times larger than the two co analogues .
this result is consistent with the 2.5 times larger @xmath66 values of the fe compounds , in the context of the fermi liquid theory , meaning @xmath78 is proportional to the square of the effective mass of the quasi - particles due to the strong correlation effect , whereas @xmath66 is proportional to the effective mass . in the point of view of spin fluctuation theory , nearly fm metals manifest fermi liquid behaviors at low temperature region with @xmath78 values enhanced by spin fluctuations.@xcite
zn@xmath1 ( r = nd and sm ) compounds , divided by applied field
@xmath79 oe . inset : applied field ( @xmath79 oe ) divided by the magnetizations of rco@xmath0zn@xmath1 ( r = nd and sm ) as a function of temperature.,title="fig:",scaledwidth=45.0% ] + before discussing the heavy rare earth compounds ( r = gd - yb ) , the results of thermodynamic measurement on ndco@xmath0zn@xmath1 and smco@xmath0zn@xmath1 will be briefly examined .
figure [ figndsm1 ] shows the temperature dependent magnetization data ( divided by the applied field @xmath80
oe ) for ndco@xmath0zn@xmath1 and smco@xmath0zn@xmath1 .
neither of them manifest any sign of magnetic ordering above 2 k. the temperature dependent @xmath32 for ndco@xmath0zn@xmath1 shows a cw behavior [ @xmath33 with @xmath81 , @xmath82 k and @xmath83 emu / mol .
the value of the effective moment is close to the theoretical values for the hund s rule ground state of the @xmath2 electrons of nd@xmath3 ion ( @xmath84 ) . on the other hand , though the magnetization of smco@xmath0zn@xmath1 drops with increase temperature , it does so in a distinctly non - cw manner .
this behavior is not unexpected in sm containing compounds @xcite , and is most likely due to the thermal population of the first excited hund s rule multiplet .
zn@xmath1 ( r = nd , sm and y).,title="fig:",scaledwidth=45.0% ] + specific heat data for ndco@xmath0zn@xmath1 and smco@xmath0zn@xmath1 are shown in fig .
[ figndsm2 ] along with data for yco@xmath0zn@xmath1 for comparison .
the low temperature upturn in the ndco@xmath0zn@xmath1 data below 2 k may be due to a lower temperature magnetic ordering or a schottky anomaly due to the cef splitting .
the specific heat data for smco@xmath0zn@xmath1 manifest a broad peak around 4 k , which is most likely due to the cef splitting of the hund s rule ground state of sm@xmath3 . both ndco@xmath0zn@xmath1 and smco@xmath0zn@xmath1 data increase much faster above 10 k , and remain more than 10 j / mol k larger above 25 k ( not shown here ) , compared with the data for the non - magnetic analogue yco@xmath0zn@xmath1 . on the other hand ,
the calculated results of the cef splitting for the hund s rule ground state of nd@xmath3 ion in a point charge model show the splitting energy levels within 25 k ( see table [ table3 ] below ) .
this large difference indicates that , at this point , the magnetic part of @xmath85 for ndco@xmath0zn@xmath1 and smco@xmath0zn@xmath1 can not be well estimated , since the @xmath85 data of yco@xmath0zn@xmath1 is not an adequate approximation of the non - magnetic background and unfortunately the laco@xmath0zn@xmath1 analogue did not form .
zn@xmath1 ( r = gd - yb ) compounds , divided by applied field @xmath86 oe.,title="fig:",scaledwidth=45.0% ] + temperature dependent magnetization data ( divided by the applied field @xmath86
oe ) for rco@xmath0zn@xmath1 ( r = gd - yb ) are presented in fig .
[ figallco1mh ] .
in addition to the previously reported , afm ordered gdco@xmath0zn@xmath1 with the nel temperature @xmath87 k@xcite , tbco@xmath0zn@xmath1 also shows afm ordering with @xmath88 k , which also clearly manifests itself in the specific heat data ( shown below in fig .
[ figallco3cp ] ) .
the rest of the members ( r = dy - yb ) do not show magnetic ordering above 2 k. due to the relatively low density of state at fermi level [ @xmath89 for the y and lu analogues@xcite and large r - r separation , such low temperature magnetic ordering for the @xmath2 local moments coupled via the ruderman - kitter - kasaya - yosida ( rkky ) interaction is not unexpected .
oe ) divided by the magnetizations of rco@xmath0zn@xmath1 ( r = gd - yb ) as a function of temperature.,title="fig:",scaledwidth=45.0% ] + figure [ figallco2hm ] shows the temperature dependent @xmath32 for r = gd - tm and yb members of the rco@xmath0zn@xmath1 series .
all the members , including ybco@xmath0zn@xmath1 , manifest cw behavior [ @xmath33 with negligible small @xmath90 ( @xmath91emu / mol ) and the values of @xmath41 close to the theoretical values for the hund s ground state of the @xmath2 electronic configurations ; all the values of @xmath42 are close to 0 , consistent with the low magnetic ordering temperatures ( table [ table1 ] ) .
.[table1 ] paramagnetic curie temperature , @xmath42 ( with @xmath92 k errors ) and effective moment , @xmath41 [ from the cw fit of @xmath36 from 50 k to 300 k ] ; nel temperature , @xmath93 for rco@xmath0zn@xmath1 compounds ( r = nd , gd - yb ) . [ cols="<,^,^,^,^,^,^,^,^",options="header " , ] for tbfe@xmath0zn@xmath1 ( @xmath86 oe ) ; ( b ) @xmath85 ; ( c ) @xmath94 and @xmath95 .
upper inset : magnetic part of specific heat .
lower inset : magnetic entropy @xmath96.,title="fig:",scaledwidth=45.0% ] + [ figtb1 ] temperature dependent @xmath34 , specific heat and resistivity data sets for tbfe@xmath0zn@xmath1 are shown in fig .
[ figtb1 ] .
the @xmath97 data are consistent with fm order below 60 k , and the magnetic phase transition manifests itself as a faint feature in @xmath85 data , indicating @xmath98 k. this is clearer in the magnetic part of specific heat ( @xmath99 ) data ( fig . [ figtb1 ] upper inset ) where @xmath4 is taken as the position of the greateat upward slope . as shown in the lower inset to fig .
[ figtb1 ] b , at @xmath4 , the magnetic entropy is close to the value for the full degeneracy of the hund s ground state of tb@xmath3 , @xmath100 .
as we shall see for the rest of the local moment members ( r = dy - tm ) , the released magnetic entropy at @xmath4 for the respective rare earth ion , is close to the full degeneracy value for their hund s ground state , except for r = tm .
the @xmath71 data manifests a change in the slope , which could be seen even more clearly on in the @xmath95 data , consistent with a @xmath101 k. zn@xmath1.,title="fig:",scaledwidth=45.0% ] + zn@xmath1 along three principle axes at 2 k. the three lines represent the calculated results based on molecular field approximation are all clustered near 9 @xmath43 and appear as a single line .
the dashed lines and the values present the extrapolate of the magnetization curves and the estimated spontaneous magnetic moments along three directions.,title="fig:",scaledwidth=45.0% ] + figure [ figtb2 ] presents a plot of @xmath102 versus @xmath32 ( an arrott plot ) isotherms near @xmath4 .
the isotherm that most closely goes though the origin is the one closest to @xmath4 , giving for this case a value of @xmath103 k , consistent with the results of the @xmath85 and @xmath71 measurements .
figure [ figtb3 ] shows magnetization versus external field data along 3 different crystallographic directions : [ 100 ] , [ 110 ] and [ 111 ] , at 2 k. all of these data sets are consistent with a low temperature fm ground state with moderate anisotropy .
the spontaneous longitudinal magnetic moment in zero applied external field , estimated by the extrapolation of the magnetization curves back to @xmath104 , yield @xmath105)=8.0 \mu _ b$ ] , @xmath106)=6.6 \mu _ b$ ] , and @xmath107)=5.7 \mu _ b$ ] .
the ratio of them is very close to @xmath108 .
such behavior indicates that the spontaneous magnetic moments along [ 111 ] and [ 100 ] directions can be understood as the projection of the one along the easy axis , [ 110 ] . at 2 k ,
the saturated moment at 55 koe along the easy axis , [ 110 ] , is @xmath109 , @xmath110 less than the value associated with the hund s ground state . for dyfe@xmath0zn@xmath1 ( @xmath86 oe ) ; ( b ) @xmath85 ; ( c ) @xmath94 and @xmath95 .
upper inset : magnetic part of @xmath85 .
lower inset : magnetic entropy.,title="fig:",scaledwidth=45.0% ] + zn@xmath1.,title="fig:",scaledwidth=45.0% ] + the low field thermodynamic and transport properties of dyfe@xmath0zn@xmath1 are shown in fig .
[ figdy1 ] .
the temperature dependent magnetization data ( fig .
[ figdy1 ] a ) suggest a fm transition below 50 k. the specific heat data show a kink associated with magnetic ordering ( fig .
[ figdy1 ] b ) , which can be seen more clearly after the subtraction of the non - magnetic background ( upper inset ) and indicates @xmath111 k. this fm transition temperature is further confirmed by a weak change in slope in @xmath71 ( associated with the low temperature loss of spin disorder scattering ) , indicating @xmath112 k. given that the loss of spin disorder scattering in intermetallics often scales with de gennes parameter @xcite , the feature we find in @xmath71 below @xmath4 becomes fainter and fainter as r progresses from gd to tm .
these values of @xmath4 are consistent with the result of the arrott plot analysis , from which a value of @xmath111 k can be inferred(fig .
[ figdy2 ] ) .
it is worth noticing that the specific heat data show a faint shoulder near 10 k , which appears to be a broad peak after the background subtraction , and is coincident with a slope change feature in @xmath71 data .
as seen below , such anomaly below @xmath4 in @xmath85 and @xmath71 data also appears for the members of r = ho , er and tm . those anomalies are likely due to the magnetic excitation energy spectrum associated with the hund s rule multiplet of r@xmath3 ions in their fm states .
( further discussion will be presented below . )
zn@xmath1 at 2 k along three principle axes .
the solid lines represent the calculated result based on molecular field approximation ( see data analysis part below ) .
the dashed lines and the values present the extrapolate of the magnetization curves and the estimated spontaneous magnetic moments along 3 directions.,title="fig:",scaledwidth=45.0% ] + the 2 k field dependent , magnetization isotherms for dyfe@xmath0zn@xmath1 are shown in fig .
[ figdy3 ] . compared to tbfe@xmath0zn@xmath1 , the magnetization curves for dyfe@xmath0zn@xmath1 reveal a slightly more complicated , anisotropic behavior .
the magnetization along [ 100 ] direction manifests one metamagnetic phase transition near 12 koe . above this transition ,
the magnetization along [ 100 ] direction is essentially the same as that for the field along the easy [ 111 ] axis .
the spontaneous longitudinal magnetization along the three directions , @xmath106)=9.1 \mu _ b$ ] , @xmath105)=7.4 \mu _ b$ ] , and @xmath107)=5.3 \mu _ b$ ] , have a ratio very close to @xmath113 .
these results indicate that @xmath105)$ ] and @xmath107)$ ] can be seen as the projection of @xmath106)$ ] .
the metamagnetic phase transition along [ 100 ] can be understood as the process of a classical spin reorientation in a cubic symmetry coordination . as in the case for gdfe@xmath0zn@xmath1 and tbfe@xmath0zn@xmath1 ,
the saturated moment of dyfe@xmath0zn@xmath1 at 55 koe , @xmath114 , is slightly less than the value of the hund s ground state value , @xmath115 . for hofe@xmath0zn@xmath1 ( @xmath86 oe ) ; ( b ) @xmath85 ; ( c ) @xmath94 and @xmath95 .
upper inset : magnetic part of @xmath85 data .
lower inset : magnetic entropy.,title="fig:",scaledwidth=45.0% ] + zn@xmath1.,title="fig:",scaledwidth=45.0% ] + figure [ figho1 ] presents the low field thermodynamic and transport data from measurements on hofe@xmath0zn@xmath1 .
the anomalies associated with the fm transition in hofe@xmath0zn@xmath1 in the specific heat and resistivity data are relatively weak .
the specific heat anomaly can be associated with @xmath116 k , and the @xmath95 data show faint anomaly at this temperature ( fig .
[ figho1 ] ) .
the @xmath4 value is determined as @xmath117 k from @xmath85 data , as well as @xmath118 k from @xmath71 data .
this determinate @xmath4 value is consistent with the result of the arrott plot analysis ( fig .
[ figho2 ] ) , which gives @xmath119 k. zn@xmath1 at 2 k along three principle axes .
the solid lines represent the calculated result based on molecular field approximation ( see data analysis part below ) .
the dashed lines and the values present the extrapolate of the magnetization curves and the estimated spontaneous magnetic moments along 3 directions.,title="fig:",scaledwidth=45.0% ] + the low temperature magnetic isotherms for hofe@xmath0zn@xmath1 ( fig .
[ figho3 ] ) manifest similar , but obviously larger , anisotropy to the ones for tbfe@xmath0zn@xmath1 .
the ratio of the spontaneous magnetization , @xmath105):m([111]):m([100 ] ) = 9.1 \mu _
b:7.0 \mu _
b:6.1 \mu _
b$ ] is close to the ratio of @xmath108 .
this ratio is consistent with the projection of the local moment from the easy [ 110 ] axis onto the [ 111 ] and [ 100 ] axes . in the external field of 55 koe
, the magnetization along the easy axis , [ 110 ] , reaches the value of @xmath120 , very close to the value of the hund s ground state , @xmath115 . for erfe@xmath0zn@xmath1 ( @xmath86 oe ) ; ( b ) @xmath85 ; ( c ) @xmath94 and @xmath95 .
upper inset : magnetic part of @xmath85 data .
lower inset : magnetic entropy.,title="fig:",scaledwidth=45.0% ] + zn@xmath1.,title="fig:",scaledwidth=45.0% ] + the low field thermodynamic and transport properties of erfe@xmath0zn@xmath1 are shown in fig .
[ figer1 ] .
the specific heat data show a kink near 18 k ( fig . [ figer1 ] b ) , which can be seen more clearly after the background subtraction ( upper inset ) and indicates @xmath121 k. the resistivity data show no clear anomaly at this temperature.(fig .
[ figer1 ] ) the released magnetic entropy reaches 21 j / mol k at @xmath4 , 90% of the one associated with the hund s ground state of er@xmath3 , @xmath122 ( fig . [ figer1 ] lower inset ) .
although @xmath71 data manifest no anomaly at @xmath4 , we will see below that the weak anomaly associated with magnetic ordering can be blow up after the background [ @xmath71 for lufe@xmath0zn@xmath1 ] subtraction . the arrott plot for erfe@xmath0zn@xmath1 ( fig .
[ figer2 ] ) , although showing non - linear , isothermal curves , demonstrates @xmath124 k with little ambiguity .
the non - linear feature is not unexpected for the @xmath2 local moment systems associated with the cef induced anisotropy .
@xcite zn@xmath1 at 2 k along three principle axes .
the solid lines represent the calculated result based on molecular field approximation ( see data analysis part below ) .
the dashed lines and the values present the extrapolate of the magnetization curves and the estimated spontaneous magnetic moments along 3 directions.,title="fig:",scaledwidth=45.0% ] + the magnetic anisotropy of erfe@xmath0zn@xmath1 is reminiscent of that of dyfe@xmath0zn@xmath1 : both have the same easy and hard magnetization orientations , [ 111 ] and [ 110 ] respectively , as well as the metamagnetic transition along the [ 100 ] direction ( fig . [ figer3 ] ) .
the ratio of the spontaneous longitudinal magnetic moments , @xmath106):m([110]):m([100 ] ) = 7.4 \mu _
b:5.9 \mu _
b:4.2 \mu _
b$ ] is also close to the ratio of @xmath125 . for tmfe@xmath0zn@xmath1 ( @xmath86 oe ) ; ( b ) @xmath85 ; ( c ) @xmath94 and @xmath95 .
upper inset : magnetic part of @xmath85 data .
lower inset : magnetic entropy.,title="fig:",scaledwidth=45.0% ] + the low field magnetization , specific heat and resistivity data for tmfe@xmath0zn@xmath1 are shown in fig . [ figtm1 ] .
the temperature dependent magnetization data suggest a fm transition below 10 k ( fig . [ figtm1 ] a ) . however , the specific heat data for tmfe@xmath0zn@xmath1 only manifest one broad peak at 4.5 k ( fig .
[ figtm1 ] b ) , which is less like the anomalies associated with @xmath4 for r = gd - er , and more like a schottky anomaly associated with a cef splitting .
the resistivity data also show anomaly below 5 k ( fig .
[ figtm1 ] c ) .
however , at this point , it is difficult to determine whether this anomaly is associated with the magnetic ordering or the cef splitting of the @xmath2 electrons of tm@xmath3 ions .
as we can see below , after the subtraction of the nonmagnetic background , the anomaly associated with the loss of the spin disorder scattering can be seen more clearly .
zn@xmath1.,title="fig:",scaledwidth=45.0% ] + for tmfe@xmath0zn@xmath1 , the arrott plot analysis provides the reliable criterion for @xmath4 determination .
figure [ figtm2 ] shows that @xmath4 can be determined as @xmath126 k without any ambiguity . at this temperature ,
the magnetic entropy is 15 j / mol k , only 70% of the value of fully released entropy of hund s ground state of tm@xmath3 , @xmath127 ( fig .
[ figtm1 ] upper inset ) .
zn@xmath1 at 2 k along three principle axes .
the solid lines represent the calculated result based on molecular field approximation ( see data analysis part below ) . the dashed lines and the values
present the extrapolate of the magnetization curves and the estimated spontaneous magnetic moments along 3 directions.,title="fig:",scaledwidth=45.0% ] + the low temperature magnetic isotherms for tmfe@xmath0zn@xmath1 manifest the same easy and hard axis as tb and ho members , [ 110 ] and [ 111 ] , respectively ( fig .
[ figtm3 ] ) .
the spontaneous longitudinal magnetic moments along the three principle axes are all close to @xmath128
. such a result may be due to the relatively low value of @xmath4 , which makes the spontaneous magnetic moment less anisotropic at 2 k. the saturated moment along the easy axis reaches @xmath129 at 55 koe , @xmath130 less than the value of the hund s ground state , @xmath131 .
( a ) and resistivity ( b ) for ybfe@xmath0zn@xmath1 and ybco@xmath0zn@xmath1 ( @xmath86 oe).,title="fig:",scaledwidth=45.0% ] + the physical properties of ybfe@xmath0zn@xmath1 and ybco@xmath0zn@xmath1 have been discussed in ref .
they manifest typical heavy fermion behavior with large electronic specific heat , 520 mj / mol k@xmath132 and 7900 mj / mol k@xmath132 , respectively .
figure [ figyb1 ] shows temperature dependent susceptibility and resistivity data for these two yb compounds .
the susceptibility data for ybfe@xmath0zn@xmath1 manifest a broad , kondo - type peak about 20 k , indicating a clear loss of local moment behavior , whereas the susceptibility for ybco@xmath0zn@xmath1 shows cw behavior down to 1.8 k ( fig . [ figallco2hm ] ) , associated with the effective moment value @xmath133 . above @xmath134 k , @xmath36 for ybfe@xmath0zn@xmath1 manifests a cw behavior with an effective moment of @xmath135 , close to the value of the hund s ground state of yb@xmath3 , @xmath136 ( see fig . [ figallfe2hm ] ) .
the resistivity data for ybfe@xmath0zn@xmath1 show a broad shoulder about 30 k , whereas for ybco@xmath0zn@xmath1 , the resistivity data shows a kondo resistance minimum about 50 k and a clear coherent peak about 2 k. these apparently different behaviors for these two yb - based heavy fermion compounds with virtually identical structures has been explained as the result of significantly different kondo temperatures : @xmath137 k and 1.5 k for fe and co compounds , respectively .
and @xmath138 ( a ) , the maximum value on @xmath139 ( b ) with respect to the de gennes factor for rfe@xmath0zn@xmath1 ( r = gd - tm).,title="fig:",scaledwidth=45.0% ] + as shown in fig .
[ figdis1 ] a , the @xmath4 values of rfe@xmath0zn@xmath1 compounds ( r = gd - tm ) scale fairly well with the de gennes factor , @xmath140 , consistent with a rkky interaction .
all of the @xmath42 values for each compounds are smaller than their respective @xmath4 values ( for r = er and tm , the values of @xmath42 are even negative ) .
these small @xmath42 values are consistent with the deviation of @xmath36 from the cw law ( fig .
[ figallfe2hm ] ) .
as observed in the case of pseudo - ternary compounds gd@xmath15y@xmath141fe@xmath0zn@xmath1 , such deviation can be explained as a result of increasing coupling between the local moments embedded in the strongly temperature dependent , polarizable matrix , yfe@xmath0zn@xmath1 or lufe@xmath0zn@xmath1 .
@xcite previous studies show that the magnetization of gdfe@xmath0zn@xmath1 at base temperature is nearly isotropic with a slightly reduced saturated moment ( @xmath142 less than the value of the hund s rule ground state of gd@xmath3 ) . for r
= tb - tm , the magnetization anisotropy at base temperature is significant , and correlates with the easy and hard axes of the respective rco@xmath0zn@xmath1 analogues .
such behavior indicates the anisotropy of the rfe@xmath0zn@xmath1 ( r = tb - tm ) compounds may mainly be due to the cef effect on the r@xmath3 ions .
the @xmath50 curves at 2 k manifest divided behavior with r = tb , ho and tm on one hand and r = dy and er on the other hand : for r = tb , ho and tm , the magnetization processes are gradual along all 3 principal axes ; for r = dy and er , the magnetization data along [ 100 ] direction shows clear and sharp metamagnetic transition .
both types of magnetization processes ( gradual increase and metamagnetic transition ) are common for the fm ordered @xmath2 local moments with cef anisotropy associated with the r in a cubic point symmetry , and can be understood in terms of the purification of the cef split @xmath2 electronic wave function due to the zeeman effect of the external field , and the rotation of the local moment .
@xcite given that tb@xmath3 and tm@xmath3 , as well as dy@xmath3 and er@xmath3 ions have same total @xmath2 electronic hund s rule ground state quantum number ( @xmath143 and @xmath144 respectively ) , the similar magnetic anisotropy indicates similar cef effect for the two sets of rare earth ions , respectively . in order to better understand the magnetic anisotropy of rfe@xmath0zn@xmath1 compounds ( r = tb - tm ) , the cef effect acting on the r ions must to be considered .
however , multiple difficulties associated with the strongly polarizable back ground [ y(lu)fe@xmath0zn@xmath1 ] as well as the strong magnetic interaction , make the determination of the cef parameters hard .
for example , in order to reduce the magnetic interaction , the magnetic r@xmath3 ions were placed into a dilute coordination , r@xmath15y@xmath141fe@xmath0zn@xmath1 or r@xmath15lu@xmath141fe@xmath0zn@xmath1 .
a fm ground state has been found even for very dilute magnetic r concentration : it was found that tb@xmath145y@xmath146fe@xmath0zn@xmath1 , dy@xmath145y@xmath146fe@xmath0zn@xmath1 as well as ho@xmath147y@xmath148fe@xmath0zn@xmath1 manifest fm ordering above 2 k. for such small @xmath149 , the background subtraction ( magnetization and/or specific heat of yfe@xmath0zn@xmath1 or lufe@xmath0zn@xmath1 ) , as well as the uncertainty of @xmath149 , make the fitting process unreliable .
y@xmath147fe@xmath0zn@xmath1 ( a ) and tm@xmath147y@xmath147fe@xmath0zn@xmath1 ( b ) along three principle axes at 1.85 k. the solid lines present the calculated magnetization by using the single - ion hamiltonian and the cef coefficients determined from the respective co members.,title="fig:",scaledwidth=45.0% ] + on the other hand , due to the very similar r coordination and the lattice parameters for fe and co series , the cef parameters determined from rco@xmath0zn@xmath1 compounds should be close to those for the rfe@xmath0zn@xmath1 compounds , for respective r members .
figure [ figertmfe ] shows that the anisotropic magnetization isotherms for the pseudo - ternary fe compounds , er@xmath147y@xmath148fe@xmath0zn@xmath1 and tm@xmath147y@xmath148fe@xmath0zn@xmath1 , which still manifest a paramagnetic state at the base temperature , are close to the calculated results from the determined cef parameters of the respective co compounds .
the calculated results also fairly well mimic the crossing behavior of the magnetization along [ 110 ] and [ 100 ] directions for r = er , as well as along the [ 111 ] and [ 100 ] directions for r = tm .
for all three directions , the calculated results are slightly larger than the experimental ones , which is most likely due to the @xmath150 uncertainty of the nominal @xmath149 value .
the larger magnetization for er@xmath147y@xmath148fe@xmath0zn@xmath1 than the calculated results below 10 koe is consistent with residual fm interactions between the er@xmath3 local moments .
the magnetization along the three axes for the all fe compounds were calculated based on the molecular field approximation in a self - consistent manner . in the single - ion hamiltonian for the r@xmath3 ions ( eqn .
[ eqn:3 ] ) , with the molecular field approximation , the magnetic interaction term is written as : @xmath151 where @xmath152 is the molecular field .
it obeys the self - consistent condition : @xmath153 @xmath154 where @xmath155 and @xmath156 are the eigenvalues and eigenenergies of the nth eigenfunction ; @xmath157 is the molecular field constant which can be obtained from the ordering temperature : @xmath158 .
the calculated magnetizations were compared with the experimental results in figs .
[ figtb3 ] , [ figdy3 ] , [ figho3 ] , [ figer3 ] and [ figtm3 ] .
all these calculated magnetization values are obviously larger than the experimental results .
this difference is probably due to ( i ) the molecular field approximation over - estimating the molecular field constant as well as the internal field , and ( ii ) the induced moments from the fe site aligning in an antiparallel manner with respect to the r@xmath3 local moments ( as in the case of gdfe@xmath0zn@xmath1 @xcite ) . for rfe@xmath0zn@xmath1
( r = gd - tm).,title="fig:",scaledwidth=45.0% ] + figure [ figdis2 ] shows the magnetic part of specific heat as a function of @xmath159 for rfe@xmath0zn@xmath1 ( r = gd - tm ) .
the magnetic ordering temperature ( @xmath4 ) of r = gd - er members manifests itself as the position of maximum slope , with a decreasing sharpness as r varies from gd to er .
tmfe@xmath0zn@xmath1 does not appear to have any anomaly in the @xmath99 data at @xmath4 .
below @xmath4 , the data sets for r = dy - tm show a broad peak , which shifts closer to its @xmath4 as r varies from dy to tm , whereas the data for @xmath160 show no explicit peak .
if the broad peaks are corresponding to the magnetic excitation energy spectrum associated with cef effect , then the relative positions of these peaks to @xmath4 , to some extend , indicate the ratio of the energy scales of the cef splitting ( for a single ion ) to the magnetic interaction .
the shift of the peak position as r varies from dy to tm indicates that the energy scale of the magnetic order decreases relative to the cef splitting .
such phenomena is consistent with the analysis on the magnetic part of entropy : as shown in the insets of figs .
[ figtb1 ] , [ figdy1 ] , [ figho1 ] , [ figer1 ] and [ figtm1 ] , tb , dy and ho compounds manifest fully released @xmath96 at their @xmath4 ; whereas er and tm compounds still release part of @xmath96 above their @xmath4 , which indicates that , unlike r = gd - ho members , the cef splitting for the @xmath2 electronic configuration of the tm@xmath3 and er@xmath3 may extend above magnetic ordering temperature .
zn@xmath1 ( r = tb - tm ) .
the arrows present the @xmath4 values for rfe@xmath0zn@xmath1 with respective r.,title="fig:",scaledwidth=45.0% ] + based on the assumption that the fe and co series have similar cef splitting ( for a similar r ions ) , the comparison between the magnetic ordering temperature and the cef splitting for different r ions is qualitatively diagrammatized in fig .
[ figlevels ] .
the levels represent the single ion , cef splitting of the hund s ground state of @xmath2 electronic configuration of r@xmath3 , determined from rco@xmath0zn@xmath1 and the arrows represent the @xmath4 values of rfe@xmath0zn@xmath1 .
the @xmath4 value is comparable with the highest energy level of cef splitting for r = ho . for r
= er and tm , the @xmath4 values is about @xmath161 and @xmath162 of the highest cef levels , respectively .
this diagram , though it can not be used to determine the precise energy splitting of the rfe@xmath0zn@xmath1 compounds ( the cef levels have been strongly modulated and mixed by the interaction energy ) , is qualitatively consistent with the specific heat measurements , and indicates that , at least for tmfe@xmath0zn@xmath1 , the cef energy splitting already happens well above its @xmath4 . in summary
, is appears plausible that , due to extremely similar liganal environments , equivalent members of the rfe@xmath0zn@xmath1 and rco@xmath0zn@xmath1 series have similar cef splitting schemes . versus @xmath163 , ( b ) : @xmath164 versus @xmath163 , ( c ) :
@xmath164 versus @xmath159 for rfe@xmath0zn@xmath1 ( r = gd - tm ) .
inset : the blow up @xmath164 data for tmfe@xmath0zn@xmath1 .
the arrow presents the fm ordering temperature.,title="fig:",scaledwidth=45.0% ] + further insight can also be gained from a careful revisiting of the transport data .
the total resistivity of rfe@xmath0zn@xmath1 ( r = gd - yb ) can be written as : @xmath165 where @xmath166 is associated with scattering from the @xmath2 moments and the spin fluctuation of itinerant electrons . as seen in fig .
[ figdis3 ] a , for the whole series above 250 k , the resistivity data sets show essentially linear behavior with slopes differing by less than 12% , within the estimated dimension error ( @xmath167% ) of these bar - like - shape samples .
these similar , high temperature behaviors indicate that , in the high temperature limit , the magnetic scattering is saturated , and the phonon scattering is essentially similar for the whole series ( due to the very dilute nature of the r ions ) . therefore , the magnetic contribution to the resistivity can be estimated by ( 1 ) subtracting residue resistivity , @xmath168 ( 2 ) normalizing the high temperature slope of all @xmath71 to that of lufe@xmath0zn@xmath1 and then ( 3 ) subtracting the @xmath169 data from the normalized data .
the result is written as : @xmath170 as shown before , the subtraction background @xmath171 already includes the scattering associated with the spin fluctuation of itinerant electrons .
thus , @xmath164 will not only include the scattering from the @xmath2 moments , but will also include scattering associated with the interaction between the @xmath2 moments and itinerant electrons .
figure [ figdis3 ] b and c show @xmath164 versus temperature , as well as normalized temperature ( @xmath159 ) for r = gd - tm . for r
= gd - er , a pronounced upward cusp , whose height decreases from gd to er , is centered about @xmath4 , whereas tmfe@xmath0zn@xmath1 manifests a broad feature and only very weak anomaly around its @xmath4 ( see the blow - up inset of fig .
[ figdis3 ] ) . as shown in fig .
[ figdis1 ] b , the maximum values of the cusps for different r scale with the de gennes factor , which indicates that the decrease of @xmath164 with @xmath163 below @xmath4 is the result of a loss of spin disorder scattering of conduction electrons , associated with the @xmath2 local moment . however , as found in the pseudo - ternary compounds gd@xmath15y@xmath141fe@xmath0zn@xmath1,@xcite the decreasing behavior of @xmath164 with increasing @xmath163 above @xmath4 is more conspicuous and must come from a different conduction electron scattering process ( simple models of @xmath71 due to magnetic scattering can not explain this anomaly @xcite ) . giving that rt@xmath0zn@xmath1 compounds only manifest this behavior when the local moments are embedded in the highly polarizable background ( gdco@xmath0zn@xmath1 does not show this behavior @xcite ) , this anomaly is thought to be associated with the spin fluctuation of the @xmath11 electrons . also appearing in the resistivity of rco@xmath0 ( r = gd - tm)@xcite , this decreasing behavior of @xmath164 with increasing @xmath163 above @xmath4 has been explained as the result of the increase of the spin fluctuation of @xmath11 electrons , which is provided by the increasing , nonuniform fluctuating @xmath2-@xmath5 electron exchange interaction , as the temperature approaches @xmath4 in the paramagnetic state .
since both y(lu)co@xmath0 and y(lu)fe@xmath0zn@xmath1 are classical examples of nffls , such an anomaly could be associated with these strongly correlated electron systems . on the other hand , considering that the hund s ground state of tm@xmath3 has been significantly split above fm ordering for tmfe@xmath0zn@xmath1 , it is not unexpected that the conduction electron scattering process manifests a different behavior associated with the cef effect . finally , the nearly fm compounds : yfe@xmath0zn@xmath1 and lufe@xmath0zn@xmath1 also merit some further discussion . shown in fig .
[ figylu1 m ] , the low field susceptibility ( h = 10 koe ) manifests a maximum near 6 k and 8 k for yfe@xmath0zn@xmath1 and lufe@xmath0zn@xmath1 respectively .
such a maximum in the temperature dependent susceptibility also appears for other examples of nearly fm compounds .
for example , pd manifests @xmath172 k @xcite ; yco@xmath0 and luco@xmath0 manifests @xmath173 k@xcite ; and tibe@xmath0 manifests @xmath174 k@xcite .
another interesting phenomena in nearly fm materials is the so - called itinerant electron metamagnetism ( iem ) , which is an applied magnetic field induced , first order , phase transition between a paramagnetic state and spin polarized state @xcite .
experimentally , iem has been observed for yco@xmath0 and luco@xmath0 around 70 t. @xcite within the framework of landau theory , the maximum in temperature dependent susceptibility is thought to be related to iem.@xcite the magnetic part of the free energy @xmath175 can be writen as the function of the magnetic moment @xmath176 : @xmath177 where @xmath178 , @xmath179 and @xmath180 are the landau expansion coefficients .
as shown by shimizu @xcite , the condition for the existence of iem is : @xmath181 , @xmath182 , @xmath183 and @xmath184 . within the framework of the spin fluctuation theory ,
yamada @xcite generalized this work by introducing a temperature dependent function of the mean square amplitude of spin fluctuations .
these theoretical works demonstrated that the existence of iem is associated with the maximum in @xmath36 by means of the factor of @xmath185 , which can be estimated as : @xmath186^{-1}. \label{eqn:12}\ ] ] furthermore , the iem can only happen below @xmath35 .
these results seem to be consistent with the experimental results in various itinerant electronic systems . @xcite according to the eqn .
[ eqn:12 ] , the values of @xmath185 can be estimated as 310 and 72 for yfe@xmath0zn@xmath1 and lufe@xmath0zn@xmath1 respectively ( @xmath187 at 10 koe ) , which are much larger than the region of the existence of iem , indicating that iem may not exist .
indeed , recent measurements on a part of the lufe@xmath0zn@xmath1 sample used for the magnetization data in fig .
[ figylu1 m ] in a pulse magnetic field up to 55 t at 0.3 k , show no evidence of metamagnetic transition . in nearly fm materials ,
no evidence of iem appears for tibe@xmath0,@xcite , which also manifests a relative low value of @xmath35 . from these points of view ,
y(lu)fe@xmath0zn@xmath1 and tibe@xmath0 may represent the examples of nffls different from yco@xmath0 and luco@xmath0 .
this lack of an iem sheds further light on the magnetic properties of the local moment bearing , rfe@xmath0zn@xmath1 ( r = gd - tm ) compounds . as shown before , all the members manifest 2nd order paramagnetic to ferromagnetic phase transitions .
this behavior is different from that seen in the rco@xmath0 ( r = gd - tm ) system : the magnetic phase transitions of r = dy - tm members for rco@xmath0 are 1st order whereas r = gd and tb members have 2nd order transitions@xcite .
this difference is not difficult to explain in landau theory : unlike y(lu)co@xmath0 , the host of y(lu)fe@xmath0zn@xmath1 lack of ability to show iem and therefore can not be induced to show metamagnetic transition by any molecular field associated with the @xmath2 local moments .
rfe@xmath0zn@xmath1 and rco@xmath0zn@xmath1 ( r = gd - lu , y ) demonstrate diverse magnetic properties .
the conspicuous differences between these two related series are mainly associated with the conduction electron polarizability of the host ( non - magnetic ) compounds . yfe@xmath0zn@xmath1 and lufe@xmath0zn@xmath1 manifest similar , nearly ferromagnetic properties . when the @xmath2 local moments are embedded in this highly polarizable medium , the rfe@xmath0zn@xmath1 ( r = gd - tm ) series shows highly enhanced fm ordering temperatures .
in contrast , yco@xmath0zn@xmath1 and luco@xmath0zn@xmath1 manifest normal , pauli paramagnetic behaviors . in a related manner , gdco@xmath0zn@xmath1 and
tbco@xmath0zn@xmath1 show low temperature afm ordering , and the magnetic properties for rco@xmath0zn@xmath1 ( r = dy - tm ) are more strongly influenced by the cef splitting of the r ions ( the dominant energy scale ) .
the cef coefficients determined for the co series are consistent with the observed anisotropies of the fe series , indicating that the cef splitting of the r - ions is similar for the all zn frank - kasper polyhedra that fully encompass the r - site . on the other hand , ybfe@xmath0zn@xmath1 and ybco@xmath0zn@xmath1 manifest different heavy fermion behaviors .
the authors thank j. frederich for growing some of the compounds , l. tan for laue x - ray measurements , e. d. mun , y. janssen and r. prozorov for helpful discussions .
ames laboratory is operated for the u.s .
department of energy by iowa state university under contract no .
de - ac02 - 07ch11358 .
this work was supported by the director for energy research , office of basic energy sciences .
figure [ figappentb ] shows the magnetization ( at h = 1000 oe ) and zero applied field resistivity for three batches of tbfe@xmath0zn@xmath1 , which were synthesized from different initial molar ratio of starting elements , tb : fe : zn = 2:3:95 , 2:4:94 and 2:5:93 . the ferromagnetic ordering temperatures , determined as @xmath188 k , @xmath189 k and @xmath190 k by resistivity measurements for the three samples , increase as the initial concentration of fe in the solution increases .
we also found similar features for r = gd and er , but the variation of their @xmath4 values are less than in the tb case ( fig .
[ figappen2 ] ) .
comparative , single crystal x - ray diffraction measurements performed on the samples , albeit inconclusive , indicated that the crystallographic difference were perhaps due to subtle variations of occupancy of the fe site.@xcite the main difficulty associated with the x - ray diffraction measurement was completely resolving the mixed site occupancies for zn and fe which have similar atomic number values . recently ,
carefully prepared , pieces of tbfe@xmath0zn@xmath1 samples with same geometric dimension , from the starting elements , tb : fe : zn = 2:3:95 and 2:5:93 , were measured using single - crystal neutron diffraction.@xcite this measurement showed that the fe site has @xmath191% deficiency for the 2:3:95 sample .
all these crystallographic measurements indicate the sensitivity of the magnetic properties to the small disorder for rfe@xmath0zn@xmath1 in this family which has local moments submerged in a highly polarizable conduction electron matrix . as shown in fig .
[ fig1poly ] , the distance between the rare earth ion and the zn nns , as well as nnns , is close to 3 ; whereas the distance with the next next nearest neighbors ( nnnn , 6 zn in @xmath192 site ) is larger than 5 . due to this isolated , cage - like coordinate of rare earth ions , the effect of ions other than this cn-16 frank - kasper polyhedron can be neglected in the calculation of the cef coefficients , based on the point charge model . appreciably , the ligands of the rare earth ions in the c-15 laves compounds ( rni@xmath0 ) form a same polyhedron , whose cef coefficients have been calculated by b. bleaney @xcite , based on the point charge model .
therefore , one can directly cite the results : where @xmath195 and @xmath196 are the charge of the nn and nnn ions ( @xmath197 for zn@xmath198 ) , @xmath199 and @xmath200 is the distance between the r ion and the two sets of ions , @xmath201 and @xmath202 are the mean fourth and sixth powers of the electronic radius for the @xmath2-electrons , and @xmath63 and @xmath66 are the steven multiplicative factors . extracting the values of @xmath201 and @xmath202 from ref.@xcite , @xmath63 and @xmath66 values from ref.@xcite , and @xmath199 and @xmath200 values from the results of single crystal x - ray diffraction@xcite , one can calculate the @xmath203 and @xmath204 values ( shown in table [ table3 ] ) . | magnetization , resistivity and specific heat measurements were performed on solution - grown , single crystals of rfe@xmath0zn@xmath1 and rco@xmath0zn@xmath1(r = y , nd , sm , gd - lu ) . whereas luco@xmath0zn@xmath1 and yco@xmath0zn@xmath1 manifest unremarkable , metallic behavior , lufe@xmath0zn@xmath1 and yfe@xmath0zn@xmath1 manifest behaviors characteristic of nearly ferromagnetic fermi liquids . when the well defined @xmath2 local moments ( gd@xmath3 - tm@xmath3 ) were embedded into this strongly polarizable host ,
they all manifest enhanced , ferromagnetic ordering and the values of @xmath4 for rfe@xmath0zn@xmath1 ( r = gd - tm ) scale with the de gennes factor . in addition , the data on the rfe@xmath0zn@xmath1 compounds indicate a small crystal electronic field ( cef ) effect , compare with the interaction energy scale . on the other hand , the local moment bearing members of rco@xmath0zn@xmath1 ( r = nd , sm , gd - tm ) manifest
weak magnetic interactions and the magnetic properties for r = dy - tm members are strongly influenced by the cef effect on the r ions .
the magnetic anisotropy and specific heat data for the co series were used to determine the cef coefficient of r ion in the cubic coordination .
these cef coefficients , determined for the co series , are consistent with the magnetic anisotropy and specific heat data for the fe series , which indicates similar cef effects for the fe and co series .
such analysis , combined with specific heat and resistivity data , indicates that for r = tb - ho , the cef splitting scale is smaller than their @xmath4 values , whereas for erfe@xmath0zn@xmath1 and tmfe@xmath0zn@xmath1 the @xmath2 electron loses part of its full hund s rule ground state degeneracy above @xmath4 . ybfe@xmath0zn@xmath1 and ybco@xmath0zn@xmath1 manifest typical , but distinct heavy fermion behaviors associated with different kondo temperatures . |
the aim of this paper is to study the dynamics in the neighborhood of unstable periodic orbits in a rotating three dimensional ( hereafter 3d ) autonomous hamiltonian system of galactic type .
instability of periodic orbits in such a dynamical system is of three kinds : simple , double and complex instability .
the case of complex instability was studied in a recent paper by katsanikas et al ( katsanikas et al . 2011 ) . here
we study the two other cases .
furthermore , we examine the stickiness effects in our 3d system .
this is the first time that the stickiness phenomenon is _ explicitly _ investigated in association with these kinds of instability ( simple and double ) in a 3d autonomous hamiltonian system . for this purpose
we consider the intersections of the orbits by a space of section . in a 3d autonomous hamiltonian system
the space of section has 4 dimensions .
so , for the visualization of the 4d spaces of section it is convenient to use the method of color and rotation ( patsis & zachilas 1994 ) .
rotation is used for understanding the geometry of a 3d projection and color is used to visualize the 4th dimension .
3d rotating autonomous hamiltonian systems are of special importance in galactic dynamics .
they represent in a realistic way rotating galaxies , in which rotation plays a key role in the dynamics on the equatorial plane , as well as in the third dimension .
the most important family for the vertical morphology of rotating disk galaxies is a family called x1v1 by ( patsis et al .
the x1v1 family is introduced at the transition from stability to simple instability ( @xmath0 ) as the stability of the central , planar family x1 undergoes a @xmath1 transition within a small jacobi integral interval at the vertical 2:1 resonance region ( for a definition of resonances in rotating galactic potentials see contopoulos pages 379 and 422 ) . at the @xmath2 transition , another family of periodic orbits ( hereafter p.o . ) ,
x1v2 , is bifurcated from x1 as simple unstable ( skokos et al .
2002 ) . for larger jacobi constants x1v2 becomes double unstable .
thus , it possesses both kinds of instabilities we want to study . in this paper
we focus on x1v2 and we examine the structure of the phase space in its neighborhood , when it is simple or double unstable .
however , in order to investigate in depth the dynamics in the neighborhood of double unstable periodic orbits we will also study a case referring to z - axis orbits .
the importance of the dynamical phenomena at the 2:1 vertical resonance is due to the role of the associated families in shaping the vertical morphology of the 3d disks .
they form to a large extent the peanut - shaped structure , which is characteristic of many disk galaxies observed edge - on ( patsis et al 2002 ) .
in addition , chaotic motion in the neighborhood of unstable p.o.s has been proven to be significant for the dynamics of rotating galactic disks _ on _ their equatorial planes ( for a review see e.g. contopoulos 2009 ) . in that respect the knowledge of the structure of phase space in the neighborhood of 3-dimensional p.o . in galactic systems
is the first step for the investigation of a possible contribution of chaotic orbits to the observed vertical profiles of this type of galaxies .
the space of section that we use in this paper for investigating the orbital behavior close to x1v2 orbits is defined by @xmath3 and we consider intersections with @xmath4 .
hereafter by `` intersections '' we will refer to intersections , which fulfill these properties , until we start discussing the dynamics close to z - axis . then we will change space of section .
every consequent in our cross section is determined by 4 cartesian coordinates , e.g. @xmath5 . in the method of color and rotation we project the consequents in a 3d subspace and observe them from different viewpoints . in the text
the viewpoints are given in spherical coordinates .
the unit for the distance @xmath6 of the consequents of the surface of section from the observer corresponds to the longest side of the bounding box of the figure ( more details can be found in katsanikas and patsis 2010 ) . in practice
we rotate the 3d figure in the computer screen in order to find the most appropriate viewpoint for the presentation on the 2d surface of the pages of a paper .
we select the viewpoint that represents best the morphology of the structures formed by the consequents in the 3d projections of the 4d space of section in order to facilitate the reader to follow our description .
then we color every point according the value of its 4th coordinate . for this reason we use the `` mathematica '' package and
we map the interval of the values of the fourth coordinate into @xmath7 $ ] . * a well defined structure in all 3d projections and a smooth color variation along them is evidence for regular motion . *
mixing of colors indicates chaotic behavior of the consequents in 4d ( patsis & zachilas 1994 , katsanikas & patsis 2010 ) the linear stability of the periodic orbits is studied , following the analysis by broucke ( broucke 1969 ) in the case of the elliptic restricted three body problem and the method used by hadjidemetriou ( hadjidemetriou 1975 ) in the three body problem .
the method has been also described in a recent earlier paper of us in the series of papers studying the dynamics in the neighborhood of p.o . in 3d rotating galactic potentials ( katsanikas & patsis 2010 ) .
the characteristic equation in a 3d autonomous hamiltonian system is of the form @xmath8 . and has 4 eigenvalues @xmath9 , @xmath10 , @xmath11 , @xmath12 , where @xmath13 , @xmath14 , with @xmath15 and @xmath16 .
thus , the stability of the p.o . depends on the two stability indices @xmath17 and @xmath18 and on the quantity @xmath19 . for @xmath20 and @xmath21 , all four eigenvalues are on the unit circle in the complex plane and the orbit is called stable @xmath22 ( see e.g. contopoulos 2002 , contopoulos & magnenat 1985 )
. then we have three kinds of instability : 1 .
simple instability @xmath23 : @xmath24 and @xmath25 or @xmath26 , ( two eigenvalues on the unit circle and two on the real axis ) 2 .
double instability @xmath27 ) : @xmath28 and @xmath29 and @xmath30 ( four eigenvalues on the real axis ) 3 .
complex instability ( @xmath31 ) ( all four complex eigenvalues are off the unit circle in the complex plane ) : @xmath32 ( for details see contopoulos 2002 , p.286 )
. hereafter we will indicate the periodic orbits as @xmath33 , @xmath34 , @xmath35 and @xmath36 according to their stability .
a generalization of the three kinds of instability in hamiltonian systems of @xmath37 degrees of freedom can be found in skokos ( skokos 2001 ) . in section 2
we present our hamiltonian system . in section 3
we discuss the stability of a family of p.o .
we call x1v2 . in section 4
we study the orbital behavior close to simple unstable orbits of x1v2 , in section 5 we investigate the dynamical behavior in the neighborhood of double unstable orbits of x1v2 , as well as of the z - axis families , in section 6 we compare the diffusion of the chaotic orbits away from the p.o . and
finally in section 7 we discuss and enumerate our conclusions .
the hamiltonian of our system in cartesian coordinates is : @xmath38 where @xmath39 is the potential : @xmath40 ^ 2)^{1/2}}- % \nonumber\\ \frac{gm_{2}}{(x^2 + \frac{y^2}{q_a^2 } + [ a_{2}+(\frac{z^2}{q_b^2}+b_{2}^2 ) ^{1/2}]^2)^{1/2 } } % \nonumber\\\ ] ] this potential in its axisymmetric form @xmath41 can be considered as an approximation of the potential of the milky way ( miyamoto & nagai 1975 ) . in our units
, the distance @xmath42=1 corresponds to 1 kpc .
the velocity unit corresponds to 209.64 @xmath43 .
the jacobi constant @xmath44 corresponds to 43950 @xmath45 and we will call it `` energy '' . for the parameters we have used the following values : @xmath46 . for the angular velocity we used the value @xmath47 in our units .
the parameters @xmath48 determine the geometry of the disks , while @xmath49 are scaling factors .
the specific values of the geometrical parameters do not have a particular physical meaning .
the dynamical phenomena we study in our paper are typical for rotating triaxial systems .
the @xmath50 value we use here is different from the one we used in ( katsanikas & patsis 2010 ) in order to have larger regions in the parameters space with non - escape orbits close to double unstable periodic orbits ( section 5 ) .
in order to study the evolution of the stability of a family of periodic orbits as a parameter of our system varies we use the stability diagram ( contopoulos 2002 p.287 ) .
the stability diagram gives the values of the coefficients @xmath17 and @xmath18 for successive periodic orbits in a family by changing only one parameter of the system ( in our case @xmath51 ) , while keeping the other parameters constant . in fig .
[ dsimple1 ] we observe in such a diagram how the family of periodic orbits x1v2 ( skokos et al .
2002 ) bifurcates from the 2d central family x1 on the plane @xmath52 at the @xmath53 transition that happens in the vertical 2:1 resonance region ( contopoulos 2002 p.390 ) .
this bifurcation occurs for @xmath54 .
we observe that the stability parameters @xmath17 , @xmath18 of the family x1v2 decrease as @xmath51 increases .
the value of @xmath18 is initially about @xmath55 at @xmath56 ( left side of diagram ) .
when @xmath18 becomes @xmath57 we have a transition of x1v2 from simple to double instability ( @xmath58 ) ( for @xmath59 ) and the x1v2 periodic orbits become double unstable .
the @xmath51 interval of x1v2 , where it is simple unstable ( @xmath60 ) and the @xmath51 interval of x1v2 where it is double unstable ( @xmath61 ) , will be called hereafter regions u and du respectively ( fig .
[ dsimple1 ] ) .
first we study the orbital behavior of the x1v2 family in the region u. in the following applications we use the set of projections @xmath62 and @xmath63 . however , we find qualitatively similar results by using the @xmath64 and @xmath65 projections respectively .
the use of the two projections gives us all needed information to understand the structure of the distribution of the consequents in the 4d space by means of the color - rotation method , as we will see below .
the initial conditions to be studied are chosen in two different ways .
first we investigate the dynamical behavior of the orbits in the neighborhood of the simple unstable periodic orbit x1v2 at @xmath66 by varying the values and the direction ( @xmath67 , @xmath68 , @xmath69 or @xmath70 ) of the perturbation .
secondly , we consider a constant perturbation @xmath71 and we apply this perturbation to the initial conditions of several periodic orbits of the x1v2 family in the region u for different @xmath51 values .
we start by keeping @xmath51 constant and investigate as a sample case the dynamical behavior very close to the point where we have the @xmath53 transition of the x1 family and the bifurcation of the family x1v2 as @xmath34 .
the energy @xmath66 is indicated with an arrow in fig .
[ dsimple1 ] .
the periodic orbit has initial conditions @xmath72 . in order to study the orbital behavior in its neighborhood we added a small perturbation in the @xmath73-direction equal to @xmath74 .
we call this perturbed orbit `` u '' .
the first 2580 consequents of orbit `` u '' define a ribbon of the form of a double loop in the @xmath62 subspace ( fig .
[ dsimple2 ] ) . in the bottom of the figure we give also from left to right the 2d projections @xmath75 and @xmath76 in order to help the reader understand the topology of the structure .
the two lobes of the double loop are inclined with respect to each other and form an angle around @xmath77 in the @xmath78 projection ( middle panel in the lower part of fig .
[ dsimple2 ] ) .
the @xmath78 projection clearly also shows that the thickness of the lobes is very small ( but not zero ) and this is the reason we describe them as ribbons .
essentially the branches of the lobes are thin tubes .
the 4d representation of this structure , colored according to the normalized @xmath73 values and using @xmath62 for the 3d spatial projection is given in fig .
[ dsimple2 ] ( top ) .
for the 3d projection we use here a point of view defined by the angles @xmath79 .
we observe that this ribbon has a smooth color variation from red to violet , in agreement with structures found in the neighborhood of simple periodic @xmath23 orbits by patsis and zachilas ( 1994 ) .
the structure of the phase space close to simple unstable periodic orbits has been studied for the first time by magnenat ( 1982 ) .
however , magnenat used 2d projections and these did not allow the full understanding of the 4d structure of the double loop objects . in the present case
, our 4d representation allows us to understand the details of this structure . by inspection of fig .
[ dsimple2 ] we infer that the color variation along the two branches is such that at the intersection region we have the meeting of two different shades .
this means that in this region the points have different values in the 4th dimension and there is no real intersection in the 4d space of section .
consequently the ribbon - like double loop structure has not an intersection in 4d .
we observe that the colors at the region of the intersection are not only different , but that they are at the two extremes of our color scale , i.e. violet and red ( see right side of fig .
[ dsimple2 ] ) .
the colors in this region are clustered around two values @xmath80 very close to the @xmath73 initial condition of the periodic orbit @xmath81 .
figure [ dsimple2 ] reflects the symmetry of the galactic potential we study . due to this symmetry , the families of p.o . that bifurcate at the vertical resonances from the planar x1 family come in pairs with mirror morphology with respect to the equatorial plane of the system in the configuration space , i.e. x1v1 and x1v1@xmath82 , x1v2 and x1v2@xmath82 etc . the p.o .
of x1v2 and x1v2@xmath82 at every @xmath51 differ only in the sign of their @xmath83 initial conditions . in the 3d @xmath62 projection we use in fig .
[ dsimple2 ] , the initial conditions of x1v2 and x1v2@xmath82 are identical and the p.o .
is in the center of the intersection region of the two branches of the double loop ribbon ( indicated with an arrow ) .
the clustering of the colors around the two values @xmath84 in the central region is due to the coexistence of x1v2 and x1v2@xmath82 .
the consequents drift away from the central region of the double loop by deviating from the two @xmath85 initial conditions with different colors .
the first 2580 consequents are along the double loop structure , thus they indicate a regular rather than chaotic motion .
we reach the same conclusion by using other 3d projections for the spatial coordinates of the consequents . in fig .
[ dsimple3 ] the first 2580 consequents of the orbit `` u '' in the 3d projection @xmath63 form again a ribbon - like surface , however this time with a single loop morphology . color is determined by the @xmath86 value of the consequents and we observe a smooth color variation along the ribbon . in this case
the two representatives of the p.o x1v2 and x1v2@xmath82 are at two different positions in the @xmath63 space , since they have different @xmath83 initial conditions .
their positions are indicated with arrows in fig .
[ dsimple3 ] .
the regions close to both of them are colored red ( same @xmath86 values ) and departing from them , red becomes orange , then yellow , green , light blue , blue and finally violet .
the ribbon is a 4d object . from both figs .
[ dsimple2 ] and [ dsimple3 ] we conclude that the first 2580 consequents form a well - defined ribbon - like surface that has no self - intersections in 4d
. a similar analysis can be done by means of the two other 3d projections @xmath87 and @xmath65 , which give 4d structures morphologically similar to @xmath62 and @xmath63 respectively .
it is evident that the orbit `` u '' in the neighborhood of a @xmath34 x1v2 p.o . for 2580 consequents
is confined in a particular region of the space of section with dimensions ( @xmath88 .
as we will see below , for larger integration times the orbit diffuses to a larger region of the phase space .
the structures we find in the phase space indicate a sticky behavior and this will be further analyzed in section 4.4 .
cc + cc + we note that although in general we have different colors in the intersections of double loop type structures , like the one in fig .
[ dsimple2]a , in exceptional cases we can have also real intersections in the 4d space .
such a case is encountered in the neighborhood of @xmath34 z - axis orbits in 4d spaces of section defined by @xmath52 .
the z - axis orbits lie entirely on the rotational , z axis , of the system .
they have initial conditions @xmath89 and are discussed in detail in section 5.2 in the context of the study of double instability .
if we compute a larger number of intersections we observe a drastic change in the topology of the ribbon - like surface .
the points deviate from the well - defined 4d structures that we described above and form a cloud of points around them in the 3d projections with mixed colors in their 4d representations .
however , in the particular case we study here , for about 3580 intersections , i.e. for 1000 consequents more than those building the structures in figs .
[ dsimple2 ] and [ dsimple3 ] , in the 3d projections that include the @xmath90 and @xmath73 coordinates we discern a structure ( i.e. a greek letter @xmath91 rotated by @xmath92 ) surrounded by a ring .
this can be seen in fig .
[ dsimple7 ] , where we depict the consequents of the orbit `` u '' in a @xmath63 projection and color them according to their @xmath86 values .
this new structure in the 3d projection of the phase space is maintained for several thousands of intersections more .
we have again a confinement of the consequents in a restricted volume of the phase space , which is larger than previously .
after 3580 intersections the consequents extend within a region ( @xmath93 . in the new situation we have on one hand the appearance of structures in certain projections which indicates order , and on the other hand mixing of colors , which indicates chaoticity .
cc + the origin of the observed structure can be understood best in the ( @xmath94 ) projection ( fig .
[ x1v1v2 ] ) .
we remind that we study the structure of phase space in the region u , which starts with the bifurcation of the family x1v2 at the @xmath53 transition of x1 in the vertical 2:1 resonance region ( fig .
[ dsimple1 ] ) . for smaller energies in the resonance region , at @xmath95
, we have the @xmath0 transition of x1 and the introduction in the system of the stable family x1v1 ( and its symmetric family x1v1@xmath96 ) .
thus for @xmath66 , we have , apart from the two simple unstable p.o . x1v2 and x1v2@xmath96 , the two stable x1v1 and x1v1@xmath96 and the planar stable x1 .
the presence of all these simple periodic orbits determines the orbital behavior of `` u '' in the following way : in the central part of fig .
[ x1v1v2]a we observe a red `` ellipse with corners '' , which forms within the first 2580 intersections and which is the projection of the structure of fig ; [ dsimple3 ] in the ( @xmath94 ) plane .
the initial conditions of x1v2 and x1v2@xmath96 are located at the upper and lower corner of the red elliptical curve respectively .
the blue ellipse inside the red one is the projection of a rotational torus ( vrahatis et al .
1997 ) belonging to a quasiperiodic orbit found by perturbing x1 in the @xmath73 direction . during the first 2580 consequents the orbit `` u '' stays sticky to the outer 4d rotational tori of x1 .
the outermost rotational tori belonging to x1 are broader than the ribbon - like tori of the `` u '' orbit . in fig .
[ x1stick ] we give two characteristic projections from the point of view @xmath97 in spherical coordinates . in ( a )
we have the @xmath98 and in ( b ) the @xmath99 projection . in both of them
we overplot the 2580 consequents of `` u '' ( red ) and one of the outermost rotational tori around x1 ( black ) .
cc + + cc + + by continuing the integration of the orbit `` u '' for more than 2580 intersections we see that the next 550 intersections form a -like structure with two lobes on the sides of the red elliptical structure ( fig .
[ x1v1v2]a , indicated with a couple of white arrows ) .
these consequents surround the rotational tori around the stable p.o . x1v1 and x1v1@xmath82 , which are very thin and appear in this projection like invariant curves ( green and magenta ) .
the rotational tori of these two families are still in the phase space region close to the @xmath34 p.o .
x1v2 , since they have been bifurcated at a nearby @xmath51 from x1 ( fig .
[ dsimple1 ] ) .
the initial conditions of x1v1 and x1v1@xmath82 are in the middle of the rotational tori , marked with black symbols .
considering more than 3130 consequents we find that the next 1370 intersections stay trapped in a ring in the @xmath100 projection , which in its turn surrounds all structures we have found in the phase space up to now .
we point to this ring with two black arrows in fig .
[ x1v1v2]a ( right above and left below ) . between the two lobes surrounding the rotational tori around x1v1 and x1v1@xmath82 and the ring structure ,
we observe a chain of eight stability islands ( colored black ) , which are the @xmath100 projections of the corresponding rotational tori of a stable 8-periodic orbit . finally , after 4500 intersections
, the consequents start diffusing into a larger volume of the phase space .
figure [ x1v1v2]a depicts the space of section after 5000 intersections .
after 15000 intersections the consequents have occupied all available phase space ( fig .
[ x1v1v2]b ) .
however , the denser regions clearly show us where the orbits have been trapped for longer times .
we have to compare the structures described up to now with the 4d space of section we presented in fig .
[ dsimple7 ] .
the projections in fig .
[ x1v1v2 ] are part of fig .
[ dsimple7 ] . we observe them as we rotate the @xmath63 diagram in our computer screen . since fig .
[ dsimple7 ] consists of 3580 consequents , apart from the ribbon it includes also the consequents forming the -like structure around the rotational tori of x1v1 and x1v1@xmath82 as well as part of the consequents that form the `` ring '' in fig .
[ x1v1v2 ] . in fig .
[ dsimple7 ] we emphasize on the distribution of the consequents in the 4th dimension . we observe that in fig .
[ dsimple7 ] the color , representing the normalized @xmath86-coordinate values , is dominated by green shades .
this means that although we do not have a smooth color variation as in figs .
[ dsimple2],[dsimple3 ] the consequents stay confined in a restricted volume of phase space .
this happens indeed .
the @xmath86 values for the first 3130 intersections are in the range @xmath101 , while within the next 450 intersections this range is almost doubled . due to this expansion of the range of @xmath86 values , in fig .
[ dsimple7 ] are introduced the orange - red and blue - violet consequents .
the fact that in 3d projections we find structure and that the color variation is dominated by a particular shade , indicates a weakly chaotic behavior of the orbit `` u '' .
this information is included in the 4d representation of the phase space as it is given in fig .
[ dsimple7 ] . by combining figs .
[ dsimple7 ] and [ x1v1v2 ] on a computer screen , we reveal not only the weakly chaotic character of the orbit , but also the regions where it sticks during the integration time .
this shows the efficiency of the color - rotation method .
one can calculate for the orbit `` u '' the finite - time lyapunov characteristic number , @xmath102 , which is defined by @xmath103 where @xmath104 and @xmath105 are the distances between two points of two nearby orbits at times @xmath106 and @xmath107 respectively ( see e.g. skokos 2010 ) . as we can see in fig .
[ lcn ] the changes we have found in the orbital behavior of `` u '' do correspond to characteristic changes of the slope of the @xmath102 curve .
cc + + during the first 2580 intersections , indicated by the left arrow in fig .
[ lcn]a the curve decreases towards zero .
then during the next 450 intersections , until the time pointed by the right arrow in fig .
[ lcn]a when the consequents form the -like structure with the lobes around the x1v1 and x1v1@xmath82 tori , it continues to decrease but the curve has a much smaller slope .
after that it starts increasing and its evolution after this point until @xmath108 can be followed in fig .
[ lcn]b ( note the different scales in the vertical axes ) .
@xmath102 increases , then levels off , and has a decreasing part until the left arrow in fig .
[ lcn]b . which corresponds to about 4180 intersections .
at this point we have an expansion of the consequents in phase space , which can be understood in the @xmath109 projection ( not shown in a figure ) . beyond the left arrow in fig .
[ lcn]b the @xmath102 curve increases . then , between fig .
[ lcn]b and [ lcn]c the curve decreases and in fig . [ lcn]c it increases again . finally , for longer times of integration , when the orbit explores all the available phase space , the curve levels off at a value @xmath110 ( fig . [ lcn]c ) the right arrow in fig .
[ lcn]b corresponds to 4500 intersections and the region to the right of this arrow corresponds to the expansion of the consequents from the ring to the whole available volume of phase space as we observe in fig .
[ x1v1v2]b .
although @xmath102 gives globally the variation of the volumes of the phase - space filled by the consequents , by means of the color - rotation representation we can see in detail the specific regions where the consequents are trapped and the coordinate along which they diffuse to occupy larger volumes of the phase space .
this information is essential in galactic dynamics for building self - consistent models using the orbital theory . summarizing the investigation of the orbital behavior of `` u '' in the neighborhood of a simple unstable periodic orbit
we underline that it is characterized by two sticky periods , one associated with the rotational tori of x1 and a second sticky period associated with the rotational tori of x1v1 and x1v1@xmath82 . during the latter sticky period we observe the structure and the ring in the @xmath111 projection . in the following section
we examine how general is this behavior . a dynamical behavior similar to that of the orbit `` u '' is observed by perturbing the @xmath83 initial condition of x1v2 up to @xmath112 ( i.e. an absolutely largest @xmath113 ) .
we find a similar behavior if we perturb @xmath114 by at most @xmath115 .
initial conditions that have the same @xmath116 , @xmath117 and @xmath118 with x1v2 and @xmath83 that deviates by @xmath119 , give rotational tori around x1 ( see katsanikas & patsis 2010 ) .
another class of orbits in the neighborhood of x1v2 is characterized by immediate stickiness around the x1v1 and x1v1@xmath96 rotational tori , without building initially a `` ribbon '' .
we find them by keeping the same energy constant ( @xmath66 ) and computing orbits by adding the perturbations given in table 1 to the initial conditions of x1v2 .
in all these cases we do not observe the well defined ribbon - like surface that we have seen in figs .
[ dsimple2 ] and [ dsimple3 ] .
the orbits , starting close to the p.o .
x1v2 , depart directly from its immediate neighborhood and stick around the rotational tori of x1v1 and x1v1@xmath96 , before they expand and explore all available phase space .
they form -shaped structures surrounded by rings as in fig .
[ dsimple7 ] without the central elliptical structure we observe in fig .
[ x1v1v2 ] . in table 1
we give the ranges of the perturbations of the initial conditions for which we have observed this orbital behavior . for deviations from the p.o . in the @xmath73 direction in the interval @xmath120
we reach the eight tori of the stability islands ( fig .
[ x1v1v2 ] ) . by perturbing an initial condition
we keep all others equal to the corresponding initial condition of the p.o . for perturbations larger than those in table 1 in the @xmath86 , @xmath114 or @xmath73-direction we do not find any stickiness but the consequents form clouds around the initial conditions of the p.o .
these clouds are not confined in a volume smaller than the total available phase space .
they occupy always larger volumes as the integration time increases .
we have to take into account that for larger @xmath69 perturbations than the listed in table 1 we first fall on initial conditions of the rotational tori of x1v1 and x1v1@xmath82 and then for even larger perturbations we find again the chaotic sea .
the second way we followed to explore the structure of phase space around x1v2 , is by varying @xmath51 , keeping the same perturbation in its initial conditions .
we did so by adding a @xmath121 perturbation in its @xmath116 initial conditions , for @xmath51 s in the region u. for @xmath122 the orbital dynamics of the orbits are as in the case for @xmath66 that we described in section 4.1 .
however , for @xmath123 a cloud of points surrounds the initial conditions of x1v2 in the 4d phase space .
this means that in all 3d projections we observe clouds surrounding x1v2 from the beginning of the calculation and also a mixing of colors .
let us now try to understand the orbital behavior of the orbits we have studied in the previous section by investigating the structure of the asymptotic curves of x1v2 in the 4d spaces of section .
the asymptotic curves are computed by mapping a line segment in the space of section on an associated dilating eigenvector , starting very close to a simple unstable periodic orbit x1v2 ( see e.g. magnenat 1981 ) .
the example we present is again for @xmath66 .
we have already studied the structure of the phase space in the neighborhood of this simple unstable periodic orbit in sections 4.1 and 4.2 .
the initial conditions of the p.o . have been given in section 4.1 .
we consider the unstable eigenvector + @xmath124 and the corresponding eigenvalue @xmath125 .
we compute the asymptotic curves of the unstable manifold and we study the morphology of its asymptotic curves .
the initial segment is composed of the initial conditions @xmath126 of the form : @xmath127 first we compute the asymptotic curves for @xmath128 and we call this part of the unstable manifold , part a. the initial segment for part a is composed of @xmath129 points ( for @xmath130 with a step @xmath131 ) .
for each point , 50 consequents were calculated . the basic criterion for choosing the values of the perturbations and
the number of the consequents is to stay on the asymptotic curves with sufficient accuracy .
the monodromy matrix conserves its symplectic character ( determinant @xmath132 ) with an accuracy at least of the order of @xmath133 . in fig .
[ dman1 ] we observe the first 50 consequents of part a of the cc + unstable manifold in the @xmath62 subspace of the 4d space of section .
the first 2 consequents are very close to the periodic orbit .
then , we discern the 3rd consequent ( indicated with an arrow ) .
the next 13 consequents follow the direction of the black arrows , starting to the right and upwards , and they form a loop . after that , we return back to the neighborhood of the periodic orbit .
then , the 16th consequent as well as the following 7 remain again very close to the periodic orbit . from the 24th consequent and for the next 14 consequents
we observe the formation of another loop , below the first one , following the directions that are indicated by the arrows .
the final 12 consequents are again very close to the periodic orbit .
the evolution in time of an initial segment of initial conditions leads to the formation of a ribbon - like double loop structure , similar to the one in fig .
[ dsimple2 ] . the same holds for the @xmath63 3d projection of the manifold , which has a similar structure with the corresponding projection of the orbit `` u '' ( fig .
[ dsimple3 ] ) . as in figs .
[ dsimple2],[dsimple3 ] color on the asymptotic curves varies smoothly , so the 4d morphology is as for the orbit `` u '' .
for @xmath134 in the initial segment of the initial conditions the asymptotic curves evolve in the way we describe below and form the second part of the unstable manifold .
we call this second part , part b. the initial segment of initial conditions for part b is composed by @xmath135 points ( for @xmath136 with a step @xmath133 ) . for every point we computed this time 20 consequents . in fig .
[ dman6 ] we depict the first 60 points of cc + the initial segment in the @xmath62 projection and color them according to their @xmath73 values .
the figure helps us understand that the asymptotic curves of part b of the unstable manifold are 4d curves , which wind around the point corresponding to the x1v2 p.o . in the space of section .
note that we have a smooth color variation along the curve from light blue to red , following the direction of the successive numbered arrows .
evolving the orbits for a larger number of points ( consequents ) from the initial segment ( 400 instead of 60 points ) we observe that the asymptotic curves of part b oscillate rather far in the @xmath137 projection ( fig .
[ dman7a ] ) .
for example in the @xmath90-direction these cc + oscillations extend between @xmath138 . considering the evolution of all @xmath135 consequents of the initial segment the asymptotic curves
fill roughly the 4d space @xmath139 @xmath140 ( fig .
[ dman7a ] ) .
the figure is complicated and one can discern apart from the formation of loops at the upper left corner in fig .
[ dman7a ] , a ring / disk structure that has at its center the initial conditions of the p.o . as in the case of the orbits we have discussed in subsections 4.1 and 4.2
, we also study the asymptotic curves by combining several 3d projections in which we color their points according to the fourth coordinate . using the @xmath63 3d projection and coloring the points according to their @xmath86 values
the asymptotic curves are depicted in the following way . in fig .
[ dman1011 ] we give the first 400 consequents totally ( 20 points on the initial segment evolved for 20 intersections ) .
they form an asymptotic curve that oscillates in the 3d @xmath63 projection . along the asymptotic curve
the color variation is smooth ( from light blue to red to dark blue and then back to light blue ) , which means that the asymptotic curve is a 4d curve .
evolving further the asymptotic curve in time we observe that the oscillations become more complicated but the smooth color variation continues along the curve .
cc this clearly shows that the oscillations , which have been calculated by magnenat ( magnenat 1982 ) in 2d projections , occur in the 4d space . in order to study the large scale evolution of part b of the unstable manifold
we evolve next 400 points of the initial segment .
the result is depicted in fig .
[ dman12a ] .
cc + after one oscillation on the left side of fig .
[ dman12a]a we have several oscillations in the right part of the figure .
the oscillations become longer than those of the previous case ( fig .
[ dman1011 ] ) and they extend to larger absolute values of @xmath73 in the 3d subspace @xmath63 ( fig .
[ dman12a]a ) . by evolving 20000 points of the initial segment ,
the asymptotic curves form a -shaped surface in the @xmath63 projection ( fig .
[ dman12a]b ) consisting of many parallel filaments .
the filamentary character of this structure is due to the parallel oscillations of the asymptotic curves in the @xmath63 subspace .
there is an obvious similarity of fig .
[ dman12a ] with figs .
[ dsimple7 ] and [ x1v1v2 ] .
the projection of the asymptotic curves in five of the six 2d projections show clouds of points that seem to fill densely the available phase space .
however , in the @xmath100 projection we get a clear insight of how the curves evolve in time . in this projection ( fig .
[ dman14 ] ) we see that the asymptotic curves of part b of the unstable manifold depart from the p.o . in the unstable direction ( arrow 1 )
and then they follow the arcs as indicated by the arrows 2,3,4,5 and 6 .
the asymptotic curves oscillate along a direction vertical to the stable eigendirection .
cc + the morphology we encounter for the case of the stable manifold is similar to that of the asymptotic curves of the unstable invariant manifold .
below we discuss the morphology of the asymptotic curves for other values of the jacobi constant in the region u of simple unstable x1v2 periodic orbits .
we describe now the invariant manifolds in the neighborhood of a simple unstable periodic orbit for values of @xmath51 @xmath141 .
they have again two parts . as the energy increases , for values larger than @xmath66 , the asymptotic curves of part a form a ribbon , double loop , surface in the 3d subspace @xmath62 .
as an example we give the case with @xmath142 ( fig .
[ eman1 ] ) .
although the lobes of the double loop are distorted , they form an acute angle in their @xmath78 projection like the corresponding lobes of the orbit `` u '' ( lower middle panel in fig .
[ dsimple2 ] ) .
we observe also that the consequents wind around the periodic orbit ( as the consequents in fig .
[ dman6 ] ) when they approach the periodic orbit .
the manifold has a similar structure like the one we have encountered in the central region of `` u '' in the same 3d projection . also , in these higher energies , the asymptotic curves of part a form again a circular ribbon - like surface in the 3d subspace @xmath63 similar to the orbit in fig . [ dsimple3 ] .
the increase of energy causes oscillations of the asymptotic curves of part a , in the 3d subspace @xmath143 . in the interval
@xmath144 the invariant manifolds do not have two different forms . along both directions
the asymptotic curves have the same morphology .
the asymptotic curves resemble the asymptotic curves of part b of the unstable invariant manifolds for @xmath66 ( the case that we described in the section 4.3 ) .
cc + from the investigation of the spaces of section and the numerical calculation of the asymptotic curves we presented in the two previous subsections it becomes evident that the phenomenon of stickiness not only is present but characterizes the dynamical behavior of many orbits that start in the neighborhood of the simple unstable orbit x1v2 .
as we know , stickiness in chaos refers to chaotic orbits that stay in a particular region of the phase space , before escaping to large distances . in 2d
autonomous hamiltonian systems stickiness appears near the borders of stability islands , as well as near the asymptotic curves of the unstable manifolds for a long time ( contopoulos & harsoula 2008 , contopoulos & harsoula 2010 ) . in our paper
we explore this phenomenon in the 4d space of section of our 3d system and we try to explain it in combination with the structure of the associated manifolds .
it is the first time that stickiness in chaos is studied in detail in the neighborhood of simple unstable periodic orbits of a 3d autonomous hamiltonian system . in order to study the phenomenon of stickiness in our system we consider first the representative example of a periodic orbit in the region u of the x1v2 family for @xmath66 . if we add a negative perturbation in the @xmath73 direction ( as we mentioned in section 4.2 ) , or add a small positive perturbation in the @xmath114 direction ( for @xmath145 ) to the initial conditions of the periodic orbit x1v2 , the orbits we find have a dynamical behavior similar to the one of the orbit `` u '' ( section 3.1 ) . by comparing the orbital behavior of all orbits having a morphology like `` u '' , with the form of the unstable manifolds we conclude that the initial conditions of `` u '' are very close to the unstable eigendirection of the manifold .
both the consequents of `` u '' and the asymptotic curves wind around the rotational tori associated with x1 .
the consequents of `` u '' follow the asymptotic curves and stick in the region dictated by the manifolds .
for example in figs .
[ dstic1]a and [ dstic1]b the first 2580 consequents of the orbit `` u '' ( depicted by blue color ) stick close to the asymptotic curves of part a of the unstable manifold ( drawn with red color ) .
this number of consequents correspond to a time of 124 periods of rotation of our system , a very large time interval in galactic dynamics ( of the order of several hubble times ) .
cc + + for a larger number of consequents , the points depart from part a of the unstable manifold .
the relation between the consequents of `` u '' and the unstable manifold is best understood again in the 3d projection @xmath63 . as we can see in this projection in fig .
[ dstic3 ] , the first 1000 consequents , after the departure of the consequents from part a of the unstable manifold , remain close to the asymptotic curves of part b of the unstable manifold .
this is the mechanism that creates in phase space the lobes of the and the ring structure we observe as denser regions in the @xmath111 projection in fig .
[ x1v1v2 ] .
these denser regions are created due to the phenomenon of stickiness around the rotational tori of x1v1 and x1v1@xmath82 .
the lobes of the structure , indicated with white arrows in fig . [ x1v1v2 ] , are formed by consequents that stay almost on the ( red ) asymptotic curves of fig .
[ dstic3 ] .
however , even after the departure of the consequents from the lobes they stay close to the x1v1 and x1v1@xmath82 rotational tori , because they are trapped in the immediate neighborhood of the eight islands of stability we observe as small black line segments in fig . [ x1v1v2 ] . in this projection
the figure resembles what is found in 2d systems close to the last kam curve ( contopoulos 2002 ) . however , we should have in mind that in the 4d space of section of our 3d hamiltonian system consequents can be found on both sides of a torus , moving through the additional dimensions .
cc in section 4.2 we said that for the values of perturbation listed in table 1 , the orbits go directly around the rotational tori of x1v1 and x1v1@xmath82 .
all these orbits have also a sticky behavior at the borders of the x1v1 and x1v1@xmath82 4d stability islands , which we can always see best in the 3d projection @xmath63 and the 2d projection @xmath100 . in all other projections
the sticky regions are discernible as denser regions of the phase space embedded in clouds .
the @xmath63 projection has the advantage that in a certain range of viewing angles only a few consequents from the cloud are projected in the regions , which look `` empty '' ( actually they are occupied by the rotational tori around the stable orbits ) .
our study of the dynamical behavior in the neighborhood of the simple unstable orbits , shows that the characteristic double loop structures that have been already found by magnenat ( 1982 ) and patsis and zachilas ( 1994 ) etc .
, as well as the -shaped structures and rings we observe in fig . [ x1v1v2 ] , are associated with stickiness at the borders of rotational tori of stable families that exist in the system .
the consequents follow the manifolds that wind around the tori .
thus it would be more accurate to associate these structure of the phase space mainly with the @xmath53 transitions , as they are not expected to exist away from such a transition point or after a @xmath146 transition .
double instability in rotating galactic potentials usually appears for large energies . in the x1 orbital tree ( skokos et al 2002 ) @xmath35 appears in higher order bifurcations , when a stable 3d bifurcation of x1 becomes first simple unstable and then double unstable , or when a simple unstable bifurcation of x1 becomes double unstable .
frequently we find @xmath35 p.o . at energies
for which nearby orbits escape .
thus the selection of the appropriate case for investigating the phase space structure is crucial .
here we consider first the dynamics in the neighborhood of orbits of the x1v2 family in the region du ( fig .
[ dsimple1 ] ) . however , in order to examine the dependence of the dynamical behavior on the energy of the orbit , we study also the phase space close double unstable z - axis orbits .
these are p.o . along the rotation axis
z in rotating triaxial potentials ( heisler et al .
1982 ) .
the periodic orbits of the x1v2 family are du for large energies ( fig .
[ dsimple1 ] ) . at first we study the case for @xmath147 .
the periodic orbit has initial conditions @xmath148 . by adding even a tiny perturbation ( e.g. @xmath149 or@xmath150 ) to its initial conditions , along any direction
, we find clouds of points in all 3d projections of the 4d space of section these are 4d clouds of points since we find scattered points in all 3d projections and mixing of colors when we apply the method of color and rotation .
their consequents occupy a volume in the 4d space of section , lying roughly in the region @xmath151
. then we explore the structure of the phase space in the neighborhood of the double unstable periodic orbits of x1v2 in the region du from @xmath61 , for the same fixed value of the perturbation , @xmath152 .
we find again 4d clouds of points as in the case of the double unstable periodic orbit at @xmath147 .
we study the morphology of the asymptotic curves of the unstable manifold in the neighborhood of the double unstable periodic orbits in the region du and we present this morphology in the neighborhood of the double unstable periodic orbit for @xmath147 ( fig .
[ dunst2 ] ) .
the computation of the asymptotic curves is similar to that of the case of the simple unstable periodic orbits ( section 4.3 ) .
the only difference is that we have now two unstable eigenvectors
. these are @xmath153 and @xmath154 .
the corresponding eigenvalues are @xmath155 and @xmath156 .
the initial conditions @xmath157 , that we use for the computation of the asymptotic curves , are given by : @xmath158 we computed the asymptotic curves , in two cases which we named `` i '' and `` ii '' respectively . in the first case , we kept @xmath159 constant and equal to zero and we used @xmath160 as a parameter .
in the second case , we kept @xmath161 and we varied @xmath159 . in case `` i '' we have taken values @xmath162 and @xmath163 with a step @xmath131 ( 2000 initial conditions ) . in the second case `` ii
'' we took values @xmath161 and we varied @xmath159 in the interval @xmath164 with a step @xmath165 ( 13500 initial conditions ) . in both cases `` i '' and `` ii '' we computed only 3 consequents for every initial condition . here
we present the morphology of the asymptotic curves in the 3d @xmath62 projection .
we found in our numerical experiments , that the morphology of the asymptotic curves in all 3d projections is similar .
as it was expected the consequents along the asymptotic curves in case `` i '' ( with the large eigenvalue @xmath166 ) move away faster from the periodic orbit than the consequents along the asymptotic curves in case `` ii '' ( with the small eigenvalue @xmath167 ) .
the consequents of the unstable invariant manifold in case `` i '' depart from the periodic orbit by following the black arrows of fig .
[ dunst2 ] .
they form first a loop in the 3d projections of the 4d space of section below the periodic orbit .
then the consequents approach again the periodic orbit but with a different color ( yellow ) than the one of the initial part which is violet .
afterwards the asymptotic curve performs a larger loop before leaving the box of the frame . in addition
we observe a smooth color distribution from violet ( close to the p.o . ) to blue , green , yellow and red and then follows a color oscillation from red to green and back again .
these oscillations correspond to oscillations of @xmath73 in the upper part of the color spectrum ( right part of fig .
[ dunst2 ] ) .
this smooth color variation along the curve indicates that these asymptotic curves are 4-dimensional curves .
cc + in both cases `` i '' and `` ii '' the asymptotic curves deviate soon from the eigendirections . in fig .
[ dunst2a ] we compare the asymptotic curves `` i '' ( blue color ) and `` ii '' ( red color ) that start along the unstable eigendirections that correspond to the large @xmath168 and small @xmath169 eigenvalues respectively .
the morphologies of these curves are different .
the orbits we start integrating close to these curves remain sticky to them only for a short time ( 0.4 periods of rotation of our system ) before moving away from them .
soon they fill in a chaotic way a large region of the phase space .
we have found similar morphologies for all cases of stable and unstable manifolds we have examined in the du region .
cc + this is a dynamical behavior typical for @xmath35 periodic orbits of the x1 tree .
however , it is not necessarily typical for the dynamics in the neighborhood of @xmath35 periodic orbits in general .
we investigated the orbital behavior close to double unstable p.o .
in several cases in our system .
below we describe the dynamics close to z - axis @xmath35 p.o.s , which differs in many aspects from what we have found in the neighborhood of the @xmath35 x1v2 p.o.s we presented up to now .
as already mentioned , the orbits of the z - axis family lie entirely on the rotational , z axis , of the system .
thus , we consider now the @xmath52 surface as surface of section with @xmath170 .
earlier studies have shown that this family becomes double unstable in slowly rotating triaxal systems ( martinet & de zeeuw 1988 , patsis & zachilas 1990 ) .
we investigated the stability of this family in the potential ( 2 ) rotating very slowly with @xmath171 .
for this value of @xmath50 the z - axis family has large regions of double instability ( du ) . in this case , as we vary the energy , the z - axis family for large negative values of @xmath51 is stable and for @xmath172 it becomes simple unstable . at this transition
bifurcates the stable family of p.o .
of stable anomalous orbits , abbreviated to sao ( heisler et al 1982 ) , and its symmetric family with respect the yz - plane .
then the z - axis family , for @xmath173 , becomes double unstable and it remains du for larger values of @xmath51 . in the @xmath174 4d space of section
the initial conditions of the z - axis orbits are @xmath175 .
for @xmath176 we present the case of the orbit in the neighborhood of the z - axis p.o .
, which deviates from it in the @xmath114 direction by @xmath177 .
we call this orbit `` duz1 '' . in the 2d @xmath109 projection
we observe the first 250 consequents being distributed around the p.o .
( blue points in fig .
[ zdu1]a ) .
these points do not show clearly any particular structure .
the next 3950 consequents however , form an `` @xmath178''-figure .
we plot with red color 950 of them in fig .
[ zdu1]a . in the same figure
the periodic orbits existing in this energy are given with black symbols .
in the center of the `` @xmath178''-figure structure the @xmath179 symbol indicates the @xmath35 z - axis orbit , while in the central regions of the two lobes of this structure the black dots ( on the left and the right ) indicate the location of sao and its symmetric p.o .
, which are surrounded by projections of rotational tori around them .
if we continue to integrate the orbit , we find that after more than 4200 intersections the consequents diffuse and occupy all available phase space .
then the @xmath109 projection is as we see in fig .
[ zdu1]b .
cc + + on one hand we have the formation of a structure discernible in projections of the phase space . on the other hand , the consequents that form the `` @xmath178''-figure structure of fig .
[ zdu1]a are not on a smooth 4d surface .
the application of the color - rotation method shows that we can hardly observe a smooth color variation along any direction .
these are indications of a chaotic behavior .
the finite time @xmath180 index points also to the chaotic character of the orbit .
for most of the time when the consequents in the @xmath109 ( or the @xmath181 ) projection stay on the `` @xmath178''-shaped figure , the index varies around @xmath182 and finally decreases until the time pointed by an arrow in fig .
this time corresponds to 4200 consequents . in fig .
[ zlia]b we see how @xmath102 evolves as the consequents start filling all available phase space .
it increases continuously up to @xmath183 and beyond .
the range of perturbations for which we find the `` @xmath178''-figure is registered in table 2 . outside the range given for
the @xmath86 coordinate we find orbits sticky to rotational tori around the sao and its symmetric orbit . for absolutely larger perturbations in the @xmath114 ,
@xmath184 or @xmath185-direction we encounter three types of orbits , namely ( a ) orbits represented by clouds of points , ( b ) orbits sticky to tori around the stable periodic orbit x1v1 and ( c ) orbits sticky to chains of rotational tori belonging to p.o . of higher multiplicity .
this orbital behavior is related with the unstable manifold of the z - axis orbit as follows .
the computation of the manifolds is similar to that of the case of the double unstable periodic orbits of x1v2 .
the unstable eigenvectors are @xmath186 and @xmath187 .
the corresponding eigenvalues are @xmath188 and @xmath189 .
the initial conditions @xmath190 , which we use for the computation of the asymptotic curves , are given by : @xmath191 we calculate the manifold by using for @xmath160 the values @xmath192 with a step @xmath165 and for every value of @xmath160 we consider for @xmath159 the values @xmath193 again with a step @xmath165 . for every initial condition we compute 10 consequents
. then we have totally @xmath194 consequents .
this way we computed the unstable asymptotic surface we present in fig .
[ zman0 ] .
we note that now we have an asymptotic surface and not an asymptotic curve .
this is expected since we have two eigenvalues outside the unit circle ( see e.g. arnold 1988 , p. 287 ) .
the asymptotic surface is easier to be visualized than in the case of the orbit in the neighborhood of the double unstable x1v2 orbit , because now the two associated eigenvalues are very close .
cc we observe that the unstable asymptotic surface we calculated has a morphology of an `` @xmath178''-figure in the 3d projection @xmath181 of the surface of section . for the 4th dimension we color the points according to the normalized @xmath73 values .
we have a complicated but smooth color variation on the surface of the `` @xmath178''-shaped structure ( fig .
[ zman0]a ) . in fig .
[ zman0]b we give the @xmath109 projection of the same figure that shows clearly the central oscillations of the surface close to the p.o .
located at the center of this `` @xmath178''-figure structure .
a similar morphology is found for the stable asymptotic surface .
the structure we present in fig . [ zman0 ] is part of the unstable manifold . if we consider a larger number of consequents for @xmath195 and a step @xmath196 in ( 5 ) , the asymptotic surface expands considerably also in the @xmath184-direction and its @xmath109 projection is as the red points show in fig . [
we realize that _ all _ consequents of `` duz1 '' ( black color ) stick on the unstable asymptotic surface .
we have to note also that inside the lobes of the `` @xmath178''-shaped manifold structure , besides the sao rotational tori ( green elliptical curve in fig . [ zman1]b ) we find also sticky orbits , initially forming toroidal objects in this region and then diffusing to a larger volume , remaining however sticky to the surface of the manifold .
the consequents of two such orbits are given with blue and magenta colors in the right lobe of the manifold in fig .
[ zman1]b .
the existence of the asymptotic surface decelerates the diffusion of these orbits to larger phase space regions for long times .
the study of the orbits in the neighborhood of simple and double unstable periodic orbits gives us an information of practical interest namely their rate of diffusion and the possible trapping of their consequents in particular regions of the available phase space .
this would allow us to separate the orbits that support physical structures from those which do not .
for this reason we measure the diffusion of the orbits we study and we compare the results among themselves . from the orbits close to x1v2 p.o .
we have consider the following orbital behaviors : * the first type of orbits we found close to x1v2 periodic orbits is represented by double loop ribbons or by closed ribbons in the 3d projections of the 4d space of section .
an example is the orbit `` u '' presented in section 4.1 .
[ dsimple2],[dsimple3 ] ) .
the orbits of this orbital type continue beyond the initial ribbons , by forming additional dense regions in phase space after long integration times .
these dense regions are discernible in some projections as `` @xmath178''-like structures or rings ( fig . [ x1v1v2 ] ) . *
the second type of orbital behavior is represented by clouds of points in the 4d surface of section . as an example
, we consider the orbit with @xmath121 from the initial conditions of the simple unstable x1v2 for @xmath197 and we call it `` u1 '' .
* the third type of orbital behavior is represented again by clouds of points in the 4d surface of section , but this time in the neighborhood of a double unstable periodic orbit . as a typical example for the discussion to follow
, we consider the orbit deviating from x1v2 by @xmath121 at @xmath147 .
we call this orbit `` du1 '' .
in addition we include an orbit in the neighborhood of a z - axis @xmath35 p.o . , i.e. * the orbit that builds an `` @xmath178''-shaped figure before diffusing and occupying all available volume of the phase space ( fig .
[ zdu1 ] ) , which we call `` duz1 '' . in order to estimate the diffusion velocity for the above types of orbital behavior we define a diffusion time and a diffusion velocity as follows : in general the volume of phase space occupied by the consequents of a chaotic orbit increases with time until they occupy the maximum available volume restricted by the surface of the zero velocity .
the time that the orbits need to occupy the maximum volume @xmath198 , we define as `` diffusion time '' @xmath199 .
then the `` diffusion speed '' is the ratio @xmath200 the quantities @xmath198 and @xmath199 for the orbits in the neighborhood of a periodic orbit can be easily computed in two steps : 1 .
as time increases we compute the mean distance from the periodic orbit @xmath201 up to this time @xmath107 of all consequents and then we compute a mean volume @xmath202 . in other words
we compute a `` mean volume '' of the hypersphere centered on the periodic orbit with radius the mean distance @xmath201 , i.e the quantity : @xmath203 .
after time @xmath199 the consequents occupy the maximum volume @xmath198 .
this is represented by the maximum peak in a @xmath202 diagram and this allows us to compute the diffusion speed . in fig .
[ diff1]a we plot the volume occupied by the consequents of the orbit `` u '' ( figs .
[ dsimple2],[dsimple3 ] ) versus time @xmath107 for an interval @xmath204 , corresponding to 4200 intersections .
we observe that for @xmath205 in our system units ( indicated by the left arrow in fig . [ diff1]a corresponding to 2580 intersections ) @xmath202 remains close to 0.165 after some initial oscillations .
this time interval corresponds to the phase that the consequents stick to the asymptotic curves of part a of the unstable manifold forming the double loop ribbon ( section 4.1 ) .
then we observe a second interval between the two arrows in fig .
[ diff1]a , where @xmath202 increases but the slope of the curve is smaller than the slope beyond the right arrow .
the second arrow corresponds to time @xmath206 or 3130 intersections .
the dynamical behavior between the two arrows corresponds to the time during which the consequents form the lobes of the -structure and the `` ring '' in the @xmath111 projection ( fig .
[ x1v1v2 ] ) sticking on the asymptotic curves of part b of the unstable manifold ( section 4.4 ) . beyond the right arrow @xmath202 increases continuously with a characteristically steeper slope than before .
the consequents leave all structures observed in the neighborhood of x1v2 ( section 4.1 ) .
figure [ diff1]b is a continuation of fig . [ diff1]a and shows that the peak of @xmath202 corresponds to @xmath207 for @xmath208 .
the speed of diffusion is @xmath209 .
after the peak of @xmath202 in fig .
[ diff1]b the consequents arrive at distances from the periodic orbit that are smaller or equal to the maximum distance .
this means that @xmath202 decreases a little and finally , it levels off to a constant value .
this happens for all chaotic orbits .
cc + + the evolution of @xmath202 and the quantity @xmath199 change in the case of the orbit `` u1 '' . in fig .
[ diff2 ] we see that for the curve labeled `` 1 '' , corresponding to the orbit `` u1 '' , we have @xmath210 and @xmath211 .
this means that the speed of diffusion is now @xmath212 , i.e. it is 1403 times larger than the velocity of diffusion of orbit `` u '' .
the upper , red curve , labeled `` 2 '' , describes the evolution of @xmath202 for the orbit `` du1 '' in the neighborhood of the @xmath35 p.o .
x1v2 at @xmath147 .
we observe that the curve has a maximum @xmath213 and @xmath214 . in this case
the speed of diffusion is @xmath215 that is 3.65 times and 5133 times larger than the diffusion speeds of the orbits `` u1 '' and `` u '' respectively . for `` u1 '' and `` du1 '' @xmath202 increases rather fast to the peak of the diagram , indicating that we do not have any important stickiness in these cases .
finally for the case we have found stickiness in chaos close to a @xmath35 z - axis p.o .
, i.e. for the orbit `` duz1 '' , the evolution of the quantity @xmath202 is given in fig .
[ zdiff ] .
initially @xmath202 decreases to a value about 1.68 during the first 1600 consequents .
this period ends at the time indicated by arrow `` 1 '' .
this happens because the majority of the points occupy the right lobe of the `` @xmath178 '' structure , as in the @xmath109 2d projection ( fig .
[ zdu1 ] ) .
this implies that the majority of the points are included in a volume smaller than the volume occupied by the consequents forming the whole `` @xmath178''-shaped structure .
thus , the mean volume decreases .
then the next consequents complete the `` @xmath178 '' structure by populating both lobes with similar number of consequents , so @xmath202 increases until @xmath216 , which corresponds to 3200 intersections ( arrow numbered `` 2 '' ) .
then we observe a small plateau until the point indicated by the arrow labeled `` 3 '' ( @xmath217 ) .
the time at this point corresponds to 4200 consequents .
this is the number of consequents that stay on the eight - figure surface before they depart from it and occupy a larger volume of the phase space . for larger times @xmath202 increases . in fig .
[ zdiff]b we have a peak at @xmath218 for @xmath219 .
the diffusion speed is @xmath220 .
cc + + it is evident by comparing figs .
[ diff1 ] , [ diff2 ] and [ zdiff ] , that the diffusion of chaotic orbits from the neighborhood of simple or double unstable p.o . does not depend on the kind of instability of the periodic orbit .
the phenomenon of stickiness that decelerates the diffusion appears both to orbits starting close to simple as well as to double unstable periodic orbits .
when this happens we observe plateaus in the @xmath202 variation .
in this paper we studied the structure of the phase space close to simple unstable and double unstable periodic orbits in a triaxial 3d system representing a rotating disk galaxy .
we have encountered different orbital behaviors and different diffusion rates from the neighborhood of the periodic orbit .
the phenomenon of stickiness seems to be ubiquitous and does not depend on the kind of instability of the periodic orbit in the center of the region we study .
our results refer to orbits in the neighborhood of the families x1v2 ( skokos et al 2002 ) and z - axis ( martinet & de zeeuw 1988 ) and can be summarized as follows : * the orbits starting in the neighborhood of a simple unstable periodic orbit form ( a ) double loop ribbons in the 4d space , ( b ) structures discernible in some projections after long integration times , ( c ) clouds of points around the @xmath34 p.o . with mixing of colors in their 4d representations .
the two first kinds of orbital behavior appear close to the transition @xmath53 of the x1 family at the point of bifurcation of the x1v2 family and indicate the presence of the phenomenon of stickiness in the dynamics of the system .
* the asymptotic curves of the @xmath34 x1v2 p.o .
close to the @xmath53 transition of the central family of periodic orbits x1 are composed of two parts .
part a ( calculated for @xmath128 in ( 3 ) ) winds around deformed rotational tori belonging to orbits deviating from the x1 orbits in the @xmath221 or @xmath222-direction .
part b ( calculated for @xmath134 in ( 3 ) ) winds around the rotational tori of the symmetric x1v1 and x1v1@xmath82 p.o . * in both cases chaotic orbits are guided by the manifolds close to the regions occupied by the rotational tori associated with the stable periodic orbits existing in the region .
these chaotic orbits are sticky and play a very important role , especially in galactic dynamics , since the stickiness time is in many cases of the order of a hubble time or even larger . *
double loops and -shaped structures surrounded by rings that are found in several 2d and 3d projections of the 4d spaces of section in 3d hamiltonian systems are formed by orbits close to the asymptotic curves of simple unstable periodic orbits that surround the rotational tori of two nearby stable p.o . as in fig .
[ x1v1v2 ] .
such configurations appear if the parent family ( in our case the planar x1 family ) has a @xmath0 transition and a nearby @xmath53 transition . at the first transition bifurcate two symmetric stable orbits ( like x1v1 and x1v1@xmath96 ) and at the second transition we have the bifurcation of two simple unstable families ( in our case the families x1v2 and x1v2@xmath96 ) . near the @xmath53 transition of the parent family ,
the already bifurcated stable orbits are not far from the unstable orbit under study , thus the asymptotic curves of the latter families can surround their rotational tori . away from the transition points of the varying parameter ( in our case
the jacobi constant ) the shape of the asymptotic curves become complicated and the consequents of the nearby orbits form clouds in the 4d spaces of section . *
double instability in 3d rotating galactic disks appears for larger energies than simple instability in the families of the x1 tree .
the usual case is after a @xmath223 transition and in general there are no rotational tori associated with stable families in the neighborhood of the @xmath35 p.o .
typically the consequents of the orbits form clouds of points in the 4d spaces of section . * in the case of a double unstable periodic orbit there are two different pairs of eigendirections , starting at the periodic orbit .
the curves of each pair start in opposite directions . between the eigendirections
an `` eigensurface '' is formed .
however the shape of this surface is complicated and it is difficult to be visualized when the associated eigenvalues are not close to each other . * in the case of z - axis @xmath35 periodic orbits in slow pattern speed models , at low energies , we find that the eigenvalues corresponding to the unstable eigenvectors are close to each other . because of that the speeds of deviation along the two eigendirections are similar .
this allows us to visualize the unstable eigensurface corresponding to the @xmath35 periodic orbit .
we find sticky chaotic orbits whose consequents stay close to the unstable manifold for very long times ( of the order of a hubble time or more ) .
we find these sticky orbits either by perturbing the initial conditions of the z - axis @xmath35 p.o .
, or by perturbing orbits on rotational tori associated with the stable anomalous orbit ( sao ) and its symmetric family .
* among the orbits we studied we found those close to the double unstable orbits of the x1v2 family having the largest diffusion speed .
the sticky chaotic orbits close to the bifurcating point of the simple unstable x1v2 orbit and close to the double unstable z - axis orbit we have examined , have comparable diffusion speeds .
they are much slower than the diffusion speeds of the orbits in the neighborhood of x1v2 simple unstable periodic orbits away from its bifurcation point , or of the double unstable orbits of the same family that have eigenvalues of very different size .
these results will be used in a forthcoming paper to assess the role of chaotic orbits in shaping the vertical profiles of disk galaxies .
_ acknowledgments _ we thank dr . c. efstathiou for comments on an earlier draft of this paper . +
arnold , v.i . [ 1988 ] , _ geometrical methods in the theory of ordinary differential equations _ ( 2nd edition , springer - verlag ny ) .
broucke , r. [ 1969 ] `` periodic orbits in the elliptic restricted three - body problem '' , _ nasa tech .
_ 32 - 1360 , 1 - 125 . 3 .
contopoulos , g. [ 2002 ] _ order and chaos in dynamical astronomy _
( springer - verlag , new york berlin heidelberg ) .
contopoulos , g. [ 2009 ] `` ordered and chaotic orbits in spiral galaxies '' in _ chaos in astronomy _ , g. contopoulos and p.a .
patsis ( eds ) ( springer - verlag , berlin , heidelberg ) , p.3 - 22 . 5 . contopoulos , g. & harsoula , m. [ 2008 ] `` stickiness in chaos '' , _ int . j. bif . chaos _ * 18 * , 2929 - 2949 . 6 .
contopoulos , g. & harsoula , m. [ 2010 ] `` stickiness effects in chaos '' , _ celest .
astron . _ * 107 * , 77 - 92 . 7 .
contopoulos , g. & magnenat , p. [ 1985
] `` simple three - dimensional periodic orbits in a galactic - type potential '' _ celest .
* 37 * , 387 - 414 . 8 .
hadjidemetriou , j. [ 1975 ] `` the stability of periodic orbits in the three - body problem '' , _ celest .
* 12 * , 255 - 276 .
heisler , j. , merritt , d. & schwarzschild , m. [ 1982 ] `` retrograde closed orbits in a rotating triaxial potential '' , _ apj _ * 258 * , 490 - 498 . 10 .
katsanikas , m. & patsis , p.a .
[ 2011 ] `` the structure of invariant tori in a 3d galactic potential '' , _ int .
chaos _ * 21 * , 467 - 496 . 11 .
katsanikas , m. , patsis , p.a . &
contopoulos , g. [ 2011 ] `` the structure and evolution of confined tori near a hamiltonian hopf bifurcation '' , _ int . j. bif
* 21 * , 2321 - 2330 .
magnenat , p. [ 1981
] _ tudes numriques sur le computement des orbites stellaires dans des modeles galactiques a 2 ou 3 degrs de libert _ phd thesis , observatoire de geneve .
magnenat , p. [ 1982
] `` numerical study of periodic orbit properties in a dynamical system with three degrees of freedom '' , _ celest .
_ * 28 * , 319 - 343 . 14 .
martinet , l. & de zeeuw , t. [ 1988 ] `` orbital stability in rotating triaxial stellar systems '' , _ astron .
astrophys . _ * 206 * , 269 - 278 . 15 .
miyamoto , m. & nagai , r. [ 1975 ] `` three - dimensional models for the distribution of mass in galaxies '' , _ publ .
. japan _ * 27 * , 533 - 543 . 16 .
patsis , p. a. & zachilas , l. [ 1990 ] `` complex instability of simple periodic orbits in a realistic two - component galactic potential '' , _ astron .
astrophys . _ * 227 * , 37 - 48 .
patsis , p.a .
& zachilas , l. [ 1994 ] `` using color and rotation for visualizing four - dimensional poincar cross - sections : with applications to the orbital behavior of a three - dimensional hamiltonian system '' , _ int . j. bif . chaos _ * 4 * , 1399 - 1424 . 18 .
patsis , p.a . ,
skokos , ch .
& athanassoula , e. [ 2002 ] `` orbital dynamics of three - dimensional bars - iii .
boxy / peanut edge - on profiles '' , _ mon not .
soc . _ * 337 * , 578 - 596 . 19 .
skokos , ch . [ 2010 ] `` the lyapunov characteristic exponents and their computation '' , _ lect . not .
phys . _ * 790 * , 63 - 135 . 20 .
skokos , ch .
[ 2001 ] `` on the stability of periodic orbits of high dimensional autonomous hamiltonian systems '' , _ physica d _ * 159 * , 155 - 179 .
skokos , ch . ,
patsis , p.a . & athanassoula , e. [ 2002 ] `` orbital dynamics of three - dimensional bars - i .
the backbone of three - dimensional bars .
a fiducial case '' , _ mon . not .
soc . _ * 333 * , 847 - 860 . 22 .
vrahatis , m.n . ,
isliker , h. & bountis , t.c .
[ 1997 ] `` structure and breakdown of invariant tori in a 4-d mapping model of accelerator dynamics '' , _ int .
j. bif . chaos . _
* 7 * , 2707 - 2722 . | we study the dynamics in the neighborhood of simple and double unstable periodic orbits in a rotating 3d autonomous hamiltonian system of galactic type . in order to visualize the four dimensional spaces of section we use the method of color and rotation .
we investigate the structure of the invariant manifolds that we found in the neighborhood of simple and double unstable periodic orbits in the 4d spaces of section .
we consider orbits in the neighborhood of the families x1v2 , belonging to the x1 tree , and the z - axis ( the rotational axis of our system ) .
close to the transition points from stability to simple instability , in the neighborhood of the bifurcated simple unstable x1v2 periodic orbits we encounter the phenomenon of stickiness as the asymptotic curves of the unstable manifold surround regions of the phase space occupied by rotational tori existing in the region . for larger energies ,
away from the bifurcating point , the consequents of the chaotic orbits form clouds of points with mixing of color in their 4d representations . in the case of double instability , close to x1v2 orbits , we find clouds of points in the four dimensional spaces of section .
however , in some cases of double unstable periodic orbits belonging to the z - axis family we can visualize the associated unstable eigensurface .
chaotic orbits close to the periodic orbit remain sticky to this surface for long times ( of the order of a hubble time or more ) . among the orbits we studied we found those close to the double unstable orbits of the x1v2 family having the largest diffusion speed .
the sticky chaotic orbits close to the bifurcation point of the simple unstable x1v2 orbit and close to the double unstable z - axis orbit we have examined , have comparable diffusion speeds .
these speeds are much slower than of the orbits in the neighborhood of x1v2 simple unstable periodic orbits away from the bifurcating point , or of the double unstable orbits of the same family having very different eigenvalues along the corresponding unstable eigendirections . |
we thank our colleagues from the pierre auger and kascade - grande collaborations for many fruitful and stimulating discussions and comments on the paper . in particular , we thank carola dobrigkeit , maria giller and jim matthews for reading the manuscript of this article .
we are grateful to d. heck and t. pierog for providing numerical results of air shower simulations and s. knurenko , k. shinozaki , p. sokolsky , m. takeda , and g. thomson for making available data tables of experimental results .
this work has been supported in part by bmbf grant 05a08vk1 .
j. w. cronin , nucl .
suppl . 138 ( 2005 )
465491 d. r. bergman and j. w. belz , j. phys .
34 ( 2007 ) r359r400 k .- h .
kampert , j. phys .
( 2008 ) 062002 a. v. olinto , phys .
( 2000 ) 329348 g. sigl , ann .
( 2003 ) 117141 l. anchordoqui , t. paul , s. reucroft , and j. swain , int . j. mod .
a18 ( 2003 ) 22292366 r. j. protheroe and r. w. clay , publ .
21 ( 2004 ) 122 a. m. hillas , ann .
22 ( 1984 ) 425444 .
d. f. torres and l. a. anchordoqui , rept .
67 ( 2004 ) 16631730 p. bhattacharjee and g. sigl , phys .
327 ( 2000 ) 109247 p.
k. f. grieder , _ cosmic rays at earth : researcher s reference , manual and data book _ , amsterdam ,
netherlands : elsevier 2001 .
b. n. afanasiev _
et al . _ , in proceedings of the tokyo workshop on techniques for the study of extremely high energy cosmic rays , edited by m. nagano ( institute for cosmic ray research , university of tokyo , tokyo , japan ) , p.35 , 2003 .
t. stanev , r. engel , a. mcke , r. j. protheroe , and j. p. rachen , phys .
d62 ( 2000 ) 093005 f. w. stecker and m. h. salamon , astrophys .
j. 512 ( 1999 ) 521526 t. yamamoto , k. mase , m. takeda , n. sakaki , and m. teshima , astropart .
( 2004 ) 405412 e. khan _ et al .
_ , astropart .
( 2005 ) 191201 d. allard _
_ , jcap 0609 ( 2006 ) 005 z. fodor , s. d. katz , and a. ringwald , proc . of physics beyond the standard model
: beyond the desert 02 ( 2002 ) 263267 and hep - ph/0210123 .
e. g. berezhko , astrophys .
j. 684 ( 2008 ) l69l71 p.
l. biermann _
et al . _ , to appear in proc .
of cosmic ray int . seminar ( cris 2008 ) " , malfa ( salina island - italy ) on september 15 - 19 , 2008 ( 2008 ) and arxiv:0811.1848 [ astro - ph ] .
e. waxman , phys .
lett . 75 ( 1995 )
386389 m. vietri , astrophys .
j. 453 ( 1995 ) 883889 k. murase , k. ioka , s. nagataki , and t. nakamura , phys .
rev . d78 ( 2008 )
023005 v. berezinsky , m. kachelrie , and a. vilenkin , phys .
79 ( 1997 ) 43024305 r. aloisio , v. berezinsky , and m. kachelrie , phys .
d74 ( 2006 ) 023516 j. r. ellis , v. e. mayes , and d. v. nanopoulos , phys .
d74 ( 2006 ) 115003 c. t. hill , nucl .
b224 ( 1983 ) 469 .
r. j. protheroe and p. a. johnson , astropart .
phys . 4 ( 1996 )
253 v. berezinsky , p. blasi , and a. vilenkin , phys .
d58 ( 1998 ) 103515 s. sarkar and r. toldra , nucl .
phys . b621 ( 2002 ) 495520 s. r. coleman and s. l. glashow , phys .
d59 ( 1999 ) 116008 j. r. ellis , n. e. mavromatos , and d. v. nanopoulos , phys . rev .
d63 ( 2001 ) 124025 m. jankiewicz , r. v. buniy , t. w. kephart , and t. j. weiler , astropart .
21 ( 2004 ) 651666 s. t. scully and f. w. stecker , astropart .
( 2008 ) 220225 w. bietenholz , arxiv:0806.3713 [ hep - ph ] , and references therein .
m. ahlers , a. ringwald , and h. tu , astropart .
24 ( 2006 ) 438450 i. f. m. albuquerque , g. r. farrar , and e. w. kolb , phys .
d59 ( 1999 ) 015021 m. kachelrie , d. v. semikoz , and m. a. tortola , phys . rev .
d68 ( 2003 ) 043005 m. kachelrie , comptes rendus physique 5 ( 2004 ) 441452 t. stanev , astrophys .
j. 479 ( 1997 ) 290 p. g. tinyakov and i. i. tkachev , astropart .
( 2002 ) 165172 j. alvarez - muniz , r. engel , and t. stanev , astrophys .
j. 572 ( 2001 ) 185201 p. p. kronberg , rep .
prog . phys .
57 ( 1994 ) 325 .
g. sigl , f. miniati , and t. a. ensslin , phys .
d70 ( 2004 ) 043007 k. dolag , d. grasso , v. springel , and i. tkachev , jcap 0501 ( 2005 ) 009 d. v. semikoz and g. sigl , jcap 0404 ( 2004 ) 003 d. seckel and t. stanev , phys .
95 ( 2005 ) 141101 g. gelmini , o. kalashev , and d. v. semikoz , j. exp .
theor . phys .
106 ( 2008 ) 10611082 d. allard , e. parizot , and a. v. olinto , astropart .
( 2007 ) 6175 j. k. becker , phys .
( 2008 ) 173246 t. k. gaisser , _ cosmic rays and particle physics _ , cambridge univ . press , cambridge , 1990
h. p. vankov , n. inoue , and k. shinozaki , phys .
d67 ( 2003 ) 043002 p. homola _ et al .
_ , comput .
commun . 173
( 2005 ) 71 p. homola _ et al .
_ , astropart .
( 2007 ) 174184 m. risse and p. homola , mod .
lett . a22 ( 2007 )
749766 j. matthews , astropart .
( 2005 ) 387397 .
j. alvarez - muniz , r. engel , t. k. gaisser , j. a. ortiz , and t. stanev , phys .
rev . d66 ( 2002 )
033011 c. meurer , j. blmer , r. engel , a. haungs , and m. roth , czech . j. phys .
56 ( 2006 ) a211 j. linsley and a. a. watson , phys .
46 ( 1981 ) 459463 .
drescher , m. bleicher , s. soff , and h. stoecker , astropart .
( 2004 ) 8794 d. heck , j. knapp , j. capdevielle , g. schatz , and t. thouw , wissenschaftliche berichte fzka 6019 , forschungszentrum karlsruhe , 1998 . c. pajares , d. sousa , and r. a. vazquez , phys .
86 ( 2001 ) 16741677 j. dias de deus , m. c. espirito santo , m. pimenta , and c. pajares , phys .
96 ( 2006 ) 162001 j. alvarez - muniz _
et al . _ ,
( 2007 ) 271277 a. m. hillas , j. phys .
g8 ( 1982 ) 14611473 .
j. abraham _ et al . _
( pierre auger collab . ) , phys .
lett . 101 ( 2008 )
061101 m. ave _
et al . _ ,
( 2003 ) 6175 m. ave , j. knapp , m. marchesini , m. roth , and a. a. watson , proc . of 28th int .
cosmic ray conf . , tsukuba ( 2003 ) 349 .
m. nagano , k. kobayakawa , n. sakaki , and k. ando , astropart .
( 2003 ) 293309 j. w. belz _
et al . _
( flash collab . ) , astropart .
25 ( 2006 ) 129139 p.
colin _ et al . _ ( macfly collab . ) , astropart . phys .
27 ( 2007 ) 317325 t. waldenmaier , j. blmer , and h. klages , astropart .
29 ( 2008 ) 205222 m. ave _
et al . _
( airfly collab . ) , astropart .
( 2007 ) 41 r. abbasi _ et al .
_ , astropart .
( 2008 ) 7786 f. arqueros , j. hrandel , and b. keilhauer ( eds . ) , nucl .
instrum . meth . a 597 ( 2008 )
f. arqueros , f. blanco , and j. rosado , nucl .
a597 ( 2008 ) 9498 m. risse and d. heck , astropart .
( 2004 ) 661 h. m. j. barbosa , f. catalani , j. a. chinellato , and c. dobrigkeit , astropart .
22 ( 2004 ) 159166 d. kuempel , k. h. kampert , and m. risse , astropart . phys . 30 ( 2008 ) 167174 r. u. abbasi
_ et al . _
( hires collab . ) , astropart .
( 2007 ) 512520 b. fick _ et al . _
( pierre auger collab . ) , 28th int .
cosmic ray conf .
, tsukuba ( 2003 ) 449 m. unger , b. r. dawson , r. engel , f. schssler , and r. ulrich , nucl .
a588 ( 2008 ) 433441 m. giller and g. wieczorek , astropart .
31 ( 2009 ) 212219 .
r. abbasi _ et al . _
( hires collab . ) , astropart .
25 ( 2006 ) 7483 s. y. benzvi _
et al . _ ,
instrum . meth .
a574 ( 2007 )
171184 b. keilhauer , j. blmer , r. engel , h. o. klages , and m. risse , astropart .
( 2004 ) 249261 d. j. bird _
_ ( fly s eye collab . ) , astrophys .
j. 441 ( 1995 ) 144150 .
h. falcke _ et al . _
( lopes collab . ) , nature 435 ( 2005 ) 313316 d. ardouin _ et al . _
( codalema collab . ) , astropart .
26 ( 2006 ) 341350 a. m. van den berg ( pierre auger collab . ) , proc .
of 30th int . cosmic ray conf .
, merida 5 ( 2007 ) 885 and arxiv:0708.1709 [ astro - ph ] .
w. gorham _
et al . _
( anita collab . ) , phys .
99 ( 2007 ) 171101 t. huege and h. falcke , astron .
astrophys . 412
( 2003 ) 1934 t. huege , r. ulrich , and r. engel , astropart .
( 2007 ) 392405 t. huege , r. ulrich , and r. engel , astropart . phys . 30 ( 2008 )
96104 r. engel , n. n. kalmykov , and a. a. konstantinov , int .
a21s1 ( 2006 ) 6569 .
o. scholten , k. werner , and f. rusydi , astropart .
( 2008 ) 94103 k. werner and o. scholten , astropart .
( 2008 ) 393411 t. gousset , j. lamblin , and s. valcares , astropart .
phys . 31 ( 2009 ) 5262 .
m. amenomori _ et al . _
( tibet as@xmath179 collab . ) , astrophys .
j. 678 ( 2008 ) 11651179 t. antoni _ et al . _
( kascade collab . ) , astropart .
24 ( 2005 ) 125 j. c. arteaga - velazques _
et al . _
( kascade - grande collab . ) , to appear in proc . of 15th int .
symposium on very high - energy cosmic ray interactions , paris , 2008 .
t. abu - zayyad _
et al . _
( hires - mia collab . ) , astrophys .
j. 557 ( 2001 ) 686699 t. abu - zayyad _
et al . _
( hires - mia collab . ) , phys .
84 ( 2000 ) 4276 r. abbasi _ et al . _
( hires collab . ) , phys .
( 2008 ) 101101 k. asakimori _ et al .
_ , proc . of 24th int
cosmic ray conf . ,
rome 2 ( 1995 ) 707 .
r. aloisio _ et al .
_ , astropart .
( 2007 ) 7691 v. berezinsky , j. phys .
( 2008 ) 012001 j. n. bahcall and e. waxman , phys .
b556 ( 2003 ) 16 t. yamamoto ( pierre auger collab . ) , proc . of 30th int .
cosmic ray conf . ,
merida 4 ( 2007 ) 335 r. u. abbasi _
et al . _
( hires collab . ) , phys .
b619 ( 2005 ) 271280 v. berezinsky , a. gazizov , and s. grigorieva , astro - ph/0302483 .
v. berezinsky , a. z. gazizov , and s. i. grigorieva , phys .
d74 ( 2006 ) 043005 d. de marco , p. blasi , and a. v. olinto , astropart .
( 2003 ) 5365 m. teshima , talk given at the _ xxxiii international symposium on multiparticle dynamics ( ismd ) _ , cracow , poland , sep 5 - 11 , 2003 , 2004 .
a. s. chou , phys .
d74 ( 2006 ) 103001 m. nagano , d. heck , k. shinozaki , n. inoue , and j. knapp , astropart .
( 2000 ) 277294 h .- j .
drescher and g. r. farrar , astropart .
( 2005 ) 372381 r. engel ( pierre auger collab . ) , proc . of 30th int .
cosmic ray conf .
, merida 4 ( 2007 ) 385 and arxiv:0706.1921 [ astro - ph ] . c. song _ et al . _
( hires collab . ) , astropart .
( 2000 ) 713 t. abu - zayyad
_ et al . _
( hires collab . ) , astropart .
( 2005 ) 157174 r. u. abbasi
_ et al . _
( hires collab . ) , phys .
92 ( 2004 ) 151101 r. abbasi _ et al . _
( hires collab . ) , astropart .
phys . 27 ( 2007 ) 370381 b. r. dawson _
et al . _
( pierre auger collab . ) , proc .
of 30th int . cosmic ray conf .
, merida 4 ( 2007 ) 425 and arxiv:0706.1105 [ astro - ph ] .
m. t. dova , m. e. mancenido , a. g. mariazzi , t. p. mccauley , and a. a. watson , nucl .
suppl . 122
( 2003 ) 235238 m. t. dova , m. e. mancenido , a. g. mariazzi , t. p. mccauley , and a. a. watson , astropart .
( 2004 ) 597607 k. shinozaki ( agasa collab . ) , nucl .
151 ( 2006 ) 310 .
m. ave , j. a. hinton , r. a. vazquez , a. a. watson , and e. zas , phys . rev .
d65 ( 2002 ) 063007 m. t. dova , m. e. mancenido , a. g. mariazzi , t. p. mccauley , and a. a. watson , astropart .
( 2004 ) 597607 .
f. a. aharonian and a. m. atoyan , astron .
362 ( 2000 ) 937952 j. abraham _ et al . _
( pierre auger collab . ) , astropart .
( 2008 ) 243256 j. abraham _ et al . _
( pierre auger collab . ) , astropart .
( 2007 ) 155168 m. ave _
et al . _ ,
d 65 ( 2002 ) 063007 .
m. risse _ et al .
_ , astropart .
( 2004 ) 479490 m. ave , j. a. hinton , r. a. vazquez , a. a. watson , and e. zas , phys .
85 ( 2000 ) 22442247 r. aloisio , v. berezinsky , and m. kachelrie , phys .
d69 ( 2004 ) 094023 j. abraham _ et al . _
( pierre auger collab . ) , phys .
( 2008 ) 211101 o. kalashev _ et al .
d 66 ( 2002 ) 063004 .
x. bertou , p. billoir , o. deligny , c. lachaud , and a. letessier - selvon , astropart .
( 2002 ) 183193 j. l. feng , p. fisher , f. wilczek , and t. m. yu , phys .
88 ( 2002 ) 161102 e. zas , new j. phys . 7 ( 2005 )
130 c. aramo _ et al . _ ,
astropart . phys . 23
( 2005 ) 6577 d. fargion , prog .
57 ( 2006 ) 384393 g. miele , s. pastor , and o. pisanti , phys .
b634 ( 2006 ) 137142 d. gora , m. roth , and a. tamburro , astropart .
( 2007 ) 402413 .
n. hayashida _
et al . _
( agasa collab . ) , astropart .
( 1999 ) 303311 j. a. bellido , r. w. clay , b. r. dawson , and m. johnston - hollitt , astropart .
( 2001 ) 167175 m. aglietta _ et al . _
( pierre auger collab . ) , astropart .
27 ( 2007 ) 244253 r. abbasi _ et al . _
( hires collab . ) , astropart .
( 2004 ) 111 j. d. swain , proc . of 10th marcel grossmann meeting on recent developments in theoretical and experimental general relativity , gravitation and relativistic field theories (
mg x mmiii ) , rio de janeiro ( 2004 ) 721740 and astro - ph/0401632 . r. u. abbasi _ et al . _
( hires collab . ) , astropart .
( 2004 ) 139149 r. u. abbasi _
et al . _
( hires collab . ) , astrophys .
j. 610 ( 2004 ) l73 r. u. abbasi _
et al . _
( hires collab . ) , astrophys .
j. 623 ( 2005 ) 164170 s. mollerach ( pierre auger collab . ) , proc . of 30th int .
cosmic ray conf .
, merida 4 ( 2007 ) 279 and arxiv:0706.1749 [ astro - ph ] .
m. kachelrie and d. v. semikoz , astropart .
26 ( 2006 ) 1015 b. revenu _ et al . _
( pierre auger collab . ) , proc . of 29th int .
cosmic ray conf . , pune ( 2005 ) 75 p. g. tinyakov and i. i. tkachev , jetp lett .
74 ( 2001 ) 15 p. g. tinyakov and i. i. tkachev , jetp lett .
74 ( 2001 ) 445448 d. s. gorbunov , p. g. tinyakov , i. i. tkachev , and s. v. troitsky , astrophys . j. 577 ( 2002 ) l93 n. w. evans , f. ferrer , and s. sarkar , phys .
d67 ( 2003 ) 103005 p. tinyakov and i. tkachev , phys .
d69 ( 2004 ) 128301 n. w. evans , f. ferrer , and s. sarkar , phys .
d69 ( 2004 ) 128302 b. e. stern and j. poutanen , astrophys .
j. 623 ( 2005 ) l33l36 r. u. abbasi _
et al . _
( hires collab . ) , astrophys .
j. 636 ( 2006 ) 680684 m. veron - cetty and p. veron , astron .
455 ( 2006 ) 773 .
d. s. gorbunov , p. g. tinyakov , i. i. tkachev , and s. v. troitsky , jetp lett .
80 ( 2004 ) 145148 d. harari ( pierre auger collab . ) , proc . of 30th int .
cosmic ray conf . , merida 4 ( 2007 ) 283 and arxiv:0706.1715 [ astro - ph ] .
v. s. berezinsky , s. i. grigorieva , and b. i. hnatyk , astropart .
( 2004 ) 617625 v. berezinsky , a. z. gazizov , and s. i. grigorieva , phys .
b612 ( 2005 ) 147153 t. wibig and a. w. wolfendale , j. phys .
g31 ( 2005 ) 255264 d. hooper , s. sarkar , and a. m. taylor , astropart .
( 2007 ) 199212 h. takami , k. murase , s. nagataki , and k. sato , arxiv:0704.0979 [ astro - ph ] . a. a. watson , proc . of conf .
on thinking , observing and mining the universe , sorrento , italy ( 2003 ) d. allard , n. g. busca , g. decerprit , a. v. olinto , and e. parizot , jcap 0810 ( 2008 ) 033 d. hooper , a. taylor , and s. sarkar , astropart . phys . 23 ( 2005 ) 1117 l. a. anchordoqui , h. goldberg , d. hooper , s. sarkar , and a. m. taylor , phys .
d76 ( 2007 ) 123008 n. g. lehtinen , p. w. gorham , a. r. jacobson , and r. a. roussel - dupre , phys .
d69 ( 2004 ) 013008 a. w. strong , i. v. moskalenko , and o. reimer , astrophys .
j. 613 ( 2004 ) 956961 s. w. barwick _
et al . _
( anita collab . ) , phys .
96 ( 2006 ) 171101 t. stanev , p. l. biermann , j. lloyd - evans , j. p. rachen , and a. watson , phys .
lett . 75 ( 1995 )
30563059 i. zaw , g. r. farrar , and j. e. greene , arxiv:0806.3470 [ astro - ph ] .
m. kachelrie , nucl .
suppl . 165
( 2007 ) 272279 m. kachelrie and p. d. serpico , phys .
b640 ( 2006 ) 225229 m. kachelrie , p. d. serpico , and m. teshima , astropart .
26 ( 2006 ) 378386 t. antoni _ et al . _
( kascade collab . ) , j. phys .
g : nucl . part .
( 2001 ) 1785 . | this review summarizes recent developments in the understanding of high - energy cosmic rays .
it focuses on galactic and presumably extragalactic particles in the energy range from the knee ( @xmath0 ev ) up to the highest energies observed ( @xmath1 ev ) .
emphasis is put on observational results , their interpretation , and the global picture of cosmic rays that has emerged during the last decade .
cosmic rays are ionized atomic nuclei reaching the earth from outside the solar system .
although already discovered in 1912 , their sources and propagation mechanisms are still subject of intense research . during the last decade
significant progress has been made and a consistent picture of cosmic - ray observations begins to evolve .
this review describes recent progress in the exploration of the sources and propagation of high - energy cosmic rays , focusing on observational results and the emerging global picture .
many reviews of cosmic - ray theory and observations are available in the literature .
most of these reviews concentrate on different aspects and energy ranges of high - energy cosmic rays which , in turn , are presented in much more detail than possible here . for example , covering the knee energy region , experimental data are compiled in @xcite and a comparison of observations and model predictions can be found in @xcite .
the data in the range between the knee and the ankle , and their interpretation are discussed in @xcite .
focusing on the upper end of the cosmic - ray spectrum , measurement techniques and observations are reviewed in @xcite .
more theoretical and phenomenological aspects of the physics of ultra high - energy cosmic rays are subject of the reviews @xcite and different source scenarios are discussed in depth in @xcite ( acceleration scenarios ) and @xcite ( non - acceleration scenarios ) .
an exhaustive compilation of experimental results of the full cosmic - ray energy range can be found in @xcite and a recent review , emphasizing measurement and analysis techniques , is given in @xcite . in this article
, we discuss high - energy cosmic - ray measurements covering the energy range from the knee to the highest energies . by concentrating mainly on observational results of the last decade and their implications for our overall understanding of cosmic rays ,
this review is complementary to the other articles . the exploration of cosmic rays is mainly driven by new experimental findings .
hence , we begin this review with a short historical overview , followed by an introduction to the physics of high - energy cosmic rays ( ) . in the energy region of interest cosmic rays are measured indirectly with large detector installations below the atmosphere , registering secondary particles produced in extensive air showers , initiated by high - energy cosmic rays .
in , basic properties of air showers are introduced and major detection techniques are discussed .
recent experimental results concerning the flux of cosmic rays , their elemental composition , and studies of anisotropies in their arrival directions are presented in sections [ energysec ] to [ anisosec ] . the global picture evolving from these measurements and their impact on the present understanding of the origin of high - energy cosmic rays
is emphasized in .
the importance of the understanding of high - energy hadronic interactions for the interpretation of air shower data is underlined in .
concluding remarks ( ) complete the review .
cosmic rays were discovered in the year 1912 by v.f .
hess during several ascends with hydrogen - filled balloons up to altitudes of 5 km @xcite .
he measured the ionization rate of air as function of altitude .
electrometers served as standard devices to measure ionizing radiation at this time @xcite .
hess found an increase of ionizing radiation with increasing height and he concluded that radiation penetrates from outer space into the atmosphere .
for the discovery of the cosmic radiation , v.f .
hess has been awarded the nobel price in 1936 . during the subsequent years w. kolhrster made further ascends with improved electrometers , measuring the altitude variation of the ionization up to heights of 9 km @xcite . in 1929
w. bothe and w. kolhrster measured coincident signals in two geiger - mller counters @xcite . placing absorber material in between the two counters they also measured the absorption characteristics of the radiation .
they concluded that the `` hhenstrahlung '' ( or cosmic radiation ) is of corpuscular nature , i.e. consists of charged particles .
similar conclusions were drawn from measurements by j. clay , who showed that the intensity of cosmic rays depends on the ( magnetic ) latitude of the observer @xcite .
this was a clear indication that a large fraction of cosmic radiation consists of charged particles .
kolhrster continued his work with geiger - mller tubes operated in coincidence . in february 1938
he reported the discovery of coincident signals between two tubes with distances as far as 75 m @xcite .
he concluded that the tubes were hit by secondary particles or showers generated by cosmic rays in the atmosphere . in the late 1930s p. auger undertook investigations of cosmic radiation at the jungfraujoch , switzerland at 3500 m.a.s.l .
he used wilson chambers and geiger - mller tubes separated by large distances and operated in coincidence @xcite .
similar to kolhrster , auger concluded that the registered particles are secondaries generated in the atmosphere , originating from a single primary cosmic ray . in the 1940s the origin of the primary radiation
could be revealed with measurements on balloons at high altitudes .
m. schein showed that the positively charged primary particles were mostly protons @xcite .
cloud chambers and photographic plates were carried into the stratosphere and it was found that cosmic rays are made up of fully ionized atomic nuclei moving at speeds closely to that of light @xcite .
many nuclei of the periodic table up to @xmath2 were found and their relative abundances determined .
hydrogen and helium occur most frequently , and the distribution in mass of the heavier nuclei appeared to be similar to that in the solar system .
elements more massive than iron or nickel were found to be very rare . since the mid 1940s large detector arrays were installed to measure extensive air showers . for most investigations detectors with a large surface and a short time resolution were required .
early detectors comprised geiger - mller counters , progress in the development of photomultipliers lead to the application of scintillation counters and the newly available detectors .
it was found that the energy spectrum of cosmic rays follows a power law @xmath3 over a wide range in energy . in 1958 g.v .
kulikov and g.b .
khristiansen measured the integral electron number spectrum in air showers using an array of hodoscope counters @xcite .
they recognized a kink in the spectrum around @xmath4 particles , corresponding to primary energies of several pev ( @xmath0ev ) .
this structure is now known as the `` knee '' in the energy spectrum . since that time there is an ongoing debate about the origin of this structure . in the 1960s
the air shower array of the m.i.t .
group at volcano ranch , new mexico was the largest cosmic - ray detector .
the set - up comprised 20 stations equipped with scintillation counters , set up on a triangular grid , covering a total area of 12 km@xmath5 . in 1962 the first event with an energy of about @xmath6 ev has been recorded with this array @xcite .
bigger air shower arrays were built ( sugar @xcite , haverah park @xcite , yakutsk @xcite , and agasa @xcite ) and , after some initial attempts , the first successful fluorescence light detector , called fly s eye , was set up in utah @xcite . with these detectors , another feature of the cosmic - ray flux
first discussed in @xcite was firmly established in the early 1990s and is now known as the `` ankle '' @xcite .
finally it should be mentioned that , in the early years , particle physics was done mainly by studying cosmic rays . in the 1930s
investigations of the cosmic radiation lead to the discovery of new elementary particles such as the positron @xcite or the muon @xcite .
the pion was discovered exposing nuclear emulsions to cosmic radiation at mountain altitudes in 1947 @xcite .
new unstable hadrons were found in cosmic - ray interactions in balloon - borne emulsion chambers @xcite in 1971 which were later , after the discovery of charm particles , identified as d mesons @xcite .
also a number of exotic phenomena were observed @xcite , none of which could be confirmed in accelerator experiments . over time
a standard description of cosmic rays evolved , which is briefly sketched in the following section .
[ abundance ] the solar system is permanently exposed to a flux of highly energetic ionized atomic nuclei the cosmic rays .
their energies extend from the mev range to at least @xmath6 ev . the differential energy spectrum of all cosmic - ray particles is depicted in .
it falls steeply as function of energy , decreasing by about a factor 500 per decade in energy .
the flux decreases from more than 1000 particles per second and square meter at gev energies to about one particle per m@xmath5 and year at a pev , and further to less than one particle per km@xmath5 and century above 100 eev .
the strong decrease in flux poses a big experimental challenge and our knowledge about the particles and their origin is more and more limited with increasing energy ( and decreasing flux ) . at sub - gev energies individual isotopes
are measured with small detectors in outer space and individual elements can be resolved with balloon - borne detectors in the tev regime . at energies exceeding 100 tev
large detection areas are required to collect a suitable number of particles in a reasonable time . at present ,
such detectors are realized at ground level only and secondary particles generated in the atmosphere ( the extensive air showers ) are registered . at pev energies , groups of elements could be resolved , while at the highest energies even a classification into `` light '' and `` heavy '' particles becomes already an experimental challenge .
the energy spectrum follows a power law @xmath7 over a wide energy range , indicating non - thermal acceleration processes .
the spectrum is rather featureless as can be inferred from .
however , small structures become clearly visible when the ordinate is multiplied with some power of the particle energy , as shall be discussed below ( see e.g. ) .
the spectral index is @xmath8 at energies up to several pev .
then a steepening is observed , the so - called _ knee _ , with @xmath9 at higher energies . a further steepening , the _ second knee _ , occurs around @xmath10ev .
finally , at about @xmath11ev , at the _ ankle _ , the spectrum flattens again .
the abundance of elements in cosmic rays is shown in as function of the nuclear charge number .
all elements of the periodic table have been found in cosmic rays .
for the relatively abundant elements up to nickel , energy spectra for individual elements have been measured @xcite .
abundances as obtained by several experiments at about 1 gev / n are depicted .
the cosmic - ray composition is compared to the abundance of elements in the solar system .
overall , both distributions look very alike .
however , there exist certain differences , which reveal information on the acceleration and propagation of cosmic rays .
the light elements lithium , beryllium , and boron as well as the elements below iron ( @xmath12 ) and below lead ( @xmath13 ) are more abundant in cosmic rays than in the solar system .
they are assumed to be produced in spallation processes of the more abundant particles of the cno , iron , and lead groups during the journey of cosmic rays through the galaxy .
hence , they are frequently referred to as secondary cosmic rays . as the spallation cross section of the relevant nuclei is known at gev energies , the ratio of secondary to primary cosmic rays is used to infer the propagation path length of cosmic rays in the galaxy .
an example is the boron - to - carbon ratio which has been measured as function of energy @xcite
. the ratio decreases as function of energy which is frequently explained in leaky box models by a rigidity - dependent = p / z$ ] . ]
decrease of the path length of cosmic rays in the galaxy @xmath14 .
typical values are @xmath15 g / cm@xmath5 , @xmath17 , and @xmath18 gv as reference rigidity .
cosmic - ray particles are assumed to propagate in a diffusive process through the galaxy , being deflected many times by the randomly oriented magnetic fields ( @xmath19 @xmath20 g ) .
the nuclei are not confined to the galactic disc , they propagate in the galactic halo as well .
the scale height of the halo has been estimated with measurements of the @xmath21be/@xmath22be - ratio by the isomax detector @xcite to be a few kpc .
the abundance of radioactive nuclei in cosmic rays measured with the cris instrument yields a residence time in the galaxy of about @xmath23 years for particles with gev energies @xcite .
the energy density of cosmic rays amounts to about @xmath24 ev/@xmath25 , a value comparable to the energy density of the visible star light @xmath26
ev/@xmath25 , the galactic magnetic fields @xmath27 ev/@xmath25 , or the microwave background @xmath28 ev/@xmath25 .
the power required to sustain a constant cosmic - ray intensity can be estimated as @xmath29 erg / s , where @xmath30 is the residence time of cosmic rays in a volume @xmath31 ( the galaxy and the galactic halo ) . with a rate of about three supernovae per century in a typical galaxy
, the energy required could be provided by a small fraction ( @xmath32 ) of the kinetic energy released in supernovae .
this had been realized already in 1934 by baade and zwicky @xcite .
the actual mechanism of acceleration remained mysterious until fermi @xcite proposed a process that involved interaction of particles with large - scale magnetic fields in the galaxy . eventually , this lead to the currently accepted model of cosmic - ray acceleration by the first - order fermi mechanism that operates in strong shock fronts which are powered by supernova explosions and propagate from a supernova remnant ( snr ) into the interstellar medium @xcite .
diffusive , first - order shock acceleration works by virtue of the fact that particles gain an amount of energy @xmath33 at each cycle , where a cycle consists of a particle passing from the upstream ( unshocked ) region to the downstream region and back . at each cycle
, there is a probability that the particle is lost downstream and does not return to the shock .
higher energy particles are those that have remained longer in the vicinity of the shock and so have had time to achieve higher energy . after a time @xmath34 the maximum energy achieved is @xmath35 , where @xmath36 refers to the velocity of the shock .
this results in an upper limit , assuming a minimal diffusion length equal to the larmor radius of a particle of charge @xmath37 in the magnetic fields @xmath38 behind and ahead of the shock . using typical values of type ii supernovae exploding in an average interstellar medium yields @xmath39ev
@xcite .
more recent estimates give a maximum energy up to one order of magnitude larger for some types of supernovae @xcite .
it has also been suggested that the cosmic rays itself interact with the magnetic fields in the acceleration region , yielding to an amplification of the fields , which in turn results in much higher energies that can be reached during the acceleration process @xcite . with this
mechanism cosmic rays are supposedly accelerated up to @xmath40 ev .
information on the composition at the source can be obtained from measurements of the abundance of refractory nuclei .
they appear to have undergone minimal elemental fractionation relative to one another . comparing
the derived abundance at the source to the abundance in the solar system reveals that the two samples exhibit a striking similarity over a wide range @xcite .
when uncertainties are taken into account , the abundances of particular isotopes are consistent with being within 20% of the solar values .
this indicates that cosmic rays are accelerated out of a sample of well mixed interstellar matter .
hence , cosmic rays are `` regular '' matter , but accelerated to extremely high energies .
the understanding of the origin of the knee in the energy spectrum is commonly thought to be a cornerstone for the understanding of the origin of ( galactic ) cosmic rays .
many approaches are discussed in the literature @xcite .
a popular explanation is that the knee is associated with the upper limit of acceleration by galactic supernovae , while the ankle is associated with the onset of an extragalactic population that is less intense but has a harder spectrum that dominates at sufficiently high energy .
another popular explanation is leakage of particles from the galaxy . at energies in the gev regime measurements
indicate a decreasing path length of cosmic rays in the galaxy .
extrapolating this to higher energies indicates that above a certain energy cosmic rays are not contained in the galaxy ( or the galactic halo ) anymore . in a simple picture ,
this can be understood since the larmor radius of a proton in the galactic magnetic field @xmath41 becomes with increasing energy comparable to and finally exceeds the thickness of the galactic disk .
if the knee is caused by the maximum energy attained during the acceleration process or due to leakage from the galaxy the energy spectra for individual elements with charge @xmath42 would exhibit a cut - off ( or a knee ) at an energy @xmath43 , with the cut - off energy @xmath44 for protons .
the sum of the flux of all elements with their individual cut - offs makes up the all - particle spectrum . in this picture
the knee in the all - particle spectrum is related to the cut - off for protons and the steeper spectrum above the knee is a consequence of the subsequent cut - offs for all elements , resulting in a relatively smooth spectrum above the knee . since the abundance of ultra - heavy nuclei ( see ) at gev energies is very low as compared to iron the end of the galactic component is often assumed to be at energies around @xmath45 . however , recently it has been proposed that ultra - heavy elements may play an important role at high energies @xcite which yields a value of @xmath46 for the end of the galactic component , coinciding with the second knee at @xmath47ev .
another interesting question is that of a natural end of the spectrum at high energy .
already 40 years ago it has been realized that interactions of cosmic rays with photons of the cosmic microwave background ( cmb ) would result in a cut - off of the spectrum above @xmath48 ev @xcite .
all hadronic particles suffer significant energy losses during propagation above this energy .
protons interact with background photons forming mainly a @xmath49 resonance @xcite and nuclei are broken up due to photodisintegration @xcite .
a compilation of energy loss lengths of protons and various nuclei is shown in fig .
[ fig : gzkplots ] ( ) @xcite .
the energy loss length of photons depends on the flux of the universal radio background ( urb ) which is not well known ( see , for example , discussion in @xcite ) .
depending on the assumptions on the urb , the energy loss length is significantly smaller than or comparable to that of hadronic particles .
s the short energy loss lengths indicate that cosmic rays with energies above @xmath6ev should come from sources within a @xmath50mpc sphere .
astrophysical sources within our galaxy are disfavored .
even though rapidly spinning young neutron stars could be thought of accelerating particles to the highest energies observed @xcite , it would be difficult to explain the apparent isotropic arrival direction of uhecrs to energies beyond @xmath51 ev . as cosmic rays of energy greater than @xmath52ev
are no longer confined by galactic magnetic fields , it is natural to assume that they are produced by extragalactic sources .
considering diffusive shock acceleration , which is thought to accelerate cosmic rays in snrs , the magnetic field strength @xmath38 in the source and the size @xmath53 of the source region are related to the maximum acceleration energy by @xcite @xmath54 where @xmath55 is the shock velocity in units of @xmath56 and @xmath42 is the particle charge .
this relation is shown in fig .
[ fig : gzkplots ] ( ) for various astrophysical objects .
the list of the very few viable candidate sources includes active galactic nuclei ( agn ) @xcite , radio lobes of fr ii galaxies @xcite , and gamma - ray bursts ( grbs ) @xcite ( for a review of astrophysical sources , see @xcite ) .
many alternative , non - acceleration scenarios called top - down models have been proposed . in these models ,
uhecrs are produced in decays of super - heavy objects such as super - heavy dark matter @xcite , cryptons @xcite , or topological defects @xcite .
all of these models postulate new particle physics and predict typically high gamma - ray fluxes at ultra - high energy @xcite .
finally there are propagation model scenarios in which the gzk energy loss processes are evaded or shifted to higher energies .
examples are violation of lorentz invariance @xcite , the @xmath42-burst model @xcite or postulation of new particles with properties similar to protons @xcite
. reviews of the different scenarios can be found in @xcite .
measurements of the arrival direction distribution , primary mass composition and flux will be the key ingredients to solving the puzzle of uhecr sources .
given an expected angular deflection of only a few degrees for particles above @xmath57ev in our galaxy @xcite and the existence of large cosmic voids with negligible magnetic fields @xcite , high statistics measurements should finally allow cosmic - ray astronomy and reveal correlations with sources or source regions .
similarly , knowing the composition of uhecrs will restrict the classes of source models .
for example , a mixed composition would exclude top - down models .
another very important source of complementary information is given by secondary particle fluxes , i.e. gamma - rays and neutrinos , produced in uhecrs sources and during propagation ( see , for example , @xcite ) . in the following
we will introduce some general properties of extensive air showers that are employed in cosmic - ray measurements .
detailed presentations of this subject can be found in @xcite .
showers initiated by photons or electrons ( called em . showers henceforth ) are governed mainly by the particle production processes ( i ) bremsstrahlung of electrons and ( ii ) pair production of electrons by photons .
in addition to radiative energy losses , electrons are subject to ionization energy loss .
the total energy loss @xmath58 of electrons can be written as @xmath59 where @xmath60 is the ionization energy loss given by the bethe - bloch formula which depends logarithmically on energy . the _ radiation length _ ,
@xmath61 , depends on the material the shower evolves in and is @xmath62g / cm@xmath5 in air .
particle multiplication and ionization energy loss are competing processes in showers .
the energy at which an electron loses the same energy due to ionization and particle production is called _ critical energy _ , @xmath63mev . some properties of em .
showers can already be understood within the very simple model of heitler @xcite . in this model
it is assumed that the incoming particle interacts in the atmosphere after traveling a depth distance @xmath64 and produces two new particles with half the energy of the initial particle .
these two new particles again interact at a distance @xmath64 from their production point .
after @xmath65 generations of successive interactions the number of particles is @xmath66 . the number of particles as function of depth @xmath67 can be written as @xmath68 .
the production of new particles continues until the energy of the secondary particles is smaller than the critical energy @xmath69 , below which absorption processes dominate over further particle multiplication .
the maximum number of particles in the shower is @xmath70 and the depth of shower maximum is given by @xmath71 , with @xmath72 being the primary particle energy .
heitler s model is certainly an over - simplified picture of an air shower but illustrates two important features .
the number of particles at shower maximum is approximately proportional to the primary energy and the depth of shower maximum increases logarithmically with energy .
approximate formulae for the longitudinal shower size profile and the lateral particle distribution of em .
showers have been derived from cascade theory @xcite .
considering only shower particles of energy @xmath73 , the depth of shower maximum is given by @xmath74 .
accounting for the energy distribution of particles in a shower this expression becomes ( here written for electrons in a photon - induced shower ) @xmath75 .\ ] ] the increase of the depth of shower maximum per decade in energy is called _ elongation rate _ , @xmath76 .
the elongation rate of em.showers is energy - independent , @xmath77g / cm@xmath5 . for energies larger than @xmath69 ,
the energy spectrum of secondary particles in a shower follows approximately a power law @xmath78 with where @xmath79 denotes the shower age parameter .
the shower age is often phenomenologically defined as @xmath80 .
based on the solutions of the one - dimensional cascade equations given in @xcite , greisen developed a compact and still often used parametrization of the mean longitudinal shower size profile ( number of charged particles ) @xcite @xmath81 a recent derivation of this expression is given in @xcite . in data analysis ,
a function proposed by gaisser and hillas @xcite is often used to fit measured shower profiles @xmath82 the particle density in dependence on the distance to the shower core ( lateral distance ) is mainly determined by multiple coulomb scattering of electrons . for electrons of low energy @xmath73 , the increase of variance of the effective scattering angle per traversed depth @xmath83 is approximately given by @xmath84 , with @xmath85mev @xcite .
results of detailed calculations of the lateral particle distribution by nishimura and kamata @xcite were parametrized by greisen @xcite as @xmath86 with @xmath87 $ ] being a normalization constant .
eq .
( [ eq : nkg ] ) is called nishimura - kamata - greisen ( nkg ) function .
the lateral distribution of the shower particles scales with the molire unit @xmath88g / cm@xmath5 and hence depends on the local air density .
the effect of varying atmospheric density along the shower track can approximately be taken into account by calculating the molire unit not at observation hight but 2 - 3 radiation lengths higher up in the atmosphere @xcite
. at very high energy , two additional processes become important and change the characteristics of em . showers .
first of all , the amplitudes of subsequent interactions of photons or electrons , which are independent at low energy , have to be added coherently @xcite as the interaction length becomes comparable to the separation of subsequent interactions .
the resulting effect is known as landau - pomeranchuk - migdal ( lpm ) effect and leads to the suppression of new particle production in certain kinematic regions @xcite . in air ,
the lpm effect typically becomes important at energies above @xmath52ev : shower - to - shower fluctuations of em .
showers increase drastically and the depth of maximum is shifted deeper into the atmosphere .
another important ultra high - energy process is that of magnetic pair production and bremsstrahlung @xcite .
gamma - rays of energies exceeding @xmath57ev can interact with the geomagnetic field of the earth .
such interactions typically take place a thousand kilometers above the atmosphere . mainly due to magnetic bremsstrahlung
, a shower of more than 100 secondary photons and a few electrons is formed , which interact in the atmosphere simultaneously .
simulations of this effect can be found in @xcite . as the primary energy is shared by many secondary particles ,
the lpm effect hardly influences such showers . due to the superposition of many lower - energy em .
showers , shower - to - shower fluctuations of converted primary photons are significantly reduced .
the dependence of the geomagnetic pre - shower effect on the local arrival direction can be used to search for photons in a model - independent way , see @xcite .
most of the differences between photon and hadron - initiated showers are related to the fact that hadronic showers develop a significant muon component whereas there are very few muons in purely em .
showers .
furthermore , hadronic multiparticle production is characterized by large event - to - event fluctuations . again a model similar to that from heitler is useful to understand the basic properties of hadronic showers ( see , for example , @xcite ) .
for simplicity we assume that a hadronic interaction of a particle with energy @xmath73 produces @xmath89 new particles with energy @xmath90 , two thirds of which being charged particles @xmath91 ( charged pions ) and one third being neutral particles @xmath92 ( neutral pions ) .
neutral particles decay immediately into em . particles ( @xmath93 ) .
charged particles re - interact with air nuclei if their energy is greater than some typical decay energy @xmath94 or decay otherwise .
the number of generations of hadronic interactions , @xmath65 , follows from @xmath95 .
supposing that one muon is produced in the decay of each charged particle , one gets @xmath96 with @xmath97 .
the numerical values for @xmath98 and @xmath94 depend on the muon energy threshold and are given in @xcite for different hadronic interaction models .
the number of muons produced in an air shower increases almost linearly with primary energy and depends on the air density ( through @xmath94 ) and the charged and total particle multiplicities of hadronic interactions . also the energy transferred to the em
. shower component can be estimated within this simple model . in each hadronic interaction , a certain fraction of the initial energy
is transferred to the em .
shower component .
after @xmath65 generations the energy in the hadronic and em .
components is given by @xmath99 simulations show that the number of generations of charged pions is typically about 5 to 6 @xcite and increases slightly with primary shower energy . correspondingly the fraction of energy transferred to the em .
component increases from about @xmath100% at @xmath0ev to @xmath101% at @xmath6ev .
the depth of shower maximum of a hadron - induced shower is given by that of the em .
shower component , @xmath102 .
the first hadronic interaction produces em .
particles of energy @xmath103
. therefore one can write in lowest order approximation @xmath104 where @xmath105 is the hadronic interaction length . from eq .
( [ eqxmax ] ) follows the _ elongation rate theorem _ , stating that hadronic showers have always an elongation rate equal to or smaller than that of em .
showers @xcite @xmath106 the coefficients are @xmath107 and @xmath108 ( see also discussion in @xcite for the next higher order term ) .
if the primary particle is a nucleus , one can use the _ superposition model _ to deduce the main shower characteristics @xcite . in this model , a nucleus with mass @xmath109 and energy @xmath72 is considered as @xmath109 independent nucleons with energy @xmath110 .
the superposition of the individual nucleon showers yields @xmath111 to a good approximation , while the number of charged particles at shower maximum is independent of the primary hadron , both , the number of muons and the depth of maximum do depend on the mass of the primary particle .
the heavier the shower - initiating particle the more muons are expected for a given primary energy .
in addition , the superposition of @xmath109 independent showers naturally explains why the shower - to - shower fluctuations are smaller for shower initiated by nuclei as compared to proton showers .
detailed shower simulations confirm the basic shower properties discussed so far .
quantitative predictions depend on the details of modeling particle production and transport @xcite .
whereas em.interactions are rather well understood within perturbative qed , hadronic multiparticle production can not be calculated within qcd from first principles .
phenomenological models have to be used to describe the final states of hadronic interactions .
the parameters of these models are determined by comparing the model predictions with accelerator measurements . reflecting the different methods for describing data , low- and high - energy interaction models are distinguished .
the former ones are typically based on the picture of intermediate resonance formation and decay as well as parametrizations of data .
the latter ones are involving the production of color strings and their fragmentation . for high energy interactions ( @xmath112 gev ) ,
the hadronic interaction models often used in air shower simulations are dpmjet ii.55 and iii @xcite , epos @xcite , qgsjet 01 @xcite and qgsjet ii @xcite , as well as sibyll 2.1 @xcite .
these models reproduce accelerator data reasonably well but are characterized by different extrapolations above a center - of - mass energy @xmath113 tev ( @xmath114ev ) , leading to very different shower predictions at high energy @xcite .
the situation is different at low energy where more measurements from fixed target experiments are available .
there , one of the main problems is the extrapolation of measurements to the very forward phase space region close to the beam direction and the lack of measurements of pion - induced interactions with light nuclei @xcite . at low energy ,
models based on data parameterizations and/or microscopic models such as fluka @xcite , gheisha @xcite , urqmd @xcite , or target @xcite are used .
differences of the predictions of these models are important for the number of muons @xcite , in particular at large lateral distances . in figs .
[ fig : longi]-[fig : nenmumodels ] , we illustrate some shower properties discussed only qualitatively so far by showing the results of detailed monte carlo simulations done with corsika @xcite and conex @xcite .
the longitudinal profile of a typical proton shower of @xmath51ev is shown in fig . [ fig : longi ] ( ) . in the region of the shower maximum , less than 1% of the charged shower particles are muons .
the electromagnetic component is absorbed much faster than the muonic one .
as muons are mainly produced in hadronic interactions , their number and lateral distribution can be used as composition - sensitive observables .
the predicted muon distributions depend on the assumptions on hadron production in air showers .
this is demonstrated in fig .
[ fig : nenmumodels ] where the expected number of muons and electrons is shown for showers initiated by proton , iron , and gamma - ray primaries , as calculated with different interaction models @xcite .
the energy dependence of the total muon number and ratio between the muon numbers of proton and iron showers is in agreement with the expectations from the superposition model . at high energy ,
the discrimination power of electron - muon number measurements is subject to large systematic uncertainties due to our limited understanding of hadronic multiparticle production . in fig .
[ fig : longi ] ( ) , we show the mean depth of shower maximum , @xmath115 , for different primary particles . the difference between proton and iron showers is in agreement with the predictions of the superposition model .
whereas the expected mean depth of maximum depends again sensitively on the chosen hadronic interaction model , the fluctuations are rather model - independent at a given energy . the elongation rate of em .
showers obtained with detailed model simulations coincides with that predicted in em.cascade theory @xcite .
the increase of the em . elongation rate at high energy is caused by the lpm effect .
also , at ultra high - energy , photon interaction with the geomagnetic field can lead to a negative elongation rate .
the elongation rates found for hadronic showers within different models ( @xmath116g / cm@xmath5 at @xmath51ev ) can be qualitatively explained with the elongation rate theorem @xcite .
an increase of the observed @xmath117 faster than or comparable to the elongation rate of em .
showers is a model - independent signature for a transition to a lighter composition .
however , over a limited range in energy , drastic changes in the characteristics of hadronic multiparticle production can also lead to an elongation rate comparable to that of em .
showers @xcite .
finally , it should be mentioned that the em .
shower component exhibits a number of universality features that are independent of the primary hadron type and also rather insensitive to the primary energy .
for example , near the shower core , the electron energy distribution is a universal function of shower age and the angular distribution of electrons depends almost exclusively on the electron energy and only slightly on shower age @xcite .
furthermore , the longitudinal shower profile ( particle flux ) at a given lateral distance can be well parametrized by an universal function of the depth of shower maximum and shower size at maximum @xcite .
this reflects the universality of the lateral distribution of electromagnetic particles if the lateral distance is measured in moliere units @xcite .
the classical set - up to measure air showers is an array of scintillation detectors , registering charged particles from the shower . in each detector
the density of charged particles ( mostly electrons , positrons , but also some converted photons ) is measured . from this information the lateral distribution of the electromagnetic component is inferred .
this yields information on the position of the shower core and the total number of particles in the shower .
measurements of the arrival times of the particles with a resolution of a few ns allow to reconstruct the orientation of the shower plane and thus , perpendicular to it , the arrival direction of the primary cosmic ray .
due to the large number of secondary particles it is usually sufficient to cover only a small fraction of the total area with detectors .
typical values range from 1.2% for the kascade array to @xmath118 for the auger array .
examples for arrays in the knee region are the eas - top experiment @xcite , where 37 stations of scintillation detectors were distributed over an area of @xmath119 m@xmath5 , located above the gran sasso underground laboratory at an altitude of 2005 m , or the kascade experiment @xcite , where 252 detector stations are placed on a rectangular 13 m grid , covering an area of @xmath120 m@xmath5 .
detectors that operate at higher energy are the scintillator array of the kascade - grande experiment @xcite , or the agasa experiment , which covered an area of about 100 km@xmath5 with 111 scintillation counter stations @xcite .
another technique utilized to measure charged particles are water detectors , like the surface array of the pierre auger observatory with its 1600 detectors distributed over an area of 3000 km@xmath5 @xcite .
they are relatively deep ( typically @xmath121 m ) compared to scintillators with typical thicknesses of the order of several cm .
consequently they have a larger response to showers with large zenith angles , e.g. the fwhm of the declination distributions were @xmath122and @xmath123for the volcano ranch ( scintillators ) @xcite and haverah park ( water ) @xcite experiments respectively .
this results in a much larger sky coverage of a water experiment compared to a scintillator array .
they are also suitable for the measurement of horizontal showers that can be used to detect neutrinos at the highest energies .
various techniques are applied for the detection of muons in air showers .
frequently , particle counters are covered with absorbers of lead , iron , or soil with a thickness of a few tens to suppress the electromagnetic component , as in the kascade @xcite or agasa @xcite experiments .
muons can also be identified via their trajectories in tracking devices , like the hegra crt detectors @xcite or the muon tracking detector of kascade @xcite . in underground laboratories ,
well shielded by rock , soil , water , or ice absorbers with a thickness corresponding to several 1000 mw.e .
( meter water equivalent ) , muons with thresholds in the tev regime are registered @xcite .
the energy of hadrons is usually determined with calorimeters .
the principle idea is to absorb an incoming particle and to measure the dissipated energy .
examples are the hadron calorimeters of the eas - top @xcite or kascade experiments @xcite .
the direction of the shower axis and hence of the primary cosmic ray is obtained by measurements of the arrival time of the shower front in the detector plane .
the shower direction is the normal to the reconstructed , slightly curved shower front .
typical detectors have a time resolution in the range from 0.5 ns up to a few ns .
the angular resolution ( 68% value ) of the kascade detector field decreases from 0.55for small showers with @xmath124 to 0.1for showers with @xmath125 @xcite .
the angular resolution of the auger surface detectors decreases from values around 2.2for showers with energies below 4 eev to values from 0.5to 1 , depending on the zenith angle , for higher energies ( @xmath126 eev ) @xcite . to obtain the position of the shower core and the number of particles ,
the particle densities are measured and an appropriate lateral distribution function is fitted to the data .
the number of particles is calculated by integration of the measured lateral distribution .
historically , the choices of parameterizations of the electron and muon lateral distributions were influenced by a review by greisen @xcite .
the nishimura - kamata - greisen ( nkg ) function became the standard function to describe the particle density @xmath127 for a shower with the number of electrons @xmath128 and shower age @xmath79 as function of distance @xmath129 to the shower , see eq .
@xcite .
traditionally , a fixed molire unit @xmath130 m is used and the parameter @xmath79 is limited to @xmath131 . to parameterize the lateral distribution of muons , greisen suggested the function @xcite @xmath132 to describe the density of muons @xmath133 as function of distance @xmath129 to the shower axis for a shower with a total muon number @xmath134 .
the greisen radius is @xmath135 m. the kascade group found that the lateral distributions of all three major shower components ( electromagnetic , muonic , and hadronic ) can be parameterized using the nkg function @xcite .
fitting simultaneously the parameters @xmath128 , @xmath136 , and @xmath79 , the measured lateral distributions of the electromagnetic component can be reproduced with an accuracy of about 1% , yielding @xmath137 m. the experimental muon distributions are described with an accuracy of 5% using @xmath138 m. finally , for the lateral distribution of hadrons with energies above 50 gev a value @xmath139 m has been found .
the nkg function has been analytically developed to describe pure electromagnetic showers . for hadron - induced air showers
it exhibits shortcomings describing the measured electron lateral distributions , most obvious at large core distances .
investigations of the kascade group showed that the measured electron lateral distributions for showers with energies up to @xmath40ev and core distances up to 200 m can be described better using a modified nkg function @xcite @xmath140 optimizing the parameters with monte carlo data , the values @xmath141 and @xmath142 have been obtained , when @xmath143 m is used for the scale parameter . for showers in the eev energy range
, the agasa group uses the parameterization @xmath144 to describe the lateral distribution of charged particles up to distances of several km from the shower axis @xcite , inspired by a function suggested by linsley @xcite .
the parameters are @xmath145 , @xmath146 and a molire unit @xmath147 m for near vertical showers with @xmath148 .
another parameterization for the electron lateral distribution has been suggested by the haverah park collaboration @xcite .
it is also applied in the auger observatory @xcite .
the signal @xmath149 in a water detector is parameterized as function of distance @xmath129 to the shower core as @xmath150 for @xmath151 m and @xmath152 at larger distances .
the shape parameter @xmath153 varies with zenith angle , while the parameter @xmath154 and the scale radius @xmath155 m are fixed .
the position of the shower core is determined by the kascade group with an uncertainty of 5 m for showers with an electron number @xmath124 improving to less than 1 m for large showers ( @xmath156 ) @xcite .
the error in the reconstructed number of electrons decreases in the same electron number interval from 18% to less than 4% .
similar values were obtained in the eas - top experiment @xcite . in the electron number range from @xmath157 to @xmath158 ,
the error in the position of the shower core improves from 7.5 to 2 m and the uncertainty in the number of electrons decreases from @xmath159 to about 10% .
one of the most important parameters to characterize a shower is the energy of the primary particle .
various methods are discussed in the literature to obtain this value . for kascade , it has been found that , at sea level , the number of muons with energy above 230 mev in the range from 40 m to 200 m from the shower axis is a good measure for the primary energy independent of the mass of the primary particle @xcite . with its relatively high detector density
the lateral distribution of muons is measured very well in the radius range from 40 to 200 m. extrapolating beyond these limits would introduce uncertainties related to the ( less well known ) shape of the lateral distribution . hence , the number of muons is reconstructed using a distance range only , in which detectors are present .
another method is to use the correlation between the number of electrons and muons reconstructed .
for example the casa - mia group uses the relation @xmath160 to estimate the primary energy @xcite .
similarly , the kascade - grande experiment obtained the relation @xmath161 to estimate the primary energy as function of the observed number of electrons ( @xmath162 mev ) and muons ( @xmath163 mev ) at sea level for showers with zenith angle @xmath164 @xcite .
a similar method has been applied by the agasa group @xcite . here , the particle density as measured 600 m from the shower core is used to estimate the primary energy . at this distance the fluctuations in the lateral distribution are found to be relatively small @xcite .
the measured value is corrected for attenuation due to the zenith angle and converted to a value @xmath165 for vertically incident showers .
the conversion to primary energy via the relation @xmath166=2.23\times10^{17}\cdot s_0(600)^{1.02}[\mbox{m}^2]$ ] yields at @xmath6 ev a difference between proton and iron induced showers of about 10% and a similar difference is obtained using two different codes to describe hadronic interactions in the atmosphere , namely qgsjet and sibyll @xcite .
the methods described so far depend on simulations of the shower development in the atmosphere . to avoid this uncertainty
another method is being applied in the auger observatory @xcite .
it makes use of the constant intensity method , which relies on the fact that primary cosmic rays arrive isotropically .
the value @xmath167 of the measured signal in a water detector at a distance of 1000 m from the shower axis is used .
the dependence of this parameter on the depth in the atmosphere ( which varies as @xmath168 of the shower zenith angle @xmath164 ) is obtained from the measured data . using this dependence , the actually measured signal is converted to the value @xmath169 , representing the signal for a shower with a zenith angle of 38 .
finally , the primary energy is estimated as @xmath170=0.149\cdot s_{38}^{1.08}$ ] using the fluorescence telescopes as optical calorimeters , which define the energy scale in a direct way ( see ) @xcite . to estimate the mass of the shower - inducing primary particle
the following array observables are used : the electron - to - muon number ratio , the arrival time distribution of the particles , the curvature of the shower front , and the slope of the lateral distribution .
the method applied most frequently is the measurement of the electron - to - muon ratio at ground level . plotting the number of electrons @xmath128 and muons @xmath134 in a plane
as shown in , we find an energy axis ( in the direction of the main diagonal ) and ( almost perpendicular to it ) a mass axis .
using a heitler cascade model the ratio of electrons to muons at shower maximum can be estimated , yielding the relation @xcite @xmath171 this illustrates that @xmath172 is a function of @xmath173 .
it can be estimated that an uncertainty in @xmath174 of about 16% results in an uncertainty of about one unit in @xmath173 .
also the shape of the shower front or the arrival time distribution of the particles at ground level is utilized as an indirect estimate of the depth of the shower maximum . for heavy nuclei ,
muons are produced earlier in the shower development and reach the ground also earlier as compared to the electromagnetic component whose particles follow the shower core and branch off to large lateral distances only very late before reaching the detector array .
it is understood that the narrower pulse profiles correspond to higher production altitudes , which in turn lead to similar arrival times in the detectors ( see e.g. @xcite ) .
many particles in the shower disc travel with relativistic velocities through the atmosphere .
approximately one third of the charged particles emit light in the forward direction @xcite , the angle in air at sea level amounts to 1.3only .
electrons ( and positrons ) are the most abundant charged species in air showers . due to their relatively low threshold ( 21 mev at sea level ) ,
they contribute mostly to the light in air showers . at present , for the detection of light , two techniques are applied : integrating detectors , in principle consisting of arrays of photomultipliers inside light collecting cones , looking upwards in the sky , and imaging detectors or telescopes , composed of large area collection mirrors and a camera with segmented read - out .
optical detectors such as detectors and fluorescence detectors ( described in the next section ) can only be operated during clear moon - less nights to obtain reliable data .
this restricts their duty cycle to about 10% .
the basic idea of integrating detectors is to measure the lateral distribution of the light with an array of photomultipliers distributed over a large area on ground level . to enlarge the collection area ,
the pmts are installed inside light collecting cones ( winston cones ) .
such observations yield the lateral density distribution of photons .
it can be parameterized by the empirical relation @xmath175 ) ; & 30~\mbox{m}<r\le120~\mbox{m } \\
c_{120}\cdot ( r/120~\mbox{m})^{-\beta } ; & 120~\mbox{m}<r\le350~\mbox{m},\\ \end{array}\right.\ ] ] a combination of an exponential and a power law function @xcite .
@xmath176 is the light intensity at 120 m distance from the shower core , @xmath79 the exponential inner slope , and @xmath177 the outer slope .
the energy of the primary particle is strongly correlated with the photon density at 120 m , @xmath176 grows approximately as @xmath178 .
the average depth of the shower maximum is approximately linearly related to the exponential slope @xmath79 .
hence , from the measurements both , energy and mass of the primary particle can be derived .
the latter through the dependence of the average depth of the shower maximum on the primary particle mass .
this technique was pioneered by the airobicc experiment on la palma island @xcite .
another example is the blanca instrument , which was located at the dugway proving ground in utah , usa and operated in coincidence with the casa experiment at the same site @xcite .
blanca consisted of 144 angle - integrating detectors which recorded the lateral distribution of air shower light .
a detector array is also installed in the tunka valley close to lake baikal in siberia @xcite , consisting of 25 detectors that cover an area of 0.1 km@xmath5 .
it is planned to extend this installation with 133 optical detectors covering an area of about 1 km@xmath5 @xcite .
cosmic - ray events within the field of view of an imaging atmospheric telescope produce a focal plane image which corresponds to the direction and intensity of light coming from the air shower . when the direction of the air shower core and the distance of the shower axis from the telescopes are known , simple geometry can be used to reconstruct the light received from each altitude of the shower .
the amount of light produced is strongly correlated with the number of electrons in the shower and is used to estimate the shower size as a function of depth in the atmosphere from which the location of the shower maximum can be determined .
this procedure is essentially geometrical and has the advantage of being almost independent of numerical simulations expect for the calculation of the angular distribution of light around the shower axis .
large telescopes are used to reconstruct air showers initiated by primary gamma rays in tev @xmath179-ray astronomy .
the telescopes and analysis procedures are designed to effectively suppress the much more abundant ( up to a factor of 1000 ) hadron induced showers .
the technique was established by the pioneering work of the whipple telescope @xcite . among the presently largest installations
are the h.e.s.s .
@xcite , magic @xcite , and veritas @xcite telescopes .
an example for a telescope , optimized for the reconstruction of hadron induced showers was the dual imaging experiment ( dice ) @xcite .
it was located inside the casa - mia array in dugway , utah ( usa ) and comprised two telescopes , each equipped with a 2 m diameter spherical mirror viewed by an array of 256 pmts . at very high energy ( @xmath180ev ) , the fluorescence light technique can be used to measure directly the longitudinal profile of air showers .
this technique is based on the detection of fluorescence light emitted by nitrogen molecules that are excited by charged particles traversing the atmosphere .
there are two transitions of electronic states of the nitrogen molecule , called 2p and 1n for historical reasons , that lead in combination with the change of the vibrational and rotational states of the molecule to several fluorescence emission bands .
a recent spectrometer measurement @xcite of these bands is shown in fig .
[ fig : fyspectrum ] .
the bands are labeled with the electronic transition type ( 2p or 1n ) and the change of the vibration quantum number @xmath181 , for example , 2p(@xmath182 ) .
most of the fluorescence light emission is found in the wave length range between 300 and 400 nm ( near uv ) .
the lifetime of the excited states of nitrogen is of the order of 10ns and the fluorescence light is emitted isotropically .
the fluorescence yield , the number of photons produced per deposited energy , depends on the gas mixture in the atmosphere and atmospheric conditions .
collisions between molecules give rise to competing de - excitation processes ( collisional quenching , see , for example , discussion in @xcite ) .
the importance of quenching increases with pressure and almost cancels the density dependence of the energy deposit per unit length of particle trajectory .
this results in a weakly height dependent rate of about 4 - 5 fluorescence photons produced per meter and charged shower particle at relevant altitudes . during the last years
several experiments have been set up for measuring the fluorescence yield under different atmospheric conditions @xcite .
recent progress is reviewed in @xcite .
still , the fluorescence yield is currently known only to a precision of about 15% @xcite as end - to - end calibrated data are not yet available .
recent measurements confirm the expectation that the fluorescence yield is independent of the energy of the exciting particle . only at very low energy ,
deviations are expected @xcite .
therefore , if the atmospheric dependence of the fluorescence yield is taken into account , the fluorescence technique allows a calorimetric measurement of the energy deposited in the atmosphere .
simulations show that about 90% of the total shower energy is converted to ionization energy and , hence , is accessible for detection @xcite . the average ratio between the energy deposited in the atmosphere and the primary particle energy is shown in fig . [ fig : fyspectrum ] ( ) .
it depends on the primary particle type and energy and , to some extent , also on the model used in the simulations .
however , as most of the shower energy is transferred to em .
particles , the model dependence corresponds to an uncertainty of only a few percent of the total energy . in case of a gamma - ray as a primary , about 99% of the energy is deposited in the atmosphere .
a complete reconstruction of a shower profile with the fluorescence technique requires the determination of the geometry of the shower axis , the determination of the light fraction , and the correction for the wavelength dependent atmospheric absorption of light . in shower observations with one fluorescence telescope ( monocular observation ) ,
a shower - detector plane is given by the image of the light track .
the orientation of the shower within this plane follows from the time sequence of the pmt signals @xcite .
the angular uncertainty of the orientation of the shower - detector plane depends on the resolution of the fluorescence camera and the length of the measured track .
typically a resolution of the order of @xmath183 is obtained . in general , the reconstruction resolution of the angle within the shower - detector plane is much worse and varies between @xmath184 and @xmath185 ( for example , see @xcite ) . the reconstruction accuracy can be improved considerably by measuring showers simultaneously with two telescopes ( stereo observation ) .
showers observed in stereo mode can be reconstructed with an angular resolution of about @xmath186 @xcite .
a similar reconstruction quality is achieved in hybrid experiments that use surface detectors to determine the arrival time of the shower front at ground @xcite . in fluorescence measurements ,
the light signal of air showers can be considered as a highly asymmetric background contribution but can also be exploited as an independent signal @xcite . knowing the longitudinal shower profile , the light contribution to the detected signal can be estimated using parametrized electron energy distributions and models for the angular distribution of the emitted light @xcite .
in general , fluorescence detectors require continuous monitoring of atmospheric conditions , in particular the measurement of the wavelength dependent mie scattering length and detection of clouds @xcite .
the temporal variations of the density profile of the atmosphere can lead to additional systematic uncertainties of shower reconstruction , in particular of the depth of shower maximum @xcite .
the first fully functional air shower fluorescence detector was the fly s eye experiment in utah that began taking data in 1982 and was operated for 10 years @xcite .
the fly s eye detector was a setup of two stations , fly s eye i ( 67 spherical mirrors of 1.5 m diameter , viewed by 880 pmts in total ) and fly s eye ii ( 8 mirrors viewed by 120 pmts ) .
the fly s eye i detector had a total field of view ( fov ) of @xmath187 in azimuth and @xmath188 in zenith . with the smaller fov of about @xmath188 in azimuth and an elevation range from @xmath189 to @xmath190 degrees ,
the fly s eye ii station was designed to measure showers in coincidence with fly s eye i. in october 1991 the shower of the highest energy measured so far , @xmath191ev , was detected with fly s eye i @xcite .
the successor to the fly s eye experiment , the high resolution fly s eye ( hires ) , took data from 1997 ( hires i ) and 1999 ( hires ii ) to 2006 .
the largest data set of hires is that of monocular observations with hires i , a telescope consisting of a ring of mirrors with a fov from @xmath192 to @xmath193 in elevation and full azimuth .
hires ii is built up of two rings of mirrors covering elevation angles up to @xmath194 .
the optical resolution of the hires detectors is with @xmath195 per pmt much higher than that of fly s eye .
the pierre auger observatory combines the observation of fluorescence light using imaging telescopes with the measurement of particles reaching ground level in a `` hybrid approach '' @xcite .
the southern auger observatory ( near malargue , argentina ) is the world s largest air shower detector and comprises 1600 polyethylene tanks set up in an area covering 3000 km@xmath5 .
each water detector has 3.6 m diameter and is 1.55 m high , containing 12 m@xmath196 of high - purity water .
a radio system is used to provide communication between each station and a central data acquisition system .
four telescope systems overlook the surface array .
a single telescope system comprises six telescopes , overlooking separate volumes of air .
each telescope has a camera with 440 pmt pixels , whose field of view is approximately 1.5 .
one camera overlooks a total field of view of 30azimuth @xmath197 28.6elevation . in the northern hemisphere ,
the telescope array ( ta ) is located in millard county , utah , usa @xcite .
it covers an area of 860 km@xmath5 and comprises 576 scintillator stations and three fluorescence detector sites on a triangle with about 35 km separation , each equipped with twelve fluorescence telescopes .
an independent measurement technique to observe air showers is provided by means of detection of radio - frequency electromagnetic waves emitted from showers .
coherent radio emission generated by extensive air showers was theoretically predicted by askaryan in 1961 @xcite and experimentally discovered by jelly in 1965 at a frequency of 44 mhz @xcite . over a period of time
this phenomenon has been considered as an interesting alternative to traditional methods of detection of high - energy cosmic rays with energy greater than @xmath40ev . in the 1960s and 1970s the experimental and theoretical efforts in this direction had only limited success @xcite .
modern experiments , such as codalema @xcite and lopes @xcite , aim at studies of radio emission from air showers using modern , improved instruments .
first break - throughs have been achieved @xcite . at present , also activities are under way to install prototype antenna systems at the site of the pierre auger observatory in argentina to investigate the possibility for radio detection of air showers at the highest energies @xcite . for a review of recent developments ,
see @xcite .
in addition to experimental difficulties there remain questions concerning the quantitative radio emission theory .
several mechanisms of radio emission generation in air have been identified after the pioneering work of askaryan where the coherent radiation of the charge - excess was put forward @xcite .
this radiation is very strong for showers developing in dense media @xcite . in the case of air showers
there is also an alternative radiation due to the acceleration of charged shower particles in the earth s magnetic field .
it is called geosynchrotron mechanism and has been recently investigated in detail @xcite .
the interrelation between these two essential mechanisms is not clear at present .
hence , also combined efforts are in progress , performing accurate radio emission calculations within the framework of a unified approach @xcite .
the all - particle energy spectrum extending from @xmath198 ev up to the highest energies is shown in .
the flux as obtained from direct measurements above the atmosphere ( represented in the figure through results from atic , proton , and runjob ) extends smoothly to high energies in the air shower detection regime .
the all - particle spectrum can be approximated by a broken power law @xmath199 with a spectral index @xmath200 below @xmath201 ev . at this energy ,
the _ knee _ , the spectral index changes to @xmath9 . in the following
we consider in more detail two energy regions : galactic cosmic rays up to energies of about @xmath40 to @xmath52 ev and the extragalactic component at higher energies .
many groups published results on the all particle energy spectrum from indirect measurements in the knee region ( @xmath202 ev ) .
the spectra obtained are compiled in .
the ordinate has been multiplied by @xmath203 .
the individual measurements agree within a factor of two in the flux values and a similar shape can be recognized for all experiments with a at energies of about 4 pev .
also shown are results for the all - particle flux as obtained by direct observations above the atmosphere approaching energies up to 1 pev . in the region of overlap , the results from direct and indirect measurements are in reasonable agreement .
typical values for the systematic uncertainties of the absolute energy scale of air shower experiments are about 15 to 25% . renormalizing the energy scales of the individual experiments to match the all - particle spectrum obtained by direct measurements in the energy region up to almost a pev requires correction factors of the order of 10% @xcite . a remarkable result , indicating that behind an absorber of 11 hadronic interaction lengths or 30 radiation lengths the energy of the primary particle is determined with an absolute error of the order of 10% .
one should keep in mind that the experiments investigate different air shower components , are situated at different atmospheric depths , and use different interaction models to interpret the observed data .
nevertheless , the systematic differences are relatively small and the all - particle spectrum seems to be well known . : cosmic - ray energy spectra for four groups of elements , from top to bottom : protons , helium , cno group , and iron group
. results from direct measurements above the atmosphere by ams @xcite , atic @xcite , bess @xcite , caprice @xcite , heat @xcite , @xcite , imax @xcite , jacee @xcite , mass @xcite , @xcite , runjob @xcite , rich - ii @xcite , @xcite , @xcite , sokol @xcite , @xcite , and fluxes obtained from indirect measurements by kascade electrons and muons for two hadronic interaction models @xcite and single hadrons @xcite , eas - top ( electrons and muons ) @xcite and single hadrons @xcite , grapes-3 interpreted with two hadronic interaction models @xcite , hegra @xcite , mt .
chacaltaya @xcite , mts .
fuji and kanbala @xcite , tibet burst detector ( hd ) @xcite and as@xmath179 ( hd ) @xcite .
results from direct measurements above the atmosphere by atic @xcite , bess @xcite , caprice @xcite , heat @xcite , @xcite , imax @xcite , jacee @xcite , mass @xcite , @xcite , rich - ii @xcite , runjob @xcite , @xcite , sokol @xcite , @xcite , and fluxes obtained from indirect measurements by kascade electrons and muons for two hadronic interaction models @xcite , grapes-3 interpreted with two hadronic interaction models @xcite , mts .
fuji and kanbala @xcite , and tibet burst detector ( hd ) @xcite .
results from direct measurements above the atmosphere by atic ( c+o ) @xcite , crn ( c+o ) @xcite , tracer ( o ) @xcite , jacee ( cno ) @xcite , runjob ( cno ) @xcite , sokol ( cno ) @xcite , and fluxes obtained from indirect measurements by kascade electrons and muons @xcite , grapes-3 @xcite , the latter two give results for two hadronic interaction models , and eas - top @xcite .
results from direct measurements above the atmosphere by atic @xcite , crn @xcite , heao-3 @xcite , @xcite , @xcite , tracer @xcite ( single element resolution ) and @xcite , @xcite , jacee @xcite , runjob @xcite , sokol @xcite ( iron group ) , as well as fluxes from indirect measurements ( iron group ) by eas - top @xcite , kascade electrons and muons @xcite , grapes-3 @xcite , and h.e.s.s .
direct light @xcite .
the latter three experiments give results according to interpretations of the measured air - shower data with two hadronic interaction models , namely qgsjet and sibyll .
the gray solid lines indicate spectra according to the @xcite . up to about a @xmath0ev direct measurements have been performed with instruments above the atmosphere . as examples ,
results for primary protons , helium , and iron nuclei are compiled in .
recently , also indirect measurements of elemental groups became possible , as discussed below and the results of the kascade and eas - top experiments are shown in the figures as well . also results from other air shower experiments
are shown .
hegra used an imaging telescope system to derive the primary proton flux @xcite .
spectra for protons and helium nuclei are obtained from emulsion chambers exposed at mts .
fuji and kanbala @xcite .
the tibet group performs measurements with a burst detector as well as with emulsion chambers and an air shower array @xcite .
grapes-3 uses the correlation between the registered number of electrons and muons to derive energy spectra for mass groups @xcite .
the h.e.s.s .
telescope system derived for the first time an energy spectrum measuring direct light @xcite .
this light is emitted by the primary nuclei in the atmosphere before its first interaction , i.e.before the air shower begins @xcite .
results for iron nuclei are shown .
over the wide energy range depicted , the flux as obtained by direct measurements is smoothly continued to higher energies with the results of air shower measurements . despite of the experimental uncertainties and systematic differences between different experiments and different interpretations of air shower data using various air shower models , a clear picture of the spectra for elemental groups
is evolving .
it is evident that the in the all - particle spectrum is caused by a depression of the flux of light elements .
the measurements follow power laws with a cut - off at high energies .
the spectra according to the are indicated in the figures as lines .
it can be recognized that the measured values are compatible with cut - offs at energies proportional to the nuclear charge @xmath204 pev @xcite .
there are several detector installations that were designed to measure cosmic rays at the highest energies . at the high - energy end , the total aperture and observation time determine the statistics of expected events . at low energy ,
the acceptance range is given by the employed detection technique and typically the distance between the detectors .
a compilation of the integrated aperture ( i.e. total exposure ) reported by experiments with data above @xmath51ev is shown in fig .
[ fig : exposure ] . at ultra high - energy , about two times
more events are expected in the auger data set than hires has collected in monocular or stereo mode .
the yakutsk experiment is expected to have slightly more than a fourth of the statistics of hires i and about one half of that of agasa .
the measurements of the cosmic - ray fluxes obtained with these exposures are shown in fig .
[ fig : flux3-all ] ( error bars indicate statistical uncertainties only ) . in case
several analyses of the same data set are available , only the most recent results are included in the plot .
the shaded area , depicting the results of the analysis of the haverah park data @xcite , accounts for some systematic uncertainties by assuming extreme elemental compositions , either fully iron or proton dominated .
the highest energy point ( fly s eye monocular observation ) corresponds to the highest energy event @xcite . for sake of clarity
upper limits are not shown .
it is common to present the data multiplied by @xmath205 to enhance deviations from a @xmath206 power law . in this representation ,
characteristic features such as the second knee at about @xmath207ev and the ankle at @xmath208ev are better visible .
however , it should be emphasized that scaling the flux with energy to some power ( e.g. @xmath209 or @xmath210 ) is misleading and does not reflect correctly the uncertainties of the measurements . in this presentation , the statistical uncertainties can not be separated from the systematic energy calibration uncertainties .
the importance of the systematic uncertainty of the experimental energy scale is demonstrated in fig .
[ fig : uhecr - flux3-selected ] in which fluxes with the nominal energy scale of the experiments are compared with that after a model - motivated energy shift has been applied @xcite . after shifting the energy scales of the experiments , a very good overall agreement of the different measurements
is obtained . in particular ,
the ankle in the cosmic - ray spectrum is clearly seen .
the good agreement is a non - trivial observation as the position of the ankle and the overall flux change in a correlated way in the @xmath205-representation .
the data sets of the hires and auger measurements provide evidence for a flux suppression at ultra high - energy .
the statistical significance of the flux suppression is difficult to specify unambiguously .
if one compares to a power - law flux a significance of more than 5 @xmath211 is found in each of the data sets @xcite .
both spectra can be well described by models with uniformly distributed sources and a gzk suppression @xcite .
a similarly good agreement between gzk model predictions and the yakutsk data was shown by several authors ( for example , @xcite ) .
only the agasa data seem to disfavor a flux suppression . assuming uniformly distributed sources of uhe protons and treating the normalization of the expected energy spectrum as free parameter agasa expects to observe 1.8 events above @xmath6ev with 11 actually detected .
this corresponds to a @xmath212 deviation from the gzk cutoff spectrum @xcite .
other assumptions on the shape of the gzk proton spectrum lead to the prediction of 2.4 expected events @xcite , corresponding to a deviation of @xmath213 from the gzk cutoff hypothesis . due to uncertainties in the absolute energy scale and partially also low statistics ,
the differences between the spectra of the different experiments are of limited statistical significance only @xcite . it should be kept in mind that a sudden and drastic change of composition from hadrons to photons could give observational results similar to a suppression of the flux @xcite observed with fluorescence detectors , but not with surface arrays .
so far there are no indications for very deeply penetrating showers that would be expected in such a case .
given the importance of the absolute energy assignment to a reconstructed shower we will summarize the current systematic uncertainties below .
the energy reconstruction of showers detected with the agasa array is based on the scintillator signal @xmath214 at 600 m from the shower core , where shower - to - shower fluctuations are the smallest and the relation between the signal and the primary energy is almost composition independent @xcite . the systematic error of energy assignment is analyzed in @xcite in detail , see also discussion @xcite .
agasa finds a total systematic uncertainty of the energy assignment of about @xmath215% .
the main sources of uncertainty are related to shower phenomenology and the simulation of the relation of @xmath214 to the primary particle energy . in particular
, the observed discrepancy between the surface detector signal at 1000 m from the shower core and the fluorescence - based calorimetric energy measurement reported by auger @xcite indicates that currently available shower simulations do not allow to obtain an absolute energy scale with an systematic uncertainty smaller than @xmath21620% .
therefore it is not surprising that discrepancies between the experimentally observed attenuation length for @xmath214 and that expected from simulations hamper a re - analysis of the agasa data @xcite .
the shower energy determination applied in hires is based on the track length integral @xmath217 where @xmath218 is a fit to the shower profile using the gaisser - hillas function ( [ eq : gaisserhillas ] ) and @xmath219 denotes the mean ionization energy deposit @xcite . with hires being a fluorescence detector , the energy reconstruction is closely related to properties of the atmosphere which is serving as calorimeter . at the same time
, atmospheric properties also determine the aperture of the detector .
the hires flux measurements ( hires i and hires ii mono ) are found to have similar systematic uncertainties @xcite .
the main contributions to the systematic uncertainty of the energy scale are the absolute calibration of the pmts ( @xmath220% ) , the limited knowledge of the air fluorescence yield ( @xmath221% ) , and atmospheric conditions ( @xmath222% ) . about @xmath220% uncertainty results from the rescaling of the measured calorimetric energy to obtain the total shower energy @xcite ,
see also .
adding the individual contributions in quadrature , the overall systematic uncertainty of the energy reconstruction amounts to @xmath223% @xcite .
in contrast to surface arrays , the aperture of fluorescence detectors has to be determined by simulations .
sources of uncertainty are here varying atmospheric conditions , simulation of shower profiles and detector trigger thresholds , and the primary cosmic ray composition .
the uncertainty due to varying atmospheric conditions , mainly that of the vertical aerosol optical depth ( vaod ) , has been estimated to contribute to the aperture uncertainty @xmath224% @xcite . in a recent study
the other , simulation - related sources of uncertainty were found not to contribute significantly to the overall flux uncertainty of @xmath225% @xcite .
the technique employed in the auger measurement of the flux combines the advantages of surface detector arrays with that of fluorescence detectors @xcite .
the surface array operates with almost 100% duty cycle and the aperture can be calculated in a rather straight - forward way for energies well above the trigger threshold .
fluorescence telescopes allow the direct measurement of the calorimetric shower energy , however , their duty cycle is only about 13% . using a set of well - reconstructed hybrid events ,
one can calibrate the energy estimator for surface detector data in an almost model - independent way .
this is done in a two - step process .
first the shower signal at 1000 m , @xmath167 , is corrected for attenuation to that of an equivalent shower of @xmath190 zenith angle , @xmath169 . to avoid any possible bias from simulations this
is done with the constant intensity method by requiring the same number of showers per unit of @xmath226 . in the second step @xmath169
is converted to total shower energy .
the aperture of the auger detector increased continuously during construction and has an uncertainty of less than @xmath227% .
the systematic uncertainty of the energy scale coming from the fluorescence energy measurement is estimated to be @xmath228% .
the main contributions to this uncertainty are coming from the uncertainty of the fluorescence yield ( @xmath229% ) , the calibration of the fluorescence telescopes ( @xmath220% ) , and the reconstruction method ( @xmath220% ) @xcite . the described calibration procedure for relating s(1000 ) to
the primary particle energy leads to an uncertainty of @xmath230% at @xmath51ev increasing to @xmath224% at @xmath6ev .
[ compsec ] at energies below @xmath231 ev the abundance of individual elements has been measured with detectors above the atmosphere . at higher energies
this is presently not possible due to the low flux values and the large fluctuations in the development of extensive air showers .
thus , in the past , mostly the mean mass has been investigated . an often - used quantity to characterize
the composition is the mean logarithmic mass , defined as @xmath232 , @xmath233 being the relative fraction of nuclei of mass @xmath234 .
experimentally , @xmath235 is obtained applying two methods : ( i ) the quantity is proportional to the ratio of the number of electrons and muons registered at ground level @xmath236 , see and ( ii ) @xmath235 is proportional to the observed depth of the shower maximum , according to the relation @xmath237 , see .
hence , the maximum of an iron induced shower should be about 150 2 higher up in the atmosphere as compared to a proton induced shower ( @xmath238 )
. recent measurements of the average depth of the shower maximum are compiled in , covering energies from @xmath239 to almost @xmath240 gev .
the experimental results are compared to predictions of the average depth of the shower maximum from simulations for primary protons and iron nuclei .
the corsika code @xcite has been used with the hadronic interaction model qgsjet 01 @xcite , qgsjet ii-3 @xcite , sibyll 2.1 @xcite , and epos 1.6 @xcite .
there are significant differences between the predictions of the different models concerning the absolute values of .
the differences become important when the model predictions are compared to the experimental data to derive information on the elemental composition of cosmic rays . below @xmath241 gev
the values obtained by different experiments exhibit a common trend , they seem to increase faster as function of energy than the simulations , which implies that the average composition would become lighter as function of energy . above the knee ( @xmath242 gev )
the measured values flatten up to about @xmath243 gev , indicating an increase of the average mass in this energy range , as expected from sequential breaks in the energy spectra for individual elements , seen already in . finally , above @xmath243 gev
the measured data exhibit about a constant slope for as function of energy .
the slope is slightly steeper than the predicted slope for iron nuclei for all models shown .
but , on the other hand , a comparison to the predicted proton values is not conclusive , while models like qgsjet 01 favor an extremely light composition at the highest energies , models like dpmjet 2.5 give a hint to an intermediate average mass . knowing the average depth of the shower maximum for protons @xmath238 and iron nuclei @xmath244 from simulations , the mean logarithmic mass is derived in the superposition model of air showers from the measured @xmath245 values using @xmath246 .
the corresponding @xmath235 values , obtained from the results shown in , are plotted in ( ) as function primary energy using the hadronic interaction model qgsjet 01 to interpret the observed data . for comparison , also
results of direct measurements are shown ( jacee and runjob ) . in the figure three energy ranges
may be distinguished for the indirect measurements .
below about @xmath241 gev the individual experiments seem to indicate a decrease of @xmath235 with energy , while above this energy up to about @xmath243 gev an increase with energy is exhibited . at the highest energies @xmath247 gev ,
again a decrease with energy can be stated .
results of measurements of electrons , muons , and hadrons at ground level interpreted with the hadronic interaction code qgsjet 01 are compiled in ( ) .
they yield a clear increase of the mean logarithmic mass as function of energy .
there seems to be some tension between the results obtained through the observation of the average depth of the shower maximum shown in the top panel and values derived from particle ratios measured at ground level depicted in the bottom panel @xcite . in particular , at energies below the knee ( @xmath242 gev ) the decrease of @xmath235 as derived from some measurements is not visible in the particle ratio results .
using the latest version of qgsjet ii does not change the situation qualitatively .
the threefold structure of the results obtained is about the same as using qgsjet 01 .
the main difference are the absolute @xmath235 values which are shifted upwards by about 0.8 units for qgsjet ii-3 with respect to qgsjet 01 . using the hadronic interaction model sibyll 2.1 yields about the same values as for qgsjet ii . using lower inelastic hadronic cross sections in the qgsjet 01 code and larger values for the elasticity of hadronic interactions the discrepancies between measurements and particle ratios at ground
can be reduced @xcite .
in addition to the mean mass as discussed above , it is interesting to investigate the energy spectra for individual elements or at least groups of elements .
information on the flux of primary protons can be inferred from the measurement of the flux of unaccompanied hadrons at ground level . with the kascade hadron calorimeter
the energy spectrum of single hadrons close to sea level has been measured in the energy range from 100 gev up to 50 tev @xcite . based on simulations using the corsika code with the hadronic interaction model
qgsjet 01 @xcite the energy spectrum of primary hadrons in the energy range from 100 gev to 1 pev has been derived . over the whole four decades in energy
it can be described by a single power law .
the kascade experiment used the number of electrons and muons ( @xmath248 mev ) measured in the scintillator array to reconstruct energy spectra for five primary elemental groups @xcite .
starting point of the analysis is the correlated frequency distribution of the number of electrons @xmath128 and the number of muons @xmath134 .
unfolding algorithms were applied to derive energy spectra for elemental groups . for the analysis the primary particles h , he , c , si , and fe were chosen as representatives for five mass groups .
details of the analysis and the used unfolding methods can be found in ref .
@xcite .
the spectra obtained are presented in . to check the influence of the description of hadronic interactions in the atmosphere on the result
, the same experimental data were unfolded using two interaction models , namely qgsjet and sibyll .
the corresponding results are displayed in the figure .
the resulting all - particle spectra for both models show a knee at about 4 pev and coincide within their statistical errors .
the decrease of light elements across the knee , i.e. the occurrence of knee - like features in the light element spectra is revealed independently of the used simulation code , as can bee seen in . in contrast , the spectra of silicon and iron groups differ significantly and look quite unexpected .
this can be understood by judging the ability of the simulations to describe the data .
it turns out that both interaction models fail to reproduce the overall correlation between @xmath249 and @xmath250 as observed in the data . in the case of qgsjet simulations
the predictions are incompatible with the data in the low energy regime ( simulations look too heavy ) , for sibyll incompatibility occurs at higher energies ( simulations look too light ) . summarizing this analysis , for the first time energy spectra for groups of elements
were reconstructed from air shower data .
the spectra indicate that the knee in the all - particle spectrum is due to fall - offs in the light element spectra resulting in a heavier composition above the knee . the eas - top collaboration combined several detection systems to obtain information on the energy spectra of primary cosmic rays .
the results are summarized in .
the proton spectrum in the energy range 0.5 - 500 tev @xcite has been obtained from the measurement of unaccompanied hadrons with a calorimeter , taking into account the contribution of helium nuclei as obtained by direct measurements @xcite .
the proton spectrum is described over the whole energy range by a single power law .
the helium and cno fluxes in the energy region from 80 to 200 tev have been studied from measurements of the light and tev muons registered with the underground macro experiment @xcite .
primaries are selected through their energy / nucleon by means of the tev muon information .
the shower energy is inferred from the measurement of the light yield at distances from 125 to 185 m from the shower core .
the flux for p+he at 80 tev and for p+he+cno at 250 tev was obtained . by subtracting the measured proton flux
the following values were calculated : @xmath251 m@xmath252sr@xmath253s@xmath253gev@xmath253 and @xmath254 m@xmath252sr@xmath253s@xmath253gev@xmath253 .
the all - particle energy spectrum is obtained from the measured shower size spectra in the knee region @xcite , showing the angular ( i.e. depth ) dependence of the knee position .
the knee is observed at @xmath255 in the vertical direction , corresponding to a primary energy @xmath256 ev , and intensity of @xmath257 m@xmath252s@xmath253sr@xmath253 with about 20% uncertainty .
the obtained power law indices of the energy spectrum are : @xmath258 and @xmath259 , below and above the knee , respectively .
a composition analysis at knee energies was performed for vertical showers ( @xmath260 ) through measurements of the number of electrons @xmath128 and the number of muons with energies above 1 gev recorded in the muon detector at core distances @xmath261 m ( @xmath262 ) .
the experimental @xmath262 distributions , measured in different shower size intervals are fitted with simulated data to obtain energy spectra for groups of elements .
intrinsic fluctuations and measurement accuracies allow a three component analysis : light ( constructed either with protons , and a mixture of 50% proton and 50% helium ) , intermediate ( cno ) , and heavy ( fe ) .
the shaded areas in indicate the energy dependence of the flux thus obtained for the three components .
the tibet air shower array has the advantage to be located at high altitude ( 4300 m , 606 2 ) .
it comprises a scintillation counter array as well as emulsion chambers and burst detectors .
the data were used to derive spectra for primary protons and helium nuclei , see @xcite .
however , one has to keep in mind that only a few hundred events remain after quality cuts and are included in the analysis , which may indicate that the results are limited by their statistical significance .
similar to kascade and eas - top , the grapes-3 experiment uses the correlation between the registered number of electrons and muons to derive the flux for individual elemental groups .
spectra for protons , helium , as well as the cno , silicon , and iron groups have been obtained @xcite , see . in 1993 observations with fly s eye @xcite gave the first indication of a systematic change of the cosmic - ray mass composition at very high energy . analyzing the mean depth of shower maximum , @xmath263 , a change from an iron dominated composition at @xmath40ev to a proton dominated composition at @xmath264ev was found . however , an analysis entirely based on the mean @xmath265 is strongly model dependent ( e.g. @xcite ) . in fig .
[ fig : mean - xmax-2008 ] ( ) a compilation of measurements of @xmath115 is shown together with model predictions . adopting the qgsjet 01 model @xcite the conclusions of @xcite
still hold , though a mixed composition is expected at @xmath40ev . on the other hand , on the basis of models like sibyll 2.1 @xcite or dpmjet 2.55 @xcite a much more moderate change of the composition
is derived .
the model ambiguity of the interpretation of @xmath115 can be resolved to some degree by studying the measured distribution of @xmath266 @xcite .
in contrast to the old measurements of fly s eye @xcite and yakutsk @xcite the hires data indicate a change from an iron - like to a proton dominated composition already at @xmath52ev .
the two independent measurements are consistent in the overlap region .
the large elongation rate of the low - energy data of @xmath267g / cm@xmath5 @xcite can only be understood in terms of a change of composition ( see ) . any model with scaling violations
will predict a change to a lighter composition @xcite .
also the muon density measured with the mia detector @xcite in the hires - mia setup indicates a change from a heavy to light composition .
the observed muon densities , however , are higher or similar than the expectation for iron primaries and not compatible with medium or light nuclei @xcite .
the new auger data on the average depth of shower maximum @xcite are , within the systematic uncertainties of about @xmath268g / cm@xmath5 in good agreement with the published hires data . on the other hand , if one would just analyze auger data , a break in the elongation rate is found at a higher energy , at about @xmath269ev .
the elongation rate is @xmath270g / cm@xmath5 below this break and @xmath271g / cm@xmath5 above @xcite .
the highest energy point of the auger data is smaller than expected from a constant elongation rate .
it is calculated from 13 observed showers and could indicate a transition to heavier elements as one would expect in case of a rigidity - dependent maximum energy of sources . the right panel of fig .
[ fig : compositionmuons ] shows a compilation of composition measurements based on particle detectors , as published in .
the akeno result given here is re - scaled according to the predictions of the qgsjet model @xcite.ev appeared to be in contradiction to the fly s eye composition interpretation @xcite , see also the analysis in @xcite . ]
the muon density of ultra high - energy showers measured with agasa was analyzed in . comparing the data with model predictions from aires @xcite and qgsjet 01 the following limits on a two component composition
were derived : less than 35% iron in the energy range @xmath272ev and less than 76% at higher energy ( 90% c.l . )
@xcite .
a re - analysis of haverah park data was done in .
the authors use the sensitivity of the steepness of the lateral particle distribution to the shower development height , which in turn depends on the depth of shower maximum as composition sensitive observable .
it was found that the predictions of corsika with the qgsjet 98 model give a good description of the data if a two - component composition with about ( @xmath273)% iron is used in the energy range from @xmath274 to @xmath52ev . at higher energy ( from @xmath52 to @xmath275ev ) , indications are seen for a transition to a lighter composition ( see fig .
[ fig : compositionmuons ] ) .
this is supported by the number of inclined showers with @xmath276ev that have triggered the haverah park array @xcite .
the data analyzed in @xcite agree well with simulations assuming all primary particles are protons , though no mass composition study was done . on the other hand ,
a first study of the time structure of haverah park showers with zenith angles less than 45@xmath277 finds a more iron - dominated composition in the same energy range @xcite .
also a re - analysis of volcano ranch data similar to the haverah park favors a large fraction of iron @xcite . using @xmath278 showers in the energy range from @xmath279 to @xmath51ev a fraction of @xmath280% ( @xmath281% ) iron is found for a two component composition and qgsjet 98 ( qgsjet 01 ) .
the discrepancy between muon density - based composition measurements and others based on features of the longitudinal profile underlines the shortcomings of the hadronic interaction models currently available .
there seems to be a systematic deficit of muons predicted in simulations in comparison with data @xcite .
first progress in addressing this problem has been made in @xcite by increasing the number of pair - produced baryons in the simulation . using gamma - ray telescopes , such as magic , h.e.s.s . , or
veritas , photons with energies of about 50 tev have been observed from several sources in the sky , e.g.@xcite . at higher energies various air shower experiments searched for gamma rays .
air showers induced by primary photons develop an almost pure electromagnetic cascade .
experimentally they are identified by their relatively low muon content or their relatively deep shower maximum . since
mostly electromagnetic processes are involved in the shower development , the predictions are more reliable and do nt suffer from uncertainties in hadronic interaction models .
results up to energies of @xmath282 ev are summarized in ( ) @xcite .
there are no indications of a substantial fraction of gamma - rays in the high - energy cosmic - ray flux . the longitudinal profile of the highest energy event observed by fly s eye ( @xmath283ev ) @xcite has been studied by several groups .
comparing the measured shower profile with monte carlo simulations shows that this event is well described by a hadronic primary particle @xcite .
however , due to the large reconstruction uncertainty of the atmospheric depth of the shower profile , a photon can not be excluded @xcite .
the deeply penetrating muon component of inclined showers of hadronic origin is employed in an analysis of haverah park data in @xcite . using the primary cosmic - ray flux parametrization of @xcite less than 48% of the observed events above @xmath51ev can be photons ( 95% c.l . ) . at energies above @xmath284ev this limit
is 50% . based on the analysis of muons observed in high - energy showers at agasa the following upper limits
were derived in @xcite : 34% , 59% and 63% for primary energies above @xmath51 , @xmath285 , and @xmath57ev , respectively ( 95% c.l . ) .
a separate analysis of the 6 highest energy events of agasa was performed in @xcite . using a new method that accounts for the arrival direction of each individual shower , a limit of 67% at 95% c.l .
could be derived for @xmath286ev .
this new analysis method was also employed in a recent study of auger shower longitudinal profile data @xcite .
no more than 16% photons are expected at 95% c.l .
it should be noted that this limit as well as the one derived in @xcite are independent of assumptions on hadronic multiparticle production at very high energy .
it relies only on the simulation of photon showers which is much better under control than that of hadron - induced showers .
in , the scintillator signals of 10 showers above @xmath6ev from agasa and yakutsk were analyzed to derive a photon fraction limit .
the authors report the strongest limit on the photon fraction for showers above @xmath6ev available so far , less than @xmath287% photons at 95% c.l . to obtain this limit ,
new energies are assigned to the considered showers , partially being very much different from those originally reconstructed .
this shows the importance of simulating and understanding detector effects which can only be done in a limited way in such studies
. a confirmation by the agasa and yakutsk experiments would be very important to establish this limit beyond doubt .
a compilation of recent upper limits on the contribution of photons to the all - particle flux is shown in @xcite .
the best photon limits are the latest results of the pierre auger observatory @xcite setting rather strong limits on the photon flux .
they are based on measurements with the auger surface detectors , taking into account observables sensitive to the longitudinal shower development , the signal rise time , and the curvature of the shower front .
the photon fraction is smaller than 2% , 5.1% , and 31% above energies of @xmath51 , @xmath288 , and @xmath289 ev , respectively with 95% confidence level . in top - down scenarios for high - energy cosmic rays , the particles are decay products of super - heavy objects .
the decay process yields relatively high fluxes of photons , a typical feature of such models @xcite .
several predictions are shown in the figure @xcite .
these scenarios are strongly disfavored by the recent auger results .
it should also be noted that the upper limits are already relatively close to the fluxes expected for photons originating from the gzk effect @xcite , shown in the figure as shaded area .
the detection of ultra high - energy cosmic neutrinos is a long - standing experimental challenge .
many experiments are searching for such neutrinos , and there are several ongoing efforts to construct dedicated experiments to detect them @xcite .
their discovery would open a new window to the universe @xcite .
however , so far no ultra high - energy neutrinos have been detected .
due to interactions in the source region or during the propagation processes , ultra high - energy cosmic rays are expected to be accompanied by ultra high - energy neutrinos .
the neutrinos are produced with different abundances for the individual flavors , e.g. pion decay leads to a ratio @xmath290 .
however , due to neutrino oscillations the ratio expected at earth is @xmath291 . to discriminate against the huge hadronic background in air shower detectors , neutrino candidates
are identified as nearly horizontal showers with a significant electromagnetic component @xcite .
the pierre auger observatory is sensitive to earth - skimming tau neutrinos that interact in the earth s crust .
tau leptons from @xmath292 charged - current interactions can emerge and decay in the atmosphere to produce a nearly horizontal shower with a significant electromagnetic component .
recent results from the pierre auger observatory together with upper limits from other experiments are presented in .
assuming an @xmath293 differential energy spectrum , the auger collaboration derived a limit at 90% confidence level of @xmath294 gev cm@xmath252 s@xmath253 sr@xmath253 in the energy range between @xmath296 and @xmath297 ev @xcite . according to top - down models for ultra high - energy cosmic rays a large flux of ultra high - energy neutrinos
is expected . as an example , the predictions of a model @xcite are shown in the figure as well . this model is disfavored by the recent upper limits .
it should also be noted that the current experiments are only about one order of magnitude away from predicted fluxes of gzk neutrinos .
[ anisosec ] the search for anisotropies in the arrival direction of cosmic rays on different angular scales can contribute to the understanding of the cosmic - ray origin , in particular the identification of source regions or individual sources .
large - scale anisotropies are connected to the propagation process of cosmic rays in the galaxy , while small - scale anisotropies would be a hint towards cosmic - ray sources .
however , it has to be considered that the larmor radius of protons with an energy around 1 pev in the galactic magnetic field ( @xmath298 g ) is of the order of 0.4 pc .
hence , it is not expected to find any point sources for galactic cosmic rays .
the situation changes for the highest energies , see below .
an excess of charged particles with energies above @xmath0 ev from the direction of a snr ( monogem ring , @xmath299 pc ) has been reported @xcite and later withdrawn @xcite .
this supernova remnant with an age of about @xmath239 yr has been suggested as possible single source of galactic cosmic rays @xcite .
a signal could not be confirmed neither by the kascade experiment @xcite investigating cosmic rays with energies above 0.3 pev nor by the tibet experiment , looking for pev @xmath179-ray emission @xcite .
the tibet air shower experiment has performed a northern sky survey , looking for tev @xmath179-ray point sources in a declination range from 0to 60@xcite .
a small excess is found , most likely caused by well known @xmath179-ray sources such as in the crab nebula and mrk 421 . at higher energies ,
the kascade experiment has performed a detailed search for point sources , covering the whole visible sky at energies @xmath300 pev @xcite .
the visible sky has been divided in cells with a size of 0.5 .
two distributions have been investigated , all events and a selection of muon poor showers .
the latter have been investigated in order to look for potential gamma rays .
they would manifest themselves in air showers with no or little muons only .
both investigations indicate an isotropic distribution of the arrival direction of cosmic rays . special attention has been given to the region of the galactic plane , as well as to the vicinity of known snrs and tev-@xmath179-ray sources .
no significant excess could be found in either sample .
the super - kamiokande experiment investigated large - scale anisotropies for cosmic rays with energies around 10 tev @xcite .
the experiment registers muons from air showers with an energy threshold of 0.8 tev .
an anisotropy map in celestial coordinates is obtained by binning the data in pixels of @xmath301 . the result after applying a smoothing algorithm is shown in ( ) .
each cell represents the relative deviation from an isotropic distribution .
a @xmath302 excess ( `` taurus excess '' ) is found with an amplitude of @xmath303 at right ascension @xmath304 and declination @xmath305 . on the other hand ,
a deficit ( `` virgo deficit '' ) is found with an amplitude of @xmath306 at @xmath307 and @xmath308 .
the angular difference between the taurus excess and the virgo deficit amounts to about 130 .
a large - scale anisotropy would be expected if the earth moves relative to a cosmic - ray rest system ( compton getting effect @xcite ) .
for such a scenario the angular difference between maximum and minimum flux should be 180 . within the relatively large angular uncertainties
the anisotropy observed by the super - kamiokande experiment would be compatible with a compton getting effect caused by a velocity of about 50 km / s relative to the rest frame .
this velocity is smaller than the rotation speed of the solar system around the galactic center ( @xmath309 km / s ) .
this implies that the rest frame of cosmic rays ( presumably the galactic magnetic fields ) is co - rotating with the galaxy .
the tibet experiment reported anisotropies in the same regions on the sky @xcite . the observed intensity for different energy thresholds
is displayed in ( ) . for energies below 12 tev the anisotropies show little dependence on energy , whereas above this energy anisotropies fade away , consistent with measurements of the kascade experiment in the energy range from 0.7 to 6 pev @xcite .
a compton getting effect caused by the orbital motion of the solar system around the galactic center would cause an excess at @xmath310 , @xmath311 and a minimum at @xmath312 , @xmath313 with an amplitude of 0.35% .
however , the measurements at 300 tev yield an anisotropy amplitude of @xmath314 , consistent with an isotropic cosmic - ray intensity .
hence , a galactic compton getting effect can be excluded with a confidence level of about @xmath315 .
this implies , similar to the result of the super - kamiokande experiment , that galactic cosmic rays co - rotate with the local galactic magnetic field environment .
the tibet experiment finds also an excess with a 0.1% increase of the cosmic ray intensity in the direction of the cygnus region ( @xmath316 , @xmath317n ) @xcite . in the same region
an excess of @xmath179-rays with energies around 10 tev is seen by the milagro experiment @xcite .
recently , the milagro experiment reported also an excess of ( charged ) cosmic rays with a significance greater than 12@xmath211 @xcite .
presently , different explanations are discussed in the literature : the excess could be related to the geminga pulsar as local cosmic - ray source @xcite or cosmic rays from a local source could reach the earth via a magnetic mirror @xcite . at energies around the knee
the rayleigh formalism has been applied by several groups to characterize the large - scale anisotropy .
the two - dimensional distribution of the arrival directions of cosmic rays is reduced to one coordinate because of the limited field of view ( typical experiments cover only a fraction of the whole sky ) and the small amplitudes expected from theory . a first order approximation of the multipole expansion of the arrival directions of cosmic rays is a harmonic analysis of the right ascension values of extensive air showers .
applying the rayleigh formalism to the right ascension distribution of extensive air showers measured by kascade yields no hints of anisotropy in the energy range from 0.7 to 6 pev @xcite .
this accounts for all showers , as well as for subsets containing showers induced by predominantly light or heavy primary particles
. also other experiments , like mt .
norikura @xcite , eas - top @xcite , akeno @xcite , and adelaide @xcite have derived rayleigh amplitudes .
some experiments find anisotropies , however the phases do not agree between the different results . hence , it seems to be more likely that all amplitudes derived should be considered as upper limits .
different groups have searched for large - scale anisotropies but no confirmed deviation from an isotropic arrival direction distribution has been found .
a dipole amplitude that is compatible with full isotropy was found in an analysis of more than 135000 showers with energies from @xmath318 to @xmath207ev of the yakutsk array @xcite .
also no significant anisotropy is found in the agasa data @xcite in the energy range between @xmath40 to @xmath207ev . in the energy range around @xmath52ev , however , agasa found an excess of showers coming from directions near the galactic center and the cygnus region @xcite .
furthermore , analysis of sugar data @xcite indicates an excess of cosmic rays coming from a similar region . the haverah park and
yakutsk arrays are located too far north to be able to see the excess regions of agasa and sugar . with the galactic center being in the field of view of the southern auger observatory , it is ideally suited to search for these possible source regions , though the energy threshold for reaching 100% acceptance of the auger surface array is just below @xmath208ev .
recently , the auger collaboration has performed an analysis of their low - energy showers from the direction of the galactic center . in the energy range from @xmath319ev to @xmath208ev
, no abnormally over - dense regions were found in the neighborhood of the galactic center @xcite .
there are 506 ( 413.6 ) events found ( expected ) in the agasa data set for a circle of @xmath320 around @xmath321 right ascension and @xmath322 declination , corresponding to an excess ratio of 1.2 . in the same region ,
the auger collaboration observed 2116 events while 2170 were expected . at energies above @xmath208ev , the large - scale structure of
the arrival direction distribution appears , within the limited statistics of the agasa array , isotropic @xcite .
this finding agrees with that of the hires collaboration , performing a global anisotropy search based on @xmath323 events observed by hires i in monocular mode @xcite . by combining data from arrays of the northern and southern
hemispheres a full sky anisotropy study is done in @xcite .
considering in total 99 showers from agasa and sugar with @xmath324 no large - scale anisotropy is found .
the data set of the auger observatory analyzed in @xcite contains more than 7000 events with @xmath325ev .
no large scale anisotropy is found .
the upper limit to the amplitude of a dipole - like anisotropy in right ascension is 3% with 95% cl @xcite .
small angle clustering could be an indication for point sources . at energies above @xmath326ev , clustering at an angular scale of @xmath327
has been reported by the agasa collaboration @xcite . this result could not be confirmed .
an analysis of a hires i data set corresponding to an exposure similar to that of agasa did not reveal any evidence for small scale clustering @xcite .
also no significant clustering is seen in a data set from hires stereo observations of more than 270 showers with @xmath276ev @xcite and in the combined agasa - hires stereo data set with @xmath328ev @xcite .
clustering at the angular scale of the agasa signal is also not found in the auger data @xcite .
uchihori et al . included showers above @xmath326ev from all four surface arrays of the northern hemisphere in their small scale correlation analysis @xcite .
the combined data set is found to contain many clusters , however , the statistical significance is low ( @xmath329% ) . on the other hand ,
there are indications for a correlation with the super galactic plane . restricting the considered arrival directions to the range of @xmath330 off the super galactic plane the chance probability for finding doublets and triplets decreases to the order of 1% .
studies of small angle correlations with the yakutsk array are difficult because of the angular resolution of the shower axis reconstruction , which is about @xmath331 @xcite .
nevertheless , at much lower energy , clusters of the arrival directions of showers in the energy range @xmath332ev were reported in @xcite .
the direction of these clusters seem to support a correlation with the super galactic plane . dividing the observed cosmic - ray showers into isotropic and cluster components this correlation
can be enhanced significantly @xcite .
a recent search for correlations at a medium angular scale has been carried out in @xcite .
publicly available data sets from agasa , hires , sugar , and yakutsk with @xmath333ev were combined by adjusting their energy scales to shift the ankle in all data sets to the same energy .
in addition events with @xmath334ev from haverah park , volcano ranch , and fly s eye were considered . in this set of 107 events in total , a signal at a scale of @xmath335 is found in the autocorrelation function .
results from the pierre auger observatory also indicate some clustering on intermediate scales with an angular separation of order of 15to 25@xcite .
the medium angle correlation reported in @xcite has a 2% chance probability to originate from an isotropic distribution .
there is a long history of searches for correlations with astrophysical point sources such as colliding galaxies and powerful radio galaxies .
it appears almost impossible to assess unambiguously the chance probability of such correlations .
first of all , highly incomplete catalogs of astrophysical objects necessarily have to be used in these analyses .
secondly , the penalty factor for scans of several catalogs , selecting different classes of objects , distance and angular ranges , and other parameters can not be calculated reliably .
part of these problems can be avoided by defining a prescription before analyzing a data set .
this has been done by the auger collaboration @xcite .
a correlation with bl lacertae , at a distance exceeding the gzk energy loss length , has been claimed for the agasa and yakutsk high - energy data ( @xmath336ev ) @xcite .
the significance of this correlation is highly debated @xcite as there might be `` hidden '' trials involved that can not be corrected for with a monte carlo simulation .
these correlations were tested with the independent data set of hires stereoscopic observations @xcite .
none of the previous claims could be confirmed .
recently , a @xmath337 correlation between bl lacertae objects of the veron - cetty & veron catalog @xcite with hires data ( @xmath338ev ) was pointed out @xcite .
this correlation was confirmed at a nominal significance of about @xmath339% not only for the high - energy part but also for the entire set of hires stereo data @xcite .
an analysis of data from the pierre auger observatory did not confirm a correlation of the arrival direction of cosmic rays with the positions of bl lacertae objects in the southern hemisphere @xcite .
a breakthrough in the anisotropy searches is the correlation discovered by the auger collaboration @xcite .
the arrival directions of the very highest energy events ( @xmath340ev ) are found to be correlated with the positions of active galactic nuclei ( agns ) from the veron - cetty & veron catalog @xcite . out of 27 events observed with an integrated aperture of 9000 km@xmath5 sr yr ,
20 are correlated with agns within an angular distance of @xmath341 .
the correlation was initially found in an exploratory search with different catalogs ( 12 out of 15 events were correlated ) .
a prescription was set up to verify or reject the correlation hypothesis using an independent data set .
of the next 8 events that were detected , 6 were correlated with agns within the prescribed phase space , corresponding to a chance probability of incorrectly accepting the hypothesis of correlation of less than 1% . after accepting the correlation hypothesis ,
the auger group made a scan to refine the correlation parameters and found an energy threshold of @xmath342ev , an `` source '' distance of less than @xmath343 ( @xmath344mpc ) and a maximum angular difference of @xmath345 as optimal parameters .
if one does not account for the penalty factor due to different trials and the parameter scans , the nominal chance probability of being compatible with isotropy of @xmath346 would be obtained .
a sky map of the measured arrival direction distributions is shown in fig .
[ fig : agn ] using equatorial coordinates .
the relative exposure of the auger observatory is indicated by the shaded regions .
the veron catalog of agns is not a complete catalog .
as expected , in the direction of the galactic plane the density of detected agns is lower than in other directions .
the nearby agns are very good tracers of the local matter distribution .
in particular , the super galactic plane is clearly visible .
a recent update of the super galactic plane position even improves the correlation between agns and this matter over - density @xcite .
the correlation of the arrival direction distribution of uhecrs with nearby agns is also reflected by the autocorrelation function which shows some indications of anisotropy in the 15to 25range , as one would expect from the typical thickness of the super galactic plane . given the limited statistics one can not conclude from the found correlation that agns are sources of uhecrs .
subsequent studies of the published highest energy events of the auger observatory revealed correlations with the large scale structure of galaxies @xcite .
a correlation of uhecrs with the large scale structure in the cosmological neighborhood is also found in the yakutsk data set @xcite .
the hires collaboration has used the correlation parameters published by auger to perform a search in their stereo data set . to obtain the same energy threshold as used in the auger analysis , they scaled their reconstructed energies down by @xmath220% @xcite . using all stereo data an exposure of roughly 4000 km@xmath5sryr
is obtained at the highest energy ( see ) . for a total of 13 events two associations with agns were found , while 3.2 such correlations are expected for an isotropic arrival direction distribution .
no correlation signal is found .
also the autocorrelation function of the highest energy events from hires is perfectly in agreement with the expectations for isotropy .
the arrival directions of the selected 13 events are shown in fig .
[ fig : agn ] as well .
the exposure distribution of the hires data set is very similar to that of the auger observatory , but north exchanged with south .
the discrepancy between the auger and hires results are currently not understood , but it is clear that a possible difference of the energy scale of the two experiments could lead to such effects .
in addition the reconstruction resolution has to be very good to reproduce the very sharp threshold of the correlation found in auger data .
an independent data set of similar size as the published one will allow to test the anisotropy signal .
[ astrosec ] the measurements indicate that the knee in the all - particle energy spectrum is caused by a break in the spectra for the light elements , yielding an increase of the mean mass of cosmic rays in this energy region .
several scenarios are discussed in the literature as possible origin for the knee , see e.g.@xcite . in the following ,
a current astrophysical picture of the origin of high - energy cosmic rays is sketched , based on recent observations .
one of the most popular explanations for the origin of the knee is that the spectra at the source exhibit a break .
the bulk of cosmic rays is assumed to be accelerated in strong shock fronts of snrs @xcite .
the finite lifetime of a shock front ( @xmath119 a ) limits the maximum energy attainable for particles with charge @xmath42 to @xmath347 pev .
many versions of this scenario have been discussed @xcite .
the models differ in assumptions of properties of the snrs like magnetic field strength , available energy , ambient medium , etc .
the differences of the predicted @xmath235 can be inferred from a. while older models @xcite limit the maximum energy to about 0.1 pev , recent ideas @xcite , taking into account latest observations of snrs , predict maximum energies above 1 pev . in such a model sufficient energy
is released from snrs to explain the observed spectra .
a special case of snr acceleration is the single source model @xcite , which predicts in the knee region pronounced structures in the all - particle energy spectrum , caused by a single snr .
such structures can not be seen in the compilation of . in the literature also
other acceleration mechanisms , like the acceleration of particles in @xmath179-ray bursts , are discussed @xcite .
they differ in their interpretation of the origin for the knee .
the approach by plaga , assuming fermi acceleration in a `` cannon ball '' is not compatible with the measured @xmath235 values , see b. a different interpretation of acceleration in the cannon ball model yields at the source a cut - off for individual elements proportional to their mass due to effects of relativistic beaming in jets .
the predictions of the actual model are compatible with recent data @xcite .
however , it remains to be clarified how a detailed consideration of the propagation processes , e.g. , in a diffusion model , effects the cut - off behavior observed at earth .
gamma - ray bursts as a special case of supernova explosions are proposed @xcite to accelerate cosmic rays from 0.1 pev up to the highest energies ( @xmath1 ev ) . in this approach
the propagation of cosmic rays is taken into account and the knee is caused by leakage from the galaxy leading to a rigidity dependent cut - off behavior .
also frequently discussed is the idea that the knee is a propagation effect .
the propagation is accompanied by leakage of particles from the galaxy . with increasing energy
it becomes more and more difficult to confine the nuclei to the galaxy . as mentioned above
, the path length decreases as @xmath348 .
such a decrease will ultimately lead to a complete loss of the particles , with a rigidity dependent cut - off of the flux for individual elements .
many approaches have been undertaken to describe the propagation process @xcite .
the leaky box model @xcite and the anomalous diffusion model @xcite yield cut - offs significantly weaker than the data shown in and @xcite .
the propagation as described in diffusion models @xcite yields @xmath235-values which are presented in c. the models are based on the same principal idea @xcite , but take into account different assumptions on details of the propagation process , like the structure of galactic magnetic fields etc .
this results in a more or less strong cut - off for the flux at the individual knees and , accordingly , in a more or less strong increase of @xmath235 .
the observed break of the spectra is relatively sharp .
it seems to be difficult to generate such a behavior by a cut - off at the source or due to propagation only .
more likely seems to be a combined approach assuming a break of the spectra at the source and leakage from the galaxy , as e.g. discussed in @xcite . during the propagation phase ,
reacceleration of particles has been suggested at shock fronts in the galactic wind @xcite .
also this mechanism yields a rigidity dependent cut - off .
another hypothetical explanation for the knee are interactions of cosmic rays with background particles like massive neutrinos @xcite or photodisintegration in dense photon fields @xcite .
such models appear to be excluded with a high level of confidence .
the interactions would produce a large amount of secondary protons , which results in a light mass composition at high energies , not observed by the experiments , see d. furthermore , a massive neutrino , proposed in @xcite can be excluded by measurements of the wmap and 2dfgrs experiments @xcite .
a completely different reason for the knee is the idea to transfer energy in nucleon - nucleon interactions into particles , like gravitons @xcite or extremely high - energy muons @xcite , which are not observable ( or not yet observed ) in air shower experiments .
the latter proposal seems to be excluded by measurements of the baikal experiment @xcite setting upper limits for the flux of muons above @xmath239 gev .
no point sources of charged cosmic rays were found in the knee region .
the observed ( large - scale ) anisotropy amplitudes in the energy region of the knee are compared to model predictions in .
two versions of a leaky - box model @xcite , with and without reacceleration are shown .
leaky - box models , with their extremely steep decrease of the path length as function of energy ( @xmath349 ) , yield relatively large anisotropies even at modest energies below @xmath350 gev , which seem to be ruled out by the measurements , see also @xcite . the measured values are almost an order of magnitude smaller . on the other hand , a diffusion model @xcite , which is based on the idea of ref .
@xcite , predicts relatively small values at low energies and a modest rise only . in the figure
, predictions for pure protons and iron nuclei are shown together with calculations for a mixed composition .
the predicted rayleigh amplitudes are compatible with the measured values .
this may indicate that diffusion models are a realistic description of cosmic - ray propagation in the galaxy at pev energies .
in addition to information extracted from measurements of charged particles , important hints towards the origin of ( hadronic ) cosmic rays may be derived from observations of high - energy @xmath179-rays .
recent observation of the h.e.s.s .
experiment improve significantly the knowledge about galactic cosmic - ray sources .
for the first time a spatially resolved image of a supernova remnant has been obtained with multi - tev gamma rays as shown in ( ) @xcite .
the shell type supernova remnant rx j1713.7 - 3946 has been studied in detail @xcite .
the image of the remnant has been divided into 14 regions to study the energy spectrum of gamma rays up to eight tev .
the indices of the power law spectra vary between @xmath351 and @xmath352 in the different fields with an overall index of the remnant @xmath353 .
the spectral indices are very close to the spectral steepness expected from fermi acceleration at strong shocks .
the morphology of the tev @xmath179-ray image agrees well with emissions in the 1 - 3 kev x - ray regime as measured by the asca satellite .
the photon energy spectrum of the supernova remnant rx j1713 is presented in ( ) .
measurements in various energy ranges ( atca at radio wavelengths , asca x - ray , egret gev @xmath179-ray , cangaroo and h.e.s.s .
tev @xmath179-ray ) are compared to predictions of a model by berezhko & vlk @xcite .
the solid line below @xmath350 ev indicates synchrotron emission from electrons ranging from radio frequencies to the x - ray regime .
the observed synchrotron flux is used to adjust parameters in the model , which in turn , is used to predict the flux of tev @xmath179-rays .
the solid line above @xmath350 ev reflects the spectra of decaying neutral pions , generated in interactions of accelerated hadrons with material in the vicinity of the source ( hadron + ism @xmath354 ) .
this process is clearly dominant over electromagnetic emission generated by inverse compton effect and non - thermal bremsstrahlung , as can be inferred from the figure .
the results are compatible with a nonlinear kinetic theory of cosmic - ray acceleration in supernova remnants and imply that this supernova remnant is an effective source of nuclear cosmic rays , where about 10% of the mechanical explosion energy are converted into nuclear cosmic rays @xcite . further quantitative evidence for the acceleration of hadrons in supernova remnants is provided by measurements of the hegra experiment @xcite of tev @xmath179-rays from the snr cassiopeia a @xcite and by measurements of the h.e.s.s .
experiment from the snr `` vela junior '' @xcite . in conclusion
, it may be stated that a standard picture of the origin of galactic cosmic rays seems to emerge from the data .
the measurements seem to be compatible with the assumption that ( hadronic ) cosmic rays are accelerated at strong shock fronts of supernova remnants .
the particles propagate in a diffusive process through the galaxy . as origin for the knee a combination of the maximum energy attained in the acceleration process and leakage from the galaxy seems to be favored .
different scenarios are discussed in the literature for the transition from galactic to extragalactic cosmic rays .
the transition most likely occurs at energies between @xmath40 and @xmath52 ev .
the flux for elemental groups of the model of hillas is shown in @xcite .
the spectra are constructed with rigidity - dependent knee features at high energies .
reviewing the properties of cosmic rays accelerated in snrs , and using the fluxes as derived by the kascade experiment ( marked as component @xmath109 in ) hillas finds that the obtained all - particle flux ( dashed line , marked with @xmath355 ) is not sufficient to explain the observed all - particle flux , see @xcite .
hillas proposes a second ( galactic ) component to explain the observed flux at energies above @xmath282 ev , marked as component `` b '' in the figure .
an extragalactic component , marked as @xmath356 , dominates the all - particle spectrum above @xmath51 ev , for details see @xcite . very similar is the model proposed by wibig and wolfendale with a transition at higher energies between @xmath52 and @xmath51 ev @xcite . in this model
the galactic cosmic - ray flux extends to higher energies .
thus , a significant contribution of the extragalactic component is required beyond @xmath52 ev only . another possibility to match
the measured all - particle flux is a significant contribution of ultra - heavy elements ( heavier than iron ) to the all - particle spectrum at energies around @xmath10ev @xcite , as illustrated in ( ) .
the figure shows spectra for elemental groups with nuclear charge numbers as indicated , derived from direct and indirect measurements according to the @xcite .
the sum of all elements is shown as solid line and is compared to the average experimental all - particle flux in the figure . in this approach
the second knee is caused by the fall - off of the heaviest elements with @xmath42 up to 92 .
it is remarkable that the second knee occurs at @xmath357 , the latter being the energy of the first knee . in this scenario
a significant extragalactic contribution is required at energies @xmath358 ev . in the model of berezinsky and collaborators
@xcite , the dip in the all - particle spectrum between @xmath52 and @xmath51 ev , see ( ) , is interpreted as a structure caused by electron - positron pair production on cosmic microwave background photons @xmath359 . assuming a power law injection spectrum with a spectral index between @xmath200 ( without cosmological source evolution ) and @xmath360 ( with cosmological source evolution ) , the spectrum can be described for @xmath361ev with a proton - dominated composition @xcite .
the shape of the dip is confirmed by data of the akeno , agasa , hires , yakutsk , and fly s eye detectors after energy - rescaling @xcite . below a characteristic energy @xmath362
ev the spectrum flattens and the steeper galactic spectrum becomes dominant at @xmath363 .
the transition energy @xmath364 approximately coincides with the position of the second knee @xmath365 observed in the all - particle spectrum .
the critical energy @xmath69 is determined by the energy @xmath366 ev , where adiabatic and pair - production energy losses are equal .
thus , the position of the second knee is explained in this scenario by proton energy losses on cosmic microwave background photons .
the extragalactic component required in the poly - gonato model is somewhere between scenarios 1 and 2 shown in fig .
[ polygonato ] ( ) .
it should be emphasized that the pair production mechanism requires the primary particles to be dominated ( @xmath367 ) by protons @xcite .
traditionally , the ankle is interpreted as the characteristic signature for the transition between galactic and extragalactic cosmic rays @xcite .
in such a scenario , extragalactic cosmic rays dominate the flux above about @xmath51ev .
this picture of the transition to extragalactic cosmic rays is supported by the pioneering observations of the fly s eye experiment that the composition changes at about @xmath208ev @xcite .
new observations by hires - mia and hires find a rather sharp transition from a heavy to a light composition at much lower energy , @xmath368ev .
it is clear that the hires data are difficult to understand within a model in which naturally heavy elements should dominate the end of the spectrum of galactic cosmic rays just below @xmath51ev .
if one assumes that extragalactic cosmic rays are accelerated in processes qualitatively similar to those in our galaxy then , at injection , the composition of extragalactic cosmic rays should be similar to that of cosmic rays at lower energy .
indeed , model calculations show that a mixed or even predominantly heavy source composition could , after taking propagation effects into account , be compatible with existing data @xcite . on the other hand , the model by berezinsky _
et al . _ predicts a proton - dominated composition at energies as low as @xmath52ev .
one of the advantages of this model is the natural explanation of the energy and the shape of the ankle . to obtain a good description of the ankle , there should not be more than @xmath369% he in the extragalactic cosmic - ray flux @xcite . this could be interpreted as indication for either strong magnetic fields in the accelerating shock fronts or top - down source scenarios , which predict proton - dominated fluxes at not too high an energy .
understanding the nature of the ankle in the cosmic - ray spectrum has direct implications on the spectrum at much higher energy .
for example , if the @xmath370 pair production model is confirmed one can conclude that ( i ) extragalactic cosmic rays are mainly protons , ( ii ) sources are cosmologically distributed , ( iii ) there should be a gzk suppression of the flux , ( iv ) an arrival direction correlation with local sources can be expected , and ( v ) the injection spectrum of extragalactic cosmic rays is rather steep ( @xmath371 ) .
finally it should be noted that the neutrino flux contains complementary information for distinguishing between different model scenarios for the ankle @xcite .
many authors assume extragalactic particles to be nuclei of intermediate to light mass .
the discrepancy of elemental compositions derived from mean depth of shower maximum and electron / muon number measurements , however , makes it impossible to use currently available composition measurements as reliable criteria to disfavor models ( see also discussion in @xcite ) .
this is most strikingly seen in the prototype hires - mia measurements @xcite . whereas the mean depth of maximum data clearly showed the transition to a proton - dominated composition ( qgsjet - based interpretation )
, the muon density at 600 m of the same showers appeared to correspond to primaries as heavy or even heavier than iron .
a pure iron composition is obviously not compatible with data .
the rather wide distribution of @xmath372 can not be described with iron primaries only .
the measurements of the hires experiment and the pierre auger observatory have given evidence for a suppression of the flux at energies exceeding @xmath289 ev @xcite .
the question arises whether this steepening is due to the gzk effect , due to the maximum energy achieved during the acceleration processes , or due to both , a gzk energy loss process and an upper energy limit in the sources .
the support for the existence of the gzk effect is provided by the correlation of the arrival directions with agn , which imply a strong anisotropy of the arrival direction distribution .
the anisotropy appears sharply at an energy of 57 eev . at this energy , the flux measured by
the pierre auger observatory is about 50% lower than expected from a power law extrapolation from lower energies .
thus , there seems to be a connection between the steepening in the spectrum and the agn correlation .
however , if the observed suppression is due to the gzk effect one would expect either a light or rather heavy elemental composition above the gzk threshold .
intermediate mass nuclei are expected to break up very rapidly in interactions with the photons of the 3-k microwave background .
the relative abundance of secondary protons produced during propagation according to a recent model @xcite is displayed in ( ) . observations of the average depth of the shower maximum at the highest energies indicate a mixed composition , see e.g. . the expected flux at earth according to a propagation model of highest energy cosmic rays by allard et al . @xcite is presented in ( ) .
in this article different scenarios for the properties of the sources are discussed , like different elemental compositions and different spectral indices for the energy spectra at the sources .
shows the result assuming a mixed elemental composition at the sources .
the contributions of individual elemental groups to the all - particle spectrum are shown .
the shape of the energy spectrum measured by the pierre auger observatory is well represented by the all - particle spectrum derived from the model .
another important source of information are fluxes of particles produced during the propagation of cosmic rays in the universe .
fluxes of neutrinos and gamma - rays due to the interaction of ultra high - energy cosmic rays with the background photon radiation provide complementary information for discriminating models of uhecrs @xcite . in measured upper limits for the neutrino flux
have been compared to a top - down model @xcite .
the latest auger data disfavor this model . photons of ultra high energy are easier to detect but their energy loss distance is very short .
nevertheless the measurement of the photon fraction in the primary cosmic - ray flux is one of the most promising methods of distinguishing between different source scenarios of extragalactic cosmic rays @xcite .
recent experimental upper limits strongly disfavor predictions of top - down models @xcite , see ( ) . due to the down - cascading of photons in the extragalactic radiation background ,
fluxes of gev photons are also a complementary source of information .
for example , neutrino and photon fluxes of a z - burst model are discussed in ref .
@xcite .
both , the neutrino flux limit by the forte satellite @xcite and the egret diffuse extragalactic photon flux analysis @xcite severely constrain this model . recently , also even more stringent limits were set by the anita experiment @xcite , ruling out most of the parameter space of the z - burst model . in summary
, top - down models are disfavored by recent experimental results of various experiments .
one of the key questions in the field of high - energy astroparticle physics is the understanding of the observed anisotropy above the gzk energy threshold . the angular scale of about @xmath192 would favour protons if the agns within this correlation angle are indeed the sources of uhecrs @xcite .
this would be at variance with the auger data on the mean depths of shower maximum .
most likely , the agn correlation has to be interpreted merely as a signature of anisotropy and correlation with the nearby matter distribution @xcite .
first of all , the rate of misidentification of a potential source along the line of sight of the arrival direction of a cosmic ray is very large .
secondly , the agns within the correlation window are found to be often less powerful seyfert 2 galaxies @xcite .
thirdly , the agns in our cosmological neighborhood are strongly clustered , making it difficult to distinguish between a true agn correlation and a correlation with larger scale structures . instead of assuming a single source in each of the directions of the measured uhecrs
, one can assign most of the arrival directions to about three sources or source regions @xcite .
this model scenario would require rather weak galactic magnetic fields and particles in the mass range up to carbon . in all anisotropy studies even at the highest energies , the galactic magnetic fields play a central role @xcite .
knowing their structure to sufficient detail would , for example , allow the determination of the charge of uhecrs .
other interesting applications are the search for the cosmological compton - getting effect , a 0.6% dipole anisotropy that is expected for cosmological sources @xcite .
a much higher number of uhecrs has to be collected for detailed anisotropy studies of the required sensitivity .
if uhecrs are confirmed to be protons , the data can be used for proton astronomy including studies of energy spectra of individual sources and magnetic field spectroscopy , e.g. @xcite .
this will , of course , require very large - aperture installations as the relevant energy range is just in the gzk suppression region .
there are strong indications for shortcomings in the shower simulations , probably due to limitations of modeling hadronic interactions .
detailed studies of the shower development in the atmosphere have been performed with the kascade multi - detector set - up and interaction models have been improved @xcite . a particularly valuable tool to test high - energy interaction models are correlations between different shower components @xcite . some years ago
several models like sibyll1.6 , dpmjet2.5 , or nexus2 failed to describe the measurements of particular correlations . on the other hand , for modern models like qgsjet01 , sibyll2.1 , or dpmjet 2.55 ,
the kascade measurements are compatible with predictions for various correlations between the electromagnetic , muonic , and hadronic components , i.e. the measurements are bracketed by the extreme assumptions of primary protons and iron nuclei @xcite . in previous analyses pure proton or iron compositions have been assumed as extreme cases .
however , at present , more detailed analyses are performed @xcite .
they take into account the spectra for elemental groups as obtained from investigations of the electromagnetic and muonic components ( as discussed above , see ) .
these investigations reveal deviations between measurements and simulations for the hadronic component of the order of 10% to 20% @xcite .
the situation is similar at very high energy .
for example , the mean @xmath266 of the hires - mia data is not consistent with the measured muon densities of the same events @xcite .
the conclusions on mass composition from haverah park data are different if the time structure of the shower front is used instead of the muon yield that determines the rate of inclined showers .
the auger data indicate that the energy scale derived from surface detector simulation seems to be of the order of 25% higher than that obtained from fluorescence measurements @xcite . in the foreseeable future soft multi - particle production will not be calculable within qcd .
therefore the modeling of cosmic - ray interactions will continue to depend strongly on the input from accelerator experiments that is needed to tune phenomenological particle production models .
measurements at both , fixed target and collider experiments can substantially contribute to reducing the uncertainties of the models and , hence , determine the composition of cosmic rays .
high - energy interactions are very important for the overall shower profile but hadronic multiparticle production is least known in this energy range .
this is illustrated in fig .
[ fig : modelslhc ] in which predictions of hadronic interaction models are shown for proton - proton collisions at @xmath373tev .
the acceptance ranges for different detector components of the cern cms @xcite detector for the lhc @xcite are also shown .
every ultra high - energy air shower contains many sub - showers of lower energy .
for example , the slope of the lateral distribution of particles in a shower is a measure of the mass of the primary particle . even for high - energy showers of @xmath51ev ,
this slope is very sensitive to assumptions on hadronic multiparticle production in the energy range of ten to a few hundred gev @xcite .
the energy distribution of hadronic interactions , in which at least one meson was produced that in turn decayed to a muon that reached sea level , has a maximum in the range between 80 and 200 gev .
most of the interactions are induced by pions ( 70% ) and nucleons ( 20% ) .
in addition to accelerator measurements of hadronic multiparticle production @xcite , measurements and understanding of air shower data at lower energy are very important to tune and validate the used hadronic interaction models @xcite .
the all - particle flux of cosmic rays is reasonably well known up to the highest energies .
recent measurements by the hires and auger collaborations established a gzk - like suppression of the flux at energies exceeding @xmath374 ev . in the knee region the mean mass of cosmic rays is found to increase as function of energy .
the knee is caused by sequential breaks in the spectra of individual elements , starting with the light elements . at present , a rigidity dependence of the cut - off energies for the individual elements is likely but not yet clear beyond doubt . above @xmath40 ev
the situation becomes very uncertain .
almost no data are available in the important energy range where the galactic cosmic - ray component is expected to end ( @xmath375 ev ) , at present there are only limited experimental efforts in this region . at the highest energies ( @xmath376 ev ) ,
several experiments indicate a light to mixed composition with a strong dependence on the model used to describe high - energy interactions in the atmosphere .
large scale anisotropies have been found at low energies , being compatible with the movement of the earth around the sun ( compton getting effect ) . in the knee region ,
the anisotropies disappear , indicating that the rest frame of galactic cosmic rays co - rotates with the galaxy . at ultra - high energies anisotropy measurements
provide independent information on the composition due to deflections in the galactic magnetic fields .
the results of the pierre auger observatory indicate a correlation between the arrival direction of cosmic rays and the super galactic plane .
the energy of the transition from galactic to extragalactic particles is discussed controversially .
the scenarios considered limit the transition energy to the range between @xmath40 and @xmath51 ev .
precise composition measurements in this region will be decisive .
the knowledge about the cosmic - ray composition is presently limited by the uncertainties in the hadronic interaction models used to describe the air shower development .
the ambiguities can not be resolved by cosmic - ray measurements solely .
collaboration with experiments at the lhc and fixed target experiments is mandatory to improve the understanding of multiparticle production in the ( extreme ) forward region , thus providing reliable simulation codes .
independent and complementary information about the origin of high - energy cosmic rays can be obtained by measurements of high - energy gamma rays and neutrinos ( multi messenger approach ) .
recent measurements of tev gamma rays from supernova remnants give strong hints for this object class as sources of galactic cosmic rays .
measurements of neutrino and photon fluxes are very useful to distinguish between different scenarios for the transition from galactic to extragalactic cosmic rays .
they are important to establish a gzk feature beyond doubt and provide composition information at the highest energies .
in the near future new cosmic - ray detectors will provide additional information on the transition from galactic to extragalactic cosmic rays ( kascade - grande @xcite , icetop / icecube @xcite , extensions of auger - south @xcite , extension of the telescope array @xcite ) , the anisotropy of the arrival directions and composition of cosmic rays at the highest energies on the whole sky ( auger north @xcite ) as well as super gzk particles ( euso @xcite , a recovery of the cosmic - ray flux beyond the gzk resonance ) .
a promising new detection technique , the measurement of radio emission from air showers , is presently developed in the pierre auger observatory and the lofar project .
this technique is expected to allow efficient cosmic - ray measurements in future large - aperture experiments . in foreseeable future
no significant improvement of hadronic interaction models is expected .
therefore , the range of direct measurements should be extended to energies approaching the knee .
ideal would be an access - type @xcite experiment in outer space with exposure times of several years .
more experimental efforts are needed in the energy region where the galactic component is expected to end . at ultra - high energies
the anisotropy studies are limited by statistics only .
therefore , large aperture experiments are needed with full - sky coverage such as the pierre auger project ( with observatories in the southern and northern hemisphere ) . |
multiple - input multiple - output ( mimo ) systems that employ non - orthogonal space - time block codes ( stbc ) from cyclic division algebras ( cda ) for arbitrary number of transmit antennas , @xmath0 , are quite attractive because they can simultaneously provide both _ full - rate _
( i.e. , @xmath0 complex symbols per channel use , which is same as in v - blast ) as well as _ full transmit diversity _ @xcite . the @xmath1 golden code is a well known non - orthogonal stbc from cda for 2 transmit antennas @xcite .
high spectral efficiencies of the order of tens of bps / hz can be achieved using large non - orthogonal stbcs .
for example , a @xmath2 stbc from cda has 256 complex symbols in it with 512 real dimensions ; with 16-qam and rate-3/4 turbo code , this system offers a high spectral efficiency of 48 bps / hz . decoding of non - orthogonal stbcs with such large dimensions , however , has been a challenge .
sphere decoder and its low - complexity variants are prohibitively complex for decoding such stbcs with hundreds of dimensions . in this paper
, we present a probabilistic data association ( pda ) based algorithm for decoding large non - orthogonal stbcs from cda . key attractive features of this algorithm are its low - complexity and near - ml performance in systems with large dimensions ( e.g. , hundreds of dimensions ) .
while creating hundreds of dimensions in space alone ( e.g. , v - blast ) requires hundreds of antennas , use of non - orthogonal stbcs from cda can create hundreds of dimensions with just tens of antennas ( space ) and tens of channel uses ( time ) . given that 802.11 smart wifi products with 12 transmit antennas at 2.5 ghz are now commercially available @xcite ( which establishes that issues related to placement of many antennas and rf / if chains can be solved in large aperture communication terminals like set - top boxes / laptops ) , large non - orthogonal stbcs ( e.g. , @xmath2 stbc from cda ) in combination with large dimension near - ml decoding using pda can enable communications at increased spectral efficiencies of the order of tens of bps / hz ( note that current standards achieve only @xmath3 bps / hz using only up to 4 transmit antennas ) .
pda , originally developed for target tracking , is widely used in digital communications @xcite-@xcite .
particularly , pda algorithm is a reduced complexity alternative to the a posteriori probability ( app ) decoder / detector / equalizer
. near - optimal performance has been demonstrated for pda - based multiuser detection in cdma systems @xcite-@xcite .
pda has been used in the detection of v - blast signals with small number of dimensions @xcite-@xcite . to our knowledge
, pda has not been reported for _ decoding non - orthogonal stbcs with hundreds of dimensions _ so far .
our results in this paper can be summarized as follows : * we adapt the pda algorithm for decoding non - orthogonal stbcs with large dimensions . with i.i.d fading and perfect csir
, the algorithm achieves near - siso awgn uncoded ber and near - capacity coded ber ( within about 5 db of the theoretical capacity ) for @xmath4 stbc from cda , 4-qam , rate-3/4 turbo code , and 18 bps / hz . * relaxing the perfect csir assumption , we report results with a training based iterative pda decoding / channel estimation scheme . the iterative scheme is shown to be effective with large coherence times . *
relaxing the i.i.d fading assumption by adopting a spatially correlated mimo channel model ( proposed by gesbert et al in @xcite ) , we show that the performance loss due to spatial correlation is alleviated by using more receive spatial dimensions for a fixed receiver aperture . * finally
, the performance of the pda algorithm is compared with that of the likelihood ascent search ( las ) algorithm we recently presented in @xcite-@xcite .
the pda algorithm is shown to perform better than the las algorithm at low snrs for higher - order qam ( e.g. , 16-qam ) , and in the presence of spatial correlation .
consider a stbc mimo system with multiple transmit and receive antennas .
an @xmath5 stbc is represented by a matrix @xmath6 , where @xmath7 and @xmath8 denote the number of transmit antennas and number of time slots , respectively , and @xmath9 denotes the number of complex data symbols sent in one stbc matrix .
the @xmath10th entry in @xmath11 represents the complex number transmitted from the @xmath12th transmit antenna in the @xmath13th time slot .
the rate of an stbc is @xmath14 .
let @xmath15 and @xmath16 denote the number of receive and transmit antennas , respectively .
let @xmath17 denote the channel gain matrix , where the @xmath10th entry in @xmath18 is the complex channel gain from the @xmath13th transmit antenna to the @xmath12th receive antenna .
we assume that the channel gains remain constant over one stbc matrix and vary ( i.i.d ) from one stbc matrix to the other .
assuming rich scattering , we model the entries of @xmath18 as i.i.d @xmath19 .
the received space - time signal matrix , @xmath20 , can be written as @xmath21 where @xmath22 is the noise matrix at the receiver and its entries are modeled as i.i.d @xmath23 , where @xmath24 is the average energy of the transmitted symbols , and @xmath25 is the average received snr per receive antenna @xcite , and the @xmath10th entry in @xmath26 is the received signal at the @xmath12th receive antenna in the @xmath13th time - slot .
consider linear dispersion stbcs , where @xmath11 can be written in the form @xcite @xmath27 where @xmath28 is the @xmath12th complex data symbol , and @xmath29 is its corresponding weight matrix . the received signal model in ( [ systemmodel ] )
can be written in an equivalent v - blast form as @xmath30 where @xmath31 , @xmath32 , @xmath33 , @xmath34 , @xmath35 whose @xmath12th entry is the data symbol @xmath28 , and @xmath36 whose @xmath12th column is @xmath37 , @xmath38 .
each element of @xmath39 is an @xmath40-pam/@xmath40-qam symbol .
let @xmath41 , @xmath42 , @xmath39 , @xmath43 be decomposed into real and imaginary parts as : @xmath44 further , we define @xmath45 , @xmath46 , @xmath47 , and @xmath48 as @xmath49^t , \\ \hspace{4 mm } { \bf x}_r = [ { \bf x}_i^t \hspace{2 mm } { \bf x}_q^t ] ^t , \hspace{4 mm } { \bf n}_r = [ { \bf n}_i^t \hspace{2 mm } { \bf n}_q^t ] ^t.\end{aligned}\ ] ] .
\hspace{10 mm } ( \mbox{9.a})\ ] ] ] now , ( [ systemmodelvec2 ] ) can be written as @xmath50 henceforth , we work with the real - valued system in ( [ systemmodelreal ] ) . for notational simplicity ,
we drop subscripts @xmath51 in ( [ systemmodelreal ] ) and write @xmath52 where @xmath53 , @xmath54 , @xmath55 , and @xmath56 .
we assume that the channel coefficients are known at the receiver but not at the transmitter .
let @xmath57 denote the @xmath40-pam signal set from which @xmath58 ( @xmath12th entry of @xmath59 ) takes values , @xmath60 .
now , define a @xmath61-dimensional signal space @xmath62 to be the cartesian product of @xmath63 to @xmath64 .
the ml solution is then given by @xmath65 whose complexity is exponential in @xmath9 .
we focus on the detection of square ( i.e. , @xmath66 ) , full - rate ( i.e. , @xmath67 ) , circulant ( where the weight matrices @xmath68 s are permutation type ) , non - orthogonal stbcs from cda @xcite , whose construction for arbitrary number of transmit antennas @xmath7 is given by the matrix in eqn.(9.a ) given at the bottom of this page . in ( 9.a ) , @xmath69 , @xmath70 , and @xmath71 , @xmath72 are the @xmath73 data symbols from a qam alphabet . when @xmath74 , the code in ( 9.a ) is information lossless ( ill ) , and when @xmath75 and @xmath76 , it is of full - diversity and information lossless ( fd - ill ) @xcite .
high spectral efficiencies with large @xmath7 can be achieved using this code construction . however , since these stbcs are non - orthogonal , ml detection gets increasingly impractical for large @xmath7 .
consequently , a key challenge in realizing the benefits of these large stbcs in practice is that of achieving near - ml performance for large @xmath7 at low decoding complexities .
the ber performance results we report in sec .
[ sec4 ] show that the pda - based decoding algorithm we propose in the following section essentially meets this challenge .
in this section , we present the proposed pda - based decoding algorithm for square qam .
the applicability of the algorithm to any rectangular qam is straightforward . in the real - valued system model in ( [ systemmodelii ] ) , each entry of @xmath59 belongs to a @xmath77-pam constellation , where @xmath40 is the size of the original square qam constellation .
let @xmath78 denote the @xmath79 constituent bits of the @xmath12th entry @xmath58 of @xmath59 .
we can write the value of each entry of @xmath59 as a linear combination of its constituent bits as @xmath80 let @xmath81 , defined as @xmath82^t\hspace{-1 mm } , \hspace{-0mm}\ ] ] denote the transmitted bit vector . defining @xmath83 $ ] , we can write @xmath59 as @xmath84 where @xmath85 is the @xmath86 identity matrix . using ( [ b2seq ] ) , we can rewrite ( [ systemmodelii ] ) as @xmath87 where @xmath88 is the effective channel matrix .
our goal is to obtain @xmath89 , an estimate of the @xmath90 vector . for this
, we iteratively update the statistics of each bit of @xmath90 , as described in the following subsection , for a certain number of iterations , and hard decisions are made on the final statistics to get @xmath89 .
the algorithm is iterative in nature , where @xmath91 statistic updates , one for each of the constituent bits , are performed in each iteration .
we start the algorithm by initializing the a priori probabilities as @xmath92 , @xmath93 and @xmath94 . in an iteration ,
the statistics of the bits are updated sequentially , i.e. , the ordered sequence of updates in an iteration is @xmath95 .
the steps involved in each iteration of the algorithm are derived as follows .
the likelihood ratio of bit @xmath96 in an iteration , denoted by @xmath97 , is given by @xmath98 denoting the @xmath99th column of @xmath100 by @xmath101 , we can write ( [ revsystemmodelii ] ) as @xmath102 where @xmath103 is the interference plus noise vector . to calculate @xmath104 ,
we approximate the distribution of @xmath105 to be gaussian , and hence @xmath106 is gaussian conditioned on @xmath96 . since there are @xmath107 terms in the double summation in ( [ ext_1 ] ) , this gaussian approximation gets increasingly accurate for large @xmath0 ( note that @xmath108 ) .
since a gaussian distribution is fully characterized by its mean and covariance , we evaluate the mean and covariance of @xmath106 given @xmath109 and @xmath110 . for notational simplicity ,
let us define @xmath111 and @xmath112 .
it is clear that @xmath113 .
let @xmath114 and @xmath115 , where @xmath116 denotes the expectation operator .
now , from ( [ ext_1 ] ) , we can write @xmath117 as @xmath118 similarly , we can write @xmath119 as @xmath120 next , the @xmath121 covariance matrix @xmath122 of @xmath106 given @xmath96 is given by @xmath123 \nonumber \\ & & \hspace{-7 mm } \big [ { \bf n } + \hspace{-1 mm } \sum_{l=0}^{2k-1 } \hspace{-1 mm } \sum_{{\stackrel{m=0}{m\neq q(i - l)+j}}}^{q-1}\hspace{-3 mm } { \bf h}_{ql+m } ( b_l^{(m ) } - 2p_l^{m+}+ 1)\big]^t\bigg\}. \end{aligned}\ ] ] assuming independence among the constituent bits , we can simplify @xmath122 in ( [ covpm1 ] ) as @xmath124 using the above mean and covariance expressions , we can write the distribution of @xmath106 given @xmath125 as @xmath126 similarly , @xmath127 is given by @xmath128 using ( [ pdf_p1 ] ) and ( [ pdf_m1 ] ) , @xmath129 can be written as using @xmath130 and @xmath131 , @xmath132 is computed using ( [ llr_i_j_k ] ) .
now , using the value of @xmath132 , the statistics of @xmath96 is updated as follows . from ( [ llr_i_j_k ] ) , and using @xmath133 , we have @xmath134 and @xmath135 as an approximation , dropping the conditioning on @xmath106 , @xmath136 and @xmath137 using the above procedure , we update @xmath138 and @xmath139 for all @xmath140 and @xmath94 sequentially .
this completes one iteration of the algorithm ; i.e. , each iteration involves the computation of @xmath130 and equations ( [ exp1 ] ) , ( [ exm1 ] ) , ( [ covpm2 ] ) , ( [ ext_fin1 ] ) , ( [ llr_i_j_k ] ) , ( [ appx1 ] ) , and ( [ appx2 ] ) for all @xmath141 .
the updated values of @xmath138 and @xmath139 in ( [ appx1 ] ) and ( [ appx2 ] ) for all @xmath141 are fed back to the next iteration .
the algorithm terminates after a certain number of such iterations . at the end of the last iteration
, hard decision is made on the final statistics to obtain the bit estimate @xmath142 as @xmath143 if @xmath144 , and @xmath145 otherwise . in coded systems ,
@xmath97 s are fed as soft inputs to the decoder .
the most computationally expensive operation in computing @xmath104 is the evaluation of the inverse of the covariance matrix , @xmath122 , of size @xmath121 which requires @xmath146 complexity , which can be reduced as follows .
define matrix @xmath147 as @xmath148 at the start of the algorithm , with @xmath149 and @xmath150 initialized to 0.5 for all @xmath141 , * d * becomes @xmath151 .
we note that when the statistics of @xmath152 is updated using ( [ appx1 ] ) and ( [ appx2 ] ) , the @xmath147 matrix in ( [ ckdef ] ) also changes .
a straightforward inversion of this updated @xmath147 matrix would require @xmath146 complexity .
however , we can obtain the @xmath153 from the previously available @xmath153 in @xmath154 complexity as follows .
since the statistics of only @xmath96 is updated , the new @xmath147 matrix is just a rank one update of the old @xmath147 matrix .
therefore , using the matrix inversion lemma , the new @xmath153 can be obtained from the old @xmath153 as @xmath155 where @xmath156 where @xmath157 and @xmath158 are the new @xmath159i.e .
, after the update in ( [ appx1])@xmath160 and ( [ appx2 ] ) @xmath160 and old ( before the update ) values , respectively . it can be seen that both the numerator and denominator in the 2nd term on the rhs of ( [ ckcorr ] ) can be computed in @xmath154 complexity .
therefore , the computation of the new @xmath153 using the old @xmath153 can be done in @xmath154 complexity .
_ computation of @xmath161 : _ using ( [ ckdef ] ) and ( [ covpm2 ] ) , we can write @xmath122 in terms of @xmath147 as @xmath162 we can compute @xmath163 from @xmath153 at a reduced complexity using the matrix inversion lemma , which states that substituting @xmath164 , @xmath165 , @xmath166 , and @xmath167 in ( [ cinv_cmp2 ] ) , we get @xmath168 which can be computed in @xmath154 complexity . _
computation of @xmath117 and @xmath119 : _ computation of @xmath104 involves the computation of @xmath117 and @xmath119 also . from ( [ exm1 ] ) , it is clear that @xmath119 can be computed from @xmath117 with a computational overhead of only @xmath169 . from ( [ exp1 ] ) , it can be seen that computing @xmath117 would require @xmath170 complexity . however , this complexity can be reduced as follows .
define vector @xmath171 as @xmath172 using ( [ exp1 ] ) and ( [ ekdef ] ) , we can write @xmath173 * u * can be computed iteratively at @xmath169 complexity as follows .
when the statistics of @xmath152 is updated , we can obtain the new @xmath171 from the old @xmath171 as @xmath174 whose complexity is @xmath169 .
hence , the computation of @xmath117 in ( [ ekdef ] ) and @xmath119 in ( [ ekdef2 ] ) needs @xmath169 complexity .
the listing of the proposed pda algorithm is summarized in the table - i in the next page . ' '' '' table - i : proposed pda - based algorithm listing ' '' '' \1 .
@xmath175 , @xmath176 , @xmath177 .
\2 . @xmath178 .
@xmath179 : number of iterations \4 .
@xmath180 ; @xmath181 is the iteration number \5 . for @xmath12 = 0 to @xmath182
\6 . for @xmath13 = 0 to @xmath183 \7 .
@xmath184 \8 .
@xmath185 \9 .
@xmath186 \10 .
@xmath187 \11 .
@xmath188 , @xmath189 \12 .
@xmath190 \13 .
@xmath191 \14 .
@xmath192 , @xmath193 \15 .
@xmath194 \16 .
@xmath195 \17 .
@xmath196 \18 .
end ; end of for loop starting at line 5 \19 .
if ( @xmath197 ) goto line 21 \20 .
@xmath198 , goto line 5 \21 .
@xmath199 @xmath200 \22 .
@xmath201 \23 . terminate ' '' ''
we need to compute @xmath202 at the start of the algorithm .
this requires @xmath203 complexity .
so the computation of the initial @xmath153 in line 2 requires @xmath204 .
based on the complexity reduction in sec .
[ complex ] , the complexity in updating the statistics of one constituent bit ( lines 7 to 17 ) is @xmath154 .
so , the complexity for the update of all the @xmath91 constituent bits in an iteration is @xmath203 .
since the number of iterations is fixed , the overall complexity of the algorithm is @xmath204 .
for @xmath205 , since there are @xmath9 symbols per stbc and @xmath206 bits per symbol , the overall complexity per bit is @xmath207 .
in this section , we present the simulated uncoded / coded ber of the pda algorithm in decoding non - orthogonal stbcs from cda ) and ill ( @xmath74 ) stbcs with pda decoding were almost the same . here , we present the performance of ill stbcs . ] .
number of iterations in the pda algorithm is set to @xmath208 in all the simulations . _
pda versus las performance with 4-qam : _ in fig .
[ fig1 ] , we plot the uncoded ber of the pda algorithm as a function of average received snr per rx antenna , @xmath25 , in decoding @xmath209 , @xmath210 , @xmath2 stbcs from cda with @xmath205 and 4-qam . perfect channel state information at the receiver ( csir ) and i.i.d fading are assumed .
for the same settings , the performance of the las algorithm in @xcite-@xcite with mmse initial vector are also plotted for comparison . from fig .
[ fig1 ] , it is seen that * the ber performance of pda algorithm improves and approaches siso awgn performance as @xmath205 is increased ; e.g. , performance close to within about 1 db from siso awgn performance is achieved at @xmath211 uncoded ber in decoding @xmath2 stbc from cda having 512 real dimensions , and this illustrates the ability of the pda algorithm to achieve excellent performance at low complexities in large non - orthogonal stbc mimo . * with 4-qam ,
pda and las algorithms achieve almost the same performance . _
pda versus las performance with 16-qam : _ figure [ fig2 ] presents an uncoded ber comparison between pda and las algorithms for @xmath2 stbc from cda with @xmath212 and 16-qam under perfect csir and i.i.d fading .
it can be seen that the pda algorithm performs better at low snrs than the las algorithm .
for example , with @xmath210 and @xmath2 stbcs , at low snrs ( e.g. , @xmath213 db for @xmath2 stbc ) , pda algorithm performs better by about 1 db compared to las algorithm at @xmath214 uncoded ber . _ turbo coded ber performance of pda : _ figure [ fig3 ] shows the rate-3/4 turbo coded ber of the pda algorithm under perfect csir and i.i.d fading for @xmath4 ill stbc with @xmath215 and 4-qam , which corresponds to a spectral efficiency of 18 bps / hz . the theoretical minimum snr required to achieve 18 bps / hz spectral efficiency on a @xmath215 mimo channel with perfect csir and i.i.d fading is 4.3 db ( obtained through simulation of the ergodic capacity formula @xcite ) . from fig .
[ fig3 ] , it is seen that the pda algorithm is able to achieve vertical fall in coded ber within about 5 db from the theoretical minimum snr , which is a good nearness to capacity performance .
we relax the perfect csir assumption by considering a training based iterative pda decoding / channel estimation scheme .
transmission is carried out in frames , where one @xmath216 pilot matrix ( for training purposes ) followed by @xmath217 data stbc matrices are sent in each frame as shown in fig .
one frame length , @xmath218 , ( taken to be the channel coherence time ) is @xmath219 channel uses .
the proposed scheme works as follows @xcite : @xmath220 obtain an mmse estimate of the channel matrix during the pilot phase , @xmath221 use the estimated channel matrix to decode the data stbc matrices using pda algorithm , and @xmath222 iterate between channel estimation and pda decoding for a certain number of times . for @xmath4 stbc from cda , in addition to perfect csir performance , fig .
[ fig3 ] also shows the performance with csir estimated using the proposed iterative decoding / channel estimation scheme for @xmath223 and @xmath224 .
2 iterations between decoding and channel estimation are used . with @xmath224 ( which corresponds to large coherence times , i.e. , slow fading ) the ber and bps / hz with estimated csir get closer to those with perfect csir .
_ effect of spatial mimo correlation : _ in figs .
[ fig1 ] to [ fig3 ] , we assumed i.i.d fading . but spatial correlation at transmit / receive antennas and the structure of scattering and propagation environment can affect the rank structure of the mimo channel resulting in degraded performance @xcite .
we relaxed the i.i.d . fading assumption by considering the correlated mimo channel model in @xcite , which takes into account carrier frequency ( @xmath225 ) , spacing between antenna elements ( @xmath226 ) , distance between tx and rx antennas ( @xmath227 ) , and scattering environment . in fig .
[ fig5 ] , we plot the ber of the pda algorithm in decoding @xmath4 stbc from cda with perfect csir in @xmath220 i.i.d . fading , and @xmath221 correlated mimo fading model in @xcite . it is seen that , compared to i.i.d fading , there is a loss in diversity order in spatial correlation for @xmath228 ; further , use of more rx antennas ( @xmath229 ) alleviates this loss in performance .
we can decode _ perfect codes _
@xcite,@xcite of large dimensions also using the proposed pda algorithm . | non - orthogonal space - time block codes ( stbc ) from cyclic division algebras ( cda ) having large dimensions are attractive because they can simultaneously achieve both high spectral efficiencies ( same spectral efficiency as in v - blast for a given number of transmit antennas ) _ as well as _ full transmit diversity .
decoding of non - orthogonal stbcs with hundreds of dimensions has been a challenge . in this paper , we present a probabilistic data association ( pda ) based algorithm for decoding non - orthogonal stbcs with large dimensions .
our simulation results show that the proposed pda - based algorithm achieves near siso awgn uncoded ber as well as near - capacity coded ber ( within about 5 db of the theoretical capacity ) for large non - orthogonal stbcs from cda .
we study the effect of spatial correlation on the ber , and show that the performance loss due to spatial correlation can be alleviated by providing more receive spatial dimensions .
we report good ber performance when a training - based iterative decoding / channel estimation is used ( instead of assuming perfect channel knowledge ) in channels with large coherence times . a comparison of the performances of the pda algorithm and the likelihood ascent search ( las ) algorithm ( reported in our recent work ) is also presented . |
recent observations of high - redshift ( @xmath19 ) type ia supernovae ( sne ia ) provide the backbone of the body of evidence that we live in an accelerating universe whose content is dominated by vacuum energy ( riess et al .
1998 ; perlmutter et al . 1999 ) .
the observational evidence for an accelerating universe is that high-@xmath20 sne ia are @xmath2130% dimmer than expected in an open universe ( i.e. , @xmath22=0.3 , @xmath23 ) .
the two most likely sources to obscure distant sne ia and affect their interpretation are dust and evolution .
the tell - tale signature of extinction by galactic - type dust , reddening , has not been detected in the amount required to provide @xmath24 mag for high-@xmath20 sne ia ( riess et al .
1998 ; perlmutter et al . 1999 ) . yet
the cosmological implications of the observed faintness of high-@xmath20 sne ia are so exotic as to merit the consideration of dust with more unusual properties .
a physical model of dust composed of larger grains ( @xmath25 0.1 @xmath26 m ) has been posited by aguirre ( 1999a , b ) to provide a non - cosmological source of extinction with less reddening .
component of this so - called `` gray '' dust , if neglected , would add too much dispersion to be consistent with the observed luminosities ( riess et al .
however , aguirre ( 1999a , b ) has shown that a uniformly distributed component of _ intergalactic _ gray dust with a mass density of @xmath27 could explain the faintness of high-@xmath20 sne ia without detectable reddening and without overproducing the far - infrared ( far - ir ) background .
previous data do not rule out this possibility .
indeed , significant _ interstellar _ extinction in the hosts of high-@xmath20 sne ia is still favored by some ( totani & kobayashi 1999 ) .
rest - frame evolution is the other potential pitfall in using high-@xmath20 sne ia to measure the cosmological parameters .
the lack of a complete theoretical model of sne ia including the identification of their progenitor systems makes it difficult to access the expected evolution between @xmath28 and 0.5 ( livio 1999 ; umeda et al .
1999 ; hflich , wheeler , & thielemann 1998 ) . an impressive degree of similarity has been observed between the spectral and photometric properties of nearby and high-@xmath20 sne ia ( schmidt et al .
1998 ; perlmutter et al . 1998 , 1999 ; riess et al .
1998 ; filippenko et al . 2000
see also riess et al . 1999c ; drell , loredo , & wassermann 1999 ) . however , it is not known what kind or degree of change in the observable properties of sne ia would indicate a change in the expected peak luminosity by 30% .
for that reason it has been necessary to compare a wide range of observable characteristics of nearby and high-@xmath20 sne ia to search for a complement to a luminosity evolution . near - ir
observations of high-@xmath20 sne ia can provide constraints on both sources of cosmological contamination . a physical model of gray intergalactic dust , such as that proposed by aguirre ( 1999a , b )
, still induces some reddening of sn light which can be detected in the wavelength range between optical and near - ir light .
in addition , near - ir observations provide a view of the behavior of high - redshift sne ia in a window previously unexplored .
specifically , normal , nearby sne ia exhibit a second infrared maximum about a month after the initial peak .
we can increase our confidence that high-@xmath20 sne ia have not evolved by observing this second maximum ; its absence would indicate a change in the physics sne ia across redshift with potentially important cosmological consequences .
we obtained ground - based @xmath29-band and space - based optical observations of sn 1999q ( @xmath2 ) to initiate a study of the systematic effects of dust and evolution on high-@xmath20 sne ia . in
2 we descibe our observations , in 3 their analysis , and in 4 their interpretation .
our high-@xmath20 supernova search team ( hzt ) has an ongoing program to discover and monitor high - redshift sne ia ( schmidt et al .
sn 1999q was discovered on jan 18 , 1999 using the ctio 4-m blanco telescope with the bernstein - tyson camera ( http://www.astro.lsa.umich.edu/btc/btc.html ) as part of a 3-night program to search for high-@xmath20 sne ia using well - established methods .
high signal - to - noise ratio spectra of sn 1999q obtained with the keck - ii telescope indicated that this was a typical sn ia at @xmath20=0.46 shortly before @xmath6-band maximum ( garnavich et al . 1999 ; filippenko et al . 2000 ) .
rest - frame @xmath6 and @xmath30-band photometry using custom filters was obtained for sn 1999q from observatories around the world .
the _ hubble space telescope _
( _ hst _ ) monitored the @xmath6 and @xmath30 light curves of sn 1999q from @xmath21 10 to 35 days after @xmath6 maximum ( rest - frame ) using the wfpc2 and the f675w and f814w filters in the course of 6 epochs .
combined , these data provide excellent coverage of the rest - frame @xmath6 and @xmath30 light curves from a few days before to 60 days after @xmath6 maximum ( in the rest frame ; clocchiatti et al .
2000 ) .
in addition , the sn was observed in the near - ir ( @xmath29-band ) for 5 epochs between 5 and 45 days after @xmath6 maximum ( in the rest frame ) .
the first observation employed the european southern observatory s 3.5-m new technology telescope equipped with the son of isaac ( sofi ) infrared camera spectrograph ( http://www.ls.eso.org ) ; subsequent observations used the keck ii telescope equipped with the near - infrared camera ( nirc ; matthews & soifer 1994 ) . due to the high sky brightness in the ir , many dithered , short images were obtained and combined to avoid saturating the detector .
care was taken to maintain a detected sky flux level of @xmath2110,000 counts , a regime where both sofi and nirc exhibit less than 0.5% non - linearity ( http://www.ls.eso.org ; http://www.keck.hawaii.edu ) . using the procedure described by garnavich et al .
( 1998 ) , we subtracted an empirical point - spread function ( psf ) scaled to the brightness of the sn from each _ hst _ observation . a coadded image of total length 7200 seconds in both f675w and f814w revealed no trace of host galaxy light to more than 5 mag below the peak brightness of the supernova ( i.e. , @xmath31 27 ) .
the host of sn 1999q is likely to be intrinsically faint or of very low surface brightness similar to the host of sn 1997ck ( garnavich et al .
1998 ) . due to the negligible contribution of host galaxy light to the images
, we have made no correction for contamination to the measured supernova light .
assuming that the restframe @xmath32 color of the host galaxy is no redder than that of early - k - type dwarfs ( @xmath32=1.0 for k0 ) , this same practice is well justified for our measurements of sn light in the @xmath29 band .
we conservatively adopt a systematic uncertainty of 0.03 mag in the sn photometry ( and 0.02 mag uncertainty in the colors ) to account for any remaining bias . the procedure described by schmidt et al .
( 1998 ) , garnavich et al .
( 1998 ) , and riess et al .
( 1998 ) was followed to calibrate the measured magnitudes of sn 1999q on the johnson @xmath6 and @xmath30 passband system .
similar steps were performed to calibrate the observed @xmath29-band magnitudes of sn 1999q onto the rest - frame cousins @xmath1-band system , though a few exceptions are noted here .
on three photometric nights we observed the secondary near - ir standards of persson et al .
because these secondary standards are solar analogues ( 0.4 @xmath33 @xmath34 @xmath33 0.8 ) , one can transform these stars from the `` persson system '' to that of nirc or sofi by calculating the photometric difference of spectrophotometry of the sun between these systems .
we found these differences to be quite small ( @xmath330.02 mag ) and this correction negligible . in practice the true transmission curve in @xmath29
is dictated by the natural opacity of atmospheric h@xmath35o and nightly variations are generally larger than differences between different facility @xmath29-passbands .
for this reason we observed secondary standards in close temporal proximity to the sn field . due to the inherent non - linearity of airmass extinction corrections in @xmath29 , the field of sn 1999q
was observed at airmasses within 0.05 of the persson et al .
( 1998 ) standards to avoid the need for airmass corrections . assuming typical @xmath29-band extinction of 0.1 mag per airmass ( krisciunas et al .
1987 ) , errors of @xmath21 0.005 mag are introduced without explicit extinction corrections .
cross - band @xmath36-corrections ( kim , goobar , & perlmutter 1996 ) were calculated using spectrophotometry of sn 1994d ( which extend redward of 9100 ; richmond et al .
1995 ) to transform the @xmath29-band magnitudes of sn 1999q to the cousins @xmath1-band . at the redshift of sn 1999q ( @xmath2 ) , observed @xmath29-band light is an excellent match to rest - frame cousins @xmath1 , and the @xmath36-correction was determined to be @xmath180.93 @xmath37 mag with no apparent dependence on supernova phase or color .
the rest - frame @xmath1 photometry of sn 1999q is given in table 1 .
of fundamental importance is our ability to correctly transform the observed @xmath29-band photometry to restframe @xmath1-band .
schmidt et al . ( 1998 ) discusses the derivations of optical zeropoints from numerous spectrophotometric stars which also have ubvri photometry .
applying the same spectrophotometry to cross - band @xmath36-corrections removes the dependence of the transformed photometry on the observed band s zeropoint .
unfortunately , this type of data is not available for the j - band .
we have therefore calibrated our @xmath29-band data using persson et al .
standards , who are fundamentally tied to the elias ( 1982 ) standards and we adopt an appropriate @xmath29-band zeropoint uncertainty of @xmath38=0.05 mag .
a unique photometric signature of typical sne ia is a resurrection of the luminosity at infrared wavelengths about a month after the initial maximum .
this feature is present ( with the most exquisite photometry ) in the @xmath30 band , grows into a `` shoulder '' in @xmath39 , and increases to a second local maximum in @xmath1 ( ford et al .
1993 ; suntzeff 1996 ; hamuy et al . 1996a ; riess et al .
this second maximum is also readily apparent at near - ir wavelengths ( @xmath40 and @xmath36 ; elias et al .
1981 ; jha et al .
no other type of sn exhibits this feature .
the secondary maximum is thought to result from the escape of radiation from the core of the supernova at long wavelengths .
resonant scattering from lines is the dominant source of opacity . at short wavelengths ( i.e. , @xmath33 5000 ) line blanketing traps radiation ; the resonance lines at longer wavelengths are fewer and further between providing escape routes for the trapped radiation ( spyromilio , pinto , & eastman 1994 ) .
wheeler et al.(1998 ) argue that this effect in itself would not explain the nonmonotic behaviour of the j - band light curve .
no model as yet fully explains the shape and timing of the infrared light curves of sne . the location and strength of the second @xmath1-band maximum is a diagnostic of the intrinsic luminosity of sne ia ( riess , press , & kirshner 1996 ; hamuy et al .
sne ia with typical peak luminosity ( i.e. , @xmath41 mag ) crest again in @xmath1 about 30 days after @xmath6 maximum .
dimmer sne ia reach their second peak earlier .
for example , sn 1992bo was @xmath210.5 mag fainter than a typical sn ia and reached its second peak in @xmath1 at @xmath2120 days after @xmath6 maximum ( maza et al .
for very subluminous sne ia this second maximum is completely absent , merging into the phase of the initial decline ( e.g. , sn 1991bg ; filippenko et al .
the physics detailing the formation of this feature also indicates that its magnitude and timing are sensitive to explosion parameters ( e.g. , ejecta composition ) which determine the peak luminosity ( spyromilio , pinto , & eastman 1994 ) . in figure 1 the relative rest - frame @xmath1-band magnitudes of sn 1999q ( @xmath2 )
are compared to a luminosity sequence of nearby sne ia .
the phase of the observations of sn 1999q was determined by multicolor light - curve shape ( mlcs ; riess et al .
1996 ; riess et al .
1998 ) fits to the @xmath6 and @xmath30 light curves and have an uncertainty ( @xmath42 ) of less than 2 days .
the observation times were also corrected for @xmath43 time dilation ( leibundgut et al .
1996 ; goldhaber et al . 1997 ; riess et al .
although the precision and sampling of the rest - frame @xmath1-band data are not high , they are sufficient to indicate that this high-@xmath20 sn ia retains significant luminosity at @xmath2130 days after @xmath6 maximum , consistent with the phase of the second @xmath1-band peak of typical sne ia and inconsistent with either very subluminous or moderately subluminous sne ia . using the @xmath6 and @xmath30 light - curve shapes of sn 1999q as a luminosity indicator
, we find its distance modulus to be @xmath44=42.67@xmath45 mag , consistent with previous sne ia favoring a cosmological constant ( riess et al .
if instead we consider the shape of the @xmath1-band light curve as an independent luminosity indicator , we find that this high-@xmath20 sn ia ( and presumably other high-@xmath20 sne ia ) is not consistent with being subluminous by 0.5 - 0.6 mag as needed to indicate a universe closed by ordinary matter .
more precise data will be needed to differentiate between an open and @xmath46-dominated universe solely on the basis of @xmath1-band light curve shapes . to measure the @xmath47 colors of sn 1999q we used the mlcs fits to the @xmath6 light curve to determine the expected @xmath6 magnitudes at the time of the ir observations .
due to the exquisite _ hst _ photometry in rest - frame @xmath6 , this process adds little uncertainty to the @xmath47 magnitudes . the milky way ( mw ) dust maps from schlegel , finkbeiner , & davis ( 1998 ) predict a reddening of @xmath48=0.021 mag in the direction of sn 1999q .
we subtracted the expected galactic reddening of the rest - frame @xmath47 light ( observed as @xmath49 ) of sn 1999q , 0.037 mag , from the measured colors .
any remaining reddening results from extragalactic sources . in figure 2 the measured @xmath47 magnitudes of sn 1999q are compared to a custom @xmath47 curve predicted from the mlcs fits to the @xmath6 and @xmath30 light - curve shapes ( riess et al .
1996 , 1998 ) .
the smaller uncertainties shown here result from photon statistics and were determined empirically ( schmidt et al . 1998 ) .
a significant , additional source of uncertainty is the intrinsic dispersion of sne ia @xmath47 colors around their custom mlcs model .
this intrinsic dispersion is determined empirically by measuring the variance of 30 nearby sne ia around their mlcs fits ( riess et al .
1996 , 1998 ) and varies from 0.1 to 0.3 mag depending on the sn ia age .
although the observed residuals from the model prediction are correlated for time separations of less than 3 days , correlated errors are insignificant for the larger differences in time between the observations of sn 1999q .
the larger uncertainties shown in figure 2 for the @xmath47 photometry of sn 1999q include the intrinsic uncertainties .
the measured @xmath3 for sn 1999q is @xmath180.09@xmath50 mag .
the error includes the systematic uncertainties of the @xmath36-corrections and the zeropoint of the @xmath29-band system , although the dominant sources of error are the photometry noise and the intrinsic dispersion in sn ia @xmath47 colors .
this value is consistent with no reddening of this high-@xmath20 sn ia .
if galactic - type dust rather than a cosmological constant were the sole reason that @xmath51 sne ia are 30% fainter than expected for an open universe ( i.e. , @xmath52 ) , then the @xmath3 of sn 1999q should be 0.25 mag ( savage & mathis 1979 ) .
this alternative to an accelerating universe ( see totani & kobayashi 1999 ) is inconsistent with the data at the 99.9% confidence level ( 3.4@xmath4 ) .
the reddening required for the sne ia data to be consistent with a universe closed by matter is ruled out at the @xmath2599.99% ( 5.1@xmath4 ) confidence level . despite the low precision of this data set , the wavelength range of the @xmath47 colors results in the ability to rule out extinction by galactic - type dust from sn 1999q alone with similar confidence as from the entire set of @xmath34 color data of riess et al .
( 1998 ) and perlmutter et al . ( 1999 ) .
the reduced amount of reddening by `` gray '' dust grains ( i.e. , @xmath25 0.1 @xmath26 m ) as proposed by aguirre ( 1999a , b ) is more difficult to detect .
the amount of gray dust needed to supplant the cosmological constant as the cause of the dimming of high-@xmath20 sne ia would result in an @xmath3=0.17 or 0.14 mag for a composition of graphite or graphite / silicate , respectively ( aguirre 1999a , b ) .
these possibilities are moderately inconsistent with the data at the 99.0% ( 2.6@xmath4 ) and 97.7% ( 2.3@xmath4 ) confidence levels , respectively .
the reddening provided by enough of such dust to change the cosmological forecast to favor a universe closed by matter is ruled out at the 99.97%(3.7@xmath4 ) and 99.90%(3.3@xmath4 ) confidence levels , respectively .
the weakest constraint comes from assuming the smallest amount of the grayest type of dust which is consistent at the 68% ( 1@xmath4 ) confidence level with an open universe ( i.e. , @xmath53=0.2 mag ) .
this dust is inconsistent with the data at the 94% ( 1.9 @xmath4 ) confidence level ( although the true inconsistency of this scenario is derived from the product of the two individual likelihoods , i.e. , 98% or 2.3@xmath4 ) .
although these results disfavor the existence of the proposed levels of gray dust , more data are needed to strengthen this important test . because it is difficult to assess all sources of uncertainty in our model for the sn ia @xmath47 color evolution , we also performed a monte carlo simulation of the measurement of @xmath3 for sn 1999q . using all nearby sne
ia which are not spectroscopically peculiar ( see branch , fisher , & nugent 1993 ) nor photometrically extreme ( @xmath54 ; phillips 1993 ) and whose @xmath47 colors were well observed , we generated a standard , unreddened @xmath47 template curve using individual reddening estimates from phillips et al .
( 1999 ) and schlegel et al .
( 1998 ) for nearby sne ia .
we then randomly selected five observations from a random member of the sample and perturbed the observations to match the photometric noise in the sn 1999q observations . from 10,000 such synthetic measurements we generated a distribution of measured @xmath3
whose shape should match the probability density function for the single @xmath3 measurement of sn 1999q . compared to the @xmath47 template curve ,
sn 1999q has an @xmath3 = @xmath180.12 mag .
the distribution of synthetic @xmath3 values is asymmetric and implies an uncertainty in the measurement for sn 1999q of @xmath55 mag and @xmath56 mag ( including the systematic uncertainties from @xmath36-corrections and the @xmath29-band zeropoint ) .
the results are consistent with no extragalactic reddening and inconsistent with galactic and gray dust reddening at nearly identical ( though marginally higher ) confidence levels as the mlcs fits .
the strength of this method is that it samples real sn ia data in the same manner as the observations of sn 1999q and therefore incorporates the intrinsic and correlated uncertainties in the @xmath47 colors of sne ia .
two teams have independently concluded that the observed faintness of high-@xmath20 sne ia indicates that the expansion of the universe is accelerating and that dark energy dominates the energy density of the universe ( riess et al .
1998 ; perlmutter et al . 1999 ) .
however , as a well - known adage reminds us , `` extraordinary claims require extraordinary evidence . ''
alternative explanations such as evolution in supernova luminosities or dust are no more exotic than a cosmological constant and must be rigorously tested .
a @xmath2130% opacity of visual light by dust is the best quantified and therefore most readily testable alternative to a cosmological constant ( aguirre 1999a , b ; totani & kobayashi 1999 ) .
measurements of @xmath34 colors indicate that this quantity of galactic - type dust is not obscuring high-@xmath20 sne ia ( riess et al .
1998 ; perlmutter et al . 1999 ) and the @xmath47 observations presented here bolster this evidence .
however , observations of neither sne ia nor other astrophysical objects previously ruled out a similar opacity by intergalactic gray dust ( aguirre 1999a , b ) . the observations presented here
do disfavor a gray intergalactic medium providing this opacity , but additional data are needed to strengthen these conclusions .
indeed , a more precise measurement of @xmath3 or @xmath57 could constrain either the total optical depth of dust in the intergalactic medium or alternately push the minimum size of such grains into an unphysical domain ( aguirre 1999a , b ) .
it may even be possible to use such measurements to constrain the contribution to the far - ir background by emission from the intergalactic medium ( aguirre 1999a , b ) .
measurements of gravitational lens systems have also been used as a probe of the high-@xmath20 extinction law and disfavor significant interstellar gray dust ( falco et al .
1999 ; mcleod 1999-except in molecular clouds ) .
to date , our inability to formulate a complete theoretical description of sne ia makes it impossible to either predict the degree of expected luminosity evolution between @xmath28 and 0.5 or to identify an observation which would conclusively determine whether the luminosity of sne ia are evolving ( but see hoeflich , thielemann & wheeler 1998 ) .
an empirical recourse is to compare all observable properties of nearby and high-@xmath20 sne ia with the assumption that if the luminosity of sne ia has evolved by @xmath2130% other altered characteristics of the explosion would be visible as well .
the detection of such a change would cast doubt on the reliability of the luminosity distances from high-@xmath20 sne ia . a continued failure to measure any difference between sn ia near and
far would increase our confidence ( though never prove ) that evolution does not contaminate the cosmological measurements from high-@xmath20 sne ia .
comparisons of high signal - to - noise ratio spectra of nearby and high-@xmath20 sne ia have revealed remarkable similarity ( riess et al .
1998 ; perlmutter et al . 1998 , 1999 ; filippenko et al .
2000 ) . because the spectrum provides a detailed record of the conditions of the supernova in the atmosphere ( i.e. , temperature , abundances , and ejecta velocities ) ,
spectral comparisons are expected to be particularly meaningful probes of evolution .
further , comparisons of time sequences of spectra reveal no apparent differences as the photosphere recedes in mass ( filippenko et al .
2000 ) , indicating that the striking resemblence between distant and nearby sne ia is not merely superficial , but endures at deeper layers .
however , these comparisons still require the rigor of a quantitative approach to determine whether or not the two samples are statistically consistent .
the distributions of light - curve shapes at high and low redshift are statistically consistent ( riess et al .
1998 ; perlmutter et al . 1999 ) .
however , drell et al . ( 1999 ) have noted that different approaches to quantifying the shape of the light curves may not be statistically consistent , so more attention needs to be focused on these light - curve shape comparisons .
the colors of pre - nebular supernovae should provide a useful probe of luminosity evolution , indicating changes in the approximate temperature and hence the thermal output of the explosion .
the @xmath34 colors of nearby and high-@xmath20 sne ia were found to be consistent by perlmutter et al .
the same consistency was found here for the @xmath47 colors .
however , neither this work nor the @xmath34 color measurements by riess et al .
( 1998 ) can rule out the possibility that high-@xmath20 sne ia could be excessively blue ( falco et al .
1999 ) ; more data are needed to explore this possibility . the time interval between explosion and maximum light ( i.e. , the risetime )
is expected to be a useful probe of the ejecta opacity and the distribution of @xmath58ni .
the initial comparison of the risetime of nearby ( riess et al .
1999b ) and high - redshift sne ia ( goldhaber 1998 ; groom 1998 ) found an apparent inconsistency ( riess et al . 1999c ) .
further analysis of the scp high - redshift data by aldering , nugent , & knop ( 2000 ) , however , concludes that the high - redshift risetime was somewhat larger and far more uncertain than found by groom ( 1998 ) and that the remaining difference in the risetime could be no more than a @xmath212.0 @xmath4 chance occurence .
the weight of the evidence suggests no significant evolution of the observed sne ia , but more observations are needed to allay remaining reasonable doubts .
perhaps the best indication that sne ia provide reliable distances at high redshifts comes from sne ia in nearby early - type and late - type galaxies .
these galaxies span a larger range of metallicity , stellar age , and interstellar environments than is expected to occur for galaxies back to @xmath59 . yet after correction for the light - curve - shape / luminosity relationship and extinction , no significant hubble diagram residuals are seen which correlate with host galaxy morphology .
this suggests that our distance estimates are insensitive to variations in the supernova progenitor environment ( schmidt et al .
however , the evidence remains circumstantial and does not rule out the possibility that a characteristic of all progenitors of nearby sne ia differs for high-@xmath20 sne ia .
further observations , especially those in the near - ir bands , can better constrain the potential contamination of the cosmological conclusions from sne ia posed by dust and evolution . further , rest - frame @xmath1 band measurements of nearby sne ia show less dispersion in intrinsic luminosity and extinction making this an attractive band for future observations ( hamuy et al .
measurements of sne ia at @xmath60 should even discriminate between the effects of a cosmological constant and those of a monotonically increasing , but unidentified systematic errors ( filippenko & riess 1999 ) .
continuing studies of high-@xmath20 sne ia should ultimately provide the extraordinary evidence required to accept ( or refute ) the accelerating universe .
we wish to thank alex athey and s. elizabeth turner for their help in the supernova search at ctio .
we have benefited from helpful discussions with anthony aguirre , stefano casertano , and ed moran .
we thank the following for their observations or for attempts to obtain useful data a. dey , w. danchi , s. r. kulkarni , & p. tuthill .
the work at u.c .
berkeley was supported by the miller institute for basic research in science , by nsf grant ast-9417213 , and by grant go-7505 from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas5 - 26555 .
support for ac was provided by the national science foundation through grant # gf-1001 - 95 from aura , inc .
, under nsf cooperative agreement ast-8947990 and ast-9617036 , and from fundacin antorchas argentina under project a-13313 .
this work was supported at harvard university through nsf grants ast-9221648 , ast-9528899 , and an nsf graduate research fellowship .
cs acknowledges the generous support of the packard foundation and the seaver institute .
based in part on observations collected at the european southern observatory , chile , under proposal 62.h-0324 . * fig 1.*-relative rest - frame @xmath1-band light curve of a high - redshift sn ia , sn 1999q ( @xmath2 ) , and a sequence of 3 nearby sne ia with different light - curve shapes and peak luminosities . the light curve of sn 1999q is consistent with that of typical nearby sne ia which reach their second maxima at @xmath2130 days after @xmath6 maximum ( e.g. , sn 1995d ; diamonds , riess et al .
the data are inconsistent with sne ia which are subluminous at visual peak by @xmath210.5 mag and reach the second maximum at @xmath2120 days after @xmath6 maximum ( e.g. , sn 1992bo ; asterisks , maza et al . 1994 ) .
sn 1999q bares no resemblence to the rapid decline ( without a second maximum ) of very subluminous sne ia ( e.g. , sn 1991bg ; circles , filippenko et al .
1992b ) .
* fig 2.*-the color evolution , @xmath47 , and the color excess , @xmath3 , of a high - redshift sn ia , sn 1999q , compared to the custom mlcs template curve with no dust and enough dust ( of either galactic - type or grayer ) to nullify the cosmological constant .
the smaller error bars are from photometry noise ; the larger error bars include all sources of uncertainty such as intrinsic dispersion of sn ia @xmath47 color , @xmath36-corrections , and photometry zeropoints .
the data for sn 1999q are consistent with no reddening by dust , moderately inconsistent with @xmath53=0.3 mag of gray dust ( i.e. , graphite dust with minimum size @xmath5 m ; aguirre 1999a , b ) and @xmath53=0.3 mag of galactic - type dust ( savage & mathis 1979 ) .
cccc + jd & age@xmath61 & @xmath1 & @xmath62 + ( 2451000 + ) & ( day ) & ( mag ) & ( mag ) + 204.2 & + 6.2 & 23.93(0.14 ) & -0.62(0.15 ) 216.4 & + 14.5 & 23.95(0.17 ) & 0.11(0.18 ) 239.3 & + 30.2 & 24.35(0.14 ) & 1.30(0.15 ) 243.3 & + 32.9 & 24.16(0.14 ) & 1.66(0.15 ) 261.3 & + 45.3 & 24.59(0.19 ) & 1.65(0.20 ) + + | we have measured the rest - frame @xmath0 and @xmath1-band light curves of a high - redshift type ia supernova ( sn ia ) , sn 1999q ( @xmath2 ) , using _
hst _ and ground - based near - infrared detectors .
a goal of this study is the measurement of the color excess , @xmath3 , which is a sensitive indicator of interstellar or intergalactic dust which could affect recent cosmological measurements from high - redshift sne ia .
our observations disfavor a 30% opacity of sn ia visual light by dust as an alternative to an accelerating universe .
this statement applies to both galactic - type dust ( rejected at the 3.4@xmath4 confidence level ) and greyer dust ( grain size @xmath5 m ; rejected at the 2.3 to 2.6 @xmath4 confidence level ) as proposed by aguirre ( 1999 ) .
the rest - frame @xmath1-band light curve shows the secondary maximum a month after @xmath6 maximum typical of nearby sne ia of normal luminosity , providing no indication of evolution as a function of redshift out to @xmath7 .
an expanded set of similar observations could improve the constraints on any contribution of extragalactic dust to the dimming of high - redshift sne ia .
adam g. riess , alexei v. filippenko , michael c. liu@xmath8 , peter challis , alejandro clocchiatti , alan diercks , peter m. garnavich@xmath9 , craig j. hogan@xmath10 , saurabh jha@xmath9 , robert p. kirshner@xmath9 , b. leibundgut , m. m. phillips , david reiss@xmath11 , brian p. schmidt , robert a. schommer@xmath12 , r. chris smith@xmath13 , j. spyromilio@xmath14 , christopher stubbs@xmath11 , nicholas b. suntzeff@xmath12 , john tonry , patrick woudt@xmath15 , robert j. brunner , arjun dey , roy gal@xmath16 , james graham@xmath17 , james larkin , steve c. odewahn@xmath16 , ben oppenheimer@xmath17 subject headings : supernovae : general@xmath18cosmology : observations |
narrow line seyfert type 1s ( nls1s ) are small mass , high mass accretion rate active galactic nuclei ( agn ) . like other agn , their fastest
, highest amplitude variability is at x - ray energies , but their lower mass means that this variability is on shorter timescales , making it easier to study over a typical @xmath1 day observation ( e.g. ponti et al 2010 ) .
all the nls1s show rapid x - ray variability ( leighly 1999 ) , but some also show deep dips in the x - ray light curve .
these dips coincide with the appearance of high energy complexity in the 2 - 10 kev spectra , either gradual curvature or strong features around the fe k@xmath2 line energy .
gallo ( 2006 ) termed these complex nls1s to distinguish them from the simple nls1s which do not show dips and have relatively power law like spectra from 2 - 10 kev .
two different models have been proposed to explain the deep dips and associated spectral complexity : partial covering and relativistic reflection . in the partial covering model
, the dips are caused by low ionisation material moving into the line of sight , increasing the absorption at the iron edge energy at 7.1 kev .
this material can only partially cover the source as some fraction of the flux at low energies is still seen ( e.g. inoue et al 2003 ; turner et al 2007 ; miyakawa et al 2012 ) .
conversely , in the relativistic reflection model the dips are caused by an extremely compact x - ray source on the spin axis of the black hole approaching the event horizon .
the resulting strong light bending focusses the intrinsic continuum away from the observer ( producing the drop in flux ) so it instead strongly illuminates the very inner disc . for high spin black holes the resulting spectrum can be dominated by highly smeared relativistic reflection , marked by a strong but extremely broad and skewed fe k@xmath2 line ( e.g. fabian et al 2004 ; miniutti & fabian 2004 ; fabian et al 2009 ) . in both models , the complex and simple nls1s are intrinsically similar , and can change from one to the other ( as observed : gallo 2006 ) depending on whether there is absorption along the line of sight , or in the reflection model , whether the compact x - ray source is close to the horizon .
both absorption and reflection models can fit the observed 0.3 - 10 kev spectra , as fitting complex models over a limited bandpass is highly degenerate .
however a recent breakthrough is in the use of variability to break some of these degeneracies .
in particular , the new techniques which can identify a lagged signal are clearly well matched to the reflection models , where there is a strong prediction that the reflected emission should lag behind the intrinsic continuum variability on the light travel time ( fabian et al 2009 ; uttley et al 2014 ) .
detection of a very short lag time ( @xmath3 s ) in the complex nls1 1h0707 - 495 is taken as unequivocal support for the high spin relativistic reflection picture , as this implies distances of the source from the disc of @xmath4 for a @xmath5 black hole .
the simple nls1 pg1244 + 026 shows a much longer lag time of @xmath0 s ( alston , done & vaughan 2014 ) , consistent with the source being somewhat further from the horizon .
lags are derived by comparing variability in a soft and hard band , and show much more structure than just the reverberation signal .
the lags change as a function of the timescale of variability .
the soft leads the hard for long timescale variability , then this lag decreases with decreasing variability timescale until it goes negative ( soft lagging ) for the fastest fluctuations .
this characteristic soft lead decreasing with frequency was first seen in the ( much higher signal to noise ) black hole binaries ( bhbs , e.g. miyamoto et al 1988 ; nowak et al 1999 ) , where it is now interpreted as the signature of fluctuations propagating through an inhomogeneous accretion flow ( kotov et al 2001 ; arevalo & uttley 2006 ) .
slow variability starts at large radii , where the spectrum is softer .
it then propagates down through the entire flow until it gets to the centre , modulating the hardest spectral region .
faster variability is only produced closer to the black hole .
it has a shorter distance to travel to reach the inner regions so takes a shorter time before it modulates the hardest x - ray component from the centre , i.e. this characteristic lag pattern is produced by a correlation of spectrum and variability with radial position .
but this also means that the soft and hard bands both contain some emission from the softer spectrum produced at large radii , and the harder spectrum produced at smaller radii .
there is just more of the softer ( large radii ) spectrum in the soft band than in the hard ( and conversely , more of the harder , small radii spectrum in the hard band than in the soft ) .
thus each band contains the same components , but with different weighting , so most of the variability is perfectly correlated , with only a very small fraction which is lagged .
this is seen in bhbs in a cross - correlation of hard and soft bands , which always peaks at zero lag , with just a small asymmetry ( torii et al 2011 ) .
the important point for agn is that the spectral content of the hard and soft bands matters , and that the lags are diluted if the same component appears in both bands ( miller et al 2010 ; uttley et al 2014 )
. thus lags between light curves in two different energy bands have to be interpreted with a model of the spectral components which contribute to each band .
even the simple nls1 pg1244 + 026 has a spectrum which requires multiple components .
these can either be modelled by a disc , low temperature comptonisation and high temperature comptonisation , together with its ( weak and only weakly smeared ) reflection from a disc , or with disc , high temperature comptonisation and its ( strong , and strongly smeared ) reflection from a disc ( jin et al 2013 ) , or with a steep power law ( presumably from the jet ? ) together with high temperature comptonisation and its ( strong and strongly smeared ) reflection from a disc ( kara et al 2014 ) .
using variability breaks degeneracies , and the combined constraints from energy spectra , power spectra in hard and soft bands , high frequency covariance spectra from the 4 - 10 kev light curve ( the spectrum of the fast variability which correlates with the 4 - 10 kev light curve ) and the lag - frequency spectrum strongly support the first model . in this model , slow fluctuations in the disc propagate down to modulate faster variability in the soft excess ( dominated by the low temperature compton emission ) , which propagates down into the high temperature compton emission from the x - ray corona , which then reflects from and is reprocessed by the disc .
none of the models with strong , strongly smeared relativistic reflection could adequately match all the spectral - timing data for this source ( gardner & done 2014 , hereafter gd14 ) .
in particular the fact relativistic reflection contributes strongly to both the hard and soft bands , results in nearly identical hard and soft band power spectra and high coherence between hard and soft bands up to high frequencies , in disagreement with observations of pg1244 + 026 .
this model had no constraints on spin , as the disc component came from @xmath6 , the soft excess from @xmath7 and the coronal size was assumed to be @xmath8 .
the soft x - ray excess is a composite of a separate low temperature comptonisation component from the disc ( which carries the required propagation signal ) and the mostly thermalised reprocessed emission from non - reflected x - ray illumination of the disc .
hence the soft band contains components which carry the soft lead and the soft lag , but identifying the lag with reprocessing rather than reflection removes the requirement for extreme relativistic smearing as the thermalised emission is already smooth , as seen in the data , quite unlike ionised reflection which has strong atomic features .
nustar data ( akn 120 : matt et al 2014 ) and long term x - ray / uv variability ( mrk 509 : mehdipour et al 2011 ; petrucci et al 2013 ; boissay et al 2014 ) favour the low temperature comptonisation model for the soft x - ray excess in standard broad line seyfert 1s .
this removes the requirement of high spin derived from assuming that this component is instead from ionised reflection ( patrick et al 2011 ; see crummy et al 2006 ) , but we note that the issue is still controversial , and probably complex ( lohfink et al 2012 )
. with this model , the lag in pg1244 + 026 does not require high spin but neither does it rule out a source at @xmath9 above the event horizon of a high spin black hole , still allowing the range from simple to complex behaviour to be due to the height of the source .
however , the broad band spectral energy distribution _ does _ rule out a standard disc around a high spin black hole .
the observed optical / uv continuum constrains the mass accretion rate onto the black hole given its mass , so the total luminosity is only dependent on the efficiency i.e. on black hole spin .
a high spin black hole produces too much luminosity which is very difficult to hide even in the uv - euv interstellar absorption gap ( done et al 2013 ) .
advection is unlikely to change this conclusion , firstly as this source is not strongly super - eddington , and secondly as the most recent global radiative mri ( but newtonian ) disc simulations of a super - eddington flow at @xmath10 show that this is still almost as radiatively efficient as a standard thin disc due to vertical convection ( jiang , stone & davis 2014 ) .
thus this source is most likely low spin , so should never show the extremely deep dips which are characteristic of the complex nls1 , as a low spin black hole can not produce extreme light bending .
this is because the steep emissivity index required can only be produced when the illuminating source height is @xmath4 ( dovciak et al 2014 ) , which for a low spin bh is below the horizon .
this suggests that simple to complex is more than just the source height , and points instead to spin as the potential difference between the two classes .
yet many nls1s show transitions between simple and complex behaviour ( gallo 2006 ; miller et al 2008 ) so spin can not be the origin of the difference either .
the accretion flow in steady state should only depend on black hole mass , spin , and mass accretion rate , plus inclination with respect to the disc which can change its appearance .
all the nls1s have similar mass and mass accretion rate ( within a factor of @xmath11 ) so if the difference between simple and complex is neither source height nor spin , then inclination is the last parameter to try .
these sources are all around eddington , so it seems unlikely that they are actually described by a flat disc as assumed in the relativistic reverberation models .
the global mri simulation at @xmath10 showed a complex flow structure , with a large scale height radiation pressure driven wind from the inner disc carrying away 30% of the input mass accretion rate ( jiang et al 2014 ) .
this wind is likely to be less strong in pg1244 + 026 since this source is only @xmath12 .
uv line driving is unlikely to help since the low mass and high mass accretion rate of nls1s mean that their disc is too hot for its photosphere to have the required opacity to uv line transitions ( hagino et al 2014 ) .
hence strong mass loss is not expected in the nls1s , but some turbulent , clumpy , failed eddington wind in the inner disc could easily form ( jiang et al 2014 ) .
high inclination angles have a high probability of a clump intersecting the line of sight , while low inclination angles are mostly free of obscuration , giving a potential mapping from complex to simple nls1 as a function of inclination . the spectral signature of this time dependent absorption could be complex , depending on the amount of source occulted and the ionisation state of the absorber . here
we explore the effect of occultations on the lag - frequency behaviour .
we show that occultations produce a soft lead when relativistic doppler effects are taken into account .
we couple orbiting clump occultations with the full spectral timing model developed in gd14 to explain the properties of the simple nls1 pg1244 + 026 .
we show that multiple occultations of this intrinsic propagation and reprocessing / reflection model can replicate the change in observed lag - frequency properties from a simple to a complex nls1 .
in particular the soft lead resulting from occultations dilutes the maximum measured reverberation lag and shifts it to higher frequencies , as observed .
many issues still remain for this to be a full model of the complex nls1 .
nevertheless , this shows a promising alternative geometry for the complex nls1 .
[ cols= " < , > " , ]
we now investigate whether the addition of occultations can change the timing properties of a simple nls1 so that they appear more typical of complex nls1s .
that is , can the effect of occultations reduce the maximum measured reverberation lag from @xmath13s to nearer @xmath14s and shift it to higher frequencies ? in the previous sections we assumed constant flux from the underlying accretion flow .
we now replace this static model with the time dependent model of gd14 shown in fig .
the disc , which is at the largest radii , generates the slowest fluctuations .
these propagate down to the soft excess , which is at smaller radii and generates its own slightly faster fluctuations .
the fluctuations in soft excess emission therefore consist of the slow fluctuations from the disc , delayed by some lag related to the propagation time , modulated by the faster fluctuations generated in the soft excess .
these fluctuations then propagate down to the corona , which generates even faster fluctuations .
the hard coronal emission therefore shows fluctuations on a whole range of timescales , as it responds to mass accretion rate fluctuations propagating down from all radii .
a fraction of these central hard x - rays will illuminate the cooler soft excess and disc components .
some of this illuminating flux will be reflected , the rest will thermalise and be reprocessed .
12 shows our spectral decomposition now including these reflected and reprocessed components . these come from a fit to the time
averaged spectrum of the simple nls1 pg1244 + 026 ( obs i d : 0675320101 , see gd14 ) , shown in black data points . for simplicity
we assume all reflection / reprocessing occurs on the soft excess ( @xmath15 ) .
thus the fluctuations in the reflected / reprocessed emission follow the coronal fluctuations ( fig .
11a ) , except for the very fastest fluctuations which are smoothed out by the range of light travel time delays . hence the reflected and reprocessed fluctuations are a lagged and smoothed version of the hard coronal fluctuations .
the soft excess therefore consists of intrinsic emission from the accretion flow ( dashed green line , fig .
12 ) , which varies slowly due to intrinsic mass accretion rate fluctuations in the soft excess and those that have propagated inwards from the disc , and reprocessed emission ( dotted green line , fig . 12 ) which follows the faster coronal fluctuations .
gd14 showed that this model can reproduce all the observed timing properties of the simple nls1 pg1244 + 026 .
we use this model to describe the emission from the underlying accretion flow as a function of time , and now add the effect of occulting clouds .
we fix the cloud parameters to @xmath16 , @xmath17s and @xmath18 .
a transit time of @xmath19s implies an orbital radius of @xmath20 .
hence we reduce our transit radius from @xmath21 in fig .
1a to @xmath20 and take into account the circular motion of the clouds . since the clouds are launched so close to the central x - ray source we allow them to be highly ionised and use the magenta transmission spectrum shown in fig .
9a ( @xmath22 , @xmath23 ) .
@xmath20 is consistent with the clouds being launched as part of a failed eddington ( radiation pressure driven ) wind from the soft excess region as sketched in fig .
radiation pressure lifts material from the accretion flow , which forms clumps as it rises ( takeuchi et al 2014 ) .
as soon as the optical depth of the clumps becomes @xmath24 , some of the material is self - shielded from x - ray photons .
the mass of the clump is still the same but the radiation pressure on it is now less .
if the source is not strongly super - eddington , the radiation pressure is not strong enough to expel the material so it falls back to the disc , resulting in a failed rather than outflowing wind .
the whole turbulent large scale height region is the source of the soft excess .
propagation of fluctuations occurs through the disc regions which are the source of the intrinsic emission , while the bulk of the reflected and reprocessed emission comes from the turbulent clouds .
we assume the turbulent velocity is less than the orbital velocity ( @xmath25 ) and that the clouds remain largely intact on the timescale of a single transit ( although @xmath26 may be sufficient to shred them on longer timescales , stripping off material before what remains falls back to the disc ) . for a source at high inclination
, these clouds will intercept the line of sight to the central regions .
as the clouds transit the line of sight , we assume they obscure the intrinsic disc , intrinsic soft excess and coronal emission .
we do not obscure the reflected or reprocessed emission , since we assume these are predominantly from the clouds .
13 shows the resulting soft and hard band light curves ( left and right respectively ) .
the top panels show the original simple nls1 model with no occultations .
these light curves have power spectra that match the hard and soft band power spectra of the simple nls1 pg1244 + 026 .
in the subsequent panels we increase the number of occulting clouds ( @xmath27 , @xmath28 ) .
the occultations are most obvious in the hard band , where the flux drops are conspicuously narrower than in the soft band , due to the smaller physical size of the corona compared to the more extended soft band components .
these occultations add power to the light curve .
the most heavily occulted hard band light curve ( bottom right panel ) shows peaks and deep troughs more typical of a complex nls1 .
14 shows the power spectra , lag - frequency spectra and coherence between hard and soft bands for the same three simulations . comparing the hard band power spectra ( fig .
14a , dashed lines ) of the original model ( black ) with the most heavily obscured model ( magenta ) shows that occultations have increased the power at @xmath29hz by almost one and a half orders of magnitude .
hard band power spectra of complex nls1s routinely show similarly high power at these frequencies .
thus occultations are more than capable of increasing the hard band high frequency power from @xmath30 typical of a simple nls1 to @xmath31 as is typical of a complex nls1 .
the power increase in the soft band ( solid lines ) is much smaller ( @xmath32 half an order of magnitude ) .
however we have only included occultations at @xmath18 .
higher latitude occultations would add additional power to the soft band , however this power would be uncorrelated with the hard band .
14b shows the lag as a function of frequency between the hard and soft bands .
the black points show the original simple nls1 model , with a strong reverberation lag of @xmath13s at @xmath33hz , matching that seen in pg1244 + 026 . as the number of occultations increases ( blue to magenta ) ,
the maximum measured reverberation lag decreases from @xmath34 to @xmath14s and increases in frequency from @xmath33 to @xmath35hz .
this much shorter reverberation lag , at higher frequency , is much more typical of those seen in complex nls1s such as 1h0707 - 495 . in our model , this is a direct result of the soft leads introduced by the occultations at low frequencies , diluting the reverberation lag and shifting its minimum to higher frequencies .
this is also in good agreement with the findings of kara et al ( 2013 ) , who showed that the reverberation lag of the complex nls1 iras 132243809 is much shorter and at higher frequency during low flux periods ( when in this scenario it would be more obscured ) than high flux periods .
14c shows the coherence between hard and soft bands . as the number of occultations increases , the coherence drops slightly , particularly at low frequencies ( from @xmath36 to @xmath37 ) .
this is due to the slightly different shaped flux drops in the hard and soft bands .
the sudden drop in coherence at @xmath38hz , characteristic of the simple nls1 pg1244 + 026 , becomes less obvious in the more obscured simulations .
the coherence functions of complex nls1s do show a more gradual drop in coherence with increasing frequency , so this is not in disagreement with the data . in fig .
12 , in grey , we show an example of the total spectrum when the source is highly absorbed . we assume the clouds are the source of the reflected and reprocessed emission , hence these components are not absorbed .
consequently the low flux spectrum is dominated by reprocessed emission at low energies ( @xmath1kev ) and reflected emission at high energies this gives a total spectrum that is no longer power - law like but instead shows strong curvature , with a large soft excess and a strong iron emission line , but the smoking gun signature of highly ionised occultation is the presence of highly ionised fe k@xmath2 absorption lines at 6.7 and 6.95 kev .
the strength of these features is probably underestimated in our model as the zxipcf model used assumes a turbulent velocity of only 200 km / s , which is probably much smaller than expected in a failed wind structure from the inner disc .
strong highly ionised absorption lines are detected in the deep dip states of several complex nls1 such as mrk 766 ( miller et al 2007 ) , mcg6 - 30 - 15 ( miller et al 2008 ) and mrk 335 ( gallo et al 2013 ) .
it is possible that even higher turbulent velocities in the failed wind could merge the 6.7 and 6.95 kev absorption lines into each other , and into the absorption edge , which could produce the more dramatic drop at k@xmath2 in 1h0707 - 495 ( hagino et al , in preparation ) .
our assumption that the clouds do not occult the reflected and reprocessed emission sets a limit to the amount of absorption present in our model .
complex nls1s often show rather stronger drops at low energy , which in this scenario would require that the clouds do also occult part of the failed wind structure or that our reprocessed emission is overestimated due to our models taking only the model flux rather than weighting this by the instrument response .
nonetheless , this model demonstrates the potential of this scenario to match the spectral variability , given a more sophisticated prescription for the reflected / reprocessed flux .
we have shown that occultation of the accretion flow can introduce lags between the hard and soft energy bands .
in particular the occultations generate soft leads at a frequency related to the transit time .
for a transit time of @xmath39s , corresponding to an orbital radius of @xmath20 , these lags predominate at low frequencies ( @xmath40hz for a @xmath41 bh ) .
when combined with a model for the intrinsic variability of the accretion flow , these low frequency soft leads act to dilute the negative reverberation lag , reducing the maximum measured lag and shifting it to higher frequencies , reproducing the trend seen in the data from simple to complex nls1 .
this can plausibly be produced by increasing inclination with respect to a clumpy , turbulent structure above the inner disc .
since nls1s are high accretion rate sources , it is likely that the inner regions of the disc will become puffed up to a large scale height where the local accretion rate exceeds the eddington limit ( e.g. jiang et al 2014 ) , producing a turbulent structure capable of obscuring the innermost regions .
low inclination sources are rarely occulted ( simple nls1 ) , while high inclination ones have multiple occultations ( complex nls1 ) .
the assumed higher inclination for complex nls1s also explains the observed association of extreme ( deep dip ) spectra with warm absorbers .
this is not causal , in that the warm absorbers are not distorting our view of the extreme spectra ( e.g. chiang & fabian 2011 ) , but is instead a consequence of high inclination so that the line of sight is more likely to intercept a wind driven from the torus / flattened broad line region ( blr ) .
importantly , the occultations superimpose highly ionised absorption lines at fek@xmath2 in the dips , as are seen in the data .
we show the evolution of the lag - frequency with increased occultations for a fixed black hole mass of @xmath42 , but this should also depend on black hole mass .
the intrinsic lags / leads in the spectral components should scale simply with mass , as should the occultation timescale .
however , mass for nls1s is hard to determine accurately , as they are accreting close to eddington .
masses estimated from line widths assume the blr clouds are virialised
. however the effective gravity experienced by the clouds will be reduced due to radiation pressure from the central source , leading to an underestimate for masses of nls1s ( marconi et al 2008 ) .
inclination is also another uncertainty , as the blr velocity field is not completely virialised , but contains a clear equatorial component ( e.g. collin et al 2006 ; kollatschny & zetzl 2013 ; pancoast et al 2014 ) . any equatorial component to the velocity field
will be suppressed in low inclination ( simple ) nls1s , so their masses will be systematically biased towards higher values compared to high inclination ( complex ) nls1s .
leighly ( 1999 ) give the fwhm for the h@xmath43 line widths in pg1244 + 026 and 1h0707 - 495 as 830 and 1050 km / s , respectively .
these are corrected for feii and have the narrow h@xmath43 component subtracted assuming that this is @xmath44 the [ oiii ] line intensity ( see also leighly & moore 2004 ) .
both sources have intrinsic optical luminosities which are very similar , so these give masses which are @xmath45 and @xmath46 , respectively ( nikolajuk et al 2009 ) .
the two object masses may be even closer if the inclination dependence discussed above is important . an eddington correction to the mass of pg1244
+ 026 increases the mass estimate to @xmath42 ( done et al 2013 ) , but this should also be similar for 1h0707 - 495 . hence , while there are large uncertainties on masses for nls1s , these two objects should be very similar . at this larger mass , pg1244 +
026 is at the eddington limit for a low spin black hole ( done et al 2013 ) . a lower mass and/or higher spin pushes the system to higher eddington fractions , so making it even less likely that the disc is flat . our model is more of a pilot study than a complete description .
obvious improvements are to include general relativistic effects of light bending on the disc image ( fig .
1a ) , e.g. middleton & ingram ( 2014 ) , miniutti et al ( 2014 ) . this would be most important for the central coronal regions as the far side of this small source always has a small impact parameter with the black hole .
this would make the corona appear larger , so a @xmath47 cloud may not occult the entire corona . as a consequence
the coronal flux drops would not be quite as narrow and deep and this would reduce the difference between the total amount of power added to the hard and soft bands .
complex nls1s do show more power in the hard band than the soft band ( zoghbi et al 2011 ) at all frequencies , unlike the simple nls1 pg1244 + 026 , which shows comparable power at low frequencies in the hard and soft bands .
our occultation model replicates this , since occultations add power to both bands .
however the small size of the corona necessarily adds much more power to the hard band . including light - bending ( or a smaller cloud size )
would slightly lessen this difference , in better agreement with the data .
the model presented here is in some way a composite between the previous extreme relativistic reflection models and partial covering models .
it follows the partial covering model in identifying absorption ( as opposed to light bending ) events as the origin for the deep dips , but has the occulting material be closer to the source ( ten gravitational radii rather than a few tens - hundreds ) , and be more highly ionised .
reflection does make an important contribution to the spectrum during the dips in our model , but it is not extremely smeared by relativistic effects .
instead , and in a step beyond what is modeled here , we envisage the reflector as a clumpy , turbulent , failed wind rather than a flat keplerian disc ( see also miller et al 2008 )
. the clumps may be only marginally optically thick , so their reflected / scattered emission is not quite the same shape as from @xmath48 ( miller & turner 2013 ) , and they may be embedded in hotter material which comptonises the reflected emission . additionally , the cloud itself could have complex structure due to the ionisation instability of x - ray illuminated material ( krolik , mckee & tarter 1981 ) .
the illuminated face of the cloud will be heated to the local compton temperature ( @xmath49k ) .
temperature decreases at larger depths into the cloud where scattering reduces the heating , so the density must increase to keep in pressure balance .
this lower ionisation state material has more line cooling , so the temperature drops abruptly , giving a sharp transition between a highly ionised skin and a nearly neutral core ( chevalier et al 2006 ) .
reflection from such structures , especially with a turbulent velocity field , may be a feasible way to reproduce the observed 2 - 10 kev spectra in the dips .
another difference between this model and standard relativistic reflection is that we include thermalisation of the illuminating flux .
hard x - rays which are not reflected can either heat the disc and be re - emitted as ( mostly ) thermalised radiation , or the energy can be released as lines / recombination continua .
the relative importance of these two processes depends strongly on the vertical structure of the disc .
thermalisation is more important for discs in hydrostatic equilibrium ( nayakshin et al 2001 ) but current reflection models are calculated for constant density discs ( e.g. ross & fabian 2005 ; garcia et al 2014 ) .
the high disc temperatures expected in nls1s means that this component must be important at some level in contributing to the soft x - ray excess in these objects , and since it is predominantly thermal then it has no strong soft x - rays lines which require high spin to smear them into the observed smooth continuum
. however , probably the most important effect which should be included in matching to real data is that the light curves in the soft and hard bands are weighted by the detector response rather than being simply flux integrated over the energy band as used here .
this is a key requirement to fit the model to real data , which will be the subject of a subsequent paper .
we have constructed a simple occultation model to investigate whether the change in spectral and timing properties between simple and complex nls1s can be explained by a difference in inclination with respect to a failed wind . in this scenario ,
clumps of material lifted from the inner parts of the accretion disc by radiation pressure , obscure the x - ray emission for sources seen at high inclination angles , resulting in more extreme variability and more complex spectra . associating the deep dips with occultation superimposes absorption features from fek@xmath2 on the dip spectra .
this is seen in complex nls1s such as mrk 776 ( miller et al 2007 ) and is a smoking gun for the reality of these occultation events .
we model the obscuration as a series of individual clouds of constant ionisation parameter which transit the inner accretion flow , co - rotating with the flow and obscure the underlying emission .
the underlying accretion flow emission is radially stratified , with the softest x - rays ( disc ) from the largest radii , and then the soft x - ray excess and corona at progressively smaller radii .
we find that occultations add power to the x - ray light curves over a range of frequencies related to the transit time .
occultations also introduce a lag between the hard and soft energy bands when doppler boosting of the underlying accretion flow emission is taken into account ; specifically occultations introduce a soft lead , with the hard band lagging the soft band .
we then combined our occultation model with the full spectral - timing model of gd14 which describes the accretion flow emission of the simple nls1 pg1244 + 026 .
this model also includes reprocessed emission as part of the soft x - ray excess , as well as reflection from it .
this reproduces the timing properties of pg1244 + 026 by assuming slow fluctuations are generated in the outer components and these propagate down to the corona , producing low frequency hard lags .
the high frequency soft lags ( reverberation lags ) are produced predominantly by fast coronal fluctuations being reprocessed in the soft excess wind material rather than by reflection from it .
the effect of the occultations is to dilute the negative reverberation lag and shift it to higher frequencies . by increasing the rate of occultations we can match the change in reverberation lag from the @xmath34s at @xmath50hz seen in the simple nls1 pg1244 + 026 , to the @xmath51s lag at @xmath35hz seen in the complex nls1 1h0707 - 495 .
the lag times and light travel times put into the model are the same in both the obscured and unobscured cases .
the only difference is the presence of occultations .
it is the soft leads caused by the broken symmetry of the ingress and egress of the occultations caused by doppler boosting of the disc emission which result in a shorter net reverberation lag in the obscured case .
the occultations also change the energy spectrum from a simple nls1 , with strong soft excess and a steep power law above 2 kev , to something resembling a complex nls1 , with a prominent iron line from reflection off the soft excess , highly ionised fe k@xmath2 absorption lines and strong spectral curvature .
the short @xmath51s reverberation lags have been taken as evidence for extreme relativistic reflection in complex nls1s . if 50s really is a light travel time this requires reflection from the innermost radii of an accretion disc around a highly spinning bh .
our occultation model shows that this need not be the case .
a short reverberation lag can be the result of a much longer light travel time , diluted and shifted by the soft leads introduced by occulting clouds .
in our model , reflection and reprocessing occurs between @xmath15 and puts no constraints on the spin of the bh . changing inclination then naturally explains the change from smooth spectra to complex spectra , and long lower frequency reverberation lags to shorter higher frequency reverberation lags in simple and complex nls1s .
given that nls1s are high accretion rate sources , it is quite natural to expect that the disc is not flat , that radiation pressure can lift material from the disc which will obscure the central emission for high inclination lines of sight .
this is a promising geometry to explore further .
we thank the anonymous referee for helpful comments .
eg acknowledges useful discussions with will alston and matt middleton .
cd acknowledges useful conversations with giovanni miniutti .
eg acknowledges funding from the uk stfc . | complex narrow line seyfert 1s ( nls1s ) , such as 1h0707 - 495 , differ from simple nls1s like pg1244 + 026 by showing stronger broad spectral features at fe k and larger amplitude flux variability .
these are correlated : the strongest fe k features are seen during deep dips in the light curves of complex nls1s .
there are two competing explanations for these features , one where a compact x - ray source on the spin axis of a highly spinning black hole approaches the horizon and the consequent strong relativistic effects focus the intrinsic flux onto the inner edge of a thin disc , giving a dim , reflection dominated spectrum .
the other is that the deep dips are caused by complex absorption by clumps close to the hard x - ray source .
the reflection dominated model is able to reproduce the very short 30s soft lag from reverberation seen in the complex nls1 1h0707 - 495 .
however , it does not explain the characteristic switch to hard lags on longer timescales . instead , a full model of propagating fluctuations coupled to reverberation can explain the switch in the simple nls1 pg1244 + 026 using a low spin black hole .
however pg1244 + 026 has a longer reverberation lag of @xmath0s . here we extend the successful propagation - reverberation model for the simple nls1 pg1244 + 026 to include the effect of absorption from clumps in a turbulent region above the disk .
the resulting occultations of the inner accretion flow can introduce additional hard lags when relativistic effects are taken into account .
this dilutes the soft lag from reverberation and shifts it to higher frequencies , making a smooth transition between the 200s lags seen in simple nls1s to the 30s lags in complex nls1s .
these two classes of nls1 could then be determined by inclination angle with respect to a clumpy , probably turbulent , failed wind structure on the disc .
= 1 = -0.5 cm [ firstpage ] black hole physics , accretion , x - rays : galaxies , galaxies : seyfert , galaxies : individual : pg1244 + 026 , galaxies : individual : 1h0707 - 495 . |
the energy loss of hard partons propagating through the soft medium created in heavy - ion collisions has long been regarded as a promising tool to gain information on the medium density evolution @xcite .
radiative energy loss , i.e. the idea that medium - induced radiation predominantly carries away energy from a hard parent parton has been rather successful in describing not only the hadronic nuclear suppression factor for central collisions but also the effects of changing medium geometry @xcite .
calculations within dynamical evolution models in various formalisms @xcite have improved on this result and show also agreement with measured hard back - to - back correlations @xcite and the measured suppression of protons @xcite .
neverteless , there are indications that a radiative energy loss picture fails to describe the suppression of heavy quarks as seen in the single - electron spectra measurements @xcite and energy loss due to elastic collisions with the medium @xcite has been suggested as a possible solution to this problem .
such calculations indicate a large component of elastic energy loss also for light quarks and gluons . in vacuum ,
the distinction between an elastic and a radiative process is straightforward if the number of asymptotic out - states is larger than the number of asymptotic in - states , the process is radiative .
however , for a parton propagating through a medium , no asymptotic out - states can be defined . in particular , radiated quanta do not need to be on - shell as long as they re - interact with the medium within a sufficiently short amount of timescale as set by the uncertainty principle .
it becomes thus to some degree a matter of convention if a particular process is seen as the ( elastic ) exchange of a virtual parton between hard parton and medium or as the ( inelastic ) radiation of a virtual parton from the parent where the radiated parton is subsequently absorbed by the medium .
however , while there is no sharp conceptual distinction , there are two crucially different regimes : if a parton radiated from the parent is highly spacelike , the formation time for this process is very short . on the other hand ,
the decoherence time for near on - shell radiation is very long , giving rise to interference effects and lpm suppression .
phenomenologically , the first regime leads ( in a medium with constant density ) to a linear dependence of energy loss on pathlength whereas in the latter regime a quadratic dependence appears .
it is the purpose of this paper to study observable consequences of this difference in pathlength dependence . to this end
, we first construct a class of models in which the energy loss has parametrically a linear pathlength dependence .
based on the requirement that the nuclear suppression factor @xmath1 for central collisions should be described , we point out limits for the parameter space of this model class . finally , we demonstrate with three different scenarios the consequences of linear pathlength dependence for observables which explicitly probe pathlength dependence such as @xmath1 vs. reaction plane or hard back - to - back dihadron correlations and discuss limits for the relative magnitude of an elastic energy loss component .
key quantity for the calculation of energy loss in a dynamically evolving medium is the probability distribution @xmath2 for a parton with initial energy @xmath3 to lose the energy @xmath4 for any given path through the medium ( in general , this includes a discrete part accounting for the possibility that the parton escapes without energy loss ) .
empirically , one finds that a strong dependence of @xmath2 on the initial energy @xmath3 does not seem to be favoured by the data @xcite , thus we will in the following approximate @xmath5 . from the energy loss distribution given a single path , we can define the overlap - geometry averaged energy loss probability distribution @xmath6 as @xmath7 here , @xmath8 is the angle of the outgoing parton with respect to the reaction plane and @xmath9 is the probability density for finding a hard vertex at the transverse position @xmath10 given the collision impact parameter @xmath11 .
this quantity is given by the product of the nuclear profile functions as @xmath12 where the thickness function is given in terms of woods - saxon the nuclear density @xmath13 as @xmath14 .
if we are interested in more differential observables , the relevant averaging procedure is changed , for example @xmath1 as a function of the reaction plane is obtained from @xmath15 or the back - to - back per - trigger - yield can be computed using @xmath16 which is found by replacing @xmath17 in eq .
( [ e - p_taa ] ) by @xmath18 , the conditional probability density to find a vertex in the transverse plane from which a near side hadron in the relevant trigger range was produced .
the latter quantity is best calculated in a monte - carlo ( mc ) framework as it also has explicit momentum dependence , for details see @xcite .
we calculate the momentum spectrum of hard partons in leading order perturbative qcd ( lo pqcd ) ( explicit expressions are given in @xcite and references therein ) . for reasonably hard momenta ( in practice 6 gev or more )
it can be assumed that hadronization takes place outside the medium .
then the medium - modified observables arise from a convolution of the pqcd parton spectrum with the suitably averaged energy loss probability density and a hadronization function .
for example , the medium - modified perturbative production of hadrons at angle @xmath8 can be computed from @xmath19 and the partonic cross section @xmath20 using @xmath21 with @xmath22 the fragmentation function for parton @xmath23 with momentum fraction @xmath24 at scale @xmath25 @xcite . from this
we compute the nuclear modification function @xmath1 vs. reaction plane as @xmath26 thus , all the information about the medium and the energy loss model is contained in @xmath2 .
for radiative energy loss , we have in past works obtained this quantity by evaluating the two line integrals @xmath27 and @xmath28 along the path of the parton through the medium and using the results of @xcite to convert the result to a probability distribution where we have assumed @xcite @xmath29 both the local energy density @xmath30 and the transverse flow field @xmath31 ( with @xmath32 the angle between flow vector and parton propagation ) have to be inferred from a dynamical evolution model .
for ease of comparison with published results , we use a 2-d hydrodynamical evolution model @xcite as in @xcite and a 3-d hydrodynamical model @xcite as in @xcite . in a static medium , @xmath33
and thus eq . ( [ e - omega ] ) exhibits the quadratic pathlength dependence for the energy scale parameter @xmath34 of radiative energy loss .
we do not attempt to construct a first - principles model of elastic energy loss in the following .
such an attempt faces a number of difficulties , among them the running of the strong coupling @xmath35 into a soft regime , the treatment of interactions of the high-@xmath0 parton in the hadronic phase of the medium , the precise nature of scattering partners in the qgp phase and @xmath36 conversion reactions .
instead , we try to construct a class of models which has the expected parametric dependence of energy loss probability distributions on the pathlength and infer then from the data limits for the parameters which characterize the model class . since any observable considered here involves massive geometrical averaging ( eqs .
( [ e - p_taa]),([e - p_phi ] ) ) and since furthermore the nuclear suppression factor is not very sensitive to the shape of even the averaged @xmath37 @xcite , we may safely assume that details in the modelling of @xmath38 will not influence the outcome significantly .
note that in an ideal quark - gluon plasma ( qgp ) , @xmath39 is a measure of the entropy density and hence ( up to a numerical constant ) of the number density @xmath40 of scattering centers which can be probed by a parton on its trajectory @xmath41 . if the cross section for elastic energy loss is @xmath42 and the medium density @xmath40 , the number of scatterings @xmath43 on a path @xmath44 is given by @xmath45 .
exponentiating this expression , we expect that the escape probability @xmath46 without undergoing energy loss for a parton is parametrically given by @xmath47 = \exp[- \gamma \cdot \kappa]\ ] ] where we have assumed that @xmath42 is approximately independent of @xmath41 and @xmath48 is defined in analogy with eq .
( [ e - omega ] ) as @xmath49 taking into account the flow corrections to the probed density . here
@xmath50 is a parameter with dimensions of a cross section measuring the interaction strength , and hence @xmath51 must hold to account for the different color factors of quarks and gluons .
note that in the case of radiative energy loss as calculated using the assumption of @xcite the escape probability is determined by the condition of no radiation beyond the vacuum shower evolution rather than by a no - scattering condition .
it can then be cast into the form @xcite
@xmath52\ ] ] which requires the knowledge of the spectrum of medium - induced radiation @xmath53 as a function of radiated energy @xmath54 .
this in turn depends on a decoherence condition for radiated quanta as well as lpm interference and can not easily be cast into a simple form comparable to eq . ( [ e - p0 ] ) .
a good discussion of the underlying physics can be found in @xcite .
if the parton does not escape without energy loss , it must undergo a shift in energy ( there is also the possibility that a strong shift into a thermal regime occurs , which is equivalent to an absorption of the parton ) .
it is reasonable to assume that the mean value of the shift in energy will grow linear in the number of scatterings @xmath55 as @xmath56 with @xmath57 the mean energy loss per scattering whereas the fluctuations around the mean will grow like @xmath58 . assuming a gaussian distribution , this leads to the ansatz @xmath59\ ] ] where @xmath60 is a normalization such that @xmath61 and ( [ e - elastic ] ) has to hold for quarks and gluons separately due to the different color factor .
@xmath32 is a parameter with the dimensions of a cross section times the energy shift per reaction .
this class of energy loss models is characterized by three parameters : * @xmath32 controls the mean shift in energy per expected scattering * @xmath62 governs the strength of fluctuations around this mean shift . if @xmath62 is small , the model will have a strong correlation between path ( and hence initial vertex ) and shift in energy , if the parameter is large , this correlation is lessened * @xmath50 finally determines the magnitude of the escape probability . in a microscopical model
, the actual distribution would presumably not be strictly gaussian and the three parameters would be correlated and calculable . here
, we will however pursue a different approach and see what can be inferred from the data .
our aim is to determine parameters of the elastic energy loss model such that @xmath1 for central collisions is described well and to use these parameter settings to compute observables where a different geometrical avergaing is performed .
it does not seem straightforward to choose from the model space given by all possible @xmath63 all the solutions which are compatible with @xmath1 .
however , according to @xcite , at least one parameter is fixed by the normalization of @xmath1 .
let us consider @xmath50 first : clearly , the averaged discrete escape probability @xmath64 may not be larger than @xmath1 .
on the other hand , if it is smaller than the error on @xmath1 , this term is irrelevant . via @xmath65 < r_{aa}\ ] ] this translates to ( evolution - model dependent ) constraints on @xmath50 .
let us now consider a scenario which is chosen such that @xmath64 is large and close to @xmath1 .
it follows that there must be a constraint on the shape of the continuous shift probability @xmath38 .
the essential idea is apparent from the following : averaging over both parton species and neglecting the fragmentation ( which is a subleading correction on the shape of @xmath1 in the high @xmath0 region @xcite ) , @xmath1 can be estimated as @xmath66 where the lower integration boundary does not include the discrete contribution at 0 and @xmath67 is the kinematic limit for the parton energy .
if the spectrum is approximated by a power law @xmath68 , then @xmath69 in other words , @xmath1 can be written as a constant term plus a term which increases with @xmath0 and corresponds to the integral of the energy loss probability density , weighted by a steeply falling spectrum . from this term
, sizeable contributions to @xmath1 will come if @xmath70 contains a lot of strength close to @xmath71 ( in this case the weight factor is close to unity ) or if the spectrum is flat , i.e. @xmath72 is small ( as at the lhc @xcite ) or if @xmath73 , i.e. at high @xmath0 .
thus , there is the generic expectation of a rise of @xmath1 with @xmath0 @xcite .
if @xmath50 is already close to the allowed limit , @xmath32 and @xmath62 must thus arrange in such a way that the shift term is small in order not to violate the limit .
this means that the continuous energy loss distribution can not contain much strength close to @xmath71 .
there are two possible ways this could happen ( note that the continuous part must integrate to approximately @xmath74 ) : first , the distribution could be very flat and extend to large @xmath4 , thus the contribution in any given fixed interval in @xmath4 would be small . or second
, the distribution could be comparatively narrow but peaked at some large @xmath4 . in the case of radiative energy loss , the first scenario is realized , and for rhic kinematics @xmath6 extends to @xmath75 gev @xcite
. however , this can not be so in the elastic case .
assume that along some fixed parton path elastic and radiative processes lead to the same mean energy loss .
then , a comparison of eqs .
( [ e - omega ] ) and ( [ e - kappa ] ) shows that parametrically the radiative energy loss will be smaller for all shorther paths but larger for all longer path .
thus , the dynamical range of expected energy loss generated in the model between short pathlength contributions close to the surface and long paths traversing the whole medium is vastly greater for radiative energy loss , making it very difficult to obtain a flat distribution in elastic energy loss unless @xmath62 is set to ( unnaturally ) large values .
however , when this is done , the probability distribution even for a single path is very wide , i.e. large fluctuations destroy the position - energy loss correlation and hence the tomographic information .
we investigate two different scenarios with a large discrete escape probability ( chosen such that it is comparable with the discrete quenching weights in the radiative energy loss scenario @xcite ) . the parameters of each scenario can be found in table [ t - params ] . in the first one ( scenario 1 )
we assume a large value of @xmath62 , i.e. sizeable fluctuations of the energy loss given a path , in the second one ( scenario 2 ) we choose small @xmath62 . the continuous energy loss probability distribution using the medium evolution provided by the 3-d hydrodynamics for quarks for a single path from the medium center and and the geometry - averaged distribution @xmath37 for both scenarios are shown in fig .
[ f - p_q ] .
we also investigate a third scenario ( scenario 3 ) in which the discrete escape probability is adjusted to half the value of scenario 1 and 2 . by the same argument seen above , the integral term in eq .
( [ e - r_aa_approx ] ) must contribute more , hence the probability distribution needs strength close to @xmath71 , leading to some @xmath0 dependent growth of @xmath1 .
this can only be achieved by choosing a comparatively large @xmath62 .
the resulting change in the shape of @xmath1 due to the increased rise with @xmath0 as compared with the other scenarios disfavours even larger values of @xmath50 ( or a complete absence of escape without energy loss ) .
.[t - params]parameters for the three different elastic energy loss scenarios described in the text [ cols="^,^,^,^",options="header " , ]
in fig . [ f - r_aa ] we show the resulting @xmath1 for different angles with respect to the reaction plane for central au - au collisions at 200 agev and for non - central collisions at impact parameter @xmath76 fm based on the medium description of the 3-d hydrodymnamics code @xcite . for comparison
we also include the radiative energy loss calculation . by construction ,
differences between the scenarios are not very pronounced in central collisions .
however , the extrapolation to non - central collisions is rather different in all cases .
first , all three elastic energy loss scenarios exhibit a pronounced rise with @xmath0 which is stronger than in the radiative case .
as outlined above , this is connected with the strength of @xmath77 close to @xmath78 .
however , second and possibly more impartant , the splitting between in - plane and out - of - plane emission in elastic energy loss is at most half of what is seen for radiative energy loss .
this is not an unecpected feature : in a dynamic medium , the leading initial density dilution goes as @xmath79 due to the longitudinal expansion ( changing into @xmath80 with @xmath81 and growing later as pressure gradients drive accelerated expansion of the medium ) . inserting this result int eq .
( [ e - kappa ] ) shows that elastic energy loss is dominated by early times when the medium is dense . at late times
, @xmath48 increases only logarithmically . using typical evolution models
, this implies that energy loss models with a linear pathlength distribution lead to sizeable losses only in the first 2 - 3 fm / c evolution time and become insensitive to late time behaviour .
however , if the medium becomes effectively transparent after 2 - 3 fm , a parton can not probe density gradients much larger than this .
this implies a loss of the sensitivity to in - plane vs. out - of - plane emission for some partons , and even more dramatic , a loss of sensitivity to the difference between average near - side and away side pathlength .
consistent with the expectation , the smallest splitting is observed when @xmath62 is set to large values and position - energy loss correlations are weakened in addition .
using the mc code of @xcite , we compute the strength of hard hadronic back - to - back correlations for all three scenarios of elastic energy loss and compare with the star measurement @xcite .
as can easily be seen , elastic energy loss gives a parametrically wrong result the calculation overshoots the data by a factor of more than two .
this expectation was already formulated in @xcite : in the radiative energy loss calculation , the difference between near side emission and away side emission arises from the drastic differences in average pathlength on near and away side @xcite .
however , since in all elastic scenarios only pathlengths of 2 - 3 fm can be probed before the medium becomes effectively dilute , such mechanism can not play a role .
thus , the main difference between near side and away side is that partons on the near side initially propagate into a zone of lower density and partons on the away side into higher density .
this accounts for the fact that there is additional suppression on the away side , however it is by far not as strong as in the radiative case .
if the position - energy loss correlation is further weakened ( as in scenario 3 ) the disagreement with the data is more pronounced .
it seems clear that a scenario in which energy loss is exclusively elastic is not very realistic in reality , energy loss is presumably caused by different component processes which have different pathlength dependence .
we may however estimate an upper bound for the strength of an elastic component with a simple ansatz the biggest uncertainty in the calculation of back - to - back correlations is given by the d - au baseline for the correlation strength .
if this uncertainty is combined with the uncertainty of the yield per trigger in au - au collisions and scenario 1 is chosen as being closest to the data , the 4 - 6 gev momentum bin provides the strongest constraint and marginally allows an elastic energy loss fraction @xmath83 .
however , this assumes that fragmentation is the only source of hadrons in this momentum bin which is not consistent with the fact that both radiative and collisional energy loss individually underpredict @xmath1 in this momentum window ( cf fig .
[ f - r_aa ] ) .
thus , the true upper limit on @xmath23 is even smaller . on the other hand ,
the most likely value of @xmath23 in the analysis comes out rather small as @xmath84 . in modelling the elastic energy loss ,
we have made several simplifying assumptions : first , we have assumed eikonal propagation of partons whereas in reality elastic energy loss leads to a deflection of partons .
furthermore , in writing down eq .
( [ e - elastic ] ) we have assumed that @xmath42 does nt strongly depend on the medium .
however , if @xmath85 ( and hence the interaction strength ) grows very strongly for @xmath86 this could to some extend effectively result in a deviations from a strictly linear pathlength dependence ( the same effect would however occur for radiative energy loss ) .
furthermore , we have assumed a gaussian distribution for the elastic energy loss probability .
we have checked that the results do not crucially depend on this point .
the reason is that the substantial geometrical averaging over many paths , cf .
( [ e - p_taa],[e - p_phi ] ) erases any detailed information on the shape of the distribution given a single path .
in particular , we verified that replacing a gaussian shape by a box shape with the same r.m.s width does not alter the resulting @xmath1 by more than 5% and yields the same back - to - back correlations within the statistical errors of the mc simulation .
finally , we have neglected any energy dependence of the energy loss probability distributions and hence finite energy corrections . however , while these ( and other ) effects would clearly have an effect on the energy loss probabilities ( to details of which the result is largely insensitive ) it is less clear how they could possibly compensate for the generically different pathlength dependence of the two regimes .
it appears that the general features of this analysis are rather robust and there is a physics reason why this should be so : the difference in near side and away side suppression is consistent with a quadratic pathlength dependence but not with a linear dependence of energy loss . as apparent from the rather consistent overprediction of the away side yield even for large discrete escape probability ( the best possible scenario ) , this is a quite generic statement which holds for a whole class of elastic energy loss models and does not depend on details of how the probability distribution appears microscopically .
it is hard to imagine a mechanism which would change this picture without strongly changing the parametric dependence on pathlength .
this places rather stringent limits on the relative magnitude of elastic contributions to light quark and gluon energy loss which may help to constrain microscopical models of energy loss . | the unexpectedly strong suppression of high @xmath0 heavy - quarks in heavy - ion collisions has given rise to the idea that partons propagating through a medium in addition to energy loss by induced radiation also undergo substantial energy loss due to elastic collisions .
however , the precise magnitude of this elastic energy loss component is highly controversial .
while it is for a parton inside a medium surprisingly difficult to define the difference between elastic and radiative processes rigorously , the main phenomenological difference is in the dependence of energy loss on in - medium pathlength : in a constant medium radiative energy loss is expected to grow quadratically with pathlength , elastic energy loss linearly . in this paper
, we investigate a class of energy loss models with such a linear pathlength dependence and demonstrate that they are incompatible with measured data on hard hadronic back - to - back correlations where a substantial variation of pathlength is probed .
this indicates that any elastic energy loss component has to be small . |
the theoretical proposal @xcite and experimental discovery of the topological insulators @xcite have provoked an intensive research effort in condensed matter physics .
topological insulators ( ti ) with time - reversal symmetry are generally characterized by a topological term in the electromagnetic action with a quantized coefficient@xcite .
these states have been theoretically predicted and experimentally observed in both two and three dimensions , including the two - dimensional ( 2d ) hgte / hgcdte quantum wells @xcite , and bulk three - dimensional materials bi@xmath0te@xmath1 , bi@xmath0se@xmath1 and bi@xmath4sb@xmath5 @xcite .
they exhibit robust gapless modes at boundaries , _ e.g. _ a 1d helical edge mode for 2d tis , and a 2d helical surface mode for 3d tis with odd numbers of dirac cones .
due to time reversal symmetry , backscattering is forbidden for the helical edge and surface states , and an analysis of interaction effects for the 1d helical edge modes shows they are stable against weak and intermediate strength interactions @xcite . bi@xmath0te@xmath1 and bi@xmath0se@xmath1 have been predicted to have bulk band gaps exceeding room temperature@xcite , which makes them promising for future applications .
et al _ predict that the surface states of bi@xmath0te@xmath1 consist of a single dirac cone at the @xmath6 point , and that the dirac cone evolves into a hexagonal shape at higher energy@xcite . furthermore , near the dirac point , the spin of the electron lies perpendicular to the momentum .
angle - resolved photo - emission spectroscopy ( arpes ) measurements performed on the surface of bi@xmath0te@xmath1 have confirmed these predictions in detail@xcite .
the typical shape of the fermi surface is a snowflake - like warped hexagon .
the low - energy ( 2 ) symmetry of the dirac cone is broken due to the @xmath7 symmetry of the underlying lattice@xcite , and can be modeled by a warping term in the effective model@xcite .
another powerful surface probe , spectroscopic scanning tunneling microscopy ( stm ) , is sensitive to quasi - particle interference ( qpi ) around impurities , and provides an important tool to study electronic structures in unconventional materials , such as high t@xmath8 cuprates @xcite .
it can provide information in momentum space through real space measurement with a high energy resolution .
recently , several groups have performed stm measurements on surface states of bi@xmath0te@xmath1 and bi@xmath4sb@xmath5 @xcite .
backscattering induced by non - magnetic impurities between time - reversal ( tr ) partners with opposite momenta is forbidden due to their opposite spin configurations .
this is confirmed by the real space friedel oscillation pattern and by analysis of the qpi characteristic scattering wavevector . in this paper
, we perform a detailed qpi analysis of the surface states of the topological insulator bi@xmath0te@xmath1 .
a general tr - invariant impurity potential including scalar and spin - orbit scattering components is studied using the standard @xmath2-matrix formalism .
the scattering on the iso - energy surface strongly depends on the both momentum and spin orientation .
scattering between tr partners vanishes as a consequence of tr symmetry .
the scattering is dominated by wavevectors which connect regions on the fermi surface of extremal curvature , but also accounting for spin polarization .
stm experiments@xcite have yielded rich information about the qpi structure .
in addition to the absence of backscattering , the stm experiments also observed recovered scattering@xcite at a wavevector ( @xmath9 in their , and @xmath10 in our notation ) , and an extinction@xcite ( i.e. near absence of scattering ) ( @xmath11 in their and our notation ) , both at wavevectors which do not connect tr states .
below , we offer a novel explanation of this experimental puzzle .
our results are in excellent overall agreement with the qpi experiment in bi@xmath0te@xmath1 .
the @xmath12 hamiltonian for the surface dirac cone was first derived in ref . .
the bare hamiltonian is written as @xmath13 , where @xmath14 . with the addition of the cubic warping term @xcite , h()=v ( ) + k^33^z .
the azimuthal angle of @xmath3 is @xmath15 , where the @xmath6-@xmath16 direction is taken as @xmath17 axis .
following ref .
, the quadratic terms are dropped since they do not significantly change the shape of the constant energy contour , and the characteristic energy and wavevector scales are defined as : @xmath18 and @xmath19
. this hamiltonian can be diagonalized by introducing ( ) = ( cc ( /2 ) & i e^-i(/2 ) + & + i e^i(/2 ) & ( /2 ) + ) , [ eigenfunctions ] where @xmath20 .
one then finds @xmath21 , with eigenvalues @xmath22 where e()= . in fig .
[ fig : fs](a ) we plot the isoenergy contour @xmath23 , which qualitatively reproduces the snowflake fermi surface observed in the first - principles calculation and the arpes experiment @xcite . as for the scattering process , we take _ imp= d^2kd^2kv_- ( ) ( ) .
[ hs ] for a single short - ranged scatterer we may approximate @xmath24 .
the second term corresponds to the spin - orbit scattering with the coefficient @xmath25 describing its relative strength to the potential scattering .
it is convenient to project the potential onto the eigenbasis of @xmath26 , so _ , v_0^ ( ) ( ) .
[ vkk ] for simplicity , we first consider the @xmath27 case ( pure scalar potential scattering ) , returning later to the general spin - orbit case ( @xmath28 ) . since the spectrum is particle - hole symmetric , let us focus on a definite ( positive ) sign of the energy .
the qpi will then be dominated by scatterings inside the positive energy band , whose effective scattering potential is : _ , ^(11)=v_0 .
[ v11 ] this effect also appears in the qpi analysis of the orbital - band systems where orbital hybridization brings strong momentum dependence to the scattering process @xcite .
( color online ) ( a)the iso - energy contour near the @xmath6 point for @xmath29 with snow - flake shape .
the @xmath17 and @xmath30 axes are chosen to be the @xmath6-@xmath16 and @xmath6-@xmath31 directions respectively , and @xmath19 .
the red and brown ( dark gray ) dots refer to the valley and the tip points on the contour , and the arrows indicates six representative scattering wavevectors . @xmath32 and @xmath33 are solutions of @xmath34 which are the boundary of the truncation for the @xmath35-integration used in this paper .
( b ) the spin orientations of the eigenfunctions for @xmath36 band at valley and tip points .
the dotted lines refer to the mirror - symmetric lines ( @xmath6-m ) , and the system has a three - fold rotational symmetry .
the arrow indicate the spin configuration in the @xmath37 plane and the solid circle ( cross ) refers to @xmath38 being along @xmath39 ( @xmath40 ) . at
the cusp points the spin lies only on the @xmath37 plane while @xmath38 has the largest magnitude at the valley points with staggered signs . ]
the points of extremal curvature on the fermi surface are divided into two groups , arising from the ` valleys ' ( @xmath41 , positive curvature ) and ` tips ' ( @xmath42 , negative curvature ) .
we define the complexified points @xmath43 , @xmath44 , @xmath45 , @xmath46 , @xmath47 , and @xmath48 . then from eqn .
[ v11 ] we obtain @xmath49 , @xmath50 , and @xmath51 , where @xmath52 , corresponding to scattering through the vectors @xmath53 , @xmath54 , and @xmath55 , respectively , with @xmath56 .
we also find @xmath57 , @xmath58 , and @xmath59 .
these processes are depicted in fig .
[ fig : fs](a ) .
while @xmath60 is a direct consequence of tr symmetry , the other processes through scattering vectors @xmath61 are in general finite .
their amplitude variation may be understood in terms of the spin orientation of the eigenfunctions throughout the brillouin zone , @xmath62 , depicted in fig .
[ fig : fs](b ) .
bi@xmath0te@xmath1 has the symmetry of @xmath7 , _
i.e. _ three - fold rotational symmetry plus the three reflection lines ( @xmath6-@xmath31 plus two equivalent lines ) .
therefore at the tips @xmath63 must vanish since @xmath64 is odd under the mirror operation . @xmath63 has the largest magnitude at the valleys , but with staggered signs , as shown in the figure . since scalar potential scattering does not flip electron spin , its matrix element is largest when @xmath65 is large and positive , _
i.e. _ high spin overlap .
this echoes the experimental finding of pascual _
et al._@xcite that in the qpi pattern on bi(110 ) , only the scattering processes preserving the spin orientation are visible . one major difference ,
however , betwwen bi(110 ) and bi@xmath0te@xmath1 is that the former has multiple fermi surfaces and the scattering processes preserving spin orientations do exist at finite @xmath66 , while the later only has one fermi surface and therefore no such scatterings could exist . at the tips , the spin lies in - plane , with @xmath67 , independent of the scanning energy @xmath68 .
it can be checked that @xmath69 , hence @xmath70 . for scatterings between the valleys
, @xmath65 depends crucially on @xmath63 and @xmath71 .
accounting for the valley - to - valley oscillation in @xmath72 , we conclude that as the scanning energy increases , @xmath73 grows while @xmath74 shrinks .
this simple argument gives a qualitative explanation for the absence of the @xmath53 scattering in the stm experiment @xcite .
for typical experimental parameters @xcite , @xmath75 and @xmath76 . in this case
we estimate the scalar potential scattering gives that @xmath77 .
to specifically compute the qpi image , we employ a @xmath2-matrix approach @xcite for multiband systems @xcite . in the operator basis @xmath78
, the green s function is written in matrix form as ( , , ) & = & _ 0 ( , ) _ , + _ 0 ( , ) _ , ( ) _ 0 ( , ) + [ gtg ] where the @xmath2-matrix satisfies _ , ( ) = _ , + d^2p _ , _ 0 ( , ) _ , ( ) , [ tmatrix ] and @xmath79_{ab}= \big[\omega+i\delta - e_a(\vk)\big]^{-1}\delta_{a , b}$ ] are the bare green s functions . in spectroscopic
imaging stm @xcite , the conductance ( @xmath80 ) measured by the stm is proportional to the local density of states defined as ( , ) = _ ( , ) + _ ( , ) , where @xmath81 is the local density of states for spin @xmath82 .
the qpi image in the brillouin zone @xmath83 is then obtained by performing the fourier transformation of the conductance @xmath80 . as a result
, we can calculate @xmath84 using the @xmath2-matrix formalism by : ( , ) & = & d^2r e^i ( , ) + & = & 12i d^2k [ rhoq ] where the trace is taken with respect to the matrix index . because physically stm measures the local density of states in the spin basis of @xmath85 , while our @xmath2-matrix theory here is developed in the eigenbasis of @xmath86 , the @xmath87 rotation matrices @xmath88 are introduced in the last line of eq .
[ rhoq ] to transform back to the physical spin basis . because the first term in eq
. [ gtg ] , @xmath89 contains the sum of the total density of states without the impurity , which makes it much larger than @xmath90 , we only plot @xmath91 in order to reveal weaker structures of the qpi induced by the impurity scattering .
we solve eq .
[ tmatrix ] numerically , using 2d polar coordinates . since the dominant scattering processes are between @xmath3 points on the constant energy contour @xmath92 ( we focus on @xmath93 here ) , we perform the integration within the range @xmath94 with @xmath32 and @xmath33 indicated in fig . [
fig : fs](a ) .
the resulting qpi images are plotted in fig .
[ fig : qpi ] for @xmath27 with @xmath23 fixed . for this choice of parameters , @xmath95 and @xmath96 . as shown in fig .
[ fig : qpi](a ) , @xmath97 and @xmath54 indicated by the red ( dark gray ) and green ( light gray ) circles are the strongest features while @xmath53 ( indicated by the white circle ) is almost invisible .
the reason why @xmath97 is even stronger than @xmath54 while they have comparable scalar scattering potential is due to the difference in the density of states . because the tip points shown in fig .
[ fig : fs](a ) have larger density of states than the valley points , the weights of @xmath97 is larger than those of @xmath54 , resulting in the stronger features observed for @xmath97 .
the strong features near @xmath98 correspond to small @xmath66 scatterings around the tips and valleys points , which have also be seen in experiments .
our results reproduce satisfactorily the experimental findings and are also consistent with the analysis from the spin - orientation selection rule discussed above .
as the scanning energy increases further , the surface states along the @xmath99 direction start to merge into the conduction band of the bulk states . in this case
, the tips of the constant energy contour will be mixed up with these bulk bands , which weakens the @xmath97 scattering but enhances the small @xmath100 scatterings near the @xmath6 point .
this is consistent with the experiment @xcite , showing that the area of the strong features near @xmath6 point becomes much larger after the scanning energy exceeds the bottom of the conduction band .
( color online ) the quasiparticle interference image for ( a ) @xmath27 and ( b ) @xmath101 with @xmath23 and @xmath102 . in this case ,
@xmath103 and @xmath96 .
( a ) the strongest large @xmath100 scatterings are @xmath97 and @xmath54 indicated by the red ( dark gray ) and green ( light gray ) circles ( and their symmetric points ) .
@xmath53 ( indicated by the white circle ) is too weak to be seen .
( b ) for @xmath101 , new qpi features with large momenta are visible . ]
now we briefly comment on the effect of the spin - orbit scattering given in eq .
[ hs ] which in principle exists in any realistic system .
since surface states of the topological insulator bi@xmath0te@xmath1 are two - dimensional , the spin - orbit scattering potential only has one component : ^so_imp = i c v_0 d^2kd^2kkk(- ) ( ) ^z ( ) .
backscattering is still forbidden because of the @xmath104 factor .
although @xmath64 does not flip spin , the angle - dependence @xmath104 gives rise to an additional suppression beyond that from the spin - orientation selection rule discussed in the case of scalar impurity scattering .
moreover , because the matrix element is linear in @xmath105 , the spin - orbit scattering tends to enhance the scatterings between quasiparticles with large momenta .
all these additional effects due to the spin - orbit scattering can be roughly seen in a straightforward calculation froim eq .
[ vkk ] : |v^(11)_a|a|^2 & = & |v^(11)_w|w|^2 = 0 + interferences which could lead to unusual suppressions or enhancements for some scattering wavevectors , depending not only on the magnitude and sign of @xmath25 , but also on the scanning energy @xmath68 . in fig .
[ fig : qpi](b ) we show the qpi image for @xmath101 . while the main features are still similiar to those of fig .
[ fig : qpi](a ) , new prominent features associated with larger momentum scatterings are visible .
since the matrix elements for spin - orbit scattering are larger for quasiparticles with larger momentum , this term will become more and more important as the scanning energy @xmath68 increases .
a detailed analysis of the spin - orbit scattering will be presented in a future publication . in comparison with the results in ref .
@xcite , we find that spin - orbit scattering from the impurity of the ag atom is not very important in this particular experiment .
in conclusion , we have analyzed the quasiparticle interference induced by nonmagnetic impurities on the surface of the topological insulator bi@xmath0te@xmath1 using a @xmath2-matrix approach . while the backscattering is completely forbidden by time - reversal symmetry ,
other scatterings are allowed , resulting in the qpi patterns observed in stm experiments @xcite .
we have shown further that the scattering strengths depends crucially on the spin orientations of the eigenfunctions .
since nonmagnetic impurities can not flip spin , the scalar scattering potential between two eigenstates is larger as their spin overlap is larger .
combined with the variation of the density of states , we have shown that some of the scatterings might be too weak to be seen in comparison with the strongest ones , and our results successfully reproduce the qpi patern observed in experiments .
we have further discussed the effect of the spin - orbit scattering on the qpi pattern . while the backscattering is still forbidden , we find that the spin - orbit scattering enhances several new features at large momentum , and the detailed qpi features strongly depends on the sign and strength of the spin - orbit scattering potential
. we are grateful to xi chen , liang fu , aharon kapitulnik , qin liu , xiaoliang qi , qikun xue for insightful discussions .
cw and wcl are supported by aro - w911nf0810291 .
s cz is supported by the department of energy , office of basic energy sciences , division of materials sciences and engineering , under contract de - ac02 - 76sf00515 . | the quasiparticle interference of the spectroscopic imaging scanning tunneling microscopy has been investigated for the surface states of the large gap topological insulator bi@xmath0te@xmath1 through the @xmath2-matrix formalism .
both the scalar potential scattering and the spin - orbit scattering on the warped hexagonal isoenergy contour are considered . while backscatterings are forbidden by time - reversal symmetry ,
other scatterings are allowed and exhibit strong dependence on the spin configurations of the eigenfunctions at @xmath3 points over the isoenergy contour .
the characteristic scattering wavevectors found in our analysis agree well with recent experiment results . |
estimating extreme risks in a multivariate framework is highly connected with the estimation of the extremal dependence structure . this structure can be described _ via _ the stable tail dependence function ( s.t.d.f . )
@xmath0 , first introduced by @xcite . for any arbitrary dimension @xmath1 , consider a multivariate vector @xmath2 with continuous marginal cumulative distribution functions ( c.d.f . )
the s.t.d.f .
is defined for each positive reals @xmath4 as @xmath5\\[-8pt]\nonumber & & \qquad = l(x_1,\ldots , x_d).\end{aligned}\ ] ] assuming that such a limit exists and is nondegenerate is equivalent to the classical assumption of existence of a multivariate domain of attraction for the componentwise maxima ; see , for example , @xcite , chapter 7 .
the previous limit can be rewritten as @xmath6 = l(x_1,\ldots , x_d),\ ] ] where @xmath7 denotes the multivariate c.d.f . of the vector @xmath2 , and for @xmath8 , @xmath9 .
consider a sample of size @xmath10 drawn from @xmath7 and an intermediate sequence , that is to say a sequence @xmath11 tending to infinity as @xmath12 , with @xmath13 .
denote by @xmath14 a vector of the positive quadrant @xmath15 and by @xmath16 the @xmath17th order statistics among @xmath10 realizations of the margins @xmath18 .
the empirical estimator of @xmath19 is obtained from ( [ eql ] ) , replacing @xmath7 by its empirical version , @xmath20 by @xmath21 , and @xmath22 for @xmath23 by its empirical counterpart @xmath24,n}$ ] , so that @xmath25 + 1,n}~\mathrm{or}~\ldots~\mathrm{or}~x^{(d)}_i \geq x^{(d)}_{n-[kx_d]+1,n } \ } } .\ ] ] see @xcite for pioneering works on this estimator . under suitable conditions , it can be shown ( see section [ secnotation ] ) that the estimator @xmath26 has the following asymptotic expansion : @xmath27 where @xmath28 is a continuous centered gaussian process , @xmath29 is a function that tends to 0 at infinity and @xmath30 is a continuous function . in particular@xmath31
can be approximated in distribution by @xmath32 , provided that @xmath33 tends to 0 as @xmath10 tends to infinity .
this condition imposes a slow rate of convergence of the estimator @xmath26 , so one would be interested in relaxing this hypothesis . as a counterpart ,
as soon as @xmath34 tends to a nonnull constant @xmath35 , an asymptotic bias appears and is explicitely given by @xmath36 .
the aim of this paper is to provide a procedure that reduces the asymptotic bias .
the latter will be estimated and then subtracted from the empirical estimator .
this kind of approach has been considered in the univariate setting for the bias correction of the extreme value index with unknown sign by @xcite .
refer also to @xcite ( @xcite ) @xcite , @xcite and @xcite for previous contributions on this problem .
note finally that the case of dependent sequences has been recently studied by @xcite .
the nonparametric estimation of the extremal dependence structure has been widely studied in the bivariate case ; see , for instance , @xcite , @xcite , @xcite , @xcite , @xcite and @xcite .
bias correction problems in the bivariate context received less attention than in the univariate setting .
to the best of our knowledge , it seems to be reduced to @xcite and @xcite , who consider the estimation of bivariate joint tails , which differs slightly from our task . as for the multivariate framework
, @xcite introduces the empirical estimator .
general approaches under parametric assumptions on the function @xmath0 have been developed , for example , by @xcite , @xcite , @xcite and @xcite .
apparently , no procedure correcting the bias can be found in the literature for dimension greater than two .
the objective of this article is to fill this gap .
note that our method does not only consist of applying the univariate bias procedure at several points .
indeed , the bias is no longer a parametric function , so that the new feature is mainly the fact that we are able to estimate and then subtract a function with an unknown form .
two families of asymptotically unbiased estimators of the s.t.d.f . are proposed , and their theoretical behaviors are studied .
a practical advantage of these new estimators is that they can be aggregated , thus reducing the variability .
the paper is organized as follows : section [ secnotation ] contains hypotheses and first results .
the bias reduction procedure is described in section [ secprocedure ] , and the main theoretical results are presented therein .
several theoretical models are exhibited in section [ secexamples ] that satisfy the required assumptions .
section [ secsimu ] illustrates the performance of the new estimators on both simulated and real data .
the estimation of side components is postponed up to section [ secrho ] .
the proofs are relegated to section [ secproofs ] .
let @xmath37 be independent and identically distributed multivariate random vectors with c.d.f . @xmath7 and continuous marginal c.d.f.s @xmath38 for @xmath39 .
suppose @xmath7 is in the domain of attraction of an extreme value distribution with c.d.f .
we recall that it supposes the existence for @xmath39 of sequences @xmath41 , @xmath42 of real numbers and a c.d.f .
@xmath40 with nondegenerate marginals such that @xmath43 for all points @xmath44 where @xmath40 is continuous .
denote by @xmath45 the @xmath46th marginal c.d.f .
of @xmath40 .
it is possible to show that the domain of attraction condition can be expressed as condition ( [ eql ] ) along with the convergence of the marginal distributions to the @xmath45 s , and that @xmath47 let @xmath48 be the measure defined by @xmath49 where @xmath50 : there exists @xmath46 such that @xmath51 for any vector @xmath52 .
several conditions are now described .
the first two have been introduced by @xcite : * the _ first - order condition _ consists of assuming that the limit given in ( [ eql ] ) exists , and that the convergence is uniform on any @xmath53^d$ ] , for @xmath54 . *
the _ second - order condition _ consists of assuming the existence of a positive function @xmath29 , such that @xmath55 as @xmath56 , and a nonnull function @xmath30 such that for all @xmath44 with positive coordinates , @xmath57 - l({\mathbf x } ) \bigr\ } \nonumber\\[-8pt]\\[-8pt]\nonumber & & \qquad = m({\mathbf x}),\end{aligned}\ ] ] uniformly on any @xmath53^d$ ] , for @xmath54 . *
the _ third - order condition _ consists of assuming the existence of a positive function @xmath58 , such that @xmath59 as @xmath56 , and a nonnull function @xmath60 such that for all @xmath44 with positive coordinates , @xmath61 - l({\mathbf x})}{\alpha(t ) } - m({\mathbf x } ) \biggr\ } \nonumber\\[-8pt]\\[-8pt]\nonumber & & \qquad = n({\mathbf x}),\hspace*{-10pt}\end{aligned}\ ] ] uniformly on any @xmath53^d$ ] , for @xmath54 .
this implicitly requires that @xmath60 is not a multiple of the function @xmath30 ; see remark [ rmmn ] .
[ rkhomol ] the function @xmath0 defined by ( [ eql ] ) and that appears in ( [ eq2ndorder ] ) and ( [ eq3rdorder ] ) is homogeneous of order 1 .
we refer , for instance , to @xcite , pages 213 and 236 .
most of the estimators constructed in this paper use the homogeneity property .
note that pointwise convergence in ( [ eql ] ) entails uniform convergence on the square @xmath53^d$ ] .
see , for instance , @xcite , page 237 .
[ rmmn ] if @xmath62 for some constant @xmath63 , the relation can be reformulated as @xmath64 - l({\mathbf x } ) } { \alpha(t)(1+c\beta(t ) ) } - m ( { \mathbf x } ) \biggr\ } \\ & & \qquad = 0,\end{aligned}\ ] ] which we want to exclude .
we refer to @xcite [ ( @xcite ) , page 385 ] , to see that the same complication turns up in the one - dimensional case .
[ rkpropmn ] the functions @xmath30 and @xmath60 involved in the second and third - order conditions satisfy some usual properties ; see , for example , @xcite . more specifically , one can show that there exist nonpositive reals @xmath65 and @xmath66 such that @xmath67 ( resp . , @xmath68 ) is a regularly varying function of order @xmath65 ( resp . , @xmath69 ) , that is , @xmath70 when @xmath71 , for each positive @xmath72 . besides , @xmath30 is homogeneous of order @xmath73 , that is to say @xmath74 , for each positive @xmath75 and @xmath44 with positive coordinates . finally , the function @xmath60 is homogeneous of order @xmath76 .
[ rkai ] an interesting situation to consider is when the c.d.f .
@xmath7 is in the domain of attraction of an extreme value distribution @xmath40 with independent components , that is , @xmath77 .
such a c.d.f .
is said to have the property of asymptotic independence . in this case , the function @xmath30 is the limit of the joint tail of the distribution , and in dimension 2 , the coefficient of tail dependence @xmath78 introduced by @xcite equals @xmath79 , where @xmath65 is defined in remark [ rkpropmn ] . in this paper
, we will handle two sets of assumptions .
first consider : * the second - order condition is satisfied , so that ( [ eq2ndorder ] ) holds ; * the coefficient of regular variation @xmath65 of the function @xmath29 defined in ( [ eq2ndorder ] ) is negative ; * the function @xmath30 defined in ( [ eq2ndorder ] ) is continuous . these hypotheses allow us to get the asymptotic uniform behavior of @xmath80 , the empirical estimator of @xmath0 defined by ( [ eq1storderhat ] ) , as detailed in the following proposition .
[ propcv - ps - l ] let @xmath81 be independent multivariate random vectors in @xmath82 with common joint c.d.f . @xmath7 and continuous marginal c.d.f.s @xmath38 for @xmath39 .
assume that the set of conditions holds .
suppose further that the first - order partial derivatives of @xmath0 ( denoted by @xmath83 for @xmath39 ) exist and that @xmath84 is continuous on the set of points @xmath85 .
consider @xmath80 the estimator of @xmath0 defined by ( [ eq1storderhat ] ) where @xmath17 is such that @xmath86 .
then as @xmath10 tends to infinity , we get @xmath87 under stronger assumptions , and for some choice of the intermediate sequence , the asymptotic distribution of the previous stochastic process can be obtained after multiplication by the rate @xmath88 . for a positive @xmath89 ,
let @xmath90^d)$ ] be the space of real valued functions that are right - continuous with left - limits . now introduce the conditions : * the third - order condition is satisfied , so that ( [ eq2ndorder ] ) and ( [ eq3rdorder ] ) hold ; * the coefficients of regular variation @xmath65 and @xmath66 of the functions @xmath29 and @xmath58 defined in ( [ eq2ndorder ] ) and ( [ eq3rdorder ] ) are negative ; * the function @xmath30 defined in ( [ eq2ndorder ] ) is differentiable and @xmath60 defined in ( [ eq3rdorder ] ) is continuous .
[ propdev - asympt - l ] assume that the conditions of proposition [ propcv - ps - l ] are fulfilled and that the set of conditions hold .
consider @xmath80 the estimator of @xmath0 defined by ( [ eq1storderhat ] ) where @xmath17 is such that @xmath86 and @xmath91 .
then as @xmath10 tends to infinity , @xmath92 in @xmath90^d)$ ] for every @xmath54 , where@xmath93 the process @xmath94 above is a continuous centered gaussian process with covariance structure @xmath95= \mu\{r({\mathbf x } ) \cap r({\mathbf y})\}$ ] given in terms of the measure @xmath48 defined by ( [ eqmu ] ) and of @xmath96 : there exists @xmath46 such that @xmath97 .
[ rkasymptbias ] a difference between the previous result and theorem 7.2.2 of @xcite consists of the choice of the intermediate sequence that is larger here .
indeed , we suppose @xmath98 whereas they choose @xmath99 , which implies @xmath100 . our choice requires the more informative second - order condition ( [ eq2ndorder ] ) .
a nonnull asymptotic bias appears in our framework .
the conditions on @xmath17 , @xmath29 and @xmath58 required in proposition [ propdev - asympt - l ] are not too restrictive : because of the regular variation of @xmath29 and @xmath58 , they are implied by the choice @xmath101 , with @xmath102 .
as pointed out in remark [ rkasymptbias ] , a nonnull asymptotic bias @xmath103 appears from proposition [ propdev - asympt - l ] .
the bias reduction procedure will consist of subtracting the estimated asymptotic bias obtained in section [ subsecmethoda ] .
the key ingredient is the homogeneity of the functions @xmath0 and @xmath30 mentioned in remarks [ rkhomol ] and [ rkpropmn ] .
this homogeneity will also provide other constructions to get rid of the asymptotic bias .
equation ( [ eqasympt - dev - l ] ) suggests a natural correction of @xmath80 as soon as an estimator of @xmath103 is available . in order to take advantage of the homogeneity of @xmath0 ,
let us introduce a positive scale parameter @xmath104 which allows to contract or to dilate the observed points .
we denote @xmath105 and @xmath106 from ( [ eqasympt - dev - l ] ) one gets @xmath107 in @xmath90^d)$ ] for every @xmath54 .
equations ( [ eqdelta ] ) and proposition [ propcv - ps - l ] yield as @xmath10 tends to infinity , @xmath108 fixing @xmath104 such that @xmath109 , a natural estimator of the asymptotic bias of @xmath26 is thus @xmath110 , where @xmath111 is an estimator of @xmath65 . recall that the unknown parameter @xmath65 is the regular variation index of the function @xmath29 involved in the - second order condition . let @xmath112 be an intermediate sequence that represents the number of order statistics used in the estimator @xmath111
. assume that @xmath113 where @xmath11 is the sequence used in proposition [ propdev - asympt - l ] .
a first asymptotically unbiased estimator of @xmath114 can be defined as @xmath115 the asymptotic behavior of this estimator is provided in theorem [ thmbiasb ] and remark [ rkcp ] .
we refer the reader to section [ secrho ] for more details concerning the estimation of @xmath65
. the previous construction can be easily generalized by correcting the estimator @xmath116 instead of @xmath117 .
indeed , from ( [ eqdev - asympt - la - l ] ) one can see that the asymptotic bias of @xmath118 is @xmath119 . recall that when @xmath10 tends to infinity , one has for any positive real @xmath120 , @xmath121 thus fixing @xmath120 such that @xmath122 will help to cancel the asymptotic bias .
it yields the following asymptotically unbiased estimator of @xmath0 : @xmath123 [ thmbiasb ] assume that the conditions of proposition [ propdev - asympt - l ] are fulfilled , and consider the estimator of @xmath0 defined by ( [ defestimb ] ) .
let @xmath112 be an intermediate sequence such that @xmath124 converges in distribution .
suppose also that @xmath17 is such that @xmath125 , @xmath126 and @xmath127 . under these assumptions , as @xmath10 tends to infinity , @xmath128 in @xmath90^d)$ ] for every @xmath54 , where @xmath129 is a continuous centered gaussian process defined by @xmath130 with covariance @xmath131 = { \mathbb e}[z_l({\mathbf x } ) z_l({\mathbf y } ) ] ( 1 - b^{-1/2 } + a^{-1/2 } ) ^2 $ ] and @xmath132 .
[ rkhypo4 ] the assumption that @xmath133 converges in distribution will be reconsidered in section [ secrho ] .
[ rkcp ] theorem [ thmbiasb ] remains true when @xmath134 and thus characterizes the asymptotic behavior of the estimator given in ( [ defestima ] ) .
for this particular choice of @xmath104 , the covariance reduces to @xmath135(2 - 2^{1/{2\rho}})^2 $ ] .
the procedure of bias reduction introduced in the previous section requires the estimation of the second - order parameter @xmath65 .
it is actually possible to avoid it , making use of combinations of estimators of @xmath0 .
the asymptotic bias of @xmath136 is @xmath137 , as already noted from ( [ eqdev - asympt - la - l ] ) . making use of ( [ eqdeltacv ] ) and homogeneity of @xmath30 ,
one gets as @xmath10 tends to infinity , @xmath138 for any intermediate sequence @xmath112 that satisfies @xmath139 .
the expression @xmath140 can thus be used as an estimator of the asymptotic bias of @xmath118 .
after simplifications , this leads to a new family of asymptotically unbiased estimators of @xmath114 by substracting the estimated bias from @xmath141 , namely @xmath142 which is well defined for any real number @xmath104 such that @xmath143 .
[ thmbias - lu ] assume that the conditions of proposition [ propdev - asympt - l ] are fulfilled , and consider the estimator of @xmath0 defined by ( [ defestimc ] ) .
let @xmath112 be an intermediate sequence such that @xmath124 converges in distribution .
suppose also that @xmath17 is such that @xmath125 , @xmath126 , @xmath144 and @xmath127 .
assume moreover that the function @xmath30 never vanishes except on the axes .
then , as @xmath10 tends to infinity , @xmath145 in @xmath146^d)$ ] for every @xmath147 and @xmath54 , where @xmath148 is a continuous centered gaussian process with covariance @xmath149 $ ] given by @xmath150\times\break ( a^{-\rho } -1)^{-2 } ( a^{-\rho } - a^{-1/2 } ) ^2 $ ] .
the covariance function specified above is decreasing with respect to the parameter @xmath104 for any fixed value of @xmath65 .
this suggests at first glance to choose @xmath104 close to 1 in order to reduce the asymptotic variance of @xmath151 , but this would give a degenerate form of ( [ defestimc ] ) .
see section [ secsimu ] for practical considerations for the choice of @xmath104 .
the aim of this section is to furnish several multivariate distributions that satisfy the third - order condition ( [ eq3rdorder ] ) . for the sake of simplicity
, expressions are displayed in the bivariate setting .
we start by focusing on heavy - tailed margins . in this case , a first possible step to get the pointwise convergence is to obtain , for well - chosen positive reals @xmath152 and @xmath153 , an expansion ( for @xmath20 tending to infinity ) of the form @xmath154 with @xmath155 .
one can then identify each term involved in ( [ eq3rdorder ] ) as follows : @xmath156\\[-8pt]\nonumber n(x , y ) & = & t_3\bigl(a(x ) , b(y ) \bigr),\end{aligned}\ ] ] where @xmath157 applying @xcite [ ( @xcite ) , corollary 5.18 ] , one can check that in such a framework a form of the bivariate extreme value distribution @xmath40 is given by @xmath158 following the idea of @xcite [ ( @xcite ) , pages 276 and 286 ] consider first a norm @xmath159 , and a cone @xmath160 of @xmath161 , that is to say , a set such that if @xmath162 , then @xmath163 for every positive @xmath20 . without loss of generality ,
suppose that @xmath164 .
let @xmath165 be a bivariate random vector with probability density function given by @xmath166 where @xmath63 is a normalizing positive constant and where @xmath29 and @xmath58 are some positive real numbers such that @xmath167 .
set @xmath168 or @xmath169 , and define @xmath170 .
one can check that for @xmath171 , @xmath172 where @xmath173 , @xmath174 and @xmath175 .
the functions @xmath30 and @xmath60 are homogeneous with order given through @xmath176 .
let us discuss some particular choices of the norm : * for the @xmath177-norm and @xmath178 , the model coincides with the bivariate pareto of type ii distribution , denoted by @xmath179 in this paper , and referred to as mp@xmath180 in @xcite , page 604 . in this case , @xmath181 , and @xmath182 .
the latter s.t.d.f .
is known as the negative logistic model , introduced by @xcite ; see also @xcite , page 307 .
* when the euclidean norm is chosen , one recovers the bivariate cauchy distribution for @xmath183 , @xmath184 and @xmath185 . on the positive quadrant , that means for @xmath186
, we have @xmath187 , @xmath188 and @xmath189 . on the whole plane , which means that @xmath190 , we get @xmath191 , @xmath192 and @xmath193 .
this can also be seen as a particular case of the following item .
* the student distributions with pearson correlation coefficient @xmath194 arise choosing the norm @xmath195 , for a positive real number @xmath196 , @xmath183 , @xmath197 and @xmath198 . in this case , the integral form of the function @xmath199 can not be totally simplified , and one classically writes the s.t.d.f . as @xmath200 , \ ] ] where @xmath201 is the c.d.f .
of the univariate student distribution with @xmath202 degrees of freedom .
this dependence structure is also obtained for some elliptical models ; see , for example , [ @xcite , page 1813 ] and next subsection . * other choices for the norm would lead to other distributions .
note that one can also relax the symmetry condition , considering , for instance , the mahalanobis pseudo - norm defined by @xmath203 for a real number @xmath65 such that @xmath204 and some positive real numbers @xmath205 and @xmath206 .
consider the usual representation of the centered elliptical distribution @xmath207 , in terms of a positive random variable @xmath208 , a @xmath209 matrix @xmath210 such that @xmath211 is of full rank , and a bivariate random vector @xmath212 independent of @xmath208 , uniformly distributed on the unit circle of the plane .
assume that @xmath208 has a probability density function denoted by @xmath213 .
one can then express the probability density function of @xmath165 as @xmath214 a sufficient condition to satisfy ( [ eq3rdorder ] ) is to assume that the distribution of @xmath208 belongs to the hall and welsch class [ @xcite ] , namely , @xmath215 with positive real @xmath63 , nonnull reals @xmath216 and @xmath217 and negative reals @xmath65 and @xmath218 . one can check that , for @xmath171 , @xmath219 where @xmath220 and @xmath221 .
assuming for simplicity that @xmath222 , the s.t.d.f . can be written as @xmath223,\end{aligned}\ ] ] which is the form already obtained for the student distribution in section [ subsecresnick - style ] for @xmath224 .
see @xcite for more details .
note finally that for a general matrix @xmath225 and the special case @xmath226 , one recovers the mahalanobis pseudo - norm already mentioned in the previous subsection .
when dealing with margins that are _ not _ heavy tailed , the calculus is done directly from ( [ eq2ndorder ] ) .
the last two examples of bivariate distributions have short and light tailed margins , respectively .
consider the bivariate c.d.f .
defined for each @xmath227 by @xmath228 given in terms of a s.t.d.f .
this distribution has standard uniform univariate margins and corresponds to a particular case of archimax bivariate copulas introduced in @xcite , in which the function @xmath229 is the clayton archimedean generator with index 1 .
expanding the left - hand side term of ( [ eq2ndorder ] ) leads to , as @xmath20 tends to infinity , @xmath230 where @xmath231 this allows us to identify @xmath232 . above
, the notation @xmath233 stands for @xmath234 .
consider the c.d.f .
defined by @xmath235 for each @xmath236 , which corresponds to the bivariate extreme value distribution with gumbel univariate margins @xmath237 and symmetric logistic s.t.d.f .
@xmath238 , where @xmath239 .
this distribution was introduced in @xcite ; see , for example , @xcite , page 304 .
expanding @xmath240 $ ] leads to @xmath241 where @xmath242 this allows us to identify @xmath232 .
the identification of second and third - order terms has previously be derived by @xcite .
the purpose of this section is to evaluate the performance of the estimators of @xmath0 introduced in section [ secprocedure ] . for simplicity
, we will focus on dimension 2 , and simulate samples from the distributions presented in section [ secexamples ] .
thanks to the homogeneity property , one can focus on the estimation of @xmath243 for @xmath244 , which coincides with the pickands dependence function @xmath245 ; see , for example , @xcite , page 267 .
considering first the estimation at @xmath246 leads to the definition of aggregated versions of our estimators .
these new estimators will be both compared in terms of @xmath177-errors for @xmath0 or associated level curves .
let us start with the estimation of @xmath247 for the bivariate student distribution with 2 degrees of freedom .
this model is a particular case of sections [ subsecresnick - style ] and [ subsecelliptical ] . for one sample of size 1000
, figure [ sec - simu1-stu2 ] gives , as functions of @xmath17 , the estimation of @xmath0 at point @xmath248 by @xmath117 , @xmath249 and @xmath250 , respectively , defined by ( [ eq1storderhat ] ) , ( [ defestimb ] ) and ( [ defestimc ] ) . for the last two estimators ,
the parameters have been tuned as follows : @xmath251 , @xmath252 and @xmath65 estimated using ( [ defrho - hat ] ) with @xmath253 .
these values have been empirically selected based on intensive simulation , and will be kept throughout the paper . for the bivariate @xmath254 law based on a sample of size 1000 . ]
one can check from figure [ sec - simu1-stu2 ] that the empirical estimator @xmath117 behaves fairly well in terms of bias for small values of @xmath17 . besides
, the bias is efficiently corrected by the two estimators @xmath249 and @xmath250 .
since the bias almost vanishes along the range of @xmath17 , one can think about reducing the variance through an aggregation in @xmath17 ( via mean or median ) of @xmath249 or @xmath250 .
this leads us to consider the two following estimators : @xmath255 where @xmath10 is the sample size and @xmath256 is an appropriate fraction of @xmath10 .
their performance will be compared to those of the family @xmath257 .
simplified notation @xmath258 will be used instead of @xmath259 . because any s.t.d.f .
@xmath0 satisfies @xmath260 , the competitors have been corrected so that they satisfy the same inequalities .
if @xmath256 satisfies the condition imposed on @xmath261 in theorems [ thmbiasb ] and [ thmbias - lu ] , then the aggregated estimators @xmath262 and @xmath263 would inherit the asymptotic properties of @xmath249 and @xmath250 .
indeed , all the estimators jointly converge , since they are based on a single process .
[ remmixture ] in the following simulation study , @xmath256 is arbitrarily fixed to @xmath264 .
such a choice is open to criticism since it does not satisfy the theoretical assumptions mentioned in the previous remark .
but it is motivated here by the fact that the bias happened to be efficiently corrected , even for very large values of @xmath17 , as already illustrated on figure [ sec - simu1-stu2 ] .
note , however , that such a choice would not be systematically the right one . in presence of more complex models such as mixtures ,
@xmath256 should not exceed the size of the subpopulation with heaviest tail . to illustrate this point ,
take , for example , the bivariate c.d.f .
@xmath265 , where @xmath40 is the c.d.f . of the bivariate @xmath266 model , and @xmath267 is the uniform c.d.f . on @xmath268
then the s.t.d.f . is @xmath269 , and only @xmath270 of the data belong to the targeted domain of attraction , so @xmath256 should not exceed @xmath271 .
classical criteria of quality of an estimator @xmath272 of @xmath273 are the absolute bias ( abias ) and the mean square error ( mse ) defined by @xmath274 where @xmath60 is the number of replicates of the experiment and @xmath275 is the estimate from the @xmath276th sample .
note that what we call _ abias _ is also referred as _ mae _ ( for mean absolute error ) in the literature .
figure [ figabias - mse - st ] plots these criteria in the estimation of @xmath247 for the bivariate @xmath254 model when @xmath277 and @xmath278 . in the bivariate @xmath254 model
when @xmath277 as a function of @xmath17 . ]
figure [ figabias - mse - st ] exhibits the strong dependence of the behavior of @xmath279 in terms of @xmath17 , as well as the efficiency of the bias correction procedures .
the estimator @xmath280 given by ( [ defestimb ] ) outperforms the estimator @xmath281 defined by ( [ defestimc ] ) , no matter the value of @xmath17 .
moreover , the abias and mse curves associated to @xmath280 almost reach the minimum of those of @xmath279 . finally , the aggregated version @xmath262 answers surprisingly well to the estimation problem of the s.t.d.f .
first , its performance is similar to the best reachable from the original estimator @xmath80 .
second , it gets rid of the delicate choice of a threshold @xmath17 ( or would at least simplify this choice ; see remark [ remmixture ] ) .
these comparisons have also been made for five other models obtained from section [ secexamples ] .
the results are very similar to the ones obtained for the bivariate @xmath254 distribution and are therefore not presented .
the comparisons are now handled not only at a single point , but for the whole function using an @xmath177-error defined as follows : @xmath282 where @xmath89 is the size of the subdivision of @xmath268 $ ] .
figure [ figl1-norm - l ] gives the boxplots based on @xmath283 realizations of @xmath284 and @xmath258 for @xmath285 in the case of six bivariate models : * first row : cauchy and @xmath254 models ; * second row : @xmath266 model and symmetric logistic model with @xmath286 ; * third row : archimax model with logistic generator @xmath287 and mixed generator @xmath288 .
-error of function @xmath0 for the estimators @xmath289 and @xmath258 .
first row : bivariate cauchy model ( left ) and bivariate @xmath254 model ( right ) .
second row : bivariate @xmath266 model ( left ) and bivariate symmetric logistic model ( right ) .
third row : bivariate archimax model with logistic ( left ) and mixed generator ( right ) . ] as already observed in figure [ figabias - mse - st ] , the estimator @xmath262 is again very competitive compared to the best element of @xmath258 , no matter the choice of model . recall that the value of @xmath17 leading to the best @xmath80 depends crucially on the model and is consequently unknown in practice , which invites any users to apply this new procedure .
the estimator @xmath290 is definitely less competitive compared to @xmath262 .
given these results we will not pursue with the @xmath290 estimator in the rest of this paper , and will focus our attention on the behavior of @xmath262 .
a natural step is now to compare the performance of our best estimator @xmath262 with an existing competitor , recently introduced by @xcite . in his work
, peng provides a data - driven method which chooses the threshold via estimating a s.t.d.f .
another interesting task is to compare @xmath262 with a convexified version of itself , since any s.t.d.f .
is a convex function ; see , for example , @xcite [ ( @xcite ) , section 8.2.2 ] or @xcite [ ( @xcite ) , section 6.1.5 ] .
note that a general convexification procedure has been proposed in dimension 2 by @xcite ; see also some alternative suggestions in @xcite . in order to take maximal advantage from this simulation study ,
the three different models implemented have been considered in two versions for each : the first model is the gaussian one , simulated with pearson s correlation coefficient @xmath291 .
the gaussian model is a particular case of elliptical distributions ( see section [ subsecelliptical ] ) , which illustrates the asymptotic independent situation ; cf .
remark [ rkai ] .
the second model is the bivariate symmetric logistic one , introduced in section [ subsecasymlog ] , with two different strengths of dependence ( close to independence on the left column , stronger dependence on the right column ) .
the third model is the bivariate student family , introduced in sections [ subsecresnick - style ] and [ subsecelliptical ] as a particular case .
two strengths of dependence have also been chosen , close to asymptotic independence on the left column and stronger dependence on the right column .
our results , summarized in figure [ figl1-norm - l - rev1 ] , will thus exhibit in particular how the performance in the estimation of the s.t.d.f .
depends on the distance to the asymptotic independence case .
the @xmath292-axis scale has been fixed for all the six cases so that one can measure that the estimation of the s.t.d.f .
is a more ambitious problem under asymptotic independence .
however , our estimator @xmath262 has still nice properties when comparing it to the empirical estimator @xmath279 .
the convex version @xmath293 performs quite equivalently as @xmath262 .
a reason for this is that by construction our estimator is actually not far from a convex function .
so balancing the cost of convexifying with the benefit in the performance motivates the simple use of @xmath262 .
finally , regarding peng s estimator @xmath294 , one observes that this estimator is an interesting alternative to the original family @xmath295 , which , however , never outperforms our proposal .
-error of function @xmath0 for the estimators @xmath296 and @xmath297 .
first row : bivariate normal model with correlation @xmath206 : @xmath298 ( left ) and @xmath299 ( right ) .
second row : bivariate symmetric logistic@xmath300 model : @xmath301 ( left ) and @xmath286 ( right ) .
third row : bivariate @xmath302 model : @xmath303 ( left ) and @xmath304 ( right ) . ]
let us illustrate in this subsection the question of estimating an arbitrarily chosen failure probability @xmath305 or @xmath306 , where @xmath165 comes from the @xmath266 model , so that @xmath305 or @xmath307 .
samples of size @xmath277 are considered .
thus empirical estimation will be useless for evaluating the probability of exceeding such extreme values for @xmath308 or @xmath309 , and an extrapolation based on extreme value theory is thus needed .
first assume that it is known that the margins are standard pareto .
this probability can be approximated by @xmath310 which naturally comes from ( [ eql ] ) , the projection on the simplex and the homogeneity of @xmath0 .
estimating the unknown parameter @xmath311 with our candidate @xmath262 and the original family @xmath258 gives several boxplots ( based on 500 replicates ) that are presented in figure [ figproba ] .
the comparison of these estimates is again favorable to @xmath262 . or @xmath312 when @xmath165 is drawn from the @xmath266 model with sample size @xmath277 and assuming margins to be known . ]
[ figproba ] we also investigated the possible use of a second - order term in the approximation of the probability @xmath305 or @xmath313 , making use of the following estimators @xmath314 the results were so similar to those obtained in figure [ figproba ] that we chose to skip them .
second , when the margins are not assumed to be known , the estimation of @xmath315 and @xmath316 can be reached by the pot method [ see , e.g. , @xcite , section 7.4 ] for several values of a threshold . after the study of mean residual life plots and quantile plots ,
the thresholds have been fixed to be @xmath317 and @xmath318 for @xmath319 .
the pot estimates deduced with these thresholds are , respectively , denoted by @xmath320 and @xmath321 .
the probability @xmath305 or @xmath322 is then approximated by @xmath323 estimating on each repetition the unknown parameter @xmath324 with our candidate @xmath262 and the original family @xmath297 gives several boxplots ( based on 500 replicates ) presented in figure [ figprobabis ] . or @xmath312 when @xmath165 is drawn from the @xmath266 model with sample size @xmath277 and estimating margins by pot method . ]
it seems clear that the uncertainty on the margins @xmath325 and @xmath326 has much more influence than that of the s.t.d.f . @xmath0 .
such findings corroborate previous studies ; see , for example , @xcite and @xcite .
another nice representation of a function of several variables is through its level sets . in the case of the function @xmath0 ,
it consists of looking ( for any positive real @xmath63 ) at sets of the form @xmath328 . from homogeneity property ,
it is characterized by @xmath329 following @xcite [ ( @xcite ) , page 245 ] , the boundary of this set can be written as @xmath330 \bigr\}.\ ] ] the estimation of @xmath331 is naturally obtained by replacing @xmath0 by any estimator , and this is done here for the estimators @xmath262 and @xmath258 .
figure [ figneptune ] ( left ) exhibits the bias phenomenon ( as @xmath17 increases ) induced by @xmath80 in the estimation of the @xmath327-curve .
the bias factor on @xmath80 is illustrated with @xmath332 and @xmath333 .
the correction of the bias with @xmath262 is effective . as in the previous section ,
the comparison of the different estimators is provided in terms of a global criterium based on the @xmath177-norm , given by @xmath334 figure [ figl1-norm - q ] displays the boxplots of this measure , based on @xmath283 realizations and for @xmath285 under the six bivariate models given in the previous section .
-curve for the bivariate @xmath254 law based on a sample of size 1000 .
right : estimated @xmath327-curve for the wave heights data introduced in @xcite . ]
the estimation of the @xmath327-curve based on the original estimator @xmath335 is strongly sensitive to the choice of @xmath17 : the bias ( resp . , the variability ) is an increasing ( resp . , decreasing ) function of @xmath17 .
the performances of @xmath262 is similar to that of the best @xmath80 , which is unknown in practice .
these features corroborate the conclusions drawn in section [ subsecl1 ] . to close this section , let us illustrate the @xmath327-curve estimation on the wave heights data set of @xcite , page 207 .
as explained therein , wave height ( hmo ) and still water level ( swl ) have been recorded during 828 storm events on the dutch coast .
the analogous figure 7.2 from @xcite is reported in figure [ figneptune ] ( right ) .
even if the two curves are not so close , the conclusion remains the same : the estimated boundary is concave , which indicates that the high values of the two variables are dependent .
-error of @xmath327-curve for the estimators @xmath336 and @xmath258 .
first row : bivariate cauchy model ( left ) and bivariate @xmath254 model ( right ) .
second row : bivariate @xmath266 model ( left ) and bivariate symmetric logistic model ( right ) .
third row : bivariate archimax model with logistic ( left ) and mixed generator ( right ) . ]
in this section , we focus on the estimation of the function @xmath30 coming from the second - order condition ( [ eq2ndorder ] ) and on the estimation of its homogeneity parameter @xmath73 . a possible way to estimate @xmath65 is to use on each margin one of the techniques developed in the univariate setting ; see , for example , @xcite or @xcite .
other methods make use of the multivariate structure of the data ; see , for example , @xcite and also @xcite in a slightly different framework .
the construction described here takes likewise advantage of the multivariate information of the sample . with this purpose
, the following proposition shows that a variable of interest is the ratio of two terms @xmath337 , defined by ( [ eqdelta ] ) .
[ proplim - quotient ] assume that the conditions of proposition [ propcv - ps - l ] are fulfilled and fix positive real numbers @xmath75 and @xmath338 .
assume moreover that the function @xmath30 never vanishes except on the axes .
then , as @xmath10 tends to infinity , for every @xmath147 and @xmath54 , @xmath339 if the requirement that the function @xmath30 is either positive , or negative in the positive quadrant does not hold , one could consider the integral of @xmath340 over the set @xmath341 s.t .
@xmath342 and prove a result like lemma [ lemdelta ] for this statistic .
then the integral of @xmath343 appears in the denominator in proposition [ proplim - quotient ] instead of @xmath30 itself , and the sign of @xmath30 does not matter
. this will be part of a future work .
a family of consistent estimators of the parameter @xmath65 can be derived from proposition [ proplim - quotient ] .
@xmath344 the following property can be obtained from the asymptotic expansion given in proposition [ propdev - asympt - l ] .
[ prop3estim - rho ] assume that the conditions of proposition [ propdev - asympt - l ] are fulfilled , and fix positive real numbers @xmath75 and @xmath338 .
consider the estimator of @xmath65 defined by ( [ defrho - hat ] ) .
assume moreover that the function @xmath30 never vanishes except on the axes .
then , as @xmath10 tends to infinity , @xmath345 in @xmath146^d)$ ] for every @xmath147 and @xmath54 , with @xmath346 figure [ figbox - rho ] illustrates the finite sample behavior of this estimator of @xmath65 for a collection of bivariate models introduced in section [ secexamples ] , for which the true value of @xmath65 is equal to @xmath3471 . given by ( [ defrho - hat ] ) using samples of size 1000 drawn from six models : @xmath254 ; @xmath266 ; symmetric logistic with @xmath286 ; archimax model with logistic generator with @xmath348 ; archimax model with mixed generator .
red line indicates the true value of @xmath349 . ]
these boxplots show that the estimator performs reasonably well in median , no matter the choice of model , but the uncertainty is rather important .
fortunately this seems from simulation studies to have only minor influence on the estimation of @xmath0 . recall that from ( [ eqdev - asympt - la - l ] ) the asymptotic bias of @xmath136 is given by @xmath350 . in order to circumvent an estimation of the term @xmath351 ,
a renormalization is needed , focusing , for instance , on the estimation of @xmath352 where @xmath353 .
thanks to ( [ eqdeltacv ] ) , this ratio can be consistently estimated by @xmath354 as soon as @xmath17 is a well - chosen intermediate sequence .
the asymptotic normality can also be derived from analogous arguments to those used in the proof of proposition [ prop3estim - rho ] .
details are not presented here for the sake of simplicity .
figure [ figl1-norm - m ] summarizes the behavior of the estimator of the curve @xmath355 through boxplots of the @xmath177-error , defined as in ( [ eqnorm1 ] ) .
we observe from this figure that the best estimation is reached for large values of @xmath17 .
this feature does not depend on the degree of asymptotic dependence in the symmetric logistic model , nor on the strength of the bias of the original estimator @xmath80 detected on figure [ figl1-norm - l ] .
these graphs confirm that the asymptotic bias is remarkably well estimated for large values of @xmath17 .
this helps to understand why the bias subtraction is accurate for large or very large choices of @xmath17 , as also commented in section [ subseccorrections ] .
-error of @xmath356-curve .
first row : bivariate logistic model with @xmath357 ( left ) and with @xmath358 ( right ) .
second row : bivariate logistic model with @xmath359 ( left ) and bivariate archimax with mixed generator ( right ) . ]
this paper deals with the estimation of the extremal dependence structure in a multivariate context . focusing on the s.t.d.f .
, the empirical counterpart is the nonparametric reference .
a common feature when modeling extreme events is the delicate choice of the number of observations used in the estimation , and it spoils the good performance of this estimator .
the aim of this paper has been to correct the asymptotic bias of the empirical estimator , so that the choice of the threshold becomes less sensitive .
two asymptotically unbiased estimators have been proposed and studied , both theoretically and numerically .
the estimator defined in section [ subsecmethodb ] proves to outperform the original estimator , whatever the model considered .
its aggregated version defined in section [ subseccorrections ] appears as a worthy candidate to estimate the s.t.d.f .
proof of proposition [ propcv - ps - l ] denote by @xmath360 the uniform random variables @xmath361 for @xmath39 .
introducing @xmath362 allows us to rewrite @xmath80 as the following : @xmath363,n } , \ldots , \frac{n}{k}u^{(d)}_{[kx_d],n } \biggr).\ ] ] write @xmath364,n } , \ldots , \frac{n}{k}u^{(d)}_{[kx_d],n } \biggr ) \\ & & \quad\qquad { } - \frac{n}{k } \bigl [ 1- f\bigl\{f_1^{-1}\bigl(1-u^{(1)}_{[kx_1],n } \bigr),\ldots , f_d^{-1}\bigl(1-u^{(d)}_{[kx_d],n } \bigr ) \bigr\}\bigr ] \\ & & \quad\qquad{}+ \frac{n}{k } \bigl [ 1- f\bigl\{f_1^{-1 } \bigl(1-u^{(1)}_{[kx_1],n}\bigr),\ldots , f_d^{-1 } \bigl(1-u^{(d)}_{[kx_d],n}\bigr ) \bigr\}\bigr ] \\ & & \quad\qquad { } - l \biggl ( \frac{n}{k } u^{(1)}_{[kx_1],n } , \ldots , \frac{n}{k}u^{(d)}_{[kx_d],n } \biggr ) \\ & & \quad\qquad { } + l \biggl(\frac{n}{k } u^{(1)}_{[kx_1],n } , \ldots , \frac { n}{k}u^{(d)}_{[kx_d],n } \biggr ) - l({\mathbf x}),\end{aligned}\ ] ] and denote @xmath365 [ resp . ,
@xmath366 and @xmath367 the first line ( resp . , second and third lines ) of the right - hand side .
applying @xcite [ ( @xcite ) , proposition 7.2.3 ] leads to @xmath368 in @xmath90^d)$ ] for every @xmath54 and for any intermediate sequence , where @xmath94 is a continuous centered gaussian process with covariance structure specified in proposition [ propdev - asympt - l ] . due to the skorohod construction we can write @xmath369 which implies , since @xmath370 , @xmath371 again for any intermediate sequence , the proof of @xcite [ ( @xcite ) , theorem 7.2.2 ] ensures the convergence for @xmath39 @xmath372 } \biggl{\vert}\sqrt{k } \biggl ( \frac{n}{k}u^{(j)}_{[kx],n } -x \biggr ) + w_l(x{\mathbf e}_j)\biggr{\vert}\to0 \qquad\mbox{a.s.},\ ] ] and finally @xmath373
as previously , this yields @xmath374 since the intermediate sequence satisfies @xmath375 , it thus remains to prove that @xmath376 the second - order condition that holds uniformly on @xmath53^d$ ] in ( [ eq2ndorder ] ) yields @xmath377,n } , \ldots , \frac{n}{k}u^{(d)}_{[kx_d],n } \biggr)\biggr{\vert}\to0 \qquad\mbox{a.s . }
\ ] ] then the result follows from @xmath378,n } , \ldots , \frac{n}{k}u^{(d)}_{[kx_d],n } \biggr)\biggr{\vert}\to0 \qquad\mbox{a.s . } , \ ] ] which is obtained combining ( [ eqeksmarg1 ] ) and the continuity of the function @xmath30 .
proof of proposition [ propdev - asympt - l ] we use the notation introduced in the proof of proposition [ propcv - ps - l ] .
thanks to the skorohod construction , we can start from ( [ eqskorohoda1 ] ) . combined with ( [ eqskorohoda3 ] ) , it is sufficient to prove the convergence @xmath379 note that the third - order condition , the uniformity on @xmath53^d$ ] of the convergence in ( [ eq3rdorder ] ) and the continuity of @xmath60 yield @xmath380,n } , \ldots , \frac{n}{k}u^{(d)}_{[kx_d],n } \biggr)+ o_{\mathbb p } \biggl(\alpha\biggl(\frac{n}{k}\biggr)\beta \biggl(\frac{n}{k}\biggr ) \biggr ) . \ ] ] thanks to ( [ eqeksmarg1 ] ) and to the existence of the first - order partial derivatives @xmath381 @xmath382 of the function @xmath30 , we have that @xmath383,n } , \ldots , \frac{n}{k}u^{(d)}_{[kx_d],n } \biggr ) - m({\mathbf x } ) \biggr\ } \\ & & \hspace*{121pt}\qquad { } + \sum_{j=1}^dw_l(x_j { \mathbf e}_j ) \partial_j m({\mathbf x } ) \biggr{\vert}\end{aligned}\ ] ] converges to 0 in probability , as @xmath10 tends to infinity .
this implies that @xmath384 which completes the proof , thanks to the choice of the intermediate sequence .
proof of theorem [ thmbiasb ] recall that @xmath385 , and denote @xmath386 .
write @xmath387 which equals , thanks to ( [ eqdev - asympt - la - l ] ) and under skorohod s construction , @xmath388 & & \quad { } - \alpha \biggl(\frac{n}{k } \biggr ) \hat b^{-\rho } m({\mathbf x } ) - \frac { b^{-1}}{\sqrt k } z_l(b { \mathbf x})+ o \biggl(\frac{1}{\sqrt{k } } \biggr ) \\[-1pt ] & & \qquad = \alpha \biggl(\frac{n}{k } \biggr ) \bigl ( \bigl(a^{-\rho } + 1 \bigr)-b^{-\rho } \bigr ) m({\mathbf x})+ \frac{1}{\sqrt k}{\accentset{\circ}{y}}_a({\mathbf x } ) \\[-1pt ] & & \quad\qquad { } + \alpha \biggl(\frac{n}{k } \biggr ) \bigl(b^{-\rho}-\hat b^{-\rho } \bigr)m({\mathbf x})+ o \biggl ( \frac{1}{\sqrt{k } } \biggr ) \\[-1pt ] & & \qquad = \alpha \biggl(\frac{n}{k } \biggr ) \bigl ( \bigl(a^{-\rho } + 1 \bigr)-b^{-\rho } \bigr ) m({\mathbf x})+ \frac{1}{\sqrt k}{\accentset{\circ}{y}}_a({\mathbf x } ) \\[-1pt ] & & \quad\qquad { } + \alpha \biggl(\frac{n}{k } \biggr)o_\mathbb{p } \biggl(\frac { 1}{\sqrt{k_\rho } \alpha(n / k_\rho ) } \biggr)+ o \biggl(\frac { 1}{\sqrt{k } } \biggr).\end{aligned}\ ] ] the first term is zero .
since both @xmath389 and @xmath29 is regularly varying with negative index , the only the last term can be put into the term @xmath390 .
finally , the covariance function follows from the equality in law as processes between @xmath391 and @xmath392 .
the proofs of theorem [ thmbias - lu ] and proposition [ prop3estim - rho ] are based on the following auxiliary result .
[ lemdelta ] assume that the conditions of proposition [ propdev - asympt - l ] are fulfilled .
then for any positive real @xmath75 , one has as @xmath10 tends to infinity , @xmath393 in @xmath90^d)$ ] for every @xmath54 .
proof of lemma [ lemdelta ] making use of the homogeneity of the function @xmath0 , write @xmath394 using the skorohod construction , it follows from equations ( [ eqasympt - dev - l ] ) and ( [ eqdev - asympt - la - l ] ) that @xmath395 tends to 0 almost surely , as @xmath10 tends to infinity
. proof of theorem [ thmbias - lu ] note that @xmath396 under a skorohod construction , lemma [ lemdelta ] allows us to write the expansions of the terms @xmath397 , @xmath398 and @xmath399 , which implies on the one hand @xmath400\\[-8pt]\nonumber & & \qquad\quad { } + \frac{1}{\sqrt{k_\rho } \alpha(n / k_\rho ) } \bigl\ { a^{-1 } z_l \bigl(a^2 { \mathbf x}\bigr ) -2 z_l(a { \mathbf x } ) + a z_l({\mathbf x } ) \bigr\ } \\ & & \quad\qquad { } + o \biggl ( \frac{1}{\sqrt{k_\rho } \alpha(n / k_\rho ) } \biggr),\nonumber\end{aligned}\ ] ] and @xmath401 on the other hand , both uniformly for @xmath402^d$ ] . combining ( [ eqdeltaterm1 ] ) and ( [ eqdeltaterm2 ] ) with equation ( [ eqasympt - dev - l ] ) , one gets @xmath403 since the last expression and equation ( [ eqdeltaterm1 ] ) are , respectively , the numerator and denominator of @xmath404 , one obtains , after simplification , @xmath405 since @xmath30 does not vanish by assumption .
the choice of the sequences @xmath17 and @xmath112 allows us to conclude since @xmath406 .
proof of proposition [ proplim - quotient ] applying lemma [ lemdelta ] , we have @xmath407 as a consequence , @xmath408 since @xmath409 by assumption .
writing @xmath410 and using twice equation ( [ eqsupdelta ] ) leads to the conclusion .
proof of proposition [ prop3estim - rho ] define @xmath411 .
lemma [ lemdelta ] used twice yields @xmath412 where @xmath413 is defined in proposition [ prop3estim - rho ] . since @xmath414 ,
the result follows straightforwardly from ( [ eqdev - asympt - q ] ) and the delta method .
we wish to thank armelle guillou for pointing out a deficiency in the original version of the paper , as well as several misprints . we thank the referees for very helpful comments . | the estimation of the extremal dependence structure is spoiled by the impact of the bias , which increases with the number of observations used for the estimation .
already known in the univariate setting , the bias correction procedure is studied in this paper under the multivariate framework .
new families of estimators of the stable tail dependence function are obtained .
they are asymptotically unbiased versions of the empirical estimator introduced by huang [ statistics of bivariate extremes ( 1992 ) erasmus univ . ] . since the new estimators have a regular behavior with respect to the number of observations , it is possible to deduce aggregated versions so that the choice of the threshold is substantially simplified .
an extensive simulation study is provided as well as an application on real data .
, + |
[ i ] second harmonic generation ( shg ) has nowadays quite a long tradition as a mean of squeezed light generation @xcite .
the preferred experimental setup has been the doubly resonant configuration as , at least in principle , permits arbitrarily large squeezing .
however , such scheme has been hampered by the technical difficulties arising from keeping the resonance in both modes simultaneously .
thus , in spite of the development of very ingenious stabilizing procedures @xcite , for the moment it has been only possible to maintain the double resonance for a few seconds . certainly , this kind of experimental delicacy can hardly surprise when dealing with the generation of non - classical states of light . in view of such difficulties ,
some experimental efforts have been recently redirected to singly resonant configurations @xcite .
although , the maximum noise suppression is then limited to a 90% @xcite , the efforts resulted in very stable intense squeezed light sources with degrees of squeezing even surpassing those reported in the doubly resonant counterparts @xcite .
this evolution highlights the importance of reducing to a minimum the technical demanding of new proposals in a so experimentally challenging field . at the same time , singly resonant optical parametric oscillation ( opo ) , the most successful method to squeeze the vacuum @xcite , has been generalized to singly resonant optical parametric amplification ( opa ) , i.e. a laser driving in the harmonic mode has been added , again showing an extraordinary stability at quite high noise suppression values in the fundamental mode @xcite . although the squeezed beams are in this case much less intense than the in shg counterpart , this setup permits a control of the phase of the squeezed quadrature , something which allowed a spectacular demonstration using quantum tomography , of the different kinds of squeezed states @xcite . in view of this experimental success
it seems timely to extend the quantum mechanical model beyond the pure phase matched cases .
more specifically , we address here quantum noise reduction in an extension of the conventional singly resonant shg to include also a coherent input in the harmonic mode as well as phase mismatch between the interacting waves .
on the other hand , an increasingly number of papers is being devoted to study quantum noise in systems combining different kinds of nonlinearities ( see , for instance , @xcite for some recent contributions ) . in particular , the combination of @xmath0 with kerr - like @xmath1 nonlinearities in c.w .
cavity systems has been quite extensively studied @xcite .
even exact full quantum results have been obtained showing , for instance , the emergence of tristability not present in the classical counterpart @xcite . with respect to the squeezing performance
the results appear as very promising at least in degenerate doubly resonant configurations @xcite . the simplest system from the implementation point of view , combining this two kinds of nonlinearities that the authors can think of is precisely a singly resonant second order nonlinear system with phase mismatched interacting waves as then , by virtue of the cascading effect , an effective kerr - like third order nonlinearity appears . in order to make the search of strong noise reduction through the parameter space affordable we are bounded to the standard linearization procedures , the only capable of yielding analytical results . inside the linear approximation perfect
squeezing is possible at dynamic instabilities .
we make use of this fact to find optimum working points showing up maximum squeezing .
they are , however , an artifact of the method as the linear approximation breaks down at the instabilities .
what matters for the practical implementation of new squeezed light sources ( our ultimate goal ) are the optimum paths through the parameter space approaching such points .
let us explain a little more what we mean .
fixed a particular parameter there will be a set of values of the remaining parameters ( including the frequency as such ) tuned up to yield the maximum noise reduction . when the chosen parameter is varied an optimum path is defined by the set of parameter values maximizing the noise reduction at any stage of the variation .
the real optimum working point will be somewhere along these paths before reaching an instability .
thus , these paths would guide the experimentalist towards the optimum working point in the real experimental setup .
the essence of our approach to find such paths will be to isolate the dynamic aspect of the squeezing behavior from the static contributions to the noise coming from the different inputs .
a simple adequate normalization will disentangle the two aspects of the quantum noise behavior .
this simplifies the analysis sufficiently to allow a characterization of the optimum paths .
another crucial issue when dealing with the squeezing performance of a system is precisely the election of the most relevant parameter to compare the different configurations with each other or with the reported related experiments .
there is no universal criterion to determine the squeezing efficiency of a given device .
an efficient setup regarding power consumption , i.e. , when compared for fixed input power could well be deceptive when compared for the same output power and perhaps inadequate to some spectroscopic applications .
however , within the state of the art of the present squeezed light generators , the main concern is to improve the squeezing figures themselves having other considerations such as the power consumption a relative importance .
under this perspective probably the parameter of utmost importance as far as c.w .
resonant systems are concerned is the energy load inside the cavity . indeed ,
the usual causes of squeezing degradation such as blue - light - induced red absorption come from an excessive mean photon number inside the cavity capable of significantly degrade the material optical response at the relevant frequencies .
these considerations will lead us to define another normalization this time useful for the evaluation of the squeezing efficiency with respect to the intra - cavity photon number .
the sketch of the article is as follows . in section [ qmm ]
the quantum mechanical model is presented . in section [ lee ]
the evolution equations are linearized , the fixed points of the system obtained and their stability studied .
section [ ss ] gives all the formulae regarding quantum noise spectra in the system . in section [ sp ] a general approach to one - mode systems
is developed which allows the definition of general criteria to characterize the optimum paths and applied to the specific case here addressed .
finally the limits of the model and possible implementations are thoroughly discussed in section [ dc ] , concluding the article with a summary of the most relevant results obtained .
[ qmm ] the system we want to address consists in a second order nonlinear medium coupling two modes of frequency @xmath2 ( fundamental ) and @xmath3 ( harmonic ) respectively and placed inside a ring cavity resonant only with the fundamental mode .
we will also assume just one input - output mirror of finite reflectivity .
the effect of phase mismatch when only the fundamental mode is driven has been experimentally studied in @xcite where bistability induced by cascading was demonstrated .
the classical evolution equation of the fundamental mode , @xmath4 , as given in @xcite , reads @xmath5 as @xmath6 being @xmath7 the length of the nonlinear medium , @xmath8 the spatial dependence of the resonator mode and @xmath9 is proportional to the second order nonlinear susceptibility ( see below ) .
splitting @xmath10 in its real and imaginary parts , eq .
( [ eq : claeqa ] ) can be recast as @xmath11\alpha + \sqrt{2 \gamma_{c } } \alpha_{in } \,.\ ] ] for a plane wave geometry [ eq : mu&ga ] @xmath12\,,\end{aligned}\ ] ] where @xmath13 and @xmath14 denote the real and imaginary part of @xmath15 . in this way
the nonlinear dynamics is divided in a nonlinear absorption ( the up conversion of photons ) and a nonlinear dispersion ( the cascading effect ) . the behavior of both parameters with @xmath16 is plotted in fig .
[ fg : fg ] .
notice that for any finite @xmath16 the nonlinear dispersion is also finite , periodically completely dominating ( null frequency doubling ) . and @xmath13 with respect to the phase mismatch . ]
quantization of eq .
( [ eq : claeqa2 ] ) is then accomplished independently for each effect . regarding the nonlinear absorption we use the two - photon model proposed in @xcite , while nonlinear dispersion is accounted for by a fourth order hamiltonian , @xmath17 as in the standard theory of optical kerr effect .
it represents a hamiltonian modification of the two - photon absorption model so that the quantum mechanical equation reads @xmath18a + 2\sqrt{\mu}\,a^{\dagger}b_{in}+\sqrt{2 \gamma_{c}}\ , a_{in } + \sqrt{2\gamma_{s}}\,w_{in}\,,\ ] ] where latin characters denote the annihilation operators for the corresponding classical ( greek characters ) modes .
two extra terms not present in the classical analog appear , namely , a white noise input , @xmath19 , accounting for the fluctuations induced by the scattering and the absorption in the crystal ( @xmath20 ) and a parametric `` gain '' term coming from the , classically empty , incoming harmonic mode , @xmath21 .
( [ eq : quant ] ) is complemented with the boundary conditions @xcite [ eq : boundary ] @xmath22 from which the output spectra can be computed .
input fields are assumed to be in coherent states .
in particular , allowing a coherent state different from the vacuum for the incoming harmonic mode we generalize the system to the case of driving both modes . in the case
@xmath23 , the squeezing properties as well as the applicability to quantum nondemoliton measurements of this system have been studied in detail in @xcite .
the used definitions for the creation operators give the following relations with the usual experimental parameters ( see appendix in @xcite ) : the input and output powers are @xmath24 and @xmath25 ; the circulating power is @xmath26 being @xmath27 the round - trip time and the single - pass power - conversion efficiency ( in w@xmath28 ) is @xmath29 .
[ lee ] defining fluctuation operators as [ eq : fluctope ] @xmath30 a linearization of eqs .
( [ eq : quant ] ) and ( [ eq : boundary ] ) yields @xmath31\delta a + \left [ 2\sqrt{\mu}\beta_{in } - ( \mu+i \gamma)\alpha^{2}\right ] \delta a^{\dagger } \nonumber \\ & + & 2\sqrt{\mu}\,\alpha^{*}\delta b_{in}+ \sqrt{2 \gamma_{c}}\,\delta a_{in } + \sqrt{2\gamma_{s}}\,w_{in}\,,\end{aligned}\ ] ] and [ eq : linboundary ] @xmath32 being @xmath33 the mean values of the corresponding input and output modes and @xmath4 a stable fixed point of the classical counterpart of eq .
( [ eq : quant ] ) , i.e. @xmath34\alpha + 2\sqrt{\mu}\,\alpha^{*}\beta_{in}+\sqrt{2 \gamma_{c}}\,\alpha_{in } \,.\ ] ] equating to zero the l.h.s . of eq .
( [ eq : class ] ) a `` state equation '' for the fixed points is obtained , namely , @xmath35 \alpha_{in}+ 2 \sqrt{\mu } \,\beta_{in}\ , \alpha_{in}^ { * } \right \ } } { ( \gamma + \mu n)^{2}+(\delta+\gamma n)^{2}-4 \mu |\beta_{in}|^{2 } } \,,\ ] ] with @xmath36 .
let be @xmath37 and @xmath38 the phases of @xmath39 and @xmath40 respectively . then , dividing both sides of eq .
( [ eq : state ] ) by @xmath41 @xmath42 \ : = \ : & & \nonumber \\ ( \delta+\gamma n ) \right ) e^{i(\phi-\varphi/2)}+ 2 \sqrt{\mu}\,|\beta_{in}|e^{-i(\phi-\varphi/2 ) } \right ] \ , .
& & \end{aligned}\ ] ] taking the squared modulus in both sides a quintic equation for @xmath43 is obtained @xmath44^{2 } - 2 \gamma_{c}|\alpha_{in}|^{2 } \left\ { ( \gamma + \mu n)^{2 } + ( \delta+\gamma n)^{2 } + 4 \mu |\beta_{in}|^{2 } + \right .
\nonumber \\ & & \left . 4 \sqrt{\mu}\,|\beta_{in}| \left [ ( \gamma+\mu n ) \cos(2 \phi-\varphi)+
( \delta+\gamma n ) \sin(2 \phi -\varphi ) \right ] \right\ } \,.\end{aligned}\ ] ] the real and the imaginary part of eq .
( @xmath45 ) determine the @xmath46 and @xmath47 as functions of the solutions of eq .
( [ eq : neqn ] ) [ eq : phases ] @xmath48 [ eq.sin ] eq .
( [ eq : neqn ] ) allows for numerical calculation of the fixed points given the input fields .
but it can be interpreted also as a linear equation for @xmath49 , i.e. @xmath50^{2 } } { |\gamma + \mu n + 2 \sqrt{\mu } |\beta_{in}| e^{i ( 2 \phi-\varphi)}|^{2 } + 4 \sqrt{\mu}\,|\beta_{in}| ( \delta+\gamma n ) \sin(2 \phi -\varphi ) } \,.\ ] ] the positive character of the r.h.s .
is not always guaranteed and therefore not for every value of the parameters a real positive @xmath43 is possible .
notice , however , that in the cases in which this happens a simultaneous change of the sign of @xmath51 and @xmath52 yields a consistent set of parameter values .
as we shall see , this fact will have useful consequences in regarding the analysis of the quantum noise behavior in the system .
the stability of the fixed points is governed by the real part of the eigenvalues of the drift matrix associated with the linearized evolution equation ( [ eq : linquant ] ) .
very simple algebra yields @xmath53 provided that the real part of both eigenvalues are negative the fixed point will be stable . with respect to the phase - matched shg case ( @xmath54 and @xmath55 , always stable ) , although both @xmath52 and @xmath51 alone tend to stabilize the dynamics , in combination are able of destabilize the system . a finite @xmath40 , on the other hand , can promote instability depending on its relative phase with respect to @xmath4 , the case @xmath56 maximizing the effect .
all of these new instabilities , however , correspond to zero eigenvalues without a finite imaginary part . in other words ,
contrary to the double resonant shg there is no hopf bifurcation and , consequently , no selfpulsing solution .
[ ss ] for a given quadrature of the electric field , @xmath57 , the squeezing spectrum is simply the noise spectrum of such a quantity , i.e. @xmath58 being @xmath59 some normalization constant and the averages are assumed stationary . as a function of the annihilation and creation operators eq .
( [ eq : sw ] ) is rewritten as @xmath60\,,\ ] ] where use has been made of the stationarity of the average and re denotes real part . from this expression
it is evident that the noise is minimized and therefore the squeezing effect maximized for a quadrature phase such as @xmath61\,,\ ] ] corresponding to a phase @xmath62 where @xmath63 is the phase of @xmath64 .
the spectrum of the conjugate quadrature ( i.e. with a phase @xmath65 corresponds to a plus sign in eq .
( [ eq : sw3 ] ) and by virtue of the heisenberg principle shows an excess noise above the vacuum . taking @xmath66 ( corresponding to vacuum noise units ) and splitting eq .
( [ eq : sw3 ] ) into a vacuum noise component plus a normally ordered part we finally arrive to @xmath67 for both the squeezing and the `` stretching '' spectra . after tedious but simple algebra , the spectra of the fundamental and second harmonic modes can be written as [ eq : spectrum ] @xmath68 where @xmath69 and [ eq : sc ] @xmath70^{2 } + 4 ( \gamma + 2 \mu n)^{2 } ( \delta + 2 \gamma n)^{2}}\ , , \label{eq : scn}\\ d & = & \left[(\gamma+2\mu n)^{2 } + ( \delta + 2 \gamma n)^{2 } - \label{eq : scd}\end{aligned}\ ] ] the correlations defining the squeezing phase @xmath63 are given by [ eq : corr ] @xmath71/d \,,\\ \label{eq : corrb } \langle \delta b_{out}(\omega)\delta b_{out}(-\omega ) \rangle & = & 8\mu\,\alpha^{2 } b \left [ \omega^{2 } + |b|^{2 } + ( \gamma + 2 \mu n)^{2 } - ( \delta + 2 \gamma n)^{2 } + i\ , 2 ( \gamma + 2 \mu n ) ( \delta + 2 \gamma n)\right ] / d \,.\end{aligned}\ ] ] the trigonometric equations for the corresponding phases are quite complicated and rather useless .
however , an interesting consequence can directly be drawn from eqs .
( [ eq : corr ] ) , namely , for detuning - s such as @xmath72 the phases are independent of @xmath2 equaling those of @xmath73 and @xmath74 respectively .
[ sp ] in the previous sections we have developed the general raw formulae regarding quantum noise in the system .
this section is devoted to the analysis of the quantum noise behavior implied by them an in particular to proceed with our program of finding optimum quantum noise reduction .
however , before any specific assessment of the quantum noise performance we will elaborate a little more on the formulae ( mainly by adequate normalizations ) in order to gain physical insight and ease our task . as an aftermath we shall obtain general results going well beyond the specifics of the system addressed here .
let us begin defining a nonlinear and a total decay rate as @xmath75 and @xmath76 respectively .
we shall scale the evolution with this total decay rate defining an dimensionless time @xmath77 . in the spectra ( [ eq : spectrum ] )
the only dependence on @xmath78 is through @xmath73 disappearing for @xmath55 .
it is also possible to restrict this dependence to such a term directly in eq .
( [ eq : linquant ] ) and eq .
( [ eq : linboundary ] ) by means of appropriate phase shifts of the modes .
all together account for @xmath79\delta c + \tilde{b}\,\delta c^{\dagger } \nonumber \\ & + & \sqrt{2 \tilde{\gamma}_{nl}}\;\delta d_{in } + \sqrt{2 \tilde{\gamma}_{c}}\;\delta c_{in } + \sqrt{2\tilde{\gamma}_{s}}\,s_{in}\,,\end{aligned}\ ] ] where the tilde represents divided by @xmath80 , @xmath81 , @xmath82 and the modes are redefined as [ eq : nfields ] @xmath83 in agreement with the previous notation @xmath84 denotes the mean value of @xmath85 .
the boundary conditions of the new modes are [ eq : nlinboundary ] @xmath86 for coherent states , the correlations of the new input modes remain as white noise but in the scaled time @xmath27 .
we shall refer to the previous formulae as the tilde normalization .
the evolution equation ( [ eq : normeq ] ) encodes the dynamic response of the intracavity system to a series of noisy input channels ( @xmath87 , @xmath88 and @xmath89 ) .
quantum mechanical consistency , i.e. , conservation of equal - time conmutators , imposes a fluctuation - dissipation relation which under this normalization reads @xmath90 the evolution equation ( [ eq : normeq ] ) along with eqs .
( [ eq : nlinboundary ] ) are now written in such a way that the input - output couplings are real - valued as in the standard input - output formalism @xcite .
this is a completely general result .
provided a well defined linearized theory in the sense of preserving equal - time conmutators we only need the adequate set of phase shifts of the input channels ( a trivial unitary transformation preserving conmutators ) making the couplings real - valued to obtain a theory formally equal to the standard input - output formalism simply because this is the theory preserving the equal - time conmutators when the couplings are real - valued .
thus , for any system with only one effective mode there is a formulation in which the intracavity field follows @xmath91\delta c + \tilde{b}\,\delta c^{\dagger } + \sum_{n=1}^{n } \sqrt{2 \tilde{\gamma}_{n } } \,\delta c_{in}^{n}\,,\ ] ] with @xmath92 the frequency scale @xmath80 defining the dimensionless time @xmath27 is just the real part of the factor multiplying @xmath93 after the phase shifts .
n-1 of the input channels will have a time - reversed counterpart corresponding to the outgoing channels fulfilling @xmath94 the remaining input channel will account for the irreversible losses .
the corresponding spectra are related with the intracavity spectra by @xmath95 where @xmath96 denotes the intracavity spectra and we have made use of the proportionality of normally ordered intracavity and outgoing correlations @xcite .
the spectra @xmath97 coincide with the spectra of the original formulation as the new outgoing modes are just a phase sift of the originals .
let us now define a sort of `` reference '' system with only one time - reversible input channel , i.e. , @xmath98\delta c + \tilde{b}\,\delta c^{\dagger } + \sqrt{2}\ , \delta c_{in}^{ref}\,,\ ] ] and @xmath99 obviously @xmath100 so that we get finally @xmath101 this is the central result of this section .
let us elaborate a little about its interpretation .
squeezing in a given output channel means that for a certain range of phase shifts the corresponding quadratures show an intensity of their fluctuations below that of the associated incoming channel ( assumed in a coherent state ) . in view of eqs .
( [ eq : gennlinboundary ] ) , the amplitude of the outgoing fluctuations is a coherent superposition of the intracavity and the incoming fluctuations .
squeezing is possible if an adequate correlation between @xmath93 and the relevant input channel is established .
but the intracavity field is nothing else than the dynamic response of the intracavity system to the incoming channels .
the input channels are uncorrelated and so the dynamic response of the intracavity system to them
. a given input channel can consequently correlate only with the dynamic response to itself .
the presence of any other input channel can only degrade the effect .
the great advantage of the tilde normalization is that makes this fact explicit . indeed
, eq . ( [ eq : scentral ] ) express the output spectra as the dynamic response of the system to an isolated input channel , i.e. , @xmath102 , scaled down by the `` static '' contribution to the noise owing to the presence of extra input channels .
the scale factor @xmath103 is just the ratio between the coupling constant of the chosen output channel and the sum of all of them .
( [ eq : scentral ] ) greatly simplifies our task of finding the optimum path to maximum noise reduction as we can center our efforts onto the simple reference system described by eqs .
( [ eq : refeq ] ) and ( [ eq : refbc ] ) .
even more interesting the results concerning the reference system will be of general applicability to any one - mode system including as such any multiply resonant system under adiabatic elimination of all the modes but one .
the normally ordered spectra of the reference system are easily calculated as @xmath104 our first step is to determine if the dynamic response is capable of a total noise suppression .
squeezing can only occur at a dynamic instability .
equaling to zero the l.h.s .
( [ eq : eigen ] ) ( the only possible unstable eigenvalue ) and after proper normalization an equation determining the instability can be written as @xmath105 written in this way an interesting parallelism with the standard opo below threshold shows up , i.e. an instability appears when the modulus of the `` losses '' coefficient equals that of the `` parametric '' coefficient , a sort of natural extension of the condition for the instability in the conventional opo for which the coefficients are real . inserting the instability condition ( [ eq : refins ] ) in @xmath106 results in @xmath107 applying lhopital s rule with respect to @xmath108
, @xmath109 equals -1 at @xmath110 , that is , perfect squeezing is obtained at the instability , again in parallel with opo .
in other words , the dynamic response of the system assuming that the condition eq .
( [ eq : refins ] ) is reachable , is capable of a complete suppression of quantum noise .
spectrum ( [ eq : sn ] ) is simple enough to permit analytical optimization .
taking partial derivative in @xmath106 with respect to @xmath111 and equaling to zero , @xmath110 appears as the optimum point whatever the values of @xmath112 and @xmath113 .
the same applies to @xmath114 when taking partial derivative with respect to @xmath112 .
notice that this last condition implies also a squeezing phase independent of the frequency .
the optimized noise obtained imposing these two conditions simplifies to @xmath115 with a minimum at the instability @xmath116 approached monotonically .
these conditions ( @xmath110 , @xmath117 and @xmath116 ) will help us in finding optimum paths .
in particular , moving @xmath113 from zero to one while maintaining @xmath118 defines an optimum path reaching the instability for the `` reference '' model .
an optimum path is defined solely by the squeezing spectrum leaving aside the `` stretching '' one .
it is important to study also the accompanying excess noise on the conjugate quadrature for it could invalidate in practice the optimum path if this excess noise is unbearable high .
the minimal excess noise production imposed by the heisenberg principle corresponds to @xmath119 .
it is a perfect complementary relation between quadratures : the deamplification of fluctuations in a given quadrature must equal the amplification of fluctuations in the conjugate . in that case
we are dealing with a minimum uncertainty state ( mus ) for those quadratures , the text - book definition of a squeezed state . adding 1 to eq .
( [ eq : sn ] ) and after some minor algebra @xmath120 straightforward algebra leads to @xmath121 .
the excess noise is minimum again in parallel to the standard opo system with real coefficients .
as discussed in the introduction we are here principally interested in the squeezing behavior with respect to the photon number @xmath43 .
unfortunately the tilde normalization is inappropriate to such a task as the frequency scale depends on @xmath43 itself .
it is far more convenient to use @xmath122 as the time scale instead of @xmath123 and to normalize the photon number as @xmath124 .
in complete parallelism to the tilde normalization we have then , @xmath125\delta c + \left [ \sqrt{k_{r}}\,\eta_{in } - ( k_{r}+i k_{i } ) m \right ] \ , \delta c^{\dagger } \nonumber \\ & + & 2 \sqrt{m k_{r}}\;\delta d_{in } + \sqrt{2 \hat{\gamma}_{c}}\;\delta c_{in } + \sqrt{2\hat{\gamma}_{s}}\,s_{in}\,,\end{aligned}\ ] ] where the hat represents divided by @xmath126 and @xmath127 which represents the harmonic mode input amplitude normalized to the value at the standard opo threshold .
the spectra ( [ eq : spectrum ] ) become now [ eq : hatspectrum ] @xmath128 with [ eq : hatsc ] @xmath129^{2 } + 4 ( 1 + 2 k_{r } m)^{2 } ( \hat{\delta } + 2 k_{i } m)^{2}}\ , , \label{eq : hatscn}\\ \hat{d } & = & \left [ ( 1 + 2 k_{r } m)^{2}+(\hat{\delta}+ 2 k_{i } m)^{2 } - |\hat{b}|^{2 } - \hat{\omega}^{2 } \right ] ^{2}+ 4 ( 1 + 2 k_{r } m)^{2 } \hat{\omega}^{2 } \ , , \label{eq : hatscd}\end{aligned}\ ] ] and @xmath130
. we shall refer to the above formulae as the hat normalization .
to refer the frequency to the cavity decay constant with the subsequent re - normalization of the system parameters , i.e. the hat normalization , is quite a standard procedure in the literature .
some care must be taken when studying the quantum noise behavior as a function of @xmath43 ( or @xmath131 ) .
it is not a free parameter of the problem as would be the input fields or the phase mismatch but it is in a nonlinear relation with them .
we need , therefore , to check that the proposed values of @xmath43 are indeed a solution of eq .
( [ eq : neqn ] ) .
fortunately the spectra ( [ eq : spectrum ] ) do not depend on the overall sign of @xmath132 and therefore the conclusions reached in section [ lee ] about the existence of @xmath133 permit a safe variation of @xmath43 in search of strong noise reduction provided the stability of the corresponding fixed points . applying eq .
( [ eq : scentral ] ) to the fundamental mode @xmath134 it is clear that the best performance corresponds to @xmath135 , that is , either @xmath136 or @xmath137 , as then @xmath138 maximizes to @xmath139 ( the escape efficiency of the cavity ) .
the case @xmath136 corresponds to the very well known case of squeezed vacuum generation . for @xmath140 and finite @xmath43
the system is formally equivalent to a resonant optical kerr effect system whose quantum noise behavior has been amply studied previously @xcite .
the condition ( [ eq : refins ] ) reduces for @xmath140 to @xmath141 , the well known turning points of optical dispersive bistability @xcite but with the nonlinear dispersion induced by cascading . indeed , such cascading
induced bistability has been experimentally demonstrated in @xcite . rewriting it within the hat normalization the condition reads @xmath142 where @xmath143 as it is evaluated at @xmath144 .
once @xmath138 is independent of @xmath16 and @xmath43 , the optimum path with respect to @xmath131 ( the only remaining free parameter ) is determined solely by the reference system .
it corresponds to increase @xmath131 till @xmath145 ( where the condition ( [ eq : inskerr ] ) is reached ) while maintaining @xmath146 and @xmath147 .
[ fg : sakerr ] displays the evolution of both the maximum squeezing and the maximum excess noise following such a path for three values of the escape efficiency , namely , 0.9 , 0.99 and the ideal 1 .
the noise is expressed in db s with respect to the vacuum noise
. a heisenberg limited excess noise appears in such a case as a specular image of the squeezing .
the instability is signaled by the divergence in the excess noise . above it ,
the curves shown are not physical as they correspond to unstable fixed points .
the case @xmath148 in fig .
[ fg : sakerr ] shows an excellent behavior with an almost heisenberg limited excess noise till near the instability . .
the curves above the divergences are not physical . ]
[ fg : sakcomp ] illustrates the idea of optimum path by comparing the @xmath149 plot of fig . [ fg : sakerr ] against various cases with fixed values of @xmath150 . below @xmath145 , for a given @xmath131 the maximum squeezing is obtained when @xmath114 as expected .
above @xmath145 it is not possible to reach the minimum noise of @xmath151 fulfilling @xmath114 . .
above the divergences the results are not physical . ] being @xmath140 the formulae simplify enough for allowing a simple expression for the squeezing phase . more specifically , from @xmath152 , @xmath153 .
on the other hand substituting eq .
( [ eq : state ] ) in eq .
( [ eq : boundarya ] ) results in @xmath154 , giving a squeezing phase relative to that of the output field of @xmath155 at the instability @xmath156 and for low @xmath157 it approaches @xmath158 .
there is a possibility of an extra control of the squeezing phase not present in the conventional kerr effect system by making use of the harmonic mode .
taking a finite @xmath159 but low enough so that @xmath160 and at the same time a @xmath40 high enough to imply @xmath161 , still we will have @xmath162 while the squeezing phase relative to the output field will depend on both the modulus and the relative phase between the input fields . in practice , however , maintaining @xmath163 very low could imply a exceedingly high @xmath40 in order to have a @xmath164 intense enough for a significant influence on the final phases .
of course , the instability point would accordingly depend on @xmath40 .
there is no hope of any behavior similar to the reported in @xcite as a competition between second and third order nonlinearities needs @xmath165 , opening the fundamental mode to the fluctuations of the input harmonic mode with strong deleterious effects . at most
, some remnants of the enhanced efficiency coming from the competition between nonlinearities can be observed for low @xmath166 .
then , as shown in fig
. [ fg : sa ] , the best working point is not necessarily located at @xmath167 , i.e. maximum squeezing is obtained with a finite mismatch . for the harmonic mode eq .
( [ eq : scentral ] ) yields @xmath168 now , the situation is the complete opposite : the performance is favored by a finite @xmath159 in order to have a non - zero @xmath166 and a large @xmath43 to approach the ratio @xmath169 to one .
in fact , under ideal conditions of perfect dynamic noise suppression and no absorption and scattering losses ( @xmath170 ) , the squeezing in both modes are complementary in the sense of @xmath171 a direct consequence of the fluctuation - dissipation relation ( [ eq : genadd1 ] ) .
this complementarity has been previously reported for the doubly resonant degenerate parametric oscillator @xcite .
the maximum squeezing available for the harmonic mode whatever the dynamic response of the system is easily obtained by setting @xmath172 to -1 in eq .
( [ eq : sbtilde ] ) , that is , @xmath173 this static contribution to the noise is now nonlinear in the sense that it depends on the phase mismatch and @xmath131 . an immediate consequence of eq . ( [ eq : ilimit ] ) is the possibility of an arbitrarily large quantum noise reduction for any finite value of @xmath13 .
the 1/9 limit of the conventional phase - matched shg is therefore due to a failure of the setup to maximize the dynamic response of the system .
let us center then , firstly in the shg - like case with @xmath55 as it includes the above mentioned conventional setup ( the experiments in @xcite and @xcite ) .
the instability points are now given by ( directly in the hat normalization ) @xmath174 both kinds of nonlinearities ( dispersive and absorptive ) are in this case necessary as the factor @xmath175 needs to be positive to allow @xmath176 to be real .
the phase - matched case is therefore excluded . as a function of the phase mismatch . ]
[ fg : g-3f ] shows @xmath175 as a function of the phase mismatch and indeed near above @xmath177 it is positive .
optimum approaches to ( [ eq : hatinsb ] ) are now more difficult to evaluate than in the fundamental mode as both the dynamic processing of the noise and the static contribution from the noise inputs ( encoded in @xmath169 ) depend on @xmath131 and @xmath16 . with respect to to @xmath131
is clear that the static part is optimized at @xmath178 .
this limit can be approached letting @xmath179 and @xmath180 ( i.e. @xmath114 ) .
@xmath113 reduces in this case of @xmath55 to @xmath181 showing a monotonic increasing behavior with respect to @xmath131 from 0 to the maximum ( at @xmath182 ) @xmath183 \,.\ ] ] notice that for @xmath184 it consistently equals 1 .
we have then , both @xmath185 and the fastest approach to 1 of @xmath113 when @xmath182 .
therefore , the squeezing along an optimum path with respect to @xmath16 is given by substituting eq .
( [ eq : bmax ] ) in the spectrum ( [ eq : sopt ] ) and then the obtained @xmath186 in eq .
( [ eq : sbtilde ] ) .
obviously , in a real experiment @xmath131 can be large but always finite .
let us take as a `` large '' @xmath131 one giving a @xmath187 around 20 db as in the case @xmath148 of fig .
[ fg : sakerr ] .
this corresponds to @xmath188 .
figure [ fg : sbkm50 ] displays @xmath189 as a function of the phase mismatch in such a case . to illustrate the modulation exerted by @xmath187
we have take this time @xmath150 equal to the real part of eq .
( [ eq : hatinsb ] ) plus a very small number . in this way
the plot remains valid for the whole range of the @xmath190 . while eq .
( [ eq : hatinsb ] ) is complex the condition @xmath191 is almost fulfilled and above the instability the noise suppression reduction follows @xmath187 . again
the pernicious effect of the instability regarding the excess noise has a very short range . for comparison @xmath187
is also depicted .
the optimum path with respect to @xmath131 is much more complicated to find because the intricate dependence of @xmath14 and @xmath13 with respect to the phase mismatch .
figure [ fg : sboptbin0 ] has been generated finding numerically the minima of @xmath192 while scanning the range of @xmath131 .
for comparison the phase - matched case is also depicted showing an asymptotic behavior towards @xmath193 . for low values of @xmath131
the effect of @xmath169 overwhelms the dynamic response so that the best value corresponds to maximize @xmath13 .
as soon as the two curves depart from each other the dynamic response dominates the behavior and the minimum noise is at the instability as in fig .
[ fg : sbkm50 ] . at this stage
the optimum path begins to follow the instability all the time .
it should be taken then , as a mathematical limit . however , in view of fig .
[ fg : sbkm50 ] , before reaching it , bearable values of the excess noise are accessible with a slight diminution of the squeezing . ) for the shg like case compared with the phase - matched shg case . ]
it is worth to mention that a squeezing as large as 48% induced by cascading has been very recently reported @xcite .
the cascading was due , however , to a detuning of the pump mode in a triply resonant non - degenerate opo with a much lower finesse for the pump mode rather than by phase mismatch .
under such conditions a cascaded @xmath1 is also induced leading ideally to perfect squeezing in the pump mode .
although a finite mismatch allows to reach @xmath187 , the overall optimum working point corresponding to @xmath194 is out of reach .
the question arises then , of if it is possible to fulfill eq .
( [ eq : refins ] ) at @xmath194 .
a glance at the definition of @xmath73 suggests it should be possible adding a driving to the harmonic mode .
in such a case @xmath169 simplifies to @xmath195 obviously independent of @xmath196 . constructing an optimum path with respect to the harmonic input reduces then , to set @xmath118 .
phase matching along with @xmath114 implies @xmath197 , @xmath198 and @xmath199 so that the instability condition ( [ eq : refins ] ) simplifies to @xmath200 a perfectly achievable condition .
we can further optimize by choosing the phase of @xmath196 adequately to approach @xmath201 to one as much as possible .
the extreme cases correspond to @xmath196 real , i.e. , @xmath202 and @xmath203 . from eqs .
( [ eq : phases ] ) and ( [ eq : ain ] ) it is easy to check that they correspond respectively to @xmath204 and @xmath205 .
the negative case maximizes @xmath201 .
it has been previously reported in @xcite .
taking squared modulus of eq .
( [ eq : boundaryb ] ) the negative case appears as promoting harmonic output power while the converse is valid for the positive .
the squeezing phase is also easy to calculate in this case . in particular , given the correlation ( [ eq : corrb ] ) , @xmath63 is determined by the phase of @xmath74 ( independent of @xmath2 as @xmath191 ) , something proportional to @xmath206 the corresponding squeezing phases are @xmath207 for the negative case while for the positive case it changes from @xmath207 to @xmath208 at @xmath209 .
on the other hand , the output harmonic amplitude is proportional to ( see eq .
( [ eq : boundaryb ] ) ) @xmath210 consequently , the relative squeezing phase for the negative @xmath196 is @xmath177 , i.e. , amplitude squeezing .
the positive case is more complicated .
it remains equal @xmath177 ( amplitude squeezing ) till @xmath211 . above this value
it changes to @xmath212 depending on the sign of @xmath213 yielding in any case phase squeezing . at a first glance
, it appears there is a sudden change from amplitude to phase squeezing when the input phases are fix to @xmath205 and @xmath214 passes through @xmath131 .
it is not so however , as at this point @xmath215 and the state collapses to a coherent state with no squeezing .
the situation is clearly depicted in figure [ fg : sbin50 ] where @xmath216 are displayed as a function of @xmath196 assumed real .
the r.h.s . of
the plot corresponds to @xmath217 while the l.h.s .
to @xmath204 and for negative ordinates it should be considered as an optimum path with respect to @xmath196 .
the behavior is completely symmetric with respect to @xmath209 where both the squeezing and the excess noise equal that of the vacuum . ) .
the curves are not physical above the divergences . ]
the optimum path with respect to @xmath131 is now given by @xmath187 at @xmath218 .
figure [ fg : sm ] is the equivalent to fig . [ fg : sboptbin0 ] for the new situation .
it represents the maximum efficiency as far as quantum noise reduction is concerned the system can yield in any way with respect to @xmath131 . the improvement with respect to the standard phase - matched shg as well as to the optimized shg is certainly high . ) in nonlinear second order singly resonant device . for comparison
the phase - matched and optimized shg cases are also shown . ]
[ dc ] two are the main purposes of the present work . on one side , to gain physical insight about the origins of quantum noise in singly resonant systems . on the other , to explore their potential as squeezed light sources .
in such a task we have used a model including all the relevant physics we wanted to address but simple enough to be tractable .
the results shown in the previous sections certainly reveal a high potential of the studied configurations .
an evaluation of the limits of the model in reproducing the real physical situation as well as a discussion of possible implementations , seems , therefore , in order .
one obvious idealization of the model is to assume perfectly coherent inputs neglecting the excess noise of real lasers something expected to have deleterious effects at low frequencies .
white and coworkers @xcite have developed an analytical approach to this problem resulting in an impressive agreement with the experiments .
as expected , the excess noise completely destroys the squeezing at low frequencies . in their experiments , however , the deleterious effect was restricted to only 7 mhz by adding a mode cleaner to the system , the spectrum coinciding with the ideal one out of this range .
even better , in @xcite the laser noise was shot - noise - limited down to 1 mhz , again using an external mode cleaner .
considering as sensible the assumption of coherent states for the input modes as well as a value of @xmath131 around 3 ( we will see below it looks like the case ) our main concern about the fundamental mode results summarized in fig .
[ fg : sakerr ] refers to the feasibility of the chosen escape efficiencies .
the ratio @xmath219 is difficult to maximize in a resonant mode because , by its own resonant nature , @xmath220 must be rather low .
thus , in @xcite it was only of 0.52 , while in @xcite it was 0.36 .
even in @xcite , a doubly resonant system specifically designed to squeeze the fundamental mode , the escape efficiency was around 0.9 , limiting the maximum squeezing achievable to 90% ( in practice , a 52% of noise reduction was reached ) .
it appears , then , that nowadays the @xmath149 should be taken rather as an ideal illustrative case . in contrast , the ultimate limit for the noise suppression in the harmonic mode ( eq . ( [ eq : ilimit ] ) ) is pushed up by the fundamental mode photon number , opening a way to bypass the usual untouchable limit imposed by the escape efficiency of the cavity ( as in the fundamental mode )
. therefore , the squeezing in the harmonic mode can be arbitrarily large under the ideal assumption that the energy load inside the cavity can be also arbitrarily large .
however , this is not totally true as the model does not take into account the losses in the harmonic mode which necessarily limit the degree of noise suppression .
we can estimate this limitation assuming the absorption in one single pass through the nonlinear material equivalent to the effect of a beam splitter with the adequate reflectivity .
taking an absorption of 0.6%/cm as in @xcite and a length of 1 cm , the equivalent reflectivity would be of @xmath221 .
the spectrum after the beam splitter is given by @xmath222 .
setting @xmath223 and @xmath224 , the ultimate squeezing achievable is precisely @xmath225 , i.e. - 22 db . in other words , the chosen value of @xmath188 in figs .
[ fg : sbkm50 ] and [ fg : sbin50 ] represents more or less the maximum the model can stand without the inclusion of the harmonic mode losses . of course , we still can not assume @xmath188 as a realistic limit for the state of the art devices as @xmath131 depends not only on the intra - cavity photon number but on the ratio @xmath226 between the nonlinearity and losses .
this ratio must be high enough in order to prevent a degradation of the nonlinear optical response in the system as commented in the introduction . besides , this ratio scales down the power available in the external sources . in view of these complications ,
probably the most reliable way of setting the physical scale of @xmath131 is to compare the results with the reported experiments . in @xcite
the quoted noise reduction was of -5.2 db .
setting to zero @xmath16 and @xmath40 in eq .
( [ eq : hatspectrumb ] ) corresponding to phase - matched shg , a -5.2 db squeezing results at @xmath227 , far from the @xmath188 limit .
fortunately , the limit ( [ eq : ilimit ] ) grows up quite quickly for low @xmath228 ( see fig .
[ fg : sm ] ) .
thus , 10 db of noise suppression are reached at @xmath229 , no such unthinkable value .
however , -15 db of noise reduction requires @xmath230 , while a -20 db figure is at the @xmath188 limit , an order of magnitude higher .
new nonlinear materials seem the only possibility for such high squeezing degrees .
a promising via consists in the use of resonant nonlinearities in asymmetric quantum wells ( aqw ) .
huge nonlinearities have been demonstrated in frequency doubling experiments and even a tuning of the nonlinearity with a d.c .
field @xcite . obviously , also the absorption is enhanced by the resonance .
this can be a problem as the ratio @xmath226 could at the end of the day not be increased . to asses
this possibility requires quite a detailed analysis out of the scope of the present work .
we can foreseen , however , a promising advantage in the fact that the losses in the harmonic mode have little influence on the performance . by maintaining a strong two photon resonance but relaxing the one photon counterpart ( tuning with a d.c . field or by an adequate energy level engineering )
, the nonlinearity would be certainly enhanced while the losses at the fundamental mode would not increase so strongly , thus enhancing @xmath226 . with only one passage through the cavity of the harmonic mode and taking into account
that a very thin layer of material is capable of shg @xcite , the corresponding deleterious effect can not be very large .
even a more exciting possibility comes from the recent experimental demonstrations of absorption inhibition in aqw induced by quantum interference @xcite .
the absorption transparency and the resonant enhancement can be combined using an adequate quantum well engineering leading to very efficient frequency doublers ( see @xcite , where precisely a scheme only resonant at the harmonic mode is proposed ) .
these are certainly promising perspectives but we should not dismiss the improvements arising at the range of the present nonlinear crystals performances . let us center then , around @xmath227 . as shown in the previous section the best strategy corresponds to drive both modes with relative phases @xmath231 ( negative @xmath196 ) and @xmath117 . in fig .
[ fg : sqm5 ] the noise behavior till @xmath232 is displayed for various `` distances '' to the instability ( [ eq : etains ] ) . even at half the instability @xmath196 value ,
the squeezing at @xmath227 grows from -5.1 db ( 69% ) to -7.2 db ( 80% ) .
the excess noise , on the other hand , rapidly increases at low @xmath228 but it also saturates quickly to bearable values .
the improvement , although nothing spectacular is quite substantial . in @xcite
it was not reckoned so because the noise suppression was studied as function of the input power .
given its nonlinear relation with @xmath131 the improvement is much slower with respect to this variable .
besides the squeezing , the output power is also enhanced . taking a negative @xmath196 in eq .
( [ eq : bout ] ) , the output power results in @xmath233 and thus , the harmonic mode input contributes constructively to it . as shown in fig .
[ fg : pw ] at half a way of the instability the power is nearly doubled .
although from the theoretical point of view the injection of a coherent signal in the harmonic mode looks quite harmless , the experimental implementation is not trivial .
however , the remarkable achievements in @xcite with the opa strongly support the feasibility of the idea . finally , a word of caution about the design of the device .
it is important to avoid the setting of oscillations out of the fundamental mode ( the so called subharmonic pumped opo @xcite ) , something capable of destroying the noise reduction @xcite . at a first glance , finite values of @xmath196
would favor the effect by promoting the down conversion .
but it is not necessarily so as the down conversion is encouraged only for a given range of the relative phase between the two driving fields .
thus , for the negative @xmath196 case studied above , being the harmonic output power maximized , the down conversion is minimized . to conclude ,
let us summarize the most relevant results .
firstly , for any system with only one effective mode we have given a systematic approach capable of isolating the processing of quantum noise by the dynamic response of the system .
this dynamic processing is maximized at zero frequency , zero generalized nonlinear detuning ( @xmath112 as defined in section [ sp ] ) and at a dynamic instability .
the static contributions to the noise coming from the different noisy inputs can in some cases , move the overall optimum working point away from that corresponding to maximum dynamic noise suppression . in spite of such , to have a rule to maximize
the dynamic quantum noise suppression resulted very useful to characterize the squeezing behavior when applied to a specific optical system .
in particular , for the case of a singly resonant second order nonlinear device , the squeezing at the fundamental mode is limited by the escape efficiency of the cavity , the best working point being within figures of merit of conventional nonlinear crystals . in the harmonic mode high squeezing requires new materials but it is only limited by the losses in the non - resonant harmonic mode opening the possibility of using multiple quantum wells with resonantly enhanced nonlinearities . however , with standard nonlinear crystals still is possible a substantial improvement with respect to the reported experiments by injecting a coherent driving in the harmonic mode .
besides , the output power is highly enhanced . c.
c. thanks s. schiller and specially a. g. white for useful comments and suggestions .
work supported in part by grants no .
tic95 - 0563-c05 - 03 , no . pb96 - 00819 , cicyt ( spain ) and comunidad de madrid 06t/039/96 ( spain ) . | quantum noise in a model of singly resonant frequency doubling including phase mismatch and driving in the harmonic mode is analyzed .
the general formulae about the fixed points and their stability as well as the squeezing spectra calculated linearizing around such points are given .
the use of a nonlinear normalization allows to disentangle in the spectra the dynamic response of the system from the contributions of the various noisy inputs . a general `` reference '' model for one - mode systems
is developed in which the dynamic aspects of the problem are not contaminated by static contributions from the noisy inputs . the physical insight gained permits the elaboration of general criteria to optimize the noise suppression performance . with respect to the squeezing in the fundamental mode the optimum working point
is located near the first turning point of the dispersive bistability induced by cascading of the second order nonlinear response .
the nonlinearities induced by conventional crystals appear enough to reach it being the squeezing ultimately limited by the escape efficiency of the cavity . in the case of the harmonic mode both , finite phase mismatch and/or harmonic mode driving allow for an optimum dynamic response of the system something not possible in the standard phase matched second harmonic generation .
the squeezing is then limited by the losses in the harmonic mode , allowing for very high degrees of squeezing because of the non - resonant nature of the mode .
this opens the possibility of very high performances using artificial materials with resonantly enhanced nonlinearities .
it is also shown how it is possible to substantially increase the noise reduction and at the same time to more than double the output power for parameters corresponding to reported experiments . |
linear mixed models and the model - based estimators including empirical bayes ( eb ) estimator or empirical best linear unbiased predictor ( eblup ) have been studied quite extensively in the literature from both theoretical and applied points of view . of these ,
the small area estimation ( sae ) is an important application , and methods for sae have received much attention in recent years due to growing demand for reliable small area estimates . for a good review and account on this topic , see ghosh and rao ( 1994 ) , rao ( 2003 ) , datta and ghosh ( 2012 ) and pfeffermann ( 2014 ) .
the linear mixed models used for sae are the fay - herriot model suggested by fay and herriot ( 1979 ) for area - level data and the nested error regression ( ner ) models given in battese , harter and fuller ( 1988 ) for unit - level data .
especially , the ner model has been used in application of not only sae but also biological experiments and econometric analysis . besides the noise , a source of variation
is added to explain the correlation among observations within clusters , or subjects , and to allow the analysis to ` borrow strength from other clusters .
the resulting estimators , such as eb or eblup , for small - cluster means or subject - specific values provide reliable estimates with higher precisions than direct estimates like sample means . in the ner model with @xmath0 small - clusters ,
let @xmath1 be @xmath2 individual observations from the @xmath3-th cluster for @xmath4 , where @xmath5 is a @xmath6-dimensional known vector of covariates .
the normal ner model is written as @xmath7 where @xmath8 and @xmath9 denote the random effect and samping error , respectively , and they are mutually independently distributed as @xmath10 and @xmath11 .
the mean of @xmath12 is @xmath13 for regression coefficients @xmath14 , and the variance of @xmath12 is decomposed as @xmath15 = \tau^2 + \si^2 . \label{hm0}\ ] ] which is the same for all the clusters .
however , jiang and nguyen ( 2012 ) illustrated that the within - cluster sample variances change dramatically from cluster to cluster for the data given in battese , et al .
also , the normality assumptions for random effects and error terms are not always appropriate in practice .
thus , we want to address the issue of relieving these assumptions of normal ner models in the two directions : heterogeneity of variances and non - normality of underlying distributions . in real application
, we often encounter the situation where the sampling variance @xmath16 is affected by the covariate @xmath5 . in such case ,
the variance function is a useful tool for describing its relationship .
variance function estimation has been studied in the literature in the framework of heteroscedastic nonparametric regression .
for example , see cook and weisberg ( 1983 ) , hall and carroll ( 1989 ) , muller and stadtmuller ( 1987 , 1993 ) and ruppert , wand , holst and hossjer ( 1997 ) .
thus , in this paper , we propose use of the technique to introduce the heteroscedastic variances into ner model without assuming normality of underlying distributions .
the variance structure we consider is @xmath17 namely , the setup means that the sampling error @xmath9 has heteroscedastic variance @xmath18 .
then we suggest the variance function model given by @xmath19 , where the details are explained in section [ sec : model ] .
related to this paper , jiang and nguyen ( 2012 ) proposed the heteroscedastic nested error regression model with the setup that variance @xmath20 is proportional to @xmath21 , namely @xmath22 this is equivalent to the assumption that @xmath23 and @xmath24 . for setup ( [ hm2 ] ) ,
jiang and nguyen ( 2012 ) assumed normality for @xmath8 and @xmath9 and demonstrated the quite interesting result that the maximum likelihood ( ml ) estimators of @xmath14 and @xmath25 are consistent for large @xmath0 , which implies that the resulting empirical bayes estimator estimates the bayes estimator consistently . in setup ( [ hm2 ] ) , however , there is no consistent estimator for the heteroscedastic variance @xmath21 , and the mean squared error ( mse ) of the eb can not be estimated consistently , since it depends on @xmath21 . to fix the inconsistent estimation of @xmath21 , maiti , ren and sinha ( 2014 ) suggested the hierarchical model such that @xmath21 s are random variables and @xmath26 has a gamma distribution .
maiti , et al .
( 2014 ) applied this setup to the fay - herriot model with statistics for estimating @xmath21 .
however , the resulting eb estimator and the mse can not be expressed in closed forms .
the same setup of @xmath21 was used recently by kubokawa , sugasawa , ghosh and choudhuri ( 2014 ) who derived explicit expressions of the eb estimator and the mse to second - order . in their simulation study , however , the finite sample properties of estimators of two hyper - parameters in the gamma prior distribution of @xmath21 are not so well . although the hierarchical models used in maiti , et al . ( 2014 ) and kubokawa , et al .
( 2014 ) provide consistent estimators for model parameters and predictors , both models assume parametric hierarchical structures based on normal distributions of @xmath8 and @xmath9 . however , the normality assumption is not always appropriate and another heteroscedastic models are useful for such a situation when the normality assumption does not seem to be correct .
in contrast to the existing results , the proposed model with variance function does not assume normality for either @xmath8 nor @xmath9 .
the advantage of this paper is that the mse of the eb or eblup and its unbiased estimator are derived analytically in closed forms up to second - order without assuming normality for @xmath8 and @xmath9 .
nonparametric approach to sae has been studied by jiang , lahiri and wan ( 2002 ) , hall and maiti ( 2006 ) , lohr and rao ( 2009 ) and others .
most estimators of the mse have been given by numerical methods such as jackknife and bootstrap methods except for lahiri and rao ( 1995 ) , who provided an analytical second - order unbiased estimator of the mse in the fay - heriot model .
hall and maiti ( 2006 ) developed a moment matching bootstrap method for nonparametric estimation of mse in nested error regression models .
the suggested method is actually convenient but it requires bootstrap replication and has computational burden . in this paper , without assuming the normality , we derive not only second - order biases and variances of estimators for the model parameters , but also a closed expression for a second - order unbiased estimator of the mse in a closed form .
thus our mse estimator does not require any resampling method and is useful in practical use .
also our mse estimator can be regarded as a generalization of the robust mse estimator given in lahiri and rao ( 1995 ) .
the paper is organized as follows : a setup of the proposed hner model and estimation strategy with asymptotic properties are given in section [ sec : model ] . in section [ sec : mse ] , we obtain the eblup and the second - order approximation of the mse .
further , we provide the second - order unbiased estimators of mse by the analytical calculation . in section [ sec :
sim ] , we investigate the performance of the proposed procedures through simulation and empirical studies . the technical proofs are given in the appendix .
suppose that there are @xmath0 small clusters , and let @xmath27 be the pairs of @xmath2 observations from the @xmath3-th cluster , where @xmath5 is a @xmath6-dimensional known vector of covariates .
we consider the heteroscedastic nested error regression model @xmath28 where @xmath14 is a @xmath6-dimenstional unknown vector of regression coefficients , and @xmath8 and @xmath9 are mutually independent random variables with mean zero and variances @xmath29 and @xmath18 , which are denoted by @xmath30 it is noted that no specific distributions are assumed for @xmath8 and @xmath9 .
it is assumed that the heteroscedastic variance @xmath31 of @xmath9 is given by @xmath32 where @xmath33 is a @xmath34-dimensional known vector given for each cluster , and @xmath35 is a @xmath34-dimensional unknown vector .
the variance function @xmath36 is a known ( user specified ) function whose range is nonnegative .
some examples of the variance function are given below .
the model parameters are @xmath14 , @xmath37 and @xmath35 , whereas the total number of the model parameters is @xmath38 .
let @xmath39 , @xmath40 and @xmath41 . then the model ( [ model ] ) is expressed in a vector form as @xmath42 where @xmath43 is an @xmath44 vector with all elements equal to one , and the covariance matrix of @xmath45 is @xmath46 for @xmath47 and @xmath48 . it is noted that the inverse of @xmath49 is expressed as @xmath50 where @xmath51 .
further , let @xmath52 , @xmath53 , @xmath54 and @xmath55 .
then , the matricial form of ( [ model ] ) is written as @xmath56 , where @xmath57 .
now we give some examples of the variance function @xmath58 in ( [ vf ] ) .
\(a ) in the case that the dispersion of the sampling error is proportional to the mean , it is reasonable to put @xmath59 and @xmath60 for the sub - vector @xmath61 of the covariate @xmath5 . for identifiability of @xmath35 , we restrict @xmath62 .
\(b ) consider the case that @xmath0 clusters are decomposed into @xmath34 homogeneous groups @xmath63 with @xmath64 .
then , we put @xmath65 which implies that @xmath66 note that @xmath67 for @xmath68 .
thus , the models assumes that the @xmath0 clusters are divided into known @xmath34 groups with their variance are equal over the same groups .
jiang and nguyen ( 2012 ) used a similar setting and argued that the unbiased estimator of the heteroscedastic variance is consistent when @xmath69 as @xmath70 , where @xmath71 denotes the number of elements in @xmath72 .
\(c ) log linear functions of variance were treated in cook and weisberg ( 1983 ) and others .
that is , @xmath73 is a linear function , and @xmath31 is written as @xmath74 . similarly to ( a ) , we put @xmath75 . for the above two cases ( a ) and ( b ) , we have @xmath76 , while the case ( c ) corresponds to @xmath77 . in simulation and empirical studies in section
[ sec : sim ] , we use the log - linear variance model .
as given in subsequent section , we show consistency and asymptotic expression of estimators for @xmath35 as well as @xmath14 and @xmath37 .
we here provide estimators of the model parameters @xmath14 , @xmath37 and @xmath35 . when values of @xmath35 and @xmath37 are given , the vector @xmath14 of regression coefficients is estimated by the generalized least squares ( gls ) estimator @xmath78 this is not a feasible form since @xmath35 and @xmath37 are unknown .
when estimators @xmath79 and @xmath80 are for @xmath37 and @xmath35 , we get the feasible estimator @xmath81 by replacing @xmath37 and @xmath35 in @xmath82 with their estimators . concerning estimation of @xmath37 , we use the second moment of observations @xmath12 s . from model ( [ model ] ) , it is seen that @xmath83=\tau^2+\si^2(\z_{ij}'\bga).\ ] ] based on the ordinary least squares ( ols ) estimator @xmath84 , a moment estimator of @xmath37 is given by @xmath85 with substituting estimator @xmath80 into @xmath35 , where @xmath86 . for estimation of @xmath35 ,
we consider the within difference in each cluster .
let @xmath87 be the sample mean in the @xmath3-th cluster , namely @xmath88 .
it is noted that for @xmath89 , @xmath90 which dose not include the term of @xmath8 .
then it is seen that @xmath91=\left(1 - 2n_i^{-1}\right)\si^2(\z_{ij}'\bga)+n_i^{-2}\sum_{h=1}^{n_i}\si^2(\z_{ih}'\bga),\ ] ] which motivates us to estimate @xmath35 by solving the following estimating equation given by @xmath92\z_{ij}=\0,\ ] ] which is equivalent to @xmath93=\0\ ] ] where @xmath94 . it is noted that , in case of homoscedastic case , namely @xmath95 , the estimator @xmath96 and @xmath37 reduces to the estimator identical to prasad - rao estimator ( prasad and rao , 1990 ) up to the constant factor .
note that the objective function ( [ bga ] ) for estimation of @xmath35 does not depend on @xmath14 and @xmath37 and that the estimator of @xmath37 depends on @xmath35 .
these suggest the following algorithm for calculating the estimates of the model parameters : we first obtain the estimate @xmath80 of @xmath35 by solving ( [ bga ] ) , and then we get the estimate @xmath79 from ( [ tau ] ) with @xmath97 .
finally we have the gls estimate @xmath98 with substituting @xmath80 and @xmath79 in ( [ bbe ] ) .
in this section , we provide large sample properties of the estimators given in the previous subsection when the number of clusters @xmath0 goes to infinity , but @xmath2 s are still bounded . to establish asymptotic results
, we assume the following conditions under @xmath70 .
* assumption ( a ) * * there exist @xmath99 and @xmath100 such that @xmath101 for @xmath102 .
the dimensions @xmath6 and @xmath34 are bounded , namely @xmath103 .
the number of clusters with one observaion , namely @xmath104 , is bounded . *
the variance function @xmath36 is twice differentiable and its derivatives are denoted by @xmath105 and @xmath106 , respectively .
* the following matrices converge to non - singular matrices : @xmath107 for @xmath108 and @xmath109 . *
the forth moments of @xmath8 and @xmath9 exist , namely @xmath110<\infty$ ] and @xmath111<\infty$ ] . the conditions 1 and 3 are the standard assumptions in small area estimation .
the condition 2 is also non - restrictive , and the simple variance function @xmath76 and @xmath112 obviously satisfies the assumption .
the moment condition 4 is necessary for existence of mse of the eblup , and it is satisfied by many continuous distributions , including normal , shifted gamma , laplace and @xmath113-distribution with degrees of freedom larger than 5 .
in what follows , we use the notations @xmath114 for simplicity . to derive asymptotic approximations of the estimators , we define the following statistics in the @xmath3-th cluster : @xmath115.\label{u2}\end{aligned}\ ] ] moreover , we define @xmath116 noting that @xmath117 and @xmath118 under assumption ( a )
. then we obtain the asymptotically linear expression of the estimators .
[ thm : asymp ] let @xmath119 be the estimator of @xmath120 . under assumption ( a )
, it follows that @xmath121 with the asymptotically linear expression @xmath122 where @xmath123 from theorem [ thm : asymp ] , it follows that @xmath124 have an asymptotically normal distribution with mean vector @xmath125 and covariance matrix @xmath126 , where @xmath127 is a @xmath128 matrix partitioned as @xmath129 & e[\psi_i^{\bbe}\psi_i^{\bga ' } ] & e[\psi_i^{\bbe}\psi_i^{\tau}]\\ e[\psi_i^{\bga}\psi_i^{\bbe ' } ] & e[\psi_i^{\bga}\psi_i^{\bga ' } ] & e[\psi_i^{\bga}\psi_i^{\tau}]\\ e[\psi_i^{\tau}\psi_i^{\bbe ' } ] & e[\psi_i^{\tau}\psi_i^{\bga ' } ] & e[\psi_i^{\tau}\psi_i^{\tau } ] \end{array}\right).\end{aligned}\ ] ] it is noticed that @xmath130=0 $ ] and @xmath131=0 $ ] when @xmath12 are normally distributed . in such a case , it follows @xmath132 and @xmath133 , namely @xmath14 and @xmath134 are asymptotically orthogonal .
however , since we do not assume that normality for observations @xmath135 , @xmath14 and @xmath136 are not necessarily orthogonal .
the asymptotic covariance matrix @xmath137 or @xmath127 can be easily estimated from samples .
for example , @xmath138 $ ] can be estimated by @xmath139 where @xmath140 is obtained by replacing unknown parameters @xmath141 in @xmath142 with estimates @xmath143 .
it is noted that the accuracy of estimation is given by @xmath144 from theorem [ thm : asymp ] and @xmath145 .
the estimator @xmath146 will be used to get the estimators of mean squared errors of predictors in section [ sec : mse ] .
we next provide the asymptotic properties of conditional covariance matrix given in the following corollary where the proof is given in the appendix .
[ cor : cond ] under assumption ( a ) , for @xmath102 , it follows that @xmath147 this property is used for estimation and evaluating the mean squared errors of eblup discussed in the subsequent section . moreover , in the evaluation of the mean squared errors of eblup and derivation of its estimators , we need to obtain the conditional and unconditional asymptotic bias of estimators @xmath143 .
let @xmath148 and @xmath149 be the second - order conditional asymptotic bias defined as @xmath150=&\b_{\bbe}^{(i)}(\y_i)+o_p(m^{-1 } ) , \ \
e[\bgah-\bga|\y_i]=\b_{\bga}^{(i)}(\y_i)+o_p(m^{-1 } ) , \\ & e[\tah^2-\tau^2|\y_i]=\b_{\tau}^{(i)}(\y_i)+o_p(m^{-1}).\end{aligned}\ ] ] in the following theorem , we provide the analytical expressions of @xmath148 and @xmath149 .
define @xmath151 , @xmath152 and @xmath153 by @xmath154 @xmath155\right)\right\}_r\\ & \ \ \ \ \ \ \
-\sum_{k=1}^m\sum_{j=1}^{n_k}\z_{kj}\si_{kj(2)}(\z_{kj}-2n_k^{-1}\z_{kj}+n_k^{-1}\barz_k)'\bom_{\bga\bga}\z_{kj}\bigg ] , \end{split}\ ] ] and @xmath156 where @xmath157 , @xmath158 , @xmath159 for @xmath160-th element @xmath161 of @xmath162 , @xmath163 for @xmath164 and @xmath165 are defined in the proof of theorem [ thm : cond.bias ] , and @xmath166 denotes a @xmath34-dimensional vector @xmath167 .
it is noted that @xmath168 are of order @xmath169 .
now we provide the second - order approximation to the conditional asymptotic bias .
[ thm : cond.bias ] under assumption ( a ) , we have @xmath170 where @xmath171 , @xmath172 and @xmath173 are of order @xmath174 , and @xmath175 and @xmath176 are given in @xmath177 and @xmath178 , respectively . from the above theorem
, we immediately obtain the unconditional asymptotic bias of the estimators @xmath143 by taking expectation with respect to @xmath179 given in the following corollary .
[ cor : bias ] under assumption ( a ) , it follows that @xmath180=(\b_{\bbe}',\b_{\bga}',b_{\tau})'+o(m^{-1}),\ ] ] where @xmath151 , @xmath152 and @xmath153 are given in @xmath181 .
we now consider the prediction of @xmath182 where @xmath183 is a known ( user specified ) vector and @xmath8 is the random effect in model ( [ model ] ) .
the typical choice of @xmath183 is @xmath184 which corresponds to the prediction of mean of the @xmath3-th cluster .
a predictor @xmath185 of @xmath186 is evaluated in terms of the mse @xmath187 $ ] . in the general forms of @xmath185 , the minimizer ( best predictor ) of the mse
can not be obtain without a distributional assumption for @xmath8 and @xmath9 .
thus we focus on the class of linear and unbiased predictors , and the best linear unbiased predictor ( blup ) of @xmath186 in terms of the mse is given by @xmath188 this can be simplified to @xmath189 where @xmath190 for @xmath191 . in case of homogeneous variances , namely @xmath192
, it is confirmed that the blp reduces to @xmath193 with @xmath194 as given in hall and maiti ( 2006 ) .
the blup is not feasible since it depends on unknown parameters @xmath14 , @xmath35 and @xmath37 .
plugging the estimators into @xmath195 , we get the empirical best linear unbiased predictor ( eblup ) @xmath196 for @xmath197 . in the subsequent section ,
we consider the mean squared errors ( mse ) of eblup ( [ eblup ] ) without any distributional assumptions for @xmath8 and @xmath9 .
to evaluate uncertainty of eblup given by ( [ eblup ] ) , we evaluate the mse defined as @xmath198 $ ] for @xmath134 .
the mse is decomposed as @xmath199\\ & = e\left[(\mut_i-\mu_i)^2\right]+e\left[(\muh_i-\mut_i)^2\right]+2e\left[(\muh_i-\mut_i)(\mut_i-\mu_i)\right].\end{aligned}\ ] ] from the expression of @xmath195 , we have @xmath200 which leads to @xmath201=\left(\sum_{j=1}^{n_i}\la_{ij}-1\right)^2\tau^2+\sum_{j=1}^{n_i}\la_{ij}^2\si_{ij}^2=\tau^2\eta_i^{-1}.\ ] ] for the second term , however , we can not obtain an exact expression , so that we obtain the approximation up to @xmath169 . using the taylor series expansion and theorem [ thm : asymp ] , we have @xmath202&=e\left[\left\{\left(\frac{\partial\mut_i}{\partial\bth}\right)'(\bthh-\bth)\right\}^2\right]+o(m^{-1})\\ & = \tr\left\{e\left[\left(\frac{\partial\mut_i}{\partial\bth}\right)\left(\frac{\partial\mut_i}{\partial\bth}\right)'e\left((\bthh-\bth)(\bthh-\bth)'\big | \y_i\right)\right]\right\}+o(m^{-1})\\ & = \tr\left\{e\left[\left(\frac{\partial\mut_i}{\partial\bth}\right)\left(\frac{\partial\mut_i}{\partial\bth}\right)'\right]\bom\right\}+o(m^{-1})\equiv r_{2i}(\bphi)+o(m^{-1}),\end{aligned}\ ] ] where we used corollary [ cor : cond ] and the fact that @xmath203 does not depend on @xmath204 except for @xmath179 .
the straightforward calculation shows that @xmath205 where @xmath206 so that we have @xmath207 which is of order @xmath169 .
we next evaluate the cross term @xmath208 $ ] .
this term vanishes under the normality assumptions for @xmath8 and @xmath9 , but in general , it can not be neglected . as in the case of @xmath209
, we obtain an approximation of @xmath208 $ ] up to @xmath169 . in the evaluation
, we assume that @xmath210 . to this end , we expand @xmath211 as @xmath212 it follows that @xmath213 and then , @xmath214&=e\left[\left(\frac{\partial\mut_i}{\partial\bth}\right)'(\bthh-\bth)w_i\right]+\frac12e\left[\left(\frac{\partial\mut_i}{\partial\bth}\right)'(\bthh-\bth)(\bthh-\bth)'\left(\frac{\partial\mut_i}{\partial\bth}\right)w_i\right]+o(m^{-1}).\end{aligned}\ ] ] using the expression of ( [ mu.deriv ] ) and corollary [ cor : cond ] , the straightforward calculation ( whose details are given in the appendix ) shows that @xmath215=o(m^{-1}),\ ] ] under the assumption @xmath210 .
moreover , from theorem [ thm : cond.bias ] , we obtain @xmath216=r_{31i}(\bphi,\bka)+o(m^{-1}),\ ] ] for @xmath217 where @xmath218 and @xmath219 , @xmath220 is defined as @xmath221 and @xmath222 , respectively , and @xmath223 .
the derivation of the expression of @xmath224 is also given in the appendix . from the expression ( [ r31 ] ) , it holds that @xmath225 . under the normality assumption of @xmath8 and @xmath9
, we immediately obtain @xmath226 and @xmath227 since @xmath228 .
this leads to @xmath229 , which means that the cross term does not appear in the second - order approximated mse , that is our result is consistent to the well - known result .
now , we summarize the result for the second - order approximation of the mse .
[ thm : mse ] under assumption ( a ) and @xmath230=e[\ep_{ij}^3]=0 $ ] , the second - order approximation of the mse is given by @xmath231 where @xmath232 , @xmath233 and @xmath224 are given in @xmath234 , @xmath235 and @xmath236 , respectively , and @xmath237 , @xmath238 and @xmath225 .
the approximated mse given in theorem [ thm : mse ] depends on unknown parameters .
thus , in the subsequent section , we derive the second - order unbiased estimator of the mse by the analytical and the matching bootstrap methods .
we first derive the analytical second - order unbiased estimator of the mse . from theorem [ thm :
mse ] , @xmath233 is @xmath169 , so that it can be estimated by the plug - in estimator @xmath239 with second - order accuracy , namely @xmath240=r_{2i}(\bphi)+o(m^{-1})$ ] .
for @xmath224 with order @xmath169 , if a consistent estimator @xmath241 is available for @xmath242 , this term can be estimated by the plug - in estimator with second - order unbiasedness . to this end , we construct a consistent estimator of @xmath242 using the expression of fourth moment of observations .
the straightforward calculation shows that @xmath243\\ & \ \ \ = \kappa_{\ep}n_i^{-4}(n_i-1)(n_i-2)(n_i^2-n_i-1)\left(\sum_{j=1}^{n_i}\siij^4\right)+3n_i^{-3}(2n_i-3)\left\{\left(\sum_{j=1}^{n_i}\siij^2\right)^2-\sum_{j=1}^{n_i}\siij^4\right\},\end{aligned}\ ] ] whereby we can estimate @xmath244 by @xmath245 , \end{split}\ ] ] where @xmath246 and @xmath98 is feasible gls estimator of @xmath14 given in section 2 . for @xmath219
, it is observed that @xmath247=\tau^4\ka_{v}+6\tau^2\siij^2+\ka_{\ep}\siij^4,\end{aligned}\ ] ] which leads to the estimator of @xmath219 given by @xmath248 from theorem [ thm : asymp ] , it is immediately follows that the estimators given in ( [ ka.ep ] ) and ( [ ka.v ] ) are consistent . using these estimators
, we can estimate @xmath249 by @xmath250 with second - order accuracy . finally , we consider the second - order unbiased estimation of @xmath251 .
the situation is different than before since @xmath252 , which means that the plug - in estimator @xmath253 has the second - order bias with @xmath169 .
thus we need to obtain the second - order bias of @xmath253 and correct them . by the taylor series expansion , we have @xmath254 from theorem [ thm : asymp ] .
then , the second - order bias of @xmath253 is expressed as @xmath255&-r_{1i}(\bphi)\\ & = \left(\frac{\partial r_{1i}(\bphi)}{\partial\bphi'}\right)e[\bphih-\bphi]+\frac12\tr\left\{\left(\frac{\partial^2 r_{1i}(\bphi)}{\partial\bphi\partial\bphi'}\right)e\left[(\bphih-\bphi)(\bphih-\bphi)'\right]\right\}+o(m^{-1})\\ & = \left(\frac{\partial r_{1i}(\bphi)}{\partial\bphi'}\right)\b_{\bphi}+\frac12\tr\left\{\left(\frac{\partial^2 r_{1i}(\bphi)}{\partial\bphi\partial\bphi'}\right)\bom_{\bphi}\right\}+o(m^{-1}),\end{aligned}\ ] ] where @xmath256 is the sub - matrix of @xmath127 with respect to @xmath136 , and @xmath257 is the second - order bias of @xmath258 given in corollary [ cor : bias ] .
the straightforward calculation shows that @xmath259 where @xmath260 therefore , we obtain the expression of the second - order bias given by @xmath261 with @xmath262 . noting that @xmath263 can be estimated by @xmath264 with @xmath265=b_i+o(m^{-1})$ ] from theorem [ thm : asymp ] , we propose the bias corrected estimator of @xmath251 given by @xmath266 which is second - order unbiased estimator of @xmath251 , namely @xmath267=r_{1i}(\bphi)+o(m^{-1}).\ ] ] now , we summarize the result for the second - order unbiased estimator of mse in the following theorem .
[ thm : mseest ] under assumption ( a ) and @xmath230=e[\ep_{ij}^3]=0 $ ] , the second - order unbiased estimator of @xmath268 is given by @xmath269 that is , @xmath270={\rm mse}_i+o(m^{-1})$ ] .
it is remarked that the proposed estimator of mse does not require any resampling methods such as bootstrap .
this means that the analytical estimator can be easily implemented and has less computational burden compared to bootstrap .
moreover , we do not assume normality of @xmath8 and @xmath9 in the derivation of the mse estimator as in lahiri and rao ( 1995 ) .
thus the proposed mse estimator is expected to have a robustness property , which will be investigated in the simulation studies .
we first compare the performances of eblup obtained from the proposed hner with variance functions ( hnervf ) with the conventional ner and the hner with random dispersions ( hnerrd ) proposed in kubokawa , et al .
( 2014 ) in terms of simulated mse . to this end , we consider the following data generating process : @xmath271 we take @xmath272 , @xmath273 , @xmath274 , @xmath275 . for the values of @xmath276 and @xmath277 , we consider two patterns : @xmath278 .
note that @xmath279 indicates that the true model holds the heteroscedasticity in sampling variances while @xmath280 indicates the true model has homoscedastic variance in which both hner models are overfitted .
we generate @xmath281 and @xmath282 from the uniform distribution on @xmath283 and @xmath284 , respectively , which are fixed through the simulation runs .
following hall and maiti ( 2006 ) , we consider five patterns of distributions of @xmath8 and @xmath9 , that is , m1 : @xmath8 and @xmath9 are both normally distributed , m2 : @xmath8 and @xmath9 are both scaled @xmath113-distribution with degrees of freedom @xmath285 , m3 : @xmath8 and @xmath9 are both scaled and located @xmath286 distribution , m4 : @xmath8 are @xmath9 are scaled and located @xmath286 and @xmath287 distribution , respectively , and m5 : @xmath8 are @xmath9 are both logistic distribution .
based on @xmath288 simulation runs , we calculate the mse of each area defined as @xmath289 where @xmath290 and @xmath291 are obtained values of the eblup and the true values of @xmath292 in the @xmath160-th iteration , respectively . for estimation of the variance component in the ner model ,
we use the prasad and rao estimator ( prasad and rao , 1990 ) .
the resulting simulated mse values for five distribution and two values of @xmath277 are given in figure [ comp1 ] ( in case of @xmath279 ) and figure [ comp2 ] ( in case of @xmath280 ) .
from figure 1 , it is observed that the hnervf provides least values of mse in all areas .
it is a natural result that the hnerrd provides second best prediction in terms of mse values , but the mse values are not so different from the ner model .
thus the model specification is appropriate , the eblup obtained from hnervf performs so well compared to the existing models . on the other hand , in figure [ comp2 ] , the hnervf provides little larger mse values than the hnerrd and ner in normal case ( m1 ) .
it is not surprising result since the parameter @xmath277 in the hnervf is 0 in the true model and the estimation error of @xmath277 inflates the mse values .
however , in other cases ( m2@xmath293m5 ) , the hnervf provides the close mse values to the ner and hnerrd although the true model has homoscedastic variances .
thus we may conclude that the hnervf has little disadvantages of over - specification in terms of mse values of the eblup .
( heteroscedasticity ) .
[ comp1 ] , title="fig:",width=207 ] ( heteroscedasticity ) .
[ comp1 ] , title="fig:",width=207 ] ( heteroscedasticity ) .
[ comp1 ] , title="fig:",width=207 ] + ( heteroscedasticity ) .
[ comp1 ] , title="fig:",width=207 ] ( heteroscedasticity ) .
[ comp1 ] , title="fig:",width=207 ] ( homoscedasticity ) .
[ comp2 ] , title="fig:",width=207 ] ( homoscedasticity ) .
[ comp2 ] , title="fig:",width=207 ] ( homoscedasticity ) .
[ comp2 ] , title="fig:",width=207 ] + ( homoscedasticity ) .
[ comp2 ] , title="fig:",width=207 ] ( homoscedasticity ) .
[ comp2 ] , title="fig:",width=207 ] we next investigate the finite sample performances of the mse estimators given in theorem [ thm : mseest ] .
we use the same data generating process given in ( [ dgp ] ) and we take @xmath274 , @xmath275 , @xmath294 and @xmath279 .
moreover , we equally divided @xmath272 areas into four groups ( @xmath295 ) , so that each group has five areas and the areas in the same group has the same sample size @xmath296 . following the simulation study in the previous subsection
, we again consider the five patterns of distributions for @xmath8 and @xmath9 .
the simulated values of the mse are obtained from ( [ sim - mse ] ) based on @xmath288 simulation runs .
then , based on @xmath297 simulation runs , we calculate the relative bias ( rb ) and coefficient of variation ( cv ) of mse estimators given by @xmath298 where @xmath299 is the mse estimator in the @xmath160-th iteration . in table
[ mse - sim ] , we report mean and median values of @xmath300 and @xmath301 in each group . for comparison , results for the naive mse estimator , without any bias correction , are reported in table [ mse - sim ] as well .
the naive mse estimator is the plug - in estimator of the asymptotic mse ( [ r1 ] ) , namely it is obtained by replacing @xmath37 and @xmath35 in formula ( [ r1 ] ) by @xmath79 and @xmath80 , respectively . in table
[ mse - sim ] , the relative bias is small , less than 10% in many cases .
when the underlying distributions leave from normality , the mse estimator still provides small relative bias although it has higher coefficient of variation .
the naive mse estimator is more biased than the analytical mse estimator in all groups and models , so that the bias correction in mse estimator is successful .
we now investigate empirical performances of the suggested model , the empirical bayes estimator and the second - order unbiased estimator of mse through analysis of real data .
the data used here originates from the posted land price data along the keikyu train line in 2001 .
this train line connects the suburbs in the kanagawa prefecture to the tokyo metropolitan area .
those who live in the suburbs in the kanagawa prefecture take this line to work or study in tokyo everyday .
thus , it is expected that the land price depends on the distance from tokyo .
the posted land price data are available for 52 stations on the keikyu train line , and we consider each station as a small area , namely , @xmath302 . for the @xmath3-th station ,
data of @xmath2 land spots are available , where @xmath2 varies around 4 and some areas have only one observation . for @xmath303
, @xmath12 denotes the value of the posted land price ( yen/10,000 ) for the unit meter squares of the @xmath304-th spot , @xmath305 is the time to take from the nearby station @xmath3 to the tokyo station around 8:30 in the morning , @xmath306 is the value of geographical distance from the spot @xmath304 to the station @xmath3 and @xmath307 denotes the floor - area ratio , or ratio of building volume to lot area of the spot @xmath304 .
this data set is treated in kubokawa , et al .
( 2014 ) , where they pointed out that the heteroscedasticity seem to be appropriate from boxplots of some areas and bartlet test for testing homoscedastic variance .
figure [ plp ] is the plot of the pairs @xmath308 , where @xmath309 is ols residuals given by @xmath310 .
it indicates that the residuals are more variable for small @xmath306 than for large @xmath306 , namely the variances seem functions of @xmath306 .
thus we apply the following hner model with a variance function given by @xmath311 where @xmath312 and @xmath313 . for the variance function @xmath314 , we use @xmath315 motivated from figure [ plp ] . as a submodel of ( [ plp - model ] )
, we also consider the homoscedastic variance model with @xmath316 .
then the estimated values of parameters in these two models are given in the following : @xmath317 the estimated values of @xmath318 and @xmath319 , coefficients of @xmath320 and @xmath306 , in both models are negative values which leads to the natural result that the @xmath320 and @xmath306 have negative influence on @xmath12 .
the sign of @xmath321 is negative .
this corresponds to the variability illustrated in figure [ plp ] .
the obtained values of eblup given in ( [ eblup ] ) are given in table [ plp - res ] for selected 15 areas .
to see the difference of predicted values in terms of the degree of shrinkage , we compute @xmath322 for each two model and the results are given in figure [ plp2 ] .
it is observed that @xmath323 in ner model decreases as the area sample size @xmath2 gets large .
this is because the sample mean provides better estimates of the true mean as @xmath2 gets larger , so that the sample mean does not need to be shrunk . on the other hand , @xmath323 in hnervf
is influenced by the estimated heteroscedastic variance @xmath324 as well as @xmath2 .
thus the plot in figure [ plp2 ] shows that the shrinkage degrees in hnervf has more variability than that in ner . in table [ plp - res ] and figure [ plp ]
, we also provide the estimates of squared root of mse ( smse ) given in theorem [ thm : mseest ] .
it is revealed from table [ plp - res ] that the estimates of the smse in ner get smaller as @xmath2 gets larger . on the other hands ,
the smse in hnervf do not have a similar property , because the smse in hnervf is affected by not only @xmath2 but also the heteroscedastic variance as indicated in the mse formula given in theorem [ thm : mse ] . from figure [ plp ] , we observe that the estimated smses of hnervf are smaller than that of ner in many areas .
especially , in area 47 , 49 , 50 and 51 , the smse values of hnervf are dramatically small compared to ner . in some other areas , the smess of hnervf is larger than that of ner , but the differences are not so large .
these observations and the residual plot in figure [ plp ] motivate us to utilize the hnervf in case of heteroscedastic variance explained by some covariates .
( left ) and estimated mse in hnervf and ner ( right ) . , title="fig:",width=309 ] ( left ) and estimated mse in hnervf and ner ( right ) . , title="fig:",width=309 ] against area sample size @xmath2 in hnervf and ner , title="fig:",width=309 ] against area sample size @xmath2 in hnervf and ner , title="fig:",width=309 ] .the estimated results of plp data for selected 15 areas [ cols="^,^,^,^,^,^,^,^,^,^ " , ]
in the context of small - area estimation , homogeneous nested error regression models have been studied so far in the literature . however , some real data sets show heteroscedasticity in variances as pointed out in jiang and nguyen ( 2012 ) and kubokawa , et al . ( 2014 ) . in such a case ,
the residuals often indicate that the heteroscedasticity can be explained by some covariates , which motivated us to propose and investigate the heteroscedastic nested error regression model with variance functions ( hnervf ) .
we have proposed the estimating method for the model parameters and the asymptotic properties of these estimators have been established without any distributional assumptions for error terms . for measuring uncertainty of the empirical bayes estimator , the mean squared errors ( mse ) have been approximated up to second - order , and their second - order unbiased estimators have been provided in the closed form . for estimation of mse in hnervf without distributional assumptions , we can utilize the moment matching bootstrap method proposed in hall and maiti ( 2006 ) . to derive the bootstrap estimator
, we use the following representation of mse : @xmath325 + 2e\left[(\muh_i-\mut_i)(\mut_i-\mu_i)\right],\ ] ] noting that the second and third terms are @xmath169 .
then we can establish the second - order unbiased mse estimator via three - point distribution or @xmath113-distribution for approximation of distribution of error terms .
however , the bootstrap method has computational burden , so that we did not treat in this paper .
* acknowledgments . *
+ the first author was supported in part by grant - in - aid for scientific research ( 15j10076 ) from japan society for the promotion of science ( jsps ) .
the second author was supported in part by grant - in - aid for scientific research ( 23243039 and 26330036 ) from japan society for the promotion of science .
proof of theorem 1 .
* since @xmath326 are mutually independent , the consistency of @xmath79 and @xmath80 follows from the standard argument of m - estimators , so that @xmath98 is also consistent . in what follows , we derive the asymptotic expressions of the estimators .
for the asymptotic expansion of @xmath80 defined as the minimizer of ( [ bga ] ) .
remember that the estimator @xmath80 is given as the solution of the estimating equation @xmath332=\0\ ] ] using taylor expansions , we have @xmath333 where @xmath334.\ ] ] from the central limit theorem , it follows that @xmath335 so that the second terms in the expansion formula is @xmath336 .
then we get @xmath337 under assumption ( a ) , we have @xmath338 from the independence of @xmath204 and the fact @xmath339 , we can use the central limit theorem to show that the leading term in the expansion of @xmath340 is @xmath341 .
thus , @xmath342 finally we consider the asymptotic expansion of @xmath344 . from the expression in ( [ bbe ] )
, it follows that @xmath345 since @xmath346 for @xmath347 , we have @xmath348 where @xmath349 under assumption ( a ) , we have @xmath350 for @xmath164 , whereby @xmath351 . since @xmath343 and @xmath352 as shown above , we get @xmath353 which completes the proof .
proof of corollary [ cor : cond ] .
* let @xmath354 .
note that @xmath355 does not depend on @xmath356 and that @xmath204 are mutually independent .
then , @xmath357=\frac{1}{m^2}\sum_{j=1,j\neq i}^{m}e\left[\psi_j^{\theta_k}\psi_j^{\theta_l}\right]+\frac{1}{m^2}\psi_i^{\theta_k}\psi_i^{\theta_l}\\ & = \bom_{kl}+\frac{1}{m^2}\left\{\psi_i^{\theta_k}\psi_i^{\theta_l}-e\left[\psi_i^{\theta_k}\psi_i^{\theta_l}\right]\right\}=\bom_{kl}+o_p(m^{-1}),\end{aligned}\ ] ] where @xmath358 is the @xmath359-element of @xmath127 and we used the fact that @xmath360=e[\psi_j^{\theta_k}]=0 $ ] for @xmath361 .
hence , we get the result from the asymptotic approximation of @xmath143 given in theorem [ thm : asymp ] .
proof of theorem [ thm : cond.bias ] . *
we begin by deriving the conditional asymptotic bias of @xmath80 .
let @xmath362 be the solution of the equation @xmath363=\0\ ] ] with @xmath364 . for notational simplicity
, we use @xmath365 instead of @xmath366 without any confusion and @xmath367 denotes the @xmath160-th component of @xmath365 , namely @xmath368 . define the derivatives @xmath369 and @xmath370 by @xmath371 it is noted that @xmath372 . expanding @xmath373 , we obtain @xmath374 where @xmath375 for @xmath376 it is also noted that @xmath377 so that @xmath378 is non - stochastic .
thus we have @xmath379=-(\f_{(\bga)})^{-1}\left\{e[\f(\bga;\bbe)|\y_i]+e\left[\f_{(\bbe)}(\obbeh-\bbe)\big|\y_i\right]+\frac12e[\t_1|\y_i]+\frac12e[\t_2|\y_i]\right\}+o_p(m^{-1}).\end{aligned}\ ] ] in what follows , we shall evaluate the each term in the parenthesis in the above expression .
for the first term , since @xmath204 are mutually independent and @xmath339 , we have @xmath380={1\over m } \u_{2i}.\ ] ] for evaluation of the second term , we define @xmath159 , where @xmath161 denotes the @xmath160-th element of @xmath162 .
then it follows that @xmath381=-\frac2n\sum_{k=1}^m e\left[(\y_k-\x_k\bbe)'\e_k\z_{kr}\e_k\x_k(\obbeh-\bbe)\bigg|\y_i\right]\\ & = -\frac{2}{n}\sum_{k=1,k\neq i}^me\left[(\y_k-\x_k\bbe)'\e_k\z_{kr}\e_k\x_k(\obbeh-\bbe)\big|\y_i\right]-\frac{2}{n}(\y_i-\x_i\bbe)'\e_i\z_{ir}\e_i\x_ie\left[\obbeh-\bbe\big|\y_i\right].\end{aligned}\ ] ] noting that it holds for @xmath382 and @xmath383 @xmath384=1_{\{\ell = k\}}\bsi_k , \ \ \ \ e[\obbeh-\bbe|\y_i]=\left(\x'\x\right)^{-1}\x_i'(\y_i-\x_i\bbe),\ ] ] we have @xmath385&=\sum_{\ell=1}^m\tr\left\{\e_k\z_{kr}\e_k\x_k(\x'\x)^{-1}\x_k'e\left[(\y_\ell-\x_\ell\bbe)(\y_k-\x_k\bbe)'\big|\y_i\right]\right\}\\ & = \tr\left\{(\x'\x)^{-1}\x_k'\bsi_k\e_k\z_{kr}\e_k\x_k\right\},\end{aligned}\ ] ] which is @xmath169 and @xmath386=o_p(m^{-1}).\ ] ] thus , we get @xmath387=-\frac2m\sum_{k=1}^m\sum_{j=1}^{n_k } \tr\left\{(\x'\x)^{-1}\x_k'\bsi_k\e_k\z_{kr}\e_k\x_k\right\}+o_p(m^{-1}),\ ] ] where the leading term is @xmath169 . for the third and forth terms ,
note that @xmath388 which are non - stochastic .
then for @xmath389 , @xmath390&=-\frac1n\sum_{k=1}^m\sum_{j=1}^{n_k}z_{kjr}\si_{kj(2)}^2(\z_{kj}-2n_k^{-1}\z_{kj}+n_k^{-1}\barz_k)'\bom_{\bga\bga}\z_{kj}+o_p(m^{-1}),\\ e[t_{2r}|\y_i]&=\frac2n\sum_{k=1}^m\tr\left(\x_k'\e_k\z_{kr}\e_k\x_k\v_{\rm ols}\right)+o_p(m^{-1}),\end{aligned}\ ] ] for @xmath158 , where we used corollary [ cor : cond ] and @xmath391=\v_{\rm ols}+o_p(m^{-1}),\ ] ] which follows from the similar argument in the proof of corollary [ cor : cond ]
. thus we obtain @xmath392&=-\frac1n\sum_{k=1}^m\sum_{j=1}^{n_k}\z_{kj}\si_{kj(2)}^2(\z_{kj}-2n_k^{-1}\z_{kj}+n_k^{-1}\barz_k)'\bom_{\bga\bga}\z_{kj}+o_p(m^{-1}),\\ e[\t_2|\y_i]&=\frac2n\sum_{k=1}^m\left\{\tr\left(\x_k'\e_k\z_{kr}\e_k\x_k\v_{\rm ols}\right)\right\}_r+o_p(m^{-1}),\end{aligned}\ ] ] where @xmath393 denotes the @xmath34-dimensional vector @xmath394 .
therefore , we establish the result for @xmath80 in ( [ cond.bias ] ) .
we next derive the result for @xmath79 .
let @xmath395 using the taylor series expansion , we have @xmath396 where we used the fact that @xmath397 .
the straight calculation shows that @xmath398 which are non - stochastic .
thus we obtain @xmath399=e[\tat^2-\tau^2|\y_i]+\left(\frac{\partial\tat^2}{\partial \bga}\right)'e\left[\bgah-\bga|\y_i\right]+\frac12\tr\left\{\left(\frac{\partial^2\tat^2}{\partial\bga\partial\bga'}\right)e\left[(\bgah-\bga)(\bgah-\bga)'\big|\y_i\right]\right\}\\ & + e\left[\left(\frac{\partial\tat^2}{\partial \bbe}\right)'(\obbeh-\bbe)\bigg|\y_i\right]+\frac12\tr\left\{\left(\frac{\partial^2\tat^2}{\partial\bbe\partial\bbe'}\right)e\left[(\obbeh-\bbe)(\obbeh-\bbe)'\big|\y_i\right]\right\}+o_p(m^{-1})\\ & \equiv b_{\tau 1}(\y_i)+b_{\tau 2}(\y_i)+b_{\tau 3}(\y_i)+b_{\tau 4}(\y_i)+b_{\tau 5}(\y_i)+o_p(m^{-1}).\end{aligned}\ ] ] from the expression of @xmath400 , it holds that @xmath401 for @xmath175 defined in ( [ u1 ] ) .
also , we immediately have @xmath402 for evaluation of @xmath403 , note that @xmath404 similarly to ( [ ols.formula ] ) , we get @xmath405\\ & = -\frac{2}{n}\sum_{k=1}^m\tr\left\{(\x'\x)^{-1}\x_k'\bsi_k\x_k\right\}+o_p(m^{-1}).\end{aligned}\ ] ] moreover , corollary [ cor : cond ] and ( [ ols.cond ] ) enable us to obtain the expression of @xmath406 and @xmath407 , whereby we get @xmath408 which completes the proof for @xmath79 in ( [ cond.bias ] ) .
we finally derive the result for @xmath98 . by the taylor series expansion , @xmath409 since @xmath410 from @xmath411 as shown in the proof of theorem [ thm : asymp ] . from ( [ bbet.ast ] ) , we have @xmath412 and @xmath413 let @xmath414 $ ] and @xmath415 $ ]
. then it can be shown that @xmath416=\bom_{\beta^{\ast}\ga_s}+o_p(m^{-1 } ) , \ \ \ e[(\bbet_{\ga_s}^{\ast}-\bbe)(\gah_s-\ga_s)|\y_i]=\bom_{\beta^{\ast}\tau}+o_p(m^{-1}),\ ] ] which can be proved by the same arguments as in corollary [ cor : cond ] .
thus from corollary [ cor : cond ] and the fact that @xmath417=\left(\x'\bsi^{-1}\x\right)^{-1}\x_i'\bsi_i^{-1}(\y_i-\x_i\bbe),\ ] ] we obtain the result for @xmath98 in ( [ cond.bias ] ) . * a4 .
derivation of @xmath224 .
* since @xmath179 given @xmath418 is non - stochastic , we have @xmath419&=e\left[e\left[\left(\frac{\partial\mut_i}{\partial\bth}\right)'(\bthh-\bth)w_i\bigg| v_i,\bep_i\right]\right]\\ & = e\left[e(\bthh-\bth |\y_i)'\left(\frac{\partial\mut_i}{\partial\bth}\right)w_i\right]\\ & = e\left[\b_{\bbe}^{(i)}(\y_i)'\left(\frac{\partial\mut_i}{\partial\bbe}\right)w_i\right]+e\left[\b_{\bga}^{(i)}(\y_i)'\left(\frac{\partial\mut_i}{\partial\bga}\right)w_i\right]+e\left[b_{\tau}^{(i)}(\y_i)\left(\frac{\partial\mut_i}{\partial\tau}\right)w_i\right]+o(m^{-1})\\ & \equiv r_{31i}(\bphi)+o(m^{-1}).\end{aligned}\ ] ] it is noted that @xmath420 and @xmath421=e\left[(v_i+\ep_{ij})w_i\right]=\left(\sum_{j=1}^{n_i}\la_{ij}-1\right)\tau^2+\sum_{j=1}^{n_i}\la_{ij}\siij^2=0.\ ] ] using the expression ( [ cond.bias ] ) and ( [ mu.deriv ] ) , it follows that @xmath422&=\left(\c_i-\sum_{j=1}^{n_i}\la_{ij}\x_{ij}\right)'\left(\x'\bsi^{-1}\x\right)^{-1}\x_i'\bsi_i^{-1}e\big[(\y_i-\x_i\bbe)w_i\big]=0\\ e\left[\b_{\bga}^{(i)}(\y_i)'\left(\frac{\partial\mut_i}{\partial\bga}\right)w_i\right]&=\eta_i^{-2}\sum_{j=1}^{n_i}\siij^{-2}\bde_{ij}'\left(\sum_{k=1}^{m}\sum_{h=1}^{n_k}\si_{kh(1)}^2\z_{kh}\z_{kh}'\right)^{-1}\m_{2ij}(\bphi,\bka)\\ e\left[b_{\tau}^{(i)}(\y_i)\left(\frac{\partial\mut_i}{\partial\tau}\right)w_i\right]&=m^{-1}\eta_i^{-2}\sum_{j=1}^{n_i}\siij^{-2}\bigg\{m_{1ij}(\bphi,\bka)-\t_1(\bga)'\t_2(\bga)\m_{2ij}(\bphi,\bka)\bigg\},\end{aligned}\ ] ] where @xmath423 , \ \ \ \ m_{1ij}(\bphi,\bka)=e\left[u_{1i}(y_{ij}-\x_{ij}'\bbe)w_i\right].\ ] ] to evaluate @xmath424 and @xmath425 , we first prove the following result for fixed @xmath426 .
@xmath427&=\tau^2\eta_i^{-1}\bigg[\tau^2(3-\kappa_v)+\kappa_{\ep}\siij^21_{\{j = k=\ell\}}+\siij^2(1_{\{j = k\neq \ell\}}-1_{\{j = k\}})\\ & \ \ \ \ \ + \siij^2(1_{\{j=\ell\neq k\}}-1_{\{j=\ell\}})+\si_{ik}^2(1_{\{k=\ell\neq j\}}-1_{\{k=\ell\}})\bigg ] .
\end{split}\ ] ] to show ( [ iden2 ] ) , we note that the left side can be rewritten as @xmath428+\sum_{h=1}^{n_i}\la_{ih}e\left[(v_i+\ep_{ij})(v_i+\ep_{ik})(v_i+\ep_{i\ell})\ep_{ih}\right]\ ] ] from the definition of @xmath429 .
using the fact that @xmath430 and @xmath8 are independent , the first term in ( [ iden3 ] ) is calculated as @xmath431=\ka_v\tau^4+\tau^2\left(\siij^21_{\{j = k\}}+\siij^21_{\{j=\ell\}}+\si_{ik}^21_{\{k=\ell\}}\right).\end{aligned}\ ] ] moreover , we have @xmath432=e\left[\ep_{ih}(\ep_{ij}+\ep_{i\ell}+\ep_{ik})v_i^2+\ep_{ij}\ep_{ik}\ep_{i\ell}\ep_{ih}\right]\\ & = \tau^2\si_{ih}^2\left(1_{\{h = j\}}+1_{\{h = k\}}+1_{\{h=\ell\}}\right)+\ka_{\ep}\siih^41_{\{j = k=\ell = h\}}+\siih^2\left(\siij^21_{\{j = k\neq \ell = h\}}+\siij^21_{\{j=\ell\neq k = h\}}+\si_{ik}^21_{\{j = h\neq k=\ell\}}\right),\end{aligned}\ ] ] whereby the second term in ( [ iden3 ] ) can be calculated as @xmath433,\ ] ] where we used the expression @xmath434 .
then we established the result ( [ iden2 ] ) . from ( [ iden2 ] ) , we immediately have @xmath435&=\tau^2\eta_i^{-1}\left[n_i\tau^2(3-\ka_v)+\siij^2(\ka_{\ep}-3)1_{\{j = k\}}\right]\\ & = e\big[(v_i+\ep_{ij})(v_i+\ep_{ik})^2w_i\big].\end{aligned}\ ] ] now , we return to the evaluation of @xmath424 and @xmath425 .
it follows that @xmath436\\ & = mn^{-1}\eta_i^{-1}\tau^2\big\{n_i\tau^2(3-\ka_v)+\siij^2(\ka_{\ep}-3)\big\}\end{aligned}\ ] ] and @xmath437\\ & = \frac{m}{n}\sum_{h=1}^{n_i}\z_{ih}\bigg\{e\left[(v_i+\ep_{ih})^2(v_i+\ep_{ij})w_i\right]-2n_i^{-1}\sum_{k=1}^{n_i}e\left[(v_i+\ep_{ij})(v_i+\ep_{ik})(v_i+\ep_{ih})w_i\right]\\ & \ \ \ \ + n_i^{-2}\sum_{k=1}^{n_i}\sum_{\ell=1}^{n_i}e\left[(v_i+\ep_{ij})(v_i+\ep_{ik})(v_i+\ep_{i\ell})w_i\right]\bigg\}.\end{aligned}\ ] ] using the identity given in ( [ iden2 ] ) , we have @xmath438 which completes the result in ( [ r31 ] ) .
evaluation of @xmath439 .
* since @xmath179 given @xmath418 is non - stochastic , we have @xmath440\\ & = \frac12e\left[e\left[\left(\frac{\partial\mut_i}{\partial\bth}\right)'(\bthh-\bth)(\bthh-\bth)'\left(\frac{\partial\mut_i}{\partial\bth}\right)w_i\bigg| v_i,\bep_i\right]\right]\\ & = \frac12e\left[\left(\frac{\partial\mut_i}{\partial\bth}\right)'e\left[(\bthh-\bth)(\bthh-\bth)'|\y_i\right]\left(\frac{\partial\mut_i}{\partial\bth}\right)w_i\right]=\frac12\tr\left\{\bom e\left[\left(\frac{\partial\mut_i}{\partial\bth}\right)\left(\frac{\partial\mut_i}{\partial\bth}\right)'w_i\right]\right\}+o(m^{-1}),\end{aligned}\ ] ] where we used corollary [ cor : cond ] in the last equation .
note that @xmath441&=-\eta_i^{-1}e\left[(v_i+\ep_{ij})^2v_i\right]+\sum_{h=1}^{n_i}e\left[(v_i+\ep_{ij})^2\ep_{ih}\right]=0\end{aligned}\ ] ] since @xmath230=0 $ ] and @xmath442=0 $ ] . using the expression ( [ mu.deriv ] ) of @xmath443 with the above moment results , we obtain @xmath444=0,\ ] ] which leads to @xmath445 . | the article considers a nested error regression model with heteroscedastic variance functions for analyzing clustered data , where the normality for the underlying distributions is not assumed .
classical methods in normal nested error regression models with homogenous variances are extended in the two directions : heterogeneous variance functions for error terms and non - normal distributions for random effects and error terms .
consistent estimators for model parameters are suggested , and second - order approximations of their biases and variances are derived . the mean squared errors of the empirical best linear unbiased predictors are expressed explicitly to second - order .
second - order unbiased estimators of the mean squared errors are provided analytically in closed forms . the proposed model and the resulting procedures are numerically investigated through simulation and empirical studies . |
excesses in searching for diboson resonance using boson - tagged jets were recently reported by the atlas collaboration @xcite .
it shows local excesses in the @xmath4 , @xmath0 and @xmath1 channels with significance of @xmath5 , @xmath6 and @xmath7 respectively .
similarly , the cms collaboration @xcite has reported an excess of @xmath8 significance in the dijet resonance channel and @xmath9 channel which may arise from @xmath10 with @xmath11 decaying hadronically .
these excesses may be evidences of new symmetries or new particles near @xmath12 .
since the resonances decay into two gauge boson , they should be bosonic states . possible origins of this excess were studied by several groups @xcite , where the excesses were explained as spin-1 gauge bosons @xcite in an extended gauge group , composite spin-1 resonances @xcite , spin-0 or spin-2 composite particles @xcite and extra scalar bosons @xcite .
the key points in explaining the excesses are the interactions of new resonance with the standard model ( sm ) gauge bosons , quarks and(or ) gluons , the former of which is relevant to the branching ratio of the new resonance and the latter of which is relevant to the production of the new resonance at the lhc .
one the one hand , one needs the couplings of new interactions to be large enough so as to give rise to a sizable production cross section at the lhc , on the other hand the strengths of these interactions should be consistent with current constraints of colliders and electroweak precision measurements .
these two requirements are mutual restraint .
a new resonance is not able to explain the atlas excesses if its interaction strengths are not mutually compatible with these two requirements . in this paper , we explain the atlas excesses in the stealth doublet model , where the second higgs doublet , @xmath13 , gets no vacuum expectation value , with mass near 2 tev , and only the cp - even part of @xmath13 mixes with the sm higgs boson .
we assume @xmath13 has sizable yukawa interaction with the first generation quarks , which is consistent with constraints of flavor physics .
such that the heavy cp - even higgs boson can be produced at the lhc via the yukawa interaction and decays into diboson states through the mixing with the sm higgs boson .
our numerical simulations show that one has @xmath14 by setting @xmath15 and @xmath16 , where @xmath17 is the yukawa coupling of the @xmath13 with the first generation quarks and @xmath18 is the mixing angle between two cp - even neutral states .
this result is consistent with current constraints from colliders and electroweak precision measurements .
the remaining of the paper is organized as follows : in section ii we give a brief introduction to the model .
section iii is the study of constraints on the model .
we investigate the atlas diboson excesses arising from this stealth doublet model in section iv .
the last part is the concluding remarks .
we work in the modified stealth doublet model @xcite , where the second higgs doublet gets no vacuum expectation value ( vev ) but its cp - even part mixes with the sm higgs boson . in the following , we describe the modified stealth doublet model first , and then study its implications in the atlas diboson excesses .
the higgs potential is the same as that in the general two higgs doublet model ( 2hdm ) , which can be written as @xmath19 in this paper , we assume the higgs potential is cp - conserving , so all couplings in eq.([potential ] ) are real .
only one higgs doublet gets nonzero vev in the stealth doublet model , we take it be @xmath20 .
the tadpole conditions for the electroweak symmetry breaking become @xmath21 where @xmath22 . after spontaneous breaking of the electroweak symmetry , there are two cp - even scalars @xmath11 and @xmath23 , one cp - odd scalar @xmath24 and two charged scalars @xmath25 , the mass eigenvalues of which can be written as @xcite @xmath26 the mixing angle @xmath18 between @xmath11 and @xmath23 can be calculated directly , we take it as a new degree of freedom in this paper . @xmath23
interacts with dibosons through the mixing .
we refer the reader to ref .
@xcite for the feynman rules of higgs interactions .
the yukawa interactions of @xmath20 with sm fermions are exactly the same as yukawa interactions of the sm higgs with fermions in the sm .
we assume @xmath13 has sizable yukawa coupling with the first generation quarks : @xmath27 where @xmath28 and @xmath29 .
since @xmath30 , there is almost no constraint on this yukawa coupling , and @xmath23 can be produced at the lhc via this interaction .
before proceeding to study atlas diboson excesses , let us investigate constraints on the mixing angle @xmath18 .
couplings of the sm - like higgs to other sm particles were measured by the atlas and cms collaboration .
comparing with sm higgs couplings , couplings of @xmath11 and @xmath31 to all sm states ( except @xmath32 quark ) are rescaled by @xmath33 and @xmath34 , respectively : @xmath35 where @xmath36 represents sm states .
thus signal rates of the higgs measurements relative to sm higgs expectations are the function of @xmath33 . performing a global @xmath37 fit to the higgs data given by atlas and cms
, one has @xmath38
@xcite , at the @xmath39 confidence level .
another constraint comes from the oblique parameters @xcite , which are defined in terms of contributions to the vacuum polarizations of gauge bosons .
the explicit expressions of @xmath40 and @xmath41 , which involve effects of all scalars , can be written as @xcite @xmath42 + c^2 { \cal b}_{2 } ( m_z^2 ; m_h^2 , m_{a}^{2 } ) \right .
\label{scal } & & \left.+ s^2 { \cal b}_{2 } ( m_z^2 ; m_h^2 , m_{a}^ { 2 } ) -{\cal b}_{2 } ( m_z^2 ; m_c^2,m_c^2){\over}\right\}\\ \delta t&= & { 1\over 4\pi s_w^2 m_w^2 } \ { s^2 b_{2 } ( 0 ; m_c^2 , m_h^2)+c^2 b_{2}(0;m_c^2,m_h^2 ) + b_{2 } ( 0;m_c^2,m_{a}^{2 } ) -s^2 b_{2 } ( 0 ; m_h^2 , m_{a}^2 ) \nonumber \\ & & -c^2 b_{2}(0;m_h^2,m_a^2 ) -s^2 b_{2 } ( 0;m_w^2 , m_h^2 ) + s^2 b_{2 } ( 0 ; m_w^2 , m_h^2 ) + s^2 b_{2 } ( 0 ; m_z^2,m_h^2 ) \nonumber \\ & & -s^2 b_{2}(0 ; m_z^2 , m_h^2)+m_w^2 s^2 [ b_0(0;m_w^2,m_h^2 ) -b_0 ( 0 ; m_w^2 , m_h^2 ) ] \nonumber \\&&+m_z^2 s^2 [ - b_0 ( 0,m_w^2 , m_h^2 ) + b_0(0;m_z^2,m_h^2)]-{1\over 2 } a_0(m_c^2 ) \ } \label{tcal}\end{aligned}\ ] ] where @xmath43 , @xmath44 , the expressions of @xmath45 and @xmath46 can be find in ref .
@xcite , @xmath47 and @xmath48 , @xmath49 with @xmath50 the weak mixing angle , @xmath51 and @xmath52 are masses of @xmath53 and @xmath54 bosons respectively . plane by setting @xmath55 and @xmath56 ; right panel : constraints on the masses of the charged and cp - odd neutral states from oblique parameters by setting @xmath57 and @xmath58 . ,
title="fig:",scaledwidth=46.0% ] plane by setting @xmath55 and @xmath56 ; right panel : constraints on the masses of the charged and cp - odd neutral states from oblique parameters by setting @xmath57 and @xmath58 .
, title="fig:",scaledwidth=46.0% ] the most recent electroweak fit ( by setting @xmath59 and @xmath60 ) to the oblique parameters performed by the gfitter group @xcite yields @xmath61 the @xmath62 can be written as @xmath63 where @xmath64 and @xmath65 ; @xmath66 with @xmath67 and @xmath68 . as can be seen from eqs .
( [ scal ] ) and ( [ tcal ] ) , there are four free parameters contributing to the oblique parameters , @xmath69 , @xmath70 , @xmath71 and @xmath18 . to preform electroweak fit , we set @xmath72 , which can be easily achieved by setting @xmath73 , and @xmath57 , so that only two free parameters left .
blue points in the left panel of fig.[fig : stu ] show the contribution to the @xmath40 and @xmath41 by setting @xmath74 and @xmath34 random parameters varying in the range @xmath75 and @xmath76 respectively .
the contour in the same plot shows the allowed region in the @xmath77 plane in the @xmath39 c.l .
a direct numerical calculation shows that @xmath78 . in the right panel of fig .
[ fig : stu ] we show the region that are allowed by the oblique observations in the @xmath79 plane by setting @xmath80 and @xmath56 . to summarize , electroweak precision measurements put stronger constraint on the @xmath18 even for the nearly degenerate heavy states . as the function of @xmath71 by setting @xmath81 and @xmath82
; right panel : production cross section of the heavy cp - even higgs boson at the lhc by setting @xmath82 , with solid , dotted and dashed lines correspond to @xmath83 respectively .
, title="fig:",scaledwidth=46.0% ] as the function of @xmath71 by setting @xmath81 and @xmath82 ; right panel : production cross section of the heavy cp - even higgs boson at the lhc by setting @xmath82 , with solid , dotted and dashed lines correspond to @xmath83 respectively .
, title="fig:",scaledwidth=46.0% ]
heavy scalar states in our model can be produced at the lhc through its yukawa interaction with the first generation quarks as was shown in eq .
( [ yukawah ] ) and can decay into diboson final states from the mixing with the sm - like higgs boson .
the main decay channels of @xmath23 are @xmath84 , @xmath85 , @xmath86 and @xmath1 .
the decay rates can be written as @xmath87 where @xmath88 , being the color index ; @xmath89 respectively .
we show in the left panel of fig .
[ fig : br&pro ] the branching ratios of @xmath23 by setting @xmath90 and @xmath82 , where the solid , dotted and dashed lines correspond to the branching ratios of @xmath91 , @xmath84 and @xmath85 respectively .
we plot in the right panel of fig .
[ fig : br&pro ] the production cross section of @xmath23 at the lhc .
the solid , dotted and dashed lines correspond to @xmath92 and @xmath93 , respectively . in the @xmath94 plane .
the blue dotted , black solid and green dashed lines correspond to @xmath95 respectively .
the region below the gray solid line satisfies @xmath96 .
the region below the red dot - dashed line satisfies @xmath97 .
the region below the cyan solid line has @xmath98 .
, scaledwidth=46.0% ] we show in fig .
[ fig : crosss ] the contours of @xmath99 in the @xmath100 plane .
the dashed , solid and dotted lines correspond to @xmath101 respectively
. one can get similar numerical results for the @xmath102 process .
the atlas reported number of excesses is about @xmath103 events near the @xmath12 peak .
given a luminosity of @xmath104 , one has @xmath105 for a @xmath106 @xcite selection efficiency of the event topology and boson - tagging requirements .
although large enough cross section can be produced at the lhc , the model is constrained by other lhc experimental results .
we will discuss these constraints one - by - one as follows : * the cms collaboration @xcite has reported an upper bound for the @xmath107 , where @xmath108 is a new resonance .
it gives @xmath109 .
the resonance can be the charged component of the heavy scalar doublet in our model .
its decay rate can be written as @xmath110 where @xmath111 and @xmath112 is the @xmath113 gauge coupling .
[ fig : crosss ] the numerical results by setting @xmath114 , where the region below the red dot - dashed line satisfy this constraint .
* the cp - odd component of the heavy scalar doublet can be the mediator of the process @xmath115 , which was also measured the cms collaboration .
one has @xmath116 the decay rate of @xmath117 can be written as @xmath118 where @xmath119 with @xmath50 the weak mixing angle .
@xmath24 can also decay into dijet final states with the decay rate the same as eq .
( [ dr1 ] ) .
we show in fig .
[ fig : crosss ] the numerical results , where the region to the top - right of the gray solid line are excluded by this constraint . * both atlas and cms has searched for resonances decaying into dijets .
we use @xmath120 with the acceptance @xmath121 . both the cp - even and the cp - odd heavy scalars as well as the charged scalar in our model mainly decay into dijet via the yukawa interaction .
we show in fig .
[ fig : crosss ] the region ( to the bottom - right corner of the cyan solid line ) allowed by this constraint .
since the decay rate of @xmath122 is tiny , there is almost no constraint on the model from @xmath123 resonance searches . as can be seen from fig .
[ fig : crosss ] , @xmath124 should be less than @xmath125 fb .
one has @xmath126 for @xmath127 and @xmath128 , which is consistent with the constraints of colliders and electroweak precision measurements .
no direct excess in the @xmath4 channel comes out of our model .
but the the atlas observed excess in the @xmath4 channel can be interpreted as the misidentification of the @xmath129-tagged jet owing to uncertainties of the tagging slections .
we investigated the prospects of the stealth doublet model as a possible explanation to the diboson excesses observed by the atlas collaboration .
the mass of heavy higgs boson was fixed at near @xmath12 in our study .
we showed that excesses in the @xmath0 and @xmath1 channels can be interpreted as the decay of the heavy cp - even higgs boson @xmath23 , which can be produced at the lhc via its yukawa interaction with the first generation quarks .
one needs the yukawa coupling @xmath15 and the mixing angle between two cp - even higgs bosons @xmath16 , which is consistent with precision measurements , so as to has a @xmath130 production cross section at the lhc .
constraints on the model from the exclusion limits in @xmath10 and @xmath131 channels given by cms collaboration and dijet searches was also studied , which showed the limited parameter space ( in fig .
[ fig : crosss ] ) that can be accommodated with the interpretation of the atlas diboson excesses in the same model .
we expect the running of the 13 tev lhc to tell us the detail about the diboson excesses and show us more clear hints of new physics behind this phenomena .
h. s. fukano , m. kurachi , s. matsuzaki , k. terashi and k. yamawaki , arxiv:1506.03751 [ hep - ph ] .
d. b. franzosi , m. t. frandsen and f. sannino , arxiv:1506.04392 [ hep - ph ] .
k. cheung , w. y. keung , p. y. tseng and t. c. yuan , arxiv:1506.06064 [ hep - ph ] . b. a. dobrescu and z. liu , arxiv:1506.06736 [ hep - ph ] .
j. a. aguilar - saavedra , arxiv:1506.06739 [ hep - ph ] .
y. gao , t. ghosh , k. sinha and j. h. yu , arxiv:1506.07511 [ hep - ph ] .
a. thamm , r. torre and a. wulzer , arxiv:1506.08688 [ hep - ph ] .
j. brehmer , j. hewett , j. kopp , t. rizzo and j. tattersall , arxiv:1507.00013 [ hep - ph ] . q. h. cao , b. yan and d. m. zhang , arxiv:1507.00268 [ hep - ph ] .
g. cacciapaglia and m. t. frandsen , arxiv:1507.00900 [ hep - ph ] .
t. abe , r. nagai , s. okawa and m. tanabashi , arxiv:1507.01185 [ hep - ph ] . b. c. allanach , b. gripaios and d. sutherland , arxiv:1507.01638 [ hep - ph ] .
t. abe , t. kitahara and m. m. nojiri , arxiv:1507.01681 [ hep - ph ] .
r. enberg , j. rathsman and g. wouda , phys .
d * 91 * , no .
9 , 095002 ( 2015 ) [ arxiv:1311.4367 [ hep - ph ] ] .
r. enberg , j. rathsman and g. wouda , jhep * 1308 * , 079 ( 2013 ) [ jhep * 1501 * , 087 ( 2015 ) ] [ arxiv:1304.1714 [ hep - ph ] ] . s. profumo , m. j. ramsey - musolf , c. l. wainwright and p. winslow , phys .
d * 91 * , no . 3 , 035018 ( 2015 ) [ arxiv:1407.5342 [ hep - ph ] ] . | the atlas collaboration has reported excesses in diboson invariant mass searches of new resonances around 2 tev , which might be a prediction of new physics around that mass range .
we interpret these results in the context of a modified stealth doublet model where the extra higgs doublet has a yukawa interaction with the first generation quarks , and show that the heavy cp - even higgs boson can naturally explain the excesses in the @xmath0 and @xmath1 channels with a small yukawa coupling , @xmath2 , and a tiny mixing angle with the sm higgs boson , @xmath3 .
furthermore , the model satisfy constraints from colliders and electroweak precision measurements . |
the quantum yang - baxter equation is an equation in the field of mathematical physics and it lies in the foundation of the theory of quantum groups . let @xmath1 be a linear operator , where @xmath2 is a vector space .
the quantum yang - baxter equation is the equality @xmath3 of linear transformations on @xmath4 , where @xmath5 means @xmath6 acting on the @xmath7th and @xmath8th components . +
a set - theoretical solution of this equation is a solution for which @xmath2 is a vector space spanned by a set @xmath9 and @xmath6 is the linear operator induced by a mapping @xmath10 .
the study of these was suggested by drinfeld @xcite .
etingof , soloviev and schedler study set - theoretical solutions @xmath11 of the quantum yang - baxter equation that are non - degenerate , involutive and braided @xcite . to each such solution , they associate a group called the structure group and they show that this group satisfies some properties .
they also give a classification of such solutions @xmath11 up to isomorphism , when the cardinality of @xmath9 is up to eight .
as an example , there are 23 solutions for @xmath9 with four elements , 595 solutions for @xmath9 with six elements and 34528 solutions for @xmath9 with eight elements . in this paper
, we establish a one - to - one correspondence between non - degenerate , involutive and braided set - theoretical solutions of the quantum yang - baxter equation ( up to isomorphism ) and garside presentations which satisfy some additional conditions up to t - isomorphism ( a notion that will be defined below ) .
the main result is as follows .
@xmath12 let @xmath9 be a finite set , and @xmath11 be a non - degenerate , involutive and braided set - theoretical solution of the quantum yang - baxter equation .
then the structure group of @xmath11 is a garside group .
+ @xmath13 conversely , assume that @xmath14 is a garside monoid such that : + - the cardinality of @xmath6 is @xmath15 , where @xmath16 is the cardinality of @xmath9 and each side of a relation in @xmath6 has length 2 and + - if the word @xmath17 appears in @xmath6 , then it appears only once . + then there exists a function @xmath18 such that @xmath11 is a non - degenerate , involutive and braided set - theoretical solution and @xmath19 is its structure group .
the main idea of the proof is to express the right and left complement on the generators in terms of the functions that define @xmath11 .
moreover , we prove that the structure group of a set - theoretical solution satisfies some specific constraints .
picantin defines the notion of @xmath0-pure garside group in @xcite and he shows that the center of a @xmath0-pure garside group is a cyclic subgroup that is generated by some exponent of its garside element .
let @xmath9 be a finite set , and @xmath11 be a non - degenerate , involutive and braided set - theoretical solution of the quantum yang - baxter equation .
let @xmath20 be the structure group of @xmath11 and @xmath21 the monoid with the same presentation .
then + @xmath12 the right least common multiple of the elements in @xmath9 is a garside element in @xmath21 .
+ @xmath13 the ( co)homological dimension of @xmath20 is equal to the cardinality of @xmath9 .
+ @xmath22 the group @xmath20 is @xmath0-pure garside if and only if @xmath11 is indecomposable .
point _ @xmath12 _ above means that @xmath20 is garside in the restricted sense of @xcite .
let us observe that , independently , gateva - ivanova and van den bergh define in @xcite monoids and groups of left and right i - type and they show that they yield solutions to the quantum yang - baxter equation .
they show also that a monoid of left i - type is cancellative and has a group of fractions that is torsion - free and abelian - by - finite .
jespers and okninski extend their results in @xcite , and establish a correspondence between groups of i - type and the structure group of a non - degenerate , involutive and braided set - theoretical solution . using our result , this makes a correspondence between groups of i - type and the class of garside groups studied in this paper .
they also remark that the defining presentation of a monoid of i - type satisfies the right cube condition , as defined by dehornoy in ( * ? ? ?
* prop.4.4 ) .
so , the necessity of being garside can be derived from the combination of the results from @xcite .
our methods in this paper are different as we use the tools of reversing and complement developed in the theory of garside monoids and groups and our techniques of proof are uniform throughout the paper .
it can be observed that our results imply some earlier results by gateva - ivanova .
indeed , she shows in @xcite that the monoid corresponding to a special case of non - degenerate , involutive and braided set - theoretical solution ( square - free ) has a structure of distributive lattice with respect to left and right divisibility and that the left least common multiple of the generators is equal to their right least common multiple and she calls this element the principal monomial . + the paper is organized as follows . in section @xmath23 , we give some preliminaries on garside monoids . in section @xmath24 ,
we give the definition of the structure group of a non - degenerate , involutive and braided set - theoretical solution and we show that it is garside , using the criteria developed by dehornoy in @xcite .
this implies that this group is torsion - free from @xcite and biautomatic from @xcite . in section @xmath25
, we show that the right least common multiple of the generators is a garside element and that the ( co)homological dimension of the structure group of a non - degenerate , involutive and braided set - theoretical solution is equal to the cardinality of @xmath9 . in section @xmath26
, we give the definition of a @xmath0-pure garside group and we show that the structure group of @xmath11 is @xmath0-pure garside if and only if @xmath11 is indecomposable . in section @xmath27 , we establish a converse to the results of section @xmath24 , namely that a garside monoid satisfying some additional conditions defines a non - degenerate , involutive and braided set - theoretical solution of the quantum yang - baxter equation .
finally , in section @xmath28 , we address the case of non - involutive solutions .
there , we consider the special case of permutation solutions that are not involutive and we show that their structure group is garside .
we could not extend this result to general solutions , although we conjecture this should be true . at the end of the section
, we give the form of a garside element in the case of permutation solutions .
+ this work is a part of the author s phd research , done at the technion under the supervision of professor arye juhasz .
i am very grateful to professor arye juhasz , for his patience , his encouragement and his many helpful remarks .
i am also grateful to professor patrick dehornoy for his comments on this paper .
all the definitions and results in this section are from @xcite and @xcite . in this paper , if the element @xmath29 of @xmath21 is in the equivalence class of the word @xmath30 , we say that _
@xmath30 represents @xmath29_. let @xmath21 be a monoid and let @xmath31 be elements in @xmath21 .
the element @xmath29 is _ a left divisor _ of @xmath32 if there is an element @xmath33 such that @xmath34 in @xmath21 and @xmath32 is _ a right least common multiple ( right lcm ) _ of @xmath29 and @xmath35 if @xmath29 and @xmath35 are left divisors of @xmath32 and additionally if there is an element @xmath30 such that @xmath29 and @xmath35 are left divisors of @xmath30 , then @xmath32 is left divisor of @xmath30 .
we denote it by @xmath36 . _
the complement at right of @xmath35 on @xmath29 _ is defined to be an element @xmath37 such that @xmath38 , whenever @xmath39 exists .
we denote it by @xmath40 and by definition , @xmath41 .
dehornoy shows that if @xmath21 is left cancellative and @xmath42 is the unique invertible element in @xmath21 , then the right lcm and the right complement of two elements are unique , whenever they exist @xcite .
we refer the reader to @xcite for the definitions of the left lcm and the left and right gcd of two elements .
an element @xmath29 in @xmath21 is _ an atom _ if @xmath43 and @xmath44 implies @xmath45 or @xmath46 .
the _ norm _ @xmath47 of @xmath29 is defined to be the supremum of the lengths of the decompositions of @xmath29 as a product of atoms .
the monoid @xmath21 is _ atomic _ if @xmath21 is generated by its atoms and for every @xmath29 in @xmath21 the norm of @xmath29 is finite .
it holds that if all the relations in @xmath21 are length preserving , then @xmath21 is atomic , since each element @xmath29 of @xmath21 has a finite norm as all the words which represent @xmath29 have the same length .
a monoid @xmath21 is _ gaussian _ if @xmath21 is atomic , left and right cancellative , and if any two elements in @xmath21 have a left and right gcd and lcm .
if @xmath0 is an element in @xmath21 , then @xmath0 is a _ garside element _ if the left divisors of @xmath0 are the same as the right divisors , there is a finite number of them and they generate @xmath21 .
a monoid @xmath21 is _ garside _ if @xmath21 is gaussian and it contains a garside element .
a group @xmath20 is a _ gaussian group _
( respectively a _ garside group _ ) if there exists a gaussian monoid @xmath21 ( respectively a garside monoid ) such that @xmath20 is the fraction group of @xmath21 .
a gaussian monoid satisfies both left and right ore s conditions , so it embeds in its group of fractions ( see @xcite ) . as an example , braid groups and artin groups of finite type @xcite , torus knot groups @xcite are garside groups .
* defn.1.6)[def_conditions ] let @xmath21 be a monoid .
@xmath21 satisfies : + - @xmath48 if @xmath42 is the unique invertible element in @xmath21 .
+ - @xmath49 if @xmath21 is left cancellative .
+ - @xmath50 if @xmath21 is right cancellative .
+ - @xmath51 if any two elements in @xmath21 with a right common multiple admit a right lcm .
+ - @xmath52 if @xmath21 has a finite generating set @xmath53 closed under @xmath54 , i.e if @xmath55 then @xmath56 .
2.1 ) [ gars_critere ] a monoid @xmath21 is a garside monoid if and only if @xmath21 satisfies the conditions @xmath48 , @xmath49 , @xmath50 , @xmath51 , and @xmath52 .
let @xmath9 be an alphabet and denote by @xmath57 the empty word in @xmath58 .
let @xmath59 be a partial function of @xmath60 into @xmath58 , @xmath59 is a _ complement _ on @xmath9 if @xmath61 holds for every @xmath29 in @xmath9 , and @xmath62 exists whenever @xmath63 does .
the congruence on @xmath58 generated by the pairs @xmath64 with @xmath65 in the domain of @xmath59 is denoted by `` @xmath66 '' .
the monoid _ associated with @xmath59 _ is @xmath67 or in other words the monoid @xmath68 ( with @xmath65 in the domain of @xmath59 ) .
the complement mapping considered so far is defined on letters only .
its extension on words is called _ word reversing _ ( see @xcite ) .
[ example_struct_gp ] let @xmath21 be the monoid generated by @xmath69 and defined by the following @xmath70 relations .
+ @xmath71 + then the complement @xmath59 is defined totally on @xmath72 and the monoid associated to @xmath59 , @xmath73 , is @xmath21 . as an example , @xmath74 and @xmath75
are obtained from the relation @xmath76 , since it holds that @xmath77 .
let @xmath59 be a complement on @xmath9 . for @xmath78
, @xmath59 is _ coherent _ at @xmath79 if either @xmath80 holds , or neither of the words @xmath81 exists .
the complement _
@xmath59 is coherent _ if it is coherent at every triple @xmath79 with @xmath78 .
dehornoy shows that if the monoid is atomic then it is enough to show the coherence of @xmath59 on its set of atoms .
moreover , he shows that if @xmath21 is a monoid associated with a coherent complement , then @xmath21 satisfies @xmath82 and @xmath83 ( see @xcite ) .
* prop.6.1)[atomic_coh ] let @xmath21 be a monoid associated with a complement @xmath59 and assume that @xmath21 is atomic .
then @xmath59 is coherent if and only if @xmath59 is coherent on @xmath9 . in example
[ example_struct_gp ] , we check if @xmath84 holds in @xmath21 .
we have @xmath85 and @xmath86 , so @xmath87 .
additionally , @xmath88 and @xmath89 , so @xmath90 .
at last , @xmath91 .
all the definitions and results in this subsection are from @xcite .
a set - theoretical solution of the quantum yang - baxter equation is a pair @xmath11 , where @xmath9 is a non - empty set and @xmath92 is a bijection .
let @xmath93 and @xmath94 denote the components of @xmath95 , that is @xmath96 .
a pair @xmath11 is _ nondegenerate _ if the maps @xmath97 defined by @xmath98 and @xmath99 are bijections for any fixed @xmath100 .
a pair @xmath11 is _ braided _
if @xmath95 satisfies the braid relation @xmath101 , where the map @xmath102 is defined by @xmath103 , @xmath104 .
a pair @xmath11 is _ involutive _ if @xmath105 , that is @xmath106 for all @xmath107 .
+ let @xmath108 be the permutation map , that is @xmath109 , and let @xmath110 .
the map @xmath6 is called the _
@xmath111matrix corresponding to @xmath95_. etingof , soloviev and schedler show in @xcite , that @xmath11 is a braided set if and only if @xmath6 satisfies the quantum yang - baxter equation @xmath3 , where @xmath5 means acting on the @xmath7th and @xmath8th components and that @xmath11 is a symmetric set if and only if in addition @xmath6 satisfies the unitary condition @xmath112 .
they define the _ structure group @xmath20 of @xmath11 _ to be the group generated by the elements of @xmath9 and with defining relations @xmath113 when @xmath114 .
they show that if @xmath11 is non - degenerate and braided then the assignment @xmath115 is a right action of @xmath20 on @xmath9 .
+ we use the notation of @xcite , that is if @xmath9 is a finite set , then @xmath95 is defined by @xmath116 , @xmath117 in @xmath9 . here , if @xmath118 is a finite set and @xmath119 for some @xmath120 , then we write @xmath121 instead of @xmath122 and @xmath123 .
the following claim from @xcite translates the properties of a solution @xmath11 in terms of the functions @xmath121 and it will be very useful in this paper .
[ debut_form ] ( i ) @xmath95 is non - degenerate @xmath124 @xmath125 , @xmath126 are bijective , @xmath127 .
+ @xmath13 @xmath95 is involutive @xmath124 @xmath128 and @xmath129 , @xmath130 .
+ @xmath22 @xmath95 is braided @xmath124 @xmath131 , @xmath132 , + and @xmath133 , @xmath134 .
let@xmath69 and @xmath123 .
assume + @xmath135 + assume also that the functions @xmath136 and @xmath137 are the identity on @xmath9 .
then a case by case analysis shows that @xmath11 is a non - degenerate , involutive and braided solution .
its structure group is generated by @xmath9 and defined by the @xmath70 relations described in example [ example_struct_gp ] . in this subsection , we prove the following result .
[ theo : garside ] the structure group @xmath20 of a non - degenerate , braided and involutive set - theoretical solution of the quantum yang - baxter equation is a garside group . in order to prove that the group @xmath20 is a garside group ,
we show that _ the monoid @xmath21 with the same presentation _ is a garside monoid . for that
, we use the garsidity criterion given in theorem [ gars_critere ] , that is we show that @xmath21 satisfies the conditions @xmath48 , @xmath49 , @xmath51 , @xmath52 , and @xmath50 .
we refer the reader to ( * ? ? ?
* lemma 4.1 ) for the proof that @xmath21 is right cancellative ( @xmath21 satisfies @xmath50 ) .
we first show that @xmath21 satisfies the conditions @xmath48 . in order to do that
, we describe the defining relations in @xmath21 and as they are length - preserving , this implies that @xmath21 is atomic .
[ cl_compl ] assume @xmath11 is non - degenerate .
let @xmath138 and @xmath139 be different elements in @xmath9 .
then there is exactly one defining relation @xmath140 , where @xmath141 are in @xmath9 . if in addition
, @xmath11 is involutive then @xmath142 and @xmath143 are different . for a proof of this result , see ( * ? ? ?
using the same arguments , if @xmath11 is non - degenerate and involutive there are no relations of the form @xmath144 , where @xmath145 .
we have the following direct result from claim [ cl_compl ] .
[ complement_f_defined ] assume @xmath11 is non - degenerate and involutive .
then the complement @xmath59 is totally defined on @xmath72 , its range is @xmath9 and the monoid associated to @xmath59 is @xmath21 .
moreover , @xmath21 is atomic .
now , we show that @xmath21 satisfies the conditions @xmath49 , @xmath51 and @xmath52 . from proposition [ complement_f_defined ] , we have that there is a one - to - one correspondence between the complement @xmath59 and the monoid @xmath21 with the same presentation as the structure group , so we say that @xmath21 is coherent ( by abuse of notation ) . in order to show that the monoid @xmath21 satisfies the conditions @xmath49 and @xmath51
, we show that @xmath21 is coherent ( see @xcite ) .
since @xmath21 is atomic , it is enough to check its coherence on @xmath9 ( from proposition [ atomic_coh ] ) .
we show that any triple of generators @xmath146 satisfies the following equation : @xmath147 , where the equality is in the free monoid @xmath58 , since the range of @xmath59 is @xmath9 . in the following lemma
, we establish a correspondence between the right complement of generators and the functions @xmath126 that define @xmath11 .
[ form_compl ] assume @xmath11 is non - degenerate .
let @xmath148 be different elements in @xmath9
. then @xmath149 . if @xmath150 , then @xmath151 .
but by definition of @xmath95 , we have that @xmath152 , so @xmath153 which gives @xmath154 .
[ formule ] assume @xmath11 is non - degenerate and involutive .
let @xmath155 be elements in @xmath9 . then @xmath156 since @xmath95 is involutive , we have from claim [ debut_form ] that for every @xmath157 , @xmath128 .
we replace in this formula @xmath8 by @xmath158 for some @xmath159 , then we obtain @xmath160 .
so , we have @xmath156 .
[ m_c1&c2 ] assume @xmath11 is non - degenerate , involutive and braided .
every triple @xmath161 of generators satisfies the following equation : @xmath162 .
furthermore , @xmath21 is coherent and satisfies the conditions @xmath49 and @xmath51 . if @xmath163 or @xmath164 or @xmath165 , then the equality holds trivially .
so , assume that @xmath161 is a triple of different generators .
this implies that @xmath166 and @xmath167 , since the functions @xmath126 are bijective . using the formulas for all different @xmath168 from lemma [ form_compl ]
, we have : @xmath169 and @xmath170 .
we prove that @xmath171 by showing that @xmath172 for all @xmath173 . since @xmath95 is braided , we have from claim [ debut_form ] , that @xmath174 @xmath175 .
but , from lemma [ formule ] , @xmath176 , so @xmath177 . the monoid @xmath21 is then coherent at @xmath9 but since @xmath21 is atomic , @xmath21 is coherent .
so , @xmath21 satisfies the conditions @xmath49 and @xmath51 .
now , using the fact that @xmath21 satisfies @xmath49 and @xmath51 , we show that it satisfies also @xmath52 .
assume @xmath11 is non - degenerate , involutive and braided .
then there is a finite generating set that is closed under @xmath54 , that is @xmath21 satisfies the condition @xmath52 . from claim [ cl_compl ] , for any pair of generators @xmath178
there are unique @xmath179 such that @xmath140 , that is any pair of generators @xmath178 has a right common multiple .
since from proposition [ m_c1&c2 ] , @xmath21 satisfies the condition @xmath51 , we have that @xmath138 and @xmath139 have a right lcm and the word @xmath180 ( or @xmath181 ) represents the element @xmath182 , since this is a common multiple of @xmath138 and @xmath139 of least length .
so , it holds that @xmath183 and @xmath184 , where @xmath179 .
so , @xmath185 is closed under @xmath54 .
the braid groups and the artin groups of finite type are garside groups which satisfy the condition that the right lcm of their set of atoms is a garside element .
dehornoy and paris considered this additional condition as a part of the definition of garside groups in @xcite and in @xcite it was removed from the definition .
indeed , dehornoy gives the example of a monoid that is garside and yet the right lcm of its atoms is not a garside element @xcite .
he shows that the right lcm of the simple elements of a garside monoid is a garside element , where an element is _ simple _ if it belongs to the closure of a set of atoms under right complement and right lcm @xcite .
we prove the following result : [ thm_delta_lcm_atoms ] let @xmath20 be the structure group of a non - degenerate , involutive and braided solution @xmath11 and let @xmath21 be the monoid with the same presentation .
then the right lcm of the atoms ( the elements of @xmath9 ) is a garside element in @xmath21 . in order to prove that , we show that the set of simple elements @xmath186 , that is the closure of @xmath9 under right complement and right lcm , is equal to the closure of @xmath9 under right lcm ( denoted by @xmath187 ) , where the empty word @xmath57 is added .
so , this implies that @xmath0 , the right lcm of the simple elements , is the right lcm of the elements in @xmath9 .
we use the word reversing method developed by dehornoy and the diagrams for word reversing .
we illustrate in example @xmath42 below the definition of the diagram and we refer the reader to @xcite and @xcite for more details . here , reversing the word @xmath188 using the diagram amounts to computing a right lcm for the elements represented by @xmath189 and @xmath190 ( see @xcite ) .
let us consider the monoid @xmath21 defined in example [ example_struct_gp ] .
we illustrate the construction of the reversing diagram .
+ ( a ) the reversing diagram of the word @xmath191 is constructed in figure [ des1 ] in the following way .
first , we begin with the left diagram and then using the defining relation @xmath192 in @xmath21 , we complete it in the right diagram .
we have @xmath86 and @xmath193 .
, title="fig : " ] , title="fig : " ] \(b ) the reversing diagram of the word @xmath194 is described in figure [ des2 ] : we begin with the left diagram and then we complete it using the defining relations in the right diagram . , title="fig : " ] , title="fig : " ] so , we have @xmath195 in @xmath21 and since @xmath196 and @xmath197 , it holds that @xmath198 in @xmath21 .
so , a word representing the right lcm of @xmath199 and @xmath200 is the word @xmath201 or the word @xmath202 .
we obtain from the diagram that the word @xmath203 represents the element @xmath204 and the word @xmath205 represents the element @xmath206 . in order to prove that @xmath207
, we show that every complement of simple elements is the right lcm of some generators .
the following technical lemma is the basis of induction for the proof of theorem [ thm_delta_lcm_atoms ] .
[ lem_mxin x ] @xmath12 it holds that @xmath208 .
+ @xmath13 it holds that @xmath209 , where @xmath210 , @xmath211 .
it holds that @xmath212 , so @xmath213 and this implies inductively _
@xmath214 ( see the reversing diagram ) .
+ let @xmath215 , then using the following rule of calculation on complements from ( * ? ? ?
* lemma 1.7 ) : for every @xmath216 , @xmath217 , we have inductively that @xmath218 . from _ @xmath12 _ , @xmath219 belongs to @xmath9 , so @xmath220 is in @xmath221 .
that is , @xmath13 holds .
since the monoid @xmath21 is garside , the set of simples @xmath186 is finite and its construction is done in a finite number of steps in the following way : + at the @xmath222th step , @xmath223 .
+ at the first step ,
@xmath224 @xmath225 @xmath226 . + at the second step ,
@xmath227 @xmath228 @xmath229 .
+ we go on inductively and after a finite number of steps @xmath230 , @xmath231 .
[ prop_simples ] it holds that @xmath207 .
the proof is by induction on the number of steps @xmath230 in the construction of @xmath186 .
we show that each complement of simple elements is the right lcm of some generators . at the first step , we have that @xmath232 for all @xmath233 . at the following steps
, we do not consider the complements of the form @xmath234 since these belong to @xmath9 ( see lemma [ lem_mxin x ] ) . at the second step , the complements have the following form @xmath235 or @xmath236 and these belong to @xmath221 from lemma [ lem_mxin x ] .
assume that at the @xmath237th step , all the complements obtained belong to @xmath221 , that is all the elements of @xmath238 are right lcm of generators .
at the @xmath239th step , the complements have the following form @xmath240 , where @xmath241 . from the induction assumption ,
@xmath190 is a right lcm of generators , so from lemma [ lem_mxin x ] , @xmath240 belongs to @xmath221 .
the right lcm of the set of simples @xmath186 is a garside element and since from proposition [ prop_simples ] , @xmath207 , we have that the right lcm of @xmath9 is a garside element .
we show now that the length of a garside element @xmath0 is equal to @xmath16 , the cardinality of the set @xmath9 . in order to show that , we prove in the following that the right lcm of @xmath230 different generators has length @xmath230 .
[ rem_interpret_compl_g ] when @xmath242 is not equal to the empty word , then we can interpret @xmath242 in terms of the functions @xmath243 using the reversing diagram corresponding to the words @xmath244 and @xmath29 , where @xmath245 and for brevity of notation we write @xmath246 for @xmath247 : + that is , @xmath248 and this is equal to @xmath249 , since the action on @xmath9 is a right action . having a glance at the reversing diagram
, we remark that if @xmath242 is not equal to the empty word , then none of the complements @xmath250 , @xmath251 , .. , @xmath252 can be equal to the empty word .
[ lem_compl_egal ]
let @xmath253 be all different elements in @xmath9 , for @xmath254 .
then @xmath255 is not equal to the empty word @xmath57 .
by induction on @xmath230 .
if @xmath256 , then @xmath257 , as @xmath258 .
+ now assume @xmath259 and assume by contradiction that @xmath260 . using the following rule of computation on the complement from ( * ? ? ?
* lemma 1.7 ) : @xmath261 , we have @xmath262 . from the induction assumption , @xmath259 and @xmath263 , so @xmath260 implies that @xmath264 .
assume @xmath30 represents the element @xmath265 , then from remark [ rem_interpret_compl_g ] , @xmath266 can be interpreted as @xmath267 and @xmath268 can be interpreted as @xmath249 .
the function @xmath269 is a bijective function as it is the composition of bijective functions , so @xmath270 implies that @xmath271 , but this is a contradiction .
so , @xmath272 .
[ thm_garside_length_n ] let @xmath20 be the structure group of a non - degenerate , braided and involutive solution @xmath11 , where @xmath273 and let @xmath21 be the monoid with the same presentation .
let @xmath0 be a garside element in @xmath21 .
then the length of @xmath0 is @xmath16 . from theorem [ thm_delta_lcm_atoms ]
, @xmath0 represents the right lcm of the elements in @xmath9 , that is @xmath274 in @xmath21 .
we show by induction that a word representing the right lcm @xmath275 has length @xmath230 , where @xmath276 for @xmath277 . if @xmath278 , then there are different generators @xmath141 such that @xmath279 , so @xmath280 is a relation in @xmath21 and the right lcm of @xmath281 has length @xmath23 .
assume that the right lcm @xmath282 has length @xmath283 .
then the right lcm @xmath284 is obtained from the reversing diagram corresponding to the words @xmath282 and @xmath285 . from lemma [ lem_compl_egal ] ,
@xmath286 is not equal to the empty word , so from lemma [ lem_mxin x ] it has length @xmath42 .
so , the right lcm @xmath275 has length @xmath230 and this implies that @xmath287 has length @xmath16 .
dehornoy and laffont construct a resolution of @xmath288 ( as trivial @xmath289-module ) by free @xmath289-modules , when @xmath21 satisfies some conditions @xcite .
moreover , they show that every garside group is of type @xmath290 , that is with a finite resolution ( * ? ? ?
2.9 - 2.10 ) .
charney , meier , and whittlesey show in @xcite that garside groups have finite homological dimension , using another approach . in @xcite ,
dehornoy and laffont show that whenever @xmath21 is a garside monoid then the resolution defined in @xcite is isomorphic to the resolution they define .
we use the following result from @xcite in order to show that the homological dimension of the structure group corresponding to a set - theoretical solution @xmath11 of the quantum yang - baxter equation is equal to the number of generators in @xmath9 .
* cor.3.6)[limiter_dimension ] assume that @xmath21 is a locally gaussian monoid admitting a generating set @xmath186 such that @xmath291 is closed under left and right complement and lcm and such that the norm of every element in @xmath186 is bounded above by @xmath16 .
then the ( co)homological dimension of @xmath21 is at most @xmath16 . using proposition [ limiter_dimension ] , we prove the following result : let @xmath11 be a set - theoretical solution of the quantum yang - baxter equation , where @xmath273 and @xmath11 is non - degenerate , braided and involutive .
let @xmath20 be the structure group corresponding to @xmath11 .
then the ( co)homological dimension of @xmath20 is equal to @xmath16 , the number of generators in @xmath9 .
the set of simples @xmath186 satisfies the conditions of corollary [ limiter_dimension ] and the norm of every element in @xmath186 is bounded by @xmath16 , since this is the length of the right lcm of @xmath186 ( from theorems [ thm_delta_lcm_atoms ] and [ thm_garside_length_n ] ) .
so , the ( co)homological dimension of @xmath20 is is equal to @xmath16 .
it was pointed to us by p.etingof that this result can be also proved differently : by showing that the classifying space of the structure group @xmath20 is a compact manifold of dimension @xmath16 ( as there is a free action of @xmath20 on @xmath292 ) .
picantin defines the notion of @xmath0-pure garside monoid @xmath21 in @xcite and he shows there that the center of @xmath21 is the infinite cyclic submonoid generated by some power of @xmath0 .
we find in this section conditions under which a monoid is @xmath0-pure garside in terms of set - theoretical solutions .
let @xmath186 be the set of simples and @xmath0 a garside element in @xmath21 .
the _ exponent _ of @xmath21 is the order of the automorphism @xmath293 , where @xmath293 is the extension of the function @xmath294 from @xmath186 into itself .
@xcite the monoid @xmath21 is @xmath0`-`_pure _ if for every @xmath117 in @xmath9 , it holds that @xmath295 , where @xmath296 and @xmath297 denotes the right lcm .
picantin shows that if @xmath21 is a @xmath0-pure garside monoid with exponent @xmath298 and group of fractions @xmath20 , then the center of @xmath21 ( resp . of @xmath20 )
is the infinite cyclic submonoid ( resp .
subgroup ) generated by @xmath299 .
let consider the following example , to illustrate these definitions .
[ example_deltapure ] let @xmath300 and let @xmath301 , where @xmath302 , be a non - degenerate , braided and involutive set - theoretical solution .
let @xmath21 be the monoid with the same presentation as the structure group of @xmath11 , the defining relations in @xmath21 are then : @xmath303 , @xmath304 and @xmath305 .
so , @xmath306 , @xmath307 and @xmath308 .
using the reversing diagram , we obtain inductively that @xmath309 for @xmath310 , that is @xmath21 is @xmath0-pure garside , since @xmath311 . as an example , @xmath312 , so @xmath313 and so @xmath314 that is @xmath315 and since @xmath316 ( see lemma [ lem_mxin x ] ) we have the equality .
each word @xmath317 for @xmath318 represents a garside element , denoted by @xmath0 .
the set of simples is @xmath319 .
the exponent of @xmath21 is equal to 1 , since the function @xmath294 from @xmath186 to itself is the identity . as an example
, the image of @xmath320 is @xmath321 , the image of @xmath203 is @xmath322 and so on .
so , the center of the structure group of @xmath11 is cyclic and generated by @xmath0 , using the result of picantin .
a non - degenerate , braided and involutive set - theoretical solution @xmath11 is said to be _ decomposable _ if @xmath9 is a union of two nonempty disjoint non - degenerate invariant subsets , where an _ invariant _ subset @xmath323 is a set that satisfies @xmath324 .
otherwise , @xmath11 is said to be _
indecomposable_. + etingof et al give a classification of non - degenerate , braided and involutive solutions with @xmath9 up to @xmath325 elements , considering their decomposability and other properties @xcite .
rump proves gateva - ivanova s conjecture that every square - free , non - degenerate , involutive and braided solution @xmath11 is decomposable , whenever @xmath9 is finite .
moreover , he shows that an extension to infinite @xmath9 is false @xcite .
we find a criterion for decomposability of the solution involving the garside structure of the structure group ( monoid ) , that is we prove the following result .
[ thm_deltapure_indecomp ] let @xmath20 be the structure group of a non - degenerate , braided and involutive solution @xmath11 and let @xmath21 be the monoid with the same presentation .
then @xmath21 is @xmath0-pure garside if and only if @xmath11 is indecomposable . in what follows ,
we use the notation from @xcite : for @xmath326 , @xmath327 denotes the set of elements @xmath328 for @xmath329 and @xmath330 .
we write @xmath331 for @xmath332 and @xmath333 for @xmath334 .
we need the following lemma for the proof of theorem [ thm_deltapure_indecomp ] [ lem_yyimpliqmy ] let @xmath11 be the union of non - degenerate invariant subsets @xmath323 and @xmath335
. then @xmath336 and @xmath337 .
if @xmath323 is an invariant subset of @xmath9 , then @xmath338 , since @xmath339 . from ( * ? ? ?
* proposition 2.15 ) , the map @xmath95 defines bijections @xmath340 and @xmath341 .
so , @xmath342 and @xmath343 , and this implies @xmath344 .
that is , we have that @xmath338 and @xmath344 , so @xmath345 and this implies inductively that @xmath346 ( see the reversing diagram ) .
the same holds for @xmath335 .
assume @xmath11 is decomposable , that is @xmath11 is the union of non - degenerate invariant subsets @xmath323 and @xmath335 . from lemma [ lem_yyimpliqmy ]
, we have @xmath346 and @xmath337 .
let @xmath347 and @xmath348 , then @xmath349 can not be the same as @xmath350 .
so , @xmath21 is not @xmath0-pure garside . +
assume @xmath21 is not @xmath0-pure garside .
let @xmath351 , we denote by @xmath352 the set @xmath353 from which we remove @xmath354 .
so , @xmath355 , where @xmath352 is a subset of @xmath9 from lemma [ lem_mxin x ] .
let @xmath356 be in @xmath9 , then from @xcite , either @xmath357 or the left gcd of @xmath358 and @xmath359 is @xmath57 . if @xmath357 , then @xmath360 and if the left gcd of @xmath358 and @xmath359 is @xmath57 , then @xmath361 and @xmath362 are disjoint subsets of @xmath9 . since @xmath21 is not @xmath0-pure garside ,
there exist @xmath363 , @xmath364 , .. , @xmath365 in @xmath9 such that @xmath366 , @xmath367 , .. , @xmath368 are disjoint subsets of @xmath9 .
moreover , @xmath369 , since each @xmath370 is equal to an element @xmath371 for some @xmath372 ( from the existence of left lcms ) . we show that @xmath373 is an invariant subset of @xmath9 , that is @xmath374 .
let @xmath370 and @xmath375 , then @xmath376 , for some @xmath377 . using the following rule of computation on the complement from ( * ? ? ?
* lemma 1.7 ) : @xmath378 , we have @xmath379 , that is @xmath380 belongs to @xmath373 .
in particular , if @xmath381 then @xmath382 , where @xmath383 .
so , @xmath373 is an invariant subset of @xmath9 for @xmath384 and this implies that @xmath11 is decomposable .
we establish the converse implication in the one - to - one correspondence between the garside groups and the structure groups , that is we prove the following : [ garside_struct_group ] let @xmath14 be a garside monoid such that : + @xmath12 there are @xmath15 relations in @xmath6 , where @xmath16 is the cardinality of @xmath9 , and each side of a relation in @xmath6 has length 2 .
+ @xmath13 if the word @xmath17 appears in @xmath6 , then it appears only once . +
then there exists a function @xmath18 such that @xmath11 is a non - degenerate , involutive and braided set - theoretical solution and @xmath19 is its structure group .
+ if additionally : @xmath22 there is no word @xmath385 in @xmath6 , then @xmath11 is square - free . in order to prove theorem [ garside_struct_group ] , we use the concepts of left lcm and left coherence from @xcite and @xcite , but we do not use exactly the same notations .
the notation for the _ left lcm of @xmath29 and @xmath35 _ is @xmath386 and for _ the complement at left of @xmath35 on @xmath29 _ the notation is @xmath387 , where @xmath388 .
@xcite let @xmath21 be a monoid .
the _ left coherence condition _ on @xmath21 is satisfied if it holds for any @xmath389 : @xmath390 .
this property is also called the _ left cube condition_. we show that if @xmath11 is a non - degenerate and involutive set - theoretical solution , then @xmath11 is braided if and only if @xmath9 is coherent and left coherent .
the left coherence of @xmath9 is satisfied if the following condition on all @xmath391 in @xmath9 is satisfied : @xmath392 , where the equality is in the free monoid since the complement on the left is totally defined and its range is @xmath9 .
note that as in the proof of the coherence , left coherence on @xmath9 implies left coherence on @xmath21 , since the monoid @xmath21 is atomic .
clearly , the following implication is derived from theorem [ theo : garside ] : [ braided_implies_leftcoherent ] assume @xmath11 is non - degenerate and involutive . if @xmath11 is braided , then @xmath9 is coherent and left coherent .
the proof of the converse implication is less trivial and requires a lot of computations .
before we proceed , we first express the left complement in terms of the functions @xmath125 that define @xmath11 . as the proofs are symmetric to those done in section 3.2 with the right complement we omit them .
[ compl_gauche ] assume @xmath11 is non - degenerate .
let @xmath356 be different elements in @xmath9 . then @xmath393 .
[ formule_gauche ] assume @xmath11 is non - degenerate and involutive .
let @xmath155 be elements in @xmath9 .
then @xmath394 .
[ cor_xcoherent_equations ] assume @xmath11 is non - degenerate .
if @xmath9 is coherent and left coherent , then for every @xmath395 the following equations hold : + ( a ) @xmath396 + ( b ) @xmath397 from lemma [ compl_gauche ] , we have for all different @xmath134 that @xmath398 and @xmath399 . if @xmath9 is left coherent , then for all different @xmath134 , we have @xmath400 @xmath401 @xmath402 .
if @xmath403 , the equality ( a ) holds trivially , so let fix @xmath8 and @xmath230 such that @xmath404 .
we denote @xmath405 @xmath406 and @xmath407 @xmath408 , these functions are bijective , since these are compositions of bijective functions and satisfy @xmath409 whenever @xmath410 .
it remains to show that @xmath411 and @xmath412 .
assume by contradiction that @xmath413 and @xmath414 , so there is @xmath415 such that @xmath416= @xmath417 @xmath418 , that is @xmath419 and @xmath420 .
that is @xmath421 and @xmath422 , since @xmath11 is involutive .
so , @xmath423 and @xmath424 .
since @xmath425 is bijective , this implies that there is @xmath426 such that @xmath427 @xmath428 , that is @xmath429 .
but , since @xmath430 , this contradicts the fact that @xmath11 is involutive .
so , since the functions @xmath431 are bijective , ( * ) is equivalent to ( a ) .
equation ( b ) is obtained in the same way using the coherence of @xmath9 ( see lemma [ form_compl ] ) .
[ coherence_implies_braided ] let @xmath11 be a non - degenerate and involutive set - theoretical solution .
if @xmath9 is coherent and left coherent , then @xmath11 is braided .
we need to show that the functions @xmath125 and @xmath126 satisfy the following equations from lemma [ debut_form ] : + _ @xmath12 _ @xmath132 , @xmath130 . + _ @xmath13 _ @xmath131 , @xmath130 . + _
@xmath432 , @xmath433 .
+ from lemma [ cor_xcoherent_equations ] , we have for @xmath434 that ( a ) @xmath435 .
assume @xmath436 , that is @xmath437 and we replace in formula ( a ) @xmath438 by @xmath416 and @xmath230 by @xmath439 , then we obtain @xmath440 . in order to show that _
@xmath12 _ holds , we show that @xmath441 . from lemma [ formule_gauche ]
, we have @xmath442 for every @xmath443 , so by replacing @xmath444 by @xmath439 , we obtain @xmath441 .
so , _ @xmath12 _ holds .
+ from corollary [ cor_xcoherent_equations ] , we have for @xmath445 that ( b ) @xmath397 .
assume @xmath446 , that is @xmath447 and we replace in formula ( b ) @xmath158 by @xmath416 and @xmath230 by @xmath448 , then we obtain @xmath449 . in order to show that _
@xmath13 _ holds , we show that @xmath450 . from lemma [ formule ]
, we have @xmath451 , so by replacing @xmath444 by @xmath448 , we obtain @xmath450 .
so , _ @xmath13 _ holds .
+ it remains to show that _ @xmath22 _ holds . from _
@xmath12 _ , we have for @xmath130 that @xmath132 and this is equivalent to @xmath452 . we replace @xmath7 by @xmath453 for some @xmath454 in the formula .
we obtain @xmath455 . by applying these functions on @xmath456 on both sides ,
we obtain @xmath457 .
since @xmath11 is involutive , we have @xmath458 ( see lemma [ debut_form ] ) .
so , @xmath459 . from lemma [ formule_gauche ] , we have @xmath460 for every @xmath461 , so replacing @xmath7 by @xmath462 and @xmath230 by @xmath463 gives @xmath464 .
so , @xmath465 . from _ @xmath12 _ , we have that @xmath466 , so @xmath432 , that is _
@xmath22 _ holds .
first , we define a function @xmath18 and @xmath467 functions @xmath121 for @xmath127 , such that @xmath123 in the following way : if there is a relation @xmath468 then we define @xmath469 , @xmath470 and we define @xmath471 , @xmath472 , @xmath473 and @xmath474 . if the word @xmath17 does not appear as a side of a relation , then we define @xmath475 and we define @xmath476 and @xmath477 .
we show that the functions @xmath125 and @xmath126 are well defined for @xmath127 : assume @xmath471 and @xmath478 for some @xmath479 and @xmath480 , then it means from the definition of @xmath95 that @xmath481 and @xmath482 that is the word @xmath17 appears twice in @xmath6 and this contradicts _ @xmath13_.
the same argument holds for the proof that the functions @xmath125 are well defined .
+ we show that the functions @xmath125 and @xmath126 are bijective for @xmath127 : assume @xmath471 and @xmath483 for some @xmath484 and @xmath485 , then from the definition of @xmath95 we have @xmath469 and @xmath486 for some @xmath487 , that is there are the following two defining relations in @xmath6 : @xmath468 and @xmath488 . but
this means that @xmath138 and @xmath489 have two different right lcms and this contradicts the assumption that the monoid is garside .
so , these functions are injective and since @xmath9 is finite they are bijective . assuming @xmath125 not injective yields generators with two different left lcms
so , @xmath95 is well - defined and @xmath490 is non - degenerate and from _
@xmath490 is also involutive .
it remains to show that @xmath490 is braided : since @xmath491 is garside , it is coherent and left coherent so from lemma [ coherence_implies_braided ] , @xmath11 is braided . obviously condition _
@xmath22 _ implies that @xmath11 is also square - free .
it remains to establish the one - to - one correspondence between structure groups of non - degenerate , involutive and braided set - theoretical solutions of the quantum yang - baxter equation and a class of garside groups admitting a certain presentation and in order to that we need the following terminology and claims .
a _ tableau monoid _ is a monoid @xmath14 satisfying the condition that each side of a relation in @xmath6 has length 2 .
we say that two tableau monoids @xmath14 and @xmath492 are _ t - isomorphic _ if there exists a bijection @xmath493 such that @xmath468 is a relation in @xmath6 if and only if @xmath494 is a relation in @xmath495 . clearly , if two tableau monoids are t - isomorphic then they are isomorphic and the definition is enlarged to groups . set - theoretical solutions @xmath11 and @xmath496 are _ isomorphic _ if there exists a bijection @xmath497 which maps @xmath95 to @xmath498 , that is @xmath499 .
let @xmath11 and @xmath496 be non - degenerate , involutive and braided set - theoretical solutions .
assume @xmath11 and @xmath496 are isomorphic
. then their structure groups ( monoids ) @xmath20 and @xmath500 are t - isomorphic tableau groups ( monoids ) .
conversely , if @xmath14 and @xmath501 are t - isomorphic tableau garside monoids each satisfying the conditions @xmath12 and @xmath13 from theorem [ garside_struct_group ] , then the solutions @xmath11 and @xmath496 defined respectively by @xmath14 and @xmath501 are isomorphic . clearly , the structure groups ( monoids ) @xmath20 and @xmath500 are tableau groups ( monoids ) . we need to show that @xmath20 and @xmath500 are t - isomorphic .
since @xmath11 and @xmath496 are isomorphic , there exists a bijection @xmath497 which maps @xmath95 to @xmath498 , that is @xmath502 .
so , since by definition @xmath96 , we have @xmath113 if and only if @xmath503 .
that is , if we take @xmath504 to be equal to @xmath293 we have that @xmath20 and @xmath500 are t - isomorphic .
for the converse , take @xmath293 to be equal to @xmath504 and from the definition of @xmath95 and @xmath498 from their tableau we have @xmath499 , that is @xmath11 and @xmath496 are isomorphic .
in this part , we consider a special case of set - theoretical solutions of the quantum yang - baxter equation , namely the permutation solutions .
these solutions were defined by lyubashenko ( see @xcite ) .
let @xmath9 be a set and let @xmath505 be a mapping .
a permutation solution is a set - theoretical solution of the form @xmath506 , where @xmath507 .
the solution @xmath11 is nondegenerate iff @xmath508 are bijective , @xmath11 is braided iff @xmath509 and @xmath11 is involutive iff @xmath510 . note that these solutions are defined by only two functions , while for a general set - theoretical solution the number of defining functions is twice the cardinality of the set @xmath9 .
let @xmath9 be a finite set and let @xmath505 be defined by @xmath506 , where @xmath507 are bijective and satisfy @xmath509 .
so , @xmath11 is a non - degenerate and braided permutation solution that is not necessarily involutive , as we do not require @xmath510 .
let @xmath20 be the structure group of @xmath11 and let @xmath21 be the monoid with the same presentation .
we define an equivalence relation on the set @xmath9 in the following way : + @xmath511 if and only if there is an integer @xmath230 such that @xmath512 .
we define @xmath513 and we define functions @xmath514 such that @xmath515)=[f(x)]$ ] and @xmath516)=[g(x)]$ ] , where @xmath517 $ ] denotes the equivalence class of @xmath29 modulo @xmath518 .
we then define @xmath519 by @xmath520,[y])= ( g'([y]),f'([x ] ) ) = ( [ g(y)],[f(x)])$ ] .
our aim is to show that @xmath496 is a well - defined non - degenerate , involutive and braided solution ( a permutation solution ) and that its structure group @xmath500 is isomorphic to @xmath20 . before doing this
, we illustrate the main ideas of the proofs to come with an example .
@xmath69 and let @xmath521 and @xmath522 .
then @xmath508 are bijective and satisfy @xmath523 but @xmath524 , so @xmath11 is a non - degenerate and braided ( permutation ) solution , where @xmath506 .
the set of relations @xmath6 is : + @xmath525 + using @xmath518 defined above , we have @xmath526,[x_{2}],[x_{5}]\}$ ] , with @xmath527 and @xmath528 , since in this example it holds that @xmath529 and @xmath530 .
applying the definition of @xmath498 yields @xmath531,[x_{1}])=([g(1)],[f(1)])=([2],[4])=([2],[2])$ ] and so on .
so , @xmath532,[x_{2}],[x_{5 } ] \mid [ x_{1}]^{2}=[x_{2}]^{2},[x_{1}][x_{5}]=[x_{5}][x_{2 } ] , [ x_{2}][x_{5}]=[x_{5}][x_{1}]\rangle$ ] .
note that in @xmath20 , it holds that @xmath533 and @xmath534 since many of the defining relations from @xmath6 involve cancellation and @xmath535 .
so , @xmath20 and @xmath500 have the same presentation , up to a renaming of the generators .
let @xmath541 be in @xmath9 such that @xmath539 .
if @xmath539 then it means that there is an integer @xmath230 such that @xmath542 .
if @xmath230 is odd , then let @xmath35 in @xmath9 be defined in the following way : @xmath543 .
so , from lemma [ lem : calcul_sk ] , @xmath544 .
so , there is a relation @xmath545 in @xmath20 which implies that @xmath546 in @xmath20 .
if @xmath230 is even and @xmath512 , then there is an element @xmath547 such that @xmath548 , where @xmath283 is odd .
so , from the subcase studied above , there is @xmath549 , @xmath550 , such that there is a relation @xmath551 in @xmath20 and this implies that @xmath552 in @xmath20 .
additionally , @xmath553 , so from the same argument as above there is @xmath554 such that there is a relation @xmath555 in @xmath20 and this implies that @xmath556 in @xmath20 .
so , @xmath546 in @xmath20 .
@xmath12 @xmath557 and @xmath558 are well defined , so @xmath498 is well - defined .
+ @xmath13 @xmath557 and @xmath558 are bijective , so @xmath498 is non - degenerate .
+ @xmath22 @xmath557 and @xmath558 satisfy @xmath559 , so @xmath498 is braided .
+ ( iv ) @xmath557 and @xmath558 satisfy @xmath560 , so @xmath498 is involutive . _
@xmath12 _ let @xmath541 be in @xmath9 such that @xmath511 .
we show that @xmath561 ) = f'([x])$ ] , that is @xmath562=[f(x)]$ ] .
since @xmath511 , there is an integer @xmath230 such that @xmath563 , so @xmath561)=[f(x')]=[f(fg)^{k}(x)]=[(fg)^{k}f(x)]=[f(x)]=f'([x])$ ] , using the fact that @xmath509 .
the same proof holds for @xmath564)=g'([x])$ ] . + _
@xmath13 _ assume that there are @xmath107 such that @xmath515)=f'([y])$ ] , that is @xmath565=[f(y)]$ ] . by the definition of @xmath518
, this means that there is an integer @xmath230 such that @xmath566 , that is @xmath567 , since @xmath509 . but
@xmath59 is bijective , so @xmath568 , which means that @xmath517=[y]$ ] . the same proof holds for @xmath558 , using the fact that @xmath569 is bijective .
+ _ @xmath22 _
let @xmath517\in x'$ ] , so @xmath570))=f'([g(x)])=[f(g(x))]$ ] and @xmath571)=g'([f(x)])=[g(f(x))]$ ] .
since @xmath509 , @xmath559
( iv ) _ let @xmath517\in x'$ ] , so from the definition of @xmath518 , we have @xmath572=[x]$ ] , so @xmath573)=[x]$ ] , that is @xmath574 .
[ lem : gisgarside ] let @xmath11 be a not necessarily involutive permutation solution .
let @xmath513 and let @xmath500 be the structure group corresponding to @xmath496 , as defined above .
then @xmath500 is garside .
furthermore , if @xmath468 is a defining relation in @xmath20 , then @xmath575[x_{j}]=[x_{k}][x_{l}]$ ] is a defining relation in @xmath500 .
it holds that @xmath496 is a non - degenerate , braided and involutive permutation solution , so by theorem [ theo : garside ] the group @xmath500 is garside .
assume that @xmath468 is a defining relation in @xmath20 , that is @xmath576 .
from the definition of @xmath498 , @xmath577,[x_{j}])=([g(x_{j})],[f(x_{i})])=([x_{k}],[x_{l}])$ ] , that is there is a defining relation @xmath575[x_{j}]=[x_{k}][x_{l}]$ ] in @xmath500 . note that this relation may be a trivial one
if @xmath575=[x_{k}]$ ] and @xmath578=[x_{l}]$ ] in @xmath500 we show that the group @xmath20 is isomorphic to the group @xmath500 , where @xmath500 is the structure group of @xmath496 and @xmath513 , and from lemma [ lem : gisgarside ] this implies that @xmath20 is a garside group .
let @xmath579 be the quotient map defined by @xmath580 $ ] for all @xmath370 . from lemma [ lem : gisgarside ]
, @xmath581 is an homomorphism of groups , so @xmath582 is an epimorphism .
we need to show that @xmath583 is injective .
we show that if @xmath517[y]=[t][z]$ ] is a non - trivial defining relation in @xmath500 , then @xmath113 is a defining relation in @xmath20 .
if @xmath517[y]=[t][z]$ ] is a non - trivial defining relation in @xmath500 , then since @xmath520,[y])= ( [ g(y)],[f(x)])$ ] , we have that @xmath584=[t]$ ] and @xmath565=[z]$ ] .
that is , there are @xmath585 $ ] and @xmath586 $ ] such that @xmath587 and @xmath588 .
this implies that @xmath589 , that is @xmath590 is a defining relation in @xmath20 .
it holds that @xmath591 and @xmath592 , so from lemma [ equiv_cancel ] , @xmath593 and @xmath594 in @xmath20 .
so , @xmath113 is a defining relation in @xmath20 .
+ note that if @xmath517[y]=[t][z]$ ] is a trivial relation in @xmath500 , that is @xmath517=[t]$ ] and @xmath595=[z]$ ] , then from lemma [ equiv_cancel ] @xmath596 and @xmath597 in @xmath20 and so @xmath113 holds trivially in @xmath20 .
so , @xmath583 is an isomorphism of the groups @xmath20 and @xmath500 and from lemma [ lem : gisgarside ] , @xmath20 is garside . in this subsection ,
we consider the structure group of a non - degenerate , braided and involutive permutation solution .
we claim that given the decomposition of the defining function of the permutation solution as the product of disjoint cycles , one can easily find a garside element in its structure group .
this result can be extended to non - degenerate and braided permutation solutions that are not involutive , using the construction from section @xmath598 .
let @xmath273 , and @xmath11 be a non - degenerate , braided and involutive permutation solution defined by @xmath599 , where @xmath59 is a permutation on @xmath600 .
let @xmath21 be the monoid with the same presentation as the structure group .
assume that @xmath59 can be described as the product of disjoint cycles : + @xmath601 , @xmath602 .
then @xmath12 for @xmath254 , @xmath603 in @xmath21 and this element is denoted by @xmath604 .
+ @xmath13 the element @xmath605 is a garside element in @xmath21 .
r. charney , j. meier and k. whittlesey , bestvina s normal form complex and the homology of garside groups , geometriae dedicata _ 105 _ ( 2004 ) 171 - 188 .
f. chouraqui , decision problems in tableau - groups and tableau - semigroups , phd thesis , technion , israel .
clifford , g.b .
preston , the algebraic theory of semigroups , vol.1 , ams surveys _ 7 _ , ( 1961 ) .
p. dehornoy , groupes de garside , ann .
scient . ec .
_ 35 _ ( 2002 ) 267 - 306 .
p. dehornoy , braids and self - distributivity , progress in math .
192 , birkhauser ( 2000 ) .
p. dehornoy , gaussian groups are torsion free , j. of algebra _ 210 _ ( 1998 ) 291 - 297 .
p. dehornoy , complete positive group presentations , j. of algebra _ 268 _ ( 2003 ) 156 - 197 .
p. dehornoy , y. lafont , homology of gaussian groups , ann .
fourier ( grenoble ) _
53 - 2 _ ( 2003 ) 489 - 540 .
p. dehornoy , l. paris , gaussian groups and garside groups , two generalizations of artin groups , proc .
london math .
_ 79 - 3 _ ( 1999 ) 569 - 604 .
drinfeld , on some unsolved problems in quantum group theory , lec .
notes math .
_ 1510 _ ( 1992 ) 1 - 8 .
p. etingof , t. schedler , a. soloviev , set - theoretical solutions to the quantum yang - baxter equation , duke math . j. _ 100 - 2 _ ( 1999 ) 169 - 209 .
garside , the braid group and other groups , quart .
oxford _ 20 _ no.78 ( 1969 ) 235 - 254 .
t. gateva - ivanova , a combinatorial approach to the set - theoric solutions of the yang - baxter equation , j. math .
_ 45 - 10 _ ( 2005 ) 3828 - 3858 .
t. gateva - ivanova , m. van den bergh , semigroups of @xmath606-type , j. of algebra _ 206 - 1 _ ( 1998 ) 97 - 112 .
e. jespers , j. okninski , monoids and groups of @xmath606-type , algebras and representation theory _ 8 _ ( 2005 ) 709 - 729 .
m. picantin , the center of thin gaussian groups , j. of algebra _ 245 _ ( 2001 ) 92 - 122 .
m. picantin , the conjugacy problem in garside groups , comm . in algebra
_ 29 - 3 _ ( 2001 ) 1021 - 1039 .
m. picantin , automatic structures for torus link groups , j. knot theory and its ramifications _ 12 - 6 _ ( 2003 ) 833 - 866 .
w. rump , a decomposition theorem for square - free unitary solutions of the quantum yang - baxter equation , advances in math . _ 193 _ ( 2005 ) 40 - 55 . | we establish a one - to - one correspondence between a class of garside groups admitting a certain presentation and the structure groups of non - degenerate , involutive and braided set - theoretical solutions of the quantum yang - baxter equation .
we also characterize indecomposable solutions in terms of @xmath0-pure garside groups . |
accurate estimates of mass loss rates from the satellite galaxies and globular clusters in orbit around the milky way offer the possibility of better understanding the dynamical history of the galaxy . when integrated over the entire population
, such measurements would tell us about the current accretion rate onto the galaxy .
this in turn would indicate whether the galaxy has grown significantly through the gradual process of tidal stripping and disruption of its satellites .
the direct inference of mass loss rates from individual objects would place new constraints on detailed analytic or numerical dynamical models of these systems . in addition , the measurements could be used to identify satellites which are likely to have well - populated streams of tidal debris associated with them . in particular , the formation , evolution and ultimate fate of the galactic globular clusters is a long standing puzzle in astrophysics .
dynamically , globular clusters are clean and relatively simple systems , yet on close inspection they reveal a wealth of complex behaviour , including relaxation and evaporation from internal dynamical interactions and mass loss in response to the tidal field of the milky way @xcite .
ideally we would like to be able to assess the importance of each of these processes for the evolution of individual globular clusters from observations . of particular interest for mass - segregated , relaxed clusters
is the differential mass loss of stars due to the different relative densities of low and high mass stars at various radii , and the consequence of this loss for the evolution of the present day mass function ( pdmf ) away from the initial mass function ( imf ) .
understanding the form of the imf of the current globulars is relevant for theories of star formation and globular cluster formation @xcite , and the related issue of how the galaxy was assembled .
the mass function of a stellar system @xmath0 is defined as the number of stars @xmath1 in the mass interval @xmath2 . a common functional form chosen to approximate this distribution is @xmath3
where @xmath4 is called the _ mass function index_. the pdmf @xmath4 of a cluster has been found to be related to the galactocentric radius @xmath5 and height above the galactic plane @xmath6 of the cluster ( capaccioli , ortolani & piotto 1991 ) . using simple semi - analytic models
capaccioli , piotto & stiavelli ( 1993 ) showed that these correlations could be reproduced by evolving a population of globular clusters which all have the same initial imf in the tidal field of the milky way .
similarly , in a comparison of several globular clusters piotto , cool & king ( 1997 ) found the luminosity function of ngc 6397 to be much flatter than m15 , m30 and m92 , and concluded that this could be due to the extreme dynamical evolution implied by its orbit calculated from its proper motion @xcite .
in addition mcclure et al .
( 1986 ) reported a dependence of the pdmf on metallicity , and djorgovski , piotto & capaccioli ( 1993 ) successfully incorporated this together with the trends in @xmath5 and @xmath6 into a trivariate analysis .
in contrast , in a recent study of pal 1 , rosenberg et al .
( 1997 ) found its mass function to be inconsistent with the trends in @xmath5 , @xmath6 , and metallicity and concluded that this could either be due to evaporation of low mass stars , or an intrinsically different imf .
one step towards solving the puzzle of just how far the pdmf of a cluster differs from its imf is determining how to interpret the signatures of tidal effects in observations of globular clusters and other satellites . in the past
, the limiting radii of galactic satellites ( which are assumed to correspond to the tidal radii @xmath7 imposed by the milky way ) have been interpreted using king s ( 1962 ) tidal radius formula @xmath8 where @xmath5 is the distance of the satellite from the galactic centre , @xmath9 is the mass of the satellite , and @xmath10 is the mass of the galaxy enclosed within @xmath5 . for the observed
@xmath11 this formula has been used to : ( i ) find @xmath9 from the current @xmath5 @xcite ; ( ii ) estimate the pericentre of the satellite s orbit given the value for @xmath9 implied by its internal dynamics ( oh , lin & aarseth 1995 ; irwin & hatzidimitriou 1995 hereafter ih85 ) ; and ( iii ) compare the galaxy s tidal field with the satellite s internal field @xcite .
such analyses have been complicated by the detection of ` extra - tidal ' stars around both dwarf spheroidal galaxies ( ih95 ; kuhn , smith & hawley 1996 ) and globular clusters ( grillmair et al .
1995 hereafter g95 ; for a summary of current observations of both galactic and extra - galactic globular clusters , see grillmair 1998 ) , and these morphological features have been shown to be consistent with those seen in simulations of tidally disrupting systems @xcite .
hut & djorgovski ( 1992 ) also estimated destruction rates of globular clusters by finding the extent to which their distribution of half - mass relaxation times deviates from a power law and interpreting this difference in terms of evolution of the system .
an alternative to these very direct interpretations of observations is to build models that include some representation of all the expected dynamical effects on globular clusters and integrate the cluster s evolution over the lifetime of the galaxy @xcite .
the success of these semi - analytic methods ( as with those in the previous paragraphs ) rests upon the representation of the physics involved , and there is some disagreement about the level of sophistication required for accurate predictions
one example is the theory of tidal shocking which has received renewed attention @xcite .
however , all these studies predict that the milky way s globular cluster system is currently undergoing significant evolution . in this paper , we return to the earlier philosophy of trying to understand how much we can learn directly from observations of tidal signatures .
we approach the problem by using numerical simulations to assess how successfully such features can be interpreted .
our study differs from earlier ones in several ways : ( i ) we include a mass spectrum , and begin with mass segregated models ; ( ii ) there is a one - one representation of stars in e.g. globular clusters ; and ( iii ) the tidal field of a full milky way model is included rather than idealised as giving rise to a spherical tidal boundary or being represented by a one - component potential . in our analysis , we explicitly distinguish between the bound and unbound stars in a satellite , which allows us to identify the characteristics of the extra - tidal population .
we use this analysis to develop and test methods for quantifying the mass loss rate and evolution of the mass function of clusters and satellites directly from current observations .
we present the simulation and analysis methods used throughout the paper in 2 .
the simulations are used to provide an overview of the characteristics of evolution in a tidal field in 3 .
we analyse our models from the viewpoint of an observer 4 . in 5 ,
the interpretations developed in 4 are applied to observations and used to measure the rate of destruction of the milky way s satellite system .
we summarise our results in 6 .
in our calculations , the satellite is represented as a discrete set of particles of different masses , one for each star expected in a cluster of the chosen mass and mass function .
the particular choice of cluster models in this paper follows the parameters adopted by chernoff & weinberg ( 1990 ) ( see also sigurdsson & phinney 1995 ) . the initial particle masses , positions and velocities are taken from an explicit , maximal @xmath12 realisation of an isotropic michie king model of a cluster @xcite , as described in 2.2 .
the cluster s evolution along an orbit in a three component rigid model of the galaxy ( described in 2.3 ) is followed using a self - consistent field ( scf ) code to calculate the mutual interactions of stars in the cluster ( see 2.4.1 ) .
the scf approach can not follow evolution due to close encounters of individual stars in the system , so we repeat one case with the effects of two - body encounters included via diffusion coefficients to illustrate their importance ( see 2.4.2 ) .
this study is intended to isolate the response of a cluster to tidal effects with the dual purposes of identifying observable signatures of tidal interactions and providing a set of control models to compare to future studies which will include more of the physics that might influence the cluster s evolution .
in particular , we make the following simplifications : 1 .
the distribution of particle masses is chosen to represent an evolved stellar population with a turnoff at @xmath13 .
we assume that continued stellar evolution is slow compared to the dynamical timescales involved .
this assumption will be relaxed in subsequent papers , with explicit stellar evolution done in tandem with the dynamical evolution .
the initial cluster models are truncated at some finite radius , @xmath14 , corresponding to the ` tidal radius ' of the king model .
however , @xmath14 is not set to the limiting radius expected along each orbit , as a key purpose of the study is to investigate observable effects of tidal shocks and stripping of a cluster that is not in exact equilibrium with the tidal field .
the chosen models and orbits cover a range of interaction strengths from those that only survive a few orbits to those that could last for the lifetime of the galaxy .
the scenario implicit in these initial conditions is that a cluster is formed very compact , with most of its mass well inside its actual tidal radius .
mass loss during stellar evolution subsequent to the cluster s formation leads to expansion of the cluster , until the tidal radius is reached .
here we assume that there is negligible loss of stars due to tidal effects until late in the cluster s evolution , and that we can start with a relaxed , mass segregated cluster with an exact king model profile , and follow subsequent mass loss due to tidal effects .
clearly this is an approximation to the real physics , but in order to isolate the tidal effects we are exploring , the initial conditions of the cluster must be close to equilibrium or internal dynamical evolution will instead dominate .
as noted above , the effects of two body encounters on the internal dynamical evolution have not been included in the majority of our models .
in this paper we restrict detailed discussions to effects that are insensitive to these simplifications .
for example , in 4 we use our models to test how a cluster s extra - tidal population can be used to determine its mass loss rate
we expect this to be valid because although the mass loss rate itself will depend on the internal dynamics of the system ( which we do not model exactly ) , the characteristics of the extra - tidal population are determined only by the tidal potential ( which we do model exactly ) .
conversely , we do not attempt to arrive at quantitative conclusions in our discussion of the evolution of clusters along different orbits ( 3 ) since these would be influenced by all three of the above simplifications .
[ cols="^,^,^,^,^,^,^,^,^,^ " , ] columns : ( 1 ) name ; ( 2 ) galactocentric distance ; ( 3 ) time period of circular orbit at that distance ; ( 4 ) angle between line of sight and plane perpendicular to satellite s velocity , calculated from proper motions in dauphole et al .
1996 ; ( 5 ) slope of extra - tidal star surface density profile ; ( 6 ) mass loss rate estimate from equation ( [ dfdt1 ] ) ; ( 7 ) mass loss rate estimate from equation ( [ dfdt2 ] ) ( 8) mass loss rate estimate from gnedin & ostriker ( 1997 ) . table [ globstab ] presents the results of mass loss estimation for the 12 globular clusters analysed by g95 ( using their tables 3 - 14 , kindly made available to us by c. grillmair ) .
four of these clusters have well - defined tidal tails with slopes @xmath15 ( given in column 5 ) , and in these cases both @xmath16 and @xmath17 are calculated . in the other cases ,
only the latter estimate is made as an upper limit on the mass loss rate .
the estimated limits for the mass loss rates range from a few to over 100 percent in the next gyr , again implying that the galaxy s globular cluster system will evolve substantially in the next hubble time .
this provides observational support for the many purely theoretical studies that have reached the same conclusion using semi - analytic models @xcite .
the last column shows the destruction rates predicted by gnedin & ostriker ( 1997 ) .
there is no striking correspondence between our ` observational ' results and the purely ` theoretical ' values .
however , the upper limit we calculate is only in direct contradiction with the theoretical calculation in the cases of ngc 288 and ngc 1904 . in the former case
our viewing angle is favourable for making an accurate estimate , but in the latter the value of @xmath18 is sufficiently large that we might expect the result to be confused by the debris geometry . in general , our mass loss rates are higher , but gnedin & ostriker ( 1997 ) themselves point out that their destruction rates should be taken as a lower limit since the rates could be increased with the inclusion of other effects ignored in their study ( such as a mass - spectrum ) .
the results in tables [ dsphtab ] and [ globstab ] should be treated with some caution as in most cases the observed densities of extra - tidal stars do not follow @xmath19 predicted by the simple model which we use to interpret the data . however , the original profiles were made either using star counts directly , or with additional photometry to subtract off some of the background .
since we expect the tidal debris to typically have velocities within @xmath20 of the satellite , a study of radial velocities around a cluster has the potential of refining these measurements considerably .
further in the future , the astrometric satellites sim ( the space interferometry mission ) and gaia ( the galactic astrometric imaging satellite ) promise proper motion measurements with a few @xmath21 accuracy ( or tangential velocities to better than 1 km s@xmath22 out to tens of kpc ) , and a similar advance in identifying debris .
we have also , for the sake of simplicity , restricted our discussion to annularly averaged surface densities .
clearly , more information is contained in two - dimensional surface density maps @xcite . however , these will be more sensitive to the orbital phase and the mass loss history of the satellite and would require more detailed analytic modeling to interpret .
the method proposed in 4.5 uses observations of a cluster s pdmf and extra - tidal stars to place some limits on the imf .
this is clearly a powerful tool for making progress towards understanding whether the imf is in any sense ` universal ' .
of course it is non - trivial to find the global pdmf , the local mass function in the exterior , and to detect extra - tidal stars around a cluster .
however , there are currently two examples in the literature where this has already been done m15 and m55 ( see g95 ; piotto , cool & king 1997 ; zaggia et al .
1997 ) . in the case of m15 , piotto ,
cool & king ( 1997 ) argue that mass - segregation is important only in the centre of the cluster and that their local sample is a fair representation of the global pdmf .
unfortunately , in the absence of mass segregation our method would find no evolution of the imf since it does not model differential mass - loss due to relaxation effects , but only steady stripping of the most weakly bound stars .
( since piotto et al .
1997 make the same statement about ngc 6397 , which has a very different @xmath4 from m15 , this might imply that the differences can only be due to relaxation effects or an intrinsically different imf
it is unclear whether this is a robust argument , or whether a low level of mass segregation farther out in the cluster might be sufficient to account for the differences through tidal shocking . ) in the case of m55 , zaggia , piotto & capaccioli ( 1997 ) find some evidence for extra - tidal stars , but do not give the density of this material since there are several other effects that it could be attributed to . despite these problems ,
these studies show that it is currently feasible to design future observations that can address these issues .
we summarise our main conclusions as follows : 1 . observations of extra - tidal stars and enhanced velocity dispersions in the outskirts of clusters can be attributed to tidally stripped material .
the star counts become dominated by unbound stars at the point where the slope of the surface density profile changes or the dispersion reaches a minimum .
this should not be identified with the tidal radius of the cluster since the edge of the bound population can still lie significantly beyond this radius .
the mass loss rate from a galactic satellite can be directly estimated from the population of extra - tidal stars within a few of its tidal radii , the orbital time period and our line of sight .
3 . using current observations we calculated the mass - loss rate from dwarf spheroidal satellites and globular clusters and found that both systems will undergo significant evolution in the next hubble time .
future observations should be able to place much stronger limits on these destruction rates .
mass - loss estimate together with measurements of the local and global pdmfs of a cluster can place limits on its imf .
our review of the literature suggests that it is currently observationally feasible to carry out this program .
we are grateful to piet hut , douglas heggie and the other occupants of e - building at the ias for helpful discussions , and to the ias , ioa cambridge and sterrewacht leiden for hospitality .
we thank mike irwin and carl grillmair for providing data used for our analysis . this work was supported in part by the national center for supercomputing applications , the pittsburgh supercomputer center , and the epcc ; nasa grant nagw2422 and the nsf under grants ast 9018526 , asc 9318185 , the presidential faculty fellows program ; funds from the ias , a pparc theory grant , an eu marie curie fellowship and the generous support of the british council . | we present the results of a study that uses numerical simulations to interpret observations of tidally disturbed satellites around the milky way . when analysing the simulations from the viewpoint of an observer , we find a break in the slope of the star count and velocity dispersion profiles in our models at the location where unbound stars dominate .
we conclude that ` extra - tidal ' stars and enhanced velocity dispersions observed in the outskirts of galactic satellites are due to contamination by stellar debris from the tidal interaction with the milky way .
however , a significant bound population can exist beyond the break radius and we argue that it should not be identified with the tidal radius of the satellite .
we also develop and test a method for determining the mass loss rate from a galactic satellite using its extra - tidal population .
we apply this method to observations of globular clusters and dwarf spheroidal satellites of the milky way , and conclude that a significant fraction of both satellite systems are likely be destroyed within the next hubble time .
finally , we demonstrate that this mass loss estimate allows us to place some limits on the initial mass function ( imf ) of stars in a cluster from the radial dependence of its present day mass function ( pdmf ) . epsf = = = = = = = = = = = = # 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 = = = = = = = = globular clusters : general stellar dynamics |
graphene is a monolayer of graphite with a band structure composed of two cones located at two inequivalent corners * k * and * k * of the brillouin zone at which conduction and valence bands merge .
this compound has recently received a large attention because of the unusual sequence of quantum hall states it reveals @xcite .
in contrast to conventional two - dimensional electron gas ( c2deg ) which display a quadratic dispersion law , graphene exhibits a linear dispersion law @xmath1 as a function of the momentum @xmath2 leading to a dirac s type hamiltonian with a fermi velocity @xmath3 replacing that of the light .
different one - electron band structure models , not including electron - electron interactions , lead to values @xmath4 with some variance , but this is the value which will adopted in this report .
this peculiar dispersion law has two important consequences in contrast to c2deg ( see for instance @xcite ) : ( i ) the wave functions have a spinor type character and ( ii ) under a magnetic field * b * applied perpendicular to the graphene plane , the dirac energy spectrum evolves into landau levels ( ll ) with energies given by : @xmath5 where @xmath6 scans all positive and negative integer values including zero .
magnetoplasmons ( mp ) , in a two - dimensional electron gas , are excitations between ll , known to be described in terms of excitonic transitions due to electron - electron interactions ( eei ) : they reveal a specific dispersion as a function of the two dimensional wave vector @xmath7 of the exciton . for a c2deg ,
the theory , derived in the frame of the hartree - fock ( hf ) approximation , has been first developed @xcite for integer values of the filling factor @xmath8 ( @xmath9 and @xmath10 being the two - dimensional carrier concentration and the flux quantum respectively ) .
these studies have been extended to the case of non - integer values of @xmath11 @xcite and have also included the calculation of matrix elements for the optical conductivity @xcite .
the effects of eei in graphene have recently been reported on a theoretical basis @xcite but with a different model than that of refs . @xcite and restricted to integer values of the filling factor .
we have followed here the lines of ref .
@xcite , which has been shown to reproduce * quantitatively * experimental results @xcite when they are interpreted in terms of mp excitations .
because of the kohn s theorem @xcite , the eei effects turn out to be tiny for c2deg
. however this theorem does not apply for a linear dispersion law and therefore eei are expected to induce significant effects in graphene .
indeed recent experimental investigations of the magneto - optical transitions in graphene @xcite have been interpreted with an effective velocity @xmath12 , replacing @xmath3 in eq.1 , ranging between @xmath13 to @xmath14 and showing a re - normalization of @xmath3 which here , we will show , is mainly due to electron - electron interactions . on general grounds ,
the coulomb energy characteristic of electron - electron interaction in magnetic field is @xmath15 where @xmath16 is the electronic dielectric constant of the material and @xmath17 the magnetic field length .
the magnetoplasmon approach assumes that @xmath18 is smaller than the one - electron energy transitions . in the present case of graphene @xmath19 and @xmath20
( see eq.1 ) leading to the ratio @xmath21 that is a condition better fulfilled for graphene than for gaas based c2deg and furthermore not dependent on the value of the magnetic field .
magneto - excitons should be therefore more stable in graphene than in c2deg and the approach derived for these later compounds should be valid .
the report is organized as follows : we will first describe the general formalism used to derive the mp dispersion curves ( section ii ) . we apply it to the case of filling factors @xmath22 in section iii and to the case of @xmath23 in section iv .
results will be discussed and compared to experimental results in section v. the details of calculations are reported in the appendices .
in contrast to the gaas case , graphene has two valleys which lead to the conclusion that in the absence of spin splitting and valley splitting each landau level ( ll ) is , in general , four times degenerate .
the fourfold degeneracy of the @xmath24 ll is still due to spin and valley symmetries , but two of these levels have an electron - like character and the two other ones a hole - like character .
we will restrict our analysis to the zero temperature case . because of the peculiar symmetry of the problem the wave functions have a spinor character which can be expressed in the landau gauge , with the potential vector components of the magnetic field @xmath25 and @xmath26 , as : @xmath27 where @xmath28 is the two - dimensional vector of components @xmath29 and @xmath30 , @xmath31 for @xmath24 and @xmath32 otherwise whereas @xmath33 when @xmath34 respectively .
@xmath35 is the standard normalized landau wave functions .
note that these wave functions differ from those proposed in ref .
@xcite by the phase factor due to the different gauge used . following the lines of ref .
@xcite , we call @xmath36 the creation operator of an exciton of energy @xmath37 , corresponding to a transition from ll @xmath38 with spin @xmath39 in valley @xmath40 to a ll @xmath6 of the same spin and same valley . this operator is defined as a function of @xmath41 and @xmath42 , the standard one particle creation and annihilation operators respectively , as : @xmath43 the total hamiltonian of the system is written as : @xmath44 where @xmath45 is the corresponding one electron energy of the ll @xmath46 with spin @xmath39 in valley @xmath40 .
the coulomb interactions appear in @xmath47 as : @xmath48 \widehat{f}_{\sigma_{1,i_{1}}}(\vec{\rho}_{1 } ) ] \end{split}\ ] ] where @xmath49 $ ] denotes the scalar product and @xmath50 with @xmath51 being the 2d fourier transform of the coulomb potential @xmath52 .
after some calculations we obtain an analytic expression for @xmath47 which is expressed as : @xmath53 in eq.7 , the summation is extended over the ensemble @xmath54 of ll , the ensemble @xmath55 of the y - component of the momentum , the ensemble of spin @xmath56 , both valleys @xmath57 and @xmath58 and the wavevector @xmath59 . the function @xmath60 is defined as : @xmath61 with the usual definition of the integral @xmath62 valid for @xmath63 : @xmath64 where @xmath65 are the laguerre polynomials . for @xmath66 the relation @xmath67 holds . using the random phase approximation ( rpa ) to treat the combination of creation and annihilation operators we arrive to the following expression for the exciton energies ( the notation @xmath68 representing the ground state of the system ) : @xmath69 f_{n_{2},i}^{\sigma}\nonumber\\ & & \hspace{3cm}\times{\cal{a}}_{n , n^{'},\sigma , i}^{+}(\vec{k})|0\rangle \nonumber\\ & & \hspace{-0.5cm}+\sum_{n_{2},n_{4}}\widetilde{e}_{n',n_{2},n , n_{4}}(k_{y},k_{x})(f_{n , i}^{\sigma}-f_{n',i}^{\sigma}){\cal{a}}_{n_{2},n_{4},\sigma , i}^{+ } ( \vec{k})|0\rangle\nonumber\\ & & \hspace{-0.5cm}-\sum_{n_{2},n_{3},\sigma_{2},j}\frac{\widetilde{\widetilde{v}}_{n',n_{2},n_{3},n}(-k_{x},k_{y})}{2\pi}(f_{n , i}^{\sigma}-f_{n',i}^{\sigma})\nonumber\\ & & \hspace{3cm}\times{\cal{a}}_{n_{2},n_{3},\sigma_{2},j}^{+}(\vec{k})|0\rangle \nonumber\\\end{aligned}\ ] ] where @xmath70 is the filling factor of ll @xmath6 with spin @xmath39 in valley @xmath40 .
the matrix elements @xmath71 and @xmath72 are given by : @xmath73 we note , at that level , that eq .
10 is formally equivalent to that obtained for c2deg @xcite except for the definition of the different matrix elements which here takes into account the spinor character of the wave functions . in this equation ,
the second term ( second line ) is a measure of the difference of exchange energies of the ll @xmath38 and @xmath6 .
the third line is related to the direct electron - hole coulomb interaction ( exciton binding energy ) .
both terms are involving excitons of same spin and same valley .
the last term of eq .
10 describes the simultaneous annihilation and creation of excitons at different points of the brillouin zone ( rpa contribution ) : it includes all possible transitions without restriction to spin or valley indices .
the exchange terms deserve a special attention in the present case . the corresponding expression for the exchange in eq .
10 reads as : @xmath74 where all laguerre polynomials have arguments @xmath75 . in eq.10 ( second line ) ,
the summation over @xmath76 , for these terms , has to include all ll from @xmath77 to @xmath78 .
the evaluation of the exchange contributions to the different situations is given in appendix a. solving eq.10 results in diagonalizing an hamiltonian , the size of which depends on the number of transitions which are assumed to be coupled by electron - electron interactions . in reality
this number is very large for graphene due to the existence of interband transitions but since @xmath18 is smaller than the energy of transitions , we can reasonably assume , in the spirit of the hf approximation , that eei will not couple , at first order , transitions with different one - electron energies . in that case , the problem reduces to solve the hamiltonian for each type of optical transitions which depend on the value of the filling factor .
however , even if one solves the whole problem in successive steps , one has to keep in mind that a * common * energy scale should be adopted for all transitions in order to compare such results with experimental ones . when writing the hamiltonian , using eq .
10 , for a given set of transitions in the basis @xmath79 corresponding to these transitions such that @xmath80 , we end up with a matrix which is not symmetric , as in c2deg @xcite , but in addition , here , many matrix elements are complex as this will be seen in appendix b. to make the treatment easier to follow , we adopt the same technique as used in @xcite which consists in writing the hamiltonian in a new basis @xmath81 where @xmath82 is a diagonal unitary matrix . the new hamiltonian is then expressed as : @xmath83 which is now symmetric and has only real matrix elements . in the calculations
we will neglect the spin splitting @xmath84 which is small in the case of graphene @xcite but will not change anyway the conclusions since the optical transitions conserve the spin .
we will furthermore assume in the following that there exists some valley splitting @xmath85 first suggested by gusynin _
the existence of such a valley splitting , has been recently supported by different models .
some of these models include different types of electron - phonon interactions @xcite : they all predict a linear dependence of @xmath85 with the magnetic field .
another one @xcite invokes eei with strain induced gauge field yielding to a valley splitting which varies like @xmath86 .
we assume here , for convenience , that @xmath85 is larger than @xmath84 in such a way the electrons remain in the same valley ( here the valley * k * for instance ) for filling factor @xmath87 .
this is not necessary true since experimental results @xcite tend to favor a situation , where for @xmath88 , the system becomes spin polarized .
the splitting @xmath85 will not be included in the present calculations but its consequence will be discussed in each case where its contribution could be relevant .
all the following results for energies are given in units of @xmath18 and as a function of @xmath89 .
( color on line ) schematic diagram of one - electron transitions used in the magnetoplasmon model for @xmath91 . on the left part of the figure ( a )
are shown the transitions implying the @xmath24 ll .
on the right part of the figure ( b ) are shown the first interband ( electron - hole ) transitions from the @xmath92 ll to the @xmath93 ll respectively . for @xmath87 , the transition @xmath94 in ( a ) disappears and the new transition @xmath95 appears.the splitting of ll @xmath6
mimics the spin splitting . ] for @xmath22 , the typical transitions to be analyzed are displayed in fig.1 .
there are two kinds of one electron transitions , in each valley , those which imply the @xmath24 ll ( fig.1a ) and those which correspond to interband ( electron - hole ) transitions ( in fig.1b are only represented those from ll @xmath96 to @xmath97 ) .
as already mentioned in the preceding section one can treat independently the one - electron transitions implying the @xmath24 ll and those involving the interband transitions .
the different one - electron energy transitions which are considered in this case are displayed in fig.1a for @xmath98 . in this figure
the splitting of ll @xmath6 mimics the spin - splitting @xmath84 for clarity but , as already said , this splitting is not taken into account in the present calculations .
the hamiltonian to be solved is therefore a matrix of rank 5 written , first , in the basis @xmath99 ( see fig .
1a for notations ) and then transformed according to eq
the corresponding diagonal matrix @xmath82 is denoted here @xmath100 and has the following diagonal elements : @xmath101 where @xmath102 is the partial filling factor of the spin - up @xmath24 ll and @xmath103 the polar angle of the exciton wave vector . for @xmath98 , the matrix @xmath104 to be diagonalised is expressed as : @xmath105 the different matrix elements are given in appendices a ( eq .
a1 ) and b ( eqs .b1 , b2 , b3 ) .
one can note that the eigenvalues of @xmath104 are identical for @xmath102= 0 or 1 ( that is @xmath11=1 or 2 ) whereas those for non integer values of @xmath11 are symmetric with respect to @xmath106 .
for @xmath87 the corresponding hamiltonian @xmath107 has to be written in the basis @xmath108 ( see fig1a ) , replacing @xmath102 by @xmath109 with a new diagonal matrix @xmath82 denoted now @xmath110 which has the following elements : @xmath111 where @xmath109 is the partial filling factor of the spin - down @xmath24 ll .
the corresponding expressions for the matrix elements are given in appendices a ( eq .
a1 ) and b ( eqs .b1 , b2 , b3 ) .
it turns out that the eigen values of @xmath107 are symmetric of those obtained for @xmath104 with respect to @xmath112 .
if we adopt a model where @xmath85 is smaller than @xmath84 for @xmath87 , we obtain an hamitonian with the same eigen values which shows that the mp results do not depend on this assumption .
results for the mp dispersion curves are displayed in fig.2 . for @xmath112 or @xmath113 one obtains , for the dispersion curves , a solution @xmath114 three times degenerate and one solution @xmath115 which have the following analytical expressions : @xmath116 where @xmath117 is the one - electron energy for these transitions and @xmath118 defined in appendix a ( eq . a2 ) is a quantity a priori divergent which will be discussed in section v. for non integer values of @xmath11 , the solutions @xmath114 remain twice degenerate , the high energy solution remains close to @xmath115 and two new solutions appear .
the linear dispersion near @xmath119 for @xmath115 is due to the rpa contribution entering eq .
as compared to the solutions found in c2deg for @xmath120 @xcite , this contribution is the same whereas that of the exciton binding energy is different .
the solutions for @xmath119 will be further discussed in section v below .
( color on line ) variation of the magnetoplasmon energies in units of @xmath121 as a function of @xmath122 for the transitions involving the @xmath24 ll and filling factors @xmath123 .
the dotted circles denote the degeneracy of the transitions . ]
following the lines of ref.@xcite we have also calculated , in the frame of the mp picture , the optical conductivity ( see appendix c , eq .
c4 ) which predicts that @xmath115 should be optically active in both polarizations of the light ( note that the optical vectors are proportional to @xmath124 ) .
the mp model has been derived without including the valley splitting @xmath85 : if such a splitting is introduced , for the @xmath24 ll , we expect a corresponding splitting of the optical transition independent on the relative magnitude of @xmath85 and @xmath84 . we discuss now the case of interband transitions displayed in fig.1b .
there are , in this case , eight possible one - electron transitions and the hamiltonian is written first in the basis @xmath125 and then transformed according to eq.13 . for @xmath126 , the corresponding diagonal matrix @xmath82 denoted here as @xmath127 has the following diagonal elements : @xmath128 for these transitions the symmetrized excitonic hamiltonians @xmath129 and @xmath130 have matrix elements which are given in appendices a ( eqs .
a3 , a4 ) and b ( eqs .
b4 , b5 , b6 ) .
the dispersion of mp energies , in units of @xmath18 , are displayed in fig.3 as a function of @xmath131 .
the corresponding one electron energy for these transitions is @xmath132 .
( color on line ) variation of the magnetoplasmon energies in units of @xmath121 as a function of @xmath122 for the transitions @xmath92 to @xmath93 ll and filling factor @xmath133 or @xmath113 .
the dotted circles denote the degeneracy of the transitions.the magnetoplasmon curves in the left panel are not degenerate and are the only optically active transitions . ] as for the preceding case , the solutions are identical for @xmath112 or @xmath113 and symmetric with respect to @xmath112 .
for integer values of @xmath11 , the eigen - values of the hamiltonian can be expressed analytically and arranged in two groups : ( i ) two single solutions @xmath134 displayed in the left part of the fig.3 and ( ii ) two other sets of solutions @xmath135 , three times degenerate ( see dotted circles in fig.3 ) , displayed in the right part of the figure .
they are expressed as : ( color on line ) variation of the magnetoplasmon energies in units of @xmath121 as a function of @xmath122 for the transitions @xmath136 to @xmath137 ll and filling factor @xmath133 or @xmath113 .
the dotted circles denote the degeneracy of the transitions.the magnetoplasmon curves in the left panel are not degenerate and are the only optically active transitions . ]
@xmath138 and @xmath139 where all matrix elements entering eqs .
19 and 20 are function of @xmath131 and given in appendices a ( eq . a4 ) and b ( eqs .
b5 , b6 ) .
only the solutions @xmath140 are optically active ( see appendix c , eq .
c5 ) . for non - integer values of the filling factor
the results are very close to those presented in fig.3 except for two solutions of the two groups of degenerate transitions which are no longer degenerate for @xmath119 .
in contrast to the case of transitions implying the @xmath24 ll , a splitting of the transitions equal to @xmath141 and due to electron - electron interactions is expected for @xmath119 .
note however , here , that the introduction of a valley splitting @xmath85 should only provide an additional component either linear in @xmath142 or in @xmath86 depending on the origin of this valley splitting . in this case
the corresponding hamiltonian @xmath143 has the same structure that @xmath144 and therefore only the values of matrix elements are different .
they are given in appendices a ( eqs .
a5 , a6 ) and b ( eqs .
b7 , b8 ) .
the dispersion of mp energies , in units of @xmath18 , is displayed in fig.4 as a function of @xmath145 .
the corresponding one electron energy for these transitions is @xmath146 .
the solutions are formally identical to those given in eqs . 19 and 20 with the appropriate changes for the matrix elements given in the appendices a ( eq .
a5 ) and b ( eqs .
b7 , b8 ) .
the splitting of the transitions for @xmath119 is here equal to @xmath147 .
we , now , evaluate the exciton energies for @xmath148 .
we will concentrate the report for filling factors @xmath23 .
the contributions of exchange are given in appendix a ( eq .
a7 ) . it turns out that the problem to solve is symmetric with respect to @xmath150 and therefore we will detail the treatment for @xmath151 and will note only the main changes for @xmath152 . in this case
we have to treat the problem depicted in fig.5 for the one electron energy transitions .
note that we have here two types of transitions those implying the @xmath24 ll and those between ll @xmath153 and @xmath154 . because the corresponding one electron energies are different they are treated independently .
( color on line ) schematic diagram of one - electron transitions used in the magnetoplasmon model for @xmath155 . ] for @xmath156 , we have to write , first , the hamiltonian in the basis @xmath157 and for @xmath158 in the basis @xmath159 and transformed them according to eq . 13 . for @xmath156 and @xmath158 , the corresponding diagonal matrices @xmath82 ,
denoted here as @xmath160 and @xmath161 respectively , have the following diagonal elements : @xmath162 where @xmath163 and @xmath164 are the partial filling factors of the spin - down and spin - up @xmath153 ll respectively , attached to the valley k with our convention .
one gets for the corresponding hamiltonians @xmath165 and @xmath166 the following expressions : @xmath167 and @xmath168 where the new matrix elements entering these matrices are given in appendices a ( eqs .
a7 ) and b ( eqs .
b9 , b10 ) .
the resulting dispersion curves are displayed in fig.6 .
the corresponding one - electron energies for both types of transitions are @xmath169 for the higher ones and @xmath170 for the lower ones .
( color on line ) variation of the magnetoplasmon energies in units of @xmath121 as a function of @xmath122 for different filling factors @xmath171.the dotted circles denote the degeneracy of the transitions . ] the eigen - values of the hamiltonians ( eqs . 22 and 23 )
can be expressed analytically . for @xmath156
there are two solutions identical to @xmath114 ( eq.17 ) which remain degenerate and which are not optical active .
two other non degenerate solutions denoted as @xmath172 are given by : @xmath173 \end{split}\ ] ] and a third one @xmath174 .
these transitions are all optically active with a relative weight depending on the filling factor . for @xmath158 ,
it remains one eigen - value solution identical to @xmath114 ( eq.17 ) and two groups of optically active non degenerate solutions denoted as @xmath175 and @xmath176 with the following analytical expressions : @xmath177 \end{split}\ ] ] and : @xmath178 \end{split}\ ] ] for which all matrix elements are function of @xmath131 .
the corresponding optical vectors for these transitions are given in appendix c ( eqs .
c7 and c8 ) .
( color on line ) variation of the magnetoplasmon energies in units of @xmath121 as a function of @xmath122 for different filling factors @xmath179 .
the dotted circles denote the degeneracy of the transitions . ]
it is easy to see that , in this case , the structures of the corresponding hamiltonians @xmath180 and @xmath181 are symmetric with respect to those given in eqs . 22 and 23 . at present , this is the @xmath153 ll attached to the valley k ( in our convention ) which starts to be filled and the notation @xmath182 refer to this ll . of course some of the diagonal matrix elements are changed but results are formally similar and the corresponding exciton dispersion curves are displayed in fig.6 .
all the eigen - value solutions of fig.6 can be expressed analytically : for @xmath183 one gets one solution
@xmath184 which is not optically active and two groups of optically active solutions : @xmath185 \end{split}\ ] ] and : @xmath186 \end{split}\ ] ] for which all matrix elements , dependent on @xmath131 , are given in appendix b ( eq .
b11 ) . for @xmath187 one
gets two solutions @xmath188 ( same expression as for @xmath183 ) which are not optically active , one optically active solution @xmath189 and two other optically active solutions denoted as @xmath190 : @xmath191 \end{split}\ ] ] the related matrix elements , dependent on @xmath131 , are given in appendix b ( eq .
we will not discuss the case of interband transitions for this configuration of filling factors but their corresponding hamiltonians @xmath192 and @xmath193 are modified in two respects : for both of them the exchange contributions entering the diagonal elements are different and for @xmath192 the transitions are now filling factor dependent in such a way the corresponding transitions disappear at @xmath194 .
the case of filling factors @xmath195 and following ones will not be discussed as well but the corresponding treatment is formally similar to the case @xmath196 with a different set of one electron energies , exchange contributions and matrix elements .
we are now focussing the discussion on the results obtained for @xmath119 which could be compared to magneto - optical absorption measurements .
for @xmath198 , all the hamiltonians are reduced to their diagonal elements which are given in appendix b. the reason is that all off - diagonal elements are proportional to @xmath131 or @xmath199 .
we will call the corresponding solutions , at @xmath119 , @xmath200 and @xmath201 for intra - ll transitions and interband transitions respectively .
we restrict the discussion to those solutions which are optically active .
all results are function of @xmath202 in coulomb units .
we then get the following results : for the transitions @xmath203 which involve the @xmath24 ll : @xmath204 as clearly apparent in fig . 6 and 7 ,
this transition is split for non integer values of @xmath148 .
the high energy component of this split level has an energy which increases with @xmath11 but its oscillator strength decreases with @xmath11 going to zero at integer value of @xmath11 . for the transitions @xmath205
which involve the transitions between the @xmath153 and @xmath154 ll : @xmath206 in this case also this transition is split for non integer values of @xmath11 .
the high energy component of this split level has an energy which increases with @xmath11 but its oscillator strength decreases with @xmath11 going to zero at integer value of @xmath11 . the optical active interband transitions @xmath207 which involve the transitions between the @xmath92 and @xmath93 landau levels are split by an amount @xmath208 but the mean energy @xmath209 ( eq .
19 ) has the following expression : @xmath210 whereas the mean energy for the optically active transition implying the @xmath136 and @xmath137 landau levels is expressed as : @xmath211 for a given value of the fermi velocity @xmath3 ( here equal to @xmath212 ) the energy @xmath213 is determined ( here @xmath214 ) and in eqs .
30 to 33 the only unknown parameter is @xmath118 .
as already said this quantity defined in appendix a ( eq . a2 ) is divergent .
the occurrence of such a problem is not specific of the graphene properties because it is also present in c2deg though it was not explicitly formulated .
the reason why this term appears here is that we wanted to define a * common * energy scale for intraband and interband transitions .
this will corresponds in c2deg to impose a common energy scale to cyclotron - type transitions and inter - band excitonic transitions .
there was an attempt to treat this later transitions in gaas , in an another context , but using the same theoretical model @xcite and indeed the same problem of divergence of the exchange interaction among the valence band levels was found without being able to solve it .
therefore this problem is not specific to graphene but , in that case , one can solve it , at least , in a semi - empirical way .
the divergence of @xmath118 is due to the infinite summation over ll ( see eq .
a2 ) which is physically artificial . we could then define , as was done in ref .
@xcite , a cut - off value on energy or number of ll , but this limit is quite arbitrary .
we propose to treat the problem in a semi - empirical way , using @xmath118 as a parameter fitted , for one type of transitions , to experimental data and then to deduce all the re - normalized velocities attached to the other transitions .
doing so , we _ implicitly assume _ that all the re - normalization of the velocity , for the fitted transition , is only due to electron - electron interactions neglecting any possible contribution from electron - phonon interaction which may be important in carbon based compounds .
among experimental data which could be used for this fitting , those related to magneto - transmission measurements @xcite are those which are expected to reflect the magneto - plasmon picture developed is this study .
another set of data , based on photoconductivity measurements on ex - foliated graphene @xcite can also be considered to compare results .
we will use the data of ref.@xcite , obtained on ex - foliated graphene , to fit @xmath118 to the @xmath203 transition at @xmath120 . in this reference , the re - normalized fermi velocity @xmath215 which , from eq.30 , leads to a value @xmath216 .
when injecting this value in eq .
31 , we predict a re - normalized fermi velocity @xmath217 to be compared with the corresponding experimental value @xcite @xmath218 .
the agreement is reasonable .
as seen on both experimental and theoretical grounds , the re - normalized velocity differs for different transitions .
one can then try to evaluate these velocities for other transitions .
the results are given in table 1 for integer values of @xmath11 .
the value quoted in the table , from ref.@xcite , corresponds to the transition @xmath24 to @xmath153 whereas the corresponding value for the transition @xmath219 to @xmath24 is found to be @xmath220 .
this corresponds to an asymmetry of the conduction and valence levels not taken into account in our model but also not reported in ref.@xcite for this transition . in table 1
, we have included the results of the re - normalized fermi velocity for transitions between ll @xmath221 to @xmath222 , @xmath154 to @xmath221 , @xmath223 to @xmath222 for which we have calculated the diagonal elements of the corresponding hamiltonians .
.[table 1 ] evaluation of the re - normalized velocities @xmath12 , at integer values of the filling factor , for different transitions .
[ cols="^,^,^,^ " , ] as clearly apparent , from table 1 , @xmath12 varies strongly with the transition though it seems to be relatively constant for all interband transitions .
this qualitative feature is also observed in experiments performed on epitaxial multi - layer graphene @xcite . however in these experiments the reported values of @xmath224 for all transitions is the same and equal to @xmath225 which is , at present , not understood .
one could , a priori , think that the results obtained are dependent on the value of @xmath3 adopted in the calculations .
in fact one can vary this value over a large range , for instance , from @xmath226 to @xmath227 getting values for @xmath228 to @xmath229 respectively but the quantity which enters the hamiltonians is in fact @xmath230 which * remains constant * , independent on @xmath3 and equal to @xmath231 .
this value has been adopted to calculate the dispersion curves of figs . 2 , 3 , 4 , 6 and 7 .
it is therefore not possible from experimental data on energies to determine @xmath118 but all transitions are now given with a * common * energy scale .
on the other hand the oscillator strengths of the transitions are proportional to @xmath124 and then absolute transmission measurements could in principle give information on @xmath3 and therefore on @xmath118 .
in conclusion , we have developed , within the hartree - fock approximation , a full treatment of the magnetoplasmon picture in graphene valid for a very large range of magnetic fields .
this model , applied for filling factors up to 6 , shows that the electron - electron interactions induce different effects : ( i ) for some of the transitions these interactions lead to a splitting of the optical transitions and ( ii ) they are responsible for a strong re - normalization of the fermi velocity as observed in magneto - optical experiments .
this re - normalization is found to be dependent on the type of investigated transitions .
the optical conductivity components have been evaluated showing that the oscillator strength of the optical transitions is proportional to @xmath124 and not to the square of the re - normalized velocity.the theory has been derived for all transitions with a common energy scale which should allow a direct comparison of its predictions with future experimental works .
the ghmfl is `` laboratoire conventionn lujf et linpg de grenoble '' .
the work presented here has been supported in part by the european commission through the grant rita - ct-2003 - 505474 .
we report in this appendix the explicit expressions for the contribution of the exchange energies entering the diagonal elements of the different hamiltonian matrices in units of coulomb energies .
we introduce the notation @xmath232 which characterize the exchange interaction in c2deg at @xmath112 . applying the expression given in eq .
12 we obtain successively the contribution of exchange for the different hamiltonians . to simplify the notations we will drop the superscript @xmath233 from @xmath234 meaning that all these quantities are real . for @xmath104
we obtain : @xmath235 where : @xmath236 and @xmath237 are laguerre polynomials of argument @xmath75 in this appendix .
the quantity @xmath118 diverges due to the simplifying assumption of the infinite linear dispersion of the graphene bands .
the summation has to be truncated at some level or this parameter has to be fitted to experimental data ( see section 5 ) . for @xmath107
we obtain for the exchange part the same expressions than those given in eq.a1 replacing @xmath102 by @xmath109 . for @xmath240 one
gets : @xmath241 where : @xmath242\nonumber\\ & = ( \sqrt2 + 1)c_{1}+ \delta c_{2}\nonumber\\ \delta c_{2}&= -\frac{1}{2}\sum_{m=0}^{\infty}\int_{0}^{\infty}dx e^{-x^{2}}\frac{x^{4}}{\sqrt{m+1 } } % \tag{a4}\end{aligned}\ ] ] @xmath243 in this equation also diverges like @xmath118 but @xmath244 converges to a value -0.156 .
similar expressions hold for @xmath245 when replacing @xmath102 by @xmath109 . for @xmath248 one
gets : @xmath249 where : @xmath250 @xmath251 in this equation also diverges like @xmath118 but @xmath252 converges to a value -0.467 .
similar expressions hold for @xmath253 when replacing @xmath102 by @xmath109 . comparing eqs .
a4 and a6 one can formally extend the treatment and find that , for any interband transition from ll @xmath254 to ll @xmath255 the corresponding divergent term @xmath256 entering the exchange contributions is given by @xmath257 where @xmath258 is finite .
the different contributions to the exchange for the different hamiltonians are : @xmath260 where @xmath261 . here also formally , when extending the treatment , one finds that , for any intra - ll transition from ll @xmath262 to ll @xmath263 the corresponding divergent term @xmath264 entering the exchange contributions is given by @xmath265 where @xmath266 is finite .
note however that in general @xmath266 is different from @xmath258 .
we report in this appendix the explicit expressions for the matrix elements of the hamiltonian matrices in units of coulomb energies .
results are given as a function of @xmath267 .
for simplicity we adopt the same notation @xmath268 for noting the matrix elements of all matrices but their expression is specific of the case under consideration .
all matrix elements @xmath269 and @xmath270 are evaluated using eq .
11 . for @xmath104
we obtain : @xmath271 the matrix elements @xmath272 entering the hamiltonian @xmath104 are : @xmath273 with @xmath274 and @xmath275 where @xmath103 is the polar angle of the exciton wave vector .
the matrix elements @xmath276 entering eqs . 15 and b1 are : @xmath277 \\
e_{0011}(k)&=-\frac{3k^{2}}{32}\sqrt{\frac{\pi}{2}}\phi(\frac{5}{2},3;-\frac{k^{2}}{2 } ) % \tag{b3 } \end{split}\ ] ] where @xmath278 , @xmath279 and @xmath280 is the confluent hypergeometric function .
for @xmath107 the matrix elements are identical to those given in eqs .
b1 , b2 and b3 replacing @xmath102 by @xmath109 when appropriate .
the matrix elements @xmath268 of @xmath240 are : @xmath281 where @xmath282 .
the matrix elements @xmath272 entering eq .
b4 are : @xmath283\\ % \tag{b5 } \end{split}\ ] ] with @xmath284 and @xmath285 .
the matrix elements @xmath276 entering b4 are : @xmath286\\ e_{-112 - 2}(k)= -\frac{k^{2}\sqrt{\pi}}{64}[\frac{3(1+\sqrt{2})^{2}}{\sqrt{2 } } \phi(\frac{5}{2};3;-\frac{k^{2}}{2})\\ -15(2+\sqrt{2})\phi(\frac{7}{2};3;-\frac{k^{2}}{2 } ) + \frac{105\sqrt{2}}{16}\phi(\frac{9}{2};3;-\frac{k^{2}}{2 } ) ] % \tag{b6}\end{gathered}\ ] ] where @xmath287 and @xmath288 . for @xmath245 two columns of the matrix @xmath240
are inverted but the eigen values are the same with @xmath109 replacing @xmath102 in eq
. b6 .
the matrix elements @xmath268 of @xmath248 are similar to those given in eq .
b4 when replacing @xmath289 , @xmath290 , @xmath291 , @xmath292 by @xmath293 , @xmath294 , @xmath295 , @xmath296 respectively and @xmath297 by @xmath298 .
the new matrix elements @xmath276 are here : @xmath299\\ e_{-223 - 3}(k)&=-\frac{k^{2}\sqrt{\pi}}{64}[3(5 + 2\sqrt6 ) \phi(\frac{5}{2};3;-\frac{k^{2}}{2})\\ & -15(4+\frac{3\sqrt3}{\sqrt{2}})\phi(\frac{7}{2};3;-\frac{k^{2}}{2})\\ & + 105(\frac{9}{8}+\frac{\sqrt{6}}{3})\phi(\frac{9}{2};3;-\frac{k^{2}}{2})\\ & -945\frac{6+\sqrt6}{48}\phi(\frac{11}{2};3;-\frac{k^{2}}{2})\\ & + \frac{10395}{192}\phi(\frac{13}{2};3;-\frac{k^{2}}{2 } ) ] % \tag{b8 } \end{split}\ ] ] where @xmath300 and @xmath301 . and the corresponding matrix elements @xmath272 : @xmath302 % \tag{b7 } \end{split}\ ] ] with @xmath303 and @xmath304 . for @xmath165
we obtain : @xmath305 for @xmath166 : @xmath306 for @xmath180 : @xmath307 for @xmath181 : @xmath308 with the corresponding new matrix elements entering eq .
b9 , b10 , b11 and b12 : @xmath309^{2}\nonumber\\ e_{1221}(k)&=e_{-122 - 1}(k ) % \tag{b13}\end{aligned}\ ] ] where @xmath310
in this appendix we derive , following the lines of ref.@xcite , the corresponding expressions which allow to calculate the optical matrix elements of the mp curves which enter in the optical conductivity @xmath311 which has two components : @xmath312 where the summation is performed on all mp transitions of energy @xmath313 with the corresponding eigen vector @xmath314 . in eq .
c1 , @xmath315 is the density of states of a single ll , @xmath316 and @xmath317 are optical vectors with components @xmath318 and @xmath319 respectively .
@xmath82 is the matrix used to symmetrize the hamiltonian and depends on the set of transitions which are considered ( see eqs .
16 , 18 , 21 ) .
@xmath320 denotes one of the transition @xmath6 to @xmath46 belonging to this set of transitions .
@xmath321 with @xmath322 is defined in reduced units ( @xmath323 standing for @xmath324 and @xmath325 for @xmath326 ) as : [ [ optical - vectors - for - widetildeh_1nu20-and - widetildeh_0nu10 ] ] optical vectors for @xmath104 and @xmath107 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ it can be easily verified that for @xmath336 where all transitions become degenerate , there is a sum rule such that @xmath337 independent of the filling factor whereas @xmath338 .
one therefore recover the selection rules obtained for the one - electron model .
note that the fermi velocity entering in the optical matrix elements is that existing in the _ absence _ of electron - electron interactions .
[ [ optical - vectors - for - widetildehi_1nu212-and - widetildehi_0nu112 ] ] optical vectors for @xmath240 and @xmath245 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ [ [ optical - vectors - for - widetildehi_1nu223-and - widetildehi_0nu123 ] ] optical vectors for @xmath248 and @xmath253 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ to conclude this part , it is worth comparing these results with those obtained in c2deg where the introduction of electron - electron interactions change the selection rules @xcite with respect to those obtained in the one - electron picture . in the case of graphene ,
the selection rules are in general , similar to the results obtained in the one - electron picture , except that the oscillator strength is condensed into one or two branches of the mp curves and that the strength remains proportional to @xmath124 and not to the square of the re - normalized velocity . | in the frame of the hartree - fock approximation , the dispersion of magnetoplasmons in graphene is derived for all types of transitions for filling factors @xmath0 .
the optical conductivity components of the magnetoplasmon curves are calculated .
it is shown that the electron - electron interactions lead to a strong re - normalization of the apparent fermi velocity of graphene which is different for different types of transitions . |
a classical problem in analysis is the well - known _
helmholtz s inverse problem of the calculus of variations _ : find a necessary and sufficient condition under which a ( system of ) differential equation(s ) can be written as an euler lagrangeagrange or a hamiltonian equation and , in the affirmative case , find all the possible lagrangian or hamiltonian formulations .
this condition is usually called _ helmholtz condition_. + the lagrangian helmholtz problem has been studied and solved by j. douglas @xcite , a. mayer @xcite and a. hirsch @xcite . the hamiltonian helmholtz problem has been studied and solved up to our knowledge by r. santilli in his book @xcite .
+ generalization of this problem in the discrete calculus of variations framework has been done in @xcite and @xcite in the discrete lagrangian case . for the hamiltonian case
it has been done for the discrete calculus of variations in @xcite using the framework of @xcite and in @xcite using a discrete embedding procedure .
+ in this paper we generalize the helmholtz theorem for hamiltonian systems in the case of time scale calculus using the work of @xcite and @xcite in the context of nonshifted calculus .
we recover then the discrete case @xcite when the time scale is a set of discrete points and the classical case @xcite when the time scale is a continuous interval .
+ the paper is organized as follow : in section 2 , we recall some results on time scale calculus . in section 3 ,
we give the definition of time scale embedding . in section 4
, we give a brief survey of the classical helmholtz hamiltonian problem . in section 5
, we introduce the definition of a time scale hamiltonian in the context of nonshifted calculus of variations and then we enunciate and prove the main result of this paper , the time scale hamiltonian helmholtz theorem . finally ,
in section 6 , we conclude and give some prospects .
we refer to @xcite and references therein for more details on time scale calculus .
+ we consider @xmath0 which denotes a bounded time scale with @xmath1 , @xmath2 and @xmath3 . the backward and forward jump operators @xmath4 are respectively defined by : @xmath5 where we put @xmath6 and @xmath7 .
a point @xmath8 is said to be left - dense ( resp .
left - scattered , right - dense and right - scattered ) if @xmath9 ( resp .
@xmath10 , @xmath11 and @xmath12 ) .
let @xmath13 ( resp .
@xmath14 , @xmath15 and @xmath16 ) denote the set of all left - dense ( resp .
left - scattered , right - dense and right - scattered ) points of @xmath0 .
+ the graininess ( resp .
backward graininess ) function @xmath17 ( resp .
@xmath18 ) is defined by @xmath19 ( resp .
@xmath20 ) for any @xmath8 .
we set @xmath21\rho(b),b]$ ] , @xmath22 and @xmath23 .
note that @xmath24 since @xmath3 .
let us recall the usual definitions of @xmath25- and @xmath26-differentiability .
a function @xmath27 , where @xmath28 , is said to be @xmath25-differentiable at @xmath29 ( resp .
@xmath26-differentiable at @xmath30 ) if the following limit exists in @xmath31 : @xmath32 in such a case , this limit is denoted by @xmath33 ( resp .
@xmath34 ) . let us recall the following results on @xmath25-differentiability .
[ rappeldelta1 ] let @xmath27 and @xmath29 .
the following properties hold : 1 . if @xmath35 is @xmath25-differentiable at @xmath36 , then @xmath35 is continuous at @xmath36 .
if @xmath37 and if @xmath35 is continuous at @xmath36 , then @xmath35 is @xmath25-differentiable at @xmath36 with : @xmath38 3 .
if @xmath39 , then @xmath35 is @xmath25-differentiable at @xmath36 if and only if the following limit exists in @xmath31 : @xmath40 in such a case , this limit is equal to @xmath33 .
[ rappeldelta2 ] let @xmath27 .
then , @xmath35 is @xmath25-differentiable on @xmath41 with @xmath42 if and only if there exists @xmath43 such that @xmath44 for every @xmath8 . the analogous results for @xmath26-differentiability are also valid .
a function @xmath35 is said to be rd - continuous ( resp .
ld - continuous ) on @xmath0 if it is continuous at every @xmath39 ( resp .
@xmath45 ) and if it admits a left - sided ( resp .
righ - sided ) limit at every @xmath45 ( resp .
@xmath39 ) . we respectively denote by @xmath46 and @xmath47 the functional spaces of rd - continuous functions on @xmath0 and of @xmath25-differentiable functions on @xmath41 with rd - continuous @xmath25-derivative
. we also respectively denote by @xmath48 and @xmath49 the functional spaces of ld - continuous functions on @xmath0 and of @xmath26-differentiable functions on @xmath50 with ld - continuous @xmath26-derivative .
+ let us recall the following results on the continuity of the forward and backward jump
. [ propcontinuity ] let @xmath30 .
the following properties are equivalent : 1 .
@xmath51 is continuous at @xmath36 ; 2 .
@xmath52 ; 3 .
@xmath53 .
also , let @xmath29 .
if the following properties are equivalent : 1 .
@xmath54 is continuous at @xmath36 ; 2 .
@xmath55 ; 3 .
@xmath56 .
[ propbourdin ] let @xmath57 .
if the following properties are satisfied : * @xmath51 is @xmath26-differentiable on @xmath50 ; * @xmath35 is @xmath25-differentiable on @xmath41 ; then , @xmath58 is @xmath26-differentiable at every @xmath59 with @xmath60 also , if the following properties are satisfied : * @xmath54 is @xmath25-differentiable on @xmath41 ; * @xmath35 is @xmath26-differentiable on @xmath50 ; then , @xmath61 is @xmath25-differentiable at every @xmath59 with @xmath62 let us now derive a result about @xmath25-differentiability of @xmath54 and the @xmath63-differentiability of @xmath51 . [ propinv ]
if @xmath51 is @xmath63-differentiable on @xmath50 and @xmath54 is @xmath25-differentiable on @xmath41 , then we have @xmath64 for all @xmath65 .
if @xmath51 is @xmath63-differentiable on @xmath50 and @xmath54 is @xmath25-differentiable on @xmath41 , then the following limits then exist and are finite @xmath66 and also @xmath51 is continuous on @xmath50 and @xmath54 is continuous on @xmath41 ( proposition ) . by the continuity of @xmath51 and @xmath54 ( proposition ) we have only only two cases to consider , @xmath67 or @xmath68 . if @xmath67 , then @xmath69 and @xmath70 and then we have @xmath64 . if @xmath68 , then @xmath71 and @xmath72 and then we have also the result .
let us denote by @xmath73 the cauchy @xmath25-integral defined in @xcite with the following result . for every @xmath74 ,
there exist a unique @xmath25-antiderivative of @xmath35 in sense of @xmath75 on @xmath41 vanishing at @xmath76 . in this case
the @xmath25-integral is defined by @xmath77 for every @xmath8 .
the @xmath25-integral has the following properties : * if @xmath78 and @xmath79 and @xmath80 are @xmath25-differentiable , then the following formula of integration by parts hold : @xmath81 * if @xmath78 , @xmath79 is @xmath26-differentiable and @xmath80 and @xmath54 are @xmath25-differentiable , then the following formula of integration by parts hold : @xmath82 the first integration by parts formula ( i ) is well known in the litterature of the time scale calculus . the second
( ii ) is obtain by @xmath25-integration of the leibniz formula ( * ? ? ?
* proposition 7 p.552 ) .
+ let us now remind some definitions of variationnal calculus .
let @xmath83 be a lagrangian _ i.e. _ a continuous map of class @xmath84 in its two last variables : @xmath85{lrcl}l : & \tk \times { \mathbb{r}}^n \times { \mathbb{r}}^n & \longrightarrow & { \mathbb{r}}\\&(t , x , v ) & \longmapsto & l(t , x , v ) \end{array}}\ ] ] and let @xmath86 be the following lagrangian functional : @xmath85{lrcl}\l : & \cdrd({\mathbb{t } } ) & \longrightarrow & { \mathbb{r}}\\&u & \longmapsto & \di \int_a^b l(\tau , u(\tau),u^\dd(\tau ) ) \dd \tau . \end{array}}\ ] ] we define @xmath87 to be the set of _ variations _ of @xmath86 .
a function @xmath88 is said to be a _
critical point _ of @xmath86 if @xmath89 for every @xmath90 where @xmath91 denotes the frechet derivative .
we remind what we call a strong form of the well known dubois
reymond lemma on time scales .
let @xmath92 .
then the equality @xmath93 holds for every @xmath94 if and only if there exist @xmath95 such that @xmath96 for all @xmath97 .
up to our knowledge , the classical form of the dubois
reymond lemma on time scales has not been used or proved .
we give the result and we call it the weak form of the dubois reymond lemma
. let @xmath98 .
then the equality @xmath99 holds for every @xmath100 such that @xmath101 if and only if @xmath102 for all @xmath65 .
the sufficient condition is obvious . for the necessary one : let @xmath103 which is clearly positive for all @xmath104 and vanish at @xmath105 and @xmath106 .
let @xmath107 .
we have @xmath108 as @xmath109 for all @xmath65 then necessarily we have @xmath110 for all @xmath65 .
we remind the time scale embedding as defined in @xcite to which we refer for more details . + we denote by @xmath111;{\mathbb{r}})$ ] the set of continuous functions @xmath112 \rightarrow { \mathbb{r}}$ ] .
a time scale embedding is given by specifying : * a mapping @xmath113,{\mathbb{r } } ) \rightarrow c({\mathbb{t}},{\mathbb{r}})$ ] ; * an operator @xmath114,{\mathbb{r } } ) \rightarrow \cdrd(\tk,{\mathbb{r}})$ ] , called a generalized derivative ; * an operator @xmath115,{\mathbb{r}})\rightarrow \crd({\mathbb{t}},{\mathbb{r}})$ ] , called a generalized integral operator .
we fix the following embedding : the mapping @xmath116 is obtained by restriction of functions to @xmath0 .
the operator @xmath117 is chosen to be the @xmath118 derivative , and the operator @xmath119 is given by the @xmath118-integral as follows : @xmath120 [ embeddingdofdifferential ] the @xmath25-differential embedding of an ordinary differential equation @xmath121 for @xmath122,{\mathbb{r}})$ ] and @xmath123,{\mathbb{r } } ) , { \mathbb{r}})$ ] , is given by @xmath124 for @xmath125 and @xmath126 .
[ embeddingdofintegral ] the @xmath25-integral embedding of an integral equation @xmath127 for @xmath122,{\mathbb{r}})$ ] and @xmath123,{\mathbb{r } } ) , { \mathbb{r}})$ ] , is given by @xmath128 for @xmath125 and @xmath126 .
[ embeddingfunctional ] let @xmath129 \times { \mathbb{r}}^2 \rightarrow { \mathbb{r}}$ ] be a continuous function and @xmath130 the functional defined by @xmath131 the time scale @xmath25-embedding @xmath132 of @xmath130 is given by @xmath133
this section is based on the book of @xcite to which we refer for more details .
+ we work on @xmath134 .
we denote by @xmath135 the transpose .
the symplectic scalar product @xmath136 is defined for all @xmath137 by @xmath138 where @xmath139 denote the usual scalar product and @xmath140 with @xmath141 the identity matrix on @xmath142 .
we define the @xmath143 symplectic scalar product induced by @xmath144 defined for @xmath145,{\mathbb{r}}^{2d})$ ] by @xmath146 let @xmath147,{\mathbb{r}}^{2d } ) \longrightarrow c^0([a , b],{\mathbb{r}}^{2d})}$ ] .
we define the adjoint @xmath148 of @xmath149 with respect to @xmath150 by @xmath151 a classical hamiltonian is a function @xmath152 such that for @xmath153,{\mathbb{r}}^d ) \times c^1([a , b],{\mathbb{r}}^d)$ ] we have the time evolution of @xmath154 given by the classical hamilton s equations @xmath155 we say that @xmath156 is hamiltonian .
the critical points @xmath153,{\mathbb{r}}^d ) \times c^1([a , b],{\mathbb{r}}^d)$ ] of the functional @xmath85{lrcl}\mathcal{l}_h : & c^1([a , b],{\mathbb{r}}^d)\times c^1([a , b],{\mathbb{r}}^d ) & \longrightarrow & { \mathbb{r}}\\&(q , p ) & \longmapsto & \mathcal{l}_h(q , p ) = \int_{a}^{b}l_h(q(t),p(t),\dot{q}(t),\dot{p}(t))dt \end{array}}\ ] ] satisfy the hamilton s equations where @xmath157 is the lagrangian defined by @xmath158 [ thm_hamiltonian ] we consider the differential equations associate to a vector field @xmath159 , @xmath160 it defines a natural operator which is written as @xmath161 in this case we have the hamiltonian helmholtz conditions : the operator @xmath162 has its frchet derivative self - adjoint at @xmath154 if and only if @xmath163 we have then the vector field x is hamiltonian if and only if the operator @xmath162 has his frchet derivative self - adjoint with respect to the symplectic scalar product .
+ in this case the hamiltonian associate to @xmath164 is given by @xmath165d\lambda.\ ] ] the classical hamiltonian helmholtz conditions are also the conditions to which the differential form associate to the vector field @xmath164 with respect to the symplectic scalar product is closed .
we define the @xmath166 scalar product defined for @xmath167 by @xmath168 and also the @xmath166 symplectic scalar product defined for @xmath169 by @xmath170 let @xmath171 .
we define the adjoint @xmath148 of @xmath149 with respect to @xmath172 by @xmath173 let @xmath174 denote the set @xmath175 and @xmath176 .
we consider the @xmath25-embedding of @xmath177 , @xmath178 defined by @xmath179 for all @xmath180 .
+ we assume that @xmath54 is @xmath25-differentiable on @xmath41 and @xmath51 is @xmath26-differentiable on @xmath50 .
+ a time scale hamiltonian is a function @xmath152 such that for @xmath180 we have the time evolution of @xmath154 given by the time scale hamilton s equations under the _ derivative form _ @xmath181 or under the _ integral form _ @xmath182 where @xmath183 and @xmath184 are constants .
moreover , the derivative form and the integral form are equivalent .
the critical points @xmath180 of the functional @xmath178 satisfy the time scale hamilton s equations @xmath185 or equivalently @xmath186 .
let @xmath187 .
the frchet derivative of @xmath178 at @xmath154 along @xmath188 is given by @xmath189\dd t.\ ] ] where the frchet derivative of @xmath190 at @xmath154 along @xmath188 is given by @xmath191 first we prove the critical points satisfy the hamilton s equations under the derivative form . + using the integration by parts formula @xmath192 of proposition and using the fact that @xmath35 vanish at @xmath76 and @xmath193 we obtain @xmath194\dd t.\ ] ] using the expression @xmath195 we obtain @xmath196\dd t.\end{aligned}\ ] ] by definition , if @xmath154 is a critical point of @xmath197 then we have @xmath198 and then using the weak form of the dubois
reymond lemma we obtain the time scale hamilton s equations under the derivative form . +
second using the same strategy with the integration by parts formula @xmath199 of proposition and using the strong form of the dubois
reymond lemma we obtain the time scale hamilton s equations under the integral form .
+ equivalence between @xmath185 and @xmath200 is due to the @xmath25-differentiability of @xmath54 on @xmath41 and the @xmath26-differentiability of @xmath51 on @xmath50 . indeed , using proposition we obtain for all @xmath65 @xmath201 then using proposition we obtain for all @xmath65 @xmath202^\nn&=\di \left[\int_{a}^{\sigma(t)}\frac{\partial h(q(\tau),p(\tau))}{\partial p}\dd\tau\right]^\nn \\ p^\nabla(t)&=\di \left[\int_{a}^{\sigma(t ) } -\frac{\partial h(q(\tau),p(\tau))}{\partial q}\dd\tau\right]^\nn \end{array } \right . \
.\end{aligned}\ ] ] using the @xmath63 version of proposition we obtain @xmath203 .
we consider the general system of time scale equations associate to @xmath159 defined as @xmath204 for all @xmath65 .
it defines an operator @xmath205 as follows : @xmath85{lrcl}o^{\mathbb{t}}_x : & \cdni({\mathbb{t}},{\mathbb{r}}^{2d } ) & \longrightarrow & c(\tkk,{\mathbb{r}}^{2d})\\&(q , p ) & \longmapsto & \begin{pmatrix } q^\dd - x_q(q , p ) \\ \rho^\dd p^\nabla - x_p(q , p ) \end{pmatrix } \ . \end{array}}\ ] ] a straightforward computation leads to : let @xmath187 .
the frchet derivative @xmath206 is given by @xmath207 and its adjoint @xmath208 with respect to the @xmath166 symplectic scalar product is given by @xmath209 by identification we obtain : the operator @xmath205 has its frchet derivative self - adjoint at @xmath180 if and only if the following conditions are satisfied over @xmath210 : @xmath163 now we can state the main result of this paper : [ helmholtztimescale ] the vector field x is a time scale hamiltonian if and only if the operator @xmath205 associated has his frchet derivative self - adjoint with respect to the @xmath166 symplectic scalar product .
+ in this case the hamiltonian associate to @xmath164 is given by @xmath165d\lambda\ ] ] let @xmath187 .
the frchet derivative of @xmath178 at @xmath154 along @xmath188 is given by @xmath189\dd t.\ ] ] the time scale hamiltonian helmholtz conditions implies that the frchet derivative of @xmath190 at @xmath154 along @xmath188 is given by @xmath211 which leads to @xmath212 by definition , if @xmath154 is a critical point of @xmath197 then we have @xmath198 and then using integration by parts formula @xmath192 with the weak form of the dubois
reymond lemma concludes the proof .
we proved a result on first order time scale equations which allows us to find the existence of a hamiltonian structure associate and in the affirmative case to give the hamiltonian .
our result recover both the discrete and classical case but it allows the mixing of both of them as the time scale calculus was created with such motivation @xcite .
the hamiltonian helmholtz problem was easier to prove as contrary to the lagrangian case . indeed , there is no mixing of @xmath25 and @xmath63 derivative such as @xmath213 or @xmath214 which happen in the discrete case @xcite .
+ the further extension of the hamiltonian helmholtz problem is to consider the derivative as combinations of @xmath25 and @xmath63 such that @xmath215 which is the diamond integral for which motivations and definitions can be found in @xcite , @xcite and references therein . +
another further extension of this result concern the stochastic calculus and more precisely the stochastic calculus on time scales defined in @xcite and @xcite .
the work is to define a natural notion of stochastic hamiltonian on time scales and then to give the stochastic version of theorem [ helmholtztimescale ] .
this extension is a work in progress and will be the subject of a future paper . | we derive the helmholtz theorem for hamiltonian systems defined on time scales in the context of nonshifted calculus of variations which encompass the discrete and continuous case .
precisely , we give a theorem characterizing first order equation on time scales , admitting a hamiltonian formulation which is defined with non - shifted calculus of variation .
moreover , in the affirmative case , we give the associated hamiltonian . |
symmetric macdonald polynomials @xcite are a family of multivariable orthogonal polynomials indexed by partitions , whose coefficients depend rationally on two parameters @xmath2 and @xmath3 . in the case @xmath4 they degenerate to the celebrated schur polynomials , which are central in the representation theory of both the general linear and symmetric groups .
let @xmath5 denote the monomial symmetric polynomial indexed by a partition @xmath6 , i.e. the symmetric polynomial defined as the sum of all monomials @xmath7 where @xmath8 ranges over all distinct permutations of @xmath9 .
the macdonald polynomials are defined as follows : let @xmath10 denote the macdonald inner product on power sum symmetric functions ( @xcite , chapter vi , equation ( 1.5 ) ) , where @xmath11 denotes the dominance order on partitions ( @xcite , chapter i , section 1 ) .
the macdonald polynomial @xmath12 is the unique homogeneous symmetric polynomial in @xmath13 which satisfies @xmath14 i.e. the coefficients @xmath15 of the lower degree terms are completely determined by the orthogonality conditions . up to normalization
, macdonald polynomials can alternatively be defined as the unique eigenfunctions of certain linear difference operators acting on the space of all symmetric polynomials @xcite .
they can also be expressed combinatorially as multivariable generating functions @xcite , or via symmetrization of non - symmetric macdonald polynomials that are computed from yang baxter graphs @xcite .
the purpose of this article is to report on an explicit matrix product formula for macdonald polynomials @xcite inspired by recent results on the multi - species asymmetric exclusion process , and to provide an explicit sum rule for calculating macdonald polynomials resulting from the matrix product formula @xcite . in the following
we need the polynomial representations of the hecke algebra of type @xmath16 , with generators @xmath17 given by @xmath18 where @xmath19 is the transposition operator with action @xmath20 on functions in @xmath13 .
it can be verified that the operators indeed give a faithful representation of the hecke algebra : @xmath21 in view of the relations for the generators , we can define @xmath22 unambiguously as any product of simple transpositions @xmath17 which gives the permutation @xmath23 .
[ def : masep ] for all partitions @xmath6 , let @xmath24 be a set of homogeneous degree @xmath25 polynomials in @xmath26 variables , defined by @xmath27 as explained in @xcite , the polynomials @xmath28 are related to the non - symmetric macdonald polynomials @xcite via an invertible triangular change of basis , and hence form a basis for the ring of polynomials in @xmath26 variables
. specializations of these polynomials at @xmath29 give stationary particle configuration probabilities of the multi - species asymmetric exclusion process on a ring .
summing over all @xmath24 results in a symmetric macdonald polynomial @xcite : [ le : symmetry ] let @xmath6 be a partition .
then @xmath30 suppose we have ( semi - infinite ) matrices @xmath31 and @xmath32 satisfying the following exchange relations @xmath33 for all @xmath34 . the main result of @xcite is a matrix product formula for the polynomials @xmath35 : [ th : mp ] there is an explicit representation of @xmath36 and @xmath32 with @xmath37 satisfying , and such that @xmath38 can be written as a matrix product , i.e. @xmath39 , \label{mpa}\ ] ] where @xmath8 is a permutation of @xmath6 and @xmath40 is a normalization factor which only depends on the partition @xmath6 .
it follows from lemma [ le : symmetry ] that the symmetric macdonald polynomial @xmath0 can be expressed as a sum over matrix product formulas .
the specialization of @xmath0 at @xmath41 and @xmath42 @xmath43 is the normalization of the stationary state of a multi - species asymmetric exclusion process on a ring . as a consequence of theorem [ th : mp ]
we derive an explicit sum formula for macdonald polynomials @xcite .
to formulate this result we need to prepare some notation .
let @xmath6 be a partition whose largest part is @xmath44 . for all @xmath45 ,
we define a partition @xmath46 $ ] by replacing all parts in @xmath6 of size @xmath47 with 0 . for example , for @xmath48
we have @xmath49 = ( 3,3,2,1,1,0 ) , \quad \lambda[1 ] = ( 3,3,2,0,0,0 ) , \quad \lambda[2 ] = ( 3,3,0,0,0,0 ) , \quad \lambda[3 ] = ( 0,0,0,0,0,0),\end{aligned}\ ] ] and in general @xmath50 = \lambda$ ] , @xmath51 = 0 $ ] .
[ th : sum ] the macdonald polynomial @xmath1 can be written in the form @xmath52 } } c_i \left ( \begin{array}{@{}c@ { } } \lambda[i-1 ] \medskip \\ \sigma\circ\lambda[i ] \end{array } \right ) t_{\sigma } \circ x_{\lambda[i ] } \circ \right ) 1\end{aligned}\ ] ] with coefficients . ] that satisfy @xmath53 if any @xmath54 , and otherwise @xmath55 where @xmath56 and @xmath57 we point out that the formula given in has many structural features in common with the work of kirillov and noumi @xcite . in these papers
the authors construct families of raising operators , which act on macdonald polynomials by adding columns to the indexing young diagram . in @xcite
the raising operators have an analogous form to macdonald @xmath2-difference operators , while in @xcite the raising operators are constructed in terms of generators of the affine hecke algebra . in both papers
the macdonald polynomial is obtained by the successive action of such raising operators on @xmath58 , the initial state .
it would be very interesting to find a precise connection between the results of @xcite and our formula , if one exists .
[ [ two - variable - example . ] ] two variable example .
+ + + + + + + + + + + + + + + + + + + + + let us demonstrate in the case @xmath59 .
we have @xmath49 = ( 3,1),\ \lambda[1 ] = ( 3,0),\ \lambda[2 ] = ( 3,0),\ \lambda[3 ] = ( 0,0),\end{aligned}\ ] ] and @xmath60 } = s_{\lambda[1 ] } = s_{\lambda[2 ] } = s_2 $ ] . hence @xmath61_1 & \lambda[0]_2 \\
\lambda[1]_{\rho_1 } & \lambda[1]_{\rho_2 } \end{array}\right ) \times \\ t_{\rho } \circ x_1 \circ \sum_{\pi \in s_2 } c_2\left ( \begin{array}{@{}l@{}l@ { } } \lambda[1]_1 & \lambda[1]_2 \\
\lambda[2]_{\pi_1 } & \lambda[2]_{\pi_2 } \end{array}\right ) t_{\pi } \circ x_1.\end{gathered}\ ] ] we compute each sum in turn , starting with the rightmost : @xmath62_1 & \lambda[1]_2 \\
\lambda[2]_{\pi_1 } & \lambda[2]_{\pi_2 } \end{array}\right ) t_{\pi } \circ x_1 & = x_1 + c_2\left ( \begin{array}{@{}cc@ { } } 3 & 0 \\ 0 & 3 \end{array}\right ) t_{1 } \circ x_1 = x_1 + q \left ( \frac{1-t}{1-qt } \right ) x_2.\end{aligned}\ ] ] combining with the middle sum , we find that @xmath63_1 & \lambda[0]_2 \\ \lambda[1]_{\rho_1 } & \lambda[1]_{\rho_2 } \end{array}\right ) t_{\rho } \circ \left ( x_1 ^ 2 + q\frac{1-t}{1-qt } x_1 x_2 \right ) & = x_1 ^ 2 + q\frac{1-t}{1-qt } x_1 x_2,\end{aligned}\ ] ] where we have used the fact that @xmath64 and @xmath65 . combining everything and passing to the leftmost sum , we have @xmath66
in the remainder of this note we sketch the proof of theorem [ th : mp ] .
it is relatively simple to show that if the functions @xmath38 defined in definition [ def : masep ] , with @xmath8 a permutation of a partition @xmath6 , can be written as a matrix product formula as in theorem [ th : mp ] , then the matrices need to satisfy the exchange relations . these can be conveniently rewritten . as before ,
let @xmath37 be the largest part of @xmath6 , we call @xmath67 the rank . it was shown in @xcite for @xmath68 and general @xmath67 in @xcite , see also @xcite , that are equivalent to the zamolodchikov
faddeev ( zf ) algebra @xcite , ( x , y ) = , [ eq : zfdef ] where @xmath69 is a twisted version of the @xmath70 @xmath71-matrix and @xmath72 is an @xmath73-dimensional operator valued column vector given by @xmath74 let furthermore @xmath75 denote the elementary @xmath76 matrix with a single non - zero entry 1 at position @xmath77 .
then equation can be rewritten as @xmath78 where the rank @xmath67 is again implicit , i.e. @xmath72 and @xmath79 , and the @xmath71-matrix is explicitly given by @xmath80 [ [ example - of - a - rank-1-solution - to - zf - algebra . ] ] example of a rank 1 solution to zf algebra .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + we give a simple explicit example for the case @xmath68 .
using the following functions @xmath81 c^+ & = 1-b^+ = \frac{y ( t-1)}{tx - y } , & c^- & = 1-b^- = \frac{x ( t-1)}{tx - y}. \nonumber\end{aligned}\ ] ] equation for @xmath82 and @xmath68 explicitly becomes : @xmath83 = \left [ \left ( \begin{array}{c } 1 \\ y \end{array } \right ) \otimes \left ( \begin{array}{c } 1 \\ x \end{array } \right ) \right].\end{aligned}\ ] ] for general rank @xmath67 , solutions to are more difficult , but can be recovered from the yang
baxter algebra which is given by @xmath84 = \left [ l(y)\otimes l(x)\right ] \cdot \check{r}(x , y ) , \label{eq : ybadef}\end{aligned}\ ] ] where @xmath85 is an @xmath76 operator - valued matrix .
the algebra is well - studied and many solutions for @xmath86 are known . for the application to macdonald polynomials
the elements of @xmath86 are given in terms of generators @xmath87 of the @xmath3-boson algebra : @xmath88 we can construct solutions of by rank - reducing the yang
baxter algebra in the following way .
assume a solution of the following modified yang
baxter algebra @xmath89 = \left [ \tilde{l}(y)\otimes \tilde{l}(x)\right ] \cdot \check{r}^{(r-1)}(x , y ) , \label{eq : ybadef_lowrank}\end{aligned}\ ] ] in terms of an @xmath90 operator - valued matrix @xmath91 and where the rank of the @xmath71-matrix on the right hand side is one lower than that on the left hand side .
assume also an operator @xmath92 that satisfies @xmath93 then @xmath94 gives a solution to provided that the operator entries of @xmath95 commute with those of @xmath96 , for all @xmath97 .
the usual way to ensure this commutativity is to demand that the entries of @xmath98 act on some vector space @xmath99 while @xmath100 act on a different vector spaces @xmath101 , and indeed we shall adopt this approach . [ [ rank-1-example - continued . ] ] rank 1 example continued .
+ + + + + + + + + + + + + + + + + + + + + + + + + the solution to the yang
baxter algebra corresponding to rank @xmath68 is equal to @xmath102 where the operators @xmath103 , @xmath104 and @xmath105 satisfy the @xmath3-boson relations .
we note that trivialising the @xmath3-boson by sending @xmath106 and @xmath107 , we reduce the rank , and thus obtain the solution @xmath108 given in : @xmath109 [ [ rank-2-solution - to - zf - algebra . ] ] rank 2 solution to zf algebra .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the rank @xmath110 case gives rise to operator valued solutions for @xmath111 .
the associated rank 2 solution to the yang
baxter algebra is @xmath112 where @xmath113 and @xmath114 are two commuting copies of the @xmath3-boson algebra .
the map @xmath115 and @xmath116 reduces the rank of @xmath117 by one @xmath118 where the indices of @xmath3-bosons are redundant in the final matrix , since we no longer need to distinguish between the two copies of the algebra .
indeed , we find that is satisfied @xmath119 = \nonumber \\ \left[\left ( \begin{array}{cc } 1 & \phi \\ y k & 0\\ y \phi^\dag & y \end{array } \right)\otimes \left ( \begin{array}{cc } 1 & \phi \\ x k & 0 \\ x \phi^\dag & x \end{array } \right)\right]\cdot \left ( \begin{array}{cc|cc } 1 & 0 & 0 & 0 \\ 0 & c^- & b^+ & 0 \\ \hline 0 & b^- & c^+ & 0 \\ 0 & 0 & 0 & 1 \end{array } \right).\end{aligned}\ ] ] we thus construct a solution of the zf algebra in the following way : [ eq : arank2 ] ^(2)(x ) = a_0(x ) + a_1(x ) + a_2(x ) = ^(2)(x)^(1)(x ) = 1 & + x k & 0 + x ^ & x 1 + x = 1 + x + k x + x ^+x^2 , which for @xmath120 is the matrix product solution to the stationary state of the two - species asep . [
[ general - solution . ] ] general solution
. + + + + + + + + + + + + + + + + + for completeness we include the general solution which does not look illuminating in algebraic form . a natural and intuitive combinatorial description of this solution is given in @xcite but due to lack of space we are not able to present this here .
assume @xmath121 and introduce the following family of @xmath122 operator - valued matrices @xmath123 , @xmath124 .
we index rows by @xmath125 and columns by @xmath126 $ ] introduced in section [ sec : main ] . ] , and take @xmath127 this general rank solution for @xmath120 in terms of @xmath3-bosons was recently obtained in @xcite .
a generalisation of these results that includes a spectral parameter was found earlier in @xcite and independently in the case of super - algebras in @xcite .
we also introduce a twist operator @xmath128 where we perform the reparametrization @xmath129 .
the operators @xmath87 are generators of the @xmath3-boson algebra with subscripts used to denote commuting copies of the algebra .
note that all operators in @xmath123 implicitly also carry an index @xmath130 , as we will adopt the convention that operators in @xmath123 and @xmath131 ( as well as @xmath132 and @xmath133 ) commute for @xmath134 .
we look at an explicit example for rank 2 taking @xmath135 . in this case
@xmath136 corresponds to the nonsymmetric macdonald polynomial @xmath137 @xcite , which , using the notation @xmath129 , @xmath138=(1-t^{m})/(1-t)$ ] is given by @xmath139 } ( x_1+x_2)x_3x_4 x_5x_6(x_5+x_6 ) \\
+ \frac{t^{4}[2]}{[3+u][4+u]}x_1x_2x_3x_4x_5x_6 .
\label{eq : edeltar=2}\end{gathered}\ ] ] we now verify the matrix product form @xmath140,\ ] ] for this explicit solution . from we
see that @xmath141 and using we note that @xmath32 should satisfy @xmath142 an explicit representation for the @xmath3-bosons in terms of semi - infinite matrices is given by [ eq : t - bosrep ] ( ) _ m , m = ( 1-t^m ) _ m , m-1,(^)_m , m = _ m , m+1,(k)_m , m=t^m _ m , m , and @xmath32 has the form @xmath143 up to a normalization , the nonsymmetric macdonald polynomial @xmath137 is now represented in matrix product form by @xmath144 \nonumber\\ & = x_3x_4x_5x_6 \operatorname{tr}\left[\left ( x_5 x_6 k^2 + ( x_1+x_2)(x_5+x_6)\phi k^2 \phi^\dag + x_1x_2 \phi^2 k^2 ( \phi^\dag)^2 \right)s\right],\end{aligned}\ ] ] where other terms involving unequal powers of @xmath103 and @xmath145 have zero trace . normalising with @xmath146 we finally get @xmath147 which can be shown to equal using the explicit representation .
we have derived a new explicit formula for symmetric macdonald polynomials using the matrix product formalism .
our main results are the matrix product formula theorem [ th : mp ] for the distinguished basis @xmath148 in the ring of homogeneous polynomials of degree @xmath6 . in the specialisation @xmath41 and @xmath149
these give the ( unnormalized ) stationary probabilities for the multi - species asymmetric exclusion process .
the limits @xmath150 ( or @xmath151 ) give the stationary probablities for the totally asymmetric exclusion process , and a recent derivation of the matrix product formula for those was given in @xcite using the tetrahedron equation for three - dimensional integrability .
while there are several similarities with our approach , the resulting expressions for the matrices in terms of bosonic operators in @xcite are different from those exhibited here . by lemma [ le : symmetry ] the matrix product formula for the polynomials @xmath152 leads to the result for macdonald polynomials .
we mention here that the normalization factor @xmath153 in theorem [ th : mp ] can be calculated explicitly , @xmath154 where @xmath67 is the largest part of @xmath6 . as a nontrivial corollary we obtain a new summation formula for macdonald polynomials , presented in theorem [ th : sum ] .
we give explicit examples of several of our constructions .
[ sec : ack ] we thank the galileo galilei institute and the organisers of the research program _ statistical mechanics , integrability and combinatorics _ for kind hospitality during part of this work .
it is a pleasure to thank philippe di francesco for bringing the papers @xcite to our attention , and eric ragoucy , ole warnaar and paul zinn - justin for helpful remarks and extended discussions on related topics .
jdg and mw are generously supported by the australian research council ( arc ) and the arc centre of excellence for mathematical and statistical frontiers ( acems ) .
l. d. faddeev , _ quantum completely integrable models in field theory _ , in : problems of quantum field theory , r2 - 12462 , dubna ( 1979 ) , 249299 .
l. d. faddeev , _ quantum completely integrable models in field theory _ , in : contemporary mathematical physics , vol .
ic ( 1980 ) , 107155 .
a. n. kirillov and m. noumi , _ @xmath2-difference raising operators for macdonald polynomials and the integrality of transition coefficients _ , algebraic methods and @xmath2-special functions , crm proceedings and lecture notes , vol .
* 22 * ( 1999 ) ; arxiv : q - alg/9605005 . | [ [ abstract . ] ] abstract .
+ + + + + + + + + we present a new , explicit sum formula for symmetric macdonald polynomials @xmath0 and show that they can be written as a trace over a product of ( infinite dimensional ) matrices .
these matrices satisfy the zamolodchikov
faddeev ( zf ) algebra .
we construct solutions of the zf algebra from a rank - reduced version of the yang baxter algebra . as a corollary
, we find that the normalization of the stationary measure of the multi - species asymmetric exclusion process is a macdonald polynomial with all variables set equal to one .
[ [ rsum . ] ] rsum .
+ + + + + + + nous prsentons une nouvelle formule explicite pour les polynmes symtriques de macdonald @xmath1 , et dmontrons quils peuvent tre crits comme une trace dun produit de matrices de dimension infinie .
ces matrices forment une reprsentation de lalgbre de zamolodchikov faddeev ( zf ) .
nous laborons des solutions de lalgbre zf en utilisant une version de lalgbre yang baxter de rang rduit .
comme corollaire , nous constatons que la normalisation de la mesure stationnaire du processus dexclusion asymtrique est un polynme de macdonald avec toutes les variables rendues gale un . [ [ samenvatting . ] ] samenvatting .
+ + + + + + + + + + + + + we presenteren een nieuwe , expliciete som formule voor symmetrische macdonald polynomen @xmath0 en laten zien dat deze geschreven kunnen worden als een spoor over een produkt van ( oneindig dimensionale ) matrices .
deze matrices voldoen aan de zamolodchikov faddeev ( zf ) algebra .
we construeren oplossingen van de zf algebra uit een rank - gereduceerde versie van de yang - baxter algebra .
een gevolgtrekking is dat de normalisatie van de stationaire toestand van het asymmetrische exclusie process met meerdere soorten deeltjes gegeven wordt als een macdonald polynoom waarin alle variabelen gelijk aan n worden gezet . |
linear lepton colliders of the nearest future provide the possibility of investigation of photons collisions at energies and luminosities the same to those in @xmath1 collisions @xcite .
the photon linear colliders have the great physical potential ( higgs ans susy particles searching , study of anomalous gauge boson couplings and hadronic structure of photons etc . , @xcite ) .
and performing of this set of investigations requires the of photon beam calibration as well . for this purpose one
traditionally use some of the well - known and precisely calculated reactions ( see , for example , @xmath2 , @xcite ) .
the difficulties appear in the calibration of photon beams of similar helicity due to the small magnitude of cross sections of the most qed processes . for this purpose
we consider the exclusive reaction @xmath3 as the possible candidate for a calibration channel on a linear photon collider .
the two various helicity configuration of the @xmath4-system leads to the different spectra of final particles and requires the two mechanisms of beam calibration .
we have analysed the behaviour of the @xmath5 reaction on beams with different helicities as a function of the parameters of detectors .
we have performed the detail comparison of it s cross sections on @xmath6- ( total helicity @xmath7 ) and @xmath8-beams ( @xmath9 ) .
the question to be answered is does the process on `` @xmath10 '' beams can have the same cross section as one on `` @xmath11 '' beams .
we have outlined the conditions that greatly restrict the observation of the process on `` @xmath9 '' beams , remaining the `` @xmath7 '' cross section almost unchanged .
( 300,200)(-90,0 ) ( 10 , 20)(40 , 30)14 ( 10 , 80)(40 , 70)14 ( 90 , 20)(40 , 30 ) ( 40 , 30)(40 , 70 ) ( 40 , 70)(90 , 80 ) ( 60 , 74)(80 , 60)14 ( 15,30)[c]@xmath12 ( 15,70)[c]@xmath13 ( 80,52)[c]@xmath14 ( 85,30)[c]@xmath15 ( 85,72)[c]@xmath16 ( 110 , 20)(140 , 30)14 ( 110 , 80)(140 , 70)14 ( 190 , 20)(140 , 30 ) ( 140 , 30)(140 , 70 ) ( 140 , 70)(190 , 80 ) ( 140 , 50)(170 , 50)14 ( 115,30)[c]@xmath12 ( 115,70)[c]@xmath13 ( 180,52)[c]@xmath14 ( 185,30)[c]@xmath15 ( 185,72)[c]@xmath16 ( 210 , 20)(240 , 30)14 ( 210 , 80)(240 , 70)14 ( 290 , 20)(240 , 30 ) ( 240 , 30)(240 , 70 ) ( 240 , 70)(290 , 80 ) ( 260 , 26)(280 , 40)14 ( 215,30)[c]@xmath12 ( 215,70)[c]@xmath13 ( 280,48)[c]@xmath14 ( 285,30)[c]@xmath15 ( 285,72)[c]@xmath16 ( 0,100 ) ( 10 , 20)(40 , 30)14 ( 10 , 80)(40 , 70)14 ( 90 , 20)(40 , 30 ) ( 40 , 30)(40 , 70 ) ( 40 , 70)(90 , 80 ) ( 60 , 74)(80 , 60)14 ( 15,30)[c]@xmath13 ( 15,70)[c]@xmath12 ( 80,52)[c]@xmath14 ( 85,30)[c]@xmath15 ( 85,72)[c]@xmath16 ( 110 , 20)(140 , 30)14 ( 110 , 80)(140 , 70)14 ( 190 , 20)(140 , 30 ) ( 140 , 30)(140 , 70 ) ( 140 , 70)(190 , 80 ) ( 140 , 50)(170 , 50)14 ( 115,30)[c]@xmath13 ( 115,70)[c]@xmath12 ( 180,52)[c]@xmath14 ( 185,30)[c]@xmath15 ( 185,72)[c]@xmath16 ( 210 , 20)(240 , 30)14 ( 210 , 80)(240 , 70)14 ( 290 , 20)(240 , 30 ) ( 240 , 30)(240 , 70 ) ( 240 , 70)(290 , 80 ) ( 260 , 26)(280 , 40)14 ( 215,30)[c]@xmath13 ( 215,70)[c]@xmath12 ( 280,48)[c]@xmath14 ( 285,30)[c]@xmath15 ( 285,72)[c]@xmath16 we consider the process @xmath17 where @xmath18 and @xmath19 are photon and fermion helicities .
we use the following set of invariants : @xmath20 we denote the final - state photon energy by @xmath21 , the polar angle ( between any final and initial particles in c.m.s . ) by @xmath22 and the angle between any of final particles by @xmath23 . for the differential cross - section
we introduce the normalized final - state photon energy ( c.m.s .
is used ) @xmath24 .
the differential cross section @xmath25 appears to be the energy spectrum of final - state photons .
the total helicity of the @xmath4 system is denoted by @xmath26 .
we consider the cross section @xmath27 where the phase - space volume element is defined by @xmath28 the process @xmath29 is the pure qed reaction in the born approximation . using the method of helicity amplitudes @xcite
, we calculate matrix elements : @xmath30 all the other non - vanishing amplitudes are obtained from @xmath31 by using c , p , bose and crossing ( between final and initial particles ) symmetries : @xmath32 @xmath33 @xmath34 @xmath35 @xmath36 @xmath37 due to no final - state polarizations can be measured we summarize over all the final particles helicities .
the further integration is performed numerically using the monte - carlo method @xcite .
one of the main purposes of the linear photon collider is the @xmath38-channel of the higgs boson production at energies about @xmath39 @xcite .
hence we use this value of c.m.s .
energy in our analysis .
we have performed calculations for various experimental restrictions on the parameters of final particles .
the events are not detected if @xmath22 , @xmath23 or @xmath21 are below the threshold values .
the default minimal values are @xmath40 , @xmath41 , @xmath42 , the minimal fermion energy was also assumed to be @xmath43 .
first of all we discuss the energy spectrum of final photons . as it is presented in fig .
[ f1]-[f2 ] , the differential cross section ( @xmath44 ) on j=2 beams decreases while one on j=0 beams raises with increasing of the final - state photon energy .
this leads to the conclusion that if one raise the threshold on @xmath21 , the process on j=2 beams will be greatly restricted , but the rate of j=0 events remains almost unchanged . in fig .
[ f31 ] we present the total cross section dependence on the @xmath21-cut . according to fig .
[ f31]c , one can achive the ratio @xmath45 up to @xmath46 without sufficient decreasing of @xmath47 .
next we analyse the total cross section dependence on the angle cuts ( @xmath48 and @xmath49 , see figs . [ f32]-[f33 ] ) . in @xmath6- experiments ( @xmath7 )
most of final fermions are radiated closely to the axis of initial photon ( polar axis ) and ca nt be detected .
the @xmath8- experiments ( @xmath9 ) have the large fraction of particles emitted at large angles .
but they are likely emitted as the collinear fermion - photon pairs , that also ca nt be separated in the detector . in the figs .
[ f32]c and [ f33]c we present the ratio of the total cross sections @xmath45 at certain configuration of cuts .
we conclude that it can be increased by reducing the polar angle cut @xmath48 , and by raising the collinear angle cut @xmath49 . in fig .
[ f4 ] we demonstrate the total cross section dependence on the c.m.s . energy .
it is close to the law @xmath50 due to all the invariants in calculations are proportional to @xmath38 .
the only non - linear effect is the @xmath21-cut .
we discover that the ratio of events on j=0 and j=2 beams strongly depends on the experimental cuts .
we obtained the region ( the configuration of cuts ) where the processes on the both j=0 and j=2 beams have the close cross sections .
that is the region of small polar angle cut , high collinear angle cut and high minimal energy of final - state photons . at these parameters
the total cross sections of @xmath29 in experiments using @xmath6- and @xmath8- beams appear to be the same order of magnitude . | the exclusive reaction @xmath0 is considered as the possible calibration process on the linear photon collider .
we analyse the energy spectrum and total cross section dependence on the detector parameters at various initial beam helicities .
= -30 mm = `` 117 = ' ' 11d = `` 245 = ' ' 250 1z 1t 1u 1s 2m^2 12 # 1#2 # 1 ( # 1 ) # 1#2_0^ # 2 d # 1#2_2 m ^ # 2 d # 1(#1 ) # 1 # 1#2 # 1 # 2 # 1 |
in 1948 , casimir @xcite predicted the existence of an attractive force between two perfectly conducting plates as a manifestation of the zero - point vacuum fluctuations of quantum fields .
although experimental verification of this prediction has little progress before 1997 , casimir effect has attracted more and more attention since 1970s @xcite .
it was gradually realized that this effect is closely related to other areas of physics such as quantum field theory , cosmology and condensed matter physics .
more recently , the advances in the casimir measurements and nanotechnology have stimulated more interest in casimir effect .
historically , the casimir effect was investigated for electromagnetic field ( massless vector field ) . in 1956 , lifshitz @xcite derived a formula to express the casimir force acting between two dielectric semi - infinite slabs in terms of the permittivity of the slabs @xcite .
refinements and generalizations of the lifshitz formula where the semi - infinite dielectric slabs are replaced by dielectric or magnetodielectric plates of finite thicknesses were derived in @xcite . in the limits the permittivity goes to infinity ,
lifshitz formula reproduces the result of casimir for two perfectly conducting plates . since 1970s ,
casimir effect was studied for other quantum fields such as scalar fields and spinor fields , for both massless and massive sectors .
in contrast , there are very few works that studied casimir effect of massive vector fields .
one of the incentives to study casimir effect of massive fields comes from the prevalence of spacetime with extra dimensions which were proposed to solve some fundamental problems in physics . to study a quantum field in a spacetime with extra dimensions , one can use kaluza - klein decomposition to decompose the quantum field to an infinite tower of massive fields in @xmath3 .
this approach has been intrinsically used in a number of works to study the casimir effect of scalar field or spinor field in spacetime with extra dimensions such as kaluza - klein spacetime and randall - sundrum spacetime . for electromagnetic field , the casimir effect on a pair of perfectly conducting plates in the kaluza - klein spacetime @xmath4 was studied in @xcite using the results of barton and dombey @xcite on casimir effect of massive photons . the pioneering work of barton and dombey @xcite considered the casimir effect on a pair of perfectly conducting plates due to a massive vector field . unlike the scalar and spinor case , the extension from massless vector field to massive vector field is in fact much more complicated .
the classical maxwell s equations have to be replaced by proca equations @xcite and the gauge degree of freedom is lost .
therefore , a massless vector field has two polarizations but a massive vector field has three . for a pair of infinitely large perfectly conducting plates , the contribution to the casimir energy from the two polarizations are the same .
however , the contributions to the casimir energy from the three polarizations are not the same .
another complication in generalizing casimir effect of massless vector field to massive vector field is that besides the conventional boundary conditions which require the continuities of @xmath5 and @xmath6 across interfaces , one also need to impose the continuities of the scalar potential @xmath7 and the vector potential @xmath8 @xcite . for massless vector field where we usually impose the gauge conditions
@xmath9 and @xmath10 , @xmath11 is not continuous for the family of tm modes .
this gives an obstacle to generalizing the results from the massless case to the massive case . in @xcite ,
barton and dombey showed that the eigenmodes of a massive vector field can be divided into two families of discrete modes and one family of continuum modes . in the massless limit
, the contribution to the casimir energy from the discrete modes yields the result of casimir @xcite .
the contribution to the casimir energy from the continuum modes has zero massless limit .
one thing that worth remarked is that one family of the discrete modes corresponds to the te modes in the massless case , but the other family of discrete modes is not a direct generalization of the tm modes in the massless case . as a first step to investigate the casimir effect on a pair of plates made of real materials in a spacetime with extra dimensions ,
we generalize the work of barton and dombey @xcite to a pair of plates made of real materials in some magnetodielectric background medium .
this can also be considered as generalizing the lifshitz formula for massless vector field to a formula for massive vector field . as explained above
, the generalizing of the casimir effect from massless vector field to massive vector field is nontrivial even for a pair of perfectly conducting plates .
therefore , one should expect that the result will even be more complicated when one considers real materials . as was discussed in @xcite , for dielectric plates ,
the plane waves of a massive vector field can be divided into two families of transverse waves and one family of longitudinal waves . however ,
one of the families of discrete modes for perfectly conducting plates is in fact not transverse , and the continuum modes for perfectly conducting plates is in fact not longitudinal .
therefore , it is not obvious how one should generalize the work of barton and dombey @xcite .
in fact , one can show that for plates made of real materials in a magnetodielectric background , only one family of the transverse modes can satisfy all the boundary conditions . as a result , to look for other eigenmodes of the massive vector field ,
we are forced to consider the superposition of the other family of transverse modes and the family of longitudinal modes . in @xcite
, lifshitz modeled two dielectric plates as two semi - infinite slabs which is a reasonable model when the skin depths of the materials of the plates are much smaller than the thicknesses of the plates .
the configuration of two semi - infinite slabs separated by a medium is a three - layer model .
however , since we want to consider plates made of any materials , we find it necessary not to make any assumption about the skin depths .
therefore , we would consider two plates with finite thicknesses which can be modeled by a five - layer configuration .
the layout of this article is as follows . in section [ s2 ]
, we review the proca equations for massive vector fields . in section [ s3 ]
, we divide the plane waves solutions to the proca equations into type i and type ii transverse waves , and longitudinal waves . in section [ s4 ] , we discuss the boundary conditions that should be satisfied by the potentials and fields on the interface of two plane parallel media . in section [ s5 ]
, we derive the casimir energies contributed by transverse modes of type i , and by the superposition of transverse modes of type ii and longitudinal modes .
we show that their massless limits are the te and tm contributions to the casimir energy due to a massless vector field . in section [ s6 ] ,
we discuss the limiting case where the plates become perfectly conducting . in section [ s7 ]
, we consider the limit when the plates become infinitely permeable . in section [ s8 ] , we study the case where one plate is perfectly conducting and one plate is infinitely permeable .
a massive vector field is represented by a four - vector @xmath12 , where @xmath7 is the scalar potential and @xmath13 is the vector potential . define the electric field @xmath14 and the magnetic field @xmath15 by @xmath16 consider a magnetodielectric medium with permittivity @xmath17 and permeability @xmath18 . let @xmath19 and @xmath20 be the free charge and free current in the medium . assume the usual linear relations @xmath21 and @xmath22 .
then the proca equations for the electromagnetic waves propagating in this medium are @xcite : @xmath23the first two are the well - known maxwell s equations which are automatically satisfied because of .
the conservation of free charges implies the continuity equation : @xmath24differentiate with respect to @xmath25 and apply the divergence operator to , implies that the lorentz condition @xmath26has to be satisfied . from and
, one can then derive from and the following two equations for @xmath7 and @xmath8 : @xmath27 , and are the equivalences of the proca equations for the potentials @xmath7 and @xmath8 . for a massless vector field ,
the primary quantities in the maxwell s equations are the electric and magnetic fields .
there is a gauge degree of freedom given by @xmath28for an arbitrary function @xmath29 .
the lorentz gauge is one of the gauge conditions that can be used to fix the gauge .
this gauge degree of freedom is lost in the massive case .
in fact , for massive vector field , the primary quantities are the potentials @xmath7 and @xmath8 , and the electric and magnetic fields are derived quantities .
the lorentz condition is a necessary condition followed from the conservation of charges .
consider an unbounded magnetodielectric medium with permittivity @xmath30 and permeability @xmath18 in the absence of free chargers and currents , i.e. , @xmath31 and @xmath32 . as discussed in @xcite ,
the monochromatic plane wave solutions of the proca equations can be divided into transverse waves with @xmath10 and longitudinal waves with @xmath33 . for the transverse waves
, it follows from lorentz condition that @xmath9 .
the equation for the vector potential @xmath8 becomes @xmath34 recall that for a massless vector field , the vector potential @xmath8 satisfies the equation with @xmath35 , and the conditions @xmath9 and @xmath36 , called coulomb gauge or radiation gauge , are usually imposed to remove the gauge degree of freedom .
therefore the transverse waves are in one - to - one correspondence with the massless electromagnetic waves .
since massless electromagnetic waves are usually divided into te and tm polarizations , we do the same for the massive case , but call them type i and type ii transverse waves . in the following , let @xmath37 , @xmath38 and define @xmath39 .
* a. type i transverse waves . *
the type i transverse waves are the equivalence of the te waves in the massless case : @xmath40 in this case , @xmath41 . the dispersion relation is : @xmath42
* b. type ii transverse waves . *
the type ii transverse waves are the equivalence of the tm waves in the massless case : @xmath43 in this case , @xmath44 .
the dispersion relation is : @xmath45 * c. longitudinal waves . *
the longitudinal waves are @xmath46 the dispersion relation is : @xmath47for longitudinal waves , @xmath33 implies that the magnetic field vanishes identically , i.e. , @xmath48 . on the other hand , the electric field
is given by @xmath49notice that the electric field also vanishes in the massless @xmath50 limit .
it should be observed that the longitudinal waves and transverse waves satisfy different dispersion relations , unless @xmath51 , i.e. , the waves traveled at the speed of light in the medium .
in this section , we consider the boundary conditions that must be satisfied by the potentials @xmath8 and @xmath7 and the fields @xmath14 and @xmath15 on the interface of two plane parallel media in the absence of free charges and currents .
let @xmath2 be the direction normal to the boundary of the two media .
as in the massless case , and imply that @xmath52 and @xmath53 are continuous . taking the derivative of with respect to @xmath25 gives @xmath54the lorentz condition then imply that @xmath55therefore , we find that @xmath56must be continuous . then implies that @xmath57 has to be continuous .
notice that the boundary condition that one obtains from is the continuity of @xmath58 but not the continuity of @xmath59 as in the massless case . as discussed in @xcite ,
the boundary conditions above are not enough for massive vector fields .
it was pointed out by kroll @xcite that the potentials @xmath7 and @xmath8 also have to be continuous .
therefore , the complete set of boundary conditions is the continuities of @xmath60 .
however , not all these conditions are independent . for example
, the continuities of @xmath7 and @xmath61 imply the continuity of@xmath62the continuities of @xmath63 imply the continuity of @xmath64 since gives @xmath65similarly , implies that if @xmath66 is continuous , then so is @xmath67 . on the other hand , the continuities of @xmath61 and the lorentz condition @xmath68
imply that @xmath7 is continuous if and only if @xmath69 is continuous . in conclusion , for the boundary conditions ,
it is sufficient to impose the continuities of @xmath70 and the continuity of either @xmath7 _ or _ @xmath71 .
= 0.4 in this article , we are interested in computing the casimir energy of two parallel magnetodielectric plates inside a magnetodielectric medium ( see fig . [ f2 ] ) .
we assume that the cross section of the plates are infinite . for convenience ,
we first consider a five - layer model consists of five plane - parallel layers of magnetodielectric media as shown in fig .
[ f1 ] .
= 0.4 at the end , we only consider the case where @xmath72 and @xmath73 .
the interfaces of the media are located at @xmath74 and @xmath75 , with @xmath76 .
we assume that there are two artificial boundaries at @xmath77 and @xmath78 , and let @xmath79 , @xmath80 . at the end
, we are going to let @xmath81 .
we also set @xmath82 and @xmath83 to be the thicknesses of the left and right plates that we are interested in , and set @xmath84 to be the separation between the plates .
for the casimir energy of massless electromagnetic field in the five layer model , one can consider the contribution from te modes and tm modes separately .
however , it turns out that for the massive case , one _ can not _ consider the contributions from the three types of waves separately .
in fact , the contribution from transverse modes of type i can be considered independently , but _ the contributions from the transverse modes of type ii and the longitudinal modes have to be considered _ _ together_. it is impossible to find type ii transverse modes or longitudinal modes that satisfy all the boundary conditions
. for transverse modes of type i , assume that @xmath85 the dispersion relation implies that @xmath86 the electric field @xmath14 and the magnetic field @xmath15 are given respectively by @xmath87for @xmath88 . in this case , @xmath89 are automatically continuous .
the continuities of @xmath90 imply that for @xmath91 , @xmath92the continuities of @xmath93 imply that for @xmath91 , @xmath94 = \frac{p_{t , i+1}}{\mu_{i+1}}\left[\mathfrak{a}_{i+1}e^{ip_{t , i+1}a_i}-\mathfrak{b}_{i+1}e^{-ip_{t , i+1}a_i}\right ] .
\end{split}\]]next , we have to impose some boundary conditions on the artificial boundaries @xmath95 and @xmath96 . as in the massless case
, we impose the conditions @xmath97 and @xmath44 , which gives @xmath98 from here , one can use the method we developed in our previous work @xcite . after a renormalization by the criterion that the casimir energy should vanish in the limit the plate separation @xmath99 is infinite , and passing to the limits @xmath100 ,
one finds that the te contribution to the casimir energy is given by @xmath101where @xmath102 ^ 2e^{-q_{t , l}t_l } \right ) \left(1-\left[\delta_{r}^{\text{i}}\right]^2e^{-q_{t , r}t_r}\right)}.\ ] ] here the variables have been renamed by @xmath103 , @xmath104 , @xmath105 , @xmath106 ; and @xmath107 the te contribution to the casimir force is given by @xmath108^{-1}e^{2q_{t , b}(\xi , k_\perp)a}-1\right)^{-1 } k_{\perp}dk_{\perp}d\xi.\ ] ] obviously , in the massless @xmath50 limit , one obtains the result for the massless case , where the mass @xmath109 in @xmath110 in is set to zero . for the superposition of transverse modes of type ii and longitudinal modes , we have for @xmath88 , @xmath111 f_{\mathbf{k}_{\perp},\omega}(y , z , t)\\ a_y=&\left [ - k_2 ( \mathfrak{c}_ie^{ip_{t , i}x}+\mathfrak{d}_ie^{-ip_{t , i}x})+ k_2(\mathfrak{e}_ie^{ip_{l , i}x}+\mathfrak{f}_ie^{-ip_{l , i}x})\right]f_{\mathbf{k}_{\perp},\omega}(y , z , t)\\ a_z=&\left [ - k_3 ( \mathfrak{c}_ie^{ip_{t , i}x}+\mathfrak{d}_ie^{-ip_{t , i}x } ) + k_3(\mathfrak{e}_ie^{ip_{l , i}x}+\mathfrak{f}_ie^{-ip_{l , i}x})\right]f_{\mathbf{k}_{\perp},\omega}(y , z , t)\\ \phi=&\frac{c^2 } { \omega}(p_{l , i}^2+k_{\perp}^2)(\mathfrak{e}_ie^{ip_{l , i}x}+\mathfrak{f}_ie^{-ip_{l , i}x})f_{\mathbf{k}_{\perp},\omega}(y , z , t ) \end{aligned}\right .. \end{split}\ ] ] the electric field @xmath14 and the magnetic field @xmath15 are given respectively by @xmath112f_{\mathbf{k}_{\perp},\omega}(y , z , t)\\ e_y=&i\left [ -\omega k_2 ( \mathfrak{c}e^{ip_{t , i}x}+\mathfrak{d}e^{-ip_{t , i}x } ) + \frac{k_2 m^2c^2}{{\varepsilon}_i\mu_i\omega \hbar^2 } ( \mathfrak{e}_ie^{ip_{l , i}x}+\mathfrak{f}_ie^{-ip_{l , i}x})\right ] f_{\mathbf{k}_{\perp},\omega}(y , z , t)\\ e_z=&i\left [ - \omega k_3 ( \mathfrak{c}e^{ip_{t , i}x}+\mathfrak{d}e^{-ip_{t , i}x } ) + \frac{k_3 m^2c^2}{{\varepsilon}_i\mu_i\omega \hbar^2 } ( \mathfrak{e}_ie^{ip_{l , i}x}+\mathfrak{f}_ie^{-ip_{l , i}x})\right ] f_{\mathbf{k}_{\perp},\omega}(y , z , t ) \end{aligned}\right.\end{aligned}\ ] ] @xmath113 since @xmath44 , we can call these modes tm modes .
the continuities of @xmath114 imply that for @xmath91 , @xmath115 for the artificial boundaries at @xmath95 and @xmath96 , one can impose the conditions @xmath116 .
these imply that @xmath117 as in the case of te modes , one can then show that the tm contribution to the casimir energy is given by @xmath118 with the notations @xmath119-\mu_j(i\xi)q_{t , j}(\xi , k_{\perp})[q_{t , b}(\xi , k_{\perp})^2-k_{\perp}^2 ] } { \mu_b(i\xi)q_{t , b}(\xi , k_{\perp})[q_{t , j}(\xi , k_{\perp})^2-k_{\perp}^2]+\mu_j(i\xi)q_{t , j}(\xi , k_{\perp})[q_{t , b}(\xi , k_{\perp})^2-k_{\perp}^2]},\quad j = l , r;\end{split}\]]@xmath120-q_{l , b}(\xi , k_{\perp})[q_{l , j}(\xi , k_{\perp})^2-k_{\perp}^2 ] } { q_{l , j}(\xi , k_{\perp})[q_{l , b}(\xi , k_{\perp})^2-k_{\perp}^2]+q_{l , b}(\xi , k_{\perp})[q_{l , j}(\xi , k_{\perp})^2-k_{\perp}^2]},\quad j = l , r;\\ \alpha_{j_1j_2}(\xi , k_{\perp})= & \frac{q_{l , j_1}(\xi , k_{\perp})^2-q_{l , j_2}(\xi , k_{\perp})^2}{q_{l , j_2}(\xi , k_{\perp})^2-k_{\perp}^2},\quad j_1,j_2=b , l , r;\\ \beta_{j_1j_2}(\xi , k_{\perp})= & -\frac{k_{\perp}^2}{q_{t , j_1}(\xi , k_{\perp})q_{l , j_2}(\xi , k_{\perp})}\left(1-\frac{\mu_{j_2}(i\xi)}{\mu_{j_1}(i\xi)}\frac{[q_{t , j_1}(\xi , k_{\perp})^2 -k_{\perp}^2]}{[q_{t , j_2}(\xi , k_{\perp})^2-k_{\perp}^2]}\right),\quad j_1,j_2=b , l , r;\\ r_{j_1j_2}^+(\xi , k_{\perp})= & 1+\frac{q_{t , j_2}(\xi , k_{\perp})\mu_{j_2}(i\xi)}{q_{t , j_1}(\xi , k_{\perp})\mu_{j_1}(i\xi)}\frac{[q_{t , j_1 } ( \xi , k_{\perp})^2-k_{\perp}^2 ] } { [ q_{t , j_2}(\xi , k_{\perp})^2-k_{\perp}^2]},\quad j_1,j_2=b ,
l , r;\\ \kappa_{j_1j_2}^+(\xi , k_{\perp})= & \frac{q_{l , j_1}(\xi , k_{\perp})^2-k_{\perp}^2}{q_{l , j_2}(\xi , k_{\perp})^2-k_{\perp}^2}+\frac{q_{l , j_1}(\xi , k_{\perp})}{q_{l , j_2}(\xi , k_{\perp } ) } , \quad j_1,j_2=b , l , r ; \end{split}\ ] ] @xmath121 is a @xmath122 matrix with components given by @xmath123 e^{-q_{t , b}a},\\ q_{12}= & \left[-\frac{\alpha_{bl}}{\kappa_{bl}^+}\frac{r_{lb}^+}{\kappa_{lb}^+ } \left(\delta_{l}^{\text{ii } } -e^{-2q_{t , l}t_l } \right)e^{q_{t , l}t_l } + \frac{\alpha_{lb}}{\kappa_{lb}^+ } \left(1+\delta_{l}^{\text{iii}}e^{-2q_{l , l}t_l } \right)e^{q_{l , l}t_l}\right ] e^{-q_{t , b}a},\\ q_{13}= & \left(1 -\left[\delta_{r}^{\text{ii}}\right]^2e^{-2q_{t , r}t_r}\right)e^ { q_{t , r}t_r } + \frac{\alpha_{rb}\beta_{br } } { r_{rb}^+r_{br}^+}\left(1-e^{-2q_{l , r}t_r}\right)e^{q_{l , r}t_r } , \\ q_{14}= & \frac{\alpha_{br}}{\kappa_{br}^+}\frac{r_{rb}^+}{\kappa_{rb}^+ } \left ( 1-\delta_{r}^{\text{ii } } e^{-2q_{t , r}t_r}\right)e^{q_{t , r}t_r } + \frac{\alpha_{rb}}{\kappa_{rb}^+}\left(1+\delta_{r}^{\text{iii}}e^{-2q_{l , r}t_r}\right ) e^{q_{l , r}t_r } , \\ q_{21}= & \left(1-\left[\delta_{l}^{\text{ii}}\right]^2e^{-2q_{t , l}t_l } \right ) e^{q_{t , l}t_l } + \frac{\alpha_{lb}}{r_{lb}^+}\frac{\beta_{bl}}{r_{bl}^+}\left(1-e^{-2q_{l , l}t_l } \right)e^{q_{l , l}t_l } , \\ q_{22}= & \frac { \alpha_{bl}}{\kappa_{bl}^+}\frac{r_{lb}^+}{\kappa_{lb}^+ } \left(1-\delta_{l}^{\text{ii } } e^{-2q_{t , l}t_l } \right)e^{q_{t , l}t_l } + \frac{\alpha_{lb } } { \kappa_{lb}^+}\left(1+\delta_{l}^{\text{iii}}e^{-2q_{l , l}t_l } \right ) e^{q_{l , l}t_l } , \\ q_{23}=&\left [ -\delta_r^{\text{ii}}\left(1-e^{-2q_{t , r}t_r}\right ) e^{q_{t , r}t_r } + \frac{\alpha_{rb}\beta_{br}}{r_{rb}^+r_{br}^+}\left(1-e^{-2q_{l , r}t_r}\right ) e^{q_{l , r}t_r } \right]e^{-q_{t , b}a},\\ q_{24}= & \left[-\frac{\alpha_{br}}{\kappa_{br}^+}\frac{r_{rb}^+}{\kappa_{rb}^+ } \left ( \delta_{r}^{\text{ii}}- e^{-2q_{t , r}t_r}\right)e^{q_{t , r}t_r } + \frac{\alpha_{rb}}{\kappa_{rb}^+}\left(1+\delta_{r}^{\text{iii}}e^{-2q_{l , r}t_r}\right)e^{q_{l , r}t_r } \right ] e^{-q_{t , b}a},\\ q_{31}= & \left[- \frac{\beta_{lb}}{r_{lb}^+}\left(1-\delta_{l}^{\text{ii } } e^{-2q_{t , l}t_l } \right ) e^{q_{t , l}t_l } -\frac{\beta_{bl}}{r_{bl}^+}\frac{\kappa_{lb}^+}{r_{lb}^+ } \left(\delta_{l}^{\text{iii } } + e^{-2q_{l , l}t_l } \right)e^{q_{l , l}t_l } \right ] e^{-q_{l , b}a},\\ q_{32}= & \left[-\frac { \alpha_{bl}\beta_{lb}}{\kappa_{bl}^+\kappa_{lb}^+}\left(1-e^{-2q_{t , l}t_l}\right)e^{q_{t , l}t_l } -\delta_{l}^{\text{iii}}\left ( 1- e^{-2q_{l , l}t_l } \right ) e^{q_{l , l}t_l}\right ] e^{-q_{l , b}a},\\ q_{33}= & \frac{\beta_{rb}}{r_{rb}^+}\left(1 -\delta_{r}^{\text{ii } } e^{-2q_{t , r}t_r}\right ) e^{q_{t , r}t_r } + \frac{\beta_{br}}{r_{br}^+}\frac{\kappa_{rb}^+}{r_{rb}^+}\left(1+\delta_{r}^{\text{iii}}e^{-2q_{l , r}t_r}\right)e^{q_{l , r}t_r } , \\ q_{34}= & \frac{\beta_{rb}\alpha_{br}}{\kappa_{rb}^+\kappa_{br}^+}\left ( 1- e^{-2q_{t , r}t_r } \right)e^{q_{t , r}t_r } + \left(1-\left[\delta_{r}^{\text{iii}}\right]^2e^{-2q_{l , r}t_r } \right)e^{q_{l , r}t_r},\\ q_{41}= & \frac{\beta_{lb}}{r_{lb}^+}\left(1-\delta_{l}^{\text{ii } } e^ { -2q_{t , l}t_l } \right ) e^{q_{t , l}t_l } + \frac{\beta_{bl}}{r_{bl}^+}\frac{\kappa_{lb}^+}{r_{lb}^+}\left(1+\delta_{l}^{\text{iii } } e^{-2q_{l , l}t_l } \right)e^{q_{l , l}t_l } , \\ q_{42}= & \frac{\alpha_{bl}\beta_{lb}}{\kappa_{bl}^+\kappa_{lb}^+}\left(1-e^{-2q_{t , l}t_l}\right)e^{q_{t , l}t_l } + \left(1-\left[\delta_{l}^{\text{iii}}\right]^2e^{-2q_{l , l}t_l } \right)e^{q_{l , l}t_l } , \\
q_{43}= & \left[-\frac{\beta_{rb}}{r_{rb}^+}\left(1 -\delta_{r}^{\text{ii } } e^{-2q_{t , r}t_r}\right ) e^{q_{t , r}t_r } -\frac{\beta_{br}}{r_{br}^+}\frac{\kappa_{rb}^+}{r_{rb}^+}\left(\delta_{r}^{\text{iii}}+e^{-2q_{l , r}t_r}\right)e^{q_{l , r}t_r } \right ] e^{-q_{l , b}a},\\ q_{44}= & \left[-\frac{\beta_{rb}\alpha_{br}}{\kappa_{rb}^+\kappa_{br}^+}\left ( 1 - e^{-2q_{t , r}t_r } \right)e^{q_{t , r}t_r } -\delta_{r}^{\text{iii } } \left(1-e^{-2q_{l , r}t_r}\right ) e^{q_{l , r}t_r}\right ] e^{-q_{l , b}a } ; \end{split}\ ] ] and @xmath124is obtained by taking the limit @xmath125 of @xmath121 . in general , @xmath126 is a very complicated function of the distance @xmath127 between the plates . for the casimir force acting between the plates , let @xmath128 be the minor matrix of @xmath121 obtained by deleting the @xmath129-row and @xmath130-column from @xmath121 .
then the tm contribution to the casimir force acting between the plates is given by @xmath131 in the massless @xmath50 limit , @xmath132 .
therefore , @xmath133 for all @xmath134 , and @xmath135 .
the matrices @xmath121 and @xmath136 reduce to matrices of the form @xmath137consequently , @xmath138 ^ 2e^{-2q_{t , l}t_l } \right)\left(1 -\left[\delta_{r}^{\text{ii}}\right]^2e^{-2q_{t , r}t_r}\right ) } e^{-2q_{t , b}a } , \end{split}\ ] ] where the massless limits of @xmath139 are @xmath140
one finds that the massless limit of the contribution to the casimir energy from the superposition of the transverse modes of type ii and longitudinal modes is precisely the contribution from the tm modes in the massless case .
this is another justification that we call the contribution from the superposition of the transverse modes of type ii and longitudinal modes as tm contribution .
it is interesting to note that in the massless limit , the two polarizations for the massive field naturally reduce to only one polarization for the massless field .
in this section , we study the limits of the casimir energy and casimir force when the plates become perfectly conducting .
it is well - known that this can be obtained by the limit @xmath141 . in this limit
, we find that @xmath142 @xmath143@xmath144 @xmath145}:=\lambda.\ ] ] for the te contribution , it is easy to see from that @xmath146 , and thus the te contribution to the casimir energy reduces to @xmath147 the te contribution to the casimir force is then given by @xmath148which is always attractive .
it is interesting to note that in the perfect conductor limit , the te contribution to the casimir force is independent of the thicknesses of the plates .
the tm contribution is much more complicated . with the help of computer
, one can show that in the perfect conductor limit , the tm contribution to the casimir energy is given by @xmath149where @xmath150 ^
2-[\delta+\lambda]^2e^{-2q_0t_l}\right)\left([1-\lambda]^2-[\delta+\lambda]^2e^{-2q_0t_r}\right),\]]and @xmath151-[\delta^2-\lambda^2]e^{-2q_0t_l}\right)\left([1-\lambda^2]-[\delta^2-\lambda^2]e^{-2q_0t_r}\right)e^{-2q_{t , b}a}\\ & -\left([1-\lambda][\delta-\lambda]-[1+\lambda][\delta+\lambda]e^{-2q_0t_l}\right)\left([1-\lambda][\delta-\lambda]-[1+\lambda][\delta+\lambda]e^{-2q_0t_r}\right ) e^{-2q_{l , b}a}\\ & + 4\lambda(1-\delta)\left([1-\lambda]+[\delta+\lambda]e^{-2q_0t_l}\right)\left([1-\lambda]+[\delta+\lambda]e^{-2q_0t_r}\right)e^{-q_{t , b}a}e^{-q_{l , b}a}\\ & + ( 1-\lambda)^2(\delta+\lambda)^2\left(1-e^{-2q_0t_l}\right)\left(1-e^{-2q_0t_r}\right)e^{-2q_{t , b}a}e^{-2q_{l , b}a}. \end{split}\]]for an arbitrary background medium , one can not split the casimir energy into contributions from two different polarizations . in the special case when the background material has refractive index @xmath152 equal to one ,
e.g. , if the perfectly conducting plates is placed in vacuum , then @xmath153 and @xmath154 simplifies drastically to @xmath155 ^ 2-[\delta+\lambda]^2e^{-2q_0t_l}\right)\left([1-\lambda]^2-[\delta+\lambda]^2e^{-2q_0t_r}\right)}e^{-2q_ma}\right)\\=&\left(1-e^{-2q_{m}a}\right ) \left(1-\frac{d^2\left(1-e^{-2q_0t_l}\right)\left(1-e^{-2q_0t_r}\right)}{\left(1-d^2e^{-2q_0t_l}\right)\left(1-d^2e^{-2q_0t_r}\right)}e^{-2q_ma}\right ) , \end{split}\]]where @xmath156 in this case , the tm contribution to the casimir energy can be written as a sum of two terms @xmath157 and @xmath158 .
@xmath157 is the same as the te contribution , obtained from by setting @xmath159 , and it corresponds to the massless tm contribution .
the corresponding casimir force is given by with @xmath159 , which is always attractive .
the contribution to the casimir energy from the second tm polarization is given by @xmath160 some simple computations give @xmath161 the contribution to the casimir force from the second tm polarization is then given by @xmath162since @xmath163 , @xmath164therefore the sign of @xmath165 is always negative , i.e. , it is an attractive force . comparing to , we find that the second tm contribution can be identified as the te contribution to the casimir energy of a pair of dielectric plates due to a _ massless _ electromagnetic field , where the permittivity of the dielectric plates is @xmath166this fact was observed in @xcite . in particular , in the limiting case of perfectly conducting plates in vacuum , our general formula yields the result that was obtained by barton and dombey @xcite . in order to gain more insight how the two tm polarizations split out in this special case , as was described in @xcite ,
let us return to the expression for the tm modes . in the case of perfectly conducting plates in vacuum , @xmath167 for @xmath168 and @xmath169 .
let @xmath170 for @xmath171 .
notice that @xmath172 the first boundary condition in then implies that @xmath173 . since @xmath174 , and
then imply that both the electric field and the magnetic field vanish in the plates . in the vacuum , @xmath168 ,
the potentials are given by @xmath175 f_{\mathbf{k}_{\perp},\omega}(y , z , t)\\ a_y=&- k_2 \left [ \left(\mathfrak{c}_i-\mathfrak{e}_i\right)e^{ip x}+\left(\mathfrak{d}_i-\mathfrak{f}_i\right)e^{-ip x } \right]f_{\mathbf{k}_{\perp},\omega}(y , z , t)\\ a_z=&- k_3\left [ \left(\mathfrak{c}_i-\mathfrak{e}_i\right)e^{ip x}+\left(\mathfrak{d}_i-\mathfrak{f}_i\right)e^{-ip x } \right]f_{\mathbf{k}_{\perp},\omega}(y , z , t)\\ \phi=&\frac{c^2 } { \omega}(p^2+k_{\perp}^2)(\mathfrak{e}_ie^{ipx}+\mathfrak{f}_ie^{-ipx})f_{\mathbf{k}_{\perp},\omega}(y , z , t ) \end{aligned}\right .. \end{split}\]]in the plates , @xmath171 , the potentials are given by @xmath176 for @xmath168 , renaming the variables by @xmath177or equivalently , @xmath178the formulas for the modes in the vacuum then becomes @xmath179f_{\mathbf{k}_{\perp},\omega}(y , z , t ) \end{aligned}\right .
\end{split}\]]for @xmath168 ; and the boundary conditions becomes @xmath180which follows from the continuities of @xmath7 and @xmath181 . here
for @xmath183 , @xmath184 and @xmath185 or @xmath186 and @xmath187 . for @xmath188 , @xmath186 and @xmath189 or @xmath190 and @xmath191 . substituting the third equation into the first equation , one obtains @xmath192which
just reflects the continuity of @xmath69 .
the boundary conditions thus can be decomposed into two systems , one is the system @xmath193that reflects the continuities of @xmath194 and @xmath69 , together with the conditions @xmath195 , @xmath196 at the artificial boundaries @xmath95 and @xmath96 ; and the other one is the system @xmath197that reflect the continuities of @xmath198 , together with the conditions @xmath199 , @xmath200 at the artificial boundaries @xmath95 and @xmath96 .
in fact , determines @xmath201 from @xmath202 but _ it does not determine uniquely_. one has the freedom to add any solutions satisfying @xmath203 corresponding to the trivial solution of @xmath204 and @xmath205 .
the nontrivial solutions of @xmath204 and @xmath202 satisfying correspond to the type 3 continuum modes discussed in @xcite .
for @xmath206 , choosing @xmath207eq . is satisfied , and we can fix the type 3 continuum modes to be @xmath208where @xmath168 for @xmath88 . for @xmath209 and @xmath210 ,
the modes are given by .
the boundary conditions satisfied by these modes are given by .
it is easy to verify that the contribution of these modes to the casimir energy is given by @xmath211 .
for the trivial solution @xmath212 , the potentials vanish identically on the plates . in the vacuum
, we have three systems of equations : @xmath213 = 0.49 these three systems are independent . in the limit @xmath214 and @xmath215 go to infinity , we only need to be concerned with the system with @xmath186 , which has nontrivial solutions if and only if @xmath216 , where @xmath217 .
in other words , @xmath218 and the corresponding potentials are given by @xmath219 in the region @xmath220 .
this is precisely the type 2 discrete modes discussed in @xcite .
their contribution to the casimir energy is @xmath221 .
notice that they are totally different from the transverse modes of type ii . in fig .
[ f3 ] , we plot the casimir forces as a function of mass when the background medium has constant refractive index @xmath222 and @xmath223 . in the graph , we choose @xmath224=10 nm . from the graph , we can see that the te contribution and the tm contribution to the casimir force are approximately the same when the refractive index of the background medium is one . since the first tm contribution to the casimir force is equal to the te contribution to the casimir force in this case , we find that the second tm contribution to the casimir force is negligibly small .
however when the background medium has refractive index @xmath223 , the graph shows that there is a significant difference between the te and tm contributions .
in fact , when @xmath109 gets larger , the total casimir force is dominated by the tm contribution .
in this section , we consider the limits of the casimir energy and casimir force when the plates become infinitely permeable , i.e. , @xmath225 . in this limit , @xmath226@xmath227 @xmath228from these , it is immediate to show that the te contribution to the casimir energy for infinitely permeable plates is given by , the same as in the case of perfectly conducting plates .
it follows that the te contribution to the casimir force is given by , which is always attractive . for the tm contribution to the casimir energy
, one can show that @xmath229where @xmath230\left[1-\delta^2e^{-2q_0t_r}\right]-\delta^2[1-e^{-2q_0t_l}][1-e^{-2q_0t_r}]e^{-2q_{l , b}a}\right)}{\left(1- \delta^2e^{-2q_0t_l}\right)\left(1-\delta^2e^{-2q_0t_r}\right)}\\=&(1-e^{-2q_{t , b}a})\left(1-\frac{\delta^2(1-e^{-2q_0t_l})(1-e^{-2q_0t_r})}{\left(1- \delta^2e^{-2q_0t_l}\right)\left(1-\delta^2e^{-2q_0t_r}\right)}e^{-2q_{l , b}a}\right ) .\end{split}\ ] ] therefore , the tm contribution to the casimir energy @xmath231 can be split into the sum of two terms @xmath157 and @xmath158 . the first tm contribution to the casimir energy @xmath157
is the same as the te contribution , given by .
therefore the corresponding casimir force is also always attractive .
the second tm contribution to the casimir energy and casimir force are given respectively by @xmath232when the refractive index of the background @xmath233 is a positive constant , @xmath234 .
therefore @xmath235 .
it follows that the second tm contribution to the casimir force @xmath236 is also always attractive .
this implies that the casimir force acting on a pair of infinitely permeable plates is always attractive .
notice that @xmath237where @xmath238 comparing to , we find that the second tm contribution to the casimir energy can be identified with the tm contribution to the casimir energy of a pair of dielectric plates with permittivity @xmath239 that is due to a _ massless _ vector field .
= 0.5 in fig .
[ f4 ] , we plot the casimir forces as a function of mass when the background medium has constant refractive index @xmath222 and @xmath223 . in this graph , we choose @xmath224=10 nm .
we can see there is a significant difference between the te and tm contributions to the casimir force .
in this section , we consider the limits of the casimir energy and casimir force in boyer s setup @xcite , i.e. , one plate is perfectly conducting and one plate is infinitely permeable . without loss of generality ,
assume that the plate on the left is perfectly conducting , and the plate on the right is infinitely permeable , i.e. , @xmath240 .
in this limit , @xmath226@xmath241 @xmath242}:=\lambda,\quad \frac{\tilde{\alpha}_{rb}\tilde{\beta}_{rb}}{\tilde{\kappa}_{rb}^+\tilde{r}_{rb}^+}=\frac{\tilde{\alpha}_{br}\tilde{\beta}_{br}}{\tilde{\kappa}_{br}^+\tilde{r}_{br}^+ } \;\longrightarrow 0.\]]the te contribution to the casimir energy and casimir force is then given respectively by @xmath243it follows that the te contribution to the casimir force is always repulsive . for the tm contribution
, one can show that @xmath244where @xmath245 ^ 2-[\delta+\lambda]^2e^{-2q_0t_r}\right)\\ \mathbb{w}=&\mathbb{w}_{\infty}+(1-\delta^2e^{-2q_0t_l})\left([1-\lambda^2]-[\delta^2-\lambda^2]e^{-2q_0t_r}\right)e^{-2q_{t , b}a } -\delta(1- e^{-2q_0t_l})\\&\times\left([1-\lambda][\delta-\lambda]-[1+\lambda][\delta+\lambda]e^{-2q_0t_r}\right)e^{-2q_{l , b}a}-\delta(1-\lambda)(\delta+\lambda ) ( 1- e^{-2q_0t_l})(1- e^{-2q_0t_r})e^{-2q_{t , b}a}e^{-2q_{l , b}a } \end{split}\ ] ] in this case , the tm contribution can not be split into the sum of two contributions even if we assume that the background medium have unity refractive index .
= 0.5 in fig .
[ f5 ] , we plot the casimir forces as a function of mass when the background medium has constant refractive index @xmath222 and @xmath223 . in this graph , we choose @xmath224=10 nm . from the graph , we find that the casimir force is repulsive when the background medium has refractive index @xmath222 .
however , when the background medium has refractive index @xmath223 , the casimir force can change from repulsive to attractive when the mass increases .
this shows that the change of mass can change the sign of the casimir force .
in this article , we have derived the casimir energy and casimir force acting on two parallel plates due to the vacuum fluctuations of a massive vector field .
we assume that the two parallel plates are made of real materials and they are placed in a magnetodielectric background . with maxwell s equations replaced by proca equations , we argue that the correct boundary conditions for massive vector fields are the continuities of @xmath246 and @xmath247 .
the last one is equivalent to the continuity of @xmath59 in the massless case .
not all the boundary conditions are independent . a set of independent boundary conditions is given by the continuities of @xmath248 and the continuity of either @xmath7 or @xmath249 .
the plane waves in an unbounded media can be divided into transverse waves of type i and type ii and longitudinal waves , where the transverse waves of type i and type ii are natural extensions of te waves and tm waves in the massless case . for a system of several plane parallel layers of general magnetodielectric media ,
there are transverse modes of type i that satisfy all the boundary conditions .
the contribution to the casimir energy from these modes is the natural generalization of the te contribution in the massless case , and we also call it te contribution . in general , there are no transverse modes of type ii or longitudinal modes that satisfy all the boundary conditions .
therefore these two types of modes have to be combined .
the formula of their contribution to the casimir energy is very complicated .
however , in the massless limit , the two polarizations reduce to one polarization and the tm contribution to the casimir energy in the massless case is reproduced .
therefore , we also call the contribution from the superposition of type ii transverse modes and longitudinal modes the tm contribution . in the limit where the plates become perfectly conducting , obtained by letting the permittivities of the plates tend to infinity , we write down explicitly the formulas for the te and tm contributions to the casimir energy . for general magnetodielectric background
, the tm contribution can not be split into two pieces .
however , when the refractive index of the background is equal to unity , which happens for instance when the background is vacuum , the tm contribution can be split into two pieces , corresponding to the contribution from type 2 discrete modes and contribution from type 3 continuum modes discussed in the work of barton and dombey @xcite . in other words ,
our result reproduces the result of barton and dombey @xcite for the special case they have considered . in this case , the casimir force is always attractive . in the limit where the plates become infinitely permeable , obtained by letting the permeabilities of the plates tend to infinity , the tm contribution to the casimir energy
can always be split into a sum of two terms . in this case , the casimir force is always attractive when the background has constant refractive index . to look for possible scenario that leads to repulsive force , we consider the configuration proposed by boyer @xcite , where one plate is perfectly conducting and one plate is infinitely permeable . in this case , we find that the tm contribution to the casimir force can not be split into a sum of two terms even when the background medium has unity refractive index .
numerical computation shows that the casimir force can become attractive for nonzero masses when the refractive index of the background medium is not one .
this shows that compare to the massless case , the casimir effect of a massive vector field can behave very differently . | we study the casimir effect due to a massive vector field in a system of two parallel plates made of real materials , in an arbitrary magnetodielectric background .
the plane waves satisfying the proca equations are classified into transverse modes and longitudinal modes which have different dispersion relations .
transverse modes are further divided into type i and type ii corresponding to te and tm modes in the massless case . for general magnetodielectric media
, we argue that the correct boundary conditions are the continuities of @xmath0 and @xmath1 , where @xmath2 is the direction normal to the plates . whereas there are type i transverse modes that satisfy all the boundary conditions , it is impossible to find type ii transverse modes or longitudinal modes that satisfy all the boundary conditions . to circumvent this problem , type ii transverse modes and longitudinal modes have to be considered together .
we call the contribution to the casimir energy from type i transverse modes as te contribution , and the contribution from the superposition of type ii transverse modes and longitudinal modes as tm contribution .
their massless limits give respectively the te and tm contributions to the casimir energy of a massless vector field .
the limit where the plates become perfectly conducting is discussed in detail .
for the special case where the background has unity refractive index , it is shown that the tm contribution to the casimir energy can be written as a sum of contributions from two different types of modes , corresponding to type 2 discrete modes and type 3 continuum modes discussed by barton and dombey @xcite .
for general background , this splitting does not work .
the limit where both plates become infinitely permeable and the limit where one plate becomes perfectly conducting and one plate becomes infinitely permeable are also investigated . |
understanding the statics and dynamics of elastic systems in a random environment is a longstanding problem with important applications for a host of experimental systems .
such problems can be split into two broad categories : ( i ) propagating interfaces such as magnetic domain walls@xcite , fluid invasion in porous media @xcite or epitaxial growth @xcite ; ( ii ) periodic systems such as vortex lattices@xcite , charge density waves@xcite , or wigner crystals of electrons@xcite . in all these systems
the basic physical ingredients are identical : the elastic forces tend to keep the structure ordered ( flat for an interface and periodic for lattices ) , whereas the impurities locally promote the wandering . from the competition between disorder and elasticity
emerges a complicated energy landscape with many metastable states .
this results in glassy properties such as hysteresis and history dependence of the static configuration . in the dynamics ,
one expects of course this competition to have important consequences on the response of the system to an externally applied force
. to study the statics , the standard tools of statistical mechanics could be applied , leading to a good understanding of the physical properties .
scaling arguments and simplified models showed that even in the limit of weak disorder , the equilibrium large scale properties of disordered elastic systems are governed by the presence of impurities .
in particular , below four ( internal ) dimensions , displacements grow unboundedly@xcite with the distance , resulting in rough interfaces and loss of strict translational order in periodic structures@xcite . to go beyond simple scaling arguments and obtain a more detailed description of the system
is rather difficult and at present only main two methods , each with its own shortcomings , have been developped .
the first one is to perform a perturbative renormalization group calculation on the disorder , and is valid in @xmath10 dimensions to first order in @xmath11 . in this functional renormalization group ( frg )
approach @xcite , the whole correlation function of the disorder is renormalized .
the occurence of glassiness is signalled by a non - analyticity appearing at a finite lengtscale during the flow , specifically a cusp in the force correlator .
this yields non trivial predictions for the roughness exponents of interfaces@xcite .
another approach relies on the replica method to study either the mean field limit ( i.e large number of components ) or to perform a gaussian variational approximation of the physical model . using this variational approach both for manifolds @xcite and for periodic systems@xcite , correlation functions and thermodynamic properties
could be obtained .
it confirms the existence of glassy properties , with energy fluctuations growing as @xmath12 where @xmath13 is a positive exponent . to obtain the glass phase in this method
, one must break the replica symmetry . at a qualitative level , this is in good agreement with the physical intuition of such systems as being composed of many low lying metastable states separated by high barriers .
as was clearly shown in the case of periodic manifolds , the correlation functions can be obtained by both the frg and variational approach and are found to be in very reasonable agreement , bridging the gap between the two methods @xcite . taken together , these two approaches thus provide a very coherent picture for the statics@xcite . in particular they allow to understand that although disorder leads to glassy features in both the manifold and the periodic systems , these two types of problems belong to quite different universality classes in other respects , such as the large distance behaviour of the correlations @xcite .
these properties have drastic consequences for the dynamics of driven systems in the case , important in practice , where an elastic description holds ( i.e when plastic deformations can be neglected ) .
determining the response to an externally applied force is not only an interesting theoretical question , but also one of the most important experimental issues . indeed in most systems
the velocity @xmath1 versus force @xmath0 characteristics is directly measurable and is simply linked to the transport properties ( voltage - current for vortices , current - voltage for cdw and wigner crystals , velocity - applied magnetic field for magnetic domain walls ) . in the presence of disorder it is natural to expect that , at zero temperature , the system remains pinned and only polarizes under the action of a small applied force , i.e. moves until it locks on a local minimum of the tilted energy landscape . at larger drive , the system follows the force @xmath0 and acquires a non - zero asymptotic velocity @xmath1 . in the simplest cases ,
the effect of disorder at large velocity is washed out and one recovers the viscous flow , as in the pure case . in the thermodynamic limit
, it is believed that there exists a threshold force @xmath14 separating both states , and that a dynamical transition occurs at @xmath14 called _ depinning _ , where the velocity is continuously switched on , like an order parameter of a second order transition in an equilibrium system@xcite , leading to a @xmath1@xmath0 characteristics such as the one shown in figure [ vfinsp ] .
an estimate of @xmath15 can be obtained via scaling arguments @xcite or with a criterion for the breakdown of the large velocity expansion@xcite . beyond @xmath15 , if one describes the depinning as a conventional dynamical critical phenomenon , the important quantities to determine are of course the depinning exponent @xmath16 giving the velocity @xmath17 and the dynamical exponent @xmath18 which relates space and time as @xmath19 .
an even more challenging question , and experimentally at least as relevant , is the response at finite temperature @xmath2 . in the most naive description ,
the system can now overcome barriers via thermal activation , leading to a thermally assisted flow @xcite and a linear response at small force of the form @xmath20 , where @xmath21 is some typical barrier .
it was realized @xcite that because of the glassy nature of the static system , the motion is actually dominated by barriers which _ diverge _ as the drive @xmath0 goes to zero , and thus the flow formula with finite barriers is incorrect .
well below the threshold critical force , the barriers are very high and thus the motion , usually called `` creep '' is extremely slow . scaling arguments , relying on strong assumptions such as the scaling of energy barriers and the use of statics properties to describe an out of equilibrium system , were used to infer the small @xmath0 response .
this led to a non linear response , characteristic of the creep regime , of the form @xmath22 where @xmath23 and @xmath24 is the roughening exponent for the static @xmath6-dimensional system .
obtaining a detailed experimental confirmation of this behaviour is a non trivial feat , in reasons of the range in velocity required .
although in vortex systems these highly non linear flux creep behaviours have been measured ubiquitously , it is rather difficult to obtain clean determination of the exponents , given the many regimes of lengthscales which characterize type ii superconductors @xcite .
in some recent measurement , some agreement with the creep law in the bragg glass regime was obtained@xcite .
probably the most conclusive evidence for the above law was obtained , not in vortex systems , but for magnetic interfaces . quite recently
et al._@xcite successfully fitted the force - velocity characteristics of a magnetic domain wall driven on a random substrate by a stretched exponential form @xmath25 over _ eleven _ decades in velocity .
this provided evidence not only of the stretched exponential behavior , but of the validity of the exponent as well .
given the phenomelogical aspect of these predictions and the uncontrolled nature of the assumptions made , both for the creep and for the depinning , it is important to derive this behavior in a systematic way from the equation of motion .
less tools are available than for the statics , and averages over disorder should be made using dynamical methods .
fortunately , it is still possible to use a functional renormalization group ( frg ) approach for the dynamical problem .
such an approach has been used at @xmath26 to study depinning @xcite .
it allowed for a calculation of the depinning exponents , in @xmath27 .
however this approach is still rather unsatisfactory .
the frg flow used in @xcite is essentially the static one , the finite velocity being only invoked to remove - by hand - some ambiguities and to cutoff the flow , with no real controlled way to show that this is the correct procedure .
furthermore in these approaches it is also necessary to assume , instead of deriving them from the frg , some scaling relations in order to obtain the exponents .
another rather problematic point is that , with no additional input , the method of @xcite would yield three universality classes for the depinning : two universality classes depending on the nature of the disorder ( random bond versus random field ) for manifolds and one for periodic systems , while numerics and physical arguments @xcite suggested that only two ( random field and periodic ) universality classes could exists .
in addition , since this is also intrinsically a @xmath26 ( and @xmath28 ) approach , it can not be used to tackle the creep behavior .
we propose here a single theory for describing all the regimes of a moving elastic system , including depinning and the non - zero temperature regimes .
our frg equations contain from the start the finite velocity and finite temperature .
they thus allow to address questions which are beyond the reach of either approximate scaling theories , or @xmath28 frg flow .
for the depinning we are able to determine the conditions required for the existence of a universal depinning behavior , as well as computing the depinning exponents ( and estimating @xmath15 ) .
we show in particular that only two universality classes exist ( out of the three ) for the depinning since we explicitly find that random bond systems flow to the random field universality class .
we can also extract from our equations the characteristic lengthscales of the depinning .
the main advantage of our approach is of course to address the finite @xmath29 small @xmath1 regime as well .
the method allows to _ derive _ the creep formula directly and thus allows to confirm the assumptions made on the scaling of the energy barriers .
in addition we show that the creep is followed by a depinning - like regime and determine its characteristic lengthscales .
a short account of some of these results was presented in ref . .
the paper is organized as follows : in section [ elasticitydisorder ] we present the equation of motion and the types of disorder studied here .
section [ preliminary ] is devoted to a brief review of scaling arguments and a summary of useful results from perturbation theory , presented in appendix [ app : pt ] .
section [ fieldtheory ] contains the field theoretical formulation of the problem and the associated renormalization group flow equations , derived in appendix [ app : derivation ] .
the static case is studied in subsection [ statics ] , focusing on the appearance of the cusp .
the effect of the temperature is studied in details in appendices [ app : temperature ] and [ app : ftfp ] . in the next sections , we study the depinning ( [ depinning ] ) and creep ( [ creep ] ) regimes .
both sections contain the outline of the derivation and a physical discussion .
appendix [ app : v ] is devoted to the effect of a small velocity on the frg .
we conclude in section [ conclusion ] , referring to an extension of our work proposed in appendix [ app : n>1 ] . in appendix
[ app : notations ] we fix the notations used throughout the paper .
elastic systems are extended objects which `` prefer '' to be flat or well ordered .
we are dealing with two different types of elastic systems which however can be treated in the same way .
on the one hand , _ interfaces _ , i.e. surfaces with a stiffness that makes local distortions energetically expensive , on the other hand , _ lattices _ with elastic displacements allowed about a regularly ordered configuration .
the first type is the easiest to visualize .
the interface is assumed to have no overhangs and is thus described by a height function @xmath30 defined at each point @xmath31 ( see figure [ interface ] ) .
its energy is proportional to its area @xmath32 and in the elastic limit @xmath33 , reduces to @xmath34=\int_{r } \frac{c}{2}|\nabla u|^{2}\end{aligned}\ ] ] relative to the flat @xmath35 configuration ( notations are defined in the appendix [ app : notations ] ) .
we denote by @xmath36 the stiffness , or elastic constant .
periodic structures , such as flux line lattices or charge density waves ( cdw ) , can be described by the same type of elastic hamiltonian .
for each point ( or line ) in the elastic periodic system one can introduce a ( vector ) displacement field @xmath37 that gives the shift from the reference position @xmath38 ( see figure [ interface ] ) .
the elastic energy for small displacements is given by a quadratic form in the differences @xmath39 between neighboring points and thus can be written as ( [ elastic ] ) in a continuum description ( @xmath31 being a generic point in space ) . when @xmath40 has more than one component , @xmath36 should be understood as a tensor ( see appendix [ app : n>1 ] ) .
to take the quenched disorder into account in such systems it is necessary to express the energy of the above elastic structure in the presence of impurities .
the coupling to a substrate or to local fields is easily written for interface models and is more subtle for lattices .
quite generally the coupling to disorder leads to an energy : @xmath41 which gives rise to a pinning force @xmath42 acting on the displacement @xmath30 . depending on the microscopic origin of the disorder term @xmath43 , the coupling ( [ disorgeneral ] ) leads to quite different physics . in the case of interfaces ( [ disorgeneral ] ) originates from @xmath44 in terms of the density @xmath45 .
one then usually distinguishes two cases : either `` random bond '' ( rb ) when @xmath46 is short range ( random exchange for magnetic domain walls ) , or `` random field '' ( rf ) as discussed below , where @xmath46 has long range correlations . in the case of periodic structures
, the density @xmath47 can be expressed using the set of vectors @xmath48 of the reciprocal lattice and ( [ disorgeneral ] ) originates from @xmath49 where @xmath50 the average density@xcite .
the potential @xmath51 is random , of short range @xmath52 ( e.g. point impurities for a vortex lattice or a cdw ) .
we call this case `` random periodic '' ( rp ) . in both cases , using ( [ hdisinterface],[rhointerface],[hdislattice],[rholattice ] ) and ( [ disorgeneral ] ) one obtains for the correlations of @xmath43 in ( [ disorgeneral ] ) @xmath53 where @xmath54 is a periodic function with the periodicity @xmath55 of the lattice in the periodic ( rp ) case@xcite .
the @xmath56 function is cutoff at the microscopic scale @xmath52 . for an interface
, @xmath54 has the shape shown in figure [ rbrfsketch ] . in that case , the width @xmath57 of @xmath54 is typically given by the width of the interface or the size of impurities .
the force resulting from such a random bond disorder has correlations@xcite @xmath58 as shown in figure [ rbrfsketch ] where @xmath59 the signature of such a rb disorder for the interface is that @xmath60 since @xmath61 decreases to zero at infinity .
another type of disorder occurs in the case of interfaces separating two phases , like e.g. a domain wall in a disordered magnet .
a random field couples differently to the two phases on the right and left of the interface , thus the energy resulting from the coupling to disorder involves an integral in the bulk of the system and not just _ at _ the interface position .
the correlation of the force can still be expressed by ( [ delta ] ) and @xmath21 still decreases to zero above a scale @xmath57 as shown on figure [ rbrfsketch ] .
contrarily to the rb case , @xmath62 does not vanish .
for a single component displacement field @xmath40 , the rf , of correlator ( [ delta ] ) , is still _ formally _ the derivative of a potential @xmath63 . the correlations of this fictious potential are of the form ( [ r ] ) with @xmath64 , and one has @xmath65 for @xmath66 which can be visualized as a random walk ( where @xmath40 plays the role of `` time '' and the _ random field strength _
@xmath67 is the `` diffusion constant '' ) .
contrarily to the rb for which @xmath68 is short range , @xmath68 for the rf grows at large @xmath40 as shown on figure [ rbrfsketch ] . in this paper
we study the overdamped driven motion of such elastic systems which obey @xmath69 where @xmath70 is a friction , @xmath0 is the external driving force density and @xmath71 a langevin noise .
the correlation @xmath72 defines as usual a temperature @xmath29 for this out of equilibrium system .
the long time behavior of ( [ notcomoving ] ) at zero drive @xmath73 , reduces to the thermodynamics at temperature @xmath29 . in ( [ notcomoving ] )
the bare@xcite pinning force @xmath74 is gaussian with zero average and correlator given by ( [ delta ] ) .
we will consider three universality classes for @xmath75 corresponding to an interface in a random potential ( rb ) , in a random field ( rf ) or a periodic system in a random potential ( rp ) .
physical realizations of such disorders would be respectively a random anisotropy for a magnetic domain wall @xcite , the random field ising systems @xcite and vortex lattices or cdw @xcite .
it is also useful to rewrite ( [ notcomoving ] ) in the co - moving frame at average velocity @xmath76 . in the remainder of this paper
, we switch to @xmath77 and thus study the following equation of motion @xmath78 where @xmath79 is the average pinning force and @xmath31 belongs to a @xmath6-dimensional internal space . from now on we specialize to an unidimensional displacement field @xmath30 as would be the case for an interface model or a single @xmath80 cdw .
this simpler case already captures the main physics at small velocity , investigated here .
extensions to many - component systems will be briefly discussed . before giving a quantitative treatment using renormalization group ,
let us review the qualitative arguments which have been given previously to describe the physics originating from ( [ notcomoving ] ) .
in the absence of drive , ( [ notcomoving ] ) is equivalent to the equilibrium problem at temperature @xmath29 .
the state of the system results from the competition between elasticity , pinning and thermal fluctuations .
the physics of such problems can be investigated by a host of methods @xcite and here we only recall the salient points .
temperature does not play an important role as will become clear and we begin with the @xmath26 case .
a subsystem of size @xmath38 , with displacement @xmath81 , is submitted to a typical elastic force density @xmath82 and to a typical pinning force density @xmath83 .
balancing these quantities , one obtains that elasticity wins at large scales for @xmath84 , resulting in a flat interface with a priori bounded displacements . in @xmath85 , systems of size @xmath38 smaller than the larkin length @xmath86 wander as predicted by the larkin model@xcite : @xmath87 at larger scales @xmath88 , the system wanders further than the correlation length @xmath57 of the disorder .
this simple picture breaks down and the system can be viewed as made of larkin domains of size @xmath89 , which are independently pinned .
first order perturbation theory confirms this picture below the larkin length . the static equilibrium ( equal - time ) correlation function at @xmath26 is ( see appendix [ app : pt ] ) @xmath90 the wandering computed from ( [ unsurq4 ] ) @xmath91 for @xmath92 gives back ( [ larkinw ] ) , and we recover the scaling expression ( [ larkinlength ] ) by equating the wandering to @xmath93 : @xmath94 we used that @xmath95 for @xmath96 with @xmath97/16 \sim s_{4}/\epsilon $ ] when @xmath98 .
the remarkable feature is that @xmath26 perturbation theory ( either using replicas or equilibrium dynamics ) gives that ( [ unsurq4 ] ) is exact _ to all orders _ @xcite in @xmath21 , and is identical to the correlation in the larkin model@xcite .
indeed , the naive perturbation series organizes as if the pinning energy were simply expanded in @xmath40 ( thus the pinning force is independent of @xmath40 with @xmath99 ) , resulting in a gaussian model .
in fact , due to the occurence of multiple minima beyond @xmath89 , this perturbative result is incorrect@xcite at large scale .
it can be shown , for example on discrete systems , that if a configuration @xmath100 which minimizes @xmath101=\int_{r}\left[\frac{c}{2 } ( \nabla u_{r})^{2}+v ( r , u_{r})\right]$ ] is defined on a volume larger than @xmath89 , then the hessian @xmath102 $ ] becomes singular @xcite .
such instability appears clearly in a functional renormalization group ( frg ) treatment of the problem@xcite which proves that @xmath21 becomes nonanalytic beyond the length @xmath103 , as will be discussed below .
it can also be seen within variational or mean - field treatments using replicas @xcite that replica symmetry breaking ( rsb ) is necessary to describe the physics beyond the larkin length @xmath103 . using either replicas with rsb or the frg it is possible to describe the physics at all scales and to obtain the correct roughness exponent @xmath24 defined by @xmath104 where the value of @xmath105 depends on the statics universality class@xcite .
since disorder induces unbounded displacements , the system is rough and the temperature is always formally irrelevant in @xmath106 .
it is described by a @xmath26 fixed point , characteristic of a glass phase .
an elastic system does not necessarily move under the action of a driving force .
the disorder leads to the existence of a threshold force @xmath15 at @xmath26 as shown on figure [ vfinsp ] .
a simple dimensional estimate of @xmath15 can be obtained@xcite by computing the sum of the independent pinning forces acting on the larkin domains @xmath107 and balancing it with the driving force acting on the same volume @xmath108 .
this gives @xmath109 another estimate of @xmath15 comes from the large velocity expansion @xcite of the equation of motion ( [ comoving ] ) ( from the criterion @xmath110 ) .
it coincides with ( [ fclarkin ] ) . for @xmath111
the system moves with a small velocity , and it has been proposed@xcite that depinning can be described in the framework of standard critical phenomena , with the velocity as an order parameter .
this leads to the assumption of two independent critical exponents@xcite @xmath112 and @xmath18 , defined through the correlation function in the comoving frame ( in the stationary state for @xmath113 ) @xmath114 @xmath115 for @xmath116 and @xmath117 for @xmath118 .
the dynamical roughening exponent @xmath112 close to the threshold a priori differs from its equilibrium value @xmath24 .
several related exponents can be also introduced such as : ( i ) the depinning exponent @xmath16 ; ( ii ) the correlation length exponent @xmath119 describing the divergence of the length @xmath120 defined from the equal - time velocity - velocity correlation function .
they satisfy @xmath121 numerically@xcite the motion of the system looks like a deterministic succession of avalanches of size @xmath120 with characteristic time @xmath122 . from the argument @xmath123 and the statistical tilt symmetry@xcite ( see below ) , the exponents @xmath124 and @xmath125 are usually determined from @xmath126 by the scaling relations @xmath127 to obtain these exponents analytically , one needs to perform an frg analysis of the equation of motion .
this will be discussed in more details in section [ depinning ] . at finite temperature @xmath2
, motion occurs at any drive . for low temperatures and very small drive @xmath128
one expects the motion to be very slow , and thus , although it is a dynamical problem , a qualitative understanding can be obtained by considering thermal activation over barriers determined from _ statics _ arguments .
an original estimate@xcite of such barriers led to linear , albeit activated , response .
however the effects linked to the glassy nature of the problem were understood at a qualitative level@xcite using scaling arguments .
the argument proceed as follows : systems larger than @xmath89 have a ( static ) roughness @xmath129 and hence the energy has typical fluctuations of order @xmath130 with @xmath131 the energy scale of a larkin domain . assuming that the energy landscape is characterized by a _
unique _ energy scale , and thus that the energy differences between neighbouring metastable states is the same as the energy barrier separating them as schematically shown in figure [ landscape ] , one obtains that the barriers height scale with an exponent @xmath132 . since the motion is very slow , it is usually argued that the effect of the drive is just to tilt the energy landscape , and the effective barrier becomes @xmath133 the maximum of ( [ barriertilt ] ) , obtained at @xmath134 , gives via arrhenius law the largest time spent in the valley by the thermally activated system and thus yields the velocity @xmath135 \qquad \mu=\frac{d-2 + 2 \zeta _ { \rm eq}}{2-\zeta _ { \rm eq}}\end{aligned}\ ] ] known as the _ creep _ motion , characterized by the stretched exponential with exponent @xmath4 .
note that the effective barrier given by the above formula vanishes at a scale @xmath136 which diverges as fast as @xmath137 , the typical size of a thermally activated excitation .
this elegant scaling argument leading to the creep formula relies however on strong assumptions and does not yield any information on the detailed behavior , in particular on what happens after the thermal jumps .
the fact that _ static _ barriers and valleys scale with the same exponent is already a non - trivial hypothesis about the structure of the infinite - dimensional energy landscape .
refined simulations@xcite of a directed polymer @xmath138 in @xmath139 and @xmath140 are consistent with the `` equal scaling assumption '' for this particular case , but a general proof is still lacking . the second , and
more delicate hypothesis is the validity of the arrhenius description : ( i ) the system being out of equilibrium , it is not clear that _ dynamical _ barriers can be determined purely from the statics ; ( ii ) one assumes that the motion is dominated by a _
typical _ barrier .
these assumptions can turn incorrect for some specific problems .
for example , in the case of a point moving in a one - dimensional random potential , the @xmath1-@xmath0 characteristics at low drive is _ not _ @xcite of arrhenius type .
although this @xmath141 case is peculiar since the particle has no freedom to pass aside impurities ( it is dominated at @xmath26 by the highest slope of the potential and at finite @xmath29 by the rare highest barriers ) , one should also address the question of the distribution of barriers in higher dimensions .
let us now study the equation of motion ( [ comoving ] ) using a full frg treatment .
this will enable us to describe the physics at all lengthscales and in particular the depinning and creep regime .
a natural framework for computing perturbation theory in off - equilibrium systems is the dynamical formalism@xcite .
after exponentiating the equation of motion ( [ notcomoving ] ) using a _ response field _
@xmath142 , the average over thermal noise and disorder can safely be done and yields the simple ( `` unshifted '' ) action @xmath143 disorder and thermal averages @xmath144 \rangle } = \langle a[u ] \rangle_{s_{\rm uns}}$ ] of any observable @xmath145 $ ] can be computed with the weight @xmath146 .
furthermore , response functions to an external perturbation @xmath147 added to the right hand side of ( [ notcomoving ] ) are simply given by correlations with the response field : @xmath148 i\hat { u}_{rt}\rangle = \frac{\delta}{\delta h_{rt}}\langle a[u ] \rangle $ ] .
it can be checked that causality is satisfied : @xmath149 i\hat { u}_{rt}\rangle $ ] vanishes . in the time - continuum ,
the reponse to a perturbation at time @xmath150 of an observable depending on @xmath151 is ill - defined .
we choose ito convention for the equation of motion , which ensures that equal time response functions , and hence any diagram occurring in perturbation theory containing a loop of response functions , vanish .
the continuum field theory necessarily breaks down at small scales and it becomes necessary to cut off the integrals over the modes at large @xmath152 , using a large wave vector @xmath153 .
a full summary of the notations can be found in appendix [ app : notations ] .
it proves more convenient to work in the _ comoving frame _
( i.e. with equation ( [ comoving ] ) ) .
the corresponding action is @xmath154 where the field @xmath40 satisfies @xmath155
. this condition fixes @xmath156 in ( [ eq : action ] ) .
this quantity is the ( macroscopic ) pinning force , since it shifts the viscous law @xmath157 by the amount of @xmath158 .
several exact relations can be derived directly from ( [ eq : action ] ) . for any static field @xmath159 ( vanishing at infinity ) @xmath160 performing the change of variable @xmath161 gives @xmath162 applying @xmath163 yields the exact relation @xmath164 where we denote by @xmath165 the exact response function .
this symmetry , known as statistical tilt symmetry , ensures that the elasticity is _ not _ corrected during the renormalization .
another important relation can be derived from @xmath166 this leads to the identity between the macroscopic mobility and the slope of the @xmath1-@xmath0 characteristics at any drive and any temperature : @xmath167 this exact result can also checked explicitely in the case of a particle moving in a one - dimensional environment @xcite . to extract the physical properties from the action ( [ eq : action ] )
it is necessary to build a perturbative approach in the disorder .
a particularly simple case@xcite occurs when the velocity is very large . in that case
the disorder operator in the action can be formally replaced by @xmath168 since one may neglect the @xmath169 compared to @xmath170 .
this trick suppresses the non - linearity and the remaining action is quadratic .
furthermore , at large velocity , @xmath171 can be replaced by @xmath172 and the disorder operator transforms into a temperature operator ( because it becomes local in time @xmath173 ) .
the resulting action is the dynamical action associated to the edwards - wilkinson equation @xcite describing the motion of an elastic system in a purely thermal noise @xmath174 with @xmath175 , langevin noise @xcite of additional temperature @xmath176 .
note that at @xmath26 the results at large @xmath1 coincide with the pertubative expansion in powers of the disorder .
the equal - time correlation function in the driven system with force @xmath0 crosses over from the static @xmath177 larkin behavior at small scale to a thermal @xmath178 behavior at larger scale @xmath179 with the same @xmath180 , generated at lengthscales @xmath181 .
we renormalize the theory using wilson s momentum - shell method . as the cutoff @xmath182
is reduced , corresponding to a growing microscopic scale @xmath183 in real space , the parameters of the effective action for slow fields ( whose modes @xmath152 are smaller than @xmath184 ) are computed by integration over the fast part of the fields ( whose modes @xmath152 lie between @xmath184 and @xmath153 ) .
this iterative integration gives rise to _ flow _ equations , better expressed in terms of the _ reduced _ quantities @xmath185 where @xmath186 is the surface of the unit sphere in @xmath6 dimensions divided by @xmath187 .
the exponent @xmath188 is for the moment arbitrary and will be fixed later so that the reduced parameters flow next to appropriate fixed points . in one case ( rb )
we will need a @xmath189-dependent @xmath188 , and it is understood that everywhere the rescaling factors @xmath190 ( appearing e.g. in ( [ tilde ] ) ) should then be replaced by @xmath191 .
the reduced quantities @xmath192 are homogeneous to @xmath193 and @xmath194 to @xmath40 .
the parameter @xmath195 , which plays a crucial role below , can simply be expressed as the following ratio @xmath196 of the distance ( along @xmath40 ) travelled by the center of mass of the interface during @xmath197 and the roughness @xmath198 .
we have defined @xmath199 as the characteristic relaxation time in the model renormalized up to scale @xmath38 .
the details of the renormalization procedure can be found in appendix [ app : derivation ] .
the flow equations read : @xmath200 \nonumber \\ & & \partial \ln \lambda= 2-\zeta -\int_{s>0}e^{-s}s \tilde{\delta}''(s\lambda)\nonumber \\ & & \partial \ln \tilde{t}= \epsilon-2 - 2\zeta + \int_{s>0}e^{-s}s \lambda
\tilde{\delta}'''(s\lambda ) \nonumber \\ & & \partial \tilde{f}= e^
{ -(2-\zeta ) l } c \lambda_{0}^2 \int_{s>0 } e^{-s } \tilde{\delta}'(s\lambda ) \nonumber\end{aligned}\ ] ] where @xmath201 and @xmath202 denotes @xmath203 .
this complicated set of equations require a few comments : ( i ) as for the statics @xcite it is necessary to renormalize the whole function @xmath21 , instead of just keeping few couplings as in standard field theory , ( ii ) the elasticity @xmath36 is not renormalized @xmath204 due to the statistical tilt symmetry ; ( iii ) our equations correctly show that no temperature can be generated at @xmath28 since the fluctuation dissipation theorem holds at equilibrium .
setting both @xmath26 and @xmath28 in ( [ flow ] ) gives back the simplified set of equations used in refs .
( setting only @xmath28 also yields equations found in ref . ) . but compared to the previous frg approaches of the depinning transition , our equations correctly take into account the effect of the velocity on the flow itself ( instead of being treated simply as a cutoff as in ref . ) .
other attempts@xcite to incorporate velocity and temperature in the frg equations did not obtain the first equation giving the renormalization of the disorder at @xmath2 and @xmath205 . to be able to tackle the full dynamical problem and study the depinning and the creep regime , one can not avoid keeping track of the velocity and of the temperature in the flow , as will become clear later , since they yield non - trivial effects which are unreachable by simple scaling arguments .
our flow equations allow in principle to compute the whole @xmath1-@xmath0 characteristics at low temperature . in the following we analyse them in the three regimes corresponding to the statics ( @xmath28 ) , to the depinning at zero temperature ( @xmath26 , @xmath206 ) and the creep regime ( @xmath2 , @xmath207 ) . at zero velocity ,
our approach is a dynamical formulation of the equilibrium problem .
it thus allows to recover the known results about the statics , avoiding the use of replicas .
the standard derivation of the statics using the frg consists in writing a replicated hamilonian for the elastic system pinned in a random potential with correlator @xmath208 . after averaging over @xmath43 the replicated action reads@xcite @xmath209=\frac{1}{2t}\sum_{a}\int_{r}|\nabla u^{a}_{r}|^{2}-\frac{1}{2t^{2}}\sum_{ab}\int_{r}{\sf r } ( u_{r}^{a}-u_{r}^{b})\end{aligned}\ ] ] where @xmath210 are the @xmath211 replica indices .
performing an frg analysis of ( [ replicaction ] ) yields for the flow of @xmath212 and @xmath29 ( remarkably independent of @xmath211 ) : @xmath213 with @xmath214 and @xmath215 , which are the same redefinitions as ( [ tilde ] ) with the correlator @xmath21 of the force is related to @xmath212 by ( [ deltar ] ) .
it is easy to see that ( [ flowreplica ] ) coincides with our equations ( [ flow ] ) when @xmath28 which read @xmath216 thus the two methods give the same results for the static and equal - time physical quantities .
the additional information conveyed by the flow of the friction @xmath70 in the dynamical formalism is discussed later in [ depinninglaw ] and [ depinningdiscussion ] .
the temperature in the static system is an irrelevant operator , since it decreases exponentially fast with @xmath189 .
one thus commonly restricts to the @xmath26 version of the above equations .
in that case , as is obvious from the closed equation @xmath217 the curvature @xmath218 ( see figure [ rbrfsketch ] ) of the correlator , for any initial condition , _ blows up _ at a finite length scale for @xmath85 @xmath219 which corresponds to @xmath220 when approximating @xmath221 by @xmath222 .
one thus recovers the larkin length ( [ larkinlength2 ] ) .
the blowup of the curvature of @xmath21 corresponds to the generation of a _ cusp _ singularity : @xmath21 becomes non - analytic at the origin and acquires for @xmath223 a non - zero @xmath224 .
however , the flow equation for the running non - analytic correlator _ still makes sense_. the non - analyticity just signals the occurence of metastable states .
a well - defined fixed point function @xmath225 exists for each of the rb , rf , rp cases when a suitable @xmath112 is chosen . in the rp case ,
@xmath226 so as to conserve the period @xmath55 , and the fixed point is given by@xcite @xmath227 for @xmath228 .
in the rf case , @xmath229 so as to conserve the rf strength @xmath230 and the fixed point is given by@xcite @xmath231 where @xmath232 , @xmath233 and @xmath234 ( see [ rft ] ) .
in the rb case , it has been shown @xcite by numerical integration of the fixed point equation that @xmath235 yields a physical fixed point , for which no analytical expression is available . despite the irrelevance of the temperature
, this operator has important transient effects during the flow , even if we are left asymptotically with the @xmath26 cuspy fixed point .
it can be shown ( see appendix [ app : temperature ] ) that the temperature hinders the flow from becoming singular at a finite scale .
the running correlator evolves smoothly towards its cuspy fixed point and remains analytic , as was also noticed in ref . .
as shown in appendix [ app : temperature ] , the rounding due to temperature occurs in a boundary layer of width proportional to @xmath236 around the origin .
this is confirmed by the existence of a well - defined expansion in @xmath29 ( see appendix [ nextt ] ) .
this effect is missed by simple perturbation theory that would naively suggest that the rounding occurs on a width proportional to @xmath237 .
indeed the correlation function is proportional to @xmath29 and smoothes @xmath75 by @xmath238 .
although not crucial for the statics this rounding has drastic consequences for the creep as analysed in section [ creep ] .
let us return on the differences between the static and dynamical formalisms . within the static approach ( [ flowreplica ] ) in the @xmath239 limit ,
despite the occurence of the cusp at @xmath240 , the rg equation for @xmath241 still makes sense after @xmath240 and flows to a fixed point controlled by @xmath201 .
however the physical meaning of the cusp is delicate@xcite . on the other hand , the use of the dynamical formalism allows to put @xmath26 from the beginning but adds to the problem a time dimension and the corresponding parameter , the friction @xmath242 . in this dynamical version of the problem ,
the cusp has strong physical consequences which are more immediate : after @xmath240 , the cusp generates infinite corrections to the friction .
this feature marks the onset of a non - zero threshold force at scales larger than the larkin length and signals that an infinite time is needed to go from one metastable state to another .
metastability thus appears very clearly in the dynamical formulation of the statics problem .
a simple physical picture of the cusp in the statics at @xmath26 was given in ref . .
the renormalized potential @xmath243 at scales @xmath88 develops `` shocks '' ( i.e. discontinuities of the force @xmath244 of typical magnitude @xmath245 at random positions ) .
let us now extend this description to draw the link with the critical force and to include thermal effects .
the force correlator for small @xmath246 is dominated by the configurations with a shock present between @xmath40 and @xmath247 : @xmath248 where @xmath249 denotes the probability to find a shock between @xmath40 and @xmath250 .
identifying the rhs with @xmath251 one finds , using the rescalings ( [ tilde ] ) , that the discontinuity in the force has the following scale dependence @xmath252 and can thus be identified with an `` effective critical force '' @xmath253 at scale @xmath38 , which will play a role in the following ( see subsection [ creepdiscussion ] ) . at @xmath254 , @xmath253 reduces to the true critical force @xmath14 . the renormalized problem at scale @xmath38 being the one of an interface in a potential @xmath243 with the above characteristics , one can
now easily understand the result that the cusp of @xmath255 is rounded on a width @xmath256 at @xmath2 . extending the previous argument
, one expects a rounding of a shock if the barrier between @xmath40 and @xmath247 is of order @xmath29 .
since near a shock the potential is linear of slope @xmath245 , the barrier is @xmath257 , and the thermal rounding should thus occur in a boundary layer of width @xmath40 given by @xmath258 using the rescalings ( [ tilde ] ) , this is indeed equivalent to the expression @xmath259 for the width of the boundary layer in rescaled variables found in appendix [ app : temperature ] .
at @xmath26 and @xmath260 , our flow equations give a self - contained picture of the depinning transition .
thanks to our formalism , the problem is reduced to the mathematical study of ( [ flow ] ) , which although complicated , requires no additional physical assumptions . to focus on the depinning transition , we must analyze the solutions of these equations in the regime of small velocitiy where , using ( [ tilde ] ) , @xmath261 is small .
we will examine the various regimes in the rg flow keeping in mind that @xmath195 increases monotonically with @xmath189 .
the equations ( [ flow ] ) involve averages over a range @xmath262 and thus one naturally expects that , at least at the beginning of the flow , @xmath263 remains close to the @xmath28 solution .
the two functions will differ in a boundary layer around @xmath264 of width denoted by @xmath265 .
although the precise form of the solution for @xmath266 ( e.g. whether the cusp persists at @xmath205 ) is very hard to obtain analytically , fortunately most of our results will not depend on such details .
as we discuss below , the main issue will be to decide whether @xmath267 or not , which is a well - posed mathematical question .
let us start by analyzing the flow up to the larkin scale @xmath240 of the statics , at which the cusp occurs and the corrections to the friction become singular in the @xmath28 flow .
here at @xmath268 one enters at @xmath240 a regime where @xmath269 is close to its fixed point ( see appendix [ app : before ] ) . within the boundary layer
, the effect of the velocity is to decrease the singularities of the statics .
as shown in appendix [ app : v ] , the blow - up of the curvature @xmath270 is slowed down by the velocity as @xmath271 and the same is true for the friction @xmath272 if the blurring of the singularity results in a suppression of the cusp , i.e. if @xmath273 remains analytic , one should wonder whether the @xmath274 flow can really _ remain _ close to @xmath275 since the convergence to the fixed point is crucially dependent on the existence of the non - analyticity and in particular on the term @xmath276 in the flow of @xmath277 in ( [ deltav0 ] ) .
a hint that @xmath278 can stabilize for a while at @xmath205 is obtained by noting that one has ( see ( [ slowdel0 ] ) ) @xmath279 which has indeed the correct sign to give the same effect .
hence it is natural to expect for @xmath223 that @xmath280 has reached everywhere a fixed point form except in the boundary layer .
the correction to the friction , crucial to determine the @xmath1@xmath0 characteristics , reads @xmath281 and thus depends on the values of @xmath282 for @xmath283 . to estimate this expression
, one must know whether the width @xmath284 of the boundary layer is smaller than @xmath195 or not . to summarize these preliminary remarks , the flow in the _ larkin regime _
@xmath285 is similar to the @xmath28 flow and @xmath286 is close to @xmath287 except for @xmath266 .
we will now analyze in details the flow for @xmath223 under the assumption that @xmath288 as mentionned above , the validity of ( [ rho ] ) can in principle be established by a mathematical or a numerical analysis of our equations .
it turns out that ( [ rho ] ) leads to the most physically reasonable results .
the alternative case will be discussed below . for @xmath223 , called the _ depinning regime _ , and
relying on ( [ rho ] ) , the flow of @xmath70 becomes @xmath289 the friction is renormalized downwards with a non - trivial exponent @xmath290 with @xmath291 for the rf case ( see appendix [ rft ] ) and @xmath292 for the rp case ( see appendix [ rfp ] ) .
for the random bond one would naively take the static @xmath24 .
however our flow equations show that during the larkin regime , the form of the disorder correlator evolves to a rf , and thus @xmath293 also in this case .
this non trivial effect of the transformation for the dynamical properties of a rb into a rf is discussed in details in section [ depinningdiscussion ] .
since @xmath294 keeps on growing in the depinning regime , the assumption that @xmath280 can be replaced by @xmath287 will cease to be valid .
this occurs when @xmath195 reaches the range @xmath295 of @xmath255 , correlation length of the running disorder .
this defines a scale @xmath296 given by @xmath297 . above this scale ,
one enters a regime where the corrections due to disorder are simply washed out by the velocity , since the integrals over @xmath298 in ( [ flow ] ) average completely over the details of @xmath255 .
one thus enters the _ edwards wilkinson regime_. perturbation theory ( [ ew ] ) shows that the interface is flat for these large scales for @xmath299 , the disorder leading only@xcite to the effective temperature @xmath300 .
the family of systems indexed by @xmath301 have all the same velocity @xmath1 and the same slope @xmath302 .
however they have lesser and lesser singular behavior @xmath303 .
we can thus iterate the frg flow up to a point where the theory can solved perturbatively ( e.g. above @xmath304 ) . for the depinning
, one can simply use the fact that the renormalized action at @xmath305 is gaussian and its friction @xmath306 is , from ( [ exactslope ] ) equal to the slope @xmath307 of the depinning characteristics . using the flow of @xmath308 in ( [ flow ] ) , the expressions for @xmath309 , @xmath310 with ( [ tilde ] ) and ( [ chirfrf ] ) lead to @xmath311\\ \frac{f_c}{\eta_{l_{c}}v }
\end{array}\right.\ ] ] with @xmath312 which will turn to be the depinning exponent and we have defined a characteristic force @xmath313 . note that @xmath314 is not exactly the critical force @xmath15 . solving ( [ flowham ] )
gives @xmath315 since the system at @xmath316 is nearly pure , one has @xmath317 and , integrating over @xmath1 the derivative @xmath318 , one gets @xmath319 which shows that the depinning is characterized by an exponent @xmath16 and a pinning force @xmath320 ( yet to be determined ) .
the flow of @xmath321 allows to fix the value of @xmath15 . instead of just computing @xmath15
we also show that the integration of the flow of @xmath321 provides a second way to derive the depinning law ( [ depinexp ] ) . indeed ,
as discussed below , in our formalism the term proportional to @xmath1 which was problematic in the previous approaches @xcite cancels naturally . in the theory renormalized up to @xmath316 ,
the short scale cutoff is @xmath7 and one can use fisrt order perturbation theory .
one has ( see ( [ calvf ] ) and ( [ eq : d1 ] ) ) @xmath322 since in the renormalized theory the disorder is close to @xmath323 ( with the rescalings ( [ tilde ] ) ) , and the friction hidden in the response function is such that @xmath324 matches the range of @xmath325 , the velocity disappears from ( [ lvperp ] ) which gives @xmath326 where @xmath327 is some constant and the only @xmath1-dependent quantity is @xmath316 . to connect @xmath328 to the initial parameters ,
one has to integrate the flow @xmath329 .
expanding @xmath330 in ( [ flow ] ) at small velocity and using @xmath331 one recognizes in the second term the correction to @xmath242 .
thus for @xmath332 one has @xmath333 where we dropped the sub - dominant terms in velocity .
the integration of the flow gives @xmath334 where we defined @xmath335 and ( [ pertflv ] ) , we note that quite remarkably the @xmath336 _ cancel each other_. we are left with @xmath337 we already know ( [ etavetac ] ) that @xmath338 and ( [ rvrc ] ) @xmath339 thus both terms r.h.s of ( [ f - fc ] ) scale like @xmath340 .
this leads to the following result to lowest order in @xmath341 : @xmath342 where we used the fact that @xmath343 ( rf or rb ) or @xmath292 ( rp ) and the link between @xmath344 and @xmath52 ( rf ) or @xmath55 ( rp ) stated in appendix [ app : before ] .
in addition , we assumed that @xmath345 has a regular behavior when @xmath260 , a non - trivial point which we discuss in subsection [ open ] .
the approach of the previous subsection [ depinninglaw ] allows us to obtain the characteristics of the @xmath26 depinning .
we extract the depinning exponent @xmath16 , the pinning force @xmath15 and the characteristic lengthscales from the equation of motion without any additional physical hypothesis or scaling relation .
although the depinning problem the exponent @xmath16 and the critical force were determined in previous studies@xcite , our method is an improvement in several ways . to get the depinning exponent and critical force , two main derivations exist in the litterature
one of them extends the static frg formalism to the out of equilibrium depinning problem at zero temperature @xcite , using an `` expansion '' around an unknown mean - field solution@xcite . instead of directly looking at the renormalized correlator of the disorder , the method obliges to deal with the time correlation of the force , @xmath346 in ref . .
this procedure does not allow for a precise enough calculation of the @xmath1-@xmath0 characteristics to demonstrate the cancellation of of the @xmath347 term ( in our equation ( [ fv - f0],[f - fc ] ) ) . in order to obtain a depinning exponent @xmath124 different from its `` mean - field '' value @xmath348
, it is necessary in ref . to neglect _ by hand _ in the small @xmath1 limit a term proportionnal to @xmath1 against a term proportionnal to @xmath349 with @xmath350 .
our method , that directly uses averaging over the disorder and properly takes into account the velocity in the flow of the renormalized action , allows to show explicitly the needed cancellation
. the other analytical study @xcite of depinning does not consider the renormalization before the larkin length and assumes that the singularity is fully developped beyond this lengthscale .
this amounts to take as a starting point the equation of motion _ at zero velocity _ with a cuspy correlator for the force , and the larkin length as the microscopic cutoff .
since the anomalous exponent of the friction is @xmath351 which is ill defined for a cuspy correlator , one is forced in this method to argue that it should be replaced by @xmath352 which is finite .
this prescription and scaling relations linking the roughness , the depinning and the time exponent @xmath353 , allows to extract the depinning exponent . in our method , the ambiguities that existed in ref .
to write the flow of @xmath242 beyond @xmath354 when using the zero velocity equations , and the trick @xmath355 becomes a well - defined mathematical property of our finite velocity rg equations : if ( [ rho ] ) is confirmed , our approach directly shows that the @xmath356 prescription is the correct one and allows to _ prove _ directly the scaling relations , instead of _ assuming _ them , to obtain the exponent .
furthermore the occurence of the asymptotic edwards wilkinson regime in ref . has to be put by hand as a cut in the @xmath28 rg flow .
the important correlation length @xmath7 ( denoted @xmath357 in ref . ) at which this regime takes place is thus not well under control and has to be estimated from dimensional analysis . in our case
the depinning regime is naturally cut when our @xmath194 , which tells how fast the system runs on the disorder , reaches the range of the flowing correlator .
the scale @xmath7 at which it occurs , and above which the non - linearities are washed out , can clearly be identified with the correlation length of the moving interface ( or more precisely , of the velocity - velocity correlation ) .
the physical interpretation of ( [ rvrc ] ) , i.e. , @xmath358 with @xmath359 is the following : @xmath7 is the scale at which `` avalanches '' occur in the driven deterministic sytem .
the motion proceeds in a succession of such processes , where pieces of interface of typical size @xmath7 depin over a distance @xmath360 during a time @xmath361 .
in addition to providing a clean derivation of the depinning exponents and of the critical force , our equations contain new physics that was unreachable by the previous methods .
although in principle one would expect three universality classes ( rf , rb , rp ) for the depinning exponent , it was conjectured by narayan and fisher @xcite that the roughness exponent of the system at the depinning transition for rb or rf is equal to the roughness exponent of the static rf case , @xmath362 .
this result _ can not _ be obtained by the approach of narayan and fisher or that of nattermann _ et al . _ since these authors did not include the velocity in their rg analysis , and simply treated the small @xmath1 limit as @xmath28 . on the contrary ,
our flow equations for the correlator shows directly that a rb disorder does indeed evolve during the flow towards a rf disorder , leaving only two different universality classes ( rf , rp ) for the dynamics against three for the statics ( rb , rf , rp ) .
such evolution is shown on figure [ rbrfpic ] , where an initial rb becomes _ dynamically _ a rf . in appendix
[ app : v ] we show that the correction to @xmath363 , which measures the rf strength of the disorder , grows as @xmath364 where we have used @xmath365 .
this ensures that a moving system , even at _ arbitrary small _ velocity , sees an effective _ random field _ at large scale .
our frg equations prompt for several remarks and questions . in the previous sections ,
we have examined the consequences of the property ( [ rho ] ) and established in that case that the values of the exponents were the ones proposed in refs . .
although we consider it as unlikely , we have not been able to rule out the possibility that either @xmath366 ( or even worse , @xmath367 ) and thus we should examine the consequences of a violation of property ( [ rho ] ) .
if @xmath368 , it is not excluded a priori that there exists another `` fixed point '' behavior ( e.g. with a scaling function of @xmath369 ) .
however in that case , the exponents should differ from the standard ones ( unless some hidden and rather mysterious sum rule would fix the value of the integral in ( [ etaflow ] ) ) . in the absence of an identified fixed point , it is not clear whether universality would hold . again this crucial point ( [ rho ] ) can be definitely answered by an appropriate integration of ( [ flow ] )
. thus the present approach , which clearly takes @xmath1 into account , identifies as ( [ rho ] ) the condition under which the trick used in refs . gives the correct exponents .
another intriguing point concerns the continuity between the @xmath28 and the @xmath260 problems . indeed , to derive the depinning law ( [ depinningresult ] ) we have assumed that @xmath345 remains finite as @xmath260 . however , we should recall that in the non - driven case ( @xmath28 and @xmath73 ) , @xmath370 diverges at @xmath240 and thus @xmath371@xcite .
if there is any continuity in the rg flow as @xmath260 then @xmath372 in this limit . in that case the consequence would be ( see ( [ depinexp ] ) ) a modification of the exponent @xmath373 if @xmath374 ( or weaker logarithmic multiplicative corrections ) .
we would then find for the depinning a different result from the conventional one .
since we are unable to solve analytically accurately enough the equation for @xmath242 around @xmath354 , one should resort to a numerical solution of our flow equations ( [ flow ] ) to resolve this question . using ( [ flow ] )
it is necessary to check that @xmath345 does not diverge as @xmath375 like a power of @xmath1 so as to recover the standard depinning exponent ( [ theexpo ] ) .
the question is of particular importance since , if really a finite - scale behavior , occuring near @xmath89 , would control the macroscopic asymptotic behavior , then again one could wonder whether universality would hold .
therefore , the description of depinning in terms of a standard critical phenomenon may be risky .
indeed as clearly appears in our frg approach , since the fixed point at @xmath28 is characterized by a _ whole function _
@xmath323 ( i.e. an infinite number of marginal directions in @xmath376 ) rather than a single coupling constant ( as in usual critical phenomena ) the effect of an additional relevant perturbation , here the velocity , can be more complex due the feedback of @xmath1 itself on the shape of the function during the flow .
this is particularly clear in the rb case which dynamically tranforms into rf .
we now deal with the non - zero temperature case .
the system can jump over any energy barrier and overcome the pinning forces , thus it _ moves _ with @xmath205 for any drive @xmath377 and never gets pinned .
let us now show how our equations ( [ flow ] ) allow to investigate the _ creep _ regime that occurs when the system moves very slowly with @xmath378 , at low temperature . as for the depinning , we are interested in infinitesimal velocities .
the bare @xmath379 is thus very small .
the main difference compared to section [ depinninglaw ] is that the temperature is now finite as well .
the main effect of @xmath29 is to round the cusp in the flow .
since we are interested in extremely small velocities , we will consider @xmath380 as the smallest quantity to start with .
a non - zero temperature gives thus rise to a new regime in the rg flow , where the rounding of the cusp is due to temperature and not to velocity .
this leads to the following regimes in the frg flow shown on figure [ scales ] .
we will examine the various regimes in the rg flow keeping in mind that again , @xmath195 increases monotonically with @xmath189 .
just as in the previous case , we expect a _ larkin regime _ for @xmath381 with small corrections . above @xmath354
the disorder reaches a regime where scaling is imposed by the temperature . indeed
since @xmath382 one can forget about the velocity and the frg equations are very similar to the @xmath28 and @xmath2 case . in appendix
[ app : temperature ] we show that the temperature rounds the cusp on a boundary layer @xmath383 and we obtain the explicit scaling form ( [ scalingform ] ) @xmath384 which in the statics holds at all scales larger than a scale of order @xmath240 . here , because we focus on @xmath260 , the scanning scale @xmath385 is smaller than the width of the boundary layer , and the flow of the friction reads in this regime @xmath386 the temperature being irrelevant by power counting , the initial flow of @xmath236 is @xmath387 since the anomalous correction to @xmath236 vanish as @xmath388 . here and in the following
, @xmath389 denotes the energy fluctuation exponent of the _ static _ problem .
together with ( [ chi2surt ] ) it shows that the friction grows extremely fast , like @xmath390 .
this is the _ thermal regime _ where motion only occurs via thermal activation over barriers .
the velocity is so small that the center of mass motion is unimportant and the temperature essentially flows as in the @xmath28 problem .
we have determined the flow in its initial stages , and we now determine the scale at which this behavior ceases to hold . the flow equation ( [ etacorr ] ) for @xmath370 together with the scaling function ( [ scalingform2 ] ) for @xmath391 for @xmath392 shows that ( [ chi2surt ] ) holds only until the new scale @xmath393 defined as @xmath394 for @xmath395 the temperature remains the main source of rounding of the cusp .
above that scale one must take the velocity into account .
in fact , this simple picture is not complete since , before reaching @xmath396 another phenomenon occurs , leading to another lengthscale . in the thermal regime the correction to @xmath236 due to disorder competes with the simple exponential decay and ( [ ttheta ] ) breaks down .
this physically expresses that motion in a disordered landscape generates a thermal noise ( provided some thermal noise is already present ) .
using ( [ scalingform2 ] ) , one has @xmath397 at small @xmath194 .
thus the correction to @xmath236 reverts at a scale @xmath398 such that @xmath399 . note that @xmath400 . above @xmath401
the temperature does not decrease any more due to heating by motion .
one can show using ( [ flow ] ) that @xmath236 saturates and does not vary much until the scale @xmath402 .
we call this intermediate regime @xmath403 the _ saturation regime_. we checked it using a numerical integration of the flow in this regime with the scaling form of the disorder ( [ scalingform2 ] ) .
analytically , if we suppose that after @xmath401 , the correction of @xmath236 due to disorder dominates @xmath404 , then one would have in this regime an invariant of the flow @xmath405 .
if this were true , it is clear that the flow could _ never _ realize the condition @xmath406 , possibility that is excluded on physical basis and by the numerics shown in figure [ tsat ] . despite the saturation of the temperature , ( [ chi2surt ] )
remains true after @xmath401 .
thus the friction and @xmath407 keep on growing and one finally reaches the scale @xmath402 at which the scanning length @xmath408 crosses the boundary layer width @xmath259 . above @xmath402 , a rigorous analytical analysis of ( [ flow ] ) becomes difficult .
we however expect , since the velocity controls now the boundary layer , a regime similar to the depinning regime at @xmath26 to occur .
using the same arguments than for the depinning , one obtains in that regime @xmath409 leading again to a decrease of the temperature , even slightly accelerated by a negative @xmath410 exponent .
let us call @xmath411 the _ depinning scale _ at which one enters such a depinning regime . from the above discussion
it is very reasonable to expect that one goes directly from the saturation to the depinning regime , i.e. @xmath412 . however we can not strictly rule out the possibility of an intermediate regime ( divergent @xmath413 when @xmath260 ) during which the correction to the friction goes smoothly from positive ( as in the thermal and saturation regimes ) to negative values ( depinning regime ) .
again , it would be useful to settle this point through a numerical solution of our flow equations . note that in the rf and rp cases , the exponent @xmath188 and the fixed point @xmath275 in ( [ creepdep ] ) are the same as in the statics .
however in the rb case , one have used a @xmath189-dependent @xmath188 which crosses over between @xmath414 for @xmath395 and @xmath343 for @xmath415 corresponding to the change from rb to rf fixed points @xmath275 . in the depinning regime , motion
now proceeds in a similar way than for the one studied in section [ depinning ] . here again at large enough scale , velocity will wash out the disorder for @xmath416 with @xmath316 determined by @xmath417 .
one then enters the edwards - wilkinson regime .
let us now compute from the flow ( [ flow ] ) the lenghtscales defined above ( see figure [ scales ] ) .
in the thermal regime @xmath418 one can compute @xmath419 either by integrating its flow or by equating the boundary values to their expression .
this gives @xmath420\\ \left ( \frac{t}{u_{c}}\right)^ { 3/2}\frac{f_{c}}{\eta _ { l_{c}}v}e^ { -\frac{3}{2}\theta ( l_{s}-l_{c } ) } \end{array } \right .
\nonumber\end{aligned}\ ] ] where we defined @xmath421 and @xmath422 . expressing the scales as a function of the velocity leads to @xmath423 \label{rc1}\\ \frac{\eta _ { l_{s}}}{\eta
_ { l_{c}}}&\approx & \frac{f_{c}}{\eta _ { l_{c}}v } \left(\frac{u_{c}}{t } \right)^ { 1/\mu } \ln \left[\left ( \frac{t}{u_{c}}\right)^{3/2 } \frac{f_{c}}{\eta _ { l_{c}}v}\right]^ { -\frac{3}{2}-\frac{1}{\mu } } \label{etac1}\end{aligned}\ ] ] with @xmath424 .
in the saturation regime @xmath403 we proceed in the same manner and obtain @xmath425\\ \left ( \frac{u_{c}}{t}\right)^ { 1/2 } e^ { \frac{\theta}{2 } ( l_{s}-l_{c } ) } \end{array } \right .
\nonumber\end{aligned}\ ] ] thus @xmath426^{\frac{1}{2}}\label{eta1t}\end{aligned}\ ] ] assuming @xmath427 , the depinning regime @xmath428 follows directly and @xmath429\\ \frac{1}{\epsilon } \frac{u_{c}}{t } e^ { \theta ( l_{s}-l_{c } ) } \end{array } \right .
\nonumber\end{aligned}\ ] ] leads to @xmath430\right)^{\frac{1}{(2-\zeta ) \beta } } \label{rdv}\\ \frac{\eta _ { l_{v}}}{\eta _ { l_{d}}}&\approx & \left(\frac{1}{\epsilon } \ln \left[\left ( \frac{t}{u_{c}}\right)^{3/2 } \frac{f_{c}}{\eta _ { l_{c}}v}\right]\right)^{1-\frac{1}{\beta } } \label{etadv } \end{aligned}\ ] ] with @xmath431 the depinning exponent ( and @xmath188 the dynamical roughness exponent ) .
we are now in a position to compute the characteristics @xmath303 .
we fix a small velovity @xmath1 and solve the flow equations for @xmath195 , @xmath432 and @xmath433 up to @xmath316 .
this allows to relate @xmath434 to the unknown @xmath435 .
we can now use the fact that at the scale @xmath304 , the disorder is essentially washed out and a perturbative calculation of @xmath436 is possible . solving backwards we determine @xmath435 , wich is simply @xmath437 where @xmath0 is the real force applied on the system and @xmath438 the bare friction .
the correction to @xmath158 can not be neglected during the depinning regime , thus , using @xmath439 , @xmath440 and expressing @xmath441 one has @xmath442 in the thermal regime there is essentially no correction to the flow of @xmath158 .
thus ( [ theflowf ] ) is controled by the depinning regime and one should integrate essentially between @xmath443 and @xmath304 .
in fact due to the exponentially decreasing behavior of the integrand in ( [ theflowf ] ) the whole integral depends in fact only of the behavior _ at _ the scale @xmath443 . assuming that @xmath427 , using ( [ rc1],[r1 t ] )
, one sees that @xmath444 for @xmath260 and thus one obtains @xmath445\\ \mu = \frac{d-2 + 2\zeta _ { \rm eq}}{2-\zeta _ { \rm eq}}\end{aligned}\ ] ] the prefactor in front of the exponential can not be obtained reliably at this order .
note that for the creep , contrarily to the depinning , the possible divergence of @xmath446 when @xmath260 ( and @xmath239 ) does not affect the argument of the exponential but only the prefactor .
for the depinning it was possible to recover the depinning law using both the integration of the flow of @xmath158 and of the friction @xmath242 and the relation ( [ dvdfomrom ] ) .
although one can also use in principle this method for the creep it gives poor results in this case .
indeed contrarily to the derivation involving @xmath158 one needs here the flow of @xmath242 in _ all _ regimes including the depinning regime @xmath447 , where @xmath242 is still renormalized .
since the renormalization of @xmath242 goes from large positive growth ( first like @xmath390 , then exponentially ) in the thermal / saturation regime to negative in the depinning regime ( where the system accelerates with subdiffusive @xmath448 ) a precise knowledge of the behavior around @xmath396 would be needed .
unfortunately the lack of precise analytical methods available above @xmath449 prevents from computing precisely such a crossover .
a crude estimate of the flow can thus only give a bound of the exact result .
if we use ( e.g. in the rf or rp cases ) the estimates of each regime , and the perturbative estimate of @xmath450 in the theory at @xmath316 : @xmath451 ( it will appear that @xmath452 diverges faster than @xmath453 when @xmath260 ) .
the product of ( [ etac1],[eta1t],[etadv ] ) is equal to @xmath454 .
integrated from @xmath455 to @xmath1 , it yields @xmath456\end{aligned}\ ] ] one would thus find , using the @xmath242 method a non - arrhenius law for the creep regime .
even if one can not strictly speaking exclude this result , as discussed above it is most likely an artefact of the approximate integration of the flow , and only a lower bound of the barrier height .
indeed compared to the integration of the flow of @xmath158 , this procedure is much more sensitive to the neglect of the crossover @xmath457 .
a more precise integration of the flow would very likely show a compensation between the latent growth of the friction during the decrease of @xmath458 ( for @xmath457 ) and the reduction of the friction occuring in the depinning regime @xmath428 .
note that if @xmath459 were equal to @xmath460 then , one would recover ( [ creepformula ] ) .
it would be useful to check explicitely on a numerical integration of the flow that such a cancellation does occur and verify that the @xmath242 method confirms also the result ( [ creepformula ] ) .
we also note that the precise determination of the lenghtscales for @xmath461 depend on obtaining an accurate solution of the rg flow equations . in the previous section ,
we have obtained the formulas ( [ r1 t ] ) and ( [ rdv ] ) under some assumptions about the mathematical form of the solutions of the flow in the region where @xmath462 and @xmath463 cross .
these assumptions , discussed in the previous section , should be checked further , e.g via numerical integration .
although this should not affect the creep exponent derived above , the precise determination of these length scales is important to ascertain the exact value of the scale @xmath464 ( i.e the avalanche scale discussed below ) .
since our flow equations ( [ flow ] ) include finite temperature and velocity , they allow for the first time to treat the regime of slow motion at finite temperature , directly from ( [ notcomoving ] ) . as for the depinning
we derive directly from the equation of motion the force velocity law and we obtain new physics .
the first important result is of course the creep formula itself ( [ creepformula ] ) .
our method allows to prove the main physical assumptions , reviewed in subsection [ sccreep ] , needed for the phenomelogical estimate , namely : ( i ) the equal scaling of the barriers and the valleys ; ( ii ) the fact that velocity is dominated by activation over the barriers correctly described by an arrhenius law . in our derivation such law comes directly from the integration of the flow equations in the thermal regime ; ( iii ) the fact that one can use the static exponents in the calculation of the barriers .
this appears directly in the formula ( [ creepformula ] ) but can also be seen from the fact that in the thermal regime the velocity can essentially be ignored in the flow equations .
we also recover the characteristic lengthscale predicted by the phenomenological estimate .
indeed one can identify the scale ( [ r2d2 ] , [ r1 t ] ) @xmath465 as the @xmath137 of subsection [ sccreep ] .
our equations allow to obtain additional physics in the very slow velocity regime .
in particular , we see that the slow motion consists in two separate regimes . at small lengthscales
@xmath466 the motion is controlled by thermal activation over barriers as would occur at @xmath28 .
this is the regime described by the phenomenological theory of the creep .
qualitatively , the main novel result obtained here is that the thermally activated regime is followed by a depinning regime , as shown by our equations .
this leads to the following physical picture : at the length @xmath467 , bundles can depin through thermal activation .
when they depin they start an avalanche like process , reminiscent of the @xmath26 depinning , up to a scale @xmath7 .
the propagation of the avalanche proceeds on larger scales in a deterministic way .
thus one is left with a depinning - like motion , and the size of the avalanches is determined by the natural cut of the rg ( @xmath468 ) , i.e. , the scale at which the propagating avalanche motion is overcome by the regular motion of the center of mass .
one recovers qualitatively and quantitatively some features of the @xmath26 case at intermediate scale .
the typical nucleus jumps over an energy barrier @xmath469 resulting in @xmath470 .
this jump of a region of size @xmath449 initiates an avalanche spreading over a much larger size @xmath7 which we find to be ( see ( [ rc1],[r1t],[rdv ] ) ) @xmath471 with @xmath472 and @xmath473 the critical exponents of the depinning , and @xmath474 the energy exponent of the statics .
note that the correlation length @xmath7 diverges at small drive and temperature as @xmath8 with @xmath475 and @xmath476 . to push the analogy
further one can consider that the avalanches at lengthscales @xmath477 are similar to the ones occuring in a regular @xmath26 depinning phenomenon due to an excess driving force @xmath478 . considering a minimal block size @xmath449 instead of @xmath89 for this `` creepy '' depinning , @xmath479 ,
one obtains for this effective excess force : @xmath480 linking the creepy motion at @xmath2 and the threshold depinning at @xmath26 .
as explained before , there might be an uncertainty in the value of the avalanche exponent , which could be changed by a quantity of @xmath481 . to confirm ( [ rvrcrtrc ] )
, one would need to further check the precise behaviour of the solution of the rg equations for @xmath477 .
one can understand qualitatively that the problem at scale @xmath461 looks like depinning according to ( [ eff ] ) .
the tilted barrier ( see subsection [ sccreep ] ) @xmath482 to be overcome in order to move a region of size @xmath38 ( all barriers corresponding to smaller scales having been eliminated ) , vanishes at@xcite @xmath483 . for the @xmath26 depinning problem
, one can define a scale dependent effective threshold force @xmath484 such that @xmath485 ( also defined in subsection [ statics ] ) , which corresponds to the force needed to depin scales larger than @xmath38 ( the true threshold @xmath486 being controlled in that case by the larkin length ) .
a possible scaling derivation of ( [ eff ] ) is obtained by noting that at @xmath2 , _ non - activated _ motion at scale @xmath38 occurs when the tilted barrier @xmath487 is of the order of @xmath29 .
this yields a @xmath29 dependent effective threshold force such that @xmath488 at @xmath489 , one has @xmath490 and ( [ ftcrff ] ) is identical to ( [ eff ] ) to zeroth order in @xmath341 ( i.e. @xmath348 ) .
in fact , to apply the above static barrier argument , it might be better to work in the co - moving frame where the velocity of the interface vanishes .
this amounts to replace @xmath0 by @xmath437 in the previous argument , and @xmath491 , @xmath492 gives back ( [ eff ] ) .
the crossover between thermally activated processes and depinning - like motion can also be recovered by noting that the condition @xmath493 which appears in the frg flow can be rewritten as ( using ( [ tilde ] ) and ( [ lambdavtau ] ) ) : @xmath494 where the lhs is a natural energy scale involved in the depinning due to driving effect of the center of mass . if it is much larger than @xmath29 , depinning effects dominate , while if it is smaller , the dynamics is activated .
finally many open questions still remain
. technically it would be interesting to reconciliate the two methods based on @xmath242 and @xmath158 which proved to be equivalent for the study of depinning .
in fact although the two methods should formally agree , the comparison at a given order in the rg is more subtle .
indeed @xmath495 , by integration over @xmath189 between @xmath455 and @xmath496 and derivation with respect to @xmath1 , gives back @xmath497 provided that @xmath498 .
however , one should notice that in @xmath495 , the derivative is understood at fixed parameters _ at the given scale_. the occurence of this hidden dependence in the velocity in the running parameters makes the equivalence between both approaches delicate .
however the additional term is of higher order in disorder .
thus , as pointed out above , it is very likely that a careful integration of the flow of @xmath242 should resolve this discrepancy , but this remains to be explicitely checked . as for the @xmath26 depinning , the existence of the _ depinning regime _ at @xmath411 depends on the precise form of the boundary layer in the presence of a velocity .
note that the alternative scenario discussed in section [ open ] , e.g. whether or not the depinnig regime is universal , would not affect the creep exponents , but only the subleading corrections .
we examine in this paper the dynamics of disordered elastic systems such as interfaces or periodic structures , driven by an external force .
we take into account both the effect of a finite temperature and of a finite velocity to derive the general renormalization group equations describing such systems .
we extract the main features of the analytical solution to these equations both in the case of the @xmath26 depinning ( shown on figure [ vfinsp ] ) and in the `` creep '' regime ( small applied force @xmath0 and finite temperature ) .
our rg equations , when properly analyzed , allow to recover the depinning law @xmath499 and the depinning exponent @xmath16 also obtained by other methods .
however , contrarily to previous approaches that needed additional physical assumptions , such as scaling relations among exponents or by hand regularisation , our approach is self - contained , all quantities being derived directly from the equation of motion .
it thus provides a coherent framework to solve the difficulties and ambiguities encountered in the previous analytical studies@xcite .
in addition our method allows to establish the universality classes for driven systems .
it shows _ explicitely _ that a random bond type disorder gives rise close to a random field critical behavior at the depinning .
thus the dynamics is characterized by only two universality classes ( random field ( rf ) for interfaces and random periodic ( rp ) for periodic systems ) instead of three .
since this phenomenon is an intrinsically dynamical one , it was out of the reach of the previous analytical approaches that used @xmath28 flow equations together with additional physical prescriptions using e.g. the velocity as a cutoff on the @xmath28 rg flow . of course one of the great advantages of the present set of rg equations is to allow for the precise study of the small applied force regime at finite @xmath29 , for which up to now , only phenomenological scaling arguments could be given .
our frg study confirms the existence of a creep law at small applied force @xmath500\end{aligned}\ ] ] with a creep exponent related to the static ones @xmath23 , where @xmath24 is the statics roughening exponent .
it provides a framework to demonstrate , directly from the equation of motion , the main assumptions used in the phenomenological scaling derivation of the creep namely : ( i ) the existence of a single scaling for both the barriers and the minima of the energy landscape of the disordered system ; ( ii ) the fact that the motion is characterized by an activation ( arrhenius ) law over a typical barrier . in addition , our study unveils a novel `` depinning - like regime '' within the creep phenomena , not addressed previously , even at the qualitative level since the phenomenological creep arguments did not address what happens _ after _
the thermally activated jump of the optimal nucleus .
although the velocity is dominated by the time spent to thermally jump over the barriers , our equations show that the small @xmath0 behavior consists in fact of _ two different regimes_. up to a size @xmath467 motion can only occur through thermal activation over barriers .
this is the regime described by the phenomenological approach to the creep .
the optimal nucleus of the scaling estimate is given directly by the rg derivation as @xmath501 .
remarkably , another interesting regime exists above this lenghtscale ( see figure [ dessin ] ) .
it emerges directly from our rg equations and can be given the following simple physical interpretation . in some regions of the system , bundles of size @xmath467 depin due to thermal activation .
these small events then trigger much larger ones , and the motion above @xmath467 proceeds in a _ deterministic way _ , much as the @xmath26 depinning . in particular once the initial bundle depins it triggers an avalanche up to a size @xmath464 which is given by @xmath502 where @xmath13 , @xmath16 and @xmath125 are the energy , depinning and correlation length exponents respectively .
the present study also raises several interesting questions which deserve further investigation , some of them rely on being able to obtain a more accurate solution of our flow equations .
we have shown explicitly how to recover from our equations the conventional depinning law ( and the scaling creep exponents ) .
it rested on a mathematical property , likely to hold , but not yet rigorously established , of the solution for the flow of the correlator of the disorder .
such behavior should be checked in details .
the equations being quite complicated , a numerical solution , albeit delicate , seems to be appropriate .
if the constraint ( [ rho ] ) on the flow defined in section [ depinningdiscussion ] were found to be violated , then the conventional picture of the depinning would very likely fail , as we have analyzed in detail .
a similar question arises concerning the flow of the friction @xmath242 as discussed in section [ creepdiscussion ] .
if the solution of the flow is found to depend on the precise behavior at the larkin length @xmath103 , it is likely that even universality could be questioned .
these issues are a priori less important for the first , thermally activated , part of the creep regime , but because of the existence of a second , depinning - like regime , they would also have consequences for creep . again , these question depend on the precise form of the flow and can be answered unambiguously by a detailed enough analysis of our equations .
it would also be of great interest to develop a more detailed physical picture of the crossover between thermally activated and depinning like motion since we found that both occur within the creep phenomenon .
several applications and extensions of our work can be envisioned .
first , extensions to many - dimensional displacement field ( of dimension @xmath503 ) , given in appendix [ app : n>1 ] , would be interesting to study within the methods used here .
one could check whether the approximation used in ref . yields the correct result for the @xmath503 depinning .
second , the effect of additional kpz non - linearities could be investigated .
in particular one could check the usual argument which yields that kpz terms are unimportant for the depinning@xcite since their coupling constant is proportionnal to the ( small ) velocity .
also , extensions to other types of disorder , such as correlated disorder@xcite are possible .
finally , it should allow to describe in a systematic way the the thermal rounding of the depinning , i.e. the study of the @xmath1@xmath0 characteristics for @xmath0 close to the threshold and small @xmath29 .
if one assumes that one can simply carry naive perturbation theory in @xmath29 around the @xmath26 solution of the rg flow near @xmath15 ( i.e. only keeping the contribution beyond @xmath304 ) , one is led in ( [ f - fc ] ) to an additional term proportional to @xmath504 , which readily yields the value for the thermal rounding exponent @xmath505 proposed in ref .
( i.e. a scaling form near @xmath506 and small @xmath29 for the velocity @xmath507 ) .
although this exponent seems to be consistent with starting values @xmath508 , its validity could be further checked by solving our rg flow equations at small @xmath29 .
one of us ( tg ) would like to thank the newton institute ( cambridge ) for support and hospitality .
here are some notations and conventions and diagrammatics we use in the text .
the surface of the unit sphere in @xmath6 dimensions divided by @xmath509 is denoted by @xmath510 .
the thermal average of any observable @xmath327 is @xmath511 , the disorder average is @xmath512 , and the average with the dynamical action @xmath513 $ ] is denoted by @xmath514 .
the fourier transform of a function @xmath147 of @xmath515 is @xmath516 where @xmath517 , and the inversion reads @xmath518 , where @xmath519 , @xmath520 .
the fourier transform of the correlator @xmath521 is @xmath522 in general or @xmath523 in the periodic case .
one has thus @xmath524 , where @xmath525 or @xmath526 in the periodic case .
note that @xmath527 is a real and even function of @xmath48 .
the graphs are made of the following units ( see figure [ diagfirst ] ) : a full line between points @xmath515 and @xmath528 is a correlation @xmath529 , an _ oriented line _ with an arrow from point @xmath528 to point @xmath515 is a response @xmath530 ( the arrow means that @xmath531 , for the function does not vanish by causality ) .
the vertex is represented as a dashed line linking points @xmath515 and @xmath532 .
the dashed line means that both points have the same position @xmath31 . from each point
emerges a @xmath142 field .
no arrow is needed for the full line or for the dashed line , since they are symmetric with respect to the exchange of their endpoints .
the correlation being proportional to @xmath29 vanishes at @xmath26 .
the graphs renormalizing the disorder ( see figure [ diagsecord ] ) are made of vertices and responses , and they possess two external @xmath533 lines . it can be easily seen that arrows are no more necessary since the two external @xmath142 lines provide an orientation to all the responses of the graph . indeed , due to causality , each of the external @xmath534 is root of a tree , whose branches are response functions , which are oriented in the direction of the root .
we derive here the direct perturbation theory at @xmath2 without the use of the msr formalism . to organize the perturbation series ,
let us multiply the non - linear part of the equation of motion @xmath535 by a fictious small parameter @xmath536 , which will be fixed to one at the end of the calculation .
directly on @xmath537 we can formally expand @xmath538 , @xmath539 , solve recursively the system ( [ systemalpha ] ) , even at non - zero temperature , and compute the @xmath540-expansion of every observable .
note that we added a source @xmath147 ( with no constant uniform part ) so as to compute the response function .
as the force is gaussian , the expansion of disorder averaged quantities is in powers of @xmath541 , and is in fact an expansion in powers of @xmath75 .
we denote by @xmath542 the exact correlation and by @xmath543 the exact response functions . the first iterative steps are @xmath544 and @xmath545 these are sufficient to compute to first order in @xmath21 the force , the correlation and response .
in the absence of disorder the system moves with a linear characteristics @xmath157 and one has the following correlation and response @xmath546 related by the fluctuation - dissipation theorem ( fdt ) @xmath547 . note that @xmath548 and @xmath549 do _ not _ verify fdt at @xmath205 .
to first order in @xmath75 one obtains at @xmath26 @xmath550 these results can be extended to any temperature @xmath29 : @xmath551 where we have introduced the effective vertices , non - local in time , smoothed by the temperature ( see figure [ corrrespfirst ] ) @xmath552 we now want to compute the corrections to the parameters @xmath36 , @xmath242 , @xmath158 , @xmath29 , @xmath553 so that @xmath1 , @xmath554 , @xmath555 remain unchanged while the physical ( ultra - violet ) cutoff @xmath556 on the @xmath152 integrations is reduced .
to first order in @xmath21 and @xmath29 , one obtains , @xmath557 with @xmath558 and @xmath559 , @xmath560 are the on - shell gaussian response and correlation functions , i.e. , with modes @xmath152 lying only between @xmath561 and @xmath153 . a completely different way for obtaining the perturbation expansion is presented in refs .
, as a first attempt to include thermal fluctuations in the large - velocity expansion of ref . .
it consists in splitting the displacement field into a @xmath26 part and a thermal part .
this procedure is probably only true to first order in @xmath29 and not controlled at higher @xmath29 . instead
, the method presented here is really an expansion in disorder at any @xmath29 .
although the calculation can in principle be pushed to second order , the method is too cumbersome to do it in practice ( see however ref . at @xmath26 ) .
it is easier to use the formalism of dynamical field theory as shown in appendix [ app : derivation ] .
here we give the details of the renormalization procedure used for the moving system .
we use the msr formalism with action @xmath513 $ ] given by ( [ eq : action ] ) .
having shifted the field @xmath562 so that its average vanishes @xmath563 , we can do perturbation theory with the gaussian part @xmath564= \int_{rt}\left [ i\hat { u}_{rt}\left(\eta \partial_{t}-c\nabla ^{2 } \right)u_{rt}-\eta t i\hat { u}_{rt}i\hat { u}_{rt}\right]\end{aligned}\ ] ] of the action . the gaussian correlation @xmath565 and response @xmath566 functions were defined in appendix [ app : pt ] .
the interaction part of the action contains the disorder correlator and also the pinning force @xmath567 : @xmath568&=&-\tilde{f}\int_{rt } i\hat { u}_{rt}\\ & & -\frac{1}{2}\int_{rtt'}i\hat { u}_{rt}i\hat { u}_{rt}\delta ( u_{rt}-u_{rt'}+v ( t - t ' ) ) \nonumber\end{aligned}\ ] ] the effective action for slow fields @xmath569 is given by the following cumulant expansion where the averages are computed within the gaussian part @xmath570 over the fast fields @xmath571 @xmath572&=&s_{0}[u,\hat { u}]+\langle s_{\rm i}[u+u^{>},\hat { u}+\hat { u}^ { > } ] \rangle \nonumber \\ & & -\frac{1}{2}\langle s_{\rm i}[u+u^{>},\hat { u}+\hat { u}^{>}]^{2 } \rangle _ { \rm c } + { \cal o } ( s_{\rm i}^{3})\end{aligned}\ ] ] we now turn to the computation of the first and second order terms . to first order , the corrections arise from the graph shown in figure [ fig : diagfo ] : they read @xmath573\right>=-\tilde{f}\int_{rt}i\hat{u}_{rt}\nonumber \\ -\int_{rtt'\kappa } ( i\kappa ) \delta_{\kappa } [ u ] ( r , t , t ' ) r^{>}_{0t - t ' } i\hat{u}_{rt } \label{firstord}\\ -\frac{1}{2 } \int_{rtt'\kappa}\delta_{\kappa } [ u ] ( r , t , t ' ) i\hat{u}_{rt}i\hat{u}_{rt'}\nonumber\end{aligned}\ ] ] with the shorthand notation : @xmath574 ( r , t , t')\equiv \delta_\kappa e^{i\kappa(u_{rt}-u_{rt ' } + v(t - t ' ) ) } e^{(i\kappa)^2 ( c^{>}_{00}-c^{>}_{0t - t'})}\nonumber\end{aligned}\ ] ] the term ( [ firstord ] ) appears to be the sum of a @xmath575 $ ] term and a @xmath576 $ ] term
. let us begin to deal with the first type .
a short time expansion of @xmath577 yields the following operators @xmath578 which is a correction to @xmath158 and @xmath579 which is a correction to @xmath242 .
the elasticity operator @xmath580 is _ not _ corrected and _ no _ higher gradients like @xmath581 are generated in the equation of motion .
note also that to this order , no kpz term @xmath582 is generated@xcite .
the @xmath576 $ ] term can be rewritten as the sum of @xmath583 which has the form of a disorder correlator and yields a correction to @xmath521 , and an operator quasi local in time @xmath584 which yields a correction to the @xmath585 term .
the projection of ( [ quasiloc ] ) on this thermal noise operator is @xmath586 to obtain the correction to the temperature @xmath29 , one uses @xmath587 .
an integration by parts of ( [ etacorr ] ) , thanks to fdt for the `` pure '' @xmath38 and @xmath588 , yields @xmath589 .
to summarize , @xmath590 the correction @xmath591 has the same form as the perturbative expression for @xmath158 , with opposite sign and shell - restricted functions @xmath592 . note that @xmath593
. in the infinitesimal shell limit ,
the shell - restricted functions @xmath594 which are evaluated at @xmath595 are of order @xmath596 .
the differential flow is thus given by ( [ flow ] ) .
the fast - modes average @xmath597 can be decomposed into one term with @xmath158 in factor plus the rest which does not contain @xmath158 .
the former vanishes for the following reason : the contraction of the @xmath598 with the @xmath599 or @xmath600 contained in the vertex operator involves a fast response @xmath601 . but
@xmath602 , since its modes live in the shell .
the latter is the connected average of two disorder vertices .
we now extract from it a correction to the disorder , i.e. , a term which has the form @xmath603 .
the corresponding diagrams are represented in figure [ diagsecord ] .
each diagram has two external @xmath604 @xmath605 legs , to which corresponds a functional half vertex of @xmath562 and @xmath606 respectively .
calling @xmath607 , @xmath608 the ( positive ) time arguments of both response functions , denoting @xmath609 , the diagrams have the following analytical expressions , integrated over @xmath610 : @xmath611 after another short distance expansion of @xmath612 and @xmath613 , noting that @xmath614 , a proper symmetry counting yields the term of order @xmath615 of ( [ flow ] ) .
the results obtained here are consistent with the analysis of ref . .
we give here a detailed analysis of the functional renormalization group flow at @xmath2 and zero velocity .
the temperature is an irrelevant operator and flows exponentially fast to zero .
we show however that the temperature rounds the cusp in a region of size proportional to @xmath29 around the origin and that in this boundary layer , the disorder correlator takes a _ super - universal _ ( to lowest order in @xmath341 ) scaling form .
in addition we show how to carry a systematic expansion at low @xmath29 . as temperature decreases ,
the correlator of the disorder becomes more and more pinched , and eventually reaches its zero - temperature cuspy fixed point at infinity .
we show that during the renormalization at @xmath28 with a flowing temperature @xmath616 , the cusp forms only asymptotically ( @xmath617 ) , and @xmath521 has the following scaling form in the boundary layer @xmath618 @xmath619 with @xmath620 and where @xmath621 measures the cusp .
furthermore , we show that the following expansion in temperature for the solution of the frg flow holds @xmath622 thus we obtained a fairly complete picture of the solution .
the flow equation of the value at zero of the disorder correlator is @xmath623 since @xmath624 , the convergence of @xmath625 towards @xmath626 implies that @xmath627 also converges . from the fixed point equation @xmath628
one has simply @xmath629 , and thus , @xmath630 we show here that the assumption that the curvature at zero of @xmath631 diverges like a power of the inverse temperature implies that _ all _ the derivatives at zero also diverge and that there exists a well defined and particularly simple scaling function in the boundary layer around zero . precisely , for any function @xmath632 decreasing to zero and a function @xmath633 such that @xmath634 if @xmath635 for some @xmath636 and @xmath637 , then , defining the functions @xmath638 , we obtain that every derivatives of @xmath639 at @xmath640 converge to the corresponding derivatives of @xmath641 , and that @xmath0 is the only fixed possible fixed point for @xmath639 . a simple way to see the convergence to the scaling function @xmath0 is to write the flow of @xmath639 @xmath642 and eliminate at large @xmath189 the rhs term which is subdominant ( higher order in @xmath632 ) for @xmath636 , since @xmath29 has been absorbed in the variable @xmath643 of @xmath639 .
we have used that @xmath644 .
hence the fixed point equation for @xmath639 is @xmath645 which has the solution @xmath646 above since we know that @xmath647 , @xmath648 and @xmath649 is easily checked .
this is confirmed by the study of the flow equations for the successive derivatives @xmath650 : @xmath651 from a trivial recurrence , the hypothesis @xmath652 implies that @xmath653 converges for any @xmath211 .
moreover the limit @xmath654 can be obtained from ( [ eqdesa ] ) and is @xmath655 . to fix the value of @xmath540 ( @xmath656 as strongly suggested by ( [ alpha=1 ] ) ) , we checked that the only values of @xmath657 , @xmath658 such that @xmath659 has a meaningful fixed point are @xmath660 .
for these values , the fixed point is @xmath661 .
the procedure which gives us the leading behavior in the boundary layer controlled by temperature can be extended analytically with arbitrary accuracy in an expansion to any order in @xmath29 .
we study @xmath662 with @xmath663 . for numerical purposes or for the following analytical computation , it is useful to switch to the function @xmath664 which remains quadratic at the origin when @xmath239 , since @xmath665 for @xmath2 and @xmath666 for @xmath26 .
this function flows as @xmath667 we can replace the scale @xmath189 dependence of @xmath668 by a @xmath29 dependence since @xmath29 and @xmath669 are linked by @xmath670 .
the function @xmath671 can be expanded in @xmath672 the expansion begins at @xmath673 since @xmath674 and we have @xmath675 the equation for the @xmath676 s reads @xmath677 one can solve this equation order by order in @xmath29 .
it is useful to divide @xmath21 and @xmath29 by @xmath678 and @xmath40 by @xmath679 . with these rescaled quantities , we have simply @xmath680 and thus @xmath681 .
if we knew the full behavior of @xmath682 , i.e. , the @xmath683 s , we could completely solve the system . here , we get @xmath684 where we wrote @xmath685 such that the three first lines are functions which vanish and have zero curvature at zero .
note that while @xmath686 is universal , the last term @xmath685 contains un unknown integration constant @xmath687 which presumably depends on the initial condition of the flow and is thus not universal .
indeed we observed a non - universal @xmath687 in a numerical integration of the flow of @xmath688 .
the procedure can be carried to any order in @xmath29 and the all the @xmath676 s are accessible .
the unknown coefficients of the expansion @xmath689 are similarly non - universal .
both subsections [ exfp ] and [ scalingboundary ] thus provide a rather convincing and consistent picture for the solution of the @xmath2 , @xmath28 frg equations ( awaiting a mathematical proof ) .
we present here the analytical solutions of the fixed point equations for rf and rp at fixed @xmath29 .
thanks to the exact expression of these fixed points , we are able to check the scaling form derived in appendix [ app : temperature ] within an `` adiabatic '' hypothesis where the running correlator at @xmath189 is identified with the fixed point at @xmath690 .
our families of fixed temperature fixed points ( ftfp ) give back the known fixed points at @xmath26 in both the rf@xcite and the rp@xcite cases .
however , even if we obtain the same _ form _ ( [ x2/2=y-1-lny ] ) for the rf @xmath26 fixed point as in ref . , we disagree with the scaling in @xmath11 .
we look for a fixed point of @xmath691 with _ fixed _ @xmath29 and initial random field condition @xmath692 .
since @xmath693 , a meaningful fixed point can be obtained only for @xmath343 . fixing the rf strength @xmath694 to one ,
we are led to the following problem : for any @xmath695 , find the fixed temperature fixed point function ( ftfp ) @xmath696 such that @xmath697 integrating ( [ rf1 ] ) from @xmath455 to @xmath496 yields @xmath698 , hence the ftfp has a cusp for @xmath26 and no cusp for @xmath699 . at @xmath26 , integrating ( [ rf1 ] ) from @xmath455 to @xmath40 and dividing by @xmath553 yields @xmath700 .
then , integrating again from @xmath455 to @xmath40 yields the @xmath26 ftfp , by imposing ( [ rf2 ] ) @xmath701 where the function @xmath702 is implicitely defined by@xcite @xmath703 since @xmath704 one has @xmath705 it is easy to compute the number @xmath706 .
note the behavior near @xmath455 given by @xmath707 thus @xmath708 .
note also the gaussian decrease of correlations at infinity @xmath709 .
an intriguing fact is the scaling of the @xmath710 fixed point with @xmath11 : its @xmath711 derivative at @xmath712 scales like @xmath713 at @xmath2 , there is no cusp ( @xmath714 ) and the same double integration of ( [ rf1 ] ) yields @xmath715 with @xmath716 implicitely defined by @xmath717 the value of @xmath718 is determined by condition ( [ rf2 ] ) . using @xmath719
, this condition reads @xmath720 this equation admits a unique solution @xmath721 for any @xmath722 . then there exists a unique ftfp @xmath696 for each @xmath2 . some of them are displayed in figure [ fig : fixedt ] .
note that @xmath26 in ( [ ytx],[defdel0 ] ) gives back the @xmath26 non - analytic fixed point @xmath723 ( [ del0u ] ) .
hence the set of ftfp has a nice @xmath239 limit , even if there is a qualitative difference between the cuspy @xmath26 ftfp and the analytic @xmath2 ftfps . as is obvious from their analytical expression , or from figure [ fig : fixedt ] , @xmath724 : the @xmath26 non - analytic fixed point is approached smoothly by the set of analytic fixed @xmath29 fixed points .
when @xmath29 approaches zero , the curvature of the ftfps at the origin goes to @xmath725 like @xmath726 with @xmath727 given by ( [ del00 ] ) .
we also checked that the @xmath696 converge when @xmath239 to the zero temperature fixed point with the predicted scaling form ( [ scalingform ] ) @xmath728 where @xmath729 is given by the @xmath26 ftfp equation @xmath730 .
some of the rf fixed @xmath731 fixed points are shown on the right bottom quarter of figure [ fig : fixedt ] , including the cuspy ( highest ) @xmath26 fixed point .
absorbing @xmath11 in @xmath29 and @xmath75 , we chose to plot the non - trivial solution to the most reduced problem @xmath732 to restore @xmath341 and @xmath363 , one simply has to note that the `` dimensions '' are @xmath733 and @xmath734 .
the left bottom of figure [ fig : fixedt ] shows @xmath735 as a function of @xmath29 .
this combination has a finite limit ( @xmath736 ) when @xmath239 . in the random periodic case
, the conservation of the period @xmath55 of @xmath75 requires @xmath292 . after a suitable rescaling , @xmath737 , @xmath738 and @xmath739
, the fixed point equation reads for the @xmath740-periodic function @xmath521 @xmath741 and is easily solved by quadrature , by analogy with a particle s position @xmath742 at time @xmath40 in a potential @xmath743 verifying @xmath744 .
the quadrature leads to the reciprocical function @xmath745 , parametrized by @xmath746 and @xmath29 , as a sum of two elliptic functions .
then , imposing the solution @xmath521 be @xmath740-periodic fixes @xmath746 as a function of @xmath29 .
the result is : * for @xmath747 the only solution is @xmath748 * for @xmath749 another solution arises , which resembles a cosinus function of linearly vanishing amplitude when @xmath750 .
this non - trivial solution has no cusp but becomes pinched as @xmath29 decreases ( growing curvature @xmath751 and higher harmonics ) .
as can be seen on the analytical expression ( not given here ) , @xmath752 .
in particular , it remains finite when the temperature vanishes .
* eventually for @xmath239 , the non - trivial solution uniformly tends to the zero temperature fixed point @xmath753 the temperature @xmath754 in our units is _ exactly _ the critical temperature @xmath755 of the random - field xy - model@xcite and the fixed points near @xmath756 reproduce the line of fixed points of this problem ( since we worked to second order , it is only an approximation ) . indeed in @xmath757 , the naive dimension of the temperature is zero and our ftfp has a direct physical meaning .
note that another random gradient term becomes relevant in @xmath757 but does not feed back on the flow of @xmath553@xcite .
we can now use the exact ftfps to check that an adiabatic hypothesis is consistent with the scaling form ( [ scalingform ] ) .
indeed , one can numerically check that the correlator with a flowing temperature has the ftfps have the scaling ( [ scalingform ] ) as @xmath239 . to conclude about the problem with a flowing temperature @xmath616
, it appears from these observations that no cusp occurs at finite scale for @xmath758 .
the cusp forms only asymptotically ( @xmath617 ) , with @xmath759 given by ( [ del00 ] ) and it obeys a scaling form in the boundary layer @xmath760 @xmath761 we note that the precise form of the flow of the temperature ( i.e. the value of @xmath474 ) only affects subdominant behavior ( i.e. the function @xmath685 in subsection ( [ nextt ] ) ) .
we give here a possible extension of the frg to a multi - dimensional displacement field .
this study generalizes the approach of ref . by including the effect of @xmath205 and @xmath2 in the flow . for periodic structures ,
a similar study of the multi - dimensional displacement field was shown in ref . to yield novel effects . in a @xmath762 dimensional space ,
we distinguish between the _ internal _ or _ longitudinal _ space of dimension @xmath6 , to which @xmath31 belongs , and the _ transverse _ space of dimension @xmath763 , to which @xmath40 belongs .
the elastic energy of an interface without overhangs defined by a height function @xmath30 is quadratic in @xmath764 of the form @xmath765 .
the disorder : the random bond ( rb ) case corresponds to a random potential @xmath766 , with correlations @xmath767 .
function @xmath768 is even , vanishes at @xmath264 and goes to a negative constant for @xmath66 .
the random field ( rf ) case corresponds to a force @xmath769 with correlations @xmath770 , where the @xmath771 vanish for @xmath772 .
a rb gives rise to a rf via @xmath773 and the correlators are related by @xmath774 .
note that this type of correlator deriving from a rb has @xmath775 .
finally , the random periodic case ( rp ) occurs when @xmath40 is defined up to a discrete set of translations forming a lattice of points @xmath776 , e.g. when @xmath40 is a phase , defined up to @xmath777 shifts . in this case
, the disorder is periodic and one has @xmath778 for any @xmath776 of the lattice ( or @xmath779 ) .
the overdamped dynamics is given by @xmath780 where @xmath242 is the friction tensor and @xmath112 a langevin noise , with correlations @xmath781 .
the tensor @xmath29 stands for the temperature(s ) of this out of equilibrium system .
we added a driving force @xmath782 perpendicular to the interface and a source field @xmath783 , as an external excitation . without assuming any symmetry ,
let @xmath784 and @xmath785 be the gaussian correlation and response functions .
we obtain by the same procedure as for the @xmath139 case the following first order corrections due to disorder @xmath786 using @xmath787 , @xmath788 , we write the on - shell corrections as ( with the matrix product @xmath789 ) @xmath790 the second order correction to @xmath75 reads @xmath791\nonumber\end{aligned}\ ] ] note that each of the first three terms are symmetric under @xmath792 and that the fourth is exchanged with the fifth under this symmetry .
then @xmath75 remains a correlator .
of course this second order correction to @xmath21 gives back the expression already computed for a @xmath793 interface if @xmath139 . at zero velocity ,
one gets the second derivative of the flow equation of balents and fisher @xcite .
if we assume that @xmath521 depends only on the component of @xmath40 parallel to the velocity and send @xmath1 to zero , then our expression reduces to the equations of ertas and kardar @xcite . to simplify the analysis ,
let us rely on the assumed symmetries of the system . if we suppose that the initial problem is rotationnally invariant , i.e. has @xmath794 symmetry , then the elasticity tensor @xmath36 , the friction tensor @xmath70 and the temperature tensor @xmath29 are only scalars and the force force correlator @xmath75 is covariant , _
i.e. _ , for any @xmath795 such that @xmath796 , @xmath797 . during the flow
, we expect from physical grounds that the running terms of the action will conserve their symmetries but the velocity @xmath1 which is fixed once for all selects a particular direction in transverse space .
the interesting symmetries are given by the little group of the velocity , i.e. , the transformations @xmath795 such that @xmath796 and @xmath798 .
then one may decompose the tensors on a basis involving @xmath1 ( one has only two frictions , temperatures , response and correlation functions and five@xcite @xmath799 s , functions of @xmath800 ) . unfortunately , the full problem can not be easily decoupled , even with the simplifications pointed out above .
no closed equation e.g. for the correlator restricted to displacements aligned with the velocity @xmath801 , has been found , and the problem even at zero temperature seems involved . the simplification used in ref . consists in assuming that @xmath75 does not depend on
the transverse coordinates .
this assumption reduces the problem to the @xmath139 case , and it would be interesting to solve at finite @xmath29 the behavior of transverse coordinates along the lines of our analysis .
the effect of a small velocity on the frg flow is mainly restricted to the boundary layer of width @xmath284 about the origin .
analytically , it is rather difficult to give an estimate of @xmath802 or to decide how @xmath803 precisely behaves in the boundary layer @xmath804 .
it is however possible to simplify the formidable second order correction to the disorder correlator , displayed in ( [ flow ] ) , and to obtain analytically several results , giving some hints about this behavior .
the @xmath805 term in ( [ flow ] ) is written under a form involving two integrations over @xmath298 , reflecting the presence of two response functions integrated over time .
after some integrations by part , the @xmath805 term becomes@xcite @xmath806 integrated over @xmath40 , this correction becomes @xmath807 for any non - crazy function @xmath21 , this expression is positive .
assuming that @xmath75 has no cusp , it can be safely expanded and we can check that it is of order @xmath808 : @xmath809 thus at @xmath810 , the integral of @xmath811 _ grows _ during the flow , whereas it was _ conserved _ in the statics . using ( [ oneint ] ) , one can also compare the flow of @xmath269 at small velocity to the cuspy @xmath28 flow .
in particular , one observes that the effect of the velocity is to reduce the blow - up of the curvature @xmath270 @xmath812 ( note that @xmath813 whereas @xmath814 ) .
the flow of the friction is similarly slowed down : @xmath815 the term @xmath276 in the @xmath28 flow of @xmath277 in ( [ deltav0 ] ) at @xmath28 is replaced at @xmath205 by @xmath816 which has the right sign ( using cauchy inequality ) to slow down the exponential growth of @xmath817 .
obtaining a numerical integration of the flow is a highly non - trivial quest@xcite , since all the interesting properties occur close to the origin yielding unaccurate results in real space . in fourier space ,
the number of harmonics to be retained is huge if one wants to focus on the quasi - cuspy behavior ( @xmath818 at the cuspy fixed point ) .
however , we obtained , at least at the beginning of the flow ( up to @xmath240 ) with small initial velocity , the curve shown in figure [ rbrfpic ] .
the initial condition was a rb disorder ( full line ) .
it is obvious on the snapshot ( dotted line ) close to @xmath240 that the flow transformed the rb into a rf .
we show here that at the scale @xmath240 , @xmath255 is very close to the static zero - temperature fixed point @xmath819 .
this can be checked numerically , even in the presence of a small temperature and small velocity .
analytically , one can not obtain an exact integration of the flow , but we can compare @xmath255 to the known @xmath819 by the following arguments .
let us take e.g. the rf case for which @xmath343
. for weak disorder one obtains from the integration of ( [ closed ] ) : @xmath820 where we have used @xmath821 .
we prove in appendix [ rft ] that @xmath822 verifies @xmath823 and that @xmath824 .
thus the range @xmath825 of @xmath287 verifies @xmath826 to determine the range of @xmath827 we use the fact that at the beginning of the flow one can neglect the non - linear term in the flow equation ( [ flow ] ) .
we are left with @xmath828 and thus the range of @xmath829 is simply @xmath830 a comparison of ( [ chirfrf ] ) and ( [ rangelc ] ) shows that the two ranges are similar .
furthermore , in the rf case @xmath230 is conserved by the flow at @xmath28 , and thus the similarity of ranges shows that the shape of @xmath831 is close to the shape of @xmath287 ( same integral , same range ) . note that for periodic structures we consider in this paper for simplicity only the case @xmath835 and thus the intermediate random manifold regime is absent@xcite .
the method can be straightforwardly extended to the general case .
note that although the initial problem does not distinguish between internal ( @xmath836 ) and height ( @xmath837 ) directions , an asymmetry occurs during the coarse - graining and it is useful to consider ( [ delta ] ) as a general starting point .
the large @xmath1 limit is simple only for the single component model when setting the internal space cutoff to zero . more generally , in this far from equilibrium regime
, other effects can become important , such as the generation of non - linear kpz type terms , or for periodic structures , effects related to the existence of a moving glass , as discussed elsewhere@xcite .
the divergence of the friction in the dynamical study of the statics ( @xmath73 , @xmath28 ) does not quantitatively feedback on the static observables but provides the qualitative information that the system is pinned in the sense that an infinite time is needed to escape the metastable states .
the scale @xmath838 in the frg , corresponding to the onset of depinning - like motion , is naturally identified with @xmath839 , and the frg relation @xmath840 compares with the scaling relation @xmath841 . | elastic systems driven in a disordered medium exhibit a depinning transition at zero temperature and a creep regime at finite temperature and slow drive @xmath0 .
we derive functional renormalization group equations which allow to describe in details the properties of the slowly moving states in both cases .
since they hold at finite velocity @xmath1 , they allow to remedy some shortcomings of the previous approaches to zero temperature depinning .
in particular , they enable us to derive the depinning law directly from the equation of motion , with no artificial prescription or additional physical assumptions , such as a scaling relation among the exponents .
our approach provides a controlled framework to establish under which conditions the depinning regime is universal .
it explicitly demonstrates that the random potential seen by a moving extended system evolves at large scale to a random field and yields a self - contained picture for the size of the avalanches associated with the deterministic motion . at finite temperature @xmath2
we find that the effective barriers grow with lenghtscale as the energy differences between neighboring metastable states , and demonstrate the resulting activated creep law @xmath3 where the exponent @xmath4 is obtained in a @xmath5 expansion ( @xmath6 is the internal dimension of the interface ) .
our approach also provides quantitatively a new scenario for creep motion as it allows to identify several intermediate lengthscales .
in particular , we unveil a novel `` depinning - like '' regime at scales larger than the activation scale , with avalanches spreading from the thermal nucleus scale up to the much larger correlation length @xmath7 .
we predict that @xmath8 diverges at small drive and temperature with exponents @xmath9 that we determine . 2 |
while both undirected and directed graphs are important in many applications , there are significantly more algorithmic and structural results for undirected graphs than for directed ones .
the main reason is likely to be the fact that most problems on digraphs are harder than those on undirected graphs .
the situation has begun to change : recently there appeared a number of important structural results on digraphs , see e.g. @xcite .
however , the progress was less pronounced with algorithmic results on digraphs , in particular , in the area of parameterized algorithms . in this paper , we introduce a new decomposition for digraphs and show its usefulness by solving an open parameterized problem on digraphs by bang - jensen , saurabh and simonsen @xcite .
we believe that our decomposition will prove to be helpful for obtaining further algorithmic and structural results on digraphs .
a digraph @xmath2 is an _ out - tree _
( an _ in - tree _ ) if @xmath2 is an oriented tree with just one vertex @xmath3 of in - degree zero ( out - degree zero ) .
the vertex @xmath3 is the _ root _ of @xmath4 a vertex @xmath5 of an out - tree ( in - tree ) is called a _ leaf _ if it has out - degree ( in - degree ) zero .
if an out - tree ( in - tree ) @xmath2 is a spanning subgraph of a digraph @xmath6 then @xmath2 is an _ out - branching _ ( an _ in - branching _ ) of @xmath0 .
it is well - known that a digraph @xmath0 contains an out - branching ( in - branching ) if and only if @xmath0 has only one strongly connected component with no incoming ( no outgoing ) arc @xcite .
a well - known result in digraph algorithms , due to edmonds , states that given a digraph @xmath0 and a positive integer @xmath7 , we can decide whether @xmath0 has @xmath7 arc - disjoint out - branchings in polynomial time @xcite .
the same result holds for @xmath7 arc - disjoint in - branchings .
inspired by this fact , it is natural to ask for a mixture " of out- and in - branchings : given a digraph @xmath0 and a pair @xmath8 of ( not necessarily distinct ) vertices , decide whether @xmath0 has an arc - disjoint out- branching @xmath9 rooted at @xmath10 and an in - branching @xmath11 rooted at @xmath5 .
we will call this problem arc- disjoint branchings .
thomassen proved ( see @xcite ) that the problem is np - complete and remains np - complete if we add the condition that @xmath12 .
the same result still holds for digraphs in which the out - degree and in - degree of every vertex equals two @xcite .
the problem is polynomial - time solvable for tournaments @xcite and for acyclic digraphs @xcite .
the single - root special case ( i.e. , when @xmath12 ) of the problem is polynomial time solvable for quasi - transitive digraphs is quasi - transitive if for every @xmath13 there is at least one arc between @xmath14 and @xmath15 , i.e. either @xmath16 or @xmath17 or both . ] @xcite and for locally semicomplete digraphs is locally semicomplete if for every @xmath18 there is at least one arc between @xmath19 and @xmath15 and for every @xmath20 there is at least one arc between @xmath19 and @xmath15 .
tournaments and directed cycles are locally semicomplete digraphs . ]
an out - branching @xmath21 and an in - branching @xmath22 are called _
@xmath1-distinct _ if @xmath23 .
bang - jensen , saurabh and simonsen @xcite considered the following parameterization of arc - disjoint branchings .
@xmath1-distinct branchings input : & a digraph @xmath0 , an integer @xmath1 .
+ question : & are there @xmath1-distinct out - branching @xmath21 and in - branching @xmath22 ?
they proved that @xmath1-distinct branchings is fixed - parameter tractable ( fpt)-distinct branchings means that the problem can be solved by an algorithm of runtime @xmath24 , where @xmath25 omits not only constant factors , but also polynomial ones , and @xmath26 is an arbitrary computable function .
the books @xcite are excellent recent introductions to parameterized algorithms and complexity . ]
when @xmath0 is strongly connected and conjectured that the same holds when @xmath0 is an arbitrary digraph .
earlier , bang - jensen and yeo @xcite considered the version of @xmath1-distinct branchings where @xmath21 and @xmath22 must have the same root . ] and asked whether this version of @xmath1-distinct branchings , which we call single - root @xmath1-distinct branchings , is fpt .
the problem abstracts a quite natural situation in wireless networks : such networks are best modeled as directed because of the inherent asymmetry in sending data ( a task consuming a lot of energy ) and receiving data ( a task that consumes very little energy ) .
rooted in - branchings in this network correspond to routing strategies for aggregation and rooted out - branchings for dissemination of data @xcite . in this scenario ,
the problem of finding an in- and out - branchings with few common arcs solves a natural congestion - type problem .
the first key idea of this paper is to relate @xmath1-distinct branchings to the problem of deciding whether a digraph has an out - branching with at least @xmath1 leaves via a simple lemma ( see lemma [ lemma : leaves - branch ] ) .
the lemma and the following two known results on out - branchings with at least @xmath1 leaves allow us to solve the problem of bang - jensen and yeo @xcite and to provide a shorter proof for the above - mentioned result of bang - jensen , saurabh and simonsen @xcite ( see theorem [ thm : strong ] ) .
@xcite[thm : alon ] let @xmath0 be a strongly connected digraph .
if @xmath0 has no out - branching with at least @xmath1 leaves , then the ( undirected ) pathwidth of @xmath0 is bounded by @xmath27 .
@xcite [ thm : kleaf ] we can decide whether a digraph @xmath0 has an out - branching with at least @xmath1 leaves in time @xmath28 .
the general case of @xmath1-distinct branchings seems to be much more complicated .
we first introduce a version of @xmath1-distinct branchings called @xmath1-rooted distinct branchings , where the roots @xmath3 and @xmath29 of @xmath21 and @xmath22 are fixed , and add arc @xmath30 to @xmath0 ( provided the arc is not in @xmath0 ) to make @xmath0 strongly connected .
this introduces a complication : we may end up in a situation where @xmath0 has an out - branching with many leaves , and thereby potentially unbounded pathwidth , but the root of the out - branching is not @xmath3 . to deal with this situation , our goal will be to _ reconfigure _ the out - branching into an out - branching rooted at @xmath3 . in order to reason about this process , we develop a new digraph decomposition we call the _ rooted cut decomposition_. the cut decomposition of a digraph @xmath0 rooted at a given vertex @xmath31 consists of a tree @xmath32 rooted at @xmath31 whose nodes are some vertices of @xmath0 and subsets of vertices of @xmath0 called _ diblocks _ associated with the nodes of @xmath32 . out strategy is now as follows . if @xmath32 is _ shallow _
( i.e. , it has bounded height ) , then any out - branching with sufficiently many leaves can be turned into an out - branching rooted at @xmath3 without losing too many of the leaves . on the other hand , if @xmath32 contains a path from the root of @xmath32 with sufficiently many non - degenerate diblocks ( diblocks with at least three vertices ) , then we are able to show immediately that the instance is positive .
the remaining and most difficult issue is to deal with digraphs with decomposition trees that contain long paths of diblocks with only two vertices , called _ degenerate _ diblocks . in this case , we employ two reduction rules which lead to decomposition trees of bounded height .
the paper is organized as follows . in the next section ,
we provide some terminology and notation on digraphs used in this paper . in section [ sec : scd ] , we prove theorem [ thm : strong ] . section [ sec : any ] is devoted to proving that rooted @xmath1-distinct branchings is fpt for all digraphs using cut decomposition and theorems [ thm : alon ] and [ thm : kleaf ] .
we conclude the paper in section [ sec : conclusion ] , where some open parameterized problems on digraphs are mentioned .
let us recall some basic terminology of digraph theory , see @xcite .
a digraph @xmath0 is _ strongly connected _
( _ connected _ ) if there is a directed ( oriented ) path from @xmath14 to @xmath19 for every ordered pair @xmath33 of vertices of @xmath0 .
equivalently , @xmath0 is connected if the underlying graph of @xmath0 is connected .
a vertex @xmath5 is a _ source _ ( _ sink _ ) is its in - degree ( out - degree ) is equal to zero .
it is well - known that every acyclic digraph has a source and a sink @xcite . in this paper , we exclusively work with digraphs , therefore we assume all our graphs , paths , and trees to be directed unless otherwise noted . for a path @xmath34 of length
@xmath35 we will employ the following notation for subpaths of @xmath36 : @xmath37 : = x_i\ldots x_j$ ] for @xmath38 is the _ infix _ of @xmath36 from @xmath39 to @xmath40 . for paths @xmath41 and @xmath42
we denote by @xmath43 their _
concatenation_. for rooted trees @xmath2 and some vertex @xmath44 , @xmath45 stands for the subtree of @xmath2 rooted at @xmath14 ( see figure [ fig : fins ] ) .
subtree notation @xmath45 for @xmath44 ( left ) and the fins @xmath46 for a path @xmath47 in @xmath2 ( right ) . ]
we will frequently partition the nodes of a tree around a path in the following sense ( _ cf .
_ figure [ fig : fins ] ) : let @xmath2 be a tree rooted at @xmath31 and @xmath48 a path from @xmath49 to some node @xmath50 .
the _ fins _ of @xmath36 are the sets @xmath51 defined as @xmath52 a set @xmath53 in a digraph @xmath0 is _ bi - reachable _ from a vertex @xmath31 if for all @xmath54 there exist two internally vertex - disjoint paths from @xmath31 to @xmath5 . given a digraph @xmath0 and a vertex @xmath31 , we can compute the set of vertices that are bi - reachable from @xmath31 in polynomial time using network flows .
let us prove a simple fact on a link between out / in - branchings with many leaves and @xmath1-distinct branchings , which together with a structural result of alon _ et al .
_ @xcite and an algorithmic result for the maximum leaf out - branching problem @xcite gives a short proof that both versions of @xmath1-distinct branchings are fpt for strongly connected digraphs .
[ lemma : leaves - branch ] let @xmath0 be a digraph containing an out - branching and an in - branching .
if @xmath0 contains an out - branching ( in - branching ) @xmath2 with at least @xmath55 leaves , then every in - branching ( out - branching ) @xmath56 of @xmath0 is @xmath1-distinct from @xmath2 .
we will consider only the case when @xmath2 is an out - branching since the other case can be treated similarly .
let @xmath56 be an in - branching of @xmath0 and let @xmath57 be the set of all leaves of @xmath2 apart from the one which is the root of @xmath56 .
observe that all vertices of @xmath57 have outgoing arcs in @xmath56 and since in @xmath2 the incoming arcs of @xmath57 are the only arcs incident to @xmath57 in @xmath2 , the sets of the outgoing arcs in @xmath56 and incoming arcs in @xmath2 do not intersect .
we will use the following dynamic programming result from @xcite .
[ lem : dp ] let @xmath58 be a digraph of ( undirected ) treewidth @xmath59
. then @xmath1-distinct branchings on @xmath58 can be solved in time @xmath60 .
note that if a digraph @xmath0 is a positive instance of single - root @xmath1-distinct branchings then @xmath0 must be strongly connected as an out - branching and an in - branching rooted at the same vertex form a strongly connected subgraph of @xmath0 .
[ thm : strong ] @xmath1-distinct branchings and single - root @xmath1-distinct branchings on strongly connected digraphs can be solved in time @xmath61 .
the proof is essentially the same for both problems and we will give it for single - root @xmath1-distinct branchings .
let @xmath0 be an input strongly connected digraph by theorem [ thm : kleaf ] using an @xmath28-time algorithm we can find an out - branching @xmath21 with at least @xmath55 leaves , or decide that @xmath0 has no such out - branching .
if @xmath21 is found , the instance of single - root @xmath1-distinct branchings is positive by lemma [ lemma : leaves - branch ] as any in - branching @xmath22 of @xmath0 is @xmath1-distinct from @xmath21 . in particular
, we may assume that @xmath22 has the same root as @xmath21 ( a strongly connected digraph has an in - branching rooted at any vertex ) .
now suppose that @xmath21 does not exist .
then , by theorem [ thm : alon ] the ( undirected ) pathwidth of @xmath0 is bounded by @xmath62 .
thus , by lemma [ lem : dp ] the instance can be solved in time @xmath61 .
in this section , we fix a digraph @xmath0 with terminals @xmath63 and simply talk about _ rooted out - branchings ( in - branchings ) _ whose root we implicitly assume to be @xmath3 ( @xmath29 ) . similarly , unless otherwise noted , a _ rooted out - tree ( in - tree ) _ is understood to be rooted at @xmath3 ( @xmath29 ) . the problem in which @xmath21 and @xmath22 must be rooted at @xmath3 and @xmath29
, respectively , will be called the problem . clearly , to show that both version of are fpt it is sufficient to prove the following : [ thm : main ] is fpt for arbitrary digraphs . in the rest of this section
, @xmath64 will stand for an instance of ( in particular , @xmath0 is an input digraph of the problem ) and @xmath58 for an arbitrary digraph .
let us start by observing what further restrictions on @xmath0 can be imposed by polynomial - time preprocessing .
let @xmath64 be an instance of . recall that @xmath0 contains an out - branching ( in - branching ) if and only if @xmath0 has only one strongly connected component with no incoming ( no outgoing ) arc .
as a first preprocessing step , we can decide in polynomial time whether @xmath0 has a rooted out - branching and a rooted in - branching . if not , we reject the instance .
note that this in particular means that in a non - rejected instance , every vertex in @xmath0 is reachable from @xmath3 and @xmath29 is reachable from every vertex .
next , we test for every arc @xmath65 whether there exists at least one rooted in- or out - branching that uses @xmath66 as follows : since a maximal - weight out- or in - branching for an arc - weighted digraph can be computed in polynomial time @xcite , we can force the arc @xmath66 to be contained in a solution by assigning it a weight of 2 and every other arc weight 1 .
if we verify that @xmath66 indeed neither appears in any rooted out - branching or in - branching , we remove @xmath66 from @xmath0 and obtain an equivalent instance of . after this polynomial - time preprocessing
, our instance has the following properties : there exists a rooted out - branching , there exists a rooted in - branching , and every arc of @xmath0 appears in some rooted in- or out - branching .
we call such a digraph with a pair @xmath63 _ reduced_.
lastly , by the following lemma we may assume that our instance is strongly connected by incurring a factor of two in the application of lemma [ lemma : leaves - branch ] .
[ lemma : strongly - connected - instance ] let @xmath64 be reduced and let @xmath67 be the digraph obtained from @xmath0 by adding the arc @xmath30 to it . then @xmath68 is a positive instance of if and only if so is @xmath64 .
furthermore , if @xmath67 contains an out - tree ( in - tree ) with at least @xmath7 leaves , then @xmath0 contains an out - tree ( in - tree ) with at least @xmath69 leaves .
for the first claim , simply note that any rooted out - branching of @xmath67 can not use the arc @xmath30 and the same holds for rooted in - branchings . for the second claim , assume @xmath2 is an out - tree in @xmath0 with @xmath7 leaves
. assume @xmath70 ( otherwise the claim follows trivially ) .
let @xmath71 and @xmath72 be the two out - trees obtained by deleting the arc @xmath30 from @xmath2 .
both are out - trees in @xmath0 and one of them contains at least @xmath69 leaves , as claimed .
the first claim of lemma [ lemma : strongly - connected - instance ] shows that we may assume that @xmath0 is strongly connected in .
this implies the following simple claim required for further references .
[ lem : treeext ] let @xmath0 be an input digraph of . then every rooted out - tree with @xmath73 leaves can be extended into a rooted out - branching with at least @xmath73 leaves . in summary
, we enforce the following properties for @xmath64 by polynomial- time preprocessing : 1 . every arc of @xmath0 is contained in at least one rooted in - branching or rooted out - branching , 2 .
@xmath0 is strongly connected .
we work towards the following win - win scenario : either we find an out - tree with @xmath74 leaves that can be turned into a rooted out - tree with at least @xmath55 leaves , or we conclude that every out - tree in @xmath0 has less than @xmath74 leaves .
we refer to the process of turning an out - tree into a rooted out - tree as a _
reconfiguration_. in the process we will develop a new digraph decomposition , the _ rooted cut - decomposition _ , which will aid us in reasoning about reconfiguration steps and ultimately lead us to a solution for the problem . to make the notion of a bi - reachable set easier to use ,
the decomposition will employ a slightly broader notion as follows .
let @xmath58 be a digraph with at least two vertices , and let @xmath75 such that every vertex of @xmath58 is reachable from @xmath31 .
let @xmath76 be the set of all vertices that are bi - reachable from @xmath31 .
the _ directed block ( diblock ) @xmath77 of @xmath31 in @xmath58 _ is the set @xmath78 $ ] , i.e. , the bi - reachable vertices together with all out - neighbors of @xmath31 and @xmath31 itself .
note that according to the above definition a diblock must have at least two vertices .
the following statement provides us with an easy case in which a reconfiguration is successful , that is , we can turn an arbitrary out - tree into a rooted out - tree without losing too many leaves .
later , the obstructions to this case will be turned into building blocks of the decomposition .
[ lemma : re - root ] let @xmath79 be the diblock of @xmath3 and let @xmath2 be an out - tree of @xmath0 whose root @xmath31 lies in @xmath80 with @xmath7 leaves .
then there exists a rooted out - tree with at least @xmath81 leaves .
we may assume that @xmath82 . in case
@xmath2 contains @xmath3 as a leaf , we remove @xmath3 from @xmath2 for the remaining argument and hence will argue about the @xmath83 remaining leaves . if @xmath31 is bi - reachable from @xmath3 , consider two internally vertex - disjoint paths @xmath84 from @xmath3 to @xmath31 .
one of the two paths necessarily avoids half of the @xmath83 leaves of @xmath85 ; let without loss of generality this path be @xmath36 .
let further @xmath57 be the set of those leaves of @xmath85 that do _ not _ lie on @xmath36 .
if @xmath86 , let @xmath87 .
we construct the required out - tree @xmath56 as follows : first , add all arcs and vertices of @xmath36 to @xmath56 .
now for every leaf @xmath88 , let @xmath89 be the unique path from @xmath31 to @xmath5 in @xmath85 and let @xmath90 be the segment of @xmath89 from the last vertex @xmath14 of @xmath89 contained in @xmath2 .
add all arcs and vertices of @xmath90 to @xmath56 .
observe that @xmath91 as @xmath5 can not be in @xmath56 already . since @xmath89 and thus @xmath90 contains no leaf of @xmath57 other than @xmath5 , in the end of the process , all vertices of @xmath57 are leaves of @xmath56 . since @xmath92 , the claim follows .
note that the definition of diblocks can be understood in terms of network flows .
let @xmath93 consider the vertex - capacitated version of @xmath58 where @xmath31 and @xmath5 both have capacity 2 , and every other vertex has capacity 1 , for some @xmath94 .
then @xmath5 is contained in the diblock of @xmath31 in @xmath58 if and only if the max - flow from @xmath31 to @xmath5 equals 2 .
dually , by menger s theorem , @xmath5 is _ not _ contained in the diblock if and only if there is a vertex @xmath95 such that all @xmath31-@xmath5 paths @xmath36 intersect @xmath10 .
this has the following simple consequence regarding connectivity inside a diblock .
[ lemma : bi - reachable - pair ] fix @xmath75 and let @xmath96 be the diblock of @xmath31 in @xmath58 .
then for every pair of distinct vertices @xmath97 , there exist an @xmath31-@xmath14-path @xmath98 and an @xmath31-@xmath19-path @xmath99 that intersect only in @xmath31 . if @xmath100 , then clearly the claim holds since every vertex in @xmath77 is reachable from @xmath31 .
otherwise , add a new vertex @xmath15 with arcs @xmath101 and @xmath102 , and note that the lemma holds if and only if @xmath15 is bi - reachable from @xmath31 .
if this is not true , then by menger s theorem there is a vertex @xmath103 , @xmath104 , such that all paths from @xmath31 to @xmath15 , and hence to @xmath14 and @xmath19 , go through @xmath5 .
but as noted above , there is no cut - vertex @xmath105 for @xmath31-@xmath14 paths , and no cut - vertex @xmath106 for @xmath31-@xmath19 paths .
we conclude that @xmath15 is bi - reachable from @xmath31 , hence the lemma holds .
next , we will use lemma [ lemma : bi - reachable - pair ] to show that given a vertex @xmath31 , the set of vertices not in the diblock @xmath77 of @xmath31 in @xmath58 partitions cleanly around @xmath77 .
[ lemma : partition - bottlenecks ] let @xmath75 be given , such that every vertex of @xmath58 is reachable from @xmath31 .
let @xmath107 be the diblock of @xmath31 in @xmath58 .
then @xmath108 partitions according to cut vertices in @xmath77 , in the following sense : for every @xmath109 , there is a unique vertex @xmath110 such that every path from @xmath31 to @xmath5 intersects @xmath77 for the last time in @xmath14 . furthermore , this partition can be computed in polynomial time . assume towards a contradiction that for @xmath109 there exist two @xmath31-@xmath5-paths @xmath111 that intersect @xmath77 for the last time in distinct vertices @xmath112 , respectively .
we first observe that @xmath113 , since the second vertices of @xmath114 and @xmath115 are contained in @xmath77 by definition . by lemma [ lemma
: bi - reachable - pair ] , we may assume that @xmath116 \cap p_2[r , x_2 ] = \{r\}$ ] .
but then @xmath114 and @xmath115 intersect for the first time outside of @xmath77 in some vertex @xmath117 ( potentially in @xmath118 ) .
this vertex is , however , bi - reachable from @xmath31 , contradicting our construction of @xmath77 .
hence there is a vertex @xmath119 such that every path from @xmath31 to @xmath5 intersects @xmath77 for the last time in @xmath14 , with @xmath120 , and clearly this vertex is unique .
finally , the set @xmath77 can be computed in polynomial time , and given @xmath77 it is easy to compute for each @xmath119 the set of all vertices @xmath121 ( if any ) for which @xmath14 is a cut vertex .
we refer to the vertices @xmath119 that are cut vertices in the above partition as the _ bottlenecks of @xmath77_. note that @xmath31 itself is not considered a bottleneck in @xmath77 . using these notions
, we can now define a _ cut decomposition _ of a digraph @xmath58 .
[ def : decomp ] let @xmath58 be a digraph and @xmath31 a vertex such that every vertex in @xmath58 is reachable from @xmath31 .
( @xmath31-rooted ) cut decomposition _ of @xmath58 is a pair @xmath122 where @xmath32 is a rooted tree with @xmath123 and @xmath124 , @xmath125 for each @xmath126 , is a collection of diblocks associated with the nodes of @xmath2 , defined and computed recursively as follows . 1 .
let @xmath77 be the diblock of @xmath31 in @xmath58 , and let @xmath127 be the set of bottlenecks in @xmath77 .
let @xmath128 be the corresponding partition of @xmath108 .
2 . for every @xmath129 ,
let @xmath130 be the @xmath14-rooted cut decomposition of @xmath131 $ ] .
@xmath32 is the tree with root node @xmath31 , where @xmath57 is the set of children of @xmath31 , and for every @xmath129 the subtree of @xmath32 rooted at @xmath14 is @xmath132 .
4 . @xmath133 . furthermore , for every node @xmath126 , we define @xmath134 as the set of all vertices contained in diblocks associated with nodes of the subtree @xmath132 .
figure [ fig : decomp ] provides an illustration to definition [ def : decomp ] .
[ fig : decomp ] [ lemma : decomposition - well - defined ] let a digraph @xmath58 and a root @xmath75 be given , such that every vertex of @xmath58 is reachable from @xmath31 .
then the @xmath31-rooted cut decomposition @xmath135 of @xmath58 is well - defined and can be computed in polynomial time .
furthermore , the diblocks cover @xmath136 , i.e. , @xmath137 , and for every node @xmath126 , every vertex of @xmath138 is reachable from @xmath14 in @xmath139 $ ] . by lemma [ lemma : partition - bottlenecks ] ,
the root diblock @xmath77 as well as the set @xmath140 of bottlenecks and the partition @xmath128 are well - defined and can be computed in polynomial time .
also note that for each @xmath129 , @xmath141 , and every vertex of @xmath142 $ ] is reachable from @xmath14 in @xmath143 by the definition of the partition .
hence the collection of recursive calls made in the construction is well - defined , and every digraph @xmath143 used in a recursive call is smaller than @xmath58 , hence the process terminates .
finally , for any two distinct bottlenecks @xmath144 we have @xmath145 .
thereby , distinct nodes of @xmath32 are associated with distinct vertices of @xmath58 , @xmath146 , and the map @xmath147 is well - defined .
it is also clear that the whole process takes polynomial time .
we collect some basic facts about cut decompositions .
[ lemma : cut - decomposition - props ] let a digraph @xmath58 and a vertex @xmath75 be given , and let @xmath148 be the @xmath31-rooted cut decomposition of @xmath58 .
then the following hold . 1 . the sets @xmath149 are all non - empty and partition @xmath150 .
2 . for distinct nodes
@xmath151 , if @xmath14 is the parent of @xmath19 in @xmath32 then @xmath152 ; in every other situation , @xmath153 .
3 . for every node @xmath126 ,
the following hold : [ item : cut - prop-3 ] 1 .
if @xmath19 is a child of @xmath14 in @xmath32 , then any arc leading into the set @xmath154 from @xmath155 will have the form @xmath156 where @xmath157 .
2 . if @xmath19 , @xmath158 are distinct children of @xmath14 in @xmath32 , then there is no arc between @xmath154 and @xmath159 . in particular , every arc of @xmath58 is either contained in a subgraph of @xmath58 induced by a diblock @xmath160 , or it is a _ back arc _ going from a diblock @xmath161 to a diblock @xmath160 , where @xmath14 is an ancestor of @xmath19 in @xmath32 .
for the first claim , the sets @xmath162 are non - empty by definition ; we show the partitioning claim . by lemma [ lemma : partition - bottlenecks ] , for every @xmath94 either @xmath163 or there is exactly one bottleneck @xmath119 such that @xmath164 in the construction of the decomposition . also note that in the latter case , @xmath165 since @xmath119 . applying the argument recursively and using that the diblocks cover @xmath136 , by lemma [ lemma : decomposition - well - defined ]
, we complete the proof of the partitioning claim .
for the second claim , the partitioning claim implies that if @xmath166 for distinct nodes @xmath151 , then either @xmath167 or @xmath168 , i.e. , @xmath5 must be a bottleneck .
this is only possible in the situation described . for claim 3(b ) , first consider the diblock @xmath77 and the partition @xmath169 given by lemma [ lemma : partition - bottlenecks ] .
we show that for any two distinct sets @xmath170 , @xmath171 of the partition , there is no arc between @xmath170 and @xmath171 .
suppose for a contradiction that there is such an arc @xmath172 , @xmath173 , @xmath174 .
by lemma [ lemma : bi - reachable - pair ] , there are paths @xmath98 and @xmath99 in @xmath77 that intersect only in @xmath31 , and by lemma [ lemma : decomposition - well - defined ] , there are paths @xmath175 from @xmath14 to @xmath10 in @xmath170 and @xmath89 from @xmath19 to @xmath5 in @xmath171 .
but then the paths @xmath176 and @xmath177 form two @xmath31-@xmath5 paths that are internally vertex - disjoint , showing that @xmath103 , contrary to our assumptions .
since the decomposition is computed recursively , this also holds in every internal node of @xmath32 . for claim 3(a ) , let @xmath172 be an arc such that @xmath178 and @xmath179 .
moreover , let @xmath180 and @xmath181 .
by construction of cut decomposition , there is a path @xmath182 from @xmath183 to @xmath158 in @xmath32 containing nodes @xmath14 and @xmath19 .
let @xmath184 be the second node in @xmath185 ( just after @xmath183 ) .
thus , there is a path @xmath36 from @xmath184 to @xmath5 in @xmath58 containing the vertices of @xmath185 apart from @xmath183 .
assume that @xmath186 .
then by lemma [ lemma : bi - reachable - pair ] , there is an @xmath183-@xmath10-path @xmath187 and an @xmath183-@xmath184-path @xmath188 of @xmath58 which intersect only at @xmath183 .
then @xmath189 and @xmath190 are internally vertex - disjoint paths from @xmath183 to @xmath5 .
this means that @xmath5 must be in @xmath191 , a contradiction unless @xmath192 , @xmath193 and @xmath168 . if @xmath194 , then @xmath36 and @xmath172 are internally vertex - disjoint paths from @xmath10 to @xmath5 .
this means that @xmath5 must be in @xmath195 , a contradiction unless @xmath192 and @xmath168 . as we saw , for every diblock @xmath161 , @xmath196 , any path `` into '' the diblock must go via the bottleneck vertex @xmath19 .
by induction , for any @xmath197 , every node of @xmath32 from @xmath31 to @xmath19 represents a bottleneck vertex that is unavoidable for paths from @xmath31 to @xmath5 .
more formally , the following holds in cut decompositions : [ lemma : ancestor - bottlenecks ] let @xmath198 be the cut decomposition of @xmath58 rooted at @xmath31 .
fix a diblock @xmath160 for @xmath126 .
consider a path @xmath36 in @xmath58 from @xmath31 to @xmath199 and let @xmath200 be the sequence of bottleneck vertices that @xmath36 encounters .
then @xmath201 with @xmath202 is the path from @xmath31 to @xmath14 in @xmath32 .
we prove the claim by induction over the depth @xmath203 of @xmath14 in @xmath32 .
if @xmath204 then any path from @xmath31 to @xmath103 contains @xmath31 itself and hence the base case for @xmath205 holds .
consider a diblock @xmath160 , @xmath126 where @xmath14 has distance @xmath203 to @xmath31 in @xmath32 and let @xmath19 be the parent of @xmath14 in @xmath32 .
we assume the induction hypothesis holds for diblocks at depth @xmath206 , hence it holds for @xmath161 in particular .
because @xmath207 , this implies that every path from @xmath31 to @xmath14 will contain all ancestors of @xmath14 in @xmath32 . since by construction every path from @xmath31 to a vertex @xmath199 needs to pass through @xmath14 , the inductive step holds .
this proves the claim . as an immediate consequence
, we can identify arcs in cut decompositions that can not participate in any rooted out - branching .
[ cor : marked - out - arcs ] let @xmath198 be the cut decomposition of rooted at @xmath31 and let @xmath208 be all the arcs that originate in a diblock @xmath160 and end in an ancestor @xmath5 of @xmath14 on @xmath32 . then for every out - tree @xmath2 rooted at @xmath31 we have @xmath209 .
fix a bottleneck vertex @xmath210 of the decomposition and let the arc @xmath172 be in an out - tree @xmath2 rooted at @xmath31 .
there must exist a path @xmath211 from @xmath3 to @xmath10 that is part of @xmath2 . by lemma [ lemma : ancestor - bottlenecks ] , this path will contain the vertex @xmath5 .
but then @xmath5 is an ancestor of @xmath10 in @xmath2 and therefore the arc @xmath172 can not be part of @xmath2 , which is a contradiction .
the decomposition actually restricts paths even further : a path that starts at the root and visits two bottleneck vertices @xmath33 ( in this order ) can not intersect any vertex of @xmath154 before visiting @xmath19 and can not return to any set @xmath212 , @xmath213 , after having left it .
[ lemma : between - bottlenecks ] let @xmath198 be the cut decomposition of @xmath58 rooted at @xmath31 . fix a diblock @xmath160 for @xmath126 . consider a path @xmath36 from @xmath31 to @xmath199 and let @xmath214 be the path from @xmath215 to @xmath216 in @xmath32 .
let further @xmath217 be the fins of @xmath185 in @xmath32 .
then the subpath @xmath218 \setminus \{x_{i+1}\}$ ] is contained in the union of diblocks of @xmath219 for @xmath220 . by lemma [ lemma : ancestor - bottlenecks ]
we know that the nodes of @xmath185 appear in @xmath36 in the correct order , hence the subpath @xmath218 $ ] is well - defined .
let us first show that the subpath @xmath218\setminus \{x_{i+1}\}$ ] can not intersect any diblock associated with @xmath221 . by lemma [ lemma : cut - decomposition - props ] ,
the only arcs from @xmath222 into diblocks below @xmath223 connect to the bottleneck @xmath223 itself .
since @xmath223 is already the endpoint of @xmath218 $ ] , this subpath can not intersect the diblocks of @xmath221 .
this already proves the claim for @xmath224 ; it remains to show that it does not intersect diblocks of @xmath225 for @xmath226 .
the reason is similar : since the bottleneck @xmath39 is already part of @xmath218 $ ] , this subpath could not revisit @xmath222 if it enters any diblock @xmath161 for a proper ancestor @xmath19 of @xmath39 in @xmath32 .
we conclude that therefore it must be , with the exception of the vertex @xmath223 , inside the diblocks of the fin @xmath219 .
[ cor : avoid - half ] for every vertex @xmath227 and every set @xmath228 of non - bottleneck vertices there exists a path @xmath36 from @xmath31 to @xmath10 such that @xmath229 . assume that @xmath157 and let @xmath230 be a path from @xmath202 to @xmath231 in @xmath32 . let further @xmath232 be the fins of @xmath185 in @xmath32 .
we partition the set @xmath233 into @xmath234 where @xmath235 for @xmath236 .
lemma [ lemma : between - bottlenecks ] allows us to construct the path @xmath36 iteratively : any path that leads to @xmath10 will pass through bottlenecks @xmath237 in succession and visit only diblocks associated with @xmath219 in the process . since there are two internally vertex - disjoint paths between @xmath237 for @xmath238
, we can always choose the path that has the smaller intersection with @xmath239 . stringing these paths together , we obtain the claimed path @xmath36 .
we want to argue that one of the following cases must hold : either the cut decomposition has bounded height and we can ` re - root ' any out - tree with many leaves into a rooted out - tree with a comparable number of leaves , or we can directly construct a rooted out - tree with many leaves . in both cases
we apply lemmas [ lemma : leaves - branch ] and [ lem : treeext ] to conclude that the instance has a solution .
this approach has one obstacle : internal diblocks of the decomposition that contain only two vertices .
let @xmath240 be the cut decomposition rooted at @xmath3 .
we call a diblock @xmath160 _ degenerate _ if @xmath14 is an internal node of @xmath32 and @xmath241 .
let us first convince ourselves that a long enough sequence of non - degenerate diblocks provides us with a rooted out - branching with many leaves .
[ lemma : win - non - degen - path ] let @xmath198 be the cut decomposition rooted at @xmath3 of @xmath58 and let @xmath19 be a node in @xmath32 such that the path @xmath242 from @xmath3 to @xmath19 in @xmath32 contains at least @xmath7 nodes whose diblocks are non - degenerate
. then @xmath58 contains an out - tree rooted at @xmath3 with at least @xmath7 leaves .
we construct an @xmath3-rooted out - tree @xmath2 by repeated application of lemma [ lemma : bi - reachable - pair ] .
let @xmath243 , @xmath244 , @xmath245 be a sequence of nodes in @xmath242 whose diblocks are non - degenerate , and for each @xmath246 let @xmath247 be the node after @xmath39 in @xmath242 .
we construct a sequence of @xmath3-rooted out - trees @xmath71 ,
@xmath244 , @xmath248 such that for @xmath238 , the vertex @xmath39 is a leaf of @xmath249 , and @xmath249 contains @xmath250 leaves .
first construct @xmath71 as a path from @xmath3 to @xmath243 , then for every @xmath251 we construct an out - tree @xmath252 from @xmath249 as follows .
let @xmath253 , which exists since @xmath222 is non - degenerate , and let @xmath254 , @xmath255 be a pair of paths in @xmath256 $ ] from @xmath39 to @xmath247 and to @xmath257 respectively , which intersect only in @xmath39 .
such paths exist by lemma [ lemma : bi - reachable - pair ] , and since @xmath39 is a leaf of @xmath249 , lemma [ lemma : ancestor - bottlenecks ] implies that @xmath249 is disjoint from @xmath258 . hence the paths can be appended to @xmath249 to form a new @xmath31-rooted out - tree @xmath252 in @xmath58 which contains a leaf in every diblock @xmath259 , @xmath260 .
finally , note that the final tree @xmath248 contains two leaves in @xmath261 , hence @xmath248 is an @xmath31-rooted out - tree with @xmath7 leaves .
the next lemma is the last assertion that we will use to prove shortly that is fpt for digraphs @xmath0 whose cut decomposition rooted at @xmath3 contains no degenerate diblocks .
[ lemma : win - bounded - height ] let @xmath198 be the cut decomposition of @xmath0 rooted at @xmath3 such that @xmath32 is of height @xmath203 and let @xmath2 be an out - tree rooted at some vertex @xmath31 with @xmath7 leaves
. then we can construct an out - tree @xmath262 rooted at @xmath3 with at least @xmath263 leaves .
assume that @xmath31 is contained in the diblock @xmath160 of the decomposition and let @xmath264 be a path from @xmath265 to @xmath266 in @xmath32 .
let @xmath57 be the leaves of @xmath2 and let @xmath267 . clearly , @xmath268 .
applying corollary [ cor : avoid - half ] with @xmath269 and @xmath270 , we obtain a path @xmath271 in @xmath0 from @xmath3 to @xmath31 that avoids half of @xmath272 .
we construct @xmath262 in a similar fashion to the proof of lemma [ lemma : re - root ] .
we begin with @xmath273 , then for every leaf @xmath274 , proceed as follows : let @xmath89 be the unique path from @xmath31 to @xmath5 in @xmath2 and let @xmath90 be the segment of @xmath89 from the last vertex @xmath14 of @xmath89 contained in @xmath262 .
add all arcs and vertices of @xmath90 to @xmath262 .
since @xmath89 and thus @xmath90 contains no leaf of @xmath272 other than @xmath5 , in the end of the process , all vertices of @xmath275 are leaves of @xmath262 . since @xmath276 , we conclude that @xmath262 contains at least @xmath277 leaves , as claimed .
the next lemma demonstrates that using lemma [ lemma : win - bounded - height ] and a number of other results we can prove that if the height of @xmath203 of the cut decomposition of @xmath0 is upper - bounded by a function in @xmath1 , then on @xmath0 is fpt .
this shows that to prove that in general it suffices to consider separately the cases of bounded @xmath203 and unbounded @xmath203 . to provide an appropriate bound on @xmath203 we will use further results on degenerate diblocks provided in section [ sec :
degenerate ] .
[ thm : boundedd ] let @xmath198 the cut decomposition rooted at @xmath3 of height @xmath203 .
if @xmath278 for some function @xmath279 of @xmath1 only , then we can solve on @xmath0 in time @xmath280 . by theorem [ thm : kleaf ] , in time @xmath281 we can decide whether @xmath0 has an out - branching with
at least @xmath282 leaves .
if @xmath0 has such an out - branching , then by lemma [ lemma : win - bounded - height ] @xmath0 has a rooted out - tree with at least @xmath55 leaves .
this out - tree can be extended to a rooted out - branching with at least @xmath55 leaves by lemma [ lem : treeext ] .
so by lemma [ lemma : leaves - branch ] , @xmath64 is a positive instance if and only if @xmath0 has a rooted in - branching , which can be decided in polynomial time .
if @xmath0 has no out - branching with at least @xmath282 leaves , by theorem [ thm : alon ] the pathwidth of @xmath0 is @xmath283 and thus by lemma [ lem : dp ] we can solve on @xmath0 in time @xmath280 .
( note that for the dynamic programming algorithm of lemma [ lem : dp ] we may fix roots of all out - branchings and all in - branchings of @xmath0 by adding arcs @xmath284 and @xmath285 to @xmath0 , where @xmath286 and @xmath287 are new vertices . ) here is a key notion for our study of degenerate diblocks .
let @xmath198 be a cut decomposition of @xmath0 .
we call a path @xmath185 in @xmath32 _ monotone _ if it is a subpath of a path from the root of @xmath32 to some leaf of @xmath32 .
we call a path @xmath185 in @xmath32 _ degenerate _ if it is monotone and every diblock @xmath160 , @xmath288 is degenerate .
let @xmath64 be a strongly connected reduced instance of . as observed in section [ sec : preprocessing ] , we can verify in polynomial time whether an arc participates in _ some _ rooted in- or out - branching .
let @xmath289 be those arcs that do not participate in any rooted out - branching and @xmath290 those that do not participate in any rooted in - branching .
since @xmath64 is a reduced instance , we necessarily have that @xmath291 , a fact we will use frequently in the following .
corollary [ cor : marked - out - arcs ] provides us with an important subset of @xmath292 : every arc that originates in a diblock @xmath160 of the cut decomposition and ends in a bottleneck vertex that is an ancestor of @xmath14 on @xmath32 is contained in @xmath292 .
let us first prove some basic properties of degenerate paths .
[ lemma : degen - path - props ] let @xmath293 be a degenerate path of @xmath198 of @xmath0 rooted at @xmath3 .
then the following properties hold : 1 .
every rooted out - branching contains @xmath294 , 2 .
every arc @xmath295 with @xmath296 is contained in @xmath292 , and 3 .
there is no arc from @xmath39 @xmath297 to @xmath161 in @xmath0 , where @xmath19 is a descendant of @xmath39 on @xmath32 , except for the arc @xmath298 .
every rooted out - branching contains in particular the last vertex @xmath245 of the path . by lemma [ lemma : ancestor - bottlenecks ]
, it follows that @xmath185 is contained in the out - branching as a monotone path , hence it contains @xmath294 .
consequently , no ` back - arc ' @xmath295 with @xmath296 can be part of a rooted out - branching and thus it is contained in @xmath292 . for the third property ,
note that all arcs from @xmath39 except back arcs are contained in @xmath222 .
since @xmath222 is degenerate there can be only one such arc .
for the remainder of this section , let us fix a single degenerate path @xmath293 .
we categorize the arcs incident to @xmath185 as follows : 1 .
let @xmath299 contain all ` upward arcs ' that originate in @xmath185 and end in some diblock @xmath161 where @xmath19 is an ancestor of @xmath243 , 2 .
let @xmath300 contain all ` on - path arcs ' @xmath295 , @xmath296 , and 3 .
let @xmath301 contain all ` arcs from below ' that originate from some diblock @xmath161 where @xmath19 is a ( not necessarily proper ) descendant of @xmath245 . by lemma [ lemma : degen - path - props ]
, this categorization is complete : no other arcs can be incident to @xmath185 in a reduced instance . by the same lemma
, we immediately obtain that @xmath302 .
we will now apply certain reduction rules to @xmath64 and prove in the following that they are safe , with the goal of bounding the size of @xmath185 by a function of the parameter @xmath1 .
* reduction rule 1 * if there are two arcs @xmath303 with @xmath304 , remove @xmath305 .
rule 1 is safe .
since @xmath64 is reduced , the arcs @xmath306 and @xmath305 can not be in @xmath292 .
pick any rooted out - branching @xmath2 that contains the arc @xmath305 . by lemma [ lemma : degen - path - props ]
, we have that @xmath307 , therefore we can construct an out - branching @xmath56 by exchanging the arc @xmath305 for the arc @xmath306 . since a )
no rooted in - branching contains either of these two arcs , b ) no out - branching can contain both , and c ) the number of leaves in @xmath2 and @xmath56 are the same , we conclude that @xmath308 is equivalent to @xmath64 and thus rule 1 is safe .
[ cor : few - useless - up - arcs ] let @xmath64 be reduced with respect to rule 1 . then we either find a solution for @xmath64 or @xmath309 .
let @xmath58 be the heads of the arcs @xmath310 .
since rule 1 was applied exhaustively , no vertex in @xmath58 is the head of two arcs @xmath310 ; therefore @xmath311 .
note that any arc in @xmath310 can not be contained in @xmath292 , therefore @xmath58 does not contain any bottleneck vertices . applying corollary [ cor
: avoid - half ] , we can find a path @xmath312 from @xmath3 to @xmath245 that avoids half of the vertices in @xmath58 .
thus we can add half of @xmath58 as leaves to @xmath312 using the arcs from @xmath310 .
thus if @xmath313 , we obtain a rooted out - tree with at least @xmath314 leaves ( this count includes @xmath245 ) , which by lemmas [ lemma : leaves - branch ] , [ lemma : strongly - connected - instance ] and [ lem : treeext ] imply that the original instance has a solution .
we conclude that otherwise @xmath315 .
[ lemma : a - plus - tails ] let @xmath293 be a degenerate path .
assume @xmath316 and that @xmath64 is reduced with respect to rule 1 .
let further @xmath317 be those vertices of @xmath185 that are tails of the arcs in @xmath299 .
we either find that @xmath64 has a solution or that @xmath318 . for every vertex @xmath319 with an arc @xmath320 , we construct a path @xmath321 from @xmath39 to @xmath29 that contains @xmath322 as follows : since @xmath323 , there exist a rooted in - branching @xmath324 that contains @xmath322 .
we let @xmath325 be the path from @xmath39 to @xmath29 in @xmath324 .
each path @xmath321 does not intersect vertices in diblocks of @xmath326 .
since @xmath321 leaves @xmath185 via the first arc @xmath322 , it can not use the arc @xmath298 . since this is the only arc that leads to vertices in diblocks of @xmath326 , the claim follows .
+ let us relabel the just constructed paths to @xmath327 such that they are sorted with respect to their start vertices on @xmath185 .
that is , for @xmath304 the first vertex of @xmath328 appears before the first vertex of @xmath329 on @xmath185 .
we iteratively construct rooted in - trees @xmath330 with the invariant that a ) @xmath331 has exactly @xmath250 leaves and b ) does not contain any vertex of @xmath185 below the starting vertex of @xmath328 . choosing @xmath332
clearly fulfills this invariant . to construct @xmath331 from @xmath333 for @xmath334 , we simply follow the path @xmath328 up to the first intersection with @xmath333 . since @xmath335 , such a vertex must eventually exist . by the above claim , @xmath328 does not contain any vertex below its starting vertex on @xmath185 , thus both parts of the invariant remain true . we conclude that @xmath336 is a rooted in - tree with @xmath7 leaves , where @xmath7 is the number of vertices in @xmath233 that have at least one upwards arc not contained in @xmath337 . for @xmath338 , lemmas [ lemma : leaves - branch ] , [ lemma : strongly - connected - instance ] and [ lem : treeext ] imply that the original instance has a solution . otherwise , @xmath339 . by corollary [ cor : few - useless - up - arcs ] , we may assume that @xmath340 .
taken both facts together , we conclude that either @xmath341 , or we can construct a solution . we will need the following simple observation .
[ lemma : root - below ] let @xmath293 be a degenerate path such that the in - branching root @xmath29 lies in some diblock of @xmath342 . then @xmath343 and @xmath344
. in every in - branching rooted at @xmath29 , a vertex @xmath39 , @xmath345 of @xmath185 will necessarily be the parent of @xmath346 .
accordingly , no arc originating at a vertex of @xmath185 can be part of any in - branching implying that @xmath347 . by lemma [ lemma : degen - path - props ] , @xmath300 is additionally contained in @xmath292 and thus in a reduced instance we necessarily have that @xmath344 .
[ lemma : a - zero - tails ] let @xmath293 be a degenerate path and assume that @xmath316 .
let further @xmath348 be those vertices of @xmath185 that are tails of the arcs in @xmath300 .
we either find that @xmath64 has a solution or we have @xmath349 . we will construct a rooted in - tree that contains @xmath350 arcs from @xmath300 .
since no rooted out - branching contains any such arc , this will prove that the instance is positive provided @xmath351 .
note in particular that we are not concerned with the number of leaves of the resulting tree .
first associate every vertex @xmath352 with an arc @xmath353 , where we choose @xmath354 to be the vertex closest to @xmath5 .
let @xmath355 be the heads of all arcs in @xmath299 and let @xmath356 be the tails of all arcs in @xmath299 .
let @xmath357 be the vertex that appears first on @xmath185 among all vertices in @xmath358 .
since @xmath359 , @xmath360 can not be contained in @xmath337 .
accordingly there exists a path @xmath175 from @xmath10 to @xmath29 that uses the arc @xmath360 .
note that @xmath175 leaves @xmath185 through an @xmath299-arc whose tail lies between @xmath361 and @xmath10 on @xmath185 .
note that the segment @xmath362 $ ] , by our choice of @xmath10 , does not contain any vertex of @xmath358 .
we now construct the _ seed in - tree _ @xmath363 as follows .
we begin with @xmath175 and add the arc @xmath364 for every vertex @xmath365 where @xmath366 .
next , we add every vertex @xmath367 to @xmath363 by finding a path from @xmath5 to @xmath29 and attach this path up to its first intersection with @xmath363 .
since @xmath5 lies above @xmath361 in the decomposition , this path can not intersect any vertex in @xmath358 .
we form an in - forest @xmath368 from the arcs of @xmath363 and all arcs @xmath369 , @xmath370 that are not in @xmath363 .
every in - tree @xmath371 has the following easily verifiable properties : 1 .
its root is the highest vertex in the decomposition among all vertices @xmath372 ( recall lemma [ lemma : cut - decomposition - props ] ) , 2 .
its root is either @xmath29 or a vertex @xmath354 with @xmath352 , and 3 .
every segment @xmath373 $ ] of @xmath185 contained in @xmath2 has no vertex of @xmath233 or @xmath358 with the possible exception of @xmath374 .
we will maintain all three of these properties while constructing a sequence of in - forests @xmath375 where each in - forest in the sequence will have less roots than its predecessor ( here @xmath376 stands for @xmath377 being a subgraph of @xmath378 ) .
we stop the process when the number of roots drop to one . the construction of @xmath379 from @xmath377 for @xmath226 works as follows .
let @xmath380 be the in - tree with the _ lowest _ root in the in - forest . by assumption , @xmath377 has at least two roots , thus it can not be @xmath29 and therefore , by part b ) of our invariant , is a vertex @xmath354 with @xmath352 . now from @xmath354 onwards , we walk along the path @xmath185 until we encounter a vertex @xmath15 that is either 1 ) the tail of an arc @xmath299 or 2 ) a vertex of the in - forest @xmath377 . in the former case ,
we use the segment @xmath381 $ ] to connect @xmath2 to a vertex in @xmath355 via the @xmath299-arc emanating from @xmath15 .
since all vertices in @xmath355 are part of the seed in - tree @xmath363 this preserves all three parts of the invariant and @xmath382 .
consider the second case : we encounter a vertex @xmath15 that is part of some tree @xmath383 .
let us first eliminate a degenerate case : the trees @xmath56 and @xmath2 are distinct . assume towards a contradiction that @xmath384 . in that case , there is no single vertex between @xmath354 and @xmath15 that is either the tail of an @xmath299 or @xmath300-arc .
if @xmath15 is such that @xmath385 ( for example @xmath386 ) , we already obtain a contradiction : the arc @xmath387 can not possibly be part of any in - branching rooted at @xmath29 , contradicting the fact that it is not in @xmath337 .
otherwise , @xmath388 for some @xmath389 . by assumption ,
@xmath390 , thus there exists a path @xmath391 which contains both the arc @xmath392 as well as the arc @xmath369 .
therefore , the vertex @xmath393 must lie below @xmath5 on @xmath36 .
furthermore , the path @xmath391 can not contain any vertex above @xmath354 by property a ) of the invariant .
it follows that the subpath @xmath394 $ ] is entirely contained in @xmath2by property c ) of the invariant , none of the vertices ( except @xmath5 ) in this subpath can be in @xmath233 or @xmath358 .
since the same was true for the subpath @xmath395 $ ] , we conclude that the whole subpath @xmath396 $ ] contains no vertex of @xmath233 or @xmath358 .
this contradicts our assumption that @xmath397 and we conclude that @xmath2 and @xmath56 must be distinct . + by our choice of @xmath2 , the vertex @xmath15 can not be the root of @xmath56 .
accordingly , we can merge @xmath2 and @xmath56 by adding the path @xmath398 $ ] .
this concludes our construction of @xmath379 .
since the root of @xmath56 lies above all vertices of @xmath2 , part 1 ) of the invariant remains true .
we did not change a non - root to a root in this construction , thus part 2 ) remains true .
the one segment of @xmath398 $ ] we added to merge @xmath2 and @xmath56 did not , by construction , contain any vertex of @xmath358 or @xmath233 , with the exception of the last vertex @xmath15 , hence part 3 ) remains true .
finally , we clearly have that @xmath382 and the latter contains one less root than the former .
the process clearly terminates with some in - forest @xmath399 which contains a single in - tree @xmath324 .
the root of this in - tree is necessarily @xmath29 .
note further that @xmath368 contained all @xmath350 arcs @xmath360 , @xmath357 ; therefore @xmath324 contains those arcs , too .
since all of these arcs are in @xmath292 , we arrive at the following : either @xmath349 , as claimed , or we found a rooted in - tree that avoids at least @xmath1 arcs with every rooted out - branching , in other words , as solution to @xmath64 .
taking lemma [ lemma : a - plus - tails ] and lemma [ lemma : a - zero - tails ] together , we see that only @xmath400 vertices of a degenerate path are tails of arcs in @xmath299 or @xmath300 .
the following lemma now finally lets us deal with degenerate paths : we argue that those parts of the path that contain none of these few ` interesting ' vertices can be contracted .
* reduction rule 2 * if @xmath373 \subseteq \hat p$ ] is such that no vertex in @xmath373 $ ] is a tail of arcs in @xmath401 , contract @xmath373 $ ] into a single vertex .
rule 2 is safe .
simply note that every vertex in @xmath373 $ ] has exactly one outgoing arc .
we already know that every arc of @xmath185 must be contained in every rooted out - branching , now we additionally have that all arcs in @xmath373 $ ] are necessarily contained in every rooted in - branching .
we conclude that rule 2 is safe .
we summarize the result of applying rules 1 and 2 .
[ lem : last ] let @xmath185 be a degenerate path in an instance reduced with respect to rules 1 and 2 . then @xmath402 . by lemmas [ lemma : a - plus - tails ] and [ lemma : a - zero - tails ] ,
the number of vertices on @xmath185 that are tails of either @xmath299 or @xmath300 is bounded by @xmath403 . between each pair of such vertices
, we can have at most one vertex that is not a tail of such an arc .
we conclude that @xmath404 , as claimed .
now we can prove the main result of this paper .
+ * proof of theorem [ thm : main]*. by lemma [ lemma : strongly - connected - instance ] , we may assume that @xmath0 is strongly connected .
consider the longest monotone path @xmath185 of @xmath32 . by lemma [ lemma :
win - non - degen - path ] , if @xmath185 has at least @xmath55 non- degenerate diblocks , then @xmath0 has a rooted out - tree with at least @xmath55 leaves .
this out - tree can be extended to a rooted out - branching with at least @xmath55 leaves by lemma [ lem : treeext ] .
thus , by lemma [ lemma : leaves - branch ] , @xmath64 is a positive instance if and only if @xmath0 has a rooted in - branching , which can be decided in polynomial time .
now assume that @xmath185 has at most @xmath1 non - degenerate diblocks . by lemma [ lem : last ]
we may assume that before , between and after the non- degenerate diblocks there are @xmath400 degenerate diblocks .
thus , the height of @xmath32 is @xmath405 .
therefore , by lemma [ thm : boundedd ] , the time complexity for theorem [ thm : main ] is @xmath406 .
we showed that the rooted @xmath1-distinct branchings problem is fpt for general digraphs parameterized by @xmath1 , thereby settling open question of bang - jensen _ et al .
the solution in particular uses a new digraph decomposition , the _ rooted cut decomposition _ , that we believe might be useful for settling other problems as well . we did not try to optimize the running time of the algorithm of theorem [ thm : main ] . perhaps , a more careful handling of degenerate diblocks may lead to an algorithm of running time @xmath61 . regarding the motivation by application wireless sensor networks
, it would further be interesting whether the problem can be computed or approximated in a distributed fashion .
let us complete the paper by mentioning some digraph problems , which are still open .
bang - jensen _ et al . _
@xcite asked what is the parameterized complexity of the following problem : given a strongly connected digraph @xmath0 and a number ( parameter ) @xmath1 , can we delete @xmath1 arcs from @xmath0 such that @xmath0 remains strongly connected .
earlier , bang - jensen and yeo @xcite asked whether the following problem is fpt : given a strongly connected digraph @xmath0 and a number ( parameter ) @xmath1 , can we path- contract @xmath1 arcs from @xmath0 such that @xmath0 remains strongly connected
. _ path - contraction _ of an arc @xmath407 of @xmath0 results in a digraph obtained from @xmath0 by replacing @xmath14 and @xmath19 with a new vertex @xmath5 , deleting all arcs incident to either @xmath14 or @xmath19 and adding arc @xmath172 if arc @xmath408 is in @xmath0 and arc @xmath409 if arc @xmath410 is in @xmath0 .
note that path - contraction preserves " directed paths .
consider the following problem introduced by bezkov , curticapean , dell and fomin @xcite .
for given vertices @xmath3 and @xmath29 of a digraph @xmath411 , and an integer ( parameter ) @xmath412 decide whether @xmath0 has an ( @xmath413)-path in @xmath0 that is at least @xmath1 longer than a shortest ( @xmath413)-path . if at least " is replaced by exactly " , then the problem is fpt @xcite .
however , it is unknown whether the original problem is even in xp .
k. brczi , s. fujishige , and n. kamiyama .
a linear - time algorithm to find a pair of arc - disjoint spanning in - arborescence and out - arborescence in a directed acyclic graph .
, 109(23 - 24):12271231 , 2009 . | an out - branching and an in - branching of a digraph @xmath0 are called @xmath1-distinct if each of them has @xmath1 arcs absent in the other .
bang - jensen , saurabh and simonsen ( 2016 ) proved that the problem of deciding whether a strongly connected digraph @xmath0 has @xmath1-distinct out - branching and in- branching is fixed - parameter tractable ( fpt ) when parameterized by @xmath1 .
they asked whether the problem remain fpt when extended to arbitrary digraphs .
bang - jensen and yeo ( 2008 ) asked whether the same problem is fpt when the out - branching and in - branching have the same root . by linking the two problems with the problem of whether a digraph has an out - branching with
at least @xmath1 leaves ( a leaf is a vertex of out - degree zero ) , we first solve the problem of bang - jensen and yeo ( 2008 ) .
we then develop a new digraph decomposition called the rooted cut decomposition and using it we prove that the problem of bang - jensen et al .
( 2016 ) is fpt for all digraphs .
we believe that the _ rooted cut decomposition _ will be useful for obtaining other results on digraphs . |
the _ phase retrieval problem _ of fourier analysis involves determining a function @xmath9 on @xmath2 from the modulus @xmath10 of its fourier transform @xmath11 .
this problem arises naturally and frequently in various areas of science , such as x - ray crystallography , electron microscopy , optics , astronomy , and remote sensing , in which only the magnitude of the fourier transform can be measured and the phase is lost .
( sometimes , as when reconstructing an object from its far - field diffraction pattern , it is the squared modulus @xmath12 that is directly measured . ) in 1984 , rosenblatt @xcite wrote that the phase retrieval problem arises in all experimental uses of diffracted electromagnetic radiation for determining the intrinsic detailed structure of a diffracting object . " today , the word all " is perhaps too strong in view of recent advances in coherent diffraction imaging . in any case , the literature is vast ; see the surveys @xcite , @xcite , @xcite , and @xcite , as well as the articles @xcite and @xcite and the references given there .
phase retrieval is fundamentally under - determined without additional constraints , which usually take the form of an a priori assumption that @xmath9 has a particular support or distribution of values .
an important example is when @xmath13 , the characteristic function of a convex body @xmath1 in @xmath2 . in this
setting , phase retrieval is very closely related to a geometric problem involving the _ covariogram _ of a convex body @xmath1 in @xmath2 .
this is the function @xmath14 defined by @xmath15 for @xmath16 , where @xmath17 denotes @xmath18-dimensional lebesgue measure and @xmath19 is the translate of @xmath1 by the vector @xmath20 .
it is also sometimes called the _ set covariance _ and is equal to the _ autocorrelation _ of @xmath0 , that is , @xmath21 where @xmath22 denotes convolution and @xmath6 is the reflection of @xmath1 in the origin . taking fourier transforms
, we obtain the relation @xmath23 this connects the phase retrieval problem , restricted to characteristic functions of convex bodies , to the problem of determining a convex body from its covariogram .
both the definition of covariogram and this connection extend to arbitrary measurable sets , but the reason for restricting to convex bodies will become clear . the covariogram was introduced by matheron in his book @xcite on random sets .
he showed that for a fixed @xmath24 , the directional derivatives @xmath25 , for all @xmath26 , of the covariogram of a convex body @xmath1 in @xmath2 yield the distribution of the lengths of all chords of @xmath1 parallel to @xmath27 .
this explains the utility of the covariogram in fields such as stereology , geometric tomography , pattern recognition , image analysis , and mathematical morphology , where information about an unknown object is to be retrieved from chord length measurements ; see , for example , @xcite , @xcite , and @xcite .
the covariogram has also played an increasingly important role in analytic convex geometry .
for example , it was used by rogers and shephard in proving their famous difference body inequality ( see ( * ? ? ? * theorem 7.3.1 ) ) , by gardner and zhang @xcite in the theory of radial mean bodies , and by tsolomitis @xcite in his study of convolution bodies , which via the work of schmuckenschlger @xcite and werner @xcite allows a covariogram - based definition of the fundamental notion of affine surface area . here
we effectively solve the following three problems . in each
, @xmath1 is a convex body in @xmath2 .
* problem 1 ( reconstruction from covariograms)*. construct an approximation to @xmath1 from a finite number of noisy ( i.e. , taken with error ) measurements of @xmath14 .
* problem 2 ( phase retrieval for characteristic functions of convex bodies : squared modulus)*. construct an approximation to @xmath1 ( or , equivalently , to @xmath0 ) from a finite number of noisy measurements of @xmath28 .
* problem 3 ( phase retrieval for characteristic functions of convex bodies : modulus)*. construct an approximation to @xmath1 from a finite number of noisy measurements of @xmath29 . in order to discuss our results
, we must first address the corresponding uniqueness problems . in view of ( [ prc ] ) , these are equivalent , so we shall focus on the covariogram .
it is easy to see that @xmath14 is invariant under translations of @xmath1 and reflection of @xmath1 in the origin .
matheron @xcite asked the following question , known as the _ covariogram problem _ , to which he conjectured an affirmative answer when @xmath30 .
_ is a convex body in @xmath2 determined , among all convex bodies and up to translation and reflection in the origin , by its covariogram ? _
the focus on covariograms of convex bodies is natural .
one reason is that mallows and clark @xcite constructed non - congruent convex polygons whose overall chord length distributions ( allowing the directions of the chords to vary as well ) are equal , thereby answering a related question of blaschke .
thus the information provided by the covariogram can not be weakened too much .
moreover , there exist non - congruent non - convex polygons , even ( see @xcite ) horizontally- and vertically - convex polyominoes , with the same covariogram , indicating that the convexity assumption also can not be significantly weakened .
interest in the covariogram problem extends far beyond geometry .
for example , adler and pyke @xcite ask whether the distribution of the difference @xmath31 of independent random variables @xmath32 and @xmath33 , uniformly distributed over a convex body @xmath1 , determines @xmath1 up to translations and reflection in the origin . up to a constant , the convolution @xmath34 is just the probability density of @xmath31 , so the question is equivalent to the covariogram problem . in @xcite
, the covariogram problem also appears in deciding the equivalence of measures induced by brownian processes for different base sets .
a detailed historical account of the covariogram problem may be found in @xcite .
the current status is as follows , in which determined " always means determined by the covariogram among all convex bodies , up to translation and reflection in the origin .
averkov and bianchi @xcite showed that planar convex bodies are determined , thereby confirming matheron s conjecture .
bianchi @xcite proved , by a long and intricate argument , that three - dimensional convex polyhedra are determined .
it is easy to see that centrally symmetric convex bodies are determined .
( in the symmetric case , convexity is not essential ; see ( * ? ? ?
* proposition 4.4 ) for this result , due to cabo and jensen . )
goodey , schneider , and weil @xcite proved that most ( in the sense of baire category ) convex bodies in @xmath2 are determined .
nevertheless , the covariogram problem in general has a negative answer , as bianchi @xcite demonstrated by constructing convex polytopes in @xmath2 , @xmath35 , that are not determined .
it is still unknown whether convex bodies in @xmath36 are determined .
none of the above uniqueness proofs provide a method for actually reconstructing a convex body from its covariogram .
we are aware of only two papers dealing with the reconstruction problem : schmitt @xcite gives an explicit reconstruction procedure for a convex polygon when no pair of its edges are parallel , an assumption removed in an algorithm due to benassi and dercole @xcite . in both these papers ,
all the exact values of the covariogram are supposed to be available .
in contrast , our first set of algorithms take as input only a finite number of values of the covariogram of an unknown convex body @xmath37 .
moreover , these measurements are corrupted by errors , modeled by zero mean random variables with uniformly bounded @xmath38th moments , where @xmath38 is at most six and usually four .
it is assumed that @xmath37 is determined by its covariogram , has its centroid at the origin , and is contained in a known bounded region of @xmath2 , which for convenience we take to be the unit cube @xmath39^n$ ] .
we provide two different methods for reconstructing , for each suitable @xmath40 , a convex polytope @xmath5 that approximates @xmath37 or its reflection @xmath41 .
each method involves two algorithms , an initial algorithm that produces suitable outer unit normals to the facets of @xmath5 , and a common main algorithm that goes on to actually construct @xmath5 . in the first method ,
the covariogram of @xmath37 is measured , _
multiple times _ , at the origin and at vectors @xmath42 , @xmath43 , where the @xmath44 s are mutually nonparallel unit vectors that span @xmath2 . from these measurements , the initial algorithm noisycovblaschke constructs an @xmath45-symmetric convex polytope @xmath46 that approximates @xmath47 , the so - called blaschke body of @xmath37 .
( see section [ prelim ] for definitions and notation . )
the crucial property of @xmath47 is that when @xmath37 is a convex polytope , each of its facets is parallel to some facet of @xmath47 .
it follows that the outer unit normals to the facets of @xmath5 can be taken to be among those of @xmath46 .
algorithm noisycovblaschke utilizes the known fact that @xmath48 , evaluated at @xmath49 , equals the brightness function value @xmath50 , that is , the @xmath51-dimensional volume of the orthogonal projection of @xmath37 in the direction @xmath27 .
this connection allows most of the work to be done by a very efficient algorithm , algorithm noisybrightlsq , designed earlier by gardner and milanfar ( see @xcite ) for reconstructing a @xmath45-symmetric convex body from finitely many noisy measurements of its brightness function .
the second method achieves the same goal with a quite different approach .
this time the covariogram of @xmath37 is measured once at each point in a cubic array in @xmath52^n$ ] of side length @xmath53 . from these measurements , the initial algorithm noisycovdiff(@xmath54 ) constructs an @xmath45-symmetric convex polytope @xmath46 that approximates @xmath55 , the difference body of @xmath37 .
the set @xmath56 has precisely the same property as @xmath47 , that when @xmath37 is a convex polytope , each of its facets is parallel to some facet of @xmath56 .
furthermore , @xmath56 is just the support of @xmath57 .
the known property that @xmath58 is concave ( a consequence of the brunn - minkowski inequality ( * ? ? ? * section 11 ) ) can therefore be combined with techniques from multiple regression .
algorithm noisycovdiff(@xmath54 ) employs a gasser - mller type kernel estimator for @xmath57 , with suitable kernel function @xmath54 , bandwidth , and threshold parameter .
the output @xmath46 of either initial algorithm forms part of the input to the main common algorithm noisycovlsq .
the covariogram of @xmath37 is now measured _ again
_ , once at each point in a cubic array in @xmath52^n$ ] of side length @xmath53 .
using these measurements , algorithm noisycovlsq finds a convex polytope @xmath5 , each of whose facets is parallel to some facet of @xmath46 , whose covariogram fits best the measurements in the least squares sense .
much effort is spent in proving that these algorithms are strongly consistent .
whenever @xmath37 is determined among convex bodies , up to translation and reflection in the origin , by its covariogram , we show that , almost surely , @xmath59 as @xmath60 , where @xmath61 denotes hausdorff distance .
( if @xmath37 is not so determined , a rare situation in view of the uniqueness results discussed above , the algorithms still construct a sequence @xmath62 whose accumulation points exist and have the same covariogram as @xmath37 . ) from a theoretical point of view , this completely solves problem 1 .
naturally , the consistency proof leans heavily on results and techniques from analytic convex geometry , as well as a suitable version of the strong law of large numbers .
some effort has been made to make the proof fairly self - contained , but some arguments from the proof from @xcite that algorithm noisybrightlsq is strongly consistent are used in proving that algorithm noisycovblaschke is strongly consistent .
one such argument rests on the bourgain - campi - lindenstrauss stability result for projection bodies . with algorithms for problem 1 in hand , we move to problem 2 , assuming that @xmath37 is an unknown convex body satisfying the same conditions as before .
the basic idea is simple enough : use ( [ prc ] ) and the measurements of @xmath63 at points in a suitable cubic array to approximate @xmath57 via its fourier series , and feed the resulting values into the algorithms for problem 1 .
however , two major technical obstacles arise .
the new estimates of @xmath57 are corrupted by noise that now involves dependent random variables , and a new deterministic error appears as well .
a substitute for the strong law of large numbers must be proved , and the deterministic error controlled using fourier analysis and the fortunate fact that @xmath57 is lipschitz . in the end the basic idea works , assuming that for suitable @xmath64 , measurements of @xmath63 are taken at the points in @xmath65 contained in the cubic window @xmath66^n$ ] , whose size increases with @xmath8 at a rate depending on the parameter @xmath67 .
the three resulting algorithms , algorithm noisymod@xmath68lsq , algorithm noisymod@xmath68blaschke , and algorithm noisymod@xmath68diff(@xmath54 ) , are stated in detail and , with suitable restrictions on @xmath67 , proved to be strongly consistent under the same hypotheses as for problem 1 .
our final three algorithms , algorithm noisymodlsq , algorithm noisymodblaschke , and algorithm noisymoddiff(@xmath54 ) cater for problem 3 .
again there is a basic simple idea , namely , to take two independent measurements at each of the points in the same cubic array as in the previous paragraph , multiply the two , and feed the resulting values into the algorithms for problem 2 .
no serious extra technical difficulties arise , and we are able to prove that the three new algorithms are strongly consistent under the same hypotheses as for problem 2 .
this provides a complete theoretical solution to the phase retrieval problem for characteristic functions of convex bodies . to summarize : @xmath69 for problem 1 , _ first _ use _ either _ algorithm noisycovblaschke _ or _ algorithm noisycovdiff(@xmath54 ) and _ then _ use algorithm noisycovlsq .
@xmath69 for problem 2 , _ first _ use _ either _ algorithm noisymod@xmath68blaschke _ or _
algorithm noisymod@xmath68diff(@xmath54 ) and _ then _ use algorithm noisymod@xmath68lsq .
@xmath69 for problem 3 , _ first _ use _ either _ algorithm noisymodblaschke _ or _ algorithm noisymoddiff(@xmath54 ) and _ then _ use algorithm noisymodlsq .
these results can also be viewed as a contribution to the literature on the associated uniqueness problems .
they show that if a convex body is determined , up to translation and reflection in the origin , by its covariogram , then it is also so determined by its values at certain countable sets of points , even , almost surely , when these values are contaminated with noise .
similarly , the characteristic function of such a convex body is also determined by certain countable sets of noisy values of the modulus of its fourier transform .
our noise model is sufficiently general to apply to all the main cases of practical interest : zero mean gaussian noise , poisson noise ( unbiased measurements following a poisson distribution , sometimes called shot noise ) , or poisson noise plus zero mean gaussian noise .
however , the main text of this paper deals solely with theory . with the exception of corollary [ cordiffe ] and remark [ cordifferem ] , where the method of proof leads naturally to rates of convergence for algorithm noisycovdiff(@xmath54 ) and hence for the two related algorithms for phase retrieval ,
the focus is entirely on strong consistency .
further remarks about convergence rates , sampling designs , and implementation issues have been relegated to the appendix .
much remains to be done .
we believe , however , that our algorithms will find applications .
for example , baake and grimm @xcite explain how the problem of finding the atomic structure of a quasicrystal from its x - ray diffraction image involves recovering a subset of @xmath2 called a window from its covariogram , and note that this window is in many cases a convex body .
we are grateful to jim fienup , david mason , and sara van de geer for helpful correspondence and to referees for some insightful suggestions that led to significant improvements .
[ cols= " < , < " , ] since the least squares problem ( [ obj1 ] ) is nonlinear , it is important to control the number of variables , that is , the number of facets of the approximation @xmath46 to the blaschke body @xmath47 of @xmath37 . to a large extent
, algorithm noisycovblaschke already does this ; the potential @xmath70 variables that would otherwise be required ( see @xcite ) is , as experiments show , considerably reduced .
in fact , if there is little or no noise , a linear programming version of the brightness function reconstruction program due to kiderlen ( see @xcite , where it is stated for measurements without noise ) is not only even faster , but also produces approximations @xmath46 to @xmath47 with at most @xmath71 facets . beyond this
, there is the possibility of using the pruning techniques discussed in ( * ? ? ?
* section 3.3 ) .
there is also the possibility of changing the variables in the least squares problem ( [ obj1 ] ) .
a convex polytope @xmath72 whose facet outer unit normals are a subset of a prescribed set @xmath73 of directions can be specified by the vector @xmath74 such that @xmath75 the possible advantage in using these variables arises from the fact that , by the brunn - minkowski inequality ( cf .
* section 11 ) ) , the covariogram @xmath76 turns out to be @xmath77-concave ( i.e. , @xmath78 is concave ) on its support in the combined variable @xmath79 .
one may therefore try solving the problem @xmath80 over the variables @xmath81 . by expanding the square in ( [ obj1app ] ) , approximating the sums by integrals , and using the prkopa - leindler inequality ( * ? ? ?
* section 7 ) , the objective function can be seen as an approximation to the difference of two log - concave functions .
these admittedly weak concavity properties may help .
regularization is often used to improve fourier inversion in the presence of noise .
we expect this to be of benefit in implementing the phase retrieval algorithms , where preliminary investigations indicate that regularization will allow the restriction on the parameter @xmath67 to be considerably relaxed . corresponding to the two basic approaches to reconstruction
one via the blaschke body and one via the difference body
there are two different sampling designs . for the former ,
measurements are made first at the origin and at points in a small sphere centered at the origin , and then again at points in a cubic array .
for the latter , measurements are made twice , each time at points in cubic array .
these sampling designs are a matter of convenience , at least regarding the cubic array
. it should be possible to use a variety of different sets of measurement points , at least for reconstructing from covariogram measurements , with appropriate adjustments in the consistency proofs . | we propose strongly consistent algorithms for reconstructing the characteristic function @xmath0 of an unknown convex body @xmath1 in @xmath2 from possibly noisy measurements of the modulus of its fourier transform @xmath3 .
this represents a complete theoretical solution to the phase retrieval problem for characteristic functions of convex bodies .
the approach is via the closely related problem of reconstructing @xmath1 from noisy measurements of its covariogram , the function giving the volume of the intersection of @xmath1 with its translates . in the many known situations in which the covariogram determines a convex body , up to reflection in the origin and
when the position of the body is fixed , our algorithms use @xmath4 noisy covariogram measurements to construct a convex polytope @xmath5 that approximates @xmath1 or its reflection @xmath6 in the origin .
( by recent uniqueness results , this applies to all planar convex bodies , all three - dimensional convex polytopes , and all symmetric and most ( in the sense of baire category ) arbitrary convex bodies in all dimensions . )
two methods are provided , and both are shown to be strongly consistent , in the sense that , almost surely , the minimum of the hausdorff distance between @xmath5 and @xmath7 tends to zero as @xmath8 tends to infinity . |
the paraxial propagation of the field through a distance @xmath5 is described in the phase space as the following simple transformation of the wigner distribution function [ [ bastiaans ] ] : @xmath13 this formula applied to the wigner function of the periodic field gives the following result : @xmath14 \delta(u - \frac{n+n'}{2 } u_0).\end{aligned}\ ] ] in the course of propagation the interference terms acquire additional phase shifts given by @xmath15 where @xmath16 is the talbot distance .
it is straightforward to see that for integer multiples of the talbot distance the original wigner function of the input field is reproduced . at intermediate distances ,
the phase shifts @xmath17 play a nontrivial role and the structure of the observed fresnel images becomes more complex .
nevertheless , they exhibit an interesting regular behavior at rational multiples of the talbot distance
. we will now discuss this effect in terms of the wigner distribution function .
let us denote @xmath18 , where @xmath19 and @xmath20 are coprime integers .
the main complication in eq .
( [ wxuz ] ) are phase factors @xmath17 which depend quadratically on @xmath21
. we will simplify it with the help of an observation used in the studies of quantum wave packets dynamics [ [ averbukh ] ] : the exponent @xmath22 is periodic in @xmath21 with the period @xmath23 if @xmath20 is a multiple of 4 and @xmath24 otherwise .
the quadratic phase factor can therefore be represented as a finite fourier sum : @xmath25 with certain coefficients @xmath26 , which have been analyzed in detail in ref .
[ [ averbukh ] ] .
substitution of the above expression yields : @xmath27 \nonumber \\ & & \label{wxumztkfinal } \times \exp\left[-\frac{2\pi i n'}{a } \left(x-\frac{s'a}{l}\right)\right ] \delta\left(u - \frac{n+n'}{2 } u_0\right).\end{aligned}\ ] ] the interference terms of the wigner function can now be interpreted as generated by pairs of fourier components of the source field shifted in the position space by rational fractions @xmath28 , where @xmath29 .
a simple rearrangement of the exponent arguments allows one to represent the sum over @xmath21 and @xmath30 solely in terms of the source field wigner function : @xmath31 w\left(x-\frac{(s+s')a}{2l},u;0\right).\ ] ] this is a compact summation formula for the wigner function at rational multiples of the talbot distance .
it shows , quite surprisingly , that @xmath32 is simply given by a _ finite _ sum of spatially displaced wigner functions of the source field , with some phase factors .
integration of the derived expression for the wigner function over @xmath3 yields the known formula for the field intensity distribution in the observation plane : @xmath33 which shows that the observed fresnel image can be represented as generated by a finite sum of shifted source field amplitudes [ [ guigay ] ] .
a detailed discussion of fractional talbot images can be found in ref .
[ [ berry ] ] .
we have discussed a one dimensional fresnel diffraction from a periodic grating using the phase space formalism .
we have seen in this picture , that the whole propagation of the field is encoded in the interference terms of the wigner function , generated by coherence between fourier components of the source wave .
the variety of the fresnel images is simply a result of a complicated interplay between phases of these interference terms .
the summation formula derived in this paper demonstrates that the fractional talbot effect can be also understood as an interference between components shifted in the position space by fractions of the pattern period .
let us finally note a close analogy between the talbot effect and the dynamics of various nonlinear quantum systems .
dynamical phase factors with a quadratic dependence on the summation index are a common feature of many problems .
these phase factors can be easily revealed in the equation of motion for the quantum wigner function in the @xmath34 phase space : @xmath35 w_{nn'}(x , p;0),\ ] ] where @xmath36 are the amplitudes of the initial wave function @xmath37 projected onto the energy eigenstate @xmath38 , and where @xmath39 are the cross wigner functions of the eigenstates . for many systems , their eigenergies @xmath40
are given by a quadratic polynomial of @xmath21 .
this can be either an approximate dependence , as it is for the rydberg electron wave packets [ [ parker ] ] , or a strict one , as it happens for an electromagnetic field mode in the kerr medium [ [ yurke ] ] and a particle in the infinite square well .
in all these cases , the state of the system turns out to be highly regular for times equal to rational multiples of the characteristic revival time , at which the system is again in its initial state .
this regularity consists in a formation of a finite superposition of copies of the initial state generated by a simple transformation . for the wave packets
dynamics , the components are spatially shifted [ [ averbukh ] ] ; for the kerr medium , the components are rotated in the phase space around its origin [ [ tara ] ] .
in particular , a close similarity exists between the talbot effect and the fractional revivals in the infinite square well .
the dynamics of the latter system is equivalent to an evolution in a free space of an infinite periodic sequence of displaced initial wave functions [ [ born],[stifter ] ] .
furthermore , the propagation of electromagnetic fields in the paraxial approximation is governed by the same wave equation as the quantum time evolution .
therefore , fresnel diffraction of classical electromagnetic waves on periodic gratings is a strict counterpart of the quantum dynamics of a confined particle .
the authors have benefited from discussions with j. h. eberly and from a stimulating talk given in ulm by b. rohwedder . this work has been partially supported by the polish kbn grant 2 po3b 006 11 and the deutsche forschungsgemeinshaft . | a phase space description of the fractional talbot effect , occurring in a one dimensional fresnel diffraction from a periodic grating , is presented . using the phase space formalism a compact summation formula for the wigner function at rational multiples of the talbot distance
is derived .
the summation formula shows that the fractional talbot image in the phase space is generated by a finite sum of spatially displaced wigner functions of the source field .
[ winthrop ] j. t. winthrop and c. r. worthington , `` theory of fresnel images .
i. plane periodic objects in monochromatic light , '' , 373381 ( 1965 ) .
[ bastiaans ] m. j. bastiaans , `` the wigner distribution function applied to optical signals and systems , '' , 2630 ( 1978 ) .
[ averbukh ] i. sh .
averbukh and n. f. perelman , `` fractional revivals : universality in the long term evolution of quantum wave packets beyond the correspondence principle dynamics , '' , 449453 ( 1989 ) .
[ guigay ] j. p. guigay , `` on fresnel diffraction by one - dimensional periodic objects , with application to structure determination of phase objects , '' opt .
acta * 18 * , 677682 ( 1971 ) . [ berry ] m. v. berry and s. klein , `` integer , fractional and fractal talbot effects , '' j. mod .
opt . * 43 * , 21392164 ( 1996 ) .
[ parker ] j. parker and c. r. stroud , jr .
, `` coherence and decay of rydberg wave packets , '' , 716719 ( 1986 ) .
[ yurke ] b. yurke and d. stoler , `` generating quantum mechanical superpositions of macroscopically distinguishable states via amplitude dispersion , '' , 1316 ( 1986 ) ; a. mecozzi and p. tombesi , `` distinguishable quantum states generated via nonlinear birefrigerence , '' , 10551058 ( 1987 ) . [ tara ] k. tara , g. s. agarwal , and s. chaturvedi , `` production of schrdinger macroscopic quantum - superposition states in a kerr medium , '' , 50245029 ( 1993 ) . [ born ] m. born and w. ludwig , `` zur quantenmechanik des krftefreien teilchens , '' z. phys . * 150 * , 106117 ( 1958 ) . [ stifter ] p. stifter , c. leichte , w. p. schleich , and j. marklof , `` das teilchen i m kasten : strukturen in der wahrscheinlichkeitsdichte , '' z. naturforsch . * 52a * , 377385 ( 1997 ) .
it is the purpose of this communication to investigate a one dimensional fresnel diffraction from a periodic grating and the corresponding fractional talbot effect [ [ winthrop ] ] using the wigner phase space distribution function . |
introducing spherical voids into a solid and otherwise homogeneous and isotropic material changes its mechanical properties significantly .
such materials are currently generated using templating techniques @xcite , by integrating hollow spheres in a matrix @xcite or by direct foaming @xcite .
the voids of the resulting material can be arranged in an ordered or in a disordered manner .
it is therefore of considerable interest to compare the mechanical properties of different ordered and disordered void structures , including the case where the voids overlap .
+ , b ) touching spheres at @xmath0 and c ) open - cell structure at @xmath1 . ]
when direct foaming is used , the equal - volume bubbles tend to organize themselves on a close - packed lattice in densest packing @xcite .
depending on the precise method of generation , the void arrangement may be chosen to be dominated by fcc ( face - centred cubic ) or hcp ( hexagonal close - packed ) arrangement @xcite .
this raises the practically relevant question whether one of these arrangements should be preferred over the other , due to advantageous mechanical properties of the resulting solid void material .
+ porous materials show a very rich range of non - linear mechanical behaviour , including plastic deformation , buckling and rupture . here , we concentrate on the linear elastic behaviour , corresponding to infinitesimally small strain .
the complementary problem , the elastic properties of crystalline arrangements of solid spheres , has been investigated experimentally and numerically by several authors @xcite .
the elastic properties of the void material of these packings , however , have not yet been investigated sufficiently and comparatively .
+ before computers made their breakthrough in science , the elastic properties of void material were estimated by superposition of the effects of a single void @xcite .
these methods yield good results for low void fraction
. however , with increasing void fraction , higher orders of interaction between the voids have to be taken into account @xcite .
christensen @xcite compared different micro - mechanic models available at that time . in 1992 , day et al .
@xcite developed a simple finite element method ( fem ) to calculate the elastic properties of a two - dimensional material with circular voids .
they investigated the influence of void fraction and topology separately and devised a simple analytical explanation for the calculated values .
after 1992 , increasing computer power became available for many research groups , resulting in further direct numerical simulations of the interstitial material of sphere or bubble arrangements in three dimensions @xcite . in 2006 , ni et al .
@xcite calculated the young s modulus of a simple cubic void structure and compared their results to analytical estimations of @xcite and @xcite , which are used later for comparison .
+ the agreement between analytical @xcite and numerical @xcite methods was very good .
however , the graphs of young s modulus versus void fraction depend only weakly on the structure .
thus , small derivations between the graphs raise the question , as to whether a difference results from the uncertainty of the method or rather from the structural differences of the investigated materials . in order to reliably extract the structural effects ,
one therefore needs to apply an identical numerical method to different structures , taking great care of the numerical uncertainty .
additionally , many of the available studies are confined to low or medium void fractions .
+ in this paper , a comparative study of a wide variety of dense sphere packings is carried out , revealing the influence of the structure on the elastic properties . the entire range of void fractions is considered , as illustrated in figure [ fig : intro ] .
small voids form closed - cell void material with low void fraction . retaining the regularly arranged void centres and increasing the diameter the voids touch each other at a certain void fraction @xmath2 , forming closed - packed void material . at
even higher void fractions , the voids overlap , forming open - cell void materials .
monodisperse spheres or microbubbles tend to crystallise when they become agglomerated .
this means that their centres form a periodic , crystalline lattice .
since these systems strive for densest packing , they are usually arranged in the hexagonally close - packed ( hcp ) or face - centred cubic ( fcc ) structure , both providing equally dense sphere packings @xcite . for comparison , simple cubic ( sc ) and
body centred cubic ( bcc ) arrangements are also taken into account here .
if the spheres are slightly polydisperse or if the agglomeration process is too fast to allow for relaxation , random closed - packed ( rcp ) structures are created .
+ from the different structures mentioned above , rectangular or cubic representative volume elements ( rve ) were derived which are shown in figure [ fig : rve ] . except for sc ,
the rve does not coincide with the primitive cell of the crystalline arrangements .
rather , it is the smallest cuboid cell which may be periodically combined to represent the complete structure , because the numerical method only allows for orthogonal , periodic boundaries .
parameters of the chosen rve are given in table [ tab : rves ] .
note that for fcc two different rves were applied and compared .
the cubic rve , labelled fcc , is a cube , bounded by planes in @xmath3 , @xmath4 and @xmath5 .
the hexagonal rve , labelled fcch , is a cuboid , bounded by @xmath6 , @xmath7 and @xmath8 planes .
this provides an additional test of the method applied by comparing the young s moduli of the different rve of the same structure .
this is explained in more detail in section [ sec : validation ] below .
the rcp structure is special , since it does not correspond to a crystalline lattice , but it does involve periodic boundary conditions .
the sphere positions for this case were generated using a gas - dynamic algorithm that is freely available @xcite .
drugan et al .
@xcite found , that with six spheres in an rve of disordered voids , the statistical uncertainty of the mechanical properties is below 5% . aiming for very high accuracy , here rves with 30 spheres
were generated .
the statistical uncertainty resulting from this type of rve was investigated , as reported in section [ sec : validation ] below .
+ [ cols="^,^,^,^,^,^",options="header " , ] the solid fraction @xmath9 of a void material is the ratio of the volume of solid material @xmath10 contained in a given rve with the volume @xmath11 .
the void fraction @xmath12 is the ratio between the void volume @xmath13 and the total volume of the rve . in case of separated spherical voids
, the void volume can be calculated from the sum of the volume of each spherical void contained in a given rve . in this case
, the void fraction depends on the sphere diameter @xmath14 , the lattice spacing @xmath15 , and the packing density @xmath16 for touching spheres of the structure considered @xmath17 defining the separation of two voids to be @xmath18 as displayed in figure [ fig : fig2 ] yields the last equality in equation ( [ eq : phi ] ) . , void diameter @xmath14 , and void separation @xmath19 .
depending on the void diameter the structure is a ) closed - cell , b ) indefinite or c ) open - cell . ] for void fractions above @xmath16 the spherical voids overlap , yielding negative values for @xmath19 . due to that
, equation ( [ eq : phi ] ) is not valid for overlapping voids .
the goal of the method elaborated in this study is to find an equivalent homogeneous continuum material that is , in a volume - averaged sense , elastically equivalent to the heterogeneous rve .
the method of homogenization is well understood and documented , for example in @xcite .
the linear elastic behaviour is described by hooke s law , which can be written as @xmath20 this law relatess the applied stress @xmath21 , the resulting strain @xmath22 and the stiffness tensor @xmath23 . for the equivalent homogeneous continuum material
it is required that it stores the same elastic energy @xmath24 per volume @xmath25 as the rve when applying the same global strain , i.e. @xmath26 stress , strain and the stiffness tensor are expressed in voigt s notation in the following .
the strain vector @xmath22 consists of 6 elements , which are normal strain @xmath27 , @xmath28 , @xmath29 and shear strain @xmath30 , @xmath31 and @xmath32 .
the stress vector @xmath21 contains the corresponding 6 elements . due to symmetry , the 6@xmath336 elements of @xmath23 consist of 21 independent elements . starting from a heterogeneous rve ,
these can be computed by applying 21 independent load cases and calculating the corresponding elastic energy @xmath34 .
these load cases are defined by 21 independent sets of strain @xmath22 .
one has to apply 6 sets of strain with one non - zero element @xmath35 yielding @xmath36 and 15 sets of strain with two non - zero elements @xmath35 , @xmath37 yielding @xmath38 .
note , that @xmath38 is not a tensor but the scalar value of the elastic energy corresponding to the load case @xmath35 , @xmath37 .
+ the diagonal elements @xmath39 can be determined from sets of strain with only one element @xmath35 @xmath40 the off - diagonal elements @xmath41 result from sets with two strains @xmath35 , @xmath37 @xmath42 note , that @xmath43 holds .
+ from the stiffness tensor @xmath41 one obtains the compliance tensor @xmath44 = @xmath45 by calculating its inverse .
one obtains a young s modulus @xmath46 of the rve from the first three main diagonal elements of the compliance tensor via @xmath47 these refer to the chosen basis of the compliance tensor but they can be rotated to obtain a value for any direction , as described below .
the young s modulus of the void structure , normalized by the young s modulus of the matrix material @xmath48 , is one main material parameter considered in the present study .
voigt s rule of mixture @xcite provides an upper bound for the young s modulus of a porous material @xmath49 poisson ratios @xmath50 of the rves can be extracted from the compliance tensor @xmath51 the influence of the poisson ratio @xmath52 of the solid material on the elastic properties of the void material is generally not negligible . in this study @xmath53
is used , representative of many polymers , in particular of polyurethane , which is used for many foams . only in rare cases , to allow comparison with the literature , other values are chosen .
+ the values of the stiffness and compliance tensor depend on the orientation of the corresponding basis . in order to derive the elastic properties of the material in any given direction one
has to transfer the original basis @xmath54 into a new basis @xmath55 using the transformation tensor @xmath56 @xcite .
the rotation matrix @xmath57 consists of four parts @xmath58 which can be computed from @xmath59 with @xmath60 @xmath61 subsequently , one can transform the stiffness tensor according to @xmath62 to apply a certain strain @xmath63 and to calculate the resulting elastic energy @xmath24 , the commercial software ansys fem was used .
the rve were meshed with tetrahedral elements with quadratic ansatz functions , controlling the mesh parameter @xmath64 , which is the number of grid points per void spacing @xmath15 ( see figure [ fig : fig2 ] ) .
the minimum grid resolution was @xmath65 points . in order to impose periodic boundary conditions , it is necessary to apply identical grids on opposite faces .
the periodic displacement and the periodic stress is then realized by adding restricting equations to periodic point pairs , corresponding to the desired stress and strain conditions .
the results presented in this article show that differences in the mechanical behaviour occur if voids are arranged in different ways , but that some of these differences are small . in this situation
it is important to assess the accuracy of the method which is applied and to demonstrate that the uncertainty of the data is below the differences addressed .
thus , three methods were applied in order to estimate the uncertainty of the obtained results .
+ the first method is a grid study . with increasing resolution
, a numerical solution should converge toward the exact solution .
however , since the resolution is usually limited due to limited computer power , one has to choose a resolution which yields results with sufficiently small deviation from the exact solution .
for the grid study , hcp with thin void separation @xmath66 yielding a void fraction of @xmath67 was considered .
since there is no analytical solution available for this problem , it was solved with different resolutions @xmath68 .
the resulting young s modulus were fitted with a power function , yielding an approximation for the exact solution @xmath69 and the deviation @xmath70 .
the results are shown in figure [ fig : gridstudy ] .
the method is found to be third order accurate in and direction , but only second order in direction .
the different order might arise because many of the material sheets separating two voids are oriented perpendicular to the direction and these material sheets are the critical regions in terms of grid resolution .
a given number of grid points resolves stretching deformation better than bending deformation because of the more uniform stress and strain distribution in the stretching case .
according to figure [ fig : gridstudy ] , the numerical error due to resolution is well below @xmath71 for @xmath65 .
the results reported below were obtained with @xmath65 , except if stated otherwise .
+ for hcp at @xmath72 .
the young s modulus in three different directions is analysed .
it is compared to the corresponding value @xmath69 for infinitely high grid resolution , derived from logarithmic fit of the computed resolutions .
broken lines show the slope for a second - order and a third - order method . ]
the second validation method is the comparison with results to be found in literature @xcite .
for this purpose the poisson ratio @xmath52 of the solid material was chosen to be equal to the values used in the literature .
figure [ fig : rnd_ref ] demonstrates the good agreement between the literature data and the results of the present method . in the case of the rcp structure , simulation of 45 different , randomly generated rves were performed and statistically analysed .
figure [ fig : rnd_ref ] shows the histogram of the results , the standard deviation of the young s modulus and the confidence interval for the mean value .
the measured confidence interval equals @xmath73 , which is in the same order of magnitude as the numerical uncertainty of the crystalline simulations .
the statistical average value of the young s modulus is in good agreement with @xcite
. + @xcite .
( b ) random distribution for a material with poisson ratio @xmath74 @xcite . for the present data , also
the statistical variation of samples is visualised .
both cases show good agreement with the literature values , demonstrating the applicability of our method . ]
the third method of validation is an internal sensitivity test for fcch and hcp .
as pointed out in table [ tab : rves ] , the fcc structure was calculated using two rves with different orientation .
the resulting young s moduli are different due to their dependence on orientation .
but by applying a unitary transformation to the stiffness tensor @xmath23 , the computed result can be transformed into the same coordinate system according to equation ( [ eq : kck ] ) @xmath75 in theory , @xmath76 and @xmath77 should be equal .
but due to computational uncertainty , they exhibit a small deviation .
figure [ fig : compare_e_rot ] shows the difference in the young s modulus in all directions for a gas fraction of @xmath78 .
this deviation yields another estimation of the uncertainty of the computation . in the present case
, the maximum deviation would be below @xmath79 for @xmath65 and below @xmath80 for @xmath81 .
this value is in good agreement with the numerical uncertainty derived from the grid study . at a void fraction of @xmath82 .
colour and distance to the origin indicate the young s modulus in the corresponding direction .
( c ) percentaged difference between both young s moduli in the corresponding direction , showing a maximum of 0.2 % .,scaledwidth=100.0% ] + a similar internal test can be done for hcp .
cazzani @xcite investigated the elastic properties of materials with hexagonal close - packed structure .
taking into account the symmetries he derived an exact formula for the young s modulus of hcp material in any direction @xmath83 in terms of only four entries @xmath84 , @xmath85 , @xmath86 , and @xmath87 of the compliance tensor @xmath88 n_1 ^ 2 \right\ } ^{-1 } .
\label{eq : cazzani}\ ] ] as explained above , one may calculate the young s modulus in any direction by rotating the stiffness tensor according to equation ( [ eq : kck ] ) .
this yields two independent ways to calculate the angular distribution of the young s modulus .
the resulting values of both methods can be compared , as demonstrated in figure [ fig : cazzani ] . .
colour and distance to the origin indicates the young s modulus in the corresponding direction .
( c ) percentaged relative difference between the young s moduli from a ) and b ) in the corresponding direction , showing a maximum of 0.2 % .,scaledwidth=100.0% ] the maximum deviation is below @xmath80 for @xmath81 , which is again in line with the findings above .
+ taking into account the different tests performed on the uncertainty of the method proves that overall the uncertainty of the employed numerical method for the computation of the young s modulus is less than @xmath89 for @xmath65 and less than @xmath80 for @xmath81 .
the young s modulus for the different configurations given in table [ tab : rves ] was computed for a range of void fractions @xmath90 .
the void fraction of the rve was varied by changing the sphere diameter @xmath14 while fixing the bubble centre positions .
the actual void fraction was calculated from the volume of all finite elements after meshing .
the method also allows for sphere diameters larger than the distances between the bubbles , yielding overlapping voids and thus , giving rise to open - cell structures .
the mean values of the young s modulus @xmath91 ( defined in equation [ eq : emean ] below ) are plotted in figure [ fig : e_mean ] . ) ) over a range of void fractions @xmath92 for different arrangements of spherical voids .
the solid line represents the fitting curve from equation ( [ eq : e_fit ] ) while the broken line represents voigt s upper bound . ]
generally , the mean young s moduli of different structures appear to be relatively close to each other and well below voigt s bound ( equation ( [ eq : voigt ] ) ) .
a rough estimation of the mean young s modulus of void material for a given void fraction can be extracted by least - square fitting of the data in figure [ fig : e_mean ] , yielding @xmath93 this fit implies a quadratic dependence of the mean young s modulus on the void fraction .
it goes to zero for void fractions of 0.915 , which is well below the range of rigidity - loss above @xmath94 .
this shows that for open - cell structures at high void fractions the quadratic scaling does not represent the actual dependence very well . in an associated study @xcite
we have shown that the young s modulus in the limit of rigidity - loss scales with the solid fraction to the power of 3.5 . + the structural differences manifest themselves mostly in the anisotropy which will be addressed now . to that end
, the corresponding stiffness tensor @xmath95 for each structure and void fraction was rotated according to equation ( [ eq : kck ] ) from the original basis @xmath54 into the new orthonormal basis @xmath55 @xmath96 varying @xmath97 and @xmath98 $ ] .
figure [ fig : e_winkel_kombi ] shows the dependence of @xmath99 on the angles @xmath100 and @xmath101 for selected void fractions . .
colour represents the relative young s modulus .
distance from the origin also represents the young s modulus in the corresponding direction @xmath102 cubic void arrangements , such as sc , bcc and fcch show a cubic symmetry in the angular dependence of the young s modulus .
hcp schows isotropic dependence around the direction . for rcp
only small deviation from full isotropy is found , corresponding to the finite number of voids in an rve of rcp .
the maximum value , @xmath103 , the minimum value , @xmath104 , and the mean value , @xmath91 of @xmath99 are extracted by @xmath105 } e_{11}(\alpha , \theta , \phi_v ) \nonumber \\
e_{min } ( \phi_v ) & = & \mathrm{min}_{\forall ( \alpha , \theta ) \in [ 0,2 \pi [ \times [ 0 , \pi ] } e_{11}(\alpha , \theta , \phi_v ) \nonumber \\
e_{mean}(\phi_v ) & = & \frac{1}{4 \pi } \int_0^{\pi } \int_0^{2 \pi } e_{11}(\alpha , \theta , \phi_v ) \sin(\theta ) d\alpha d\theta .
\label{eq : emean}\end{aligned}\ ] ] figure [ fig : e_max_min ] shows the maximum and minimum young s modulus for different void arrangements at different void fractions . in order to distinguish the structures more clearly , the figure displays the difference between the young s modulus and the voigt s bound @xmath106 . for each void structure
the maximum and minimum young s modulus is shown , derived from transformation of the stiffness tensor according to equation ( [ eq : emean ] ) .
the figure compares all configurations investigated.,scaledwidth=100.0% ] simple cubic shows very prominent maximum and minimum values of the young s modulus .
all the other values are close to each other . apart from sc
, the transition from closed - cell to open - cell void material seems to change the young s modulus only slightly .
fcch and fcc show nearly perfect agreement , as already observed in the validation section . from the stiffness tensor
the poisson ratio for any pair of orthogonal directions @xmath107 and @xmath108 can be derived by rotation of the basis of the tensor according to equation and subsequent application of equation . in this study , only special combinations of directions are investigated , corresponding to the symmetries of the void arrangements . to that end , a new orthonormal basis @xmath109 was defined with respect to the original basis @xmath54 @xmath110 the direction @xmath111 is fixed , parallel to the rve axis @xmath112 , defined in figure [ fig : rve ] .
the angle @xmath97 was varied , so that @xmath113 and @xmath114 cover the complete plane , orthogonal to @xmath111 .
the corresponding poisson ratios @xmath115 , @xmath116 and @xmath117 are given in figure [ fig : nu_winkel_kombi ] .
+ , @xmath116 , and @xmath117 on the direction @xmath100 of @xmath113 for different void arrangements at certain void fractions . ] for cubic void arrangements , such as sc and bcc , four - fold symmetry of @xmath117 and @xmath116 around the @xmath111 direction is visible , referring to the four - fold symmetry of a cube . for @xmath118
the basis @xmath109 coincides with the cubic axes , so that @xmath115 , @xmath117 , and @xmath116 are equal .
the poisson ratio @xmath115 is independent of the direction of @xmath113 , because four - fold symmetry in the @xmath119 layer is sufficient to assume isotropy of the resulting strain in this layer .
the hexagonal structures fcch and hcp show three - fold symmetry in the @xmath119 layer causing @xmath115 @xmath117 , and @xmath116 to be independent of the direction of @xmath113 . for rcp
only small deviations from isotropy are visible , resulting from the finite number of voids in the rve .
+ from these angular dependencies , three characteristic combinations of directions are derived .
these are @xmath120 , @xmath121 , and @xmath122 .
figure [ fig : nu_max_min ] shows the dependence of these selected poisson ratios on the void fraction and on the void arrangement . for different arrangements of spherical voids .
each plot shows one of the selected poisson ratios @xmath120 , @xmath121 , and @xmath122 .
they will be discussed in separate subsections below . ] for vanishing void fraction all poisson ratios converge to @xmath123 which is the poisson ratio of the matrix material .
it is interesting to note , that for high void fractions poisson ratios higher than @xmath124 appear , which is not possible for isotropic material but can be the case for anisotropic materials . for isotropic material , a poisson ratio higher than @xmath124 means , that the volume increases under uni - axial compression , which implies a negative bulk modulus . for anisotropic void material high poisson ratios
are possible .
the issue was investigated by ting et al .
@xcite in the framework of general anisotropic material . in zinc @xcite and cubic metals
@xcite researchers found negative poisson ratios .
however , in the present study all computed poisson ratios are positive .
but , for high void fractions close to rigidity - loss poisson ratios close to zero appear in some cases . + in the following , the different structures are discussed separately , concerning young s modulus and poisson ratio .
compared to other void arrangements the simple cubic one shows very disparate maximum and minimum values in the young s modulus . in figure
[ fig : e_winkel_kombi]a the young s modulus in different directions is plotted for a void fraction of @xmath125 , corresponding to thin material sheets between the voids .
it shows a clear maximum in the , and directions .
+ the reason for this high young s modulus can be inferred from consideration of the structure of the interstitial material shown in figure [ fig : sc_kombi1]a .
the material forms columns in the direction , supporting a load very effectively , as illustrated in figure [ fig : sc_kombi1]b . but
applying the load at an angle to the column direction , e.g. in direction ( figure [ fig : sc_kombi1]c ) , this causes shearing of the structure .
therefore , the young s modulus shows a minimum in direction .
+ . the structure corresponds to beams forming a cubic wire - frame .
( b ) mechanism of deformation when straining a cubic wire - frame .
strain along the direction causes perpendicular relaxation of the individual wires but no relaxation of the structure .
( c ) strain along the direction causes perpendicular relaxation of the structure in the direction . ]
the extremal values of the poisson ratio and their dependence on the void fraction is shown in figure [ fig : nu_max_min ] . in figure
[ fig : nu_winkel_kombi ] the dependence of the poisson ratio on the direction is presented for a void fraction of @xmath125 .
the poisson ratio @xmath115 for compression in direction is small and decreases with increasing void fraction .
this is also due to the column - like structure in this direction ( figure [ fig : sc_kombi1]a ) .
compression of the structure causes relaxation of the individual columns in transversal direction , but not relaxation of the whole structure ( figure [ fig : sc_kombi1]b ) .
buckling of the column is not relevant to the linear - elastic simulation .
+ in contrast to the direction , stress along the direction is distributed on columns in and direction .
this causes very strong relaxation in the direction because the structure is folded ( figure [ fig : sc_kombi1]c ) .
therefore , the maximum value of @xmath116 in figure [ fig : nu_max_min ] increases significantly with decreasing void fraction . for values of @xmath100 that give a maximum of @xmath116
the corresponding @xmath117 shows a minimum ( figure [ fig : nu_winkel_kombi ] ) .
+ below a void fraction of @xmath126 the voids in sc arrangement are separated , forming a closed - cell structure . above this void fraction , the voids cut through each other ( figure [ fig : sc_kombi1]a ) . crossing this critical void fraction
, the young s modulus shows a smooth reduction by about 10 % ( figure [ fig : e_max_min ] ) , because the material sheets between voids are removed .
also the poisson ratios ( figure [ fig : nu_max_min ] ) show a significant reduction crossing the void fraction @xmath127 .
in contrast to other void arrangements , many of these material sheets are oriented parallel to the cubic axes , adding substantially to the young s modulus along the cubic axes because these sheets act the same way as the columns , supporting an external load by elongation .
also , the material sheets connect the columns and stiffen the structure when the load is not applied along the column direction . + the young s modulus for the body centred cubic void arrangement ( figure [ fig : e_max_min ] ) shows a very interesting behaviour , in that at a void fraction of @xmath128 the directions for maximum and minimum young s modulus switch .
at this point the young s modulus is the same for all directions .
figure [ fig : nu_winkel_kombi ] shows , that also the poisson ratios all equal 0.32 at this void fraction .
that young s modulus and poisson ratio are independent of the angle means , that bcc at a void fraction of 0.59 is elastically isotropic .
figure [ fig : e_winkel_kombi ] shows the angular dependence of the young s modulus above and below this switching point , respectively . for lower void fraction , the maximum is oriented in direction , for higher void fraction in direction . bcc with high void fraction shows an angular distribution of the young s modulus similar to sc with maximum young s modulus along the columns in the cubic axes
. + figure [ fig : compare_columns ] shows the structure of the interstitial material .
again , the elastic behaviour can be explained with the column structure .
figure [ fig : compare_columns ] illustrates the existence of columns .
figure [ fig : compare_columns]a gives the fraction of the cross section that forms an unbroken column for different structures , orientations and void fractions . to derive the cross section of unbroken columns , the three - dimensional distribution of solid fraction of the void material
is projected in a direction parallel to the columns , yielding the two - dimensional distribution of material .
this two - dimensional distribution is shown in figure [ fig : compare_columns]c , d , f , g for bcc at different void fractions and different directions of projection . for interpretation
, they can be thought of as idealised x - ray photographies of the structure .
the share of the area that equals 1 in the two - dimensional distribution represents the cross sectional area of an unbroken column in the direction of projection . in case of a body centred cubic arrangement
this cross - sectional area behaves similar to the young s modulus . for void fractions below @xmath129 the larger area
is oriented in direction .
above @xmath130 the orientation contains stronger columns .
the value of @xmath130 coincides with the void fraction at which the elastic behaviour of bcc switches . at @xmath131
the straight columns in direction completely disappear .
the columns in direction remain up to void fractions of @xmath132 . + under different angles .
( c , d , f , g ) distribution of material in the case of bcc , projected in ( c , d , e ) and direction ( f , g , h ) at @xmath133 ( c , f ) and @xmath132 ( d , g ) , respectively .
white colour corresponds to values of 1 and marks unbroken columns . ]
the extremal values of the poisson ratio and their dependence on the angle @xmath100 is shown in figure [ fig : nu_max_min ] and [ fig : nu_winkel_kombi ] , respectively .
the latter is performed for a void fraction of @xmath134 and @xmath82 .
the poisson ratio also shows a switching of behaviour at @xmath128 .
again , bcc with high void fraction behaves qualitatively similar to sc , with high values of @xmath116 and low values of @xmath117 .
again , this can be explained with the orientation of columns in the void material . for high void fraction , columns are oriented along the cubic axes ( similar to sc ) .
stress along the cubic axes is supported by these columns , yielding low relaxation in the perpendicular direction .
stress diagonal to the column direction , e.g. along the direction , is distributed to the and column , which causes folding of the structure in direction , as sketched in figure [ fig : sc_kombi1]b .
this results in high values of @xmath116 but even lower values of @xmath117 .
bcc at low void fractions behaves completely different .
the strongest columns are oriented in , , , and direction .
stress along the direction , hence , is distributed on the columns in and .
this causes a folding of the structure in direction , yielding high values of @xmath117 but low values of @xmath116 . in order to compare fcc and hcp in a direct way ,
the hexagonal orientation of fcc , here called fcch , is considered .
as shown in figure [ fig : rve ] , the rve of fcch is bounded by planes in ( 1 0 0 ) , ( 0 1 -1 ) and ( 1 1 1 ) orientation .
the two void arrangements , fcch and hcp , have a very similar structure .
the voids are placed in layers of hexagonal ordering .
these layers are stacked in different sequences , shown in figure [ fig : hcp_fcc_column ] .
+ for both structures , the dependence of the young s modulus on the void fraction an on the direction is shown in figure [ fig : e_max_min ] and figure [ fig : e_winkel_kombi ] .
the hcp arrangement shows a clear maximum in the direction .
this effect is similar as observed with the simple cubic arrangement . in this direction ,
the structure of the interstitial material contains a straight column , marked c in figure [ fig : hcp_fcc_column]b , which supports the load very efficiently .
hcp also contains broken side - columns a that contribute only slightly to the young s modulus in direction .
+ fcch shows four maxima in the , , and direction , which is consistent with its cubic symmetry . in these directions ,
the fcch structure contains columns that are broken in every third layer , marked b in figure [ fig : hcp_fcc_column]a . as a result ,
the maximal young s moduli for fcch are a bit smaller than the maximum for hcp . to support load along these broken columns , the load has to be transferred between the columns .
this is done by shearing the material sheets between the columns . for higher void fraction ,
the material sheets become very thin so that load transfer between the broken columns causes stronger shearing deformation . supporting the load along the unbroken column c of hcp
does not rely on shearing of material sheets .
therefore , the maximum young s modulus of fcch decreases faster than the maximum of hcp with increasing void fraction , as shown in figure [ fig : e_max_min ] .
+ fcch shows broken columns b while hcp shows unbroken columns c and broken side - columns a. ] the young s modulus of hcp appears to be symmetric around the axis while for fcch it shows cubic symmetry .
however , in the ( 0 0 0 1 ) and ( 1 1 1 ) plane , respectively , the young s moduli of hcp and fcc are isotropic ( figure [ fig : e_winkel_kombi ] ) .
the isotropy of fcch and hcp in the basal plane can be related to crystal symmetry as illustrated in figure [ fig : symm ] .
macroscopic properties such as elastic moduli must conform to the external symmetry of the crystal . in the case of hcp , this includes a six - fold rotation about the axis , perpendicular to the hexagonal layers . a well known theorem states that this is sufficient to ensure that the stiffness tensor ( which is fourth rank ) is transversely isotropic @xcite , precisely the property exhibited in the calculations ( figure [ fig : e_winkel_kombi ] ) . also equation ( [ eq : cazzani ] ) derived by cazzani @xcite reflects this isotropy , yielding a dependence of the young s modulus on @xmath135 .
+ in the case of fcch , transverse isotropy is not observed , but at least young s modulus is isotropic for imposed strain in the ( 1 1 1 ) plane ( and also in the ( -1 1 1 ) , ( 1 -1 1 ) , and ( 1 1 -1 ) plane ) as observed in figure [ fig : e_winkel_kombi ] . for fcc
there is indeed no six - fold rotation included in its point group . instead , the axes , , and correspond to a three - fold symmetry around the axis . three - fold symmetry is not sufficient to ensure isotropy .
including a reflection at the plane three - fold symmetry is transformed into six - fold symmetry .
the reflection - based six - fold symmetry implies isotropy only if the direction of strain is not changed by the reflection .
apart from the axis itself , the only directions for strain which are not altered by this reflection are those which lie in the reflection plane .
hence , a limited isotropy holds for directions of strain within such a reflection plane only , as observed .
+ figure [ fig : hcp_fcch_iso ] shows the dependence of the young s modulus on the angle @xmath100 for @xmath136 .
perfect isotropy is not reproduced by the numerical method .
the fluctuations are below 0.2 % which is in line with the numerical uncertainty of the method .
additionally , fcch and hcp exhibit different values for the young s modulus in the basal plane .
this is surprising , since both structures consist of layers of hexagonally arranged voids .
the difference in the young s modulus between fcch and hcp parallel to these layers equals about @xmath89 for increased numerical resolution @xmath137 .
since the numerical uncertainty for the increased resolution was found to be smaller than @xmath80 , the difference in the young s modulus is definitely physical .
the interstitial material of both void structures consist of the same type of hexagonal layers , but the way the layers are stacked and connected is different .
this might cause a tiny difference in the transversal relaxation during deformation . .
with increasing numerical resolution @xmath64 the young s modulus becomes more isotropic .
a small difference of 1 % remains even for very high resolution .
note the small range of the vertical axis . ]
+ the extremal values of the poisson ratio and its dependence on the angle @xmath100 are shown in figure [ fig : nu_max_min ] and [ fig : nu_winkel_kombi ] , respectively , for a void fraction of @xmath72 . the ( 0 0 0 1 ) and ( 1 1 1 ) planes are isotropic in terms of the poisson ratio .
but there is a difference in the values between both structures , supporting the idea of a difference in the young s modulus between the ( 0 0 0 1 ) and ( 1 1 1 ) plane . for fcch
the poisson ratio @xmath117 is higher than for hcp .
this means , that under strain orthogonal to the direction fcch relaxes more along the direction than hcp along the direction .
consequently , fcch is softer in the ( 1 1 1 ) plane than hcp in the ( 0 0 0 1 ) plane .
still , both planes are isotropic in terms of the poisson ratio .
we have recently published a more detailed study on the relations of the elastic parameters of hcp and fcc void material and their scaling behaviour @xcite . in case of random void arrangement
the mean young s modulus , averaged over 45 different , randomly generated rves is shown in figure [ fig : e_max_min ] for different void fractions .
the mean young s modulus is relatively low , compared to periodic structures .
the reason is that in random packing column - like structures do not exist .
maximum and minimum young s modulus for a given void fraction are derived by calculating the extremal values of each rve separately and averaging them over all 45 corresponding rves .
note , that the extremal values depend on the numbers of voids within an rve . a higher number of voids causes an averaging effect , yielding less prominent extremal values . for an infinite number of voids random
packing is isotropic .
however , the extremal values give an estimation of the local variation of the mechanical properties in random packing . as an example
, figure [ fig : e_winkel_kombi ] shows the dependence of the young s modulus on the direction at @xmath138 for one arbitrarily selected rve . even though only 30 voids are considered , the variations are small , below @xmath139 , and no distinct anisotropy is evident . the poisson ratio of random void packing is shown in figure [ fig : nu_max_min ] . since random packing should be isotropic in the limit of an infinite sample , the definition of the orthonormal basis @xmath109 is arbitrary .
however , for comparison with the other structures the same algorithm is used .
again , the variation of @xmath140 is a measure for the local anisotropy of random packing .
it is below @xmath139 in the present case .
+ computation of overlapping voids was not possible for random packing with the present method , because one has to define the boundaries of the rve with sufficient distance to the void surfaces to create a suitable fe mesh .
with increasing void diameter , possible positions for these boundaries become less available .
the research question at the start of this investigation was whether a particular arrangement of spherical voids in a regular lattice would yield more advantageous properties than others , so that the fabrication process should target such an arrangement .
having achieved a level of uncertainty of 0.2 % , the present study provides very sound data for the linear elastic properties of a number of such void materials .
these are reported and a detailed comparison is undertaken .
the result is that for given solid fraction , the variation of direction - averaged values among the structures is small . in this situation
one may seek to extract useful structure - independent rules .
for example , this study gives a fitting curve for the mean young s modulus , averaged over all directions , that represents quite well a wide range of void fractions and different void arrangements .
+ the present study is confined to linear elasticity , but extension to non - linear behaviour of the interstitial material will be interesting to investigate .
non - linear behaviour is expected to be much more sensitive to the void organisation , resulting , e.g. , from buckling phenomena @xcite .
another important aspect is the growth and coalescence of voids , leading to material failure @xcite which might also be sensitive to the void organization . +
the dominant structural feature of these void materials consists of the necks confined between three or four voids , which narrow to zero thickness as rigidity - loss is approached .
these necks or columns have been used throughout this study to explain elastic behaviour of different void materials . in an associated study @xcite a beam model
is proposed which takes into account the topology of the network of necks in hcp and fcc void material and derives a scaling theory for the elastic behaviour close to rigidity - loss .
the simple beam model shows very good agreement with the detailed fe simulation in the present paper .
a notable feature of the present results is the occurrence of poisson ratios that lie outside the well known bonds for isotropic materials .
this is also reproduced and well explained with the beam model .
+ the present study is motivated by the elastic properties of solid foams . in case of monodisperse bubbles
shaped boundaries @xcite or electromagnetic fields @xcite could create regularly arranged bubble crystals .
however , the investigated geometries differ substantially from solid foam . close to a gas fraction corresponding to touching bubbles
the bubbles would remain approximately spherical , justifying the approach of spherical voids . with increasing gas fraction , bubbles in a foam deform . seeking a state of minimum surface energy they form thin
, flat lamellas between neighbouring bubbles and elongated plateau borders confined between three bubbles .
these plateau borders correspond to the necks or columns , mentioned above , but show a substantially different geometry @xcite .
this presumably yields a different elastic behaviour which is less sensitive to the solid fraction .
gibson and ashby @xcite modelled an open - cell foam structure to consist of a network of straight beams with no necks but constant beam thickness .
they found the young s modulus to scale with the square of the void fraction while necks give a scaling to the power of 3.5 , as shown in our other study @xcite .
the realistic scaling presumably lies somewhere between these extreme approximations . + for computations of more realistic foams at very high gas fractions , one could use the surface evolver @xcite to find realistic geometries .
subsequently , one may apply fe simulations similar to the present work , in order to extract the elastic behaviour of these foams .
however , it is expected that the qualitative behaviour is similar . for these extensions
the present study provides valuable reference data which clarify the physical properties and may also serve as validation data for further assessments .
we gratefully acknowledge fruitful discussions with christophe poulard .
computation time was provided by the center for information services and high performance computing ( zih ) at tu dresden .
we acknowledge support from the european research council ( erc ) under the european union s seventh framework program ( fp7/2007 - 2013 ) in form of an erc starting grant , agreement 307280-pomcaps . we acknowledge support from the european centre for emerging materials and processes ( ecemp ) at tu dresden and the helmholtz - alliance liquid metal technologies ( limtech ) .
dw acknowledges the support of sfi .
43 natexlab#1#1[1]`#1 ` [ 2]#2 [ 1]#1 [ 1]http://dx.doi.org/#1 [ ] [ 1]pmid:#1 [ ] [ 2]#2 , , , , , ( ) .
, , , , ( ) . , , , , , , ( ) .
, , , , , , , , , , , , , ( ) . , , , , ( ) .
, , , ( ) . , , , , ( ) .
, , , , , ( ) .
, , , , ( ) . , , , ( ) . , , ( ) .
, , , ( ) . , , ( ) .
, , , ( ) . , , , ( ) .
, , , , , ( ) . , , , ( ) . , , , ( ) . , , , , ( ) .
, , , , , , , , , , , ( ) . , , ( ) .
, , , , , ( ) . , , , ( ) .
, , , , , ( ) . , , , ( ) .
, , , ( ) . , , , , , ( ) . , , , , .
, , , , , , ( ) . , , ph.d .
thesis , tu dresden , .
, , , , , , , , . | in this work the linear elastic properties of materials containing spherical voids are calculated and compared using finite element simulations .
the focus is on homogeneous solid materials with spherical , empty voids of equal size .
the voids are arranged on crystalline lattices ( sc , bcc , fcc and hcp structure ) or randomly , and may overlap in order to produce connected voids . in that way
, the entire range of void fraction between 0.00 and 0.95 is covered , including closed - cell and open - cell structures . for each arrangement of voids and for different void fractions
the full stiffness tensor is computed . from this , the young s modulus and poisson ratios are derived for different orientations .
special care is taken of assessing and reducing the numerical uncertainty of the method . in that way
, a reliable quantitative comparison of different void structures is carried out . among other things
, this work shows that the young s modulus of fcc in the ( 1 1 1 ) plane differs from hcp in the ( 0 0 0 1 ) plane , even though these structures are very similar . for a given void fraction sc offers the highest and the lowest young s modulus depending on the direction .
for bcc at a critical void fraction a switch of the elastic behaviour is found , as regards the direction in which the young s modulus is maximised . for certain crystalline void arrangements and certain directions poisson ratios between 0 and 1 were found , including values that exceed the bounds for isotropic materials . for subsequent investigations the full stiffness tensor for a range of void arrangements and void fractions
are provided in the supplemental material .
void material , young s modulus , poisson ratio , foam , finite element method |
high redshift galaxy populations are now being detected back to 780 million years after the big bang ( spectroscopically confirmed:@xmath0=6.96 ; iye @xcite ) , probing into the epoch of cosmic reionization ( e.g. , fan @xcite ; hu & cowie @xcite ) .
many of these very distant galaxies show evidence for star formation activity ( e.g. , taniguchi @xcite ) . some are even found to be hyperluminous infrared galaxies ( hlirgs ; bertoldi @xcite ; wang @xcite ) with far - infrared ( fir ) luminosities exceeding 10@xmath15 , suggesting vigorous star formation and/or agn activity . to probe the earliest stages of galaxy formation and the importance of agn in this process ,
it is necessary to study the star formation characteristics of these galaxies . a good diagnostic to examine the star - forming environments in distant hlirgs are observations of molecular gas , the fuel for star formation .
the by far brightest and most common indicator of molecular gas in galaxies is line emission from the rotational transitions of carbon monoxide ( co ) , which was detected in @xmath1640 galaxies at high redshift ( @xmath0@xmath91 ; see solomon & vanden bout @xcite for a review ) .
these observations have revealed molecular gas reservoirs with masses of @xmath910@xmath17 in these galaxies , even in the highest redshift quasar known , j1148 + 5251 at @xmath0=6.42 ( walter @xcite , @xcite ; bertoldi @xcite ) .
although co is a good tracer of the total amount of molecular gas in a galaxy , due to the relatively low critical density of @xmath18@xmath2 required to collisionally excite its lower @xmath19 transitions , it is not a reliable tracer of the dense molecular cloud cores where the actual star formation takes place .
recent studies of nearby actively star - forming galaxies have shown that hydrogen cyanide ( hcn ) is a far better tracer of the dense ( @xmath20@xmath2 ) molecular gas where stars actually form ( e.g. gao & solomon @xcite , @xcite , hereafter : gs04a , gs04b ) . in the local universe
it was found that the hcn luminosity ( @xmath5 ) scales linearly ( unlike @xmath21 ) with the fir luminosity ( @xmath11 ) over 78 orders of magnitude , ranging from galactic dense cores to ulirgs ( wu et al.@xcite ) . as @xmath11 traces the massive star formation rate ( unless agn heating is significant ) , this implies that hcn is also a good tracer of star formation .
hcn has now also been detected in five galaxies at @xmath0@xmath92 ( solomon @xcite ; vanden bout @xcite ; carilli @xcite , hereafter : c05 ; wagg @xcite ; gao @xcite , hereafter : g07 ) .
adding a number of upper limits obtained for other high-@xmath0 galaxies , these observations indicate that the more luminous , higher redshift systems systematically deviate from the linear @xmath5@xmath22 correlation found in the local universe ( g07 ) , and hint at a rising slope of the relation toward high @xmath11 and/or @xmath0 . to further investigate this apparent non - linear ,
rising trend , our aim has been to extend the range of existing hcn observations beyond redshift 6 and to higher @xmath11 . in this letter , we report sensitive vla observations of emission toward the @xmath0=6.42 quasar j1148 + 5251 , the highest redshift source detected in co. a previous , less sensitive search for emission in this source has yielded no detection ( c05 ) .
we use a concordance , flat @xmath23cdm cosmology throughout , with @xmath24=71mpc@xmath25 , @xmath26=0.27 , and @xmath27=0.73 ( spergel @xcite ) .
we observed the transition line ( @xmath28ghz ) toward j1148 + 5251 using the vla in d configuration in 10tracks on 2004 june 21 and 26 ( these observations were discussed by c05 ) , and between 2007 may 01 and 13 . at the target redshift of 6.419 , the line is shifted to 23.892873ghz ( 11.96 mm ) .
the total integration time amounts to 80hr .
observations were performed in fast - switching mode using the nearby source 11534 + 49311 ( at @xmath29 distance to j1148 + 5251 ) for secondary amplitude and phase calibration .
observations were carried out under very good weather conditions with 26 antennas [ including 9 expanded very large array ( evla ) antennas in 2007 ] : the phase stability was excellent ( typically @xmath715@xmath30 phase rms for the longest baselines ) .
the qso 3c286 was observed as a primary flux calibrator .
the resulting flux scale is accurate within the standard 15% uncertainty . in 2004 , two 25mhz wide intermediate frequency bands ( ifs ) with seven 3.125mhz channels each were observed simultaneously , one centered at the line frequency ( 23.8929ghz ) , and one centered at 24.0430ghz for continuum monitoring .
this leads to an effective bandwidth of 21.875mhz for both the line and continuum observations , corresponding to 274 at 23.9ghz . for lines as broad as co ( 279 fwhm ; bertoldi @xcite )
, this setup would miss only @xmath3115% flux in the line wings . in 2007 ,
observations were carried out in continuum mode , with one 25mhz wide if centered at the line frequency ( 23.8929ghz ) , and one 50mhz wide if centered at 23.7649ghz to monitor the source s continuum . for data reduction and analysis ,
the @xmath32 package was used .
all data were mapped using the clean algorithm and ` natural ' weighting ; this results in synthesized beams of 4.0@[email protected]@xmath33 for the line map and 3.5@[email protected]@xmath33 ( @xmath1620kpc at @xmath35 ) for the continuum map ( see fig .
the final rms in the integrated line map is 15@xmath36jy beam@xmath25 , and 14@xmath36jy beam@xmath25 in the continuum map . averaging both uv datasets to a ` high - sensitivity ' continuum map leads to an rms of 11@xmath36jy beam@xmath25 .
r c c c & ( 7 @xmath37 15 ) & @xmath73.3 & 1 + & @xmath7360 & @xmath7142 & 2 + & 570 @xmath37 57 & 26.4 @xmath37 2.6 & 3 + & 2450 & 26.9 @xmath37 2.4 & 2 + & 2140 & 17.3 @xmath37 2.4 & 2 +
no emission is detected in j1148 + 5251 ( fig .
[ f1 ] , left ) .
we derive a peak flux density of ( 7 @xmath37 15)@xmath36jy beam@xmath25 at the source s co position ( see table [ tab-1 ] ) , setting an upper limit limits ; however , note that all conclusions hold for 3@xmath38 limits . ] of 30@xmath36jy to the emission line peak flux .
no continuum emission is detected at and/or close to the line frequency ( fig .
[ f1 ] , right ) . from the ` high - sensitivity ' continuum map ,
we derive a peak flux density of ( 11 @xmath37 11)@xmath36jy beam@xmath25 at the source s position , setting an upper limit of 22@xmath36jy to the 23.9ghz continuum flux density .
this is consistent with the model - predicted continuum level of @xmath318@xmath36jy ( beelen @xcite ) . from our observations ,
we derive a limit to the line luminosity of @xmath39kpc@xmath4 ( assuming a hcn / co linewidth ratio of 0.67 , i.e. , the average value for the four @xmath0@xmath92 hcn - detected quasars@xmath92 sample , but is up to 100% in nearby starburst galaxies like ngc253 ( e.g. , knudsen @xcite ) . ] ; see table [ tab-2 ] for references ) , corresponding to 13% of the co luminosity ( walter @xcite ) . using the fir luminosity derived by beelen ( @xcite ) galaxies are only accurate within a factor of 2 . ]
, we find @xmath22/@xmath5@xmath96600 .
the nondetection of in j1148 + 5251 at the depth of our observations has several implications . for the following comparison with other galaxies ( for which @xmath6 and @xmath5 are given in the ground - state transitions ) ,
we assume that the and lines are thermalized , so that @xmath40=@xmath41 and @xmath42=@xmath43 . considering the high co excitation in this source ( bertoldi @xcite ) and
the fact that both lines arise from low @xmath19 transitions , this is likely a valid assumption .
l c c l sdssj1148 + 5251 & @xmath96600 & @xmath70.13 & 1,2,3 + cloverleaf & 1305@xmath37360 & [email protected] & 4,5,6 + irasf10214 + 4724 & 2835@xmath37855 & [email protected] & 7,8,9 + vcvj1409 + 5628 & 2615@xmath37860 & [email protected] & 9,10,3 + apm08279 + 5255 & 1000 & 0.27 & 11,12,13 + smmj16359 + 6612 & 1550@xmath37550 & [email protected] & 14,15,16 + & & & + avg . ` normal ' spiral ( 33 ) & 580@xmath12 & 0.04@xmath44 & 17,14 + avg .
@xmath0@xmath450 lirg ( 23 ) & 650@xmath46 & 0.07@xmath47 & 17,14 + avg .
@xmath0@xmath450 ulirg ( 9 ) & 740@xmath13 & 0.17@xmath8 & 17,14 + avg .
@xmath0@xmath450 ` all ' ( 33 + 23 + 9 ) & 680@xmath48 & 0.05@xmath49 & 17,14 + avg .
@xmath0@xmath92 galaxy ( 5 ) & 1525@xmath14 & 0.17@xmath10 & 416 + emission is considered a good tracer for the total amount of molecular gas in a galaxy , while emission is considered a good tracer for the dense peaks of the molecular mass distribution .
the @xmath5/@xmath6 ratio thus is considered a measure for the dense fraction of molecular gas in a galaxy/@xmath6 as a tracer of the dense gas fraction . ] . on average ,
this ratio is about four to five times higher in local ulirgs compared to normal , nearby spiral galaxies [ median value of 0.17@xmath8 ( ulirgs ) vs. 0.04@xmath44 ( spirals ) ; see table [ tab-2 ] ] .
however , there does not appear to be a further increase toward the even more fir - luminous systems at @xmath0@xmath92 ( 0.17@xmath10 on average ) within the statistical uncertainties ( all values rederived from gs04ab ; g07 ) .
the upper limit for @xmath5/@xmath6 in j1148 + 5251 is 0.13 , placing it below the median but within the range of values measured for ulirgs and @xmath0@xmath92 galaxies ( fig .
our observations thus confirm the finding of g07 that galaxies brighter than @xmath11=10@xmath50 appear to have a higher fraction of dense molecular gas than normal spiral galaxies ; however , there is no indication that this fraction rises further toward the highest @xmath11 , or that it changes with redshift .
the fir luminosity is thought to originate dominantly from dust - reprocessed light of young massive stars , and thus to be a good tracer of the star - formation rate ( sfr ) in a galaxy .
star formation in galaxies takes place in dense molecular clouds that are traced well by hcn emission , resulting in a linear correlation between the fir and hcn luminosities ( wu @xcite ) .
however , g07 have found a mild increase of @xmath11/@xmath5 with increasing fir luminosity between normal spiral galaxies and ulirgs [ @xmath22/@xmath5=580@xmath12 ( spirals ) vs. 740@xmath13 ( ulirgs ) on average ; see table [ tab-2 ] ] .
this trend appears to get stronger toward the even more fir - luminous @xmath0@xmath92 systems ( @xmath11/@xmath5=1525@xmath14 on average )
. the lower limit of @xmath11/@xmath5@xmath96600 obtained for j1148 + 5251 is consistent with such a rising trend toward high-@xmath0 systems , and extends it toward higher redshift ( fig .
[ f2]b ) . due to the fact that the high-@xmath0 sources also have systematically higher @xmath11 than the nearby galaxies
, it remains unclear whether we observe a rising trend with redshift or with @xmath11 ( fig .
[ f2]c ) , or both .
krumholz & thompson ( @xcite ) argue that @xmath22/@xmath5 is expected to be higher for galaxies with a median ( molecular ) gas density @xmath51 close to or higher than the critical density @xmath52 required for excitation of the observed hcn transition than for galaxies with lower @xmath53 . in their case , they define star formation efficiency as the fraction of the mass that is converted into stars per dynamical time of the system . note that this is different than the star formation rate per unit total gas mass .
they argue that the non - linear relation between @xmath11 and @xmath6 ( e.g. , kennicutt @xcite , @xcite ; gs04b ; riechers @xcite ) arises due to the fact that co traces all gas .
the star - formation rate is then dictated by the density @xmath54 divided by the free - fall time @xmath55 ( @xmath56 ) , giving the standard schmidt - law : star formation rate @xmath57 , or @xmath11 @xmath58@xmath59 . for molecules like hcn , which only trace the small fraction of dense gas clouds directly associated with star formation in normal galaxies , @xmath55 is roughly fixed by @xmath60 .
hence the star formation rate shows a linear relationship with @xmath54 , or @xmath11 @xmath61@xmath62 .
however , in extreme galaxies , where @xmath51 in the molecular ism approaches @xmath63 , @xmath55 again becomes relevant ( i.e. , hcn emission no longer selects just the rare , dense peaks whose density is fixed by @xmath52 , but instead traces the bulk of the ism , whose density can vary from galaxy to galaxy , and thus the variation of @xmath54 and @xmath55 re - enter the calculation ) , and the relationship approaches @xmath11 @xmath61@xmath59 ( and @xmath5 @xmath64 ) .
interestingly , current data show a marginal trend for a changing power - law index at the highest luminosities of the type proposed by krumholz & thompson .
this change in power - law index from 1 to 1.5 would suggest that , in these extreme luminosity systems , @xmath51 approaches @xmath65 .
more systems at high luminosity are required to confirm this trend of changing power - law index .
like most of the @xmath0@xmath92 hcn - detected sources , j1148 + 5251 is a quasar .
it has been found that , even for such strong agn galaxies , the bulk of @xmath11 is likely dominatly heated by star formation in most cases ( e.g. , carilli @xcite ; omont @xcite ; beelen @xcite ; riechers @xcite ) .
however , based on radiative transfer models of the dust sed of j1148 + 5251 , li ( @xcite ) argue that this source may currently undergo a ` quasar phase ' , in which agn heating of the hot and warm dust contributes significantly to @xmath11 .
if correct , this may be an alternative explanation for the elevated @xmath11/@xmath5 in this galaxy .
the ( rest - frame ) ir properties ( tracing emission from hot dust ) of j1148 + 5251 are similar to those of other @xmath0@xmath96 quasars with much lower @xmath11 ( tracing emission from warm dust ) , and even to local quasars ( jiang @xcite ) .
this supports the assumption that the hot dust in j1148 + 5251 is dominantly heated by the agn ; however , the lack of a correlation between @xmath66 and @xmath11 in quasars indicates that the warm dust may still be dominantly heated by star formation .
moreover , j1148 + 5251 follows the radio - fir correlation for star - forming galaxies ( carilli @xcite ) , which also suggests a starburst origin for the dominant fraction of @xmath22 .
furthermore , one of the @xmath0@xmath92 hcn detections and some of the meaningful limits are submillimeter galaxies without a known luminous agn , but are still offset from the local @xmath11/@xmath5 relation .
it thus appears unlikely that agn heating alone can account for the higher average @xmath11/@xmath5 in the high-@xmath0 galaxy sample . even when assuming the highest @xmath11/@xmath5 of 2835 found among all hcn - detected galaxies in table [ tab-2 ] , the depth of our observations is sufficient to detect a galaxy with the redshift and @xmath11 of j1148 + 5251 ( @xmath0=6.42 ) in hcn emission at a signal - to - noise ratio of @xmath94.5 .
to first order , our lower limit thus is consistent with previous suggestions ( g07 ) that @xmath22/@xmath5 ratios in high redshift sources lie systematically above those for nearby galaxies .
the scatter around this trend is still significant , and will primarily be improved by increasing the number of hcn - detected galaxies at high @xmath0 . in addition , it will be important to improve on the main sources of error for the individual high-@xmath0 detections ( e.g. , signal - to - noise limited hcn / co linewidth ratio , accuracy of the fir sed fit , agn bias of @xmath11 ) .
the statistical and individual results , so far , would even be consistent with an even stronger increase in @xmath22/@xmath5 toward the highest @xmath0 and/or @xmath11 .
our study of j1148 + 5251 may hint at such an effect . clearly , it is desirable to obtain more sensitive observations of this source to further investigate this issue .
due to its superior collecting area and high calibrational stability , the vla is ideally suited for such a sensitive study .
although j1148 + 5251 is the most co- and fir - luminous @xmath0@xmath96 galaxy known , 80hr of vla observations were necessary to obtain the current limit . in a favourable case
, the line may have a strength of about 1.5 times the current rms . to obtain a solid 5@xmath38 detection of such a line , of order 1000hr of observations with the vla would be required .
due to improved receivers and antenna performance , the fully operational evla will be by a factor of two more sensitive to spectral lines of several 100 width ( such as in j1148 + 5251 ) , but will still require long integration times .
studies of dense gas at @xmath0@xmath96 thus appear to require an order of magnitude increase in collecting area , such as offered by future facilities like the ska phase i demonstrator ( e.g. , carilli @xcite ) , which can serve as a low frequency counterpart to the atacama large millimeter / submillimeter array ( alma ) .
the authors would like to thank philip solomon and yu gao for access to their dataset of nearby galaxies .
we also thank axel wei for helpful discussions .
d. r. thanks vernesa smoli for a great piece of code .
d. r. acknowledges support from the deutsche forschungsgemeinschaft ( dfg ) priority program 1177 . c. c.acknowledges support from the max - planck - gesellschaft and the alexander von humboldt - stiftung through the max - planck - forschungspreis 2005 .
we thank the referee for helpful comments .
beelen , a. , cox , p. , pety , j. , 2004 , a&a , 423 , 441 beelen , a. , cox , p. , benford , d. j. , 2006 , apj , 642 , 694 bertoldi , f. , carilli , c. l. , cox , p. , a&a , 406 , l55 bertoldi , f. , cox , p. , neri , r. , et al . 2003b , a&a , 409 , l47 carilli , c. l. , bertoldi , f. , rupen , m. p. , 2001
, apj , 555 , 625 carilli , c. l. , walter , f. , bertoldi , f. , 2004 , aj , 128 , 997 carilli , c. l. , solomon , p. m. , vanden bout , p. a. , 2005 , apj , 618 , 586 [ c05 ] carilli , c. l. 2006 , ska memo 70 fan , x. , carilli , c. l. , & keating , b. 2006 , ara&a , 44 , 415 gao , y. , & solomon , p. m. 2004a , apjs , 152 , 63 [ gs04a ] gao , y. , & solomon , p. m. 2004b , apj , 606 , 271 [ gs04b ] gao , y. , carilli , c. l. , solomon , p. m. , & vanden bout , p. a. 2007 , apj , 660 , l93 [ g07 ] graci - carpio , j. , garca - burillo , s. , planesas , p. , & colina , l. 2006 , apj , 640 , l135 hu , e. m. , & cowie , l. l. 2006 , nature , 440 , 1145 iye , m. , 2006 , nature , 443 , 186 jiang , l. , fan , x. , hines , d. c. , 2006 , aj , 132 , 2127 kennicutt , r. c. 1998a , apj , 498 , 541 kennicutt , r. c. 1998b , ara&a , 36 , 189 kneib , j .-
, van der werf , p. p. , knudsen , k. k. , 2004 , mnras , 349 , 1211 kneib , j .-
p . , neri , r. , smail , i. , 2005 , a&a , 434 , 819 knudsen , k. k. , walter , f. , weiss , a. , bolatto , a. , riechers , d. a. , & menten , k. 2007 , apj , 666 , 156 krumholz , m. r. , & thompson , t. a. 2007 , apj , in press ( arxiv:0704.0792 ) li , y. , hopkins , p. f. , hernquist , l. , 2007 , apj , submitted ( arxiv:0706.3706 ) omont , a. , cox , p. , bertoldi , f. , 2001 , a&a , 374 , 371 riechers , d. a. , walter , f. , carilli , c. l. , 2006a , apj , 645 , l13 riechers , d. a. , walter , f. , carilli , c. l. , 2006b , apj , 650 , 604 solomon , p. , vanden bout , p. , carilli , c. , & guelin , m. 2003 , nature 426 , 636 solomon , p. m. , & vanden bout , p. a. 2005 , ara&a , 43 , 677 spergel , d. n. , bean , r. , dor , o. , 2007 , apjs , 170 , 377 taniguchi , y. , ajiki , m. , nagao , t. , 2005 , pasj , 57 , 165 vanden bout , p. , solomon , p. , & maddalena , r. 2004 , apj , 614 , l97 wagg , j. , wilner , d. j. , neri , r. , downes , d. , & wiklind , t. 2005 , apj , 634 , l13 walter , f. , bertoldi , f. , carilli , c. l. , et al .
2003 , nature , 424 , 406 walter , f. , carilli , c. , bertoldi , f. , et al .
2004 , apj , 615 , l17 wang , r. , carilli , c. l. , beelen , a. , 2007 , aj , 134 , 617 wei , a. , henkel , c. , downes , d. , & walter , f. 2003 , a&a , 409 , l41 wei , a. , downes , d. , neri , r. , 2007 , a&a , 467 , 955 wu , j. , evans ii , n. j. , gao , y. , solomon , p. m. , shirley , y. l. , & vanden bout , p. a. 2005 , apj , 635 , l173 | we report a sensitive search for the emission line towards sdssj114816.64 + 525150.3 ( hereafter : j1148 + 5251 ) at @xmath0=6.42 with the very large array ( vla ) .
hcn emission is a star formation indicator , tracing dense molecular hydrogen gas ( @xmath1@xmath2 ) within star - forming molecular clouds .
no emission was detected in the deep interferometer maps of j1148 + 5251 .
we derive a limit for the hcn line luminosity of @xmath3k pc@xmath4 , corresponding to a hcn / co luminosity ratio of @xmath5/@[email protected] .
this limit is consistent with a fraction of dense molecular gas in j1148 + 5251 within the range of nearby ultraluminous infrared galaxies ( ulirgs ; median value : @xmath5/@xmath6=0.17@xmath8 ) and hcn - detected @xmath0@xmath92 galaxies ( 0.17@xmath10 ) .
the relationship between @xmath5 and @xmath11 is considered to be a measure for the efficiency at which stars form out of dense gas . in the nearby universe , these quantities show a linear correlation , and thus , a practically constant average ratio . in j1148
+ 5251 , we find @xmath11/@xmath5@xmath96600 .
this is significantly higher than the average ratios for normal nearby spiral galaxies ( @xmath11/@xmath5=580@xmath12 ) and ulirgs ( 740@xmath13 ) , but consistent with a rising trend as indicated by other @xmath0@xmath92 galaxies ( predominantly quasars ; 1525@xmath14 ) .
it is unlikely that this rising trend can be accounted for by a contribution of active galactic nucleus ( agn ) heating to @xmath11 alone , and may hint at a higher median gas density and/or elevated star - formation efficiency toward the more luminous high - redshift systems .
there is marginal evidence that the @xmath11/@xmath5 ratio in j1148 + 5251 may even exceed the rising trend set by other @xmath0@xmath92 galaxies ; however , only future facilities with very large collecting areas such as the square kilometre array ( ska ) will offer the sensitivity required to further investigate this question . |
the effect of nonmagnetic impurity zn in high - temperature cuprate superconductors has attracted great interests .
the impurity zn induces curie magnetism , a local staggered susceptibility in the normal state @xcite , an in - gap antiferromagetic dynamical spin susceptibility , and a virtual bound electronic state around zn in the superconducting state @xcite .
the perturbation theory for the mean - field states close to two - dimensional antiferromagnetic instability allows coexistence of the staggered - type , generalized ruderman - kittel - kasuya - yosida ( rkky ) oscillation and the local density of states of electrons with a zero bias resonance peak at finite temperatures @xcite .
however , recent numerical calculation using gtzwiller approximation indicates that the frozen staggered moments are locally reduced by zn in the superconducing state of a moderately doped two - dimensional @xmath7 model @xcite . according to this theory , the zero bias resonance peak in the local density of states and the local staggered moments are exclusive at @xmath8=0 k. it remains to be solved whether the zn strengthens @xcite or weakens @xcite local magnetic correlation around zn in the optimally doped or overdoped system . nuclear magnetic resonance
( nmr ) and nuclear quadrupole resonance ( nqr ) are powerful techniques to study the local states on crystalline imperfection , dilute exchange spin networks , friedel oscillation , rkky oscillation , or kondo screening effect @xcite .
we focus on the local electronic states near zn in yba@xmath0(cu@xmath1zn@xmath2)@xmath3o@xmath9 ( @xmath107-@xmath11 ) ( y123 ) and yba@xmath0(cu@xmath1zn@xmath2)@xmath5o@xmath6 ( y124 ) via zn - induced cu nqr satellite spectra @xcite .
y123 is a nearly optimally doped system ( oxygen content @xmath126.9 ) , while y124 is a stoichiometric underdoped one .
zn - neighbor @xmath13y nmr signals and their relaxation times have intensively been studied for y123 @xcite and for y124 @xcite . however , there is only a few direct observation of the local electronic state via zn - neighbor cu nqr / nmr signals ; a report on the satellite cu nuclear spin - lattice relaxation of a non - superconducting y124 with heavily zn doping @xcite but no report on the superconducting y123 or y124 with small concentrations of zn . in this paper
, we report on a study of satellite signals in the planar cu(2 ) nqr spectra of zn - substituted superconductors y123 and y124 . the wipeout effect on the cu nqr main signals and the growth of the zn - induced cu satellite signals indicate that the satellite but not the main signal includes the zn - neighbor cu nuclei .
we demonstrate that the satellite cu nuclear spin - lattice relaxation time is shorter than the main one in the superconducting y123 and y124 .
the local antiferromagnetic correlation near zn is enhanced even for the optimally doped system .
high quality powder samples of a pure y123 ( @xmath14=6.98 studied in ref .
@xcite ) , a slightly oxygen - deficient y123 ( @xmath14=6.89 in refs .
@xcite ) , zn - substituted y123 ( the annealed samples in @xcite ) , a pure y124 and zn - substituted y124 @xcite were prepared for the present study . the superconducting transition temperatures are @xmath15@xmath1692 k both for y123 with @xmath14=6.98 and with @xmath14=6.89 ; @xmath15=81 , 69 , and 46 k for y123 with zn content @xmath17=0.007 , 0.017 , 0.033 ; @xmath15=82 k for the pure y124 , and @xmath15=68 , 56 , 37 , and 27 k for y124 with zn content @xmath17=0.005 , 0.01 , 0.015 , 0.022 , respectively . to estimate the relative intensity of cu nqr spectra , the respective samples of y123 and of y124 with nearly the same volume were prepared .
the oxygen contents conform to the values in the previous studies .
a phase - coherent - type pulsed spectrometer was utilized to perform cu nqr experiments .
the zero - field cu nqr frequency spectra with quadrature detection were obtained by integration of cu nuclear spin - echoes with changing the frequency in a two - pulse sequence ( @xmath18/2-@xmath19-@xmath18 echo ) . a typical width of the first exciting @xmath18/2-pulse @xmath20 was about 3 @xmath21s ( the excited frequency region @xmath2283 khz from 2@xmath23=@xmath18/2 ) .
the cu nuclear spin - lattice relaxation curves were measured by an inversion recovery technique , where the nuclear spin - echo @xmath24 was monitored as a function of the time @xmath25 in a sequence of @xmath26 echo . the pulse interval for the nqr spectrum measurement was @xmath19=15 @xmath21s .
figure 1 shows @xmath27cu(2 ) nqr spectra ( upper panels ) and the @xmath28cu(2 ) nuclear spin - lattice relaxation curves @xmath29 ( lower panels ) for the oxygen - deficient y123 ( left ) , zn - substituted , nearly optimally doped y123 ( middle ) , and zn - substituted , underdoped y124 ( right ) at @xmath8=4.2 k. the cu nqr spectra are normalized by the respective values of peak intensity .
zn doping induces lower frequency tails or satellite signals both for y123 and y124 , which is known in refs .
oxygen deficiency is also known to yield similar low frequency satellite or shoulder in the cu(2 ) nqr spectrum for y123 without zn @xcite .
since y124 has no appreciable oxygen deficiency , the satellite signals result from purely zn doping effect .
= 3.7 in for oxygen deficient y123 in the lower panel of fig.1 , the satellite and the main signals show nearly the same relaxation curve @xmath30 , and their relaxation times @xmath31 are longer than a few seconds .
this is consistent with the previous reports @xcite . on the other hand , for zn - doped y123 and y124
, the zn - induced satellite signal shows a shorter relaxation time than the main signal , and their relaxation times are shorter than 1.0 s. thus , the lower frequency satellite in the cu nqr spectrum for zn - doped y123 possesses an electronic state different from that for the slightly oxygen deficient one . our observation for zn - substituted y123 and y124 that the lower frequency signals in the broad nqr spectra show the shorter @xmath31 is in parallel to that for zn - free underdoped la@xmath32sr@xmath2cuo@xmath5 @xcite .
for y124 ( y123 ) , we confirm that the satellite recovery curve @xmath30 is nearly independent of frequency over 28.4 - 28.6 mhz ( 30.1 - 30.5 mhz ) and so is the main one over 29.2 - 30.2 mhz ( 31.5 - 31.6 mhz ) . here , the excited frequency region by the pulse is @xmath2283 khz at each nqr frequency .
thus , the frequency dependence of the relaxation curve is not monotonic .
figure 2(a ) shows the @xmath27cu(2 ) nqr spectrum of y123 with zn content of @xmath17=0.033 ( @xmath15=46 k ) at @xmath8=4.2 k and the hot spot regions ( shaded bars ) excited in this spectrum to measure the isotope dependence of cu(2 ) nuclear spin - lattice relaxation curves .
figure 2(b ) shows the respective @xmath27cu(2 ) nuclear spin - lattice relaxation curves @xmath30 for the above y123 at @xmath8=4.2 k. the dashed and the solid curves are the respective least - squares fitting results for @xmath28cu(2 ) and @xmath33cu(2 ) , using a theoretical function including a stretched exponential function ( see below ) . the nuclear spin - lattice relaxation rate due to a magnetic origin is proportional to the square of the @xmath27cu nuclear gyromagnetic ratio @xmath34 . since we could not find any large deviation from the scaling with respect to the reduced time @xmath35 , the overlapping effect of the broad @xmath36cu(2 ) lines is small .
= 3.5 in figure 3 shows the @xmath28cu(2 ) nqr spectra with zn doping at @xmath8=100 k(@xmath37 @xmath15 ) for y123 with zn contents of @xmath17=0 , 0.007 , 0.017 , 0.033 ( a ) , and for y124 with zn @xmath17=0 , 0.01 , 0.015 , 0.022 ( b ) from the top to the bottom .
the normal - state cu nqr intensity is not affected by the meissner shielding effect .
the main signal intensity of the @xmath28cu(2 ) nqr spectrum around 31.5 mhz for y123 and around 29.8 mhz for y124 rapidly decreases with increasing zn concentration .
alternatively , the lower frequency tail or the broad satellite signal increases with increasing zn concentration .
the solid and the dotted curves are numerical simulations using three gaussian functions ( @xmath38 , @xmath39 , and @xmath40 ) . by using two gaussian functions , @xmath38 and @xmath39 , we can reproduce the main spectra .
the shaded gaussian @xmath40 is the satellite spectrum .
= 3.7 in in figs .
3(c ) and 3(d ) , the estimated intensity of the main spectrum @xmath41(=@xmath38+@xmath39 ) ( upward closed triangles ) and that of the satellite spectrum @xmath42(=@xmath40 ) ( open circles ) normalized by @xmath43 are plotted as functions of the plane - site zn(2 ) concentrations @xmath44=3@xmath17/2 for y123 ( c ) and of @xmath44=2@xmath17 for y124 ( d ) .
the open squares are the total intensity(=@xmath45+@xmath46+@xmath40 ) .
the values of @xmath41 and @xmath42 are estimated after @xmath47 correction using @xmath48 from fitting @xmath49s)=@xmath48exp@xmath50 $ ] with the fitting parameters @xmath48 , @xmath51 and @xmath52 .
the dashed curves are numerical calculations of cu intensity with various zn configurations , using the binomial - probability @xmath53@xmath54@xmath55 , where @xmath56 is the number of the @xmath57th nearest neighbor ( nn ) cu sites .
( 1-@xmath44 ) is the probability that a plane site is occupied by a cu atom but not a zn atom , or it is the total cu nqr intensity with zn doping .
with respect to zn configuration , the probability of finding cu atoms can be seen in the decomposition of ( 1-@xmath44)=(1-@xmath44)@xmath58@xmath53 . in fig .
4 , schematic illustrations of a @xmath28cu(2 ) nqr spectrum ( a ) and the cuo@xmath0 plane with a zn impurity ( b ) are given for a guide to the calculation .
= 3.5 in the calculated curves on the cu nqr intensity are in what follows . in figs .
3(c ) and 3(d ) , the decreasing functions with increasing @xmath44 are @xmath59= @xmath60=@xmath61 ( 1 ) , @xmath62=@xmath63 ( 2 ) , and @xmath64=@xmath65 ( 3 ) .
these are the probability of finding the cu atoms with all cu atoms in their 1st nn positions ( 1 ) , in their 12 neighboring ( up - to 3rd nn ) positions ( 2 ) , and in their 20 neighboring ( up - to 4th nn ) positions ( 3 ) . in other words ,
the diminished cu nqr signals come from the cu atoms with at least one zn atom in their 1st nn positions ( the 1st nn wipeout effect ) ( 1 ) , in their up - to 3rd nn positions ( up - to 3rd nn wipeout effect ) ( 2 ) , and in their up - to 4th nn positions ( up - to 4th nn wipeout effect ) ( 3 ) . in figs .
3(c ) and 3(d ) , the increasing functions with increasing @xmath44 are @xmath59= @xmath66=@xmath67 ( 4 ) , and @xmath68=@xmath69 ( 5 ) .
these are the probability of finding the cu atoms with one zn atom in their 4th nn positions _ or _ in the 2nd and the 3rd nn positions ( 4 ) , and in their @xmath57(=1 , 2 , 3)th nn positions ( 5 ) . the relative intensity of the satellite cu nqr spectrum of y124 in fig .
3(d ) is smaller than that of y123 in fig .
3(c ) . for zn - substituted la@xmath32sr@xmath2cuo@xmath5 , the disappearance of the cu nqr signal without any satellite signal
is observed @xcite .
thus , it is likely that the observable satellite signal depends on the carrier - doping level or the underlying magnetic correlation . as seen in figs .
3(c ) and 3(d ) , the zn - doping dependence of the total intensity for y123 ( @xmath700.017 ) and y124 ( @xmath700.022 ) agrees with that of the case ( 3 ) within the experimental accuracy , where the up - to 3rd nn cu sites to zn are unobservable .
the zn doping dependence of the main intensity for y123 ( @xmath700.017 ) and y124 ( @xmath700.022 ) agrees with that of the case ( 2 ) , where the up - to 4th nn cu sites to zn are unobservable .
the zn - nearest - neighbor cu nqr spectrum will be shifted and broadened due to local crystalline or electric deformation , being unobservable probably due to a shorter relaxation time than a few @xmath21s . for y123 ,
the satellite signal agrees with either of the 4th nn cu site or the 2nd and the 3rd nn cu site to zn [ the case ( 4 ) ] . with @xmath700.017 ,
the former assignment is consistent with those to the total and the main signals , whereas with @xmath17=0.033 , the latter is consistent with those to the total and the main ones . for y124 ,
the satellite signal agrees with the 1st , 2nd , or 3rd nn cu site to zn [ the case ( 5 ) ] .
however , this is inconsistent with the up - to 3rd nn wipeout effect on the total intensity .
thus , the satellite resonance of y124 may arise from the fraction of the 4th nn cu site to zn . in the above analysis
, one should note that the so - called all - or - nothing " model is implicitly assumed , where some nuclei are fully observable but the others are fully unobservable @xcite .
this enables us to estimate the number of observed nuclei from the observed nqr signal intensity . in the actual sample ,
however , there must be @xmath47 distribution .
figure 5 illustrates a two - pulse sequence with a cu nqr spin - echo in a rotating frame with a cu nqr frequency in the all - or - nothing " model ( a ) and in an actual system with @xmath47 distribution ( b ) .
in the case of fig .
5(b ) , some of the observable nuclear spin - echoes after the two pulses will be underestimated due to a somewhat shorter @xmath47 than the totally corrected @xmath47 decay curve exp(-2@xmath71 ) , even when the gaussian - times - exponential function empirically well describes @xmath72 .
conversely , the @xmath73 fluctuation effect due to a short @xmath31 @xcite or the motional narrowing effect @xcite on the transverse nuclear spin relaxation causes an exponential decay in a longer delay time @xmath19 but in a gaussian decay in a shorter time @xmath19 , then the total correction using a simple exponential function within the observable @xmath19 range will lead to overestimation of the extrapolated @xmath48 . for both the main and the satellite signals , such a @xmath47 distribution or dynamical effect can cause overestimation or underestimation of the intensity .
thus , one should be careful with the results from intensity analysis .
one can safely mention that the main signal does not come from the cu sites near zn but that the satellite one from the zn - neighbor cu sites .
= 3.7 in here , we briefly mention the cu nqr frequency .
the cu nqr frequencies of stoichiometic cuprate oxides have been analyzed with several models @xcite .
for non - stoichiometic or impure compounds , although there is an attempt with numerical simulation @xcite , it is more complicated , because local crystal strain @xcite or electric charge oscillation @xcite can affect the line profile of cu nqr spectrum .
thus , we do not employ a specific model to analyze how the satellite cu nqr frequency is shifted to the lower side than the main one .
we just emphasize that cu@xmath74 is a jahn - teller ion but zn@xmath74 is a non - jahn - teller ion , so that the local crystalline distortion @xcite must modify the electric field gradient tensor at zn - neighbor cu sites , differently from that at cu sites away from zn or from that in the pure y123 @xcite or y124 @xcite .
= 3.5 in figure 6 shows the main @xmath28cu(2 ) nuclear spin - lattice relaxation curve for y123 with zn concentrations @xmath17=0.017 ( a ) and 0.033 ( b ) at @xmath8=100 k. the result for y124 has been reported in ref .
@xcite . for y123 with a small zn concentration of @xmath17=0.017 , the recovery curve @xmath30 relaxes more quickly than that for pure y123 , which is consistent with the result for y124 with zn @xmath700.010 . with further doping of @xmath17=0.033 , @xmath75 1.0 ms )
relaxes more slowly than that for pure y123 , which is similar to the result for y124 with heavy doping of zn @xmath17=0.022 at @xmath8=200 k @xcite .
= 3.5 in figure 7 shows temperature dependence of the main and the satellite @xmath28cu nuclear spin - lattice relaxation curves @xmath30 for zn - substituted y123 with @xmath17=0.017 ( upper ) and y124 with @xmath17=0.01 ( lower ) . for y123 ,
the satellite signal recovers more quickly than the main one below @xmath15 , but they show nearly the same recovery curves above @xmath15 . for y124
, however , both above and below @xmath15 , the satellite signal recovers more quickly than the main one . with
further zn doping for y124 ( @xmath17=0.022 ) , the difference between the main and the satellite signals tends to be reduced ( not presented here ) , being consistent with the non - superconducting y124 @xcite .
the result in fig .
7 is evidence for different magnetic correlation between the main and the satellite signals in the superconductors with small concentrations of zn . in figs . 1 , 2 , 6 , and 7 , all the recovery curves are nonexponential functions .
if the satellite signal comes from the same @xmath57th nn cu sites by zn and if the position dependence of zn - induced relaxation process , i.e. the spatial distribution of zn - induced moments or spin density , is in axial symmetry along the @xmath44 axis , then the recovery curve must be a single exponential function .
however , the actual curves are nonexponential ones .
thus , the satellite signals consist of non - single cu sites , or the zn - induced moment or spin density shows non - axial symmetry , e.g. @xmath76-wave symmetry .
nonexponential relaxation is frequently observed in the impure materials .
we have applied a conventional impurity - induced nmr relaxation theory @xcite to the impure high-@xmath15 cuprate superconductors and quantum spin systems @xcite , which has been developed for dilute heisenberg spin systems and dilute magnetic alloys .
we believe that the analysis yields a vital , physical framework on the strongly correlated impure systems .
the experimental @xmath30 is analyzed by the exponential function times a stretched exponential function @xmath77 $ ] .
@xmath78 , @xmath79 and @xmath80 are the fitting parameters .
the numerical factor 3 is introduced to conform to the conventional expression of @xmath31 @xcite .
the solid and the dashed curves in figs .
1 , 2 , 6 , and 7 are the least - squares fitting results using this approximate function .
@xmath79 is the cu nuclear spin - lattice relaxation time due to the host cu electron spin fluctuation via a hyperfine coupling .
@xmath80 is the impurity - induced nuclear spin - lattice relaxation time , i.e. , a spatial average of 1/@xmath81 ( @xmath82 is a distance between a nuclear and an impurity - induced moment , @xmath83 is a coupling constant between the nuclear spin and the impurity - induced moment , @xmath84 is the impurity - induced spin - spin correlation function @xcite , and @xmath85 is the nuclear resonance frequency ) @xcite . for an isolated local moment on the impurity site ,
one obtains 1/@xmath86 ( a longitudinal direct dipole coupling @xmath87 , or a two - dimensional rkky interaction @xmath88 ) @xcite . for an impurity - induced moment with a local coupling @xmath89 , if @xmath90 or e@xmath91 , one obtains @xmath92 exp@xmath93 $ ] or exp(-ln@xmath94 ) ( @xmath95 is a correlation length , and @xmath96 is a constant ) @xcite . instead of including all the direct or indirect nuclear - electron interactions , we assume a _ minimal _ model with a single impurity - induced relaxation and the host cu electron spin relaxation .
the recovery curve is expressed by @xmath97\prod_i[(1-c)+c\exp(-3t / t_1({\bf r}_i))]$ ] ( @xmath98 is the cu site index ) as a function of a lot of time constants of @xmath79 , @xmath99 , @xmath100 , @xmath101 , @xmath102 . for @xmath1031 ,
the product @xmath104 $ ] is approximated by @xmath105 $ ] , and then the spatial integral leads to the stretched exponential function with the single time constant @xmath80 @xcite .
eventually , only two fitting parameters on the time constants , i.e. @xmath79 and @xmath80 are obtained .
the distribution of @xmath31 is taken into consideration through @xmath106 , although relations among @xmath106 , the @xmath47 distribution , and the broad cu nqr frequency spectrum are not straightforward . in the actual fitting
, we noticed that it is not easy to estimate the precise value of 1/@xmath79 in the case of @xmath107 .
however , zn - substitution effect on each relaxation time is fairly obvious . the slow relaxation of @xmath75 1.0 ms ) due to heavy doping of zn as in fig .
6(b ) is the experimental fact . within the present model , since @xmath108 $ ] at @xmath109 ,
then one estimates to decrease 1/@xmath79 with zn doping of @xmath17=0.033 .
this estimation is not due to delicate fitting .
= 3.5 in figure 8 shows the estimated @xmath112 and @xmath113 of the satellite and of the main signals as functions of temperature for zn - substituted y123 ( a ) and y124 ( b ) .
first , the satellite @xmath112 for y124 at @xmath8=4.2 k is an order of magnitude higher than that for y123 .
second , the temperature dependence of @xmath112 of the satellite signal is different between y123 and y124 . for y124 at lower temperatures below @xmath15 , @xmath112 increases as temperature is decreased , in contrast to @xmath8 independent or slight decrease of @xmath112 for y123 .
third , the difference in the @xmath8 dependence of @xmath112 between the main and the satellite signals is marked below @xmath15 for y123 but above @xmath15 for y124 .
fourth , the main @xmath113 above about 100 k decreases with zn doping for both systems .
fifth , the satellite @xmath113 was estimated to be @xmath114 1.0 s@xmath115k@xmath115 for y124 with zn . from these results , one can see that the local magnetic correlation around zn is different between underdoped and optimally doped samples .
the cu(2 ) nuclear spin - lattice relaxation time probes the in - plane antiferromagnetic dynamical spin susceptibility at an nmr / nqr frequency @xcite and homogeneity of the cuo@xmath0 plane in real space @xcite .
relation between our results on the cu nqr relaxation and the previous @xmath13y nmr results @xcite is not straightforward .
@xmath116 : the obvious difference in @xmath112 between the main and the satellite signals indicates an evidence for the zn - induced virtual bound state @xcite through the locally enhanced magnetic correlation @xcite . for underdoped y124 , the difference in @xmath112 may indicate that zn - induced localized moments " persist both above and below @xmath15 @xcite .
for optimally doped y123 , the below-@xmath15 difference in @xmath112 may indicate that the zn - induced , generalized rkky oscillation is extended around zn below @xmath15 but localized above @xmath15 .
this is consistent with a prediction in the perturbative mode - mode coupling theory @xcite , where the spatially extended potential scattering by zn plays a key role @xcite .
the separated frequency of the satellite cu nqr spectrum may also support such a finite range effect of zn impurity potential . according to ref .
@xcite , the 1st nn potential term @xmath117 will be larger in y124 than y123 , because the satellite @xmath112 is larger than the main one in y124 above and below @xmath15 .
the local electronic state induced by zn reflects the matrix antiferromagnetic correlation @xcite .
the optimally doped y123 still has the underlying antiferromagnetic correlation @xcite .
@xmath118 : for y124 , the absence or a small value of the satellite @xmath113 suggests that the conduction electrons in the underdoped system are expelled from the vicinity of zn atoms , or that the pseudo spin - gap effect is enhanced near zn .
a remnant of the pseudo spin - gap effect can be seen as a decrease of @xmath112 below @xmath8=100 k with @xmath17=0.005 and 0.01 in fig .
8(b ) . for y123
, nearly the same @xmath113 between the satellite and the main signals indicates that the conduction electrons uniformly distribute near and away from zn both above and below @xmath15 , in contrast to the underdoped y124 .
the pure y124 is a stoichiometric , homogeneous , underdoped system .
nevertheless , it has some inherent electronic instability , e.g. magnetic instability , which may lead to an inhomogeneous magnetic response .
the underdoped electronic state of y124 is located more close to the magnetic instability than the optimally doped one of y123 .
the above-@xmath15 decrease in @xmath113 with zn doping for both systems indicates that zn suppresses the host antiferromagnetic spin susceptibility @xcite .
this is recently reproduced by a numerical quantum monte carlo simulation with the dynamical cluster approximation based on the two - dimensional @xmath7 model @xcite .
we succeeded in measuring the zn - induced satellite cu(2 ) nqr nuclear spin - lattice relaxation curves in the superconductors y123 and y124 above and below @xmath15 in a zero external magnetic field . with doping a small amount of zn , the electron spin correlation near zn is enhanced in the optimally doped system as well as in the underdoped one .
the optimally doped y123 still has an underlying antiferromagnetic correlation .
the local electronic state induced by zn reflects it .
we would like to thank m. matsumura , h. yamagata ( yba@xmath0cu@xmath3o@xmath119 ) and y. ueda ( yba@xmath0cu@xmath3o@xmath120 ) for providing their samples and for helpful discussions , and y. ohashi , m. ogata , m. h. julien , and k. ishida for stimulating discussions .
this work was supported by the new energy and industrial technology development organization ( nedo ) as collaborative research and development of fundamental technologies for superconductivity applications .
h. yasuoka , t. shimizu , t. imai , s. sasaki , y. ueda , and k. kosuge , hyperfine interaction2 * 49 * , 167 ( 1989 ) .
h. yasuoka , s. sasaki , t. imai , t. shimizu , y. ueda , and k. kosuge , phase transitions , * 15 * , 183 ( 1989 ) .
t. imai , h. yasuoka , t. shimizu , y. ueda , k. yoshimura , and k. kosuge , physica c * 162 - 164 * , 169 ( 1989 ) .
t. imai , t. shimizu , h. yasuoka , y. ueda , k. yoshimura , and k. kosuge , _ technical report of issp _ , ser .
a , no . 2145 , june 1989 . s. fujiyama , y. itoh , h. yasuoka , and y. ueda , j. phys .
. jpn . * 66 * , 2864 ( 1997 ) .
f. j. adrian , phys .
b * 38 * , 2426 ( 1988 ) .
t. shimuzu , j. phys .
. jpn . * 62 * , 772 ( 1993 ) , _ ibid _ * 62 * , 779 ( 1993 ) .
k. hanzawa , j. phys .
. jpn . * 62 * , 3302 ( 1993 ) .
y. ohta , w. koshibae , and s. maekawa , j. phys .
61 * , 2198 ( 1992 ) .
j. yu , a. j. freeman , r. podloucky , p. herzig , and p. weinberger , phys . rev .
b * 43 * , 532 ( 1991 ) .
k. schwarz , c. ambrosch - draxl , and p. blaha , phys .
b * 42 * , 2051 ( 1990 ) .
j. haase , r. stern , d. g. hinks , and c. p. slichter , in _ stripes and related phenomena _ , edited by a. bianconi and n. l. saini , ( kluwer academic / plenum publishers , new york , 2000 ) , p. 303 .
j. haase , c. p. slichter , r. stern , c. t. milling , and d. g. hinks , physica c * 341 - 348 * , 1727 ( 2000 ) . | we studied local electronic states near zn in optimally doped yba@xmath0(cu@xmath1zn@xmath2)@xmath3o@xmath4 and underdoped yba@xmath0(cu@xmath1zn@xmath2)@xmath5o@xmath6 via satellite signals of plane - site cu(2 ) nuclear quadrupole resonance ( nqr ) spectra . from the relative intensity of cu nqr spectra ,
the satellite signals are assigned to zn - neighbor cu nqr lines .
the cu nuclear spin - lattice relaxation time of the satellite signal is shorter than that of the main signal , which indicates that the magnetic correlation is locally enhanced near zn both for the underdoped and the optimally doped systems .
the pure yba@xmath0cu@xmath5o@xmath6 is a stoichiometric , homogenous , underdoped electronic system ; nevertheless , the zn - induced inhomogeneous magnetic response in the cuo@xmath0 plane is more marked than that of the optimally doped yba@xmath0cu@xmath3o@xmath4 . epsf.sty |
separated by a hostile region of length @xmath11 .
the focus of observation is the interdependence of the critical values of @xmath12 and @xmath13 at which extinction occurs.,scaledwidth=40.0% ] need for the new experimental proposal that we present here arises from the fact that the stationary single mask experiment @xcite uses for its interpretation a number of inputs which suffer from a certain degree of uncertainty .
some of these uncertainties are about the values of @xmath6 and @xmath4 @xcite , and others about the extent to which other phenomena such as signaling @xcite that occur in bacterial movement might affect the outcome of the experiment . because absolute values of critical lengths may be difficult to obtain with an acceptable degree of accuracy ,
we propose here stationary mask observations , that focus , instead , on the _ interdependence of two critical lengths_. the essential feature of this present proposal is to have more than one mask in the petri dish , so that there are two controllable lengths , a favorable length associated with a livable region for the bacteria , and a hostile length associated with an unlivable region .
the double control possible in the proposed experiment opens the prediction space from a single value to an infinity of values .
the variation of the critical value of one of the lengths when the other is changed presents an entire _ relationship _ that is directly observable rather than a single value and therefore could lead to a much cleaner and trustworthy interpretation of the observations . the simplest system to study consists of two masks , each of length @xmath14 , separated by a distance of length @xmath11 as shown in fig . [ fig1 ] .
the previously investigated case with a single mask @xcite corresponds , obviously , to infinite @xmath13 . to test the idea behind the new proposal
, we carried out numerical studies of the full nonlinear fisher equation ( [ fisher ] ) with given values of @xmath6 , @xmath4 , @xmath9 , and @xmath5 ( 10 , 0.1 , -0.9 and 1 in appropriate units . )
all lengths were expressed as ratios to the diffusion length @xmath15 .
the numerical studies used an explicit finite differences scheme , @xmath1 being discretized in intervals of 0.1 .
the convergence to a steady state was analyzed by measuring the distance between successive solutions .
the result we found ( see fig . [ fig2 ] ) is that the critical value of the mask width @xmath12 , i.e. , the smallest value that can support a non - zero bacterial population in the steady state , is lowered for a finite intermask distance @xmath13 .
we considered 10 different values of the hostile length ( the intermask distance @xmath13 ) and varied the favorable length ( the mask width @xmath12 . ) for each @xmath12-@xmath13 pair , we started with arbitrary initial conditions and let the program run until no time dependence was discernible . and @xmath13 respectively .
the former is the width of the mask and the @xmath11 is the distance between the two masks ( see fig .
1 ) , both lengths being expressed in this plot in units of the diffusion length @xmath16 where @xmath4 is the growth rate under the masks . outside the masks @xmath4 is replaced by @xmath17 to represent the harsh effect of the ultraviolet light , the magnitude of @xmath9 being taken for the purposes of this plot to be 9 times that of @xmath4 .
we solved eq .
[ fisher ] numerically for various different values of the nonlinear parameter @xmath5 and found that the results did not depend on those values provided @xmath5 was non - zero and positive .
the shaded region represents pairs of @xmath12 and @xmath13 values that lead to bacterial extinction in the steady state .
filled circles mark the onset of extinction and are obtained from the numerical solution . the solid line constructed simply to smoothly join the circles ,
is found to coincide precisely with the prediction of our analytic theory.,scaledwidth=40.0% ] we repeated the procedure for each of several sufficiently low values of @xmath12 and increased @xmath12 until the extinction disappeared .
we also reversed the procedure starting with high values of @xmath12 and decreased them systematically until extinction appeared .
numerous runs allowed us to obtain corresponding pairs of @xmath12 and @xmath13 that mark the transition region .
the results are denoted by filled circles in fig .
the shaded area represents the extinction region and the unshaded area the parameter region in which bacterial population densities are non - zero in the steady state .
the curve passing through the numerically found transition points may be considered , at this stage of our discussion , to be simply a smooth joining trace .
we will see below that its exact shape can be accessed through our analytic theory .
it is easy to understand , on the basis of a qualitative argument , the shape and tendency of the results of the numerical solutions of the nonlinear equation .
bacteria diffuse from within the mask to the harsh region and die if they reach that region .
small values of the mask width or large values of the intermask distance result in extinction .
the extinction effect is worsened by an increase in the intermask distance for small values of that distance but the effect saturates for larger values .
hence the saturation in the curve .
all these features can be tested experimentally in our proposed set - up .
quantitative comparison with the predictions of the fisher equation are possible because we have developed an exact analytic theory of the interplay of the favorable and hostile distances as reflected in the transition .
what makes the proposed comparison with experiment significant is that , although the fisher equation , whose _ numerical _ solution has led us to set out the separation curve between the extinction region and the rest in fig .
[ fig1 ] , can not be solved exactly by analytic means , the separation curve itself can be obtained analytically .
indeed , we show below that the curve is given precisely by @xmath18\right ] .
\label{analytic}\ ] ] this prediction , which is one of the central analytic results of the present paper , coincides with the solid curve in fig .
following the ideas of ludwig et al . @xcite but applying them to the many - mask system , we consider the steady state of the fisher equation ( [ fisher ] ) and argue that if there is a transition , the quadratic term in the steady state @xmath19 can be neglected at the extinction point in favor of the linear terms since @xmath19 itself vanishes at the transition .
we are thus led to seek the solutions of @xmath20 under the two masks , and of @xmath21 outside the masks , with @xmath22 and @xmath23 .
we take the masks to lie from @xmath24 to @xmath25 . using @xmath26 and @xmath27 as constants , the most general functions for @xmath19 inside and outside the mask , respectively ,
are , as a result of the symmetry , ( we consider only the right side of the origin , since all considerations repeat unchanged on the left side by symmetry ) @xmath28 inside the mask , @xmath29 , @xmath30 in the extreme outside , in the harsh region , @xmath31 , and @xmath32 in the central region between the two masks , @xmath33 . matching the logarithmic derivative of the solution at the outer and the inner boundaries leads to @xmath34&=&\alpha_1/\alpha \nonumber \\
\tan(\alpha d-\phi)&=&-(\alpha_1/\alpha)\tanh { \alpha_1 d } , \nonumber\end{aligned}\ ] ] and elimination of @xmath35 from these two equations leads to eq .
( [ analytic ] ) quoted above .
notice how , in light of the behavior of the hyperbolic tangent , eq .
( [ analytic ] ) , one of our central results , reduces to ludwig et al.s single mask value @xcite of @xmath12 , when @xmath13 attains infinite values , and half that value when @xmath13 vanishes .
both are fully expected and natural results .
it is also instructive to rewrite the transition relation as @xmath36 where we express the favorable distance @xmath12 and hostile distance @xmath13 normalized to the growth diffusion length @xmath37 as @xmath38 and @xmath39 respectively , and the depletion parameter @xmath40 is the square root of the ratio of the destruction rate @xmath9 and the growth rate @xmath4 . and the hostile length @xmath13 respectively from eq .
( [ analytic ] ) . as in other plots ,
the lengths are expressed as dimensionless ratios to the diffusion length @xmath41 .
different curves correspond , as shown in the legend , to different values of @xmath40 , i.e. , to different intensities of the ultraviolet radiation .
note the doubling of @xmath12 as @xmath13 passes from @xmath42 to @xmath43 in all curves .
, scaledwidth=40.0% ] we display in fig .
[ fig3 ] our analytic result , eq .
( [ analytic ] ) , equivalently eq .
( [ twodimless ] ) .
as in other plots the lengths are expressed as dimensionless ratios to the diffusion length @xmath44 .
different curves correspond , as shown in the legend , to different values of @xmath40 , i.e. , to different intensities of the ultraviolet radiation .
the curve for @xmath45 from our theory is shown as the solid line in fig .
[ fig2 ] , and displays exact coincidence with the numerical findings . to be noted is the fact that the numerical solutions are of the full nonlinear equation and
have ben found for a specific @xmath5 whereas the analytic theory is linear and does not require the use of a value of @xmath5 , requiring for its application only the condition @xmath46 .
it is clear from our eq .
( [ analytic ] ) that the variation of the destructive rate @xmath9 may also be employed for useful experimental exploration , at least in principle . while this could have also been done in the one - mask scenario of ref .
@xcite , we explain the idea here in the two - mask case .
additional confirmation of the theoretical picture of the dynamics of the bacteria might be provided by observing how the critical @xmath12-@xmath13 values change with the _ intensity _ of the ultraviolet light .
even if we do not know the precise dependence of @xmath9 on the intensity , we can be fairly certain that it increases with the latter , and undergoes a saturation at high values of the intensity .
a variation of the critical value of @xmath12 with the intensity would show therefore qualitative behavior similar to the variation with @xmath9 , equivalently with the ratio @xmath47 .
the latter variation is displayed in fig .
[ fig4 ] . with one exception ,
all @xmath13 values produce the same saturation value of @xmath12 for large @xmath40 .
the exception is @xmath48 : its saturation value is one half that of the others .
mathematically , this corresponds to the hyperbolic tangent becoming @xmath49 for all non - zero values of @xmath13 for large enough @xmath50 , but vanishing if @xmath13 vanishes .
physically , this means that if there is an adverse region between the masks , bacteria will be killed on arriving there , in light of the infinitely harsh radiation , reducing the problem to a single - mask scenario with the given @xmath12 as the mask width . on the other hand , if the intermediate harsh region does not exist at all ( because @xmath48 ) , one is reduced to considering a _ single _ mask with width twice that of the given mask . ) for the dependence of the critical value of @xmath12 on the intensity of ultraviolet radiation , represented here by the value of @xmath40 , for given values of the hostile length ( half the intermask separation ) @xmath51 as shown .
all @xmath13 values produce the same saturation values of @xmath12 for infinitely harsh ultraviolet intensity except for @xmath48 : all curves reach the same higher value for large @xmath40 except the @xmath48 curve whose saturation value is one half that of the others .
see text for explanation .
the lengths are expressed as dimensionless ratios to the diffusion length @xmath15.,scaledwidth=40.0% ] our suggestion , thus , is to use both the relative variation of the favorable and hostile lengths , @xmath12 and @xmath13 , and of the intensity of ultraviolet light in the manner discussed , to check our simple quantitative predictions .
it is possible and useful to construct set - ups with _ multiple _
masks for further experimental verification . the theory for this situation along the ludwig et al .
arguments we have developed here is slightly more tedious to write down .
for ease in notation , let us adopt as we have done in eq .
( [ twodimless ] ) , the symbols @xmath52 and @xmath53 to express the favorable and hostile lengths in units of @xmath15 .
then the linear petri dish formula for an even number of masks is obtained by executing the following pseudocode .
let @xmath54 be an operator defined by the successive operations of ( i ) taking the hyperbolic tangent of what it acts on , ( ii ) multiplying the result by @xmath40 , ( iii ) taking the arctangent of the result , and ( iv ) subtracting the result from @xmath52 .
define the operator @xmath55 similarly through the successive operations of ( i ) taking the trigonometric tangent of what it acts on , ( ii ) dividing the result by @xmath50 , ( iii ) taking the hyperbolic arctangent of the result , and ( iv ) subtracting the result from @xmath56.to obtain the required formula for an even number of masks , start with the product @xmath57 .
apply @xmath54 and then @xmath55 alternately and successively so that there are as many @xmath54 s in the operation as the number of mask pairs in the system , but one more @xmath54 than @xmath55 .
( thus , do @xmath54 for a pair of masks , @xmath58 for 4 masks , @xmath59 for 6 masks , and so on . ) finally , equate the result to the arctangent of @xmath40 to get an implicit formula showing the @xmath52-@xmath53 relation .
the pseudocode can be expressed succinctly in terms of operators @xmath55 and @xmath54 by stating that the general expression for @xmath60 masks can be formally written as @xmath61 \left(\xi\delta\right)=\arctan \xi$ ] . as can be verified ,
this reproduces the 2-mask formula , eq .
( [ analytic ] ) .
the 4-mask formula is @xmath62
our suggested set - up and calculations have taken the petri dish to have infinite extension away from the two masks , an assumption that should be reasonable in light of the destructive effects of the radiation .
we now analyze another practical extension of the theory .
consider the two - mask situation in a _
circular _ petri dish , i.e. one in which the boundary conditions are periodic , the hostile distance between the edges of the masks being @xmath11 on one side as in the earlier analysis but now also @xmath63 on the other side , see fig .
let the width of the dish be small enough so that bacterial diffusion can be considered to be still 1-dimensional but over a total distance of extent @xmath64 .
we are not interested here in 2-dimensional considerations appropriate to wide dishes , an example of which may be found in ref .
@xcite for experiments with _ moving _ masks or rotating dishes . as well as @xmath12 and @xmath13.,scaledwidth=40.0% ]
we now have , as in the analysis given above , @xmath65 inside the right mask , i.e. , for @xmath29 , but to the right of the right mask , i.e. , for @xmath31 , @xmath19 is not proportional to @xmath66 as in the earlier analysis , but @xmath67 this expression , which , in the right region under consideration , where @xmath68 , can be written as @xmath69 , will hold on the left extreme as well but there it simplifies to @xmath70 .
the additional phase @xmath71 is immediately determined from the periodicity in the boundary conditions , i.e. , by matching the logarithmic derivative of the solution at @xmath72 : @xmath73 matching also at @xmath24 and @xmath74 , one arrives at the new formula for critical @xmath12 and @xmath13 , valid for a _
petri dish : @xmath75 we see that the circular dish result , eq .
( [ circular ] ) , is symmetric in the two hostile lengths @xmath76 and @xmath13 as it obviously must be , that it reduces to the result for the linear dish of infinite extent , viz . , eq .
( [ analytic ] ) , when @xmath76 becomes infinite since then the new hyperbolic tangent in eq .
( [ circular ] ) reduces to @xmath49 , and that , when @xmath76 vanishes ( equivalently when @xmath13 vanishes ) , it yields a new expression for the critical width for the single - mask case but with twice the width value .
if there were simply a single mask in a circular petri dish of extent @xmath76 and no @xmath13 , the generalization of the ludwig formula that would come out of our analysis would be @xmath77 most importantly , we see that it should be possible in principle to use the new analytic result(s ) for experimental probing of the phenomenon under consideration since one could construct and employ circular petri dishes of desirable radius .
needless to say , the added effects discussed here would not be discernible for @xmath76 values that are very large as then bacteria would die in the external regions quickly enough .
we have also derived the corresponding expressions combining circular dish geometry with the multiple mask set - ups .
specifically , the 4-mask formula takes the form , @xmath78 being @xmath76 measured in units of @xmath15 , @xmath79 whose reduction to the linear dish formula is obvious as @xmath76 becomes infinite .
we have verified all these expressions that we have derived , by comparing their predictions to numerical solutions of the fisher equation with explicit ( arbitrary ) non - zero @xmath5 s .
the proposal for experimental observations of abrupt population transitions in patches that we have presented above should be of interest for multiple reasons .
patches or spatial inhomogeneities have been studied and emphasized early on in ecology by many authors such as levin @xcite , murray @xcite and shigesada @xcite in varied contexts including traveling waves , and there have been recent contributions as well on the dirichlet conditions problem for general nonlinearity incorporating allee effects @xcite . our analysis here provides a clean proposal for experiments which derives its potential for clear interpretation from the use of two ( or more ) controllable distances of travel , one favorable and the other unfavorable .
one should be able to use it for extraction of parameter combinations in the fisher equation which has been ubiquitous in mathematical ecology .
the idea behind the present proposal should therefore find use in systems other than bacterial aggregates .
indeed , we have found similar abrupt transitions in _ infected _ rodent populations in our study of the hantavirus epidemic .
the theory for that system is , however , more complex , and can not be developed without approximation .
by contrast , the theory we have presented here is , perhaps surprisingly , exact , i.e. , analytical .
this is so in spite of the fact that analytic solutions of the fisher equation for these situations do not seem to be possible except as elliptic quadratures .
the reason for this happy state of affairs can be understood from the original arguments of ludwig et al . given in their paper on spatial patterning of the budworm @xcite , or from the lucid explanations given ,
e.g. , in a recent text on mathematical ecology @xcite . at the transition ,
the densities vanish and therefore terms of order higher than the first may be safely neglected .
our theoretical contribution is only in generalizing that analysis to multiple masks and nonlinear geometries as shown .
the observationable controllable features our proposal emphasizes , and directly utilizes , are ( i ) the interplay of the favorable and adverse lengths for bacterial traversal , ( ii ) the variation of the intensity of ultraviolet radiation which changes the interrelationship of the critical values , ( iii ) the multiple - mask set - up which reintroduces bacteria into favorable regions after they have passed into arid areas , and ( iv ) the optional use of circular geometry for the petri dish which provides one more controllable length .
we are currently involved in investigating fluctuation effects that appear from detailed monte carlo considerations and become manifest for small population densities , in constructing the theory for transitions in partially infected populations of rodents in open terrains , in analyzing the effects of static disorder in the placement and size of the masks , and in studying the effects of dynamic ( time ) variation of mask size and placement as in ref @xcite .
the bacterial experiments proposed above are , however , ready to go and we hope that observations will be made soon along the lines we have discussed . this work was supported in part by the nsf under grant no .
int-0336343 , by nsf / nih ecology of infectious diseases under grant no .
one of us ( vmk ) thanks gandhi m. viswanathan for an insightful remark that helped the completion of this work .
h. e. stanley _ introduction to phase transitions and critical phenomena _ ( oxford university press , 1971 ) .
s. h. strogatz , _ nonlinear dynamics and chaos : with applications to physics , biology , chemistry and engineering _ ( westview press , 1994 ) .
e. renshaw , _ modelling biological populations in space and time _ ( cambridge university press , 1991 ) .
a. l. lin , b. mann , g. torres , b. lincoln , j. kas , and h. l. swinney , biophys . j. * 87 * , 75 ( 2004 ) ; see also a. l. lin in _ modern challenges in statistical mechanics : patterns , growth , and the interplay of nonlinearity and complexity , _ eds . v. m. kenkre and k. lindenberg , aip conference proceedings volume 658 ( isbn 0 - 7354 - 0118 - 7 , 2003 ) .
d. r. nelson and n. m. shnerb , phys .
e * 58 * , 1383 ( 1998 ) ; k. a. dahmen , d. r. nelson and n. m. shnerb , j. math . biol .
* 41 * , 1 ( 2000 ) . v.
m. kenkre and m. n. kuperman , phys .
e. * 67 * , 051921 ( 2003 ) .
n. perry , j. r. soc .
interface * 2 * , 379 ( 2005 ) .
j. wakita , k. komatsu , a. nakahara , t. matsuyama and m. matsushita , j. phys .
jpn . * 63 * , 1205 ( 1994 ) .
b. mann , phd thesis ( univ of texas , austin , 2001 ) .
m. ballard , v. m. kenkre and m. n. kuperman , phys .
e * 70 * , 031912 ( 2004 ) .
h. c. berg , phys . today * 53 * , 24 ( 2000 ) ; e. ben - jacob , i. cohen and h. levine , adv . phys .
* 49 * , 395 ( 2000 ) .
r. a. fisher , ann .
london * 7 * , 355 ( 1937 )
. j. g. skellam , biometrika * 38 * , 196 ( 1951 ) .
h. kierstead and l. b. slobodkin , j. mar . res .
* 12 * , 141 ( 1953 ) .
m. kot , _ elements of mathematical ecology _ ( cambridge university press , cambridge , 2003 ) .
d. ludwig , d. g. aronson and h. f. weinberger , j. math .
* 8 * , 217 ( 1979 ) .
n. m. shnerb , phys .
e * 63 * , 011906 ( 2000 ) .
s. a. levin , ann .
* 7 * , 287 ( 1976 ) .
j. d. murray , _ mathematical biology_(springer - verlag , berlin , 1993 ) .
n. kinezaki , k. kawasaki , f. takasu and n. shigesada , theor .
biol . * 64 * , 291 ( 2003 ) ; n. shigesada , k. kawasaki and e. teramoto , theor .
. biol . * 30 * , 143 ( 1986 ) ; n. shigesada and k. kawasaki , _ biological invasions : theory and practice _ ( oxford university press , oxford , 1997 ) .
v. mendez and d. campos , phys .
e * 77 * , 022901 ( 2008 ) . | an explicit proposal for experiments leading to abrupt transitions in spatially extended bacterial populations in a petri dish is presented on the basis of an exact formula obtained through an analytic theory .
the theory provides accurately the transition expressions in spite of the fact that the actual solutions , which involve strong nonlinearity , are inaccessible to it .
the analytic expressions are verified through numerical solutions of the relevant nonlinear equation .
the experimental set - up suggested uses opaque masks in a petri dish bathed in ultraviolet radiation as in lin et al .
, biophys .
j. * 87 * , 75 ( 2004 ) and perry , j. r. soc .
interface * 2 * , 379 ( 2005 ) but is based on the interplay of two distances the bacteria must traverse , one of them favorable and the other adverse . as a result of this interplay feature ,
the experiments proposed introduce highly enhanced reliability in interpretation of observations and in the potential for extraction of system parameters .
phenomena displaying abrupt transitions are of special interest to a variety of sciences , including physics and biology . in some physical cases
they arise from cooperative interactions among a large number of constituents ( e.g. , molecules or spins ) , and in others from nonlinearities in interaction inherent in the system , ( e.g. , in some mechanical systems with only a few degrees of freedom . ) in the first form they are known as phase transitions @xcite , in the second as bifurcations @xcite .
abrupt phenomena also command attention in the context of extinction of populations , a subject of obvious interest to biology @xcite .
transitions in populations , therefore , constitute an exciting topic of interdisciplinary science combining physics and biology , and the present paper reports a theory and proposes an experiment in this topic .
the special feature of the proposed experiment is that it may be performed with relatively simple equipment and measurement techniques .
the system to be considered consists of bacteria in a petri dish , allowed to grow and move , spatial selection being imposed via lethal ultraviolet radiation that is incident on the dish but punctuated by opaque masks that protect the bacteria in chosen regions .
experiments under such a setup were initiated several years ago by lin et al .
@xcite who used moving masks in response to a theoretical analysis @xcite that focused on spatial disorder . a quite different experiment using stationary masks of varying sizes , and employing bacterial extinction as the key phenomenon
was then proposed @xcite .
that proposal was carried out experimentally @xcite and the predictions of ref .
@xcite were verified .
related investigations on this topic may be found in @xcite , the general problem of bacterial dynamics in petri dishes having been addressed in various papers and contexts @xcite .
the underlying assumption behind most of these studies is that the bacteria obey a simple fisher equation @xcite for the time evolution of their dynamics , their population density @xmath0 , where @xmath1 is the position in a 1-dimensional space ( the linear dimension of the petri dish ) , and @xmath2 is the time , being governed by @xmath3 here @xmath4 and @xmath5 are , respectively , the growth rate and a competition parameter arising from the resources being limited , and @xmath6 is the bacterial diffusion constant . equation ( [ fisher ] ) generally does not permit analytic solutions for arbitrary times but can be solved in the steady state explicitly in terms of jacobian elliptic functions , if the bacterial population is assumed to vanish at the edges of a finite region .
physically , this could correspond to an idealization in which , while the mask protects the bacteria from the lethal effects of the ultraviolet radiation , the latter is so potent that bacterial population must vanish everywhere outside the mask where the radiation impinges on the bacteria .
mathematically , these are dirichlet boundary conditions .
the elliptic function solution , given independently in a recent analysis @xcite , but known decades earlier through the work of skellam @xcite , and elucidated in textbooks @xcite , leads to the well - known kiss transition @xcite : the steady state bacterial population vanishes for any width of the mask lower than a critical value that , interestingly , depends on the diffusion constant @xmath6 and the growth rate @xmath4 , but not on the nonlinearity parameter @xmath5 .
the specific expression for the critical mask width is @xmath7 . combining this expression with information about bacterial diffusion constants and growth rates discussed by mann in his ph.d .
thesis @xcite , kenkre and kuperman calculated the critical mask width to be around @xmath8 , and suggested that an experiment be carried out to observe the transition .
perry @xcite followed the suggestion , and reported observing a transition width of 0.8 cm in his experiments on a non - chemotactic strain rp9535 of _ e. coli _ bacteria . since the dirichlet boundary condition is only an extreme idealization , perry argued , appropriately , that it is preferable to invoke the analysis of ludwig et al .
@xcite for quantitative verification .
that analysis assumes that the growth rate @xmath4 is negative in the region outside the mask ( and of magnitude @xmath9 ) rather than infinite as would correspond to dirichlet boundary conditions , and arrives at the critical mask width as being @xmath10 .
the dirichlet expression is recovered from this ludwig formula if the effect of the ultraviolet radiation outside the mask is `` infinitely '' lethal . |
data were obtained during the nights of 1012 december 2004 with the sparsepak ifu ( see @xcite ) .
we used the 860 l / mm grating at order 2 , which provides a spectral resolution @xmath2 4,900 , and the 316 l / mm grating at order 8 , which provides a spectral resolution @xmath3 10,000 . in the latter mode ,
the wavelength coverage includes the h@xmath4 , [ n@xmath5ii ] @xmath66548,6583 and [ s@xmath5ii ] @xmath66716,6731 emission lines .
the higher spectral resolution data are presented and discussed in this paper .
the data were reduced in iraf .
figure [ figure : image ] shows the sparsepak pointings on an h@xmath4 image of ngc 891 .
[ figure : image ]
to analyze the kinematic structure of the dig halo of ngc 891 , position - velocity ( pv ) diagrams have been constructed from the spectra of all 82 fibers in pointing ` h ' ( cf .
[ figure : image ] ) . to enhance
signal - to - noise , several fibers were averaged together for each major axis distance @xmath7 at the highest @xmath0 .
figure [ figure : image ] indicates which fibers were included in each @xmath0-range . because a bright sky line interferes with several of the h@xmath4 profiles , and to increase signal - to - noise , pv diagrams were constructed of the sum of the [ n@xmath5ii ] @xmath86583 + [ s@xmath5ii ] @xmath86716 emission lines .
contour plots of the pv diagrams are included in figure [ figure : pvdiags ] .
rotation curves were derived for each pv diagram using the envelope tracing method ( e.g. , @xcite ) . because ngc 891 is nearly edge - on ( @xmath9 ; @xcite ) , the edge of each line profile furthest from the systemic velocity ( the `` envelope '' ) corresponds to gas at the line of nodes ; this velocity is thus approximately the azimuthal velocity at that @xmath7 .
the azimuthal velocity is found by ( @xcite ) : @xmath10 where @xmath11 is the velocity at the location of the envelope , @xmath12 is the velocity resolution of the instrument , and @xmath13 is the velocity dispersion of the gas .
azimuthal velocity curves were constructed from the pv diagrams , and are shown in figure [ figure : rotcurs ] .
extinction in the midplane prevented reliable velocities from being found for @xmath14 .
[ figure : rotcurs ] it is clear from figure [ figure : rotcurs ] that the derived azimuthal velocities decrease with height .
a close inspection of the pv diagrams shows that this is not an effect of signal - to - noise variations with height .
an average azimuthal velocity gradient of 0.80 km s@xmath1 arcsec@xmath1 ( 17 km s@xmath1 kpc@xmath1 ) was calculated .
similar results were obtained by following this procedure for the [ n@xmath5ii ] @xmath86583 emission line alone ( 14 km s@xmath1 kpc@xmath1 ) .
these values are in agreement with that measured by @xcite for the neutral component , 15 km s@xmath1 kpc@xmath1 .
the envelope tracing method is sensitive to the changing signal - to - noise ratio of the data with @xmath0 . to account for this effect and the radial gas distribution ,
we have generated galaxy models .
the h@xmath4 image presented in figure [ figure : image ] was used to obtain estimates of the radial density profile at each of the heights considered in the model . because the distribution of the dig is not axisymmetric , the profiles were modified by hand to better match the shape of the data pv diagrams .
the amplitude of the radial density profile was chosen such that the signal - to - noise in the model approximately matched that in the data .
we then created model galaxies with these derived radial profiles using a version of the gipsy task galmod , modified to allow for a vertical gradient in azimuthal velocity which begins at a height @xmath15 above the midplane : @xmath16,\ ] ] where @xmath17 has units of [ km s@xmath1 arcsec@xmath1 ] , and for the models considered here , the major axis rotation curve is flat [ @xmath19 km s@xmath1 ] .
the inclusion of the parameter @xmath15 was motivated by the results of @xcite , who find that the gradient in the neutral component of ngc 891 starts at approximately @xmath20 kpc ( though that result may be a consequence of beam smearing ) .
that value of @xmath15 was used for the models shown here , but we note that the appropriate value can not be determined for our data ( we lack data below @xmath21 kpc ) .
artificial sparsepak observations were made of the models to create pv diagrams , and compared to the data ( figure [ figure : pvdiags ] ) .
the model that appears to best match the data has @xmath22 km s@xmath1 kpc@xmath1 .
a statistical analysis of difference images suggests that the best match occurs at @xmath23 km s@xmath1 kpc@xmath1 , but because the model can not perfectly reproduce the shape of the pv diagrams , comparing by eye is the most reliable way to determine the gradient . to ensure that our results are not influenced by an incorrect specification of the radial density profile , we also consider models with a flat radial density profile , and find that the same gradient yields the best match .
we conclude that the vertical gradient in azimuthal velocity for the dig halo of ngc 891 is approximately 15 km s@xmath1 kpc@xmath1 .
to attempt to understand the origin of the azimuthal velocity gradient derived in this paper , the kinematic data will be compared with the results of an entirely ballistic model of disk halo flow ( cf .
further information regarding the kinematic structure will be obtained once the lower spectral resolution pointings are reduced and analyzed .
results from these efforts will be presented in a forthcoming paper .
bershady , m. a. , et al .
2004 , pasp , 116 , 565 bershady , m. a. , et al .
2005 , apjs , 156 , 311 collins , j. a. , benjamin , r. a. , & rand , r. j. 2002 , apj , 578 , 98 fraternali , f. , et al .
2005 , in asp conf .
331 , extra - planar gas , ed .
r. braun ( san francisco : asp ) , 239 rand , r. j. , kulkarni , s. r. , & hester , j. j. 1990 , apj , 352 , l1 sofue , y. & rubin , v. 2001 , ara&a , 39 , 137 swaters , r. 1994 , doctoraal scriptie , univ . groningen | we present wiyn sparsepak observations of the diffuse ionized gas ( dig ) halo of ngc 891 .
preliminary results of an analysis of the halo velocity field reveal a clear gradient of the azimuthal velocity with @xmath0 which agrees with results for the neutral gas . the magnitude of the gradient has been determined , using two independent methods , to be approximately 15 km s@xmath1 kpc@xmath1 . |
tverberg s theorem @xcite states that given @xmath14 points in @xmath9 , it is possible to split these points into @xmath4 sets @xmath6 with intersecting convex hulls , that is with @xmath10 .
if one replaces convex hulls with affine hulls , one obtains a valid statement ( lemma [ lem : affine_tverberg ] ) , which has the advantage of being applicable to any field @xcite .
lemma [ lem : affine_tverberg ] is also easier to prove than the original tverberg s theorem .
since the proof only uses properties of closure operators , the statement does generalize to any matroid ( lemma [ lem : matroidal_tverberg ] ) . in both these cases the conclusion can be strengthened a bit : instead of @xmath15 , one can require @xmath16 .
in this paper we study the variant of tverberg s theorem for matroidal closures and show that it allows a colorful version a generalization where the original points are colored and one furthermore requires that no resulting set @xmath17 , @xmath18 contains two or more points of the same color . while the version without colors is straightforward ( * ? ? ?
* lemma 12 ) the proof of the colorful version is more subtle .
moreover , our proof method yields an efficient algorithm that finds the required sets in polynomial time . before we state our results formally ,
let us introduce some notations and terminology which will allow us to nicely present the statements and proofs .
we assume that the reader is acquainted with the basic matroid theory .
we always use the symbols @xmath4 and @xmath2 to denote non - negative integers .
we use the symbols @xmath3 , @xmath19 , @xmath20 and @xmath21 for matroidal closure , affine closure , convex hull and rank function , respectively .
if @xmath1 is a set , we consider a sequence @xmath22 of elements from @xmath1 as a set of pairs @xmath23 .
with this convention we can use the set theoretic terminology for sequences : @xmath24 is the length of the sequence , @xmath25 means that @xmath26 is a subsequence of @xmath0 , we know what it means for two subsequences to be disjoint , we can use the operation @xmath27 of ( sequence ) difference , etc .
if @xmath28 is a sequence and we need to refer to the set @xmath29 , we use the symbol @xmath30 .
if @xmath31 is a map from the subsets of @xmath1 ( for example a closure operator , rank function ) , and @xmath22 is a sequence in @xmath1 , we use a shorthand @xmath32 . to make formulas and equations shorter
, we leave out the parantheses after the operators @xmath3 , @xmath19 , @xmath20 and @xmath21 when there is no danger of confusion . a coloring of a sequence @xmath28 is any map @xmath33 into some set @xmath34 of colors , that is , @xmath35 assigns to each pair @xmath36 a color from @xmath34 .
the sequence @xmath0 is _ rainbow _ with respect to @xmath35 , if the restriction of @xmath35 to @xmath0 is injective .
let us first state the non - colorful variant of tverberg s theorem for affine hulls and its easy generalization to matroidal closures .
[ lem : affine_tverberg ] let @xmath0 be a sequence of points in an affine space @xmath37 of dimension @xmath38 . if to be finite , therefore the slightly unusual formulation . ]
@xmath39 , then there exist @xmath4 pairwise disjoint subsequences @xmath6 of @xmath0 with @xmath40 .
in fact , there are @xmath4 pairwise disjoint subsequences @xmath6 satisfying @xmath41 . [
lem : matroidal_tverberg ] let @xmath1 be a ( finitary , then there exists a finite set @xmath42 such that @xmath43 . with these addition , such terms as rank or basis can be correctly defined . ] ) matroid of rank @xmath2 with closure operator @xmath3 and @xmath0 be a sequence of points in @xmath1 with @xmath44 .
then there exist @xmath4 pairwise disjoint subsequences @xmath6 of @xmath0 satisfying @xmath7 . in (
* lemma 13 ) we only stated that there exists sets @xmath17 with @xmath45 .
the proof there implies lemma [ lem : affine_tverberg ] , and ( if one replaces @xmath19 with the closure operator @xmath3 of a matroid ) lemma [ lem : matroidal_tverberg ] . in the case of matroids of finite rank , both lemmas
can also be obtain as a direct consequence of theorem [ thm : colorful_matroidal_tverberg ] . in @xcite we applied lemma [ lem : affine_tverberg ] to homology groups over finite fields .
this enabled us to prove some inequalities for simplicial complexes embeddable into various manifolds .
our colorful matrodial tverberg ( theorem [ thm : colorful_matroidal_tverberg ] ) provides a control of the resulting sets , which enables us to further improve the bounds from @xcite . for the details of the improvement , see the author s thesis @xcite .
we are now ready to state the main results of this paper .
[ thm : colorful_matroidal_tverberg ] let @xmath1 be a matroid of a finite rank @xmath2 and @xmath0 be a sequence of non - loops in @xmath1 colored by some colors in such a way that at most @xmath2 elements of @xmath0 are colored by the first color , at most @xmath46 by the second color , at most @xmath46 by the third color , etc . if @xmath44 , then there exist @xmath4 pairwise disjoint rainbow subsequences @xmath6 of @xmath0 , such that @xmath47
furthermore , if the time required to decide whether a point @xmath48 lies in the closure of a set @xmath49 is bounded by @xmath50 , then the subsequences @xmath6 can by found in time polynomial in @xmath4 , @xmath2 , @xmath50 and @xmath24 . in the proof of theorem [ thm : colorful_matroidal_tverberg ] we encounter another version of colorful matroidal tverberg s theorem .
[ thm : colored - tverberg - special ] let @xmath1 be a matroid of a finite rank @xmath2 and @xmath0 a sequence of non - loops in @xmath1 colored by @xmath2 colors in such a way that at least @xmath4 elements of @xmath0 are colored by the first color , at least @xmath51 by the second color , at least @xmath51 by the third , , at least @xmath51 by the @xmath2th color .
then there exist @xmath4 pairwise disjoint rainbow subsequences @xmath6 of @xmath0 such that @xmath47 . furthermore
, if the time required to decide whether a point @xmath48 lies in the closure of a set @xmath49 is bounded by @xmath50 , then the subsequences @xmath6 can by found in time polynomial in @xmath4 , @xmath2 , @xmath50 and @xmath24 .
note the different conditions on the number of points of each color . in theorem [ thm : colorful_matroidal_tverberg ]
these conditions are used to ensure that we have enough colors . in theorem
[ thm : colored - tverberg - special ] we already have the right number of colors , but the conditions ensure that the length of @xmath0 is sufficient
. moreover , these results are tight : [ prop : tight ] lemma [ lem : affine_tverberg ] , lemma [ lem : matroidal_tverberg ] , theorem [ thm : colorful_matroidal_tverberg ] and theorem [ thm : colored - tverberg - special ] are sharp . to be precise , for any @xmath4 and any matroid @xmath1 of rank @xmath2 there exists a sequence @xmath0 of non - loops in @xmath1 with @xmath52 such that any division of @xmath0 into @xmath4 disjoint subsequences @xmath53 ,
, @xmath54 satisfies @xmath55 . * tverberg - type theorems in @xmath9 * + let us now compare our main results with the related theorems valid in @xmath9 . in this section
@xmath56 denotes the @xmath57-dimensional simplex .
tverberg s theorem can be stated as follows : if @xmath58 is an affine map , there are @xmath4 pairwise disjoint faces @xmath59 of @xmath60 with @xmath61 .
this is the reason why tverberg s theorem is also called affine tverberg s theorem . to avoid confusion ,
we have decided not to use the name `` affine tverberg '' for lemma [ lem : affine_tverberg ] .
if @xmath4 is a prime power , zaydin @xcite showed that the same result holds for an arbitrary continuous map @xmath62 .
the statement is known as topological tverberg .
it was a long - standing open problem , whether topological tverberg can be extended to other values of @xmath4 .
the negative answer came in 2015 , when frick ( based on the previous work of mabillard and wagner @xcite ) constructed first counterexamples @xcite .
counterexamples for other values of @xmath38 and @xmath4 followed shortly afterwards .
@xcite if @xmath4 is a prime , there is a colorful version of ( topological ) tverberg s theorem @xcite as well : suppose that the vertices of @xmath63 are colored in such a way , that no color is used more than @xmath5-times .
then for every continuous map @xmath58 , there are @xmath4 pairwise disjoint rainbow faces @xmath59 of @xmath60 with @xmath61 .
the colorful version provides more control over the resulting sets @xmath64 . even if @xmath62 is an affine map , the only known proof uses topological methods and needs the assumption that @xmath4 is prime . whether this assumption can be relaxed in the affine situation
is an open question .
moreover , the topological proof does not provide any way how to find the pairwise disjoint faces , it merely shows their existence .
we see that theorem [ thm : colorful_matroidal_tverberg ] does not require @xmath4 to be a prime number , it relaxes the conditions on the colors from topological version a bit and provides an efficient algorithm for finding the desired sets .
we also note that brny , kalai and meshulam proved another , very different tverberg type theorem for matroids @xcite , they considered continuous maps from the matroidal complex and showed the following : if @xmath65 denotes the maximal number of disjoint bases in a matroid m of rank @xmath66 , then for any continuous map @xmath62 from the matroidal complex @xmath1 into @xmath9 there exists @xmath67 disjoint independent sets @xmath68 such that @xmath69 .
we postpone the technical proofs of our main results , theorems [ thm : colorful_matroidal_tverberg ] and [ thm : colored - tverberg - special ] to the end of the paper .
first we prove proposition [ prop : tight ] showing their tightness .
the proof is a variant of the standard construction for showing that tverberg s theorem is tight .
we start with an auxiliary lemma .
[ lem : intersection ] let @xmath1 be a matroid with finite basis @xmath70 . then for any two sets
@xmath71 @xmath72 since the operator @xmath3 is monotone , the inclusion @xmath73 is obvious .
let us now prove the opposite inclusion .
let @xmath74 be an arbitrary element .
we want to show that @xmath75 . if @xmath76 is a loop , @xmath77 .
so assume that @xmath76 is not a loop .
let @xmath78 and @xmath79 be inclusion minimal subsets with @xmath80 and @xmath81 , respectively . since we assume that @xmath76 is not a loop , @xmath82 .
we will show by contradiction that @xmath83 , hence proving the claim .
if @xmath84 , we may up to symmetry assume that there is an element @xmath85 which does not lie in @xmath86
. from the inclusion minimality of @xmath87 follows that @xmath88 .
the exchange principle yields @xmath89 .
similarly @xmath90 for an arbitrary @xmath91 .
the set @xmath92 is independent being a subset of a basis @xmath70 . by construction @xmath93 . comparing the ranks of both sides and using the fact that @xmath94
, we see that @xmath95 has to belong to @xmath87 .
since @xmath95 was arbitrary , this implies @xmath96 in contradiction with @xmath87 being minimal with @xmath80 .
we can now finally prove proposition [ prop : tight ] .
let @xmath97 be a basis of the matroid @xmath1 .
it suffices to take @xmath98 let @xmath6 be disjoint subsequences of @xmath0 .
then @xmath99 where the first equality follows by inductive application of lemma [ lem : intersection ] and the second equality uses the fact that each element @xmath100 is missing in at least one sequence @xmath17 .
we also note that the assumption in theorem [ thm : colorful_matroidal_tverberg ] that there are at most @xmath4 points of the first color is necessary .
otherwise , one can consider the sequence @xmath101 in @xmath102 where the first @xmath57 elements are red and the last element @xmath103 is blue . then
although the length of @xmath0 can be arbitrary , there are no three disjoint rainbow subsequences @xmath53 , @xmath104 , @xmath105 with @xmath106 . on the other hand ,
it is not true that this condition is necessary in every matroid .
for example , consider the affine line over the field with two elements .
we begin the proof by showing that theorem [ thm : colored - tverberg - special ] implies theorem [ thm : colorful_matroidal_tverberg ] .
the reduction of theorem [ thm : colorful_matroidal_tverberg ] to theorem [ thm : colored - tverberg - special ] follows a well known pattern , a similar reduction previously appeared in the proof of the optimal colored tverberg theorem @xcite or in sarkaria s proof for the prime power tverberg theorem @xcite , see also de longueville s exposition ( * ? ? ?
nevertheless , there are subtle differences because we are working in greater generality and because we need to take algorithmic aspects into consideration .
assume that the assumptions of theorem [ thm : colorful_matroidal_tverberg ] are satisfied .
we show how to turn the sequence @xmath0 and the matroid @xmath1 with closure operator @xmath3 into a sequence @xmath26 and matroid @xmath107 with closure operator @xmath108 that satisfy the assumptions of theorem [ thm : colored - tverberg - special ] .
moreover , we construct @xmath109 and the coloring of @xmath26 in a such way that the sets @xmath110 , @xmath111 , , @xmath112 will satisfy @xmath113 iff and only if @xmath114 and the rainbowness of @xmath115 will imply that @xmath17 is rainbow .
let @xmath2 be the rank of @xmath1 and @xmath38 the number of colors used in @xmath0
. from the conditions follows that @xmath116 . if the length of @xmath0 is strictly larger than @xmath117 , we throw the superfluous elements of @xmath0 away .
this does not add a point of any color , therefore all assumptions of theorem [ thm : colorful_matroidal_tverberg ] remain preserved .
so we may assume that the length of @xmath0 is precisely @xmath117 .
we form @xmath107 from @xmath1 by adding @xmath118 new coloops that is independent on any set that does not contain @xmath76 . in other words
, we form @xmath107 as the direct sum of @xmath1 with the uniform matroid @xmath119 . ]
now we form the sequence @xmath26 by appending @xmath121 to @xmath0 .
clearly we can color the new elements of @xmath26 so that in total there are exactly @xmath4 points of the first color , and exactly @xmath51 points of every other color .
we see that @xmath26 , @xmath107 satisfy the assumptions of theorem [ thm : colored - tverberg - special ] .
it follows that there are @xmath4 rainbow subsequences @xmath122 of @xmath26 satisfying @xmath123 .
since the points @xmath124 are coloops and since each one of them was added exactly @xmath5-times , it follows that they can not contribute to @xmath125 .
consequently , @xmath126 and @xmath127 .
we conclude that @xmath128 are the required subsequences of @xmath0 .
observe that the reduction is polynomial in @xmath4 , @xmath2 , @xmath50 and @xmath24 .
now we can start with the proof of theorem [ thm : colored - tverberg - special ] .
here we describe the main idea .
we let @xmath54 be a rainbow independent subsequence of the maximal rank . in an ideal case @xmath129 and we may obtain the remaining subsequences @xmath130 by apply induction on the sequence @xmath131 inside @xmath1 . however , we may be unlucky .
it may happen that no such @xmath54 satisfies @xmath129 , see fig .
[ fig : unlucky ] .
.,title="fig : " ] + a situation in which no rainbow @xmath54 satisfies @xmath129 .
@xmath132 , @xmath3 is the affine hull , @xmath133 .
points @xmath134 use the first color ( blue ) , @xmath135 use orange , @xmath136 , @xmath137 use red .
we see that in this case we could simply take the subsequence @xmath138 and unify colors blue and red into one color ( say violet ) . then @xmath26 lives in a submatroid of rank @xmath139 and satisfies the conditions of theorem [ thm : colored - tverberg - special ] , so we may use induction .
we obtain subsequences @xmath140 of @xmath26 satisfying @xmath141 .
these are clearly also subsequences of @xmath0 .
moreover they are not only rainbow in the violet - orange coloring , but also in the original blue - orange - red coloring . in the proof
we show that if @xmath142 , we may always resolve the situation by an analogous trick .
let us now carry out the technical details .
since we promised an algorithmic solution , we describe an algorithm that finds the desired subsequences .
first we compute @xmath143 .
since instead of @xmath0 we can consider the subsequence @xmath26 formed by the elements colored by the first @xmath144 colors ( while preserving all assumptions of theorem [ thm : colored - tverberg - special ] ) , we may assume that @xmath145 .
now we find an inclusion maximal independent rainbow subsequence @xmath146 of @xmath0 .
this can clearly be done in time polynomial in @xmath4 , @xmath50 , @xmath2 and @xmath24 .
we will proceed in the proof by induction on the triple @xmath147 ( in lexicographical ordering ) .
if @xmath148 or @xmath149 the statement is trivial , so assume @xmath150 . if @xmath151 , then @xmath152 . because @xmath146 is rainbow , @xmath153 and @xmath1 satisfy the assumptions of theorem [ thm : colored - tverberg - special ] for @xmath154 . by applying induction
we obtain @xmath51 disjoint rainbow subsequences @xmath130 of @xmath153 with @xmath155 .
if we now set @xmath156 we see that @xmath6 are the desired disjoint rainbow subsequences with @xmath157 .
therefore we may assume that @xmath158 we would like to increase @xmath146 by adding a point of a color that is not yet used in @xmath146 .
unfortunately , this is not possible without replacing some points of @xmath146 first .
our algorithm uses a cycle to find out which points to replace and how . within the cycle we need to keep track of `` replacement rules '' which makes this part a bit technical .
moreover , there are three possibilities what can occur at one iteration of the cycle : either we construct a larger independent rainbow set @xmath146 , we find the desired sets @xmath6 in a smaller submatroid , or we adjust the replacement rules . [ [ the - cycle ] ] the cycle + + + + + + + + + in the @xmath159th step ( @xmath160 ) of the cycle the replacement rules consist of the following data : 1 .
set @xmath161 of colors ( this set corresponds to colors that we may use while replacing some points ) , 2 .
subsequence @xmath162 of @xmath146 ( eventually we would like to replace the subsequence @xmath162 of @xmath146 by another sequence @xmath163 ) , 3 . for each element @xmath164 whose color is in @xmath161 and which does not lie in @xmath165 a subsequence @xmath163 of @xmath0 ( we want to replace @xmath162 with @xmath163 , hence increasing the length of our subsequence by one ) to simplify the terminology , if @xmath166 is a subsequence of @xmath0 , let @xmath167 denote the set of all the colors used by elements of @xmath166 .
if @xmath168 is a set of colors , let @xmath169 be the subsequence of @xmath0 formed by all elements with color from @xmath168 .
we want the data to satisfy the following conditions : 1 .
@xmath170,[c : more - colors ] 2 .
@xmath171 for some @xmath172,[c : rkp - colors ] 3 .
@xmath173,[c : rkp - augmenting - size ] 4 .
@xmath174 and @xmath175[c : same_affine_hulls ] 5 .
@xmath176 and @xmath177[c : no - further - colors ] note that conditions [ c : rkp - colors ] and [ c : rkp - augmenting - size ] imply that @xmath163 only contains elements that have the same colors as points in @xmath162 plus one additional point that has color @xmath178 , which is not yet present in @xmath146 .
the first step ( @xmath179 ) is easy .
we set @xmath180 and let @xmath181 be all the colors of @xmath0 except for those already used in @xmath146 .
no element @xmath182 is contained in are the elements of @xmath0 whose color lies in @xmath181 and we assume that @xmath0 contains only non - loop elements . ] @xmath183 , so we need to define the set @xmath184 for every such @xmath164 .
we simply put @xmath185 .
now we check that the above defined sets satisfy all the prescribed conditions .
note that by , @xmath186 .
this together with the fact that @xmath146 is independent implies that @xmath187 . since we have @xmath2 colors , there is a color that is not used in @xmath146 . in other words
, @xmath181 is nonempty .
hence conditions [ c : more - colors][c : no - further - colors ] are satisfied trivially ( with @xmath188 in condition [ c : rkp - colors ] ) .
so suppose that the sets @xmath161 , @xmath162 and @xmath163 are already constructed . since @xmath189 there are three cases that may occur : 1 .
@xmath190,[c : lower - dimensional ] 2 .
@xmath191 or[c : stop - backward - induction ] 3 .
@xmath192 and @xmath193.[c : continue - backward - induction ] we deal with the particular cases separately : in this case , we may apply the trick we used for fig .
[ fig : unlucky ] .
let us describe it formally .
we set @xmath194 and @xmath195 .
@xmath1 has rank @xmath2 and by we know that @xmath196 .
it follows that @xmath197 and since @xmath189 , we also have @xmath198 .
condition [ c : more - colors ] implies @xmath170 , so there is a point @xmath199 .
because @xmath162 is rainbow and independent and @xmath200 , @xmath201 has @xmath144 distinct elements , say @xmath202 .
we define @xmath203 . in @xmath26
we recolor @xmath164 and all points of color @xmath204 by a new color @xmath205 . because @xmath206 ( we evaluate @xmath207 with respect to the original coloring ) , the assumption @xmath190 ( case [ c : lower - dimensional ] ) implies that @xmath26 is a sequence of elements from @xmath107 .
also in @xmath26 there are @xmath144 colors , at least @xmath4 elements of color @xmath205 and at least @xmath51 elements of all the remaining colors .
therefore , the assumptions of theorem [ thm : colored - tverberg - special ] are satisfied for @xmath208 . by induction
we obtain the desired disjoint rainbow subsequences @xmath6 of @xmath26 ( which itself is a subsequence of @xmath0 ) with @xmath209 .
these subsequences are rainbow with respect to the new coloring of @xmath26 . by the construction of the new coloring these subsequences are also rainbow in the original coloring of @xmath0 . in this case
, we construct a new independent rainbow subsequence @xmath210 with @xmath211 : we pick a point @xmath212 with @xmath213 and set @xmath214 .
before we show that such @xmath210 is a rainbow independent subsequence of size @xmath215 , we prove the following auxiliary equality : @xmath216 indeed , @xmath217 where the last equality uses the fact that @xmath174 from condition [ c : same_affine_hulls ] . because any closure operator @xmath3 satisfies @xmath218 we may rewrite the expression further to @xmath219 by condition [ c : same_affine_hulls ] @xmath220 , which reduces the equality to : @xmath221 using again , we obtain @xmath222 since @xmath189 , equation follows .
using the fact that @xmath189 , we are now ready to verify that @xmath210 is a rainbow independent subsequence with @xmath211 . * * @xmath211 : * @xmath223 . because @xmath146 is rainbow , condition , for some @xmath224 [ c : rkp - colors ] implies that the sequences @xmath225 and @xmath163 do not share any color .
in particular , they are disjoint and @xmath226 . since @xmath173 ( condition [ c : rkp - augmenting - size ] ) , @xmath227 . because @xmath189 , we have @xmath228 . * *
@xmath210 is rainbow : * @xmath163 contains one element of color that is not used in @xmath146 , otherwise it uses the same colors as @xmath162 . because @xmath229 , we see that @xmath230 uses exactly @xmath215 colors . this , together with the previous item , yields that @xmath230 is rainbow . * * @xmath210 is independent * : from the equality we get @xmath231 moreover , we have chosen a point @xmath164 which satisfies @xmath213 , so @xmath232 since @xmath146 was independent and @xmath210 has exactly one element more , the independence of @xmath210 follows .
let @xmath233 be an inclusion maximal independent rainbow subsequence of @xmath0 that contains @xmath210 .
we may now start our algorithm again but this time we replace the maximal independent rainbow subset @xmath146 by @xmath233 .
we have decreased the quantity @xmath234 and preserved @xmath2 and @xmath4 . by induction
we obtain the desired disjoint rainbow subsequences @xmath6 with @xmath235 . in this case , we show how to construct sets @xmath236 , @xmath237 and for every @xmath238 with @xmath239 we construct a subsequence @xmath240 .
we choose @xmath237 to be any inclusion minimal subsequence @xmath241 satisfying @xmath242 because we assume that @xmath192 , such set @xmath237 does exist .
we further define @xmath243 before we construct @xmath240 , we prove the following auxiliary claim : [ cl : two ] @xmath244 by condition [ c : more - colors ] , @xmath245 . by eq . , we have @xmath246 . by construction
both @xmath162 and @xmath237 are subsequences of the independent sequence @xmath146 which together with the preceding yields @xmath247 .
condition [ c : more - colors ] and the fact that we are in case [ c : continue - backward - induction ] yields @xmath248 .
since also @xmath249 , we see that @xmath250 and @xmath251 .
now we construct sets @xmath240 for all points @xmath238 satisfying @xmath252 .
let @xmath164 be such a point . by definition of @xmath237 , @xmath249
, so @xmath164 can not lie in @xmath207 .
equation implies @xmath253 . because @xmath241 is a rainbow set is rainbow ! ] , there exists a unique element @xmath254 with @xmath255 .
since we assume @xmath256 , we have @xmath257 , where the last inclusion follows from condition [ c : more - colors ] .
in particular , @xmath258 , hence @xmath259 since @xmath237 is an inclusion minimal subsequence of @xmath146 for which @xmath249 , there exists an element @xmath260 such that @xmath261 .
since @xmath262 , the exchange principle implies @xmath263 .
it easily follows that @xmath264 claim [ cl : two ] together with imply that @xmath265 .
since @xmath266 was chosen to satisfy @xmath267 , we have @xmath268 as well .
together with @xmath260 , this implies that @xmath269 is defined .
we set does depend on the choice of @xmath266 , i.e. , if we choose another @xmath260 that satisfies @xmath270 , we obtain a different set @xmath240 . ]
@xmath271 it remains to show that our assignment satisfies conditions [ c : more - colors][c : no - further - colors ] .
* * condition [ c : more - colors ] : * by , we have @xmath272 .
condition [ c : no - further - colors ] implies that @xmath161 contains a color that is not used in @xmath146 and since @xmath241 , which together with yields @xmath273 .
condition [ c : more - colors ] follows . *
* condition [ c : rkp - colors ] : * condition [ c : rkp - colors ] states that @xmath274 for some @xmath275 , in particular @xmath276 . together with the fact that elements @xmath164 and @xmath4 have the same color ( @xmath277 ) , yields @xmath278 .
if we now apply condition [ c : rkp - colors ] for @xmath269 and claim [ cl : two ] , we see that @xmath279 , where @xmath280 .
note that @xmath281 , hence @xmath282 .
condition [ c : rkp - colors ] follows . *
* condition [ c : rkp - augmenting - size ] : * by definition @xmath283 . because @xmath237 is a subset of the rainbow set @xmath146 , @xmath237 is itself rainbow .
together with @xmath284 , where @xmath285 , this implies that the sets @xmath286 and @xmath287 are disjoint . since @xmath288 ( equation ) , @xmath289 and @xmath290 ( conditions [ c : rkp - colors ] and [ c : no - further - colors ] ) , we have @xmath291 and @xmath292 . from @xmath293
follows @xmath294 .
since @xmath254 , we have @xmath295 , where the last equality uses the induction hypothesis for @xmath159 . claim [ cl : two ] then yields @xmath296 as desired . * * condition [ c : same_affine_hulls ] : * by definition ( ) @xmath297 , so we only need to verify that @xmath298 .
let us compute .
using the fact that @xmath299 from condition [ c : same_affine_hulls ] and we may rewrite @xmath300 as follows : @xmath301 now we use condition [ c : same_affine_hulls ] for @xmath159 ( @xmath302 ) .
we obtain @xmath303 where the last equality follows from . *
* condition [ c : no - further - colors ] : * by definition @xmath304 .
this implies @xmath305 .
hence @xmath306 . by the induction assumption @xmath176 . because @xmath307 is rainbow , @xmath308 .
claim [ cl : two ] then implies @xmath309 as desired . because @xmath177 and @xmath281 , we have @xmath310 as well .
it follows that we may increase @xmath159 and continue in the loop . in each step of the cycle
we either terminate and output the desired subsequences , or we construct a sequence @xmath237 whose rank is strictly larger than the rank of @xmath162 ( claim [ cl : two ] ) .
since the rank of @xmath237 is from above bounded by @xmath311 it follows that the loop terminates after at most @xmath311 iterations .
verifying that all other steps can be done in time polynomial in @xmath4 , @xmath2 , @xmath50 and @xmath24 and that they are repeated only polynomial number of times is easy .
rota basis conjecture @xcite is a well known problem in matroid theory which has a close connection to our colorful matroidal tverberg s theorem . let us restate it so that the similarity is clearly visible
. let @xmath1 be a matroid of rank @xmath2 .
let @xmath0 be a sequence of @xmath312 elements colored by @xmath2 colors such that points of each color form a basis .
do there always exist @xmath2 pairwise disjoint rainbow subsequences @xmath313 of @xmath0 with @xmath314 ? in its full generality the conjecture has only been verified for @xmath315 @xcite . the conjecture is also known to be true in several special cases @xcite .
proof of theorem [ thm : colored - tverberg - special ] indicates the difficulties that appear if one tries to proof rota s basis conjecture purely combinatorially . | in this paper we show a variant of colorful tverberg s theorem which is valid in any matroid : let @xmath0 be a sequence of non - loops in a matroid @xmath1 of finite rank @xmath2 with closure operator @xmath3 .
suppose that @xmath0 is colored in such a way that the first color does not appear more than @xmath4-times and each other color appears at most @xmath5-times .
then @xmath0 can be partitioned into @xmath4 _ rainbow _
subsequences @xmath6 such that @xmath7 .
in particular , @xmath8 .
a subsequence is called rainbow if it contains each color at most once .
the conclusion of our theorem is weaker than the conclusion of the original tverberg s theorem in @xmath9 , which states that @xmath10 , whereas we only claim that @xmath11 . on the other hand
, our theorem strengthens the tverberg s theorem in several other ways : it is applicable to any matroid ( whereas tverberg s theorem can only be used in @xmath9 ) , instead of @xmath12 we have the stronger condition @xmath7 , and we add a color constraints that are even stronger than the color constraints in the colorful version of tverberg s theorem .
recently , we used the first property and applied the non - colorful version of this theorem to homology groups with @xmath13 coefficients to obtain several non - embeddability results , for details we refer to @xcite . |
during the last 30 years the addition laws of elliptic functions stay in the focus of the studies in the nonlinear equations of mathematical physics .
a large part of the interest was drawn to the subject by the works of f. calogero @xcite , where several important problems were reduced to the elliptic addition laws .
the term `` calogero - moser model '' being widely used in literature , the papers caused a large series of publications , where on one hand more advanced problems were posed and on the other hand some advances were made in the theory of functional equations .
the `` addition theorems '' for weierstrass elliptic functions : @xmath2 played the key rle in the works of the that period . at the same time , the development of the algebro - geometric methods of solution of integrable systems @xcite employed in an essential way the addition formulas for theta functions of several variables .
the same addition formulas were needed in applications of hirota method . in @xcite the addition theorems for vector baker - akhiezer functions of several variables are obtained and a program is put forward to apply the addition theorems to problems of the theory of integrable systems , in particular , to multidimensional analogs of calogero - moser type systems .
the fundamental fact of the elliptic functions theory is that any elliptic function can be represented as a rational function of weierstrass functions @xmath3 and @xmath4 .
the corresponding result ( see @xcite ) in the theory of hyperelliptic abelian functions is formulated as follows : any hyperelliptic function can be represented as a rational function of vector functions @xmath5 and @xmath6 , where @xmath7 is the genus of the hyperelliptic curve on the jacobi variety of which the field of abelian functions is built . in particular , for the genus @xmath8 sigma - function we obtain the following trilinear differential addition theorem @xmath9\sigma(u)\sigma(v)\sigma(w)\big|_{u+v+w=0}=0.\ ] ] in the present paper we find the explicit formulas for the addition law of the vector functions @xmath0 and @xmath1 . as an application
the higher genus analogs of the frobenius - stickelberger formula are obtained .
our approach is based on the explicit construction of the groupoid structure that is adequate to describe the algebraic structure of the space of @xmath7-th symmetric powers of hyperelliptic curves .
take a topological space @xmath10 .
a space @xmath11 together with a mapping @xmath12 is called _ a space over @xmath10_. the mapping @xmath13 is called `` an anchor '' in differential geometry .
let two spaces @xmath14 and @xmath15 over @xmath10 be given .
a mapping @xmath16 is called _ a mapping over @xmath10 _ , if @xmath17 for any point @xmath18 . by _ the direct product over @xmath10 _ of the spaces @xmath14 and @xmath15 over @xmath10 we call the space @xmath19 together with the mapping @xmath20 .
the space @xmath10 together with the identity mapping @xmath21 is considered as the space over itself .
a space @xmath11 together with a mapping @xmath22 is called _ a groupoid over @xmath10 _ , if there are defined the structure mappings over @xmath10 @xmath23 that satisfy the axioms 1 .
@xmath24 provided @xmath25 2 .
@xmath26 , provided @xmath27 .
the mapping @xmath28 may not be defined for all pairs @xmath29 and @xmath30 from @xmath11 . a groupoid structure on @xmath11 over the space @xmath10 is called commutative , if @xmath31 , provided @xmath27 . a groupoid structure on the algebraic variety @xmath11 over the algebraic variety @xmath10
is called _ algebraic _ , if the mapping @xmath32 as well as the structure mappings @xmath28 and @xmath33 are algebraic .
we take as @xmath10 the space @xmath34 with coordinates @xmath35 , @xmath36 .
let @xmath37 where @xmath38 , be a polynomial in @xmath39 and @xmath40 .
define the family of plane curves @xmath41 we assume that at a generic value of @xmath42 , genus of the curve from @xmath43 has fixed value @xmath7 .
let us take as @xmath11 the universal fiber - bundle of @xmath7-th symmetric powers of the algebraic curves from @xmath43 .
a point in @xmath11 is represented by the collection of an unordered set of @xmath7 pairs @xmath44 and an @xmath45-dimensional vector @xmath42 that are related by @xmath46 , @xmath47 .
the mapping @xmath13 takes the collection to the point @xmath48 let @xmath49 be an entire rational function on the curve @xmath43 with the parameters @xmath42 .
a zero of the function @xmath49 on the curve @xmath43 is the point @xmath50 , such that @xmath51 .
the total number of zeros of the function @xmath49 is called the order of @xmath49 .
the further construction is based on the following fact .
let @xmath49 be an order @xmath52 @xmath53 , entire rational function on the curve @xmath43 .
then the function @xmath49 is completely defined ( up to a constant with respect to @xmath54 factor ) by any collection of @xmath55 its zeros .
this fact is a consequence of weierstrass gap theorem ( lkensatz ) . in particular ,
an ordinary univariate polynomial is an entire rational function on the curve of genus @xmath56 and is completely defined by the collection of all its zeros .
let us construct the mapping @xmath33 .
let a point @xmath57,\lambda\}\in \mathsf{x}$ ] be given .
let @xmath58 be the entire rational function of order @xmath59 on the curve @xmath43 defined by the vector @xmath42 , such that @xmath58 is zero in @xmath60 , that is @xmath61 , @xmath47 .
denote by @xmath62 $ ] the complement of @xmath63 $ ] in the set of zeros of @xmath58 .
denote by @xmath64 the point in @xmath11 thus obtained and set @xmath65 .
so , the set of zeros of the function @xmath58 , which defines the mapping @xmath33 , is the pair of points @xmath66 from @xmath11 and @xmath67 . the mapping @xmath33 is an involution , that is @xmath68 .
let us construct the mapping @xmath28 .
let two points @xmath69,\lambda\}$ ] and @xmath70,\lambda\}$ ] from @xmath11 be given .
let @xmath71 be the entire rational function of order @xmath72 on the curve @xmath43 defined by the vector @xmath42 , such that @xmath71 is zero in @xmath73 and @xmath74 .
denote by @xmath75 $ ] the complementary @xmath7 zeros of @xmath71 on the curve @xmath43 .
denote by @xmath76 the point in @xmath11 thus obtained and set @xmath77 .
so , the set of zeros of the function @xmath71 , which defines the mapping @xmath28 , is the triple of points @xmath78 from @xmath11 and @xmath79 .
[ universal - groupoid ] the above mappings @xmath28 and @xmath33 define the structure of the commutative algebraic groupoid over @xmath80 on the universal fiber - bundle @xmath11 of @xmath7-th symmetric powers of the plane algebraic curves from the family @xmath43 . the mapping @xmath28 is symmetric with respect to @xmath60 and @xmath64 , and thus defines the commutative operation . by the construction
the mappings @xmath32 , @xmath28 and @xmath33 are algebraic .
[ associatif ] the mapping @xmath28 is associative . let three points @xmath81 and @xmath76 be given .
let us assign @xmath82 we have to show that @xmath83 let @xmath84 be the function defining the point @xmath85 , and @xmath86 be the function defining the point @xmath87 .
consider the product @xmath88 .
it is a function of order @xmath89 with zeros at @xmath90 therefore , the function @xmath91 is an entire function of order @xmath92 with the zeros @xmath93 .
similarly , the product @xmath94 is a function of order @xmath89 with zeros at @xmath95 hence we find that @xmath96 is an entire function of order @xmath92 with the zeros @xmath97 .
the functions @xmath98 and @xmath99 have order @xmath92 and both vanish at the points @xmath100 thus by weierstrass gap theorem @xmath101 and , therefore , @xmath102 .
[ weak_inverse ] the mappings @xmath28 and @xmath33 satisfy the axiom .
let two points @xmath60 and @xmath64 be given .
assign @xmath103 we have to show that @xmath104 consider the product @xmath105 , which is the function of order @xmath89 with the zeros @xmath106 as @xmath107 the function @xmath108 is the entire function of order @xmath59 with zeros at @xmath109 , that is @xmath110 . hence it follows that @xmath111 the theorem is proved .
the lemma below is useful for constructing the addition laws on our groupoids .
[ even_part ] given @xmath112 , let us assign @xmath113 and @xmath114 , @xmath115 then @xmath116 the formula of lemma [ even_part ] is important because its left hand side depends formally on @xmath76 only , while the right hand side is completely defined by the pair @xmath81 .
the above general construction becomes effective once we fix a model of the family of curves , that is once the polynomial @xmath117 is given .
we are especially interested in the models of the form , cf .
for instance @xcite , @xmath118 where @xmath119 and the summation is carried out over the range @xmath120 , @xmath121 under the condition @xmath122 .
it is important that a model of the kind ( possibly with singular points ) exists for an arbitrary curve . at the generic values of @xmath42
a curve in such a family has genus @xmath123 in this paper we consider in detail the case @xmath124 , that is the families of hyperelliptic curves .
a hyperelliptic curve @xmath43 of genus @xmath7 is usually defined by a polynomial of the form @xmath126
in this paper we apply the change of variables @xmath127 in order to simplify the formulas in the sequel .
below we study the constructions related to the hyperelliptic curves defined by the polynomials of the form @xmath128 let us introduce the grading by assigning @xmath129 @xmath130 and @xmath131 .
then the polynomial @xmath117 becomes a homogeneous polynomial of the weight @xmath132 an entire function on @xmath43 has a unique representation as the polynomial @xmath133 where @xmath134.$ ] in such representation we have not more than one monomial @xmath135 of each weight , by definition @xmath136 .
the order of a function @xmath137 is equal to the maximum of weights of the monomials that occur in @xmath137 .
in fact , on the set of zeros of the polynomial @xmath137 we have @xmath138 . therefore , the zeros of @xmath137 that lie on the curve are defined by the roots @xmath139 of the equation @xmath140 the total number of the roots is equal @xmath141 , where @xmath142 denotes the degree of the polynomial @xmath143 in @xmath39 , which is exactly the highest weight of the monomials in @xmath137 .
in this case weierstrass gap theorem asserts that the sequence of nonnegative integers @xmath144 , @xmath145 , @xmath146 , in ascending order has precisely @xmath7 `` gaps '' in comparison to the sequence of all nonnegative integers .
all the gaps are less than @xmath59 . for a given point @xmath69,\lambda\}\in \mathsf{x}$ ] the entire function @xmath58 defining the mapping @xmath33 has the form @xmath147 in fact , as @xmath148 , any entire function of order @xmath59 does not depend on @xmath40 .
the function @xmath58 defines the unique point @xmath149,\lambda\},\ ] ] which , obviously , also belongs to @xmath11 .
let us construct the functions @xmath84 that have the properties required in lemmas [ associatif ] and [ weak_inverse ] .
define the @xmath150-dimensional row - vector @xmath151,\ ] ] which is composed of all monomials @xmath152 of weight not higher than @xmath72 ( the restriction @xmath145 applies ) .
then , up to a factor constant in @xmath54 , the function @xmath153 is equal to the determinant of the matrix composed of @xmath154 rows @xmath155 , @xmath156 and @xmath157 , @xmath158 by the construction the function @xmath153 vanishes at the points @xmath73 and @xmath74 , and is uniquely defined by this property . as @xmath159 , at fixed @xmath42 the function @xmath153 has @xmath59 zeros at the given points of the curve and the collection of zeros at @xmath75,$ ] @xmath47 , which defines the unique point @xmath76 in @xmath11 .
let @xmath160 denote the @xmath161-th symmetric power of the space @xmath162 .
a point of the space @xmath160 is an unordered collection @xmath163,$ ] @xmath164 .
consider the space @xmath165 and let us define the mapping @xmath166 take a point @xmath167,z\}\in \mathsf{s}$ ] .
denote by @xmath168 the vandermonde matrix , composed of @xmath7 rows @xmath169 , denote by @xmath170 the diagonal matrix @xmath171 and by @xmath172 the vector @xmath173 .
set @xmath174 it is clear that the domain of definition of the mapping @xmath175 is the open and everywhere dense subset @xmath176 in @xmath177 consisting of the points @xmath178,z\}$ ] such that the determinant @xmath168 does not vanish .
we define the mappings @xmath179 and @xmath180 by the following formulas : let @xmath181 and @xmath182 , then @xmath183,\lambda\})=\{[(x_i , y_i)],\lambda_2\},\ ] ] where @xmath184 @xmath185 @xmath186,z\})=\{[(\xi_i,\eta_i)],p_s(t)\}.\ ] ] by the construction , the mappings are mappings over @xmath10 .
the domain of definition of @xmath187 coincides with the domain @xmath188 of definition of the mapping @xmath175 .
let @xmath189 , then @xmath190 .
let @xmath191 , then @xmath192 .
thus we have [ birazio ] the mappings @xmath193 and @xmath187 establish the birational equivalence of the spaces @xmath11 and @xmath177 over @xmath194 the assertion of lemma [ birazio ] helps to transfer onto the space @xmath177 the groupoid over @xmath10 structure , which is introduced by theorem [ universal - groupoid ] on the space @xmath11 .
let @xmath195 .
the birational equivalence induces the mappings @xmath196 and @xmath197 that are defined by the formulas @xmath198 the mappings @xmath196 and @xmath197 define the structure of commutative algebraic groupoid over the space @xmath199 on the space @xmath177 .
let us proceed to constructing the structure of algebraic groupoid over @xmath200 on the space @xmath125 .
the classical vite mapping is the homeomorphism of spaces @xmath201 .
let us use vite mapping to construct a birational equivalence @xmath202 .
let @xmath203\in \operatorname{sym}^{g}(\mathbb{c}^2)$ ] and @xmath204 .
let us assign @xmath205 , @xmath206 and @xmath207 .
define the mapping @xmath208 and its inverse @xmath209 with the help of the relations @xmath210 note , that @xmath208 is a rational mapping , while @xmath209 is a nonsingular algebraic mapping .
using the mapping @xmath208 , we obtain the mapping @xmath211 and its inverse @xmath212 .
_ the companion matrix of a polynomial @xmath213 _ is the matrix @xmath214 where @xmath215 is the @xmath216-th basis vector in @xmath217 .
its characteristic polynomial @xmath218 is @xmath219 the companion matrices @xmath220 of the polynomials@xmath213 for @xmath221 , have the form @xmath222 note , that the companion matrix for @xmath223 is included in the companion matrix for @xmath224 as the lower right @xmath225 submatrix .
we make use of the following property of a companion matrix .
let the polynomial @xmath226 and one of its roots @xmath227 be given .
set @xmath228 .
then the relations @xmath229 hold for an arbitrary vector @xmath230 .
[ companion ] let @xmath231 then @xmath232 thus the lemma holds for @xmath233 .
one can complete the proof by induction .
the mapping @xmath234 with respect to which @xmath235 is a mapping over @xmath200 , is given by the formula @xmath236 using lemma [ companion ] one can directly verify the `` over '' property , that is , that @xmath237 for any @xmath238 .
let @xmath239 .
the birational equivalence @xmath235 induces the mappings @xmath240 and @xmath241 defined by formulas @xmath242 the mappings @xmath240 and @xmath241 define the structure of commutative algebraic groupoid over the space @xmath199 on the space @xmath125 .
in what follows we use the shorthand notation @xmath243 let @xmath244 then @xmath245 introduce the @xmath246-matrix @xmath247 that is composed of the @xmath248-matrix @xmath249 with the help of the companion matrix @xmath220 of the polynomial@xmath213 .
denote by @xmath250 the matrix composed of the first @xmath7 columns of @xmath251 , and denote by @xmath252 the @xmath253-st column of @xmath251 .
let @xmath254 and let @xmath255 , then @xmath256[additto_1 ] suppose the points @xmath257 @xmath258 , are defined .
we rewrite the function @xmath71 as a linear combination of monomials @xmath259 where @xmath260 , @xmath261.\ ] ] let us set @xmath262 we assign weights to the parameters @xmath263 by the formula @xmath264 . then @xmath265 .
let @xmath266 be the point defining any of the collections of @xmath7 zeros of the function @xmath267 consider the function @xmath268 . by the construction @xmath269
if @xmath270 let us apply lemma [ companion ] .
we obtain @xmath271 where @xmath272 . using the above notation , we come to the relation @xmath273 where @xmath274 .
suppose the polynomial @xmath213 has no multiple roots , then from the equalities @xmath275 @xmath276 one can conclude that @xmath277 by substituting the points @xmath278 and @xmath279 to , we obtain the system of @xmath72 linear equations , which is satisfied by the coefficients @xmath280 of the entire function @xmath71 .
the assertion of the theorem is the compatibility condition of the system of linear equations obtained .
the vectors @xmath280 of coefficients of the entire function @xmath71 are expressed by the formulas @xmath281^{-1}(\ell(\overline{a}_1)-\ell(\overline{a}_2)),\\ h_1(a_1,a_2)=-\frac{1}{2}\big[(\ell(\overline{a}_1)+ \ell(\overline{a}_2))-(l(\overline{a}_1)+l(\overline{a}_2))h_2(a_1,a_2)\big].\end{gathered}\ ] ] as vector functions of the points @xmath282 and @xmath283 from @xmath125 .
now , we know the coefficients @xmath280 of @xmath71 and we can give the expression of @xmath284 as a function of @xmath285 and @xmath286 .
it follows from that the following assertion holds .
@xmath287^{-1}(h_1+r_2(c^{(3)})p_{\operatorname{ev}}^{(3)}),\ ] ] where @xmath288 is the companion matrix of the polynomial @xmath289 .
let us find the explicit formula for the function @xmath153 as a function of the points @xmath282 and @xmath283 .
we introduce the @xmath290-matrix @xmath291 where @xmath292 denote by @xmath293 the matrix composed of the first @xmath154 columns of the matrix @xmath294 the entire rational function @xmath153 defining the operation @xmath295 has the form @xmath296 by a use of the formula and lemma [ even_part ] we can find @xmath297 similar to the condition of lemma [ even_part ] denote @xmath298 and @xmath299 .
one can easily show that @xmath300 .
thus , the product @xmath301 is an even function in @xmath40 . set @xmath302 therefore , @xmath303 , as a function on the curve @xmath43 , is the polynomial in @xmath39 and the parameters @xmath304 and @xmath305 .
the values of @xmath304 and @xmath305 are defined by the mapping @xmath306 according to , and @xmath307 .
namely , we have @xmath308 lemma [ even_part ] asserts that dividing the polynomial @xmath309 by @xmath310 gives the zero remainder and the quotient equal @xmath311 .
thus , the calculation is reduced to the classical algorithm of polynomial division .
[ zakon ] consider the space @xmath125 together with the mapping @xmath306 defined by as a groupoid over @xmath200 .
let @xmath312 and @xmath313 be the points from @xmath125 such that @xmath314 .
then the addition law has the form @xmath315 where the coordinates of the point @xmath316 are given by the formulas @xmath317^{-1}(h_1+r_2(c^{(3)})p_{\operatorname{ev}}^{(3)}).\end{gathered}\ ] ] [ genus1.1 ] let @xmath318 the family of curves @xmath43 is defined by the polynomial @xmath319 in the coordinates @xmath320 on @xmath321 and @xmath322 and @xmath323 the mapping @xmath324 is given by the formula @xmath325 let us write down the addition formulas for the points on the groupoid @xmath323 over @xmath321 .
set @xmath326 , @xmath327 and suppose @xmath328 .
let @xmath329 .
we have : @xmath330 @xmath331 , @xmath332 and so on .
@xmath333 thus , the function defining the operation @xmath295 has the expression @xmath334 whence , we find @xmath335 @xmath336 @xmath337 and , by , @xmath338 further , @xmath339 upon dividing the polynomial @xmath340 by the polynomial @xmath341 we find @xmath342 and , finally , we obtain the addition law of the elliptic groupoid in the following form @xmath343 one may check directly that @xmath344 let @xmath345 the family of curves @xmath43 is defined by the polynomial @xmath346 in the coordinates @xmath347 on @xmath348 and @xmath349 on @xmath350 where @xmath351 @xmath352 , and @xmath353 , the mapping @xmath354 is given by the formula @xmath355 let us write down the addition formulas for the points on the groupoid @xmath356 over @xmath348 . set @xmath357 , @xmath358 and suppose @xmath359 .
let @xmath360 @xmath361 we omit the calculation , which is carried out by the same scheme as for @xmath362 , and pass to the result .
set @xmath363 we have @xmath364 to shorten the formulas it is convenient to employ the linear differential operator @xmath365 it is important to note that @xmath366 adds unity to the weight , @xmath367 and that it is _ tangent _ to the singular set where the addition is not defined : @xmath368 let @xmath369 and @xmath370 .
note , that @xmath371 . using this notation the addition formulas
are written down as follows @xmath372
for each curve @xmath43 from the family consider the jacobi variety @xmath373 .
the set of all the jacobi varieties is _ the universal space @xmath374 of the jacobi varieties of the genus @xmath7 hyperelliptic curves_. the points of @xmath374 are pairs @xmath375 where the vector @xmath376 belongs to the jacobi variety of the curve with parameters @xmath42 .
the mapping @xmath377 that acts as @xmath378 makes @xmath374 the space over @xmath379 there is a natural groupoid over @xmath200 structure on @xmath374 . evidently , the mappings @xmath380 and @xmath381 satisfy the groupoid over @xmath200 axioms . [ [ addition - theorems - for - the - hyperelliptic - boldsymbolwp - functions ] ] addition theorems for the hyperelliptic @xmath382-functions ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ let us define the mapping @xmath383 over @xmath200 by putting into correspondence a point @xmath384 and the point @xmath385 , where @xmath386 here @xmath387 and @xmath388 is the hyperelliptic sigma - function @xcite . first , by abel theorem any triple of points @xmath390 that satisfies the condition @xmath391 corresponds to the set of zeros @xmath392 , @xmath393 , of an entire rational function of order @xmath72 on the curve @xmath43 .
namely , let @xmath207 , then @xmath394 ( for shortness , instead of indicating the end point of integration explicitly , we give only the first coordinate . )
second , for the given value @xmath395 the system of @xmath7 equations @xmath396 with respect to the unknowns @xmath397 is equivalent to the system of algebraic equations @xmath398 the roots of which are the required points @xmath397 ( see , for instance , @xcite ) .
above all note that @xmath59 hyperelliptic functions@xmath399 form a basis of the field of hyperelliptic abelian functions , i.e. , any function of the field can be expressed as a rational function in @xmath400 and @xmath401 .
the assertion of theorem [ additio_2 ] written down in the coordinates of @xmath125 takes the form of the addition theorem for the basis functions @xmath400 and @xmath401 .
one has @xmath7 functions @xmath405 and the functions are not abelian . however , by an application of abel theorem for the second kind integrals ( see @xcite ) one obtains the addition theorems for @xmath404-functions as well .
on one hand , any @xmath404-function can be represented as the sum of @xmath7 second kind integrals and an abelian function . on the other hand , an abelian sum of the second kind integrals with the end points at the set of zeros of an entire rational function @xmath137
is expressed rationally in terms of the coefficients of @xmath137 .
we employ the function computed in the variables indicated in corollary [ dura ] .
[ zeta_g ] let @xmath406 and @xmath407 then @xmath408 where @xmath409 is the rational function in @xmath410 and @xmath411 equal to the coefficient of the monomial of the weight @xmath412 in the function computed in the variables indicated in corollary [ dura ] .
we have the identity ( see @xcite,@xcite ) @xmath413 suppose that the closed path @xmath193 encloses all zeros @xmath414 of the function @xmath415 .
then we have @xmath416 because @xmath417 is a rational function on the curve and , hence , a uniform function , the total residue of @xmath418 on the riemann surface of the curve @xmath43 is zero . to write down this fact explicitly consider the parametrization @xmath419 of the curve @xmath43 near the point at infinity and denote @xmath420 .
we obtain @xmath421+\sum_{i=1}^{3 g } \mathrm{res}_{x = x_i}\!\bigg[\mathrm{d}\big(\log r_{3g}(x , y)\big)\int_{\infty}^{x}x^g\,\frac{\mathrm{d}x}{2y}\bigg]\!=0,\ ] ] which is in fact a particular case of abel theorem .
thus , the final expression is @xmath422.\ ] ] it remains to use the expansions @xmath423 to compute the residue . in the conditions of theorem [ zeta_g ] we have @xmath425 [ zeta_g-1 ] where @xmath426 and @xmath427 are the coefficients of the monomials of weight @xmath428 and @xmath429 in the function indicated in theorem [ zeta_g ] .
[ ex : g1 ] let @xmath362 .
the function @xmath430 has the form @xmath431 , where @xmath432 , cf .
example [ genus1.1 ] .
thus , theorem [ zeta_g ] gives the classic formula @xmath433 as @xmath434 and @xmath435 , cf .
example [ genus1.1 ] , theorem [ zeta_g-1 ] yields the relation @xmath436 which is the addition formula for weierstrass @xmath4-function . combining lemma [ p_gg ] with theorem [ zeta_g ] we find @xmath438 in the case @xmath362 due to the fact that @xmath434 formula gives the famous relation @xmath439 discovered by frobenius and stickelberger .
let us pass to the case @xmath440 , we have @xmath441 . note that @xmath442 .
the coefficient @xmath409 is expressed as follows , cf .
example [ genus1.1 ] , @xmath443 and the coefficient @xmath426 , respectively , @xmath444 we come to the relations @xmath445 hence , by eliminating @xmath409 and @xmath426 , we obtain the identity @xmath446 where @xmath447 and @xmath448 , provided @xmath407 [ [ trilinear - addition - theorems - for - the - hyperelliptic - sigma - functions ] ] trilinear addition theorems for the hyperelliptic @xmath449-functions ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ let us multiply the left hand side of by the product @xmath452 , then becomes the trilinear relation @xmath453\sigma(u)\sigma(v)\sigma(w)\big|_{u+v+w=0}=0,\end{gathered}\ ] ] which is satisfied by the genus @xmath8 sigma - function .
it is important to notice that the elliptic identity is equivalent to the trilinear addition law @xmath454\sigma(u)\sigma(v)\sigma(w)\big|_{u+v+w=0}=0,\end{gathered}\ ] ] which is satisfied by weierstrass sigma - function .
let us denote @xmath455 and @xmath456 .
the functions @xmath457 where @xmath409 and @xmath426 are given in example [ ex : g1 ] , vanish on the plane @xmath391 .
moreover , one can show that for any @xmath458 there exist unique polynomials @xmath459 $ ] such that @xmath460 and at least one of the polynomials @xmath461 is nontrivial .
for the hyperelliptic sigma - function of an arbitrary genus @xmath7 we propose the following hypothesis .
let @xmath462 $ ] .
consider the ring @xmath463 $ ] as a graded ring of linear differential operators .
we conjecture that there exists a collection of @xmath72 linear operators @xmath464 @xmath465 where @xmath466 , such that @xmath467 where @xmath468 and @xmath415 is the function computed in the variables indicated in corollary [ dura ] .
thus , @xmath7 operators @xmath469 @xmath47 define the trilinear relations @xmath470 note , that the assertions of theorem [ zeta_g ] , lemma [ p_gg ] , and theorem [ zeta_g-1 ] imply the relations @xmath471 buchstaber v m , enolskii v z , leykin d v , hyperelliptic kleinian functions and applications , in solitons , geometry and topology : on the crossroad , editors : buchstaber v m and novikov s p , advances in math .
sciences , ams translations , series 2 , vol .
179 , 1997 , 134 . | we construct an explicit form of the addition law for hyperelliptic abelian vector functions @xmath0 and @xmath1 . the functions @xmath0 and @xmath1 form a basis in the field of hyperelliptic abelian functions , i.e.
, any function from the field can be expressed as a rational function of @xmath0 and @xmath1 . |
disk galaxy rotation curves ; strong gravitational lensing by galaxies and clusters ; the dynamics of stars in elliptical galaxies and of gas and galaxies in clusters ; these are just examples of the various luminous tracers that probe the inner structure of dark matter halos .
such observations place strong constraints on the distribution of dark matter in these highly non - linear regions that may be contrasted directly with theoretical predictions for halo structure . given the sensitivity of such predictions to the nature of the dark matter , these observational constraints constitute provide tests that may question or even rule out particular models of dark matter
. robust prediction of the inner structure of cold dark matter ( cdm ) halos is a quintessential n - body problem , albeit one of considerable complexity due to the large overdensities and , consequently , the short crossing times involved .
indeed , only recently have computational capabilities improved to the point of allowing realistic simulation of the regions which house the luminous components of individual galaxies .
this work builds upon the pioneering efforts of frenk et al ( 1988 ) , dubinski and carlberg ( 1991 ) , and crone et al ( 1994 ) , among others , which led to the identification of a number of key features of the structure of dark matter halos assembled by hierarchical clustering .
one important result of this early work concerns the absence of a well defined central `` core '' of constant density in virialized cdm halos . in this sense ,
dark matter halos are `` cuspy '' : the dark matter density increases apparently without bounds toward the center of the halo . a second result concerns the remarkable similarity ( `` universality '' ) in the structure of dark matter halos of widely different mass .
this was first proposed by navarro , frenk & white ( 1996 , 1997 ; hereafter nfw ) , who suggested a simple fitting formula to describe the spherically - averaged density profiles of dark matter halos , @xmath11 where @xmath12 and @xmath13 are a characteristic density and radius , respectively .
the larger the mass of a halo , the lower its characteristic density , reflecting the lower density of the universe at the ( later ) assembly time of more massive systems . further simulation work of similar numerical resolution ( see , e.g. , cole & lacey 1996 , huss et al 1999 ) provided support for the nfw conclusions , although small but systematic differences began to emerge as the numerical resolution of the simulations improved ( see , e.g. , moore et al 1999 , hereafter m99 , ghigna et al 2000 , fukushige & makino 1997 , 2001 , 2003 ) .
these authors reported deviations from eq .
[ eq : nfw ] that increase systematically inward , and thus are particularly noticeable in high - resolution simulations .
in particular , fukushige & makino ( 2001 ) reported that nfw fits tend to underestimate the dark matter density within the scale radius @xmath13 .
m99 reached a similar conclusion and interpreted this result as indicating a density cusp steeper than that of the nfw profile .
these authors preferred a modified fitting function which diverges as @xmath9 near the center , @xmath14 one should note , however , that there is no consensus amongst n - body practitioners for such modified profile ( see , e.g. , klypin et al 2001 and power et al 2003 , hereafter p03 ) , and that there is some work in the literature suggesting that the central density cusp might actually be _ shallower _ than @xmath15 ( subramanian et al 2000 ; taylor & navarro 2001 ; ricotti 2003 ) .
this unsettled state of affairs illustrates the difficulties associated with simulating the innermost structure of cdm halos in a reliable and reproducible manner .
the high density of dark matter in such regions demands large numbers of particles and fine time resolution , pushing to the limit even the largest supercomputers available at present . as a result ,
many of the studies mentioned above are either of inadequate resolution to be conclusive or are based on results from a handful of simulations where computational cost precludes a systematic assessment of numerical convergence . establishing the detailed properties of the central density cusp , as well as deriving the value of its asymptotic central slope , are important for a number of reasons .
for example , steeper cusps place larger amounts of dark matter at the center , exacerbating the disagreement with observations that suggest the presence of a constant density core in low surface brightness galaxies or in strongly barred galaxies ( moore 1994 , flores & primack 1994 , mcgaugh & de blok 1998 , debattista and sellwood 1998 , van den bosch et al 2000 ) .
steep cusps would also be important for direct detection experiments for dark matter , as a possible gamma - ray annihilation signal of wimps at the galactic center would be particularly strong for @xmath9 cusps ( calcneo - roldn & moore 2000 ; taylor & silk 2003 ; stoehr et al 2002 ) . finally , the detailed structure of the central cusp is not the only focus of contrasting claims in the literature .
for example , the `` universality '' of cdm halo structure has been questioned by jing & suto ( 2000 ) , who find that the slope of the density profile at a fixed fraction of the virial radius steepens towards lower halo masses .
klypin et al ( 2001 ) , on the other hand , point out that such a systematic trend is entirely consistent with universality as originally claimed by nfw , and just reflects the mass dependence of halo characteristic density .
we address these conflicting issues here using a suite of nineteen high - resolution simulations of the formation of halos in the standard @xmath0cdm cosmogony .
halo masses are chosen in three main groups : `` dwarf '' halos with @xmath16 , `` galaxy '' halos with @xmath17 and `` cluster '' halos of mass @xmath18 .
this allows us to gain insight into the effects of cosmic variance at each mass scale , as well as to explore the mass dependence of the structure of @xmath0cdm halos .
we define the mass of a halo to be that contained within its virial radius , that is , within a sphere of mean density contrast @xmath19 .
we express the present value of hubble s constant as @xmath20 km s@xmath21 mpc@xmath21 ] .
this paper is organized as follows .
section [ sec : numexp ] describes briefly the numerical simulations ;
[ sec : res ] discusses our main results ; and we summarize our conclusions in [ sec : conc ] .
the numerical set up of our simulations follows closely the procedure described by p03 , where the interested reader may find a thorough discussion of our initial conditions generating scheme , the choice of n - body codes and integrators , as well as the criteria adopted to optimize the choice of the numerical parameters of the simulations . for completeness , we include here a brief discussion of the main numerical issues , but refer the reader to p03 for a more detailed discussion .
the simulations reported in this paper have been performed using two parallel n - body codes : gadget , written by volker springel ( springel , yoshida & white 2001 ) , and pkdgrav , written by joachim stadel and thomas quinn ( stadel 2001 ) . as discussed in p03 ,
both codes give approximately the same results for appropriate choices of numerical parameters , and neither code seems obviously to outperform the other when similar numerical convergence criteria are met .
we adopt a flat , @xmath23 @xmath0cdm cosmological model whose dynamics is dominated at present by a cosmological constant , @xmath24 .
the matter power spectrum is normalized so that the present linear rms amplitude of mass fluctuations in spheres of radius @xmath25 mpc is @xmath26 .
we assume a linear fluctuation power spectrum given by the product of the square of the appropriate cdm transfer function , @xmath27 , and a harrison - zeldovich primordial power spectrum ( i.e. @xmath28 ) .
the halo samples were drawn from three different @xmath0cdm cosmological `` parent '' simulations .
table [ tab : numexp ] lists the main numerical parameters of each of these simulations : @xmath29 is the size of the cosmological box , @xmath30 is the initial redshift , @xmath31 is the particle mass , and @xmath32 is the softening parameter , assumed fixed in comoving coordinates .
the dwarf , cluster , and most of the galaxy halos , were extracted from the simulations @xmath0cdm-512 ( yoshida , sheth & diaferio 2001 ) and sgif-128 .
these two parent simulations , both carried out within the virgo consortium , used the cdm transfer function given by cmbfast ( seljak & zaldarriaga 1996 ) , assuming @xmath33 and @xmath34 .
this transfer function is well fit by the bardeen et al ( 1986 ) fitting formula with a value of @xmath35 for the shape parameter @xmath36 .
three of the galaxy halos ( g1-g3 , see table [ tab : halpar ] ) were extracted from the parent simulation described by eke , navarro & steinmetz ( 2001 , labeled ens01 in table [ tab : numexp ] ) . that simulation used the bardeen et al ( 1986 ) fitting formula for the cdm transfer function , with @xmath37 and @xmath38 .
since completing the numerical convergence tests reported in p03 , we have developed a more flexible and powerful set of codes for setting up the resimulation initial conditions .
this resimulation software enables us to iterate the procedure in order to `` resimulate a resimulation '' , an important step for setting up appropriate initial conditions for dwarf halos .
the basic methodology employed is very similar to the methods described in p03 , with just a few minor differences .
galaxy halos g1-g3 were selected from the ens01 simulation and their initial conditions were created using the software described in p03 .
all of the other halos were set up with the new codes , following the procedure we describe below .
the first stage is to carry out , up to the redshift of interest ( typically @xmath39 ) , a `` parent '' simulation of a large , representative volume of a @xmath0cdm universe .
these parent simulations are used to select halos targeted for resimulation at higher resolution . once a halo has been selected for resimulation at @xmath39 , we trace all particles within a sphere of radius @xmath40 to the @xmath41 `` unperturbed '' configuration .
we then create a set of initial conditions with much higher mass resolution in the volume occupied by the halo particles , and resample the remainder of the periodic box at lower resolution , taking care to retain sufficient resolution in the regions surrounding the halo of interest so that external tidal forces acting on the high - resolution region are adequately represented .
the procedure involves two main steps .
firstly , we set up a uniform multi - mass distribution of particles to approximate the particle positions in the high - resolution region at @xmath41 .
this is accomplished by arranging particles either in a cubic grid or as a `` glass '' , within a cube just big enough to contain the region of interest .
either choice approximates a uniform mass distribution very accurately .
outside the cube we lay down particles on a set of concentric cubic shells , centered on the cube , which extend outwards until they fill the entire periodic volume of the parent simulation .
these concentric shells are filled with more massive particles whose interparticle separation increases approximately linearly with distance from the high resolution region . unlike the grid or glass , this arrangement does not reproduce a perfectly uniform mass distribution .
however , by populating each shell with regularly spaced particles , we obtain a configuration which is uniform enough for our purposes . in the interest of efficiency , we replace those particles in the high - resolution cube that do not end up in the selected halo with more massive particles made by combining several high - resolution ones . this procedure , particularly for the dwarf halo resimulations , significantly reduces the number of particles in the initial conditions and the run time of the subsequent simulation .
thus , each halo forms from an `` amoeba - shaped '' region consisting only of the highest resolution particles in the hierarchy . we have explicitly checked that the resampling procedure adds no extra power ; in tests , the multi - mass particle distribution remains very close to uniform over an expansion factor of up to @xmath42 .
once a multi - mass but uniform mass distribution has been created , the next step is to to add the appropriate gaussian density fluctuations .
this is done by assigning a displacement and a peculiar velocity to each particle using fourier methods . by using the same amplitude and phase for every fourier mode present in the parent simulation
, a perturbed density field essentially identical to that of the parent simulation can be reproduced . in the high - resolution cube , because the particle mass is smaller than in the parent simulation , it is necessary to add additional short wavelength modes ( with amplitudes fixed by the adopted power spectrum ) down to the nyquist wavelength of the new particle grid . to ensure that the fourier transforms needed to add this extra power are of a manageable size
, we make the additional power periodic on the scale of the central cube rather than on the scale of the parent simulation .
the longest wavelength added is typically smaller than one tenth of the side length of the original cube . as described in p03 ,
the individual components of the displacement field are generated in turn , and the displacements calculated at the particle positions by trilinear interpolation .
to set up growing modes , we use the zeldovich approximation and make the peculiar velocities proportional to the displacements .
the initial redshift , @xmath30 , of each resimulation is chosen so that density fluctuations in the high - resolution region are in the linear regime .
p03 find that convergent results are obtained when @xmath30 is high enough that the ( theoretical ) rms mass fluctuation on the smallest resolved mass scale , @xmath43 does not exceed @xmath44 ( where @xmath31 is the mass of a high - resolution particle ) .
all of our simulations satisfy this criterion .
the resimulated halos analyzed in this paper were all identified in the parent simulations by applying the friends - of - friends ( fof ) group finding algorithm ( davis et al 1985 ) with a linking length @xmath45 .
cluster - sized halos were drawn from a @xmath46 mpc simulation volume ( @xmath0cdm-512 in table [ tab : numexp ] ) .
the fof(@xmath47 ) groups were first ordered by mass and then ten consecutive entries on the list centered around a mass of @xmath48 were selected .
galaxy halos were likewise drawn from a @xmath49 mpc volume ( sgif-128 ) , with the exception of three of the halos ( g1-g3 ) which were selected from a @xmath50 mpc volume ( ens01 ) .
target dwarf halos were also found in the sgif-128 simulation .
however , because of their extremely low mass ( corresponding to 5 - 6 particles in sgif-128 ) , it was necessary to create a second `` parent '' simulation for them by resimulating a region of the sgif-128 volume at significantly higher resolution . to this end , a spherical region of radius @xmath51 mpc , with mean density close to the universal average , was selected at random within the @xmath52 mpc box .
this spherical region was then resimulated with roughly one hundred times more particles than in sgif-128 .
the target dwarf halos were identified within this spherical volume again from an fof(0.2 ) group list .
a total of eighteen halos with 450 - 550 particles ( corresponding to masses of @xmath53-@xmath54 ) were chosen .
we report results on the four halos in this list that have been resimulated to date .
high resolution initial conditions for these dwarf halos were created in an identical way to the more massive galaxy and cluster halos .
numerical parameters were chosen to ensure that all halos , regardless of mass , were resimulated at comparable mass resolution ( typically over @xmath55 particles within the virial radius at @xmath39 , see table [ tab : halpar ] ) .
we focus our analysis on the spherically - averaged mass profile of simulated halos at @xmath39 .
this is measured by sorting particles in distance from the center of each halo and arranging them in bins of equal logarithmic width in radius .
density profiles , @xmath56 , are computed simply by dividing the mass in each bin by its volume .
the cumulative mass within each bin , @xmath57 , is then used to compute the circular velocity profile of each halo , @xmath58 , as well as the cumulative density profile , @xmath59 , which we shall use in our analysis .
the center of each halo is determined using an iterative technique in which the center of mass of particles within a shrinking sphere is computed recursively until a few thousand particles are left ( see p03 for details ) . in a multi - component system , such as a dark halo with substructure
, this centering procedure isolates the densest region within the largest subcomponent . in more regular systems ,
the center so obtained approximately coincides with the centers defined by the center of mass weighted by the local density or by the gravitational potential of each particle .
we note that , unlike in nfw , no attempt has been made to select halos at a particularly quiet stage in their dynamical evolution ; our sample thus contains halos in equilibrium as well as a few with prominent substructure as a result of recent accretion events .
the analysis presented in p03 demonstrated that the mass profile of a simulated halo is numerically robust down to a `` convergence radius '' , @xmath60 , that depends primarily on the number of particles and time steps , as well as on the choice of gravitational softening in the simulation .
each of these choices imposes a minimum radius for convergence , although for an `` optimal '' choice of parameters ( i.e. , one that , for given @xmath60 , minimizes the number of force computations and time steps ) the most stringent criterion is that imposed by the number of particles within @xmath1 . in this optimal case ,
the minimum resolved radius is well approximated by the location at which the two - body relaxation time , @xmath61 , equals the age of the universe ( see hayashi et al 2003 for further validation of these criteria , but see binney 2003 for a different opinion ) . to be precise , we shall identify @xmath60 with the radius where @xmath62 equals the circular orbital timescale at the virial radius , @xmath63 .
thus , @xmath64 is defined by the following equation , @xmath65 here @xmath66 is the number of particles enclosed within @xmath67 , and @xmath68 . with this definition
, the convergence radius in our best - resolved halos , outside which @xmath69 converges to better than @xmath6 , is of order @xmath70 .
the top panels of figure [ figs : rhoprof ] show the density profiles , @xmath56 , of the nineteen simulated halos in our sample . in physical units , the profiles split naturally into three groups : from left to right , `` dwarf '' ( dotted ) , `` galaxy '' ( dashed ) , and `` cluster '' ( dot - dashed ) halos , respectively .
each profile is shown from the virial radius , @xmath1 , down to the innermost converged radius , @xmath60 ; a convention that we shall adopt in all figures throughout this paper .
the thick solid lines in the top - left panel show the nfw profiles ( eq . [ eq : nfw ] ) expected for halos in each group , with parameters chosen according to the prescription of eke , navarro & steinmetz ( 2001 ) .
note that these nfw curves are _ not _ best fits to any of the simulations , but that they still capture reasonably well the shape and normalization of the density profiles of the simulated halos .
the top right panel of figure [ figs : rhoprof ] is similar to the top left one , but the comparison is made here with the modified form of the nfw profile proposed by m99 ( eq . [ eq : mooreetal ] ) .
there is no published prescription specifying how to compute the numerical parameters of this formula for halos of given mass , so the three profiles shown in this panel are just `` eyeball '' fits to one halo in each group . like the nfw profile
, the m99 formula also appears to describe reasonably well the gently - curving density profiles of @xmath0cdm halos . figure [ figs : rhoprof ] thus confirms a number of important trends that were already evident in prior simulation work .
* @xmath0cdm halo density profiles deviate significantly from simple power laws , and steepen systematically from the center outwards ; they are shallower than isothermal near the center and steeper than isothermal near the virial radius .
* there is no indication of a well defined central `` core '' of constant density ; the dark matter density keeps increasing all the way in , down to the innermost resolved radius . * simple formulae such as the nfw profile ( eq . [ eq : nfw ] ) or the m99 formula ( eq . [ eq : mooreetal ] )
appear to describe the mass profile of all halos reasonably well , irrespective of mass , signaling a `` universal '' profile shape .
properly scaled , a dwarf galaxy halo is almost indistinguishable from a galaxy cluster halo .
we elaborate further on each of these conclusions in what follows .
are the density profiles of @xmath0cdm halos described better by the nfw formula ( eq . [ eq : nfw ] ) or by the modification proposed by m99 ( eq . [ eq : mooreetal ] ) ?
the answer may be seen in the bottom panels of figure [ figs : rhoprof ] .
these panels show the deviations ( simulation minus fit ) from the _ best _ fits to the density profiles of each halo using the nfw profile or the m99 profile .
these fits are obtained by straightforward @xmath71 minimization , assigning equal weight to each radial bin .
this is done because the statistical ( poisson ) uncertainty in the determination of the density within each bin is negligible ( each bin contains from several thousand to several hundred thousand particles ) so the remaining uncertainties are likely to be dominated by systematics , such as the presence of substructure , varying asphericity , as well as numerical error , whose radial dependence is difficult to assess quantitatively ( see p03 ) . as shown in the bottom panels of figure [ figs : rhoprof ] ,
there is significant variation in the shape of the density profile from one halo to another .
some systems are fit better by eq .
[ eq : nfw ] than by eq .
[ eq : mooreetal ] , and the reverse is true in other cases . over the radial range resolved by the simulations , @xmath56 deviates from the best fits by less than @xmath72 .
nfw fits tend to _ underestimate _ the density in the inner regions of most halos ; by up to @xmath73 at the innermost resolved point .
m99 fits , on the other hand , seem to do better for low mass halos , but tend to _ overestimate _ the density in the inner regions of cluster halos by up to @xmath74 .
we have explicitly checked that these conclusions are robust to reasonable variation in the binning used to construct the density profiles , as well as in the adopted minimization procedure .
this level of accuracy may suffice for a number of observational applications , with the proviso that comparisons are restricted to radii where numerical simulations are reliable ; i.e. , @xmath75 .
deviations from the best fits increase systematically towards the center , so it is likely that extrapolations of either fitting formula to radii much smaller than @xmath60 will incur substantial error .
we discuss below ( [ ssec : newprof ] ) possible modifications to the fitting formulae that may minimize the error introduced by these extrapolations .
many observations , such as disk galaxy rotation curves or strong gravitational lensing , are better probes of the _ cumulative _ mass distribution than of the differential density profile shown in figure [ figs : rhoprof ] . since cumulative profiles are subject to different uncertainties than differential ones , it is important to verify that our conclusions regarding the suitability of the nfw or m99 fitting formulae are also applicable to the cumulative mass distribution of @xmath0cdm halos .
the radial dependence of the spherically - averaged circular velocity profile of all halos in our series is shown in figure [ figs : vcprof ] .
as in figure [ figs : rhoprof ] , the thick solid curves in the top left ( right ) panel are meant to illustrate a typical nfw ( m99 ) profile corresponding to dwarf , galaxy , and cluster halos , respectively .
the bottom left and right panels show deviations from the _ best _ fit to each halo using the nfw or m99 profile , respectively . both profiles
reproduce the cumulative mass profile of the simulated halos reasonably well .
the largest deviations seen are for the m99 fits , but they do not exceed @xmath76 over the radial range resolved in the simulations .
nfw fits fare better , with deviations that do not exceed @xmath6 . as with the density profiles , the deviations between simulation and fits ,
although small , increase toward the center , suggesting that caution should be exercised when extrapolating these fitting formulae beyond the spatial region where they have been validated .
this is important because observational data , such as disk galaxy rotation curves , often extend to regions inside the minimum convergence radius in these simulations .
we have noted in the previous subsections that systematic deviations are noticeable in both nfw and m99 fits to the mass profiles of simulated @xmath0cdm halos .
nfw fits tend to underestimate the dark matter density near the center , whilst m99 fits tend to overestimate the circular velocity in the inner regions .
the reason for this is that _ neither _ fitting formula fully captures the radial dependence of the density profile .
we explore this in figure [ figs : slopes ] , which shows the logarithmic slope , @xmath77 , of all simulated halos , as a function of radius .
although there is substantial scatter from halo to halo , a number of trends are robustly defined .
the first trend to note is that halo density profiles become shallower inward down to the innermost resolved radius , @xmath60 ( the smallest radius plotted in figure [ figs : slopes ] ) .
_ we see no indication for convergence to a well defined asymptotic value of the inner slope in our simulated halos _ , neither to the @xmath78 expected for the nfw profile ( solid curves in figure [ figs : slopes ] ) nor to the @xmath79 expected in the case of m99 ( dotted curves in same figure ) .
the second trend is that the radial dependence of the logarithmic slope deviates from what is expected from either the nfw or the m99 fitting formulae . near @xmath60 ,
the slopes are significantly shallower than @xmath79 ( and thus in disagreement with the m99 formula ) but they are also significantly steeper than expected from nfw fits . in quantitative terms , let us consider the slope well inside the characteristic radius , @xmath5 ( where the slope takes the `` isothermal '' value , as well as the density at that radius , @xmath80 , can be measured directly from the simulations , without reference to or need for any particular fitting formula .
for the nfw profile , @xmath5 is equivalent to the scale radius @xmath13 ( see eq .
[ eq : nfw ] ) .
the density at @xmath5 is related to the nfw characteristic density , @xmath12 , by @xmath81 .
] of @xmath82 ) . for cluster halos , for example
, at @xmath83 ( @xmath84 kpc ) the average slope is approximately @xmath85 , whereas the nfw formula predicts @xmath86 and m99 predicts @xmath87 .
this is in agreement with the latest results of fukushige , kawai & makino ( 2003 ) , who also report profiles shallower than @xmath9 at the innermost converged radius of their simulations .
a best - fit slope of @xmath88 was also reported by moore et al ( 2001 ) for a dwarf galaxy halo ( of mass similar to the draco dwarf spheroidal ) , although that simulation was stopped at @xmath89 , and might therefore not be directly comparable to the results we present here .
this discrepancy in the radial dependence of the logarithmic slope between simulations and fitting formulae is at the root of the different interpretations of the structure of the central density cusp proposed in the literature .
for example , because profiles become shallower inward more gradually than in the nfw formula , modifications with more steeply divergent cusps ( such as eq .
[ eq : mooreetal ] ) tend to fit density profiles ( but not circular velocity profiles ) better in the region interior to @xmath5 .
this is _ not _ , however , a sure indication of a steeper cusp .
indeed , _ any _ modification to the nfw profile that results in a more gradual change in the slope inside @xmath5 will lead to improved fits , _ regardless _ of the value of the asymptotic central slope .
we show this explicitly below in [ ssec : newprof ] .
conclusive proof that the central density can not diverge as steeply as @xmath79 is provided by the total mass inside the innermost resolved radius , @xmath60 .
this is because , at any radius @xmath67 , the mean density , @xmath90 , together with the local density , @xmath56 , provide a robust upper limit to the asymptotic inner slope .
this is given by @xmath91 , under the plausible assumption that @xmath92 is monotonic with radius .
figure [ figs : betaprof ] shows @xmath93 as a function of radius ; clearly , except for possibly one dwarf system , no simulated halo has enough dark mass within @xmath60 to support cusps as steep as @xmath9 .
the nfw asymptotic slope , corresponding to @xmath94 , is still consistent with the simulation data , but the actual central value of the slope may very well be shallower . we emphasize again that there is no indication for convergence to a well defined value of @xmath95 : density profiles become shallower inward down to the smallest resolved radius in the simulations .
figure [ figs : slopes ] shows also that there is a well - defined trend with mass in the slope of the density profile measured at @xmath96 to @xmath97 ( the innermost point plotted for each profile ) : @xmath98 for clusters , @xmath99 for galaxies , and @xmath100 for dwarfs .
a similar trend was noted by jing & suto ( 2000 ) , who used it to argue against a `` universal '' density profile shape .
as discussed by klypin et al ( 2001 ) , this is just a reflection of the trend between the concentration of a halo and its mass .
it does _ not _ indicate any departure from similarity in the profile shape .
indeed , one does _ not _ expect the profiles of halos of widely different mass , such as those in our series , to have similar slopes at a constant fraction of the virial radius .
rather , if the density profiles are truly self - similar , slopes ought to coincide at fixed fractions of a _ mass - independent _ radial scale , such as @xmath5 . figure [ figs : rhoprofsc]a shows the striking similarity between the structure of halos of different mass when all density profiles are scaled to @xmath5 and @xmath101 .
the density profile of a dwarf galaxy halo then differs very little from that of a galaxy cluster @xmath102 times more massive .
this demonstrates that spherically - averaged density profiles are approximately `` universal '' in shape ; rarely do individual density profiles deviate from the scaled average by more than @xmath72 . in the scaled units of figure [ figs : rhoprofsc ] , the nfw and m99 profiles
are fixed , and are shown as solid and dotted curves , respectively . with this scaling ,
differences between density profiles are more evident than when best fits are compared , since the latter by definition minimize the deviations . in figure [ figs : rhoprofsc]a , for example , it is easier to recognize the `` excess '' of dark mass inside @xmath5 relative to the nfw profile that authors such as m99 and fukushige & makino ( 1997 , 2001 , 2003 ) have ( erroneously ) interpreted as implying a steeply divergent density cusp .
the similarity in mass profile shapes is also clear in figure [ figs : rhoprofsc]b , which shows the circular velocity curves of all halos in our series , scaled to the maximum , @xmath103 and to the radius where it is reached , @xmath104 .
nfw and m99 are again fixed curves in these scaled units .
this comparison is more relevant to observational interpretation , since rotation curve , stellar dynamical , and lensing tracers are all more directly related to @xmath69 than to @xmath56 .
because of the reduced dynamic range of the y - axis , the scatter in mass profiles from halo to halo is more clearly apparent in the @xmath105 profiles ; the nfw and m99 profiles appear to approximately bracket the extremes in the mass profile shapes of simulated halos .
we discuss below a simple fitting formula that , with the aid of an extra parameter , is able to account for the variety of mass profile shapes better that either the nfw or m99 formula .
although the discussion in the previous subsections has concentrated on global deviations from simple fitting formulae such as nfw or m99 , it is important to emphasize again that such deviations , although significant , are actually rather small .
as shown in figure [ figs : vcprof ] , best nfw fits reproduce the circular velocity profiles to an accuracy of better than @xmath106 down to roughly @xmath107 of @xmath1 .
although this level of accuracy may suffice for some observational applications , the fact that deviations increase inward and are maximal at the innermost converged point suggests the desirability of a new fitting formula better suited for extrapolation to regions beyond those probed reliably by simulations .
an improved fitting formula ought to reproduce : ( i ) the more gradual shallowing of the density profile towards the center ; ( ii ) the apparent lack of evidence for convergence to a well - defined central power - law ; and ( iii ) the significant scatter in profile shape from halo to halo .
after some experimentation , we have found that a density profile where @xmath108 is a power - law of radius is a reasonable compromise that satisfies these constraints whilst retaining simplicity ; @xmath109 which corresponds to a density profile of the form , @xmath110 .
\label{eq : newprof}\ ] ] this profile has finite total mass ( the density cuts off exponentially at large radius ) and has a logarithmic slope that decreases inward more gradually than the nfw or m99 profile .
the thick dot - dashed curves in figures [ figs : slopes ] and [ figs : betaprof ] show that eq .
[ eq : newprof ] ( with @xmath111 ) does indeed reproduce fairly well the radial dependence of @xmath108 and @xmath112 in simulated halos .
furthermore , adjusting the parameter @xmath113 allows the profile to be tailored to each individual halo , resulting in improved fits . indeed , as shown in figure [ figs : rhoprof_alpha ] , eq .
[ eq : newprof ] reproduces the density profile of individual halos to better than @xmath106 over the reliably resolved radial range , and that there is no discernible radial trend in the residuals .
this is a significant improvement over nfw or m99 fits , where the maximum deviations were found at the innermost resolved radius .
the best - fit values of @xmath113 ( in the range @xmath114 - @xmath115 ) show no obvious dependence on halo mass , and are listed in table [ tab : fitpar ] .
the average @xmath113 is @xmath116 and the dispersion about the mean is @xmath117 .
we note that the @xmath118 profile is not formally divergent , and converges to a finite density at the center , @xmath119 ( for @xmath120 ) .
it is unclear at this point whether such asymptotic behavior is a true property of @xmath0cdm halos or simply an artifact of the fitting formula that results from choosing @xmath121 in eq .
[ eq : newbeta ] .
the simulations show no evidence for convergence to a well - defined central value for the density , but even in the best - resolved cases they only probe regions where densities do not exceed @xmath122 .
this is , for @xmath113 in the range @xmath114 - @xmath115 , several orders of magnitude below the maximum theoretical limit in eq .
[ eq : newprof ] .
we note as well that the convergence to @xmath121 is quite slow for the values of @xmath113 favored by our fits .
indeed , for @xmath123 , the logarithmic slope only reaches a value significantly shallower than the nfw asymptotic slope at radii that are well inside the convergence radius of our simulations ; for example , @xmath124 only reaches @xmath125 at @xmath126 , corresponding to @xmath127 pc for galaxy - sized halos .
this implies that the @xmath118 profile is in practice `` cuspy '' for most astrophysical applications . establishing conclusively
whether @xmath0cdm halos actually have divergent inner density cusps is a task that awaits simulations with much improved resolution than those presented here .
figure [ figs : profcomp ] compares the density and circular velocity profiles implied by the @xmath118 formula ( eq . [ eq : newprof ] ) with the nfw and m99 profiles ( left panels ) , as well as with the fitting formula proposed by stoehr et al ( 2002 , hereafter swts ) to describe the structure of _ substructure _ halos ( right panels ) .
the top left panel of figure [ figs : profcomp ] shows that , despite its finite central density , the @xmath118 profile can approximate fairly well both an nfw profile ( for @xmath128 ) and an m99 profile ( for @xmath129 ) for over three decades in radius .
the circular velocity profile for @xmath130 is likewise quite similar to nfw s ( bottom left panel of figure [ figs : profcomp ] ) , but the similarity to the shape of the m99 @xmath105 profile is less for all values of @xmath113 .
interestingly , the @xmath105 profiles corresponding to @xmath118 resemble parabolae in a @xmath131-@xmath131 plot , and thus may be used to approximate as well the mass profiles of substructure halos , as discussed by swts .
this is demonstrated in the bottom right panel of figure [ figs : profcomp ] , where we show that the @xmath105 profiles corresponding to @xmath123 , @xmath115 , and @xmath132 , are very well approximated by the swts formula , @xmath133 ^ 2 , \label{eq : swts}\ ] ] for @xmath134 , @xmath35 , and @xmath135 , respectively .
the latter value ( @xmath136 , or @xmath137 ) corresponds to the median of the best swts fits to the mass profile of substructure halos .
note that this is quite different from the @xmath129 - @xmath115 required to fit isolated @xmath0cdm halos ( see table [ tab : fitpar ] ) .
it might actually be preferable to adopt the @xmath118 profile rather than the swts formula for describing substructure halos , since @xmath138 is monotonic with radius and extends over all space .
this is not the case for swts , as shown in the top - right panel of figure [ figs : profcomp ] .
the swts density profiles are `` hollow '' ( i.e. , the density has a minimum at the center ) , and extend out to a maximum radius , given by @xmath139 .
this is because the circular velocity in the outer regions of the swts formula fall off faster than keplerian , and therefore the corresponding density becomes formally negative at a finite radius .
the @xmath118 profile thus appears versatile enough to reproduce , with a single fitting parameter , the structure of @xmath0cdm halos and of their substructure . since @xmath118 captures the inner slopes better than either the nfw or m99 profile , it is also likely to be a safer choice should extrapolation of the mass profile beyond the converged radius prove necessary .
we end by emphasizing , however , that all simple fitting formulae have shortcomings , and that _ direct comparison with simulations rather than with fitting formulae should be attempted whenever possible_. the application of fitting formulae such as the one described above requires a procedure for calculating the characteristic scaling parameters for a given halo mass , once the power spectrum and cosmological parameters are specified .
nfw developed a simple procedure for calculating the parameters corresponding to halos of a given mass . because of the close relationship between the scale radius , @xmath13 , and characteristic density , @xmath12 , of the nfw profile and the @xmath5 and @xmath140 parameters of eq .
[ eq : newprof ] , we can use the formalism developed by nfw to compute the expected values of these parameters in a given cosmological model .
nfw interpreted the characteristic density of a halo as reflecting the density of the universe at a suitably defined time of collapse .
their formalism assigns to each halo of mass @xmath141 ( identified at @xmath39 ) a collapse redshift , @xmath142 defined as the epoch when half the mass of the halo was first contained in progenitors more massive than a certain fraction @xmath143 of the final mass . with this definition , and once @xmath143 has been chosen , @xmath144 can be computed using the press - schechter theory ( e.g. , lacey & cole 1993 ) .
the nfw model then assumes that the characteristic density of a halo ( i.e. , @xmath12 in eq . [ eq : nfw ] ) is proportional to the mean density of the universe at @xmath145 .
the redshift dependence of the characteristic density was first probed in detail by bullock et al ( 2001 , hereafter b01 ) , who proposed a modification to nfw s model in which , for a given halo mass , the scale radius , @xmath13 , remains approximately constant with redshift .
eke , navarro & steinmetz ( 2001 , hereafter ens ) , on the other hand , argued that the characteristic density of a halo is determined by the _ amplitude and shape _ of the power spectrum , as well as by the universal expansion history .
their formalism reproduces nicely the original results of nfw as well as the redshift dependence pointed out by b01 , and is applicable to more general forms of the power spectrum , including the `` truncated '' power spectra expected in scenarios such as warm dark matter ( see ens for more details ) .
we have used the ens and b01 formalisms to predict the halo mass dependence of the scaling parameters , @xmath140 and @xmath5 , and we compare the results with our simulations in figure [ figs : rm2rhom2 ] .
the ens prediction is shown by the solid line whereas the dotted line shows that of b01 .
both formalisms reproduce reasonably well the trend seen in the simulations , so that one can use either , in conjunction with eq .
[ eq : newprof ] ( with @xmath113 in the range 0.1 - 0.2 ) , to predict the structure of a @xmath0cdm halo .
a simple code that computes @xmath5 and @xmath140 as a function of mass in various cosmological models is available upon request from the authors .
existing codes that compute nfw halo parameters as a function of mass and of other cosmological parameters may also be used , noting that @xmath146 and that @xmath147 .
finally , we note that neither formalism captures perfectly the mass dependence of the characteristic density ; small but significant deviations , as well as a sizable scatter , are evident in figure [ figs : rm2rhom2 ] .
dwarf galaxy halos appear to be less concentrated than predicted by the formalism proposed by b01 ; a similar observation applies to cluster halos when compared to ens predictions .
such shortcomings should be considered when deriving cosmological constraints from fits to observational data ( see , e.g. , zentner & bullock 2002 , mcgaugh et al 2003 ) ; and suggest again that direct comparison between observation and simulations is preferable to the use of fitting formulae .
we have analyzed the mass profile of @xmath0cdm halos in a series of simulations of high mass , spatial , and temporal resolution .
our series targets halos spanning five decades in mass : `` dwarf '' galaxy halos with virial circular velocities of order @xmath148 km s@xmath21 ; `` galaxy''-sized halos with @xmath149 km s@xmath21 ; and `` cluster '' halos with @xmath150 km s@xmath21 .
each of the nineteen halos in our series was simulated with comparable numerical resolution : they have between @xmath151 and @xmath152 million particles within the virial radius , and have been simulated following the `` optimal '' prescription for time - stepping and gravitational softening laid down in the numerical convergence study of p03 .
the high resolution of our simulations allows us to probe the inner properties of the mass profiles of @xmath0cdm halos , down to @xmath153 of @xmath1 in our best resolved runs .
these results have important implications for the structure of the inner cusp in the density profile and resolve some of the disagreements arising from earlier simulation work .
our main conclusions may be summarized as follows . *
@xmath0cdm halo density profiles are `` universal '' in shape : i.e. , a simple fitting formula reproduces the structure of all simulated halos , regardless of mass . both the nfw profile and the profile proposed by m99
describe the density and circular velocity profiles of simulated halos reasonably well .
best nfw fits to the circular velocity profiles deviate by less than @xmath6 over the region which is well resolved numerically .
best m99 fits reproduce circular velocity profiles to better than @xmath76 over the same region .
it should be noted , however , that the deviations increase inwards and are typically maximal at the innermost resolved radius , a result that warns against extrapolating to smaller radii with these fitting formulae .
* @xmath0cdm halos appear to be `` cuspy '' : i.e. , the dark matter density increases monotonically towards the center with no evidence for a well - defined `` core '' of constant density .
we find no evidence , however , for a central asymptotic power - law bin the density profiles .
these become progressively shallow inwards and are significantly shallower than isothermal at the innermost resolved radius , @xmath60 . at @xmath154 , the average slope of `` cluster '' , `` galaxy '' and
`` dwarf ''
halos halos is @xmath155 , @xmath99 , and @xmath100 , respectively .
this is steeper than predicted by the nfw profile but shallower than the asymptotic slope of the m99 profile . *
the density and enclosed mass at @xmath60 may be used to derive an upper limit on any asymptotic value of the inner slope .
cusps as steep as @xmath79 are confidently ruled out in essentially all cases ; the asymptotic slope of the nfw profile ( @xmath94 ) is still consistent with our data .
the radial dependence of @xmath108 differs from that of the nfw profile , however , decreasing more slowly with decreasing radius than is predicted .
for some scalings of the nfw fitting formula to the numerical data , this shape difference appears as a dark matter `` excess '' near the center which has ( erroneously ) been interpreted indicating a steeply divergent density cusp . * a simple formula where @xmath108 is a power law of radius reproduces the gradual radial variation of the logarithmic slope and its apparent failure to converge to any specific asymptotic value ( eq . [ eq : newprof ] ) .
this formula leads to much improved fits to the density profiles of simulated halos , and may prove a safer choice when comparison with observation demands extrapolation below the innermost converged radii of the simulations .
our study demonstrates that , although simple fitting formulae such as nfw are quite accurate in describing the global structure of @xmath0cdm halos , one should be aware of the limitations of these formulae when interpreting observational constraints .
extrapolation beyond the radial range where these formulae have been validated is likely to produce substantial errors .
proper account of the substantial scatter in halo properties at a given halo mass also appears necessary when assessing the consistency of observations with a particular cosmological model .
direct comparison between observations and simulations ( rather than with fitting formulae ) is clearly preferable whenever possible .
given the computational challenge involved in providing consistent , robust , and reproducible theoretical predictions for the inner structure of cdm halos it is likely that observational constraints will exercise to the limit our hardware and software capabilities for some time to come .
this work has been supported by computing time generously provided by the high performance computing facility at the university of victoria , as well as by the edinburgh parallel computing centre and by the institute for computational cosmology at the university of durham .
expert assistance by colin leavett - brown in victoria and lydia heck in durham is gratefully acknowledged .
jfn is supported by the alexander von humboldt foundation , the natural sciences and engineering research council of canada , and the canadian foundation for innovation . | we investigate the mass profile of @xmath0cdm halos using a suite of numerical simulations spanning five decades in halo mass , from dwarf galaxies to rich galaxy clusters .
these halos typically have a few million particles within the virial radius ( @xmath1 ) , allowing robust mass profile estimates down to radii below @xmath2 of @xmath1 .
our analysis confirms the proposal of navarro , frenk & white ( nfw ) that the shape of @xmath0cdm halo mass profiles differs strongly from a power law and depends little on mass .
the logarithmic slope of the spherically - averaged density profile , as measured by @xmath3 , decreases monotonically towards the center and becomes shallower than isothermal ( @xmath4 ) inside a characteristic radius , @xmath5 .
the fitting formula proposed by nfw provides a reasonably good approximation to the density and circular velocity profiles of individual halos ; circular velocities typically deviate from best nfw fits by less than @xmath6 over the radial range which is well resolved numerically . on the other hand , systematic deviations from the best nfw fits are also noticeable . inside @xmath5 , the profile of simulated halos becomes shallower with radius more gradually than predicted and , as a result , nfw fits tend to underestimate the dark matter density in these regions .
this discrepancy has been interpreted as indicating a steeply divergent cusp with asymptotic inner slope , @xmath7 .
our results suggest a different interpretation .
we use the density and enclosed mass at our innermost resolved radii to place strong constraints on @xmath8 : density cusps as steep as @xmath9 are inconsistent with most of our simulations , although @xmath10 is still consistent with our data .
our density profiles show no sign of converging to a well - defined asymptotic inner power law . we propose a simple formula that reproduces the radial dependence of the slope better than the nfw profile , and so
may minimize errors when extrapolating our results inward to radii not yet reliably probed by numerical simulations . |
after the recent photoemission experiments@xcite on quasi one - dimensional materials , the need of understanding the dynamical spectral functions of strongly correlated electron systems has arised . while the low energy behavior is usually well described within the framework of the luttinger liquid theory,@xcite the experimentally relevant higher energies ( @xmath1 mev ) can be calculated for example by diagonalizing small clusters@xcite or by quantum monte - carlo calculations.@xcite unfortunately , both methods have limitations either given by the small size of the system or by statistical errors and use of analytic continuation . even for the bethe - ansatz solvable models , where the excitation spectra can be calculated , the problematic part of calculating the matrix elements remains : the wave functions are required , and they are simply too complicated .
there is , however , a special class of models , where the evaluation of the matrix elements is made possible through a relatively simple factorized form of the wave - function , and some results were already published by sorella and parolla@xcite for the insulating half - filled case and by the authors@xcite away from half - filling .
the dynamical , zero temperature one - particle spectral functions can be defined as the imaginary parts of the time ordered green s function : @xmath2 @xmath3 is measured in angular resolved inverse photoemission experiments and can be calculated from the lehmann representation : @xmath4 while @xmath5 is measured in the angular resolved photoemission experiments and is given by : @xmath6 here @xmath7 is the number of electrons , @xmath8 denotes the final states and @xmath9 destroys an electron with momentum @xmath10 and spin @xmath11 .
if the spectral functions are known , the time ordered green s function can be obtained from @xmath12 the special models for which the matrix elements can be calculated are : \i ) the hubbard model , defined as usual : @xmath13 in the limit @xmath14 ; \ii ) the anisotropic @xmath15 model @xmath16 in the limit @xmath17 , where @xmath18 are the usual projected operators .
actually , the hubbard model in the large @xmath19 limit can be mapped onto a strong coupling model usually identified as the @xmath15 model plus three - site terms using a canonical transformation,@xcite where @xmath20 is small ; \iii ) an extension of the @xmath15 model first proposed by xiang and dambrumenil,@xcite defined by the hamiltonian @xmath21 where @xmath22 in the exchange part of the hamiltonian ensures that two spins interact as long as there is no other spin between them . the motivation to study
this model is that , unlike the infinite @xmath19 hubbard model , there is a finite energy @xmath23 associated with spin fluctuations , and this will give us useful indications about the finite @xmath19 hubbard model . from the models defined above , the hubbard model is the most relevant one .
it plays a central role as the generic model of strongly correlated electron systems .
even though it is comparatively simple , it is very difficult to solve except for the one dimensional case , where it is solvable by bethe ansatz.@xcite unfortunately , the bethe ansatz solution is not convenient for direct computation of spectral functions , therefore an alternative approach was needed . in the limit of small @xmath19 one can use the renormalization group@xcite to show that the hubbard model belongs to the universality class of the tomonaga - luttinger model,@xcite usually referred to as luttinger - liquid.@xcite the luttinger liquids are characterized by power - law decay of correlation functions , and nonexistence of quasiparticles.@xcite the underlying conformal field theory can be used to relate the exponents to finite - size corrections of the energy and momentum.@xcite this gives consistent results not only with the renormalization group in the weak coupling regime,@xcite but also with the special case of @xmath14 , where the exponents of the static correlations could be obtained using a factorized wave function.@xcite actually , the spin - charge factorized wave function also describes the excited states as well,@xcite and it can be used to calculate the dynamical spectral functions as well .
the spectral functions obtained in this way are very educative and in some sense , unexpected .
for example , it turns out that the spectrum contains remnants of bands@xcite crossing the fermi energy at @xmath24 - the so called shadow bands .
also it gives information on the applicability of the power - law luttinger liquid correlation function.@xcite the aim of this paper is not only to give the details of the calculation , that can be useful for other correlation functions , but also to present some new results on the low energy behavior of the charge and spin part ( both for the isotropic heisenberg and xy spin model ) .
the paper is organized as follows : in section [ sec : wavefunc ] we review the factorized wave function and in section [ sec : specfunc ] we show how the spectral functions can be given as a convolution of spin and charge parts
. sections [ sec : charge ] and [ sec : spin ] are devoted to the detailed analysis of the charge and spin parts .
the relation to the results obtained from the finite - size corrections and conformal field theory is discussed in section [ sec : cft ] .
finally , in section [ sec : conclu ] we present our conclusions .
it has been shown,@xcite by using the bethe ansatz solution , that the ground state wave function of the hubbard model in the @xmath25 limit can be constructed as a product of a spinless fermion wave function @xmath26 and a squeezed spin wave function @xmath27 .
this can be alternatively seen using perturbational arguments@xcite and then extended to the @xmath15 model in the @xmath28 limit . moreover , the wave function of the excited states are also factorized:@xcite @xmath29 the spinless fermion wave function @xmath26 describes the charges and is an eigen function of @xmath7 noninteracting spinless fermions on @xmath30 sites with momenta @xmath31 where the @xmath32 are integer quantum numbers and @xmath33 .
the charge part is not fully decoupled from the spin wave function @xmath27 , as the momentum @xmath34 ( @xmath35 ) of the spin wave function imposes a twisted boundary condition on the spinless fermion wave - function ( each fermion hopping from site @xmath36 to site @xmath37 will acquire a phase @xmath38 ) to ensure periodic boundary conditions for the original problem .
the energy of the charge part is @xmath39 and the momentum reads @xmath40 , or , using eq .
( [ eq : kiq ] ) : @xmath41 on the other hand , the spin wave functions @xmath27 are characterized by the number of down spins @xmath42 , the total momentum @xmath43 , and the quantum number @xmath44 within the subspace of momentum @xmath43 .
they are eigenfunctions of the heisenberg hamiltonian @xmath45 with eigenenergies @xmath46 .
@xmath47 depends on the actual charge wave function @xmath26 . in the case of the @xmath0 hubbard model , @xmath48 where @xmath49 and @xmath50 are the operators of spinless fermions at site @xmath51 . for the ground state
@xmath52 it reads @xmath53 $ ] , where @xmath54 is the density . for the @xmath15 model : @xmath55 and for the ground state @xmath56 $ ] . for the model of xiang and dambrumenil @xmath57 and
is independent of the charge part .
the energy of the factorized wave function is then given as the sum of the charge and spin energies , with the assumption that the correct @xmath58 is chosen .
if @xmath25 or @xmath59 , then the spectrum collapses and we can assume all the spin states degenerate , simplifying considerably some of the calculations to be presented later .
furthermore , we choose @xmath7 to be of the form @xmath60 ( @xmath61 integer ) , when the ground state is unique . then in the ground - state the spinless fermion wave - function
@xmath62 is described by the quantum numbers @xmath63 and @xmath64 , so that the distribution of the @xmath65 s is symmetric around the origin and we choose the spin part as the ground - state of the heisenberg model according to ogata and shiba s prescription.@xcite this choice of the spin wave function makes the difference between the @xmath25 and @xmath66 ( the so called @xmath67 model ) limits .
the price we have to pay for such a simple wave function is that the representation of real fermion operators @xmath68 in the new basis becomes complicated . as a first step
, we can write @xmath68 as @xmath69 , where @xmath70 creates a fermion at an unoccupied site and the @xmath71 adds a fermion at an already occupied site , thus creating a doubly occupied site . @xmath72 means the spin state opposite to @xmath11 . this latter process gives contributions to the spectral functions in the upper hubbard band , @xmath73 which can be calculated in a similar way , but we will not address this issue in the present paper .
next , we define the operators @xmath74 and @xmath75 acting on the spin part of the wave function : the @xmath76 adds a spin @xmath11 to the beginning of the spin wave function @xmath77 if @xmath78 , or inserts a spin @xmath11 after skipping the first @xmath79 spins , and makes it @xmath80 long , e.g. : @xmath81 and @xmath82 . the @xmath83 is defined as the adjoint operator of @xmath74 , i.e. it removes a spin from site @xmath79 .
then , to create a fermion at the empty site @xmath84 , we need to create one spinless fermion with operator @xmath85 and to add a spin @xmath11 to the spin wave function with operator @xmath86 : @xmath87 the apparent simplicity is lost for @xmath88 .
then , apart from creating a spinless fermions with @xmath89 in the charge part , we have to consider the following two possibilities : either the @xmath84 site is empty , and with @xmath88 we create a spin at the beginning of the spin wave function with @xmath86 ; or it is occupied , and we insert a spin between the first and second spin in @xmath27 with @xmath90 .
so we end up with @xmath91 b^\dagger_1 .\ ] ] obviously we choose the @xmath84 in further calculations for its simplicity .
however , one can show that the final result does not depend on this special choice and the translational invariance is preserved even for these complicated operators .
to use the factorized wave functions in the calculation of the spectral function it is more convenient to transfer the @xmath10 dependence from the @xmath92 operator to the final state : @xmath93 and @xmath94 where the momenta of the final states are @xmath95 .
as we already pointed out , the addition of an electron to the ground state can result in a final state with or without a doubly occupied state . correspondingly , the spectral function has contributions from the upper and lower hubbard bands : @xmath96 .
we will now consider @xmath97 only . from eqs .
( [ eq : oswflhb ] ) and ( [ eq : adeclhb ] ) we get the following convolution as a consequence of the wave function factorization : @xmath98 and similarly for @xmath5 : @xmath99 @xmath100 and @xmath101 depend on the spinless fermion wave function only : @xmath102 and they are discussed in more detail in the next section ( sec .
[ sec : charge ] ) . on the other hand , @xmath103 and @xmath104 are determined by the spin wave function only : @xmath105 and are analyzed in sec .
[ sec : spin ] .
although we do not present it here , a similar analysis can be made for @xmath73 . in eqs .
( [ eq : alhbca ] ) and ( [ eq : blhbca ] ) the simple addition of the spin and charge energies is assumed . strictly speaking ,
this is only valid for the @xmath0 , @xmath28 and the model of xiang and dambrumenil for any @xmath23 . in the other cases
the dependence of @xmath58 on the charge wave function should be explicitly taken into account .
still , it is a reasonable approximation , as the important matrix elements will come from exciting a few particle - hole excitations only , which will give finite - size corrections to @xmath58 in the thermodynamic limit .
furthermore , we are neglecting the @xmath106 corrections to the effective operators@xcite and to the wave functions .
the momentum distribution function , @xmath107 can be calculated from the spectral function as @xmath108 , leading to a similar expression as used by pruschke and shiba:@xcite @xmath109 where @xmath110 and similarly @xmath111 . the local spectral function @xmath112 is given by @xmath113 where @xmath114 .
similar equation holds for @xmath115 .
to calculate @xmath100 and @xmath101 defined in eq .
( [ eq : defaomegaq ] ) , we need to evaluate matrix elements like @xmath116 , where the two states have different boundary conditions . in the ground state @xmath117 , but we will not specify @xmath118 yet . to calculate these matrix elements ,
we need the following anti - commutation relation : @xmath119/2)}{\sin([k'-k]/2 ) } .
\end{aligned}\ ] ] where @xmath10 and @xmath120 are wave - vectors with phase shift @xmath121 and @xmath122 , respectively , see eq .
( [ eq : kiq ] ) . for @xmath123
the anti - commutation relation is the usual one : @xmath124 , while for @xmath125 the overall phase shift @xmath126 due to momentum transfer @xmath127 to the spin degrees of freedom gives rise to the anderson s orthogonality catastrophe.@xcite then a typical overlap @xmath128 , where @xmath129 is the vacuum state , is given by the following determinant : @xmath130 replacing the anticommutator , the determinant above becomes @xmath131 this determinant is very similar to the cauchy determinant ( there the elements are @xmath132 instead of @xmath133 ) and it can be expressed as a product,@xcite so for the overlap we get : @xmath134 where the sign @xmath135 is for @xmath136 and @xmath137 for @xmath138 .
now we turn back to the @xmath139 .
the matrix elements in eq .
( [ eq : defaomegaq ] ) are @xmath140 2 where @xmath141 is a wave vector with phase shift @xmath122 .
here we have used that @xmath142 holds , independently of the actual quantum numbers @xmath143 and @xmath144 . similarly , for the matrix elements in @xmath145
we get : @xmath146 we are now ready to calculate the spectral functions numerically .
one has to generate the quantum numbers @xmath32 , and evaluate the energy , momentum and the expressions above . from now on
, we will consider @xmath117 .
first of all , it turns out that the following sum rules are satisfied for every @xmath43 : @xmath147 in the absence of the anderson orthogonality catastrophe , when @xmath148 , the contribution to the spectral functions comes from one particle - hole excitations only , and the spectral functions are nothing but the familiar @xmath149 .
this is not true any more when we consider @xmath150 . in that case
we get contributions from many particle - hole excitations as well .
the largest weight comes from the one particle - hole excitations , and increasing the number of excited holes , the additional weight decreases rapidly .
although from eq .
( [ eq : aqmatele ] ) we can calculate the matrix elements numerically for all the excitations of the final state , its application is limited to small system sizes ( typically @xmath151 ) .
it is due to the fact that the time required to generate all the possible states ( quantum numbers @xmath152 ) is growing exponentially . therefore , in some of the calculations we take into account up to three particle - hole excitations only .
in table . [ tab : sr ] we give the total sum rule for small sizes in a calculation where we took into account up to one , two and three particle - hole excitations .
we can see that the missing weight is really small in the approximation that includes up to three particle - hole excitations in the final state .
so , if we restrict ourselves to a finite number of particle - hole excitations and introduce the function @xmath153 \frac{\pi}{l } + \frac{q \ ! - \ !
\pi}{2l }
\right ) , \label{eq : defg}\end{aligned}\ ] ] the calculation of the spectral weight becomes simple .
the weight of the peak corresponding to a one particle - hole excitation can be given as : @xmath154\frac{\pi}{l } ) } a_q^{(0,0 ) } , \label{eq : def1ph}\ ] ] where we have removed the quantum number @xmath155 ( hole ) from and added @xmath156 ( particle ) to the set @xmath143 of the ground - state of @xmath157 fermions , so that the momentum of the final state is @xmath158 and the energy is @xmath159 , where the @xmath160 is the momentum of the ground state .
furthermore , @xmath161 is the overlap between the @xmath7 electron ground state with boundary condition @xmath162 and the @xmath80 electron ground state with boundary condition @xmath43 , and will be discussed later .
similarly , for the two particle - hole excitations we get : @xmath163\frac{\pi}{l } ) \sin^2([{\cal i}^p_{1 } \!-\ ! { \cal i}^p_{2}]\frac{\pi}{l } ) } { \sin^2([{\cal i}^p_{1 } \!-\ ! { \cal i}^h_{1}]\frac{\pi}{l } ) \sin^2([{\cal i}^p_{1 } \!-\ ! { \cal i}^h_{2}]\frac{\pi}{l } ) \sin^2([{\cal i}^p_{2 } \!-\ ! { \cal i}^h_{1}]\frac{\pi}{l } ) \sin^2([{\cal i}^p_{2 } \!-\ ! { \cal i}^h_{2}]\frac{\pi}{l } ) } a_q^{(0,0 ) } \label{eq : def2ph}\ ] ] 2 with energy and momentum @xmath164 the corresponding equations for three or more particle - hole excitations are similar to those above , but since they are long , we do not give them here . a typical plot of @xmath139
is shown in fig .
[ fig : aomkq ] .
we choose @xmath165 , which is halfway between the symmetric @xmath166 and the trivial @xmath63 case . in the figure
we can see the singularity near the fermi energy , furthermore the weights are distributed on a cosine- like band . to make it more clear , in fig .
[ fig : support ] we show the support of @xmath139 and the distribution of the weights .
now , what can we say about @xmath161 , the weight of the lowest peak ? in the ground state the quantum numbers @xmath32 and @xmath167 are densely packed , and from eq .
( [ eq : aqmatele ] ) we get @xmath168^{2 n + 1 - 2 j } \nonumber\\ & & \times \prod_{j=1}^n \left [ \sin^2 \frac{(2 j\!-\!1)\pi\ !
+ \!q}{2 l } \sin^2 \frac{(2 j\!-\!1)\pi \!-\!q}{2 l } \right]^{j - n-1 } \nonumber\end{aligned}\ ] ] from this we can conclude that @xmath161 is an even function of @xmath43 and @xmath169 .
we are not able to give a closed formula for the sum . however
, very useful information can be obtained by noticing that @xmath170 and in the thermodynamic limit , @xmath171 .\ ] ] here the @xmath43 is extended outside the brillouin zone .
now it is straightforward to get the size- and filling - dependence of @xmath161 : @xmath172 , \label{eq : a00}\ ] ] where @xmath173 eq .
( [ eq : a00 ] ) is also valid for @xmath174 , apart from the sign in the @xmath175 correction .
the @xmath176 is an even function of @xmath43 , @xmath177 , and it satisfies the second order recurrence equation @xmath178 which can be reduced to @xmath179 and it follows that @xmath180 , @xmath181 etc .
are zero . in the interval from @xmath166 to
@xmath162 it can be approximated as @xmath182 with accuracy @xmath183 .
furthermore @xmath184 . as we can see in fig .
[ fig : support ] , for low energies @xmath139 has so called towers of excitations centered at momenta @xmath185 , where @xmath186 is an integer .
the largest weights are for the peaks in the tower with @xmath187 , the next with @xmath188 ( if @xmath189 ) or @xmath190 ( if @xmath191 ) , and so on .
the lowest excitation in tower @xmath186 corresponds to a set of densely packed quantum numbers @xmath32 shifted by @xmath186 . from the definition of the momenta @xmath65 ,
this is equivalent to imposing a twist of wave - vector @xmath192 .
therefore we can introduce @xmath193 , where @xmath194 is not restricted to be in the brillouin zone , but for @xmath195 it has values outside .
we define @xmath196 to describe the @xmath186-th tower , so that @xmath100 has contributions from each of the towers : @xmath197 . furthermore , we enumerate the peaks in a given tower with indices @xmath79 and @xmath198 , so that the energy and momentum of the peaks are , from eqs .
( [ eq : kiq ] ) , ( [ eq : elhb ] ) and ( [ eq : plhb ] ) : @xmath199 where we have neglected the @xmath200 finite - size corrections . here
@xmath201 is the ` fermi energy ' , @xmath202 is the ` fermi ( charge ) velocity ' and @xmath203 is the ` fermi momentum ' of spinless fermions representing the charges . by @xmath204 we denote the weight of the peaks , and for convenience , we also introduce the relative weights @xmath205 . the weight of the first few lowest - lying peaks can be calculated explicitly by eqs .
( [ eq : defg])-([eq : def2ph ] ) , as they are given by a finite number of particle - hole excitations .
the degeneracy of each peak grows with @xmath79 and @xmath198 .
here we assumed that the dispersion relation is linear near the fermi level with velocity @xmath206 .
clearly , this picture is valid for energies small compared to bandwidth . from eq .
( [ eq : def1ph ] ) we get the relative weights @xmath207 , e.g. @xmath208 is given as : @xmath209 introducing @xmath210 , the relative weights in the thermodynamic limit simplify so that : @xmath211 and also @xmath212 holds .
note that some peaks are degenerate and therefore they are a sum of more contributions .
now , it takes only one step to get the general formula which reads ( including the finite - size corrections ) : @xmath213 , \label{eq : aii}\end{aligned}\ ] ] where @xmath214 it can also be expressed with the help of the @xmath215-function , since @xmath216 the asymptotic expansion of the @xmath215-function gives @xmath217 which is a reasonable approximation apart from the @xmath78 peak .
then , it follows that @xmath207 has a power law behavior : @xmath218 note that the exponent @xmath219 in eq .
( [ eq : a00 ] ) is also given by @xmath220 .
we can clearly see the manifestation of the underlying conformal field theory : i ) the finite - size corrections to the energy and momentum [ eqs .
( [ eq : echargefs ] ) and ( [ eq : pchargefs ] ) ] of the lowest lying peak in the tower determines the exponents of the correlation functions ; ii ) the weights in the towers are given by @xmath215-function.@xcite the spectral function @xmath196 in the thermodynamic limit is given by @xmath221 and collecting everything together , eqs .
( [ eq : a00 ] ) and ( [ eq : aiipow]-[eq : atl ] ) , for the low energy behavior of @xmath139 we get @xmath222^{\beta_{\tilde q } } \left [ \frac{\omega \!-\ ! \varepsilon_c \!-\ !
u_c ( k \!-\ ! k_{\tilde q } ) } { 4 \pi u_c \sin(\pi n ) } \right]^{\beta_{-{\tilde q } } } .
\label{eq : aqkom}\end{aligned}\ ] ] it is also worth mentioning the symmetry property @xmath223 .
the whole calculation can be repeated for the spectral function @xmath145 : @xmath224^{\beta_{-{\tilde q } } } \left [ \frac{\varepsilon_c \!-\ ! \omega \!+\ !
u_c ( k \!+\ ! k_{\tilde q } ) } { 4 \pi u_c \sin \pi
n } \right]^{\beta_{{\tilde q } } } .
\label{eq : bqkom}\end{aligned}\ ] ] we should note , however , that these expressions are restricted for the weights far from the edges of the towers , where the asymptotic expansion of the @xmath215-function , eq .
( [ eq : gasympt ] ) , is valid .
this is especially true when @xmath225 , where the correct result is @xmath226)$ ] .
in other words , for the exponents close to @xmath227 there can be a considerable deviation from the power law behavior , and the spectral weight accumulates along the edges of the towers .
this behavior can be observed in fig .
[ fig : aomkq ] , where the exponents are @xmath228 and @xmath229 .
for the local ( @xmath10-averaged ) spectral function @xmath230 the weight of the @xmath51-th peak , denoted by @xmath231 , is @xmath232 the summation gives : @xmath233 .
\nonumber\end{aligned}\ ] ] if we put it together with eqs .
( [ eq : a00 ] ) and ( [ eq : echargefs ] ) , and neglect the @xmath175 corrections , the local spectral function in the @xmath234 limit reads : @xmath235 for @xmath236 the @xmath237 should be replaced by @xmath238 .
we show @xmath239 for some selected values of @xmath43 in fig .
[ fig : aomq ] . here
we try to make some statements about @xmath240 in eq .
( [ eq : nkf ] ) .
a nave calculation in the low energy region is to sum up the weights near @xmath241 @xmath242 of course , one is aware that the summation includes high energies as well , where the equivalent for @xmath243 of eq .
( [ eq : aii ] ) is not valid any more .
however , the largest contributions come from the low energy regions and the error is not very large
. we do not want to get precise values , but rather some qualitative results . neglecting the @xmath244 corrections
, the sum gives for @xmath245 : @xmath246 and for @xmath247 the @xmath61 and @xmath248 should be replaced by @xmath249 and @xmath250 .
again , we can use the asymptotic expansion of the @xmath215-function to get @xmath251 where @xmath252 for @xmath253 and @xmath254 for @xmath255 should be taken in the argument of the sine .
it is interesting that , although the exponent of the singularity @xmath219 is the same for @xmath256 and @xmath257 , there is a strong asymmetry due to the prefactor ( a similar observation was made by frahm and korepin@xcite ) . in fig .
[ fig : abq ] this behavior is clearly observed .
for @xmath258 the correct result of @xmath259 is recovered .
to calculate @xmath103 and @xmath104 given by eqs .
( [ eq : defcomegaq ] ) , we need to know the energies and wave - functions of the spin part .
they can be calculated from the usual spin-@xmath260 heisenberg hamiltonian , see eq .
( [ eq : hseff ] ) , taking @xmath7 and @xmath157 sites ( spins ) . for the @xmath261 case the excitation spectrum of the spins collapse , and then we can use the local , @xmath262 integrated functions @xmath263 and @xmath264 .
they are related to the spin transfer function @xmath265 , defined by ogata and shiba,@xcite as was first noticed by sorella and parola.@xcite the spin transfer function gives the amplitude of removing a spin @xmath11 at site @xmath266 ( here we choose @xmath267 ) and inserting it at site @xmath51 , and can be given as @xmath268 where the operator @xmath269 permutes the spins at sites @xmath79 and @xmath270 .
then @xmath271 and @xmath272 read @xmath273 , \nonumber \\
d_{\sigma}(q ) & = & \frac{1}{n-1 } \sum_{j=0}^{n-2 } e^{i ( q+\pi ) j } \omega_{0\rightarrow j,\sigma } .
\label{eq : cdw}\end{aligned}\ ] ] in particular , @xmath274 , and it follows that @xmath275 and @xmath276 .
we are interested in these quantities for two particular cases : the isotropic heisenberg model because it is physically relevant , and the @xmath277-model because it allows analytical calculations .
we first consider the @xmath277-model because the simplicity of that case makes it more convenient to introduce the basic ideas . in this special case the spin problem can be mapped to noninteracting spinless fermions using the wigner - jordan transformation .
it means that the eigenenergies and wave functions are known , and we can calculate @xmath104 and @xmath103 analytically .
we are facing a similar problem - the orthogonality catastrophe - as when we calculated the @xmath278 , but now it comes from the overlaps between states with different number of sites . for convenience ,
we choose the spinless fermions to represent the @xmath72 spins , so that the operator @xmath86 ( @xmath279 ) only adds ( removes ) a site and does not change the number of fermions , which we fix to be @xmath280 . then we have to evaluate matrix elements like @xmath281 and @xmath282 , where in the @xmath283 the @xmath37 site is unoccupied and the fermions are on sites @xmath284 and from site @xmath285 they hop to @xmath286 skipping the @xmath287 site . for simplicity , we consider cases when the number of spin up and down fermions is odd ( @xmath7 is even ) , so that we do not have to worry about extra phases arising from the jordan - wigner transformation
. then the momentum of the ground state @xmath283 is @xmath288 .
let us denote by @xmath120 the momenta of fermions on a @xmath157 site lattice , quantized as @xmath289 and by @xmath10 the momenta of fermions on a @xmath7 site lattice , quantized as @xmath290 , where @xmath291 and @xmath292 are integers ( @xmath293 ) , and by @xmath8 and @xmath294 the operators of the spinless fermions .
the energy and momentum of the state are : @xmath295 to calculate the matrix element in @xmath296 , see eq .
( [ eq : defcomegaq ] ) , we need the following anti - commutation relation : @xmath297/2 ) } , \nonumber\end{aligned}\ ] ] and the matrix element @xmath298 is again given by a cauchy determinant , which can be expressed as a product : @xmath299^{-{n_{\bar\sigma } } } \prod_{j=1}^{n_{\bar\sigma } } \sin^2 \frac{k'_j}{2 } \prod_{j > i } \sin^2 \frac{k_j - k_i}{2 } \nonumber \\ & & \times \prod_{j > i } \sin^2 \frac{k'_j - k'_i}{2 } \prod_{i , j } \sin^{-2 } \frac{k'_i - k_j}{2 } .
\label{eq : prodc}\end{aligned}\ ] ] similarly , in the case of @xmath300 , the anticommutator is @xmath301/2 ) } , \ ] ] and the matrix element @xmath302 is equal to @xmath303^{-{n_{\bar\sigma } } } \prod_{j=1}^{n_{\bar\sigma } } \sin^2 \frac{k_j}{2 } \prod_{j > i } \sin^2 \frac{k_j - k_i}{2 } \nonumber \\ & & \times \prod_{j > i } \sin^2 \frac{k'_j - k'_i}{2 } \prod_{i , j } \sin^{-2 } \frac{k'_i - k_j}{2 } .
\label{eq : prodd}\end{aligned}\ ] ] as soon as we have the product representation , it is straightforward to analyze the low energy behavior and also to obtain numerically @xmath304 and @xmath305 for larger system sizes .
the low energy spectra of @xmath300 and @xmath296 consist of towers centered at momenta @xmath306 , where @xmath307 . to analyze the low energy behavior in the tower labeled by @xmath308
, we can proceed analogously to the charge part : the weights in the tower of excitations , @xmath309 and @xmath310 , can be calculated from eqs .
( [ eq : prodc ] ) and ( [ eq : prodd ] ) .
the energy and momentum of the state @xmath311 can be calculated from eqs .
( [ eq : espinfsxy ] ) and ( [ eq : pspinfsxy ] ) and neglecting the @xmath312 corrections they read : @xmath313 where @xmath314 and the `` fermi energy '' and the velocity of the spins are : @xmath315 and @xmath316 .
the relative weights @xmath317 can be calculated from eq .
( [ eq : prodd ] ) , e.g. : @xmath318 where @xmath319 and the other @xmath317 are similar . in the thermodynamic limit , @xmath320
, the weight @xmath321 simplifies to @xmath322 .
\label{eq : dgamma10}\ ] ] neglecting the finite - size corrections , for general @xmath311 and @xmath308 we get : @xmath323 where the exponents @xmath324 are defined in eq .
( [ eq : gammaxy ] ) and the weights again follows the prescription of the conformal theory , with strong logarithmic finite - size corrections however .
a similar analysis can be done for @xmath296 . from the above and eq .
( [ eq : defcomegaq ] ) we obtain @xmath325^{\gamma_r^- } \nonumber\\ & & \times \left[\varepsilon_{\sigma}-\omega - u_\sigma ( q - q_{r,\sigma } ) \right]^{\gamma_r^+ } \nonumber\\ & & \times \theta(\varepsilon_{\sigma}-\omega + u_\sigma |q - q_{r,\sigma}| ) \label{eq : dqom}\end{aligned}\ ] ] and @xmath326^{\gamma_r^+ } \nonumber\\ & & \times \left[\omega-\varepsilon_{\sigma } - u_\sigma ( q - q_{r,\sigma } ) \right]^{\gamma_r^- } \nonumber\\ & & \times \theta(\omega-\varepsilon_{\sigma } - u_\sigma |q - q_{r,\sigma}| ) \label{eq : cqom}\end{aligned}\ ] ] where @xmath327 are numbers which can be determined numerically . we immediately see that the @xmath328 and @xmath329 are singular at @xmath330 : @xmath331 with exponent @xmath332 and they are strongly asymmetric around @xmath333 , as we can conclude from the analog of eq .
( [ eq : bqk ] ) . for the non - magnetic case ( @xmath334 )
, the singularity is at @xmath335 for all the towers , and the exponents of the main singularity ( @xmath336 ) are @xmath337 and @xmath338 , furthermore @xmath339 . although the heisenberg model is solvable by bethe - ansatz and in principle the wave functions are known , it is too involved to give the matrix elements of @xmath103 and @xmath104 .
the simplest alternative way is exact diagonalization of small clusters and dmrg@xcite extended to dynamical properties.@xcite we have used both methods to calculate the weights for system sizes up to @xmath340 and @xmath341 , respectively . a typical distribution of the weights for @xmath103 for zero magnetization is given in fig .
[ fig : weights ] .
there are several features to be observed : i ) due to selection rules , the nonzero matrix elements are with the @xmath342 final states only ; ii ) the weight is concentrated along the lower edge of the excitation spectra in the interval @xmath343 ; iii ) there are two , almost overlapping towers visible corresponding to @xmath336 and @xmath344 .
our interpretation of the spectrum is that the weight mostly follows the dispersion of the spinon of faddeev and takhtajan,@xcite since the final states have an odd number of spins , thus there can be a single spinon in the spectrum and it has a cosine - like dispersion .
it is also surprising that for @xmath103 more than 97% and for @xmath104 more than 99% of the total weight is found in this spinon branch .
this behavior is similar to that discussed by talstra , strong and anderson,@xcite where they added two spins to the spin wave function .
we can also try to analyze the low energy behavior from the conformal field theory point of view .
namely , from the bethe - ansatz solutions the finite - size corrections to the energy are known@xcite and they are also given by eqs .
( [ eq : espinfs ] ) and ( [ eq : pspinfs ] ) apart from @xmath345 corrections , with @xmath346 for zero magnetization the velocity @xmath347 reads @xmath348 , the energy is @xmath349 and @xmath350 , and the exponents are @xmath351 and @xmath352 , very close to the @xmath277 exponents ( @xmath353 and @xmath354 , respectively ) . for arbitrary magnetization @xmath347 , @xmath355 and @xmath356
are to be calculated from integral equations.@xcite also , we check if eq .
( [ eq : dgamma ] ) is satisfied for the @xmath336 tower in fig .
[ fig : cts ] .
namely , it tells us that @xmath357 and @xmath358 , apart from finite - size corrections which we assumed to be of the same form as in the case of the @xmath277 model in eq .
( [ eq : dgamma10 ] ) .
we believe that this method can also be used to determine exponents in a more general cases as well .
another interesting point is that the exponent @xmath351 already indicates that @xmath359 vanishes , in agreement with the selection rules .
however , there is still some weight for @xmath360 , which comes from @xmath342 bound states of spinons .
we do not know the finite - size scaling of that weight , i.e. if it disappears in the thermodynamic limit or not .
now , if we recall that @xmath361 , then it follows ( see eq .
( [ eq : bqk ] ) ) that the contribution to @xmath362 for @xmath363 is strongly suppressed , and we see essentially the contributions from the @xmath364 tower .
since the contribution to @xmath305 and @xmath304 come mostly from the lower edge of excitation spectrum , we can use the approximations @xmath365 where @xmath366 is the des cloizeaux - pearson dispersion@xcite @xmath367 the @xmath328 and @xmath329 can be calculated numerically for small clusters ( typically up to @xmath368 with exact diagonalization and @xmath369 with dmrg ) for the non - magnetic case ( see refs . ) .
the @xmath370 and @xmath371 seems to have small finite - size effect , as follows from eq .
( [ eq : cdw ] ) , and the singularity in the non - magnetic case is given by @xmath372 , as already noticed by sorella and parola.@xcite we have also calculated @xmath328 and @xmath329 for the system with finite magnetization @xmath373 ( see fig . [
fig : qm ] ) . there
@xmath374 , @xmath375 and the exponents are @xmath376 and @xmath377 .
these exponents are consistent with @xmath378 and in surprisingly good agreement with the simple formula given by frahm and korepin@xcite @xmath379 valid in a large magnetic field .
the real space green s function can be calculated from the spectral functions as @xmath380 for @xmath381 and @xmath3 should be replaced by @xmath5 for @xmath382 , as follows from eq .
( [ eq : gab ] ) .
then , from eqs .
( [ eq : alhbca ] ) , ( [ eq : aqkom ] ) and ( [ eq : cqom ] ) it follows that : @xmath383 where @xmath384 was defined as @xmath385 , furthermore @xmath386 are numbers .
the charge velocity @xmath206 is the same one as in eq .
( [ eq : echargefs ] ) , while the spin velocity is @xmath387 , where @xmath347 was defined in eq .
( [ eq : espinfs ] ) .
the green s function has singularities at different momenta , depending on the actual quantum numbers @xmath186 and @xmath308 , see table . [ tab : sing ] for details . on the other hand , according to the conformal field theory,@xcite a correlation functions @xmath388 reads : @xmath389 x } } { ( x\!-\!u_c t)^{2 \delta^+_c } ( x\!+\!u_c t)^{2 \delta^-_c } ( x\!-\!u_s t)^{2 \delta^+_s } ( x\!+\!u_s t)^{2 \delta^-_s } } , \ ] ] where the exponents @xmath390 are related to the finite - size corrections : @xmath391 and @xmath392 are numbers .
the quantum numbers @xmath393 , @xmath394 , @xmath395 and @xmath396 characterize the excitations and are related to @xmath186 and @xmath308 as given in table .
[ tab : qn ] .
the @xmath397 s are the elements of the so called dressed charge matrix .
it can be calculated from bethe ansatz solution of the hubbard model , and in the large @xmath19 limit they read : @xmath398 where @xmath356 can be obtained solving an integral equation . for the non - magnetic case @xmath399 and @xmath350 .
then we are ready to identify the exponents : @xmath400 and @xmath401 , and in this way we can directly see the validity of the cft in the large-@xmath19 limit . in case of the @xmath402 model
no bethe ansatz result is known , but using the analogy with the isotropic case , the exponents are readily obtained using the substitution @xmath403 , @xmath404 , @xmath405 and @xmath406 .
to conclude , we have shown that for some special cases the spectral functions of the 1d hubbard can be calculated using the spin - charge factorized wave - function , which implies that the spectral functions are given as a convolution involving the charge and spin parts .
analytical calculations are possible for the charge part and for the spin part in the case of the @xmath277 model .
the low energy behavior turns out to be fully consistent with the predictions of the conformal field theory , i.e. the exponents are given by the finite - size corrections to the energy and momentum , and the weights are given by @xmath215-function .
based on this , we propose a new way to determine the exponents of the correlation functions .
furthermore , we argue that when the exponents of the correlation functions are close to integers , the luttinger - liquid power - law behavior of the correlation functions should be taken with care , as it comes from the asymptotic expansion of the @xmath215 function .
b. dardel _
et al . _ ,
lett . * 67 * , 3144 ( 1991 ) ; y. hwu _ et al .
b * 46 * , 13624 ( 1992 ) ; c. coluzza _ et al . _ ,
b * 47 * , 6625 ( 1993 ) ; b. dardel _ et al _ , europhys
* 24 * , 687 ( 1993 ) ; m. nakamura _
et al . _ ,
b * 49 * , 16191 ( 1994 ) . | we give the details of the calculation of the spectral functions of the 1d hubbard model using the spin - charge factorized wave - function for several versions of the @xmath0 limit .
the spectral functions are expressed as a convolution of charge and spin dynamical correlation functions .
a procedure to evaluate these correlation functions very accurately for large systems is developed , and analytical results are presented for the low energy region .
these results are fully consistent with the conformal field theory .
we also propose a direct method of extracting the exponents from the matrix elements in more general cases . 2 |
there is a lot of effort invested in the development and construction of large interferometric gravitational wave detectors such as ligo , virgo , geo , etc .
these observatories are expected to measure the gravitational waves emitted in relativistic astrophysical systems , for example in the coalescence of two compact objects . at the late stages of such systems
one can consider the situation of having two black holes or neutron stars in nearly circular orbits which are shrinking due to the loss , by the emission of gravitational waves , of energy and angular momentum . in order to estimate
how important is the radiation mechanism of angular momentum in these kind of processes , let us consider briefly the situation of two newtonian point masses , with mass @xmath0 each , in circular motion at a distance @xmath1 .
then one can calculate the ratio between total angular momentum @xmath2 and total mass square @xmath3 , obtaining @xmath4 ; where @xmath5 is the schwarzschild radius of masses @xmath0 , and we are using geometric units for which the speed of light and the gravitational constant have the unit value .
this means that for orbits for which @xmath6 , the angular momentum exceeds the kerr limit to have a final black hole , as opposed to a naked singularity .
in other words , some how , one has to account for the radiation of angular momentum before the final collapse of the system , if one is to expect a black hole at the end of the coalescence .
it is probably interesting to mention that according to newtonian dynamics , the magnitude of the velocity of these particles when they are in the @xmath7 orbit is @xmath8 of the speed of light ; which indicates that a relativistic description of these systems is required . since the angular momentum loss is crucial for the coalescence mechanism , it is important to have an accurate description of this process .
this leads to the first question which is : what is the definition of angular momentum in a relativistic system ?
there are several difficulties associated to this question .
to begin with , the spacetime appropriate for the description of a relativistic isolated system of compact objects is curved and asymptotically flat .
therefore , since the asymptotic symmetry group is the infinite dimensional bms group@xcite , the notion of angular momentum in general relativity will have to deal with the so called problem of supertranslations .
roughly speaking , among the generators of the bms group , one can distinguish a set of 6 rotations , and an infinite set of supertranslations . a normal 4-dimensional subgroup of the bms group exists which allows for the unambiguous definition of total momentum ; the so called bondi momentum .
to give perspective to this problem , let us recall that in special relativity , the angular momentum @xmath9 , the intrinsic angular momentum @xmath10 and the linear momentum @xmath11 are related by @xmath12 where @xmath13 are numeric spacetime indices and @xmath14 represents the translational freedom .
a rest frame is one in which the momentum @xmath15 has only timelike components .
given a rest reference frame in minkowski spacetime one needs to use the spacelike translation freedom appearing in expression ( [ eq : j ] ) in order to single out the center of mass reference frame . in the center of mass
frame one has @xmath16 ; that is the total momentum coincide with the intrinsic angular momentum . in general relativity , instead of having just a momentum
, we have an infinite component supermomentum vector ( associated with the infinite generators of supertranslations ) . therefore the analog of equation ( [ eq : j ] ) necessarily will forces us to deal with the notion of supermomentum and its corresponding notion of rest frames@xcite .
we have solved the problem of rest frames in the past with the construction of the so called nice sections@xcite .
given an asymptotically flat spacetime , these sections provide a well defined notion of rest frames at future null infinity .
we have also proved that they have the expected physical properties@xcite(see comments in the next section . ) . in this article
we use this construction to give a definition of intrinsic angular momentum in general relativity , solving in this way the deficiencies of previous works@xcite .
in order to define rest frames , we need to fix the notion of supermomentum .
we will use the supermomentum ` psi ' that was used in the nice section construction@xcite . let the section @xmath17 of future null infinity be characterized by the condition @xmath18 , of the bondi coordinate system @xmath19 .
then the supermomentum is given by @xmath20 where @xmath21 is the surface element of the unit sphere on @xmath17 , @xmath22 are the spherical harmonics , the scalar @xmath23 is given by @xmath24 where we are using the ghp@xcite notation and where @xmath25 is the leading order asymptotic behavior of the second weyl tensor component , @xmath26 is the leading order of the bondi shear , @xmath27 is the edth operator of the unit sphere , and a dot means partial derivative with respect to the retarded time @xmath28 .
the condition for a section to be of the nice type is that all the spacelike components of the supermomentum vanish .
that is , if @xmath29 is nice then @xmath30 for @xmath31 .
since any section @xmath29 can be obtained from an arbitrary reference section @xmath17 by a supertranslation @xmath32 , it is useful to know what is the equivalent condition on @xmath33 .
we have shown that the condition for the section determined by @xmath33 to be of the nice type is@xcite @xmath34 where the conformal factor @xmath35 can be related to the bondi momentum by @xmath36 with @xmath37 and @xmath11 is evaluated at the section @xmath38 .
also , the rest mass @xmath39 at the same section is given by @xmath40@xcite .
it has been proved that there exists a four parameter family of solutions of equation ( [ nice ] ) with the expected physical properties@xcite .
an important property is that ; if @xmath41 is a nice section and @xmath42 is another nice section which is generated from @xmath41 by a timelike translation , then @xmath42 is to the future of @xmath41 .
this is exactly what happens in minkowski space when @xmath41 is the intersection of the future null cone emanating from an interior point , let us say @xmath43 , with future null infinity ; and if @xmath42 is the intersection of the future null cone corresponding to another point @xmath44 which is in the future of @xmath43 .
there is then an analogy between rest frames at future null infinity centered on a nice section @xmath17 and rest frames of minkowski spacetime centered on a point @xmath45 .
we will define intrinsic angular momentum by using the so called ` charge integrals of the riemann tensor'@xcite .
a charge integral of the riemann tensor is a quantity ascribed to a 2-sphere @xmath17 by the integration of the 2-form @xmath46 , namely : @xmath47 where @xmath46 is expressed in terms of the curvature by @xmath48 and where a right star means right dual of the riemann tensor and the 2-form @xmath49 is defined next . in our case
@xmath17 is assumed to be a nice section@xcite of future null infinity .
how is one supposed to chose the 2-form @xmath50 ?
it was discussed in the literature@xcite that , in terms of the spinor notation , it is convenient to require @xmath51 and @xmath52 where the vector @xmath53 is a generator of asymptotic symmetries(for the details behind this and other matters see the full length paper on this subject@xcite ) .
however on a non - stationary spacetime , these equations have in general no solution at future null infinity . in spite of
that we can require this equations to be satisfied on a nice section .
let @xmath54 and @xmath55 be the spinor components of the 2-form @xmath50 .
then we require @xmath56 where @xmath57 is a spin weight 0 quantity satisfying @xmath58 and @xmath59 , @xmath60 and @xmath61 are spin weight 0 and 1 functions respectively that solve the equations @xmath62 and @xmath63 using the potential @xmath64 of the shear satisfying @xmath65 , the component @xmath66 can be expressed by @xmath67 where the spin weight 0 quantity @xmath68 satisfies @xmath69 and @xmath70 . having a timelike one - parameter family of nice sections at future null infinity , this procedure provides with a 2-form @xmath50 at future null infinity with the functional dependence @xmath71 then the charge integral becomes @xmath72 in order to pick up the intrinsic angular momentum we need to impose the center of mass condition@xcite @xmath73 this is the analog of the fixing of the translation freedom in the minkowskian case by requiring that three components of the angular momentum tensor vanish , namely @xmath74 . in this way
, there is left a unique one parameter family of nice sections , which determines the center of mass frames .
let us denote these sections with @xmath75 ; then the intrinsic angular momentum @xmath76 is defined by @xmath77 where in order to pick up the intrinsic angular momentum one must take @xmath78 and @xmath79@xcite .
several advantages exist when one uses the charge integral approach for the definition of physical quantities .
for example , although we have defined the tensor @xmath80 at future null infinity , let us consider the possibility to extend its definition to the interior of the spacetime .
assume @xmath81 is a spacelike hypersurface in the interior of the spacetime which extends up to future null infinity , and which has as boundary the section @xmath17 ; as shown in figure [ fig : inte ] .
has as boundary the section _ s _ at future null infinity .
, scaledwidth=45.0% ] then , using stokes theorem , one can express the charge integral @xmath82 as an integral on @xmath81 , namely @xmath83 one can prove@xcite that the exterior derivative of @xmath80 can be expressed by @xmath84 where @xmath85 is the einstein tensor , @xmath86 is the traceless part of @xmath87 and @xmath88 , its trace ; in other words @xmath89 let us look at equation ( [ eq : dc2 ] ) in the linearized gravity case .
then , it is noted@xcite that if the vector @xmath90 were a killing vector of the flat background metric , and @xmath86 were @xmath91 , then equation ( [ eq : qonsigma ] ) will give the conserved quantities in the context of linearized gravity .
this is telling us that the charge integrals have the appropriate meaning in the weak field limit .
also one can use stokes theorem to calculate the flux of angular momentum .
let @xmath42 be a section to the future of the section @xmath41 of future null infinity , and let now @xmath92 be the region which has as boundaries @xmath41 and @xmath42 ; see figure [ fig : flux ] .
then the flux law for angular momentum is given by @xmath93 at future null infinity has as boundary the sections @xmath41 and @xmath42 .
, scaledwidth=45.0% ] when the spacetime is stationary , it can be seen@xcite that our definition does give the intrinsic angular momentum of the spacetime ; and also that the center of mass , defined in terms of nice sections , is what one expects to be .
let us consider a radiative spacetime in which one can distinguish three stages ; starting with a stationary regime , passing through a radiating stage and ending in a stationary regime .
one can see that our construction gives the correct intrinsic angular momentum in the first and third stages ; even though the center of mass frames of the first and third stages are related in general by a supertranslation .
it should be stressed that given a point at future null infinity , which it could be considered as the event of the detection of gravitational waves , then the center of mass sections just defined provides with a unique section where all physical quantities should be defined .
a couple of previous works@xcite also provide with a prescription for the selection of unique sections at future null infinity , however they are either non - local@xcite or they do not mention the subject of supermomentum@xcite and therefore can not be considered as either rest frames nor center of mass .
it is probably worth mentioning that several estimates on the gravitational radiation of different astrophysical systems are done with the quadrupole radiation formula .
if one needs these estimates at two different times , such that back reaction has occurred due to the emission of gravitational waves between the first and second time , then the multipoles should be calculated at the corresponding center of mass for each time .
in other words , even in the use of the quadrupole radiation formula one needs to have a well defined notion of center of mass .
our work provides both notions ` center of mass ' and ` intrinsic angular momentum ' in one construction .
we would like to thank dr .
a. rizzi for ongoing discussions on his work and dr . c. kozameh for help in these matters .
we acknowledge support from secyt - unc and conicet . | there are several definitions of the notion of angular momentum in general relativity .
however none of them can be said to capture the physical notion of intrinsic angular momentum of the sources in the presence of gravitational radiation .
we present a definition which is appropriate for the description of intrinsic angular momentum in radiative spacetimes .
this notion is required in calculations involving radiation of angular momentum , as for example is expected in binary coalescence of black holes .
pacs numbers : 04.20.-q , 04.20.cv , 04.20.ha |
since abrikosov s @xcite prediction of the flux - line lattice in type - ii superconductors from ginzburg - landau ( gl ) theory @xcite , several approximate formulas for the magnetization @xmath6versus the applied magnetic field @xmath7 or average induction @xmath8 have been published @xcite . in these papers and below , the basic situation is considered where a macroscopically large , homogeneous and isotropic , long superconductor is exposed to a uniform parallel field @xmath7 . in this ideal case demagnetization effects , flux - line pinning , and surface effects may be disregarded , and thus the flux - lines are straight lines forming an ideal periodic lattice .
these results are easily extended to anisotropic superconductors ( where an anisotropic effective - mass tensor is introduced into the gl theory ) by defining an effective gl parameter @xmath9 that depends on the orientation of the flux lines ; this transformation works when @xmath7 is along a principle symmetry axis @xcite .
generalization to geometries where demagnetization effects occur , are possible by introduction of a demagnetizing factor ; but this concept works only for homogeneous specimens with the shape of an ellipsoid . in this case
the flux lines in the bulk are still straight and form an ideal flux - line lattice ( fll ) . for other specimen shapes the fll is distorted , i.e. , the orientation and density of the fll varies spatially and can be calculated only numerically @xcite .
the aim of the present paper is to compare the widely used approximate expressions for @xmath5 with the exact value obtained numerically and to give useful general analytic interpolation formulas valid in the entire ranges of @xmath7 and @xmath10 where the fll exists , namely , @xmath1 for @xmath7 , or @xmath11 , and @xmath12 for @xmath10 , where @xmath13 and @xmath14 are the lower and upper critical fields and @xmath10 is the gl parameter .
interestingly , such general formulas have not been published yet , and thus the accuracy of the commonly used expressions is not known , probably due to the difficulty of the numerical solution of the complex - valued gl equations .
early numerics @xcite used the circular cell method ( ccm ) , which approximates the hexagonal unit cell of the triangular fll ( or the quadratic unit cell of the square fll ) by a circle and the two - dimensional ( 2d ) solution by the 1d rotationally symmetric solution inside this circular cell ; both the gl function and the magnetic field are forced to have vanishing slope on this circular boundary , as the exact solution has on the boundary of the wigner - seitz cell .
this method yields the exact @xmath15 and is expected to be best at low inductions @xmath16 where the flux lines are well separated .
but surprisingly , the circular cell approximation gives very good magnetization curves at all @xmath8 ( see fig . 1 ) and even yields the exact value of the upper critical field @xmath17
. some more exact results of the ccm are listed below .
another method @xcite uses a similar circular symmetric gl order parameter and a linear superposition of circular symmetric magnetic fields to obtain excellent approximate @xmath5 , see also ref . .
an in principle exact numerical method @xcite uses periodic real trial functions for the squared gl function @xmath18 and magnetic field @xmath4 and minimizes the resulting free energy functional with respect to a finite number of fourier coefficients .
the same method was later applied @xcite to solve the microscopic bcs - gorkov theory for the properties of the fll in the entire temperature interval @xmath19 where @xmath20 is the superconducting transition temperature ( gl theory , strictly spoken , applies only close to @xmath20 ) . recently
this variational method was improved @xcite by keeping the same periodic trial functions but now solving the gl equations iteratively ; this iteration works much faster and allows to use many more fourier coefficients ( many thousands instead of only five in ref . ) .
i shall use this 2d iterative precision method of ref . for the calculation of the fll at @xmath21 . at low inductions @xmath16
this 2d method is supplemented by an iterative circular cell method presented in appendix a. this 1d method yields accurate values of @xmath22 , which then can be used in interpolation formulas . for convenience ,
i introduce the reduced fields @xmath23 , @xmath24 , @xmath25 , such that one has @xmath26 , @xmath27 , @xmath28 , and @xmath29 .
for completeness it should be mentioned that the isolated vortex @xcite and the fll @xcite have also been computed from bcs theory ( valid at all temperatures ) using the quasiclassical eilenberger theory based on energy - integrated green functions .
this method was recently extended to compute the fll structure and local density of states for s - wave @xcite , d - wave @xcite , and chiral p - wave @xcite superconductors .
very recently the gl method @xcite was generalized phenomenologically to lower temperatures and to charged vortices @xcite .
magnetization curves of the triangular fll , which coincide within line thickness with the results for the square fll and for the fll obtained from the circular cell approximation , see fig . 3 for the difference .
shown are @xmath30 versus @xmath23 ( upper left triangle ) and @xmath31 versus @xmath32 ( lower right triangle ) .
one has @xmath33 .
the lower panel shows an enlarged scale .
the solid lines show the exact numerical result .
the dotted lines show the simple interpolation eq .
( 22 ) good for @xmath34 ( upper panel ) , and the combined low and high field limit eq .
( 23 ) good for @xmath35 ( lower panel ) . ]
the properties of the fll within gl theory are calculated by minimizing the gl free energy of the superconductor with respect to the complex gl function @xmath36 and to the vector potential @xmath37 of the local magnetic field @xmath38 . in the usual reduced units @xcite ( length @xmath39 , magnetic field @xmath40 , energy density @xmath41 , where @xmath42 is the thermodynamic critical field ) the spatially averaged free energy @xmath43 of the gl theory referred to the meissner state ( @xmath44 , @xmath45 ) reads @xmath46 here @xmath47 means spatial averaging over the superconductor of volume @xmath48 . introducing the supervelocity @xmath49 and the magnitude @xmath50 of @xmath51
$ ] one may express @xmath43 as a functional of the real and gauge invariant functions @xmath52 and @xmath53 , @xmath54 in the presence of vortices @xmath55 has to be chosen such that @xmath56 has the appropriate singularities along the vortex cores , see e.g. eq.(b4 ) in app .
b. in this paper i consider the ideal periodic fll in a homogeneous ( pin - free ) large superconductor in a uniform magnetic field @xmath7 along @xmath57 . in this 2d situation one
has @xmath58 , @xmath59 , and @xmath60 . within gl theory in reduced units
the properties of this ideal fll depend only on two parameters : the gl parameter @xmath10 and the average induction @xmath61 .
the equilibrium magnetic field @xmath7 , and the magnetization @xmath6 , are obtained either from the definition @xmath62 or , more elegantly , from the virial theorem discovered by doria , gubernatis , and rainer , @xcite which in reduced units reads @xmath63 some of the properties of the fll , and all properties of the isolated flux line , may be calculated in an elegant way by the circular cell approximation @xcite as described in app .
a. in the circular cell method the hexagonal wigner - seitz cell around each flux line is replaced by a circle with radius @xmath64 and same area @xmath65 if each flux line carries one quantum of flux @xmath66 tm@xmath67 . in reduced units
one has @xmath68 and @xmath69 with @xmath70 .
the boundary conditions on the ccm circle @xmath71 are @xmath72 .
i find that the free energy of the triangular fll , @xmath73 , and its magnetization , @xmath74 , are reproduced by the ccm with high accuracy in the entire ranges of @xmath10 and @xmath8 , @xmath75 and @xmath76 .
in particular , the ccm not only yields @xmath15 ( in the limit @xmath77 ) but it also reproduces the exact upper critical field @xmath78 , and in the special case @xmath79 even the exact result @xmath80 .
these somewhat surprising features of this approximation are related to the facts that @xmath17 , and in the case @xmath81 even the entire curve @xmath82 , are _ independent _ of the detailed arrangement of the flux lines , i.e. , they are the same for triangular and square or honey - comb flls and for any other arrangement of single or multiple quanta flux lines .
another surprising finding is that the virial theorem , eq .
( 3 ) , works perfectly in the ccm .
figure 1 shows the magnetization curves @xmath0 and the equilibrium field @xmath82 of the superconductor obtained by the ccm for @xmath83 , 1 , 1.2 , 1.5 , 2 , 3 , 5 , 7 , 10 , and 20 .
in the limit @xmath84 the ccm yields the lower critical field @xmath15 , which with high accuracy is fitted by the formula @xmath85 \ , , \nonumber \\
\nonumber h_{c1 } = { h_{c1 } \over h_{c2 } } = { \ln \kappa + \alpha(\kappa ) \over 2\kappa^2 } \ , , \\
\alpha(\kappa ) = \alpha_\infty + \exp [ - c_0 - c_1 \ln\kappa - c_2 ( \ln\kappa)^2 ] \pm \epsilon % \alpha(\kappa ) = 0.49693 + \exp[-0.775\ln\kappa % \\ \nonumber -0.1303(\ln\kappa)^2 - .41577 ] \pm 0.00076 \end{aligned}\ ] ] with @xmath86 , @xmath87 , @xmath88 , @xmath89 , and @xmath90 .
this expression yields at @xmath91 the correct value @xmath92 and for @xmath93 it has the limit @xmath94 .
a simpler expression for @xmath95 , yielding an @xmath96 with error still less than 1% and with the correct limits at @xmath97 and @xmath93 , is @xmath98 the ccm in principle can not yield properties related to the different symmetries of the fll , or to its shear modulus , and it can not give the form factors ( fourier coefficients ) of the magnetic field @xmath4 that may be measured by neutron scattering .
these subtle properties can be computed by the 2d method presented in ref . and in app .
b. this effective numerical method expresses the smooth functions
@xmath99 and @xmath4 as 2d fourier series and determines the fourier coefficients by iteration .
figure 2 ( top ) shows the difference of the free energy densities of the triangular ( @xmath73 ) and square ( @xmath100 ) flls .
this difference is proportional to the shear modulus @xmath101 of the triangular fll ( the shear modulus of the unstable square fll is negative within gl theory ) by the relation @xcite @xmath102 note that this difference is very small , @xmath103 .
( by ten times ) is the difference between the free energy densities of the ccm ( @xmath104 ) and of the triangular fll plotted in fig . 2 ( bottom ) .
one has @xmath105 .
this result shows that the ccm is an excellent approximation for global properties of the fll .
both differences are largest for large @xmath10 and have a maximum near @xmath106 .
the finding @xmath107 means that the triangular fll is stable for all @xmath108 .
note that for @xmath91 one has exactly @xmath109 for all @xmath110 .
figure 3 ( top ) shows the difference between the magnetizations @xmath111 of the square fll and @xmath74 of the triangular fll .
again , this difference is small , @xmath112 and the relative difference has the limits @xmath113 .
figure 3 ( bottom ) shows the difference between the magnetization @xmath114 obtained by the ccm ( see fig . 1 ) and the exact value @xmath74 of the triangular lattice .
like with the free energy , this difference is again smaller by a factor of ten than the difference between two lattice symmetries , @xmath115 and @xmath116 .
the differences vanish exactly at @xmath91 , and also at @xmath117 , since there @xmath118 .
the _ relative _ differences ( insets in fig . 3 )
are maximum at @xmath93 .
the smallness of these differences explains why in fig .
1 the magnetization curves for all three cases @xmath74 , @xmath111 , and @xmath114 coincide within line thickness .
top : the difference of the free energy densities of the triangular ( @xmath73 ) and square ( @xmath100 ) flls in units @xmath41 , plotted versus the reduced induction @xmath23 for @xmath83 , 1 , 1.2 , 1.5 , 2 , 3 , 5 , 10 , and 200 .
this difference equals @xmath119 times the shear modulus @xmath101 of the triangular fll .
bottom : the very small difference between the free energy densities of the circular cell method ( @xmath104 ) and of the triangular fll .
note that the top and bottom plots look similar , but the scales of the ordinate differ by a factor of about ten . ]
top : the difference between the magnetizations @xmath111 of the square fll and @xmath74 of the triangular fll in units @xmath17 , plotted versus the reduced induction @xmath23 for @xmath83 , 1 , 1.2 , 1.5 , 2 , 3 , 5 , 10 , and 200 .
the inset shows the relative difference .
bottom : the difference between the magnetization @xmath114 obtained by the ccm ( see fig . 1 ) and the exact value @xmath74 of the triangular lattice .
the inset shows the relative difference . ]
comparison of the gl functions @xmath52 and magnetic inductions @xmath8 calculated for the triangular fll and from the circular cell approximation for the example @xmath120 , @xmath121 . shown
are the cross sections @xmath122 , @xmath123 along the nearest neighbor direction @xmath124 , and @xmath125 , @xmath126 along the perpendicular direction @xmath127 , and @xmath128 , @xmath129 from the ccm .
all @xmath8 are in units @xmath130 of the triangular fll .
small deviations can be seen only close to the cell boundary @xmath71 , @xmath131 . at lower @xmath110
the deviations are even smaller . ]
maximum field @xmath132 minus applied field @xmath7 , saddle - point field @xmath133 minus induction @xmath8 , and @xmath134 minus minimum field @xmath135 , for the triangular fll , plotted versus @xmath23 for @xmath136 , 1 , 1.2 , 1.5 , 2 , 3 , 5 , 7 , 10 , 20 , 50 , 100 , 200 .
the solid lines show these small differences in units @xmath137 , multiplied by appropriate functions of @xmath10 to obtain collapse of the curves near @xmath138 .
the dashed lines show the same functions multiplied by factors @xmath139 and @xmath140 such that they tend to a finite constant value near @xmath138 , cf .
( 6 ) - ( 8) . ]
figure 4 shows an example ( @xmath120 , @xmath121 ) comparing the spatial functions @xmath52 and @xmath8 of the triangular fll and from the ccm .
shown are the cross sections @xmath122 along the nearest neighbor direction @xmath124 and @xmath125 perpendicular to this , and @xmath128 from the ccm [ @xmath141 is the vortex spacing , @xmath142 . it is seen that @xmath122 and @xmath128 , and also @xmath123 and @xmath129 , coincide closely ; at lower @xmath143 the difference is smaller than the line thickness . the solutions for the square fll deviate more from the circular cell solutions .
the maximum , minimum , and saddle - point fields of the triangular fll , @xmath144 , @xmath145 , and @xmath146 , depend on @xmath110 and @xmath10 .
@xmath147 is only slightly above the equilibrium field @xmath7 , and @xmath134 and @xmath148 are close to each other and lie somewhat below the average field @xmath8 . @xmath147 and @xmath148 are shown in fig . 3 of ref . as functions of @xmath110 for several @xmath149 . in fig .
5 the small differences @xmath150 , @xmath151 , and @xmath152 are plotted versus @xmath110 , in units @xmath137 and multiplied by a function of @xmath10 such that the curves for all @xmath153 collapse at @xmath154 .
one finds for all @xmath10 near @xmath138 : @xmath155 the factor 0.069 in eqs .
( 6)(8 ) is @xmath156 where @xmath157 is the abrikosov parameter of the triangular fll .
plots of @xmath158 where @xmath159 is the field value at the boundary of the circular cell in the ccm , look similar to the plots of @xmath152 in fig . 5 ( lower panel ) ,
since the value @xmath159 lies approximately in the middle between @xmath148 and @xmath134 , see fig .
since for @xmath93 and @xmath160 the field in the vortex center equals @xmath161 , one has @xmath162 , and thus the function plotted in fig .
5 ( upper panel ) for @xmath163 tends to the limit @xmath164 , cf .
( 4 ) .
the magnetic field variance @xmath165 ^ 2 \rangle$ ] of the triangular fll for @xmath83 , 1 , 1.2 , 1.5 , 2 , 3 , 5 , 7 , 10 , 20 , 50 , 100 , 200 plotted in units of @xmath137 as @xmath166 ( solid lines ) such that the curves for all @xmath10 collapse near @xmath138 , cf .
the dashed lines show the same functions divided by @xmath167 such that they tend to a finite constant 0.172 at @xmath138 .
all curves are plotted versus @xmath168 to stretch them at small @xmath110 values and show that they go to zero linearly .
the limits , eq . ( 12 ) , for @xmath169 and @xmath170 are depicted as dash - dotted straight lines .
the upper frame 0.383 shows the approximation ( 11 ) . ]
lower panel : the exact magnetization @xmath171 of the triangular fll ( solid lines ) and the fit , eq .
( 15 ) , ( solid lines with dots ) plotted for many @xmath10 values versus @xmath168 to stretch the low field region .
shown is @xmath172 normalized to its maximum value @xmath15 occurring at @xmath173 .
the fit ( 15 ) is good for all @xmath10 and not too small @xmath174 .
upper panel : the deviation @xmath175 of the fit from the exact @xmath171 is very small when @xmath176 . the dotted lines in the lower panel show the old london approximation , eq .
( 18 ) . ]
exact magnetization of the triangular fll ( solid lines ) and the logarithmic fit , eq .
( 19 ) , ( solid lines with dots ) plotted versus @xmath177 to stretch the region at small @xmath23 .
the dotted lines show the london nearest - neighbor approximation , eq .
the dashed lines show the london expression , eq .
( c6 ) , with the sum taken over all shells up to @xmath178 vortex spacings .
both london approximations are good fits at very low @xmath110 and all @xmath10 . ]
the variance of the magnetic field is @xmath179 ^ 2 \rangle = \langle b(x , y)^2\ ! -b^2 \rangle = \!\sum_{{\bf k } \ne 0}\!\ ! b_{\bf k}^2 \ , , \end{aligned}\ ] ] where @xmath180 are the fourier coefficients of @xmath181 and @xmath182 the vectors of the reciprocal lattice of the fll ( app .
b ) . near @xmath138
the abrikosov solution of the linearized gl theory @xcite yields for all @xmath10 values @xcite @xmath183 the functions @xmath184 and @xmath185 are plotted in fig . 6 versus @xmath186 for various @xmath10 .
it can be seen that eq .
( 10 ) is a rather good approximation for the large range @xmath187 . at smaller @xmath110 the variance
@xmath188 has a maximum and then goes to zero again at @xmath173 . for small inductions @xmath189 and large @xmath10 one can use the london approximation @xmath190 . for
the appropriate cutoff at large magnitudes @xmath191 see ref . and below .
in the range @xmath192 the unity in the denominator of @xmath180 may be disregarded since @xmath193 .
thus @xmath8 drops out and @xmath194 becomes independent of @xmath110 : @xcite @xmath195 this often used formula corresponds to the upper axis in fig .
one can see that this approximation is good only for very large @xmath196 and only in the range of @xmath110 near the maximum of @xmath194 .
at very small @xmath197 both @xmath188 and @xmath198 drop linearly to zero when @xmath163 . in this range
the sum in eq .
( 9 ) can be evaluated as an integral , yielding @xmath199 this approximation is good for @xmath200 and very small @xmath110 ( @xmath201 for @xmath202 ; @xmath203 for @xmath204 ) , see the two straight lines in fig .
for @xmath200 a good approximation , with less than 5% error from @xmath138 down to the value @xmath205 where the maximum of @xmath194 occurs , is @xmath206 \,.\ ] ] this approximation is much better that the interpolation , eq .
( 3 ) of ref . .
this section presents analytic expressions which approximate the computed magnetization @xmath207 ( fig .
1 ) as a function of the induction @xmath23 or of the thermodynamic field @xmath24 . we distinguish approximations working at high or low inductions .
the linearized gl theory yields for @xmath208 abrikosov s @xmath137 solution @xcite @xmath209 where @xmath210 $ ] ( ref . and app .
b ) is the abrikosov parameter , @xmath211 ( @xmath212 ) for the triangular ( square ) fll .
the linear magnetization @xmath213 is a good approximation in the range @xmath214 , see fig . 1 .
this suggests the following fit to the exact @xmath215 : @xmath216\ , g_1(\kappa ) + \epsilon_1 , \nonumber \\
f_1(b ) = 2.50u^2 - 8.08u+0.39 , ~
u=(1-b)^{0.41 } , \nonumber \\
g_1(\kappa ) = ( 1.133 + 1.926 / \kappa^{2.25 } ) ( 2\kappa^2 -1)/ ( 2\kappa^4 ) \,.\ , \end{aligned}\ ] ] with relative error @xmath217 for @xmath176 for the triangular fll . formula ( 15 ) is a good approximation with relative error @xmath218 for all @xmath10 in the large range of fields @xmath219 , see fig .
the same expression ( 14 ) fits also the @xmath220 of the square fll , with somewhat larger error if the same functions @xmath221 , @xmath222 are used rather than the optimally fitted ones . for
the difference @xmath223 see fig .
3 . for completeness
i mention here also the london approximation @xcite which was supposed to be good in the `` intermediate field range '' @xmath224 that exists only in superconductors with extremely large @xmath10 .
within london theory the induction is ( see app .
b ) @xmath225 where the sum goes over all @xmath226-vectors with length from @xmath227 to some cutoff @xmath228 . inserting this into the london free energy density @xmath229 /(2\mu_0)$ ] and averaging over the superconductor one
gets @xmath230 the integral from @xmath231 to @xmath232 equals @xmath233 where @xmath234 is some constant and @xmath23 as above .
this yields @xmath235 with constant @xmath236 obtained by our fit to the numerical @xmath237 at @xmath238 .
this old london approximation is shown in fig .
7 as dotted lines .
one sees that this fit works only at large @xmath239 in the relatively small interval @xmath240 , i.e. at very low @xmath110 ( but not too low @xmath110 ) .
it gives @xmath241 at @xmath242 for all @xmath10 .
this fit is slightly improved by replacing @xmath243 by @xmath244 , which gives the correct @xmath241 at @xmath138 .
a much better fit in the spirit of this logarithmic approximation is ( see fig .
8) @xmath245 \ , , \nonumber \\
% g_2(b ) & = & 1.666 u + 0.273 u^2 -1.581 u^3 , ~ u=1-b \,,~~ f_2(b ) & = & 0.357 + 2.890\,b-1.581\,b^2 \,.~~ \end{aligned}\ ] ] this fit is good for @xmath246 ( error @xmath247 ) and @xmath200 ( error @xmath248 ) in the large ranges @xmath249 for @xmath250 .
these intervals of validity may also be expressed as @xmath251 ( 0.85 ) for @xmath252 ( @xmath253 ) .
all the above approximations do not describe the correct vertical slope of @xmath0 at @xmath254 , or zero slope of @xmath82 and unity slope of @xmath255 at @xmath256 .
this is achieved by the london approximation of pairwise interacting vortices described in app .
c. for very small @xmath257 one may account only for the nearest neighbor shell of six vortices in the triangular fll of spacing @xmath258 . with @xmath259 ,
( c8 ) , this yields for @xmath260 : @xmath261 , \nonumber \\ c = { a \over \lambda } = \left ( { 4\pi / \sqrt3 \over b \kappa^2 } \right)^{1/2 } .
\end{aligned}\ ] ] formula ( 20 ) correctly describes the steep diverging slope of @xmath262 or slopes @xmath263 and @xmath264 as @xmath84 and is valid for @xmath265 for @xmath266 , accidentally it also fits well @xmath237 for @xmath267 and @xmath268 , see the dotted lines in fig . 8 .
a smoother fit is obtained by the exact london expression ( c6 ) if one or three neighbor shells are included in the sum . but taking more terms in the sum improves the fit only at large @xmath10 .
accounting for neighbors up to @xmath269 lattice spacings apart ( about 5000 terms ) one gets good approximation to @xmath215 and @xmath32 for @xmath270 ( 0.02 , 0.05 ) if @xmath239 ( @xmath271 , @xmath272 ) , see fig.8 . in the limit @xmath273 the infinite sum ( c6 ) reproduced eq . ( 18 ) , i.e , the dashed curves in fig .
8 for @xmath274 then will straighten and cut the axis @xmath275 at @xmath276 ( @xmath277 in fig .
7 , @xmath278 in fig . 8) .
all the approximations for @xmath237 and @xmath259 known so far , including the above formulas , fit either the low or high field regions .
the formulas ( 15 ) and ( 20 ) [ or better , eq . ( c6 ) with the sum taken over three shells ] have a small overlap for all @xmath10 and thus , together , they fit the entire range @xmath279 [ though the good fit of the low-@xmath10 data by the london expression ( 20 ) or ( c6 ) is accidental ] . for practical purposes one may construct interpolation formulas that approximate the numerically obtained magnetization in the entire range @xmath280 .
they should satisfy the five conditions @xmath281^{-1}\,,~ \end{aligned}\ ] ] with @xmath282 from eq .
a simple expression that satisfies all these conditions is @xmath283 with @xmath284 .
formula ( 22 ) approximates the exact @xmath285 well for @xmath267 with relative deviation @xmath286 , for @xmath287 with @xmath288 , and for @xmath289 with @xmath290 , see the dotted lines in fig . 1 , top . for large @xmath10 , general interpolation formula are more difficult to construct because of the nonanalytic limiting expression eq . ( 20 ) .
one may , however , combine the @xmath291 from eq .
( 20 ) with the @xmath292 from eq .
( 19 ) using a smooth transition at @xmath293 , e.g. , with weights @xmath294 and @xmath295 $ ] , or slightly better , @xmath296 $ ] , yielding @xmath297 this interpolation between expressions ( 19 ) and ( 20 ) works well for @xmath298 with relative error @xmath299 for @xmath200 and @xmath300 for @xmath301 , see the dotted lines in fig . 1 ( bottom ) .
thus , @xmath215 in the entire ranges of @xmath110 and @xmath10 may be approximated by eq .
( 22 ) or eq .
the properties of the ideal periodic flux - line lattice in superconductors are calculated from ginzburg - landau theory for the entire ranges of gl parameters @xmath75 and inductions @xmath302 .
the differences between the free energies and magnetizations of the triangular and square vortex lattices and the values obtained by the circular cell approximation are investigated in detail .
approximate analytical expressions are given for the variance @xmath303 of the periodic induction and for the magnetization @xmath304 .
these limiting and interpolation formulas should replace previous approximate expressions that have rather limited validity .
the numerical methods presented in the appendices , in principle may be applied also to theories going beyond the isotropic gl theory considered here .
the calculation of the isolated flux line and of the fll within the circular cell method , is a cylindrically symmetric problem .
the free energy depends on the magnitude of the gl function @xmath128 and on the magnetic induction @xmath129 ( along @xmath57 ) related to the vector potential @xmath305 and supervelocity @xmath306 ( along @xmath307 ) by @xmath308 in reduced units @xmath309 , the free energy of a flux line or of the circular cell with radius @xmath64 ( @xmath310 ) averaged over this cell and referred to the meissner state ( @xmath311 , @xmath256 ) reads @xmath312 { 2\pi r\,dr \over \pi r^2}\ ] ] with @xmath313 . minimizing the functional ( a2 ) with respect to @xmath128 and @xmath306
we obtain the two gl equations , which may be written in the form @xmath314 where @xmath315 is the current density . in eq .
( a3 ) a term @xmath316 was added on both sides to improve the convergence of the iteration below .
the boundary conditions are @xmath317 an appropriate ansatz in terms of fourier series is @xmath318 for the equidistant grid @xmath319 , @xmath320 , one has the orthogonality relation @xmath321 and similar equations for @xmath322 and @xmath323 .
the gl equations ( a3 ) and ( a4 ) thus may be written in the form of equations for the fourier coefficients @xmath324 and @xmath325 : @xmath326 } \
, , \\ a_k = { 1\over k^2 + 1 } \big [ a_k + { 2\over n_r } \sum_{i=1}^{n_r } \sin kr_i \times \nonumber \\
\big ( \sum_{n'=1}^n a_{k ' } { kr\cos kr - \sin kr \over r^2 } -f^2 q \big ) \big ] \ , . \end{aligned}\ ] ] these two equations may be used to obtain the @xmath324 and @xmath325 by iteration , starting with appropriate initial values .
the iteration becomes more stable and faster if the value of the previous iteration step is added with a certain weight @xmath327 , e.g. , c=0.6 , according to the algorithm : @xmath328 with the symbols @xmath329 and @xmath330 denoting the right hand sides of eqs .
( a12 ) and ( a13 ) , respectively .
rapid convergence is achieved by iterating equations ( a14 ) , ( a15 ) alternately .
the equilibrium magnetic field @xmath7 is then obtained from eq .
( 3 ) , and the magnetization from @xmath331 r \,dr \ , . \end{aligned}\ ] ] at very large @xmath10 and very small @xmath110 a large number @xmath332 of grid points @xmath333 is needed to achieve high accuracy , @xmath334 . in this case
the accuracy with a limited number of grid points may be improved by choosing a nonequidistant grid , e.g. , @xmath335 with equidistant @xmath336 . to use the orthogonality relations
one then has to express @xmath52 , @xmath8 , and @xmath337 as fourier series in the new variable @xmath338 and also write the two gl equations in terms of the variable @xmath339 , using , e.g. , @xmath340 and @xmath341 .
this yields @xmath342 and the fourier series @xmath343 the equations for the new fourier coefficients are @xmath344 \,,\,~ \\ a_k = { 1\over k^2 + 1 } \big [ a_k + { 2\over n_r } \sum_{i=1}^{n_r } \sin ku_i \times \nonumber \\ \big ( \ ! \sum_{n'=1}^n
\!a_{k ' } { ku_i\cos ku_i - 4\sin ku_i \over u_i^2 } -4 u^2 q f^2 \big ) \big ] \,.\,~ \end{aligned}\ ] ] for better convergence a term @xmath345 was added on both sides of eq .
( a17 ) to yield ( a24 ) .
the corresponding iteration scheme using ( a14 ) , ( a15 ) needs a smaller weight @xmath346 and more iteration steps , but for large @xmath347 it is faster than the first scheme since it needs less grid points @xmath332 to reach the same accuracy .
the properties of the ideally periodic fll within gl theory may be calculated by minimizing the gl free energy of the superconductor , eq .
( 2 ) , with respect to appropriate periodic trial functions , e.g. , fourier series with a large number of terms . for the smooth function @xmath348 we write the ansatz @xmath349 with @xmath350 , @xmath351 . in all sums here and below the term @xmath352
is excluded . for vortex positions
@xmath353 the reciprocal lattice vectors are @xmath354 with @xmath355 the unit cell area and @xmath356 . for the triangular lattice one
has @xmath357 , @xmath358 , and for the square lattice @xmath359 , @xmath360 . for supervelocity @xmath53 and induction @xmath361
we choose @xmath362 here @xmath363 is the supervelocity of the abrikosov @xmath137 solution , which satisfies @xmath364 { \bf\hat z}\ , , \end{aligned}\ ] ] where @xmath365 is the 2d delta function .
this relation shows that @xmath366 is the velocity field of a lattice of ideal vortex lines but with zero average rotation .
close to each vortex center one has @xmath367 and @xmath368 with @xmath369 . in principle @xmath370
may be expressed as a slowly converging fourier series by integrating ( b4 ) using @xmath371 as in ref . .
but it is more convenient to take @xmath366 from the exact relation @xmath372 where @xmath373 is the abrikosov @xmath137 solution given by the rapidly converging series ( b1 ) with coefficients @xcite @xmath374 \end{aligned}\ ] ] for general lattice symmetry , and @xmath375 ( @xmath376 ) for the triangular lattice .
this @xmath377 is normalized to @xmath378 , which means that @xmath379 for any lattice symmetry .
another strange property of the abrikosov solution ( b6 ) is that @xmath380 , although both terms diverge at the vortex positions ; this relation follows from ( b4 ) and ( b5 ) using @xmath381 .
the useful formula ( b5 ) may be proven via the complex @xmath137 solution @xmath382 ; it means that near @xmath137 the third and fourth term in @xmath43 , eq .
( 2 ) , are identical .
approximate solutions @xmath383 and @xmath384 may be computed by using a finite number of fourier coefficients @xmath385 and @xmath386 and minimizing the free energy @xmath387 with respect to these coefficients @xcite . however , a much faster and more accurate solution method @xcite is to iterate the two gl equations @xmath388 and @xmath389 written in appropriate form .
namely , the iteration is stable and converges rapidly if one isolates a term @xmath390 on the l.h.s . and puts the remaining terms to the r.h.s . as a kind of `` inhomogeneity '' of such london - like equations , e.g. , @xmath391 with the abbreviations @xmath392 , @xmath393 , @xmath394 , and @xmath395 .
equations ( b7 ) , ( b8 ) introduced some `` penetration depths '' @xmath396 and @xmath397 ( in real units ) , which stabilize the convergence of the iteration . acting on the fourier series @xmath398 ( b1 ) and @xmath399 ( b3 ) the laplacian @xmath400 yields a factor @xmath401 ; this facilitates the inversion of ( b7 ) and ( b8 ) . using the orthonormality @xmath402 ( for @xmath403 ) one obtains from ( b1 ) , ( b2 ) @xmath404 and @xmath405 .
the convergence of the iteration is considerably improved by adding a third equation which minimizes @xmath43 , eq .
( 2 ) , with respect to the amplitude of @xmath398 , i.e. , @xmath406 .
this step gives the largest decrease of @xmath43 .
the resulting three iteration equations for the parameters @xmath407 and @xmath386 then read @xcite @xmath408 \cos{\bf k r } \rangle \over
k^2 + \bar \omega } \ , , \end{aligned}\ ] ] with @xmath409 and @xmath410 as above .
the solutions @xmath383 , @xmath411 , and @xmath412 are obtained by starting , e.g. , with @xmath413 and @xmath414 and then iterating the three equations ( b10 ) , ( b11 ) , ( b12 ) by turns until the coefficients do not change any more .
after typically 25 such triple steps , the solution stays constant to all 15 digits and the gl equations are exactly satisfied .
since all terms in ( b10 ) - ( b12 ) are smooth periodic functions of @xmath415 , high accuracy is achieved by using a regular spatial 2d grid , e.g. , @xmath416 ( @xmath417 ) and @xmath418 ( @xmath419 , @xmath420 ) with constant weights @xmath421 and @xmath422 .
these @xmath423 = 100 to 5000 grid points fill the rectangular basic area @xmath424 , @xmath425 , which is valid for any unit cell with the shape of a parallelogram .
spatial averaging @xmath426 then just means summing @xmath427 terms and dividing by @xmath427 .
best accuracy is achieved by considering all @xmath428 vectors within a half circle @xmath429 , with @xmath430 chosen such that the number of the @xmath428 is slightly less than the number @xmath427 of grid points .
the high precision of this method may be checked with the identity @xmath431 , which is valid at @xmath91 for all @xmath110 .
this relation is confirmed with an error @xmath432 .
the equilibrium field @xmath7 or reversible magnetization @xmath433 is computed from doria s virial theorem , eq .
the modified london equation for a lattice of straight vortex lines at regular positions @xmath434 ( app . b ) is @xmath435 where @xmath436 is the 2d delta function .
the solution for the magnetic field of one isolated vortex at @xmath437 is @xmath438 the modified bessel function @xmath439 has the derivative @xmath440 with the limits @xmath441 , @xmath442 , and for @xmath443 : @xcite @xmath444 for a periodic fll one obtains the fourier series @xmath4 , eq .
( 16 ) , which may also be written as a sum over isolated vortex fields , @xmath445 .
similarly , the free energy of the fll may be written as a sum of vortex self energies ( @xmath446 per unit length ) plus a double sum over all interactions between two vortices . the average energy density @xmath43 , eq.(17 )
, then reads @xmath447 for the triangular vortex lattice we write @xmath448 with @xmath449 ( @xmath141 = vortex spacing ) and @xmath450 1 , 3 , 4 , 7 , 7 , 9 , @xmath451 .
taking the derivative @xmath62 one obtains for @xmath24 with @xmath452 : @xmath453 .
\end{aligned}\ ] ] here the sum is over @xmath454 , i.e. the number of six flux lines per shell is already accounted for . equation ( c6 ) is still exact .
it works for @xmath257 ( i.e. for nonoverlapping vortex cores ) and for @xmath455 ( i.e. when the long - range interaction of vortices is purely magnetic @xcite ) . with the expansions ( c4 ) one obtains for @xmath456 : @xmath457 .
\end{aligned}\ ] ] at very small @xmath110 , namely for @xmath458 , the sum may be restricted to the nearest neighbor shell , i.e. to the first term , @xmath459 , yielding @xmath460 .
\end{aligned}\ ] ] | the magnetization curves @xmath0 for ideal type - ii superconductors and the maximum , minimum , and saddle point magnetic fields of the vortex lattice are calculated from ginzburg - landau theory for the entire ranges of applied magnetic fields @xmath1 or induction @xmath2 and ginzburg - landau parameters @xmath3 .
results for the triangular and square flux - line lattices are compared with the results of the circular cell approximation .
the exact magnetic field @xmath4 and magnetization @xmath5 are compared with often used approximate expressions , some of which deviate considerably or have limited validity .
useful limiting expressions and analytical interpolation formulas are presented . |
* funding . * we acknowledge financial support from the cnrs mission interdisciplinaire defi codim , aix - marseille university a*midex ( noanr-11-idex- 0001 - 02 ) , agence nationale de la recherche ( 10-insb-04 - 01 , 11-insb-0006 ) , iserm pc201508 , and from the national science foundation through awards phy-1068325 and phy-1203931 .
this work was partially supported by ircica usr 3380 univ .
lille - cnrs l. p. ramirez , m. hanna , g. bouwmans , h. el hamzaoui , m. bouazaoui , d. labat , k. delplace , j. pouysegur , f. guichard , p. rigaud , v. kermene , a. desfarges - berthelemot , a. barthelemy , f. prevost , l. lombard , y. zaouter , f. druon , and p. georges , opt .
express 23 , 5406 - 5416 ( 2015 ) . | we report an experimental test of single - shot polarimetry applied to the problem of real - time monitoring of the output polarization states in each core within a multicore fiber bundle .
the technique uses a stress - engineered optical element together with an analyzer and provides a point spread function whose shape unambiguously reveals the polarization state of a point source .
we implement this technique to monitor , simultaneously and in real time , the output polarization states of up to 180 single mode fibers cores in both conventional and polarization - maintaining fiber bundles .
we demonstrate also that the technique can be used to fully characterize the polarization properties of each individual fiber core including eigen - polarization states , phase delay and diattenuation .
multicore fiber ( mcf ) are important for a broad range of applications spanning over the fields of telecommunication @xcite , high power fiber lasers @xcite and medical endoscopes @xcite .
recently there has been a growing interest for coherent beam combining @xcite using mcfs both in the fields of broadband laser amplification @xcite and lens - free endoscopy @xcite . in coherent beam combining ,
an initial laser beam is , spectrally and/or spatially , split into @xmath0 beamlets that are launched and propagate into the @xmath0 individual cores of a mcf . at the exit of the bundle
, the different beamlets are recombined in a single beam and their relative phases are adjusted by a servo for coherent synthesis of an amplified ultrashort pulse or a focused spot , depending the application .
optimal temporal and/or spatial , focus at the bundle output requires to actively manage the relative beamlet phase and group delay @xcite together with their polarization state .
recent works report the use of polarization - maintaining @xmath1 mcf @xcite to ensure that the output beamlets exhibit identical polarization states for optimal spectral and/or spatial interference .
nevertheless , accurate core - by - core polarization state characterizations have not yet been carried out , neither the relative stability of these states during a fiber bend or twist .
such monitoring is important for calibration purposes and to infer the polarization origin of output power evolutions during fiber bends .
ideally one would like to monitor in real time the polarization state at the output of 10 to @xmath210000 fiber cores constituting the fiber bundle of interest .
this is quite challenging since standard polarimetry imaging techniques usually require time - sequential operations such as , for instance , analyzing the intensity of the light passing through a rotating quarter wave plate and a fixed polarizer @xcite . single shot polarimetry imaging has been also reported where the wavefront is divided and directed to different sensors @xcite , or where a clever combination of savart plates is used @xcite .
these single shot operations come with a significant technical complexity .
( a ) ) , for ( c ) horizontal , ( d ) linear at @xmath3 , ( e ) right - hand circular , ( f ) vertical , ( g ) linear at @xmath4 , ( f ) left - hand circular polarization states , where a representation of the polarization is overlaid with blue indicating linear polarization , and green and red indicating right- or left - handedness . ] a method of polarimetry was recently demonstrated that can infer complete polarization information using single - exposure images of discrete point - like light sources using an optical window subjected to stress with trigonal symmetry @xcite . this stress - engineered optical ( seo ) element @xcite provides , near the window s center fig .
[ fig1](a ) , a retardance that increases linearly with the radius and an orientation of the fast axis that precesses with the azimuth fig .
[ fig1](b ) .
the seo can be viewed as a spatially - varying wave plate : each point over the cross section of the test beam experiences a different combination of retardance and orientation .
therefore a collimated beam passing through the desired section of the seo ( the rest being blocked ) acquires an intricate polarization structure @xcite that , when observed at the focus of a lens , and after passing through a left - handed circular analyzer ( lhca ) , provides a point spread function ( psf ) whose shape unambiguously reveals the polarization state @xcite .
fig .
[ fig1](c - f ) shows measured psfs at the focal plane for the standard six reference polarization states .
as discussed in @xcite , the psf is a linear function of the stokes parameter vector , so the state of polarization ( including how polarized or unpolarized the light is ) can be inferred from the measured psf in real time , without the need for sequential measurements .
zimmerman _
et al . _
@xcite described how an extension of this principle allows one to capture polarization information from an image sampled by a pinhole array by placing the seo in the pupil ( fourier conjugate ) plane . in this letter
, we report on using seo polarimetric imaging to monitor in real time the polarization state at the output of a mcf .
we concentrate on the polarization state evolution when the fiber bundle is illuminated with rotating linear polarization states and retrieve , for each fiber core , the equivalent phase plate .
we also investigate the stability of the output polarization state when the mcf is slightly bent .
fig .
[ fig2](a ) shows a schematic of the optical setup .
a pulsed ytterbium laser ( t - pulse , amplitude systemes , 1030 nm , 50 mhz , 1 w ) is controlled in polarization by using a polarizer and a half wave plate . the laser beam is then weakly focused to illuminate the entrance facet of the tested mcf .
the output mcf facet is imaged onto a cmos camera after crossing a seo element ( placed at a fourier conjugate plane ) and a left - handed circular analyzer ( composed of a quarter wave plate and a polarizer ) .
two fused silica mcfs are considered in this work ( fig .
[ fig2](b - c ) ) , the first one consisting of 19 polarization - maintaining ( pm - mcf ) single mode fiber cores ( pitch 40 @xmath5 m , core diameter 18 @xmath5 m , outer diameter 348 @xmath5 m , length 50 cm ) , the second one ( mcf ) made up of 475 non - pm single mode fiber cores ( pitch 15.3 @xmath5 m , core diameter 3.1 @xmath5 m , outer diameter 421 @xmath5 m , length 30 cm ) . both mcfs exhibit very low cross talk , below -30 db and were held straight during the measurements .
@xmath6=25 mm , mcf multi core fiber , lenses @xmath7 and @xmath8 represent a set of lenses that image the fiber face onto the camera with a magnification of ~13x , seo stress - engineered optical element , qwp quarter wave plate , lhca left - handed circular analyzer ( b ) 19 cores pm single mode mcf ; ( c ) 475 cores non - pm single mode mcf ; scale bar : 100 @xmath5 m . ]
the system is calibrated by taking images with unpolarized illumination of the fiber input ( or with the sum of images with two orthogonal illumination polarizations ) and where polarizers are used after the fiber to create the six reference psfs for all cores , such as those shown in fig .
[ fig1](c - h ) for one core . by then performing a linear fit of the psf for each core with the corresponding reference psfs
, one can determine the stokes parameters .
figure [ fig3](a ) shows an image of the seo psfs ( grey scale ) obtained when illuminating 8 cores of the pm - mcf with linearly polarized light aligned along one of the pm - mcf principal axes .
note that the pm - mcf input facet was illuminated with a quasi - plane wave , without special core coupling optimization .
overlaid on fig .
[ fig3](a ) are the output polarization ellipses ( as directly deduced from the seo recorded psfs of @xmath9 pixels each ) .
note that these outputs are all approximately vertically polarized , a result consistent with the aligned linear polarized illumination and the pm characteristics of this mcf .
the movie in visualization 1 shows the same images for several linear input polarizations ( rotating the hwp in fig .
[ fig2](a ) ) .
clearly , two linear output polarizations are obtained when the pm - mcf is illuminated with linearly polarized input light along two orthogonal directions ( close to 0 and 90 ) .
visualization 1 also shows the trajectories over the poincar sphere of the output polarization states corresponding to the 8 cores .
these paths are all approximately great circles ( that is paths that draw the largest circle on the poincar sphere ) , and this indicates that each core behaves like a phase plate .
this phase - plate - like behavior is represented pictorially in fig .
[ fig3](b ) , where on each core , we overlay its two eigen - polarizations ( solid and dashed lines ) . the angle traced by the orange arc corresponds to the phase delay between these two eigen - polarizations states and its handedness ( right and left ) is represented by the colours green and red
. as expected , the two eigen - polarizations are mostly linear and oriented along two orthogonal axis on this pm - mcf .
note that since the jones matrices are not perfectly unitary , the eigen polarizations might not be exactly orthogonal . as for the phase delay , it arises from the differential phase between the two eigen - polarization that has been accumulated during the propagation along the fiber length and this is indeed expected to vary from core to core .
m. also see visualization 1 . ]
we now move to the non - pm 475 cores mcf , that we denote mcf for simplicity . here again the mcf facet is illuminated with a quasi - plane wave without special care to optimize core coupling .
figure [ fig4](a ) shows the recorded seo psfs ( grey scale ) , @xmath10 pixels each , together with the overlaid output polarization states for 180 cores , when the mcf is illuminated with vertically polarized light . as in the previous case ,
green and red ellipses represent right - handedness and left - handedness , respectively .
contrary to the pm - mcf , there are no apparent correlations in core to core output polarization states .
the movie in visualization 2 shows the same images when rotating the linear input polarizations , together with the trajectories over the poincar sphere of the output polarization states corresponding to each core . as previously ,
the paths follow approximately great circles over the poincar sphere surface indicating that each core acts as a phase plate . using this polarimetric information , we represent in fig .
[ fig4](b ) the two eigen - polarizations associated to each core , together with the phase delay ( orange arc ) between them . as previously , for each core , the underlying unpolarized transmitted intensity is also shown ( grey scale ) .
careful inspection of fig .
[ fig4](b ) reveals correlation between core to core eigen - polarization axis .
this residual anisotropy probably comes from the stress present in the mcf preform and is a direct result of its fabrication process .
as in the pm - mcf case the phase delay between the two eigen - polarization shows no core to core correlation and comes from the phase accumulated during propagation .
m. also see visualization 2 . ]
our seo - based polarimetric measurement can also address diattenuation in each fiber core , an effect related to the differential propagation loss arising between the two eigen - polarizations .
this can be seen in fig .
[ fig3](b ) and fig .
[ fig4](b ) where the measured diattenuation is indicated by the size of the second ( dashed ) eigen - polarization ellipse relative to the non - dashed one ( when diattenuation is absent the sizes of the dashed and non - dashed ellipse are the same ) . when diattenuation is present , the output polarization state trajectory is no longer a great circle over the poincar sphere as the input polarization is varied and describes a great circle trajectory over the poincar sphere .
the seo based polarimetry imaging allows then the full characterization of each core as a jones matrix , whose pictorial representation in terms of eigen - polarizations , diattenuation and relative phase is given in fig .
[ fig3](b ) and fig .
[ fig4](b ) .
we finally investigate the influence of a slight twist or bend on the fiber core output polarization states .
we take here full advantage of our single shot polarimetric measurement by recording video rate images of the seo psfs for 180 illuminated cores while bending the fiber .
visualization 3 shows 180 seo psfs ( grey scale ) and the associated output polarization states when the mcf is illuminated with vertically polarized light and manually wiggled and slightly bent ( radius of curvature approximately 10 cm over a fiber length of few cm ) .
visualization 4 shows the applied manual wiggling and bending .
fig .
[ fig5](a ) shows the seo psfs where we have selected four cores and display their trajectories over the poincar sphere fig .
[ fig5](b ) while applying manual bending .
visualization 5 shows a movie of the same .
quite surprisingly , the output polarization states are weakly affected by the bend suggesting that the phase delay accumulated along propagation between the two eigen - polarization states remains fairly constant in the case of wiggles and slight bends .
this clearly demonstrates that seo polarimetric imaging can monitor in real time the polarization state at the output of a mcf having a very large number of cores .
+ m. see visualizations 3 - 5 for movies of the same . ] in conclusion , we have described and tested a method of single shot polarimetry @xcite applied to the characterization of two different types of multicore fiber ( mcf ) . using a stress - engineered optical element followed by an analyzer , we can simultaneously retrieve the polarization states of the light emitted by a large number of fiber cores simply by looking at the individual images of the cores at the focus of a lens .
the technique is real - time and we have shown that it allows the full polarization characterization of each fiber core .
our approach is interesting for applications such as coherent beam combining in mcfs where the output polarization in each core has to be identical .
it also opens new directions in the field of lens - free endoscopy @xcite where performing polarimetric imaging together with full polarization and phase control at the mcf output gives access to the full vector psf control , in a way similar to what has been done in microscopy @xcite .
this would open the door for important new avenues in endoscopy including structural molecular and biological imaging @xcite . |
the decomposition of tensors @xmath5 as a sum of _ simple _ tensors ( i.e. tensors of rank @xmath6 ) is a central problem for many applications of multilinar algebra to algebraic statistics , signal theory , coding theory and others . for statistical inference ,
it is meaningful to know if a probability distribution , arising from a model , uniquely determines the parameters that produced it .
when this happens , the parameters are called _
identifiable_. the notion of _ generic _ identifiability for parametric models has been considered in @xcite and in @xcite 2.2 .
indeed , conditions which guarantee the uniqueness of this decomposition , for generic tensors in the model , are quite important in the applications .
when generic identifiability holds , the set of non - identifiable parameters has measure zero , thus parameter inference is still meaningful .
notice that many decomposition algorithms converge to _ one _ decomposition , hence a uniqueness result guarantees that the decomposition found is the chased one .
we refer to @xcite and its huge reference list , for more details .
even from a purely theoretical point of view , the study of the decomposition shows some beautiful link between multilinear algebra and projective geometry .
the present paper is devoted to study one intriguing special case , which shows an exceptional behavior . among tensors
@xmath5 whose rank has the generic value , only one example is known when we have identifiability , that is @xmath7 , @xmath8 , @xmath9 . on the contrary ,
general tensors whose rank is _ smaller _ than the generic value , often have a unique decomposition . excluding the cases of matrices ( tensors of dimension @xmath10 ) , identifiability is known to hold when the rank @xmath4 is _ small_. an evidence is given for @xmath7 by the celebrated kruskal s bound @xcite , which , for general tensors of given rank , is refined and extended in a series of papers ( see strassen s paper @xcite , the recent paper @xcite ) .
let s order the @xmath11 s so that @xmath12 . in @xcite corollary 8.4
it was proved that , with the assumption @xmath13 , the variety @xmath14 is @xmath4-identifiable if and only if @xmath15 the general tensor of rank @xmath16 has not a unique decomposition . after this result , we say that a tensor is _ unbalanced _ if @xmath13 .
this range is one unity larger than the corresponding unbalanced range considered in @xcite while studying the dimension of secant varieties of segre varieties ( see 8 of @xcite where these two notions were compared ) . in the case
@xmath17 the corresponding tensors are called _
balanced_. only few examples of balanced tensors , whose rank is smaller than the generic value , are known to be not generically identifiable .
we mention the case of tensors of rank @xmath18 in @xmath19 ( @xcite ) and , in dimension @xmath3 , tensors of rank @xmath3 in @xmath20 ( classical , see @xcite 4 ) and tensors of rank @xmath2 in @xmath21 ( @xcite , theorem 1.3 ) . a computer aided analysis ( see @xcite theor .
7.5 ) shows that when the numbers @xmath11 s grow , sporadic examples disappear , and we expect that a general balanced tensor , of rank smaller than the generic value , is identifiable . the present paper is devoted to illustrate one sporadic example , which we believe should be the last one , for balanced tensors of dimension @xmath7 .
namely , we use a geometric approach to show that general tensors of rank @xmath0 in @xmath22 are not uniquely decomposable .
notice that tensors of the mentioned type have generic rank equal to @xmath23 .
the proof of the non - uniqueness is based on the weak - defectivity principle , classically introduced by terracini ( @xcite ) .
we refer to @xcite and the introduction of @xcite for an account of the geometric reduction of the problem . in details , we prove that through @xmath0 general points of the segre variety @xmath24 , which corresponds to simple tensors in @xmath22 , one can find a special fourfold @xmath25 which is the segre - veronese image of @xmath26 , embedded by forms of type @xmath27 .
since through a general point of the span @xmath28 of @xmath25 one can find many linear @xmath29-spaces which are @xmath0-secant to @xmath25 , then by @xcite theorem 2.9 , it follows the weak defectivity and the non identifiability of our tensors .
the example is interesting also because the subvariety @xmath25 , which produces the non - identifiability of tensors of rank @xmath0 in @xmath22 , is quite complicate .
in particular , we are unable to estimate how many @xmath0-secant spaces to @xmath25 are there through a general point of the span @xmath28 . consequently , we are unable to determine how many different decomposition are there , for a general tensor @xmath30 as above .
we simply know that the number is finite , and at least @xmath2 .
let us mention that , from the geometrical point of view , the existence of the subvariety @xmath25 through @xmath0 general points of @xmath24 is proved by some
ad hoc argument .
a complete theory of special subvarieties that one can find through general points of segre varieties , seems actually far beyond our reach .
for basic facts about the geometric point of view on tensors we follow @xcite . given any irreducible projective variety @xmath31 , we denote by @xmath32 the @xmath4-th secant variety of @xmath31 , that is the zariski closure of the set @xmath33 .
@xmath32 is indeed the zariski closure of the set of elements having @xmath31-rank equal to @xmath4 . in the space @xmath34 , where @xmath35 , the ( projectification of the ) cone @xmath31 of simple tensors corresponds to the embedding of @xmath36 , via the segre map .
the ( projectification of the ) cone of tensors of rank @xmath4 is an open dense subset of the secant variety @xmath32 .
we recall from @xcite def .
2.1 the following @xmath31 is called @xmath4-identifiable if the general element of @xmath32 has a unique expression as the sum of @xmath4 elements of @xmath31 .
thus , in our notation , we say that @xmath36 is @xmath4-identifiable if the general tensors in @xmath37 of rank @xmath4 has a unique decomposition as a sum of simple tensors .
a complete list of known segre varieties @xmath38 , with @xmath39 , for which a computer based algorithm does not prove the @xmath4-identifiability , is provided in @xcite , 5 , see also @xcite 7 .
the list corresponds to the case of tensors of dimension @xmath3 , for which the algorithm can not prove the uniqueness of the decomposition . in all the examples , except for two of them ,
it is indeed well known that general tensors of rank @xmath4 have infinitely many decompositions .
the two remaining cases are listed below : @xmath40 in the first case , the effective proof that @xmath31 is not @xmath2-identifiable ( and the general tensor of rank @xmath2 has exactly @xmath41 decompositions ) is contained in @xcite , theorem 1.3 .
the latter case needs an ad hoc analysis which is the target of the present note .
our main tool is to prove the existence of particular , very degenerate subvarieties @xmath25 , through @xmath4 general points of the segre variety @xmath38 . indeed , we recall the following : [ inverseweak ] let @xmath31 be a projective , irreducible non degenerate variety of dimension @xmath42 in @xmath43 , @xmath44 .
suppose that for any general @xmath4-tuples of points @xmath45 one can find a subvariety @xmath25 of pure dimension @xmath46 containing the points @xmath47 , whose span has dimension @xmath48 assume that @xmath49 and moreover assume that through a general point of @xmath50 one finds @xmath51 @xmath4-secant @xmath52-linear spaces .
then @xmath31 is not @xmath4-identifiable . indeed through a general point of @xmath32 one finds at least @xmath53 @xmath4-secant @xmath52-linear spaces .
it is essentially theorem 2.9 of @xcite .
from this point on , we focus our attention to the vector space @xmath54 of tensors of type @xmath1 , which has dimension @xmath55 . from the projective point of view
, simple tensors in @xmath54 corresponds to points of the segre embedding of @xmath24 into @xmath56 .
we also fix the rank @xmath57 , i.e. we consider the eighth secant variety @xmath58 . we know that @xmath31 is not @xmath0-defective , so that @xmath58 has projective dimension @xmath59 ( see @xcite , 5 ) .
this means that the subvariety ( cone ) of tensors of rank @xmath0 in @xmath54 has the expected dimension @xmath60 .
[ p1p2 ] fix eight general points @xmath62 of @xmath63 and fix eight general points @xmath64 of @xmath65 .
then there exists a segre embedding @xmath66 for which the line @xmath67 contains @xmath68 for all @xmath69 .
in other words , each @xmath68 lies in @xmath70 and @xmath71 , where @xmath72 is the projection @xmath73 .
the embeddings @xmath74 are parametrized by the quotient group @xmath75 , which has dimension @xmath76 .
since the eight points @xmath68 are general , we have an @xmath0-dimensional family @xmath77 of embeddings @xmath78 for which @xmath79 . since the unique automorphism of @xmath65 which fixes four general points is the identity , as @xmath78 varies in @xmath77 , the family of @xmath61-tuples @xmath80 dominates @xmath81 .
since the group @xmath82 acts transitively on the points @xmath83 , it follows that the orbit of the set @xmath84 , under @xmath85 , dominates @xmath86 .
the claim follows .
we send @xmath26 to the three factors @xmath65 , @xmath63 and @xmath63 , by using the identity on @xmath65 and divisors @xmath88 and @xmath89 respectively .
thus , we need to prove that we can arrange this map @xmath90 so that the image passes through @xmath0 general points of @xmath31 .
the choice of eight general points in @xmath24 corresponds to the choice of @xmath0 general points in each factor . by the previous lemma , for a general choice of points @xmath91 , @xmath92 and @xmath93 , we can find divisors @xmath94 , which define segre embeddings @xmath95 of @xmath96 into @xmath63 , for which each @xmath68 ( resp .
each @xmath97 ) lies in the line @xmath67 ( resp .
@xmath98 ) .
the following lemma would be easy , provided one knows a table of @xmath61-dimensional varieties in @xmath28 , whose @xmath0-th secant order is different from @xmath6 .
since the table is missing , we need to compute directly what happens for the segre product @xmath26 .
[ nooass ] let @xmath25 be an embedding of @xmath26 into @xmath28 , through a divisor of type @xmath27 .
then through a general point @xmath102 one can draw at least @xmath2 spaces of dimension @xmath29 , which are @xmath0-secant to @xmath25 .
we consider the tangential projection from the tangent spaces at @xmath29 general points @xmath103 of @xmath25 , which is a rational map @xmath104 . by the theorem 4.2 ( vi ) of @xcite we have that the number of seven dimensional spaces which are @xmath0-secant to @xmath25 and contain a general point @xmath102 is @xmath105 , for a general choice of points @xmath103 .
so it is enough to show that @xmath106 .
a computer based algorithm , implemented in m2 @xcite , which is available in the ancillary files of the arxiv submission of this paper , shows that there exists a 7-uple @xmath107 and a point @xmath108 such that the fiber @xmath109 consists of @xmath2 reduced points .
consider the rational map @xmath110 induced by the tangential projection .
that is @xmath111 then , after resolving the indeterminacy of the map @xmath112 , we get , from the stein factorization , that the general fiber of @xmath112 consists of six points , so that @xmath106 .
computer experiments show that the base locus of @xmath113 consists of @xmath114 lines for @xmath115 ( each tangent space at a point @xmath116 meets @xmath25 in two lines ) and consists of @xmath117 lines plus @xmath61 extra points for @xmath118 .
we do not know how to prove theoretically the existence of these @xmath61 points in the base locus .
fix @xmath0 general points @xmath120 and a general point @xmath121 , so that @xmath122 is a general point of the @xmath0-th secant variety of @xmath31 . by lemma [ p1p1p2 ] ,
the eight points are contained in the image @xmath123 of a segre - veronese embedding of @xmath26 through a divisor of type @xmath27 .
@xmath25 spans a @xmath28 , which clearly contains @xmath122 , and @xmath122 is a general point of @xmath28 . by lemma [ nooass ] , one finds @xmath2 linear spaces of ( projective ) dimension @xmath29 , which are @xmath0-secant to @xmath25 and contain @xmath122 .
since these spaces are also @xmath0-secant to @xmath31 , the claim follows .
unfortunately , we can only conclude that there are _ at least _
@xmath2 decompositions .
one reason is that the lower bound with @xmath125 considered in the proof can be strict .
for example , if @xmath25 is the @xmath0-veronese embedding of @xmath65 , then ranestad and schreyer prove ( see theorem 1.7 ( iv ) of @xcite ) that a general polynomial of degree @xmath0 has exactly @xmath126 decompositions as the sum of @xmath127 powers of linear forms . on the other hand ,
the tangential projection from @xmath117 points has base locus given by the @xmath117 points themselves and so its degree is @xmath128 .
let @xmath129 be a linear system of @xmath130 matrices , with ( affine ) dimension @xmath3 .
assume that @xmath129 has rank @xmath0 , i.e. there are @xmath0 matrices of rank @xmath6 which generate all the elements of @xmath129 .
then there are at least @xmath2 sets of @xmath0 rank @xmath6 matrices , whose spans contain @xmath129 . | the identifiability of parameters in a probabilistic model is a crucial notion in statistical inference .
we prove that a general tensor of rank @xmath0 in @xmath1 has at least @xmath2 decompositions as sum of simple tensors , so it is not @xmath0-identifiable .
this is the highest known example of balanced tensors of dimension @xmath3 , which are not @xmath4-identifiable , when @xmath4 is smaller than the generic rank . |
paraparticles in quantum mechanics arise from the fact that the heisenberg - born - jordan relation for the quantum harmonic oscillator is not uniquely determined by the equations of motion @xcite and are specifically related to deformations induced by the reflection operator @xcite that appears in the calogero - vasiliev oscillator @xcite .
while paraparticles are allowed in the formalism of quantum mechanics , they have not been experimentally discovered as fundamental particles in nature @xcite .
recently , we have proposed a quantum simulation of driven even - order para - bose oscillators in a trapped - ion setup @xcite , for example , @xmath0 where the para - bose operators of order @xmath1 fulfill the following bilinear commutation relations @xcite : @xmath2 & = & 1 + ( p-1 ) \hat{\pi } , \nonumber \\
\left\ { \hat{a } , \hat{a}^{\dagger } \right\ } & = & 2 \hat{n } + p , \nonumber \\ \left[\hat{n } , \hat{a}^{\dagger } \right ] = \hat{a}^{\dagger } , & \quad & \left[\hat{n } , \hat{a } \right ] = - \hat{a}. \end{aligned}\ ] ] we have used the parity operator given by @xmath3 ; the operators @xmath4 , @xmath5 , and @xmath6 are the number , creation , and annihilation operators of the oscillator and the order parameter @xmath1 is a positive integer , @xmath7 .
the standard boson algebra showing bose - einstein statistics is recovered with order @xmath8 .
the work on para - bose particles is extensive and , for example , their coherent states are well understood @xcite . here , we are interested in the properties of a peculiar set of states provided by simple time evolution in a particular realization of our para - bose oscillators simulation , @xmath9 these states look like gilmore - perelomov coherent states but , formally , only the standard boson case , @xmath8 , yields gilmore - perelomov coherent states @xcite due to the characteristics of the para - bose algebra .
nevertheless , we will provide in the following an analytic expression for these states and show that they span an overcomplete , nonorthogonal basis for the corresponding para - bose hilbert space .
then , we will address their para - bose fock occupation distribution , and show that it follows sub- , super- , and poissonian statistics , calculated through the mandel q parameter , depending on the coherent parameter look - alike and the para - bose order .
thus , our previously proposed quantum simulation allows the engineering of highly nonclassical and semiclassical para - bose states under the restrictions imposed by experimental constraints such as dissipation .
furthermore , we will show that these states do not minimize the robertson uncertainty relation for the para - bose field quadratures , nor are their quadrature variances equal for para - bose particles beyond standard bosons .
finally , we will close with a summary and conclusion .
we can write a generalization of the states provided by time evolution of the trapped - ion simulation in a gilmore - perelomov coherent state look - alike form , @xmath10 where the parameter @xmath11 is a complex number that parametrizes the states , and the state @xmath12 is the para - bose vacuum state of order @xmath1 .
formally , only standard bosons , @xmath8 , are gilmore - perelomov coherent states @xcite , @xmath13 nevertheless , we will show that these para - bose states can be interesting on their own .
it is cumbersome but straightforward to calculate their para - bose fock state decomposition , @xmath14 where the notations @xmath15 and @xmath16 stand for the pochhammer symbol and the confluent hypergeometric function of the first kind , respectively .
as the second argument of the confluent hypergeometric function fulfills @xmath17 , we can be sure that all of the amplitudes are entire functions of @xmath1 and @xmath11 that converge for all finite @xmath11 @xcite . here
, we can also use the fact that @xmath18 for any given @xmath19 and @xmath20 to recover the standard coherent states for bosons proposed by sudarshan @xcite and glauber @xcite , @xmath21 it is possible to demonstrate that our para - bose states resolve the identity , @xmath22 this can be accomplished by substitution of the para - bose fock state decomposition presented above , @xmath23^{2 } } i_{even } \vert 2j \rangle \langle 2 j \vert \right .
\nonumber \\ & & \left .
+ \sum_{j } \frac{2^{2j+2 } j ! \left(\frac{p}{2}\right)_{j+1}}{\left[(2j+1)!\right]^{2 } } i_{odd } \vert 2j + 1 \rangle \langle 2 j + 1 \vert \right],\end{aligned}\ ] ] and reduced to prove the following integrals ; @xmath24 ^ 2}{2^{2j+1}j!\left(\frac{p}{2}\right)_{j } } , \\
i_{odd } & = & \int_{0}^{\infty } dr ~r^{4j+3 } e^{-r^2 } ~_{1}f^{2}_{1}\left(\frac{1-p}{2 } , j+\frac{3}{2},\frac{r^{2}}{2}\right ) = \frac{\left[(2j+1)!\right]^2}{2^{2j+2}j!\left(\frac{p}{2}\right)_{j+1}}\end{aligned}\ ] ] which can also be computed numerically . here
, orthogonality arises from the phase of the complex parameter @xmath11 , and the radial - like variable is its module , @xmath25 .
thus , the set of states @xmath26 with @xmath27 and @xmath28 form an overcomplete basis for the corresponding para - bose hilbert space of order @xmath1 .
the basis is also non - orthogonal with overlaps that decay with the addition of the complex parameters moduli , in a similar fashion to standard coherent states , @xmath29\end{aligned}\ ] ] where the auxiliary confluent hypergeometric product converges for finite parameters , @xmath30 in short , these experimentally feasible states are the standard coherent states for just @xmath8 and provide us with an overcomplete , nonorthogonal basis for the para - bose hilbert space of the order of @xmath31 .
our closed form for the overcomplete , nonorthogonal para - bose bases allows the study of their properties ; for example , their para - bose fock number state occupation distribution , @xmath32 as expected , we recover the standard coherent states photon distribution for @xmath8 [ fig . [
fig : fig1](a ) ] and more complex distributions for higher orders ; for example , for order @xmath33 [ fig .
[ fig : fig1](b ) ] , @xmath34 [ fig .
[ fig : fig1](c ) ] , and @xmath35 [ fig .
[ fig : fig1](d ) ] .
the effect of having para - bose particles is such that distributions for contiguous orders have significant variations with respect to each other : figs .
[ fig : fig1](a ) and [ fig : fig1](b ) and figs . [ fig : fig1](c ) and [ fig : fig1](d ) . of our overcomplete ,
nonorthogonal para - bose basis with order ( a ) @xmath8 , ( b ) @xmath33 , ( c ) @xmath34 , and ( d ) @xmath35 , and complex parameter @xmath36 . ] the mean para - bose excitation number , @xmath37 can also be given in terms of integer functions , @xmath38 involving the confluent hypergeometric function @xmath39 and the generalized hypergeometric function @xmath40 @xcite .
note that the asymptotic values for the hypergeometric functions , @xmath41 allow us to write the mean para - bose excitation number for a large complex parameter module , @xmath42 , & p = 2 ,
\\ \vert \alpha \vert^{2 } + \frac{(p-1)}{2(p-2 ) } , & p > 2 , \end{array } \right . \qquad
\vert \alpha \vert \gg 1 , \nonumber \\ \end{aligned}\ ] ] where the function @xmath43 is the digamma function @xcite .
note that in the limit of large para - bose order and large complex parameter module , we can approximate the mean para - bose excitation number by @xmath44 for @xmath45 and @xmath46 .
these asymptotic behaviors were confirmed numerically .
it is also possible to provide the variance for the mean para - bose excitation number , @xmath47 but we leave it to the appendix due to its size .
it is shorter to write its asymptotic behavior , @xmath48 - \frac{1}{16}\psi^2 \left(\frac{1}{2}\right ) \\ - \frac{1}{8 } \left[3 + \psi \left(\frac{1}{2}\right)+ \frac{1}{8 } \ln\left(2 \vert\alpha\vert^{2}\right)\right ] \ln \left(2 \vert \alpha \vert^{2}\right ) + \frac{1}{4 } + \frac{3}{8 } \psi \left(\frac{1}{2}\right ) , & p = 2 , \\
\frac{5}{2 } \vert\alpha\vert^{2 } + \frac{9}{8 } \left[\ln \left(2 \vert\alpha\vert^{2}\right ) - \psi \left(\frac{1}{2}\right)\right ] - \frac{33}{16 } , & p = 4 , \\
\vert \alpha \vert^{2}\frac{2p-3}{p-2 } + \frac{p(p-1)}{4(p-4)(p-2 ) } , & p \neq 2,4 , \end{array } \right .
\vert \alpha \vert \gg 1,\nonumber \\\end{aligned}\ ] ] and we can realize that our para - bose states with large complex parameter modulus always have a variance of the excitation number larger than its mean : @xmath49 for @xmath31 and @xmath46 .
a more complex behavior emerges outside the asymptotic limit , as shown in fig .
[ fig : fig2 ] for different coherent parameter values , @xmath50 [ fig .
[ fig : fig2](a ) ] , @xmath51 [ fig .
[ fig : fig2](b ) ] , @xmath36 [ fig .
[ fig : fig2](c ) ] , and @xmath52 [ fig .
[ fig : fig2](d ) ] , where we show the standard deviation as error bars , @xmath53 . from the numerics outside the asymptotic regime
, we can conjecture the following : first , for a given para - bose order larger than two , there exists a family of isomodular complex parameters that makes the mean value of the para - bose excitation number and its variance equal . for these complex parameters ,
lower- ( larger-)order para - bose states will have a mean excitation number that is larger ( smaller ) than its variance .
second , there exists an isomodular family of states , @xmath54 , that makes the mean value of the para - bose excitation number and its variance equal for para - bose order two , @xmath33 .
this value serves as the critical value after which the mean value of the excitation number will be smaller than its variance for any given para - bose order . obviously , for any given complex parameter and standard bosons , the mean value of the excitation number and its variance are always equal , @xmath55 for @xmath8 . in bars , and its standard deviation , @xmath53 in error bars , for some members @xmath56 of our overcomplete , nonorthogonal para - bose basis with complex parameter ( a ) @xmath50 , ( b ) @xmath51 , ( c ) @xmath36 , and ( d ) @xmath52 . ] this peculiar behavior for the mean value of the para - bose excitation number and its variance means our states will show sub - poissonian , poissonian , and super - poissonian statistics depending on the para - bose order and the complex parameter modulus .
we can see this with the mandel q parameter @xcite , @xmath57 which is no longer a constant for our para - bose states , again we provide the asymptotic for a large complex parameter modulus for the sake of space , @xmath58 + 2}{4 \vert\alpha\vert^{2 } + \ln \left(2 \vert\alpha\vert^{2}\right ) - \psi \left(\frac{1}{2}\right ) } \\ -\vert\alpha\vert^2 - \frac{1}{4 } \left[\ln \left(2 \vert\alpha\vert^2\right ) - \psi \left(\frac{1}{2}\right)\right ] - \frac{5}{2 } , & p = 2 , \\
\frac{6}{4 \vert\alpha\vert^{2}+3 } \left\{\vert\alpha\vert^{2 } + \frac{3}{4 } \left[\ln \left(2 \vert\alpha\vert^{2}\right ) - \frac{5}{2 } - \psi \left(\frac{1}{2}\right)\right]\right\ } , & p = 4 , \\
\frac{(p-1 ) \big[4\vert\alpha\vert^{2 } ( p-4 ) ( p-2 ) - p^{2 } + 11 p - 16 \big]}{2 ( p-4 ) ( p-2 ) \big[2 \vert\alpha\vert^{2 } ( p-2 ) + p-1 \big ] } , & p \neq 2,4 .
\end{array } \right .
\vert \alpha \vert \gg 1 , \nonumber \\\end{aligned}\ ] ] upon close inspection , we can realize that in the limit of the large complex parameter modulus , the para - bose fock state distribution of our states is always super - poissonian as expected , @xmath59 for @xmath46 .
the full expression for the mandel q parameter , which can be found in the appendix , allows us to rewrite the conjectures mentioned above : for a given para - bose order larger than two , there exists a family of isomodular complex parameters that yields poissonian statistics for the excitation number distribution ; lower- ( larger-)order para - bose states that belong to this isomodular complex parameter family will have sub- ( super)-poissonian statistics , shown in figs .
[ fig : fig3](a ) and [ fig : fig3](b ) .
there exists a critical family of isomodular complex parameters , @xmath54 , after which the excitation number distributions for para - bose order @xmath60 follow super - poissonian statistics and may be described with a semi - classical theory ; see figs .
[ fig : fig3](c ) and [ fig : fig3](d ) . of our overcomplete ,
nonorthogonal para - bose basis with complex parameter ( a ) @xmath50 , ( b ) @xmath51 , ( c ) @xmath36 , and ( d ) @xmath52 . ]
the mean values for the field quadratures , @xmath61 of our overcomplete , nonorthogonal para - bose basis are also integer functions , @xmath62 given in terms of confluent hypergeometric functions of the first kind , @xmath63 this auxiliary function takes a unit value for both standard bosons , @xmath64 , and large complex parameters , @xmath65 for @xmath46 , and we recover the standard boson value for the para - bose field quadratures .
the second momenta of the para - bose field quadratures can be found in the appendix and used to construct the quadratures variances , @xmath66 , & p = 2 , \\
\frac{p}{2 } + \frac{1}{2}\left(\alpha + \alpha^{\ast}\right)^{2 } \left[1 + \frac{(p-1)}{2 \vert \alpha \vert^{2 } ( p-2 ) } g(p,\vert\alpha\vert ) - h^{2}(p,\vert\alpha\vert)\right ] , & p \neq 2 , \end{array } \right . \nonumber \\ \sigma^{2}_{y } & = & \left\ { \begin{array}{ll } 1 - \frac{1}{2 } \left(\alpha^{\ast } - \alpha\right)^{2 } \big[f(\vert\alpha\vert ) - h^{2}(2,\vert\alpha\vert)\big ] , & p = 2 , \\
\frac{p}{2 } - \frac{1}{2}\left(\alpha^{\ast } - \alpha\right)^{2 } \left[1 + \frac{(p-1)}{2 \vert \alpha \vert^{2 } ( p-2 ) } g(p,\vert\alpha\vert ) - h^{2}(p,\vert\alpha\vert)\right ] , & p \neq 2 , \end{array } \right . \nonumber \\ \end{aligned}\ ] ] figure [ fig : fig4 ] show the mean value of the @xmath67 para - bose field quadrature as bars and its standard deviation @xmath68 as error bars for para - bose orders @xmath69 $ ] and complex parameter values identical to previous figures @xmath50 [ fig . [
fig : fig4](a ) ] , @xmath51 [ fig .
[ fig : fig4](b ) ] , @xmath36 [ fig .
[ fig : fig4](c ) ] , and @xmath52 [ fig .
[ fig : fig4](d ) ] . while it is possible to provide asymptotic approximations for the variances of the field quadratures , which can be found in the appendix , they portray no intelligible information beyond the behavior shown in the plots .
quadrature , @xmath70 in bars , and its standard deviation , @xmath68 in error bars , for some members @xmath56 of our overcomplete , nonorthogonal para - bose basis with complex parameter ( a ) @xmath50 , ( b ) @xmath51 , ( c ) @xmath36 , and ( d ) @xmath52 . ] , for some members @xmath56 of our overcomplete , nonorthogonal para - bose basis with complex parameter ( a ) @xmath50 , ( b ) @xmath51 , ( c ) @xmath36 , and ( d ) @xmath52 in blue ( dark gray ) bars and the lower bound for the corresponding robertson uncertainty relation in gold ( light gray ) bars . ] we can also calculate the lower bound of the robertson uncertainty relation for the field quadratures @xcite , @xmath71 which yields a simple closed form expression , @xmath72 which , again , converges to the standard boson expression for large para - bose complex parameter values , @xmath73 due to the asymptotic properties of the confluent hypergeometric function , @xmath74 the actual value for the product of the quadratures standard deviation of our overcomplete , nonorthogonal basis states is given by the following expression , @xmath75 - \frac{1}{4}\left ( { \alpha^{\ast}}^{2 } - \alpha^{2}\right)^{2 } \big[f(\vert \alpha \vert ) - h^{2}(2 , \vert \alpha \vert)\big]^{2 } , & p = 2 , \\ \frac{p^{2}}{4 } + p \vert\alpha \vert^{2 } \xi(p,\vert\alpha\vert ) - \frac{1}{4}\left({\alpha^{\ast}}^{2 } - \alpha^{2}\right)^{2 } \xi^{2}(p,\vert\alpha\vert ) , & p \neq 2 .
\end{array } \right .
\nonumber \\\end{aligned}\ ] ] where the auxiliary function is related to those defined before , @xmath76 we can compare the lower bound of the robertson uncertainty relation and the actual value for our para - bose states and realize that they never minimize the uncertainty relation for the para - bose field quadratures ; see figs .
[ fig : fig5](a ) and [ fig : fig5](d ) .
we have introduced an overcomplete , non - orthogonal basis for para - bose hilbert spaces of order @xmath1 using a structure that looks like standard boson gilmore - perelomov coherent states . these states , for even para - bose order , arise naturally in trapped ions where two orthogonal vibrational modes of the ion center of mass are involved .
our overcomplete , nonorthogonal para - bose basis shows the three types of poissonian statistics in their mean excitation number distribution ; this feature can be controlled by the para - bose order and the complex parameter modulus . in an actual ion - trap experiment ,
these could be fine tuned as they are related to the initial state and the evolution time , in that order .
furthermore , our states present relative quadrature squeezing and never minimize the uncertainty relation for the para - bose field quadratures standard deviations .
these properties could be used in a trapped - ion quantum simulation to generate tripartite states that inherit these nonclassical or semiclassical characteristics .
the authors acknowledge financial support from conacyt through doctoral grants no .
@xmath77 , no .
@xmath78 , and project grant no .
cb-2015 - 01/@xmath79 .
the second momenta of the para - bose excitation number , @xmath80 f\left(\sqrt{2}\vert \alpha\vert \right ) + \frac{9}{2 } \vert \alpha \vert^{2 } f(\vert \alpha\vert ) , & p = 4 , \\
p \vert\alpha\vert^{2 } + \vert\alpha \vert^{4 } + \frac{p-1}{6(p-4)(p-2)}\big\{3\big [ p - 1 - 2 \vert \alpha \vert^{2 } ( p-4)^{2 } \big]-\chi_{1}(p,\vert \alpha \vert ) -\chi_{2}(p,\vert \alpha \vert)\big\ } , & p \neq 2,4 , \end{array } \right .
\\ \end{aligned}\ ] ] can be written in terms of dawson integral @xmath81 @xcite and the auxiliary integer functions , @xmath82 ~_{1}f_{1}\left(\frac{3-p}{2 } , \frac{3}{2 } , 2\vert \alpha \vert^{2 } \right ) \\
\chi_{2}(p,\vert\alpha\vert ) & = & 4 \vert \alpha \vert^{2 } e^{- 2 \vert\alpha\vert^2 } ( p-3 ) \big[1 - p + 2\vert \alpha \vert^{2 } ( p + 2 ) \big ] ~_{1}f_{1 } \left ( \frac{5-p}{2 } , \frac{5}{2 } , 2\vert \alpha \vert^{2 } \right).\end{aligned}\ ] ] its asymptotic expression for a large complex parameter modulus can be given in the following form , @xmath83 + \frac{1}{8 } \ln \left(2 \vert \alpha \vert^{2}\right ) - \frac{1}{4 } + \frac{3}{8 } \psi \left(\frac{1}{2}\right ) - \frac{1}{2}\psi \left(\frac{3}{2}\right ) , & p = 2 , \\
\frac{5 \vert \alpha \vert^{2}}{2 } + \frac{9}{8 } \left [ \ln \left(2 \vert \alpha \vert^{2}\right ) -
\psi \left(\frac{1}{2}\right)\right ] - \frac{33}{16 } , & p = 4 , \\
p \vert\alpha\vert^{2 } + \vert\alpha \vert^{4 } + \frac{3(p-1)\big [ p - 1 - 2 \vert \alpha \vert^{2 } ( p-4)^{2 } \big]}{6(p-4)(p-2 ) } , & p \neq 2,4 , \end{array } \right .
\vert \alpha \vert \gg 1 .
\\ \end{aligned}\ ] ] with the full forms , we can calculate the standard deviation of the mean para - bose excitation number , @xmath84 f\left(\sqrt{2}\vert \alpha\vert \right ) + \frac{9}{2 } \vert \alpha \vert^{2 } f(\vert \alpha\vert ) - \frac{9}{2 } \vert \alpha \vert^{2 } f^{2}(\vert \alpha\vert ) , \hspace{4 cm } p = 4 , \\ \vert \alpha \vert^{2}\frac{3p-4}{p-2}-\frac{p-1}{12 ( p-4 ) ( p-2)^2}\big(\big\{3(p-4 ) ( p-1)g^{2}(p,\alpha ) + 2 ( p-2 ) [ \chi_{1}(p,\vert\alpha\vert ) + \chi_{2 } ( p,\vert\alpha\vert ) - 3 p+3]\big\ } & \\ + 12\vert \alpha \vert^{2 } ( p-4 ) ( p-2)g(p,\vert\alpha\vert)\big ) , \hspace{9 cm } p \neq 2,4 \end{array } \right . \\
\end{aligned}\ ] ] these full forms yield the following mandel q parameter , @xmath85 -\frac{5}{2 } - \vert \alpha \vert^2 f(\vert\alpha\vert ) , \ ; p = 2 , \\ \frac{3}{2 \vert \alpha \vert + 3 \sqrt{2 } f\left(\sqrt{2 } \vert \alpha \vert \right ) } \big\{3 \vert \alpha \vert \big [ f(\vert\alpha\vert ) - f^{2}\left(\sqrt{2 } \vert \alpha \vert \right ) \big ] - \sqrt{2 } \left(4 \vert \alpha \vert^2 + 3\right ) f\left(\sqrt{2 } \vert \alpha \vert \right)\big\ } , \hspace{2.5 cm } \ ; p = 4 , \\ \frac { ( p-1)}{6 ( p-4 ) ( p-2 ) \big[2\vert \alpha \vert^2 ( p-2 ) + ( p-1 ) g(p,\vert\alpha\vert ) \big ] } \big\ { - 6\left(2 \vert \alpha \vert^2 + 1\right ) ( p-4 ) ( p-2 ) g(p,\vert\alpha\vert ) \bigg . &
\\ \bigg .
-3(p-4 ) ( p-1 ) g^{2}(p,\vert\alpha\vert ) - 2 ( p-2 ) [ \chi_{1}(p,\vert\alpha\vert ) + \chi_{2}(p,\vert\alpha\vert ) - 12\vert\alpha\vert^2 ( p-4 ) -3p + 3 ] \big\ } , p \neq 2,4 .
\end{array } \right . \\
\end{aligned}\ ] ] the asymptotic values of the para - bose field quadratures for large complex parameter are : @xmath86 the exact full form of the second momenta for the field quadratures is given in the following , @xmath87 , & p \neq 2 , \end{array } \right .
\langle \hat{y}^2 \rangle & = & \left\ { \begin{array}{ll } 1 - \frac{1}{2}\left ( \alpha^{\ast } -\alpha \right)^{2 } f(\vert\alpha\vert ) , & p = 2 , \\ \frac{p}{2 } - \left(\alpha^{\ast } - \alpha\right)^{2 } \left[\frac{1}{2 } + \frac{(p-1)}{4(p-2)\vert \alpha \vert^{2 } } g(p,\vert\alpha\vert ) \right ] , & p \neq 2 , \end{array } \right .
\nonumber \\ \end{aligned}\ ] ] and has an asymptotic approximation , @xmath88\right\ } , & p = 2 , \\
\frac{p}{2 } + \left(\alpha + \alpha^{\ast}\right)^{2 } \left[\frac{1}{2 } + \frac{(p-1)}{4(p-2)\vert \alpha \vert^{2 } } \right ] , & p \neq 2 , \end{array } \right
. \nonumber \\
\langle \hat{y}^2 \rangle & \approx & \left\ { \begin{array}{ll } 1 - \frac{1}{2}\left ( \alpha^{\ast } -\alpha \right)^{2}\left\{1 + \frac{1}{4 \vert \alpha \vert^{2}}\left[\ln \left(2 \vert \alpha \vert^{2}\right)-\psi \left(\frac{1}{2}\right)\right]\right\ } , & p = 2 , \\ \frac{p}{2 } - \left(\alpha^{\ast } - \alpha\right)^{2 } \left[\frac{1}{2 } + \frac{(p-1)}{4(p-2)\vert \alpha \vert^{2 } } \right ] , & p \neq 2 , \end{array } \right .
\vert\alpha\vert \gg 1 .
\nonumber \\ \end{aligned}\ ] ] the full form for their variances can also be given in the asymptotic limit , @xmath89 - \frac{1}{2 \sqrt{\pi } \vert \alpha \vert^2 } \left(2 + \frac{1}{2 \sqrt{\pi } \vert\alpha \vert^2 } \right)\big ] , & p = 2 , \\ \frac{p}{2 } + \frac{1}{2}\left(\alpha + \alpha^{\ast}\right)^{2 } \left[\frac{(p-1)}{2 \vert \alpha \vert^{2 } ( p-2)}\right ] , & p \neq 2 , \end{array } \right . \nonumber \\
\sigma^{2}_{y } & \approx & \left\ { \begin{array}{ll } 1 - \frac{1}{2 } \left(\alpha^{\ast } - \alpha\right)^{2 } \big[\frac{1}{4 \vert \alpha \vert^2 } \left[\ln \left(2 \vert \alpha \vert^2\right)-\psi \left(\frac{1}{2}\right)\right ] - \frac{1}{2 \sqrt{\pi } \vert \alpha \vert^2 } \left(2 + \frac{1}{2 \sqrt{\pi } \vert\alpha \vert^2 } \right)\big ] , & p = 2 , \\
\frac{p}{2 } - \frac{1}{2}\left(\alpha^{\ast } - \alpha\right)^{2 } \left[\frac{(p-1)}{2 \vert \alpha \vert^{2 } ( p-2)}\right ] , & p \neq 2 , \end{array } \right .
\vert \alpha \vert \gg 1.\nonumber \\\end{aligned}\ ] ] finally , this is the full form for the product of the quadratures variances in the asymptotic limit , @xmath90 - \frac{1 } { \sqrt{\pi } } \left(2 + \frac{1}{2 \sqrt{\pi } \vert\alpha \vert^2 } \right ) & \\ \quad + \frac{1}{4}\left(\alpha^{2 } + { \alpha^{\ast}}^{2}\right)^{2 } \big[\frac{1}{4 \vert\alpha \vert^2 } \left[\ln \left(2 \vert \alpha \vert^2\right)-\psi \left(\frac{1}{2}\right)\right ] - \frac{1}{2 \sqrt{\pi } \vert \alpha \vert^2 } \left(2 + \frac{1}{2 \sqrt{\pi } \vert\alpha \vert^2 } \right)\big]^2 , & p = 2 , \\
\frac{p^{2}}{4 } + \frac{p(p-1)}{2(p-2 ) } + \frac{1}{4 } \left(\alpha^{2 } + { \alpha^{\ast}}^{2}\right)^{2 } \left[\frac{(p-1)}{2(p-2)\vert \alpha \vert^{2}}\right]^{2 } , & p \neq 2 .
\end{array } \right .
\nonumber \\\end{aligned}\ ] ] 18 ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1 '' '' @noop [ 0]secondoftwo sanitize@url [ 0 ] + 12$12 & 12#1212_12%12 @startlink[1 ] @endlink[0 ] @bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) , ed .
, @noop _ _ ( , , ) @noop * * , ( ) @noop * * , ( ) | motivated by the proposal to simulate para - bose oscillators in a trapped - ion setup [ c. huerta alderete and b. m. rodrguez - lara , phys .
rev .
a * 95 * , 013820 ( 2017 ) ] , we introduce an overcomplete , nonorthogonal basis for para - bose hilbert spaces . the states spanning these bases can be experimentally realized in the trapped - ion simulation via time evolution .
the para - bose states show both nonclassical and semiclassical statistics on their fock state distribution , asymmetric field quadrature variances , and do not minimize the uncertainty relation for the field quadratures .
these properties are analytically controlled by the para - bose order and the evolution time ; both parameters might be feasible for fine tuning in the trapped - ion quantum simulation . |
in what is usually referred to as _ ghost imaging _ the coherent imaging of an object is achieved with incoherent light upon exploiting the spatial correlations between two light beams @xcite . the object interacts with one of the beams and an image of the object is built up by scanning the other beam .
ghost imaging may be obtained either with classically correlated beams @xcite , as those obtained by splitting the light from a ( pseudo ) thermal source , or with entangled beams @xcite , as those obtained by parametric downconversion . in the latter case one
may achieve in principle higher visibility . for objects which modify only the amplitude of light an image may be obtained with a single spatially incoherent beam upon measuring the autocorrelation function in the far field , without the need of ghost imaging
this is no longer possible when the object is also modifying the phase of the beam . in particular , it is of interest to investigate ghost imaging in the extreme case of pure phase objects , i.e. objects altering only the phase information carried by the beam .
phase objects are also of intrinsic interest in quantum information processing , since they introduce reversible unitary operations .
ghost imaging of pure phase objects has been extensively analyzed theoretically and experimentally using both classically or quantum correlated beams @xcite .
related effects connected with ( nonlocal ) dispersion cancellation have been investigated as well , both in the temporal and the spatial domains @xcite . in this paper
we suggest and demonstrate experimentally a scheme to achieve coherent nonlocal compensation / superposition of pure phase objects , also paving the way for the reconstruction of the overall phase function imposed to the two beams .
our scheme is based on two - photon polarization and momentum entangled states , which are generated by spontaneous parametric downconversion and purified using a spatial light modulator .
the same device is also used to introduce arbitrary phase functions on the two beams , which represent arbitrary phase objects . in our setup the insertion of a single phase object on one of the beams reduces the purity of the state and the amount of shared entanglement , whereas the original entanglement can be retrieved by adding a suitable phase object on the other beam .
the image of both single or double phase objects can be thus obtained by scanning the coincidence counts on one of the two beams . as a possible application , we also suggest and demonstrate a protocol for quantum key distribution based on nonlocal phase compensation . the paper is structured as follows . in the next section
we describe in details our experimental setup and the properties of the two - photon entangled states that are generated . in section [
s : dpi ] we analyze in some details phase imaging and nonlocal phase compensation of phase objects .
finally , in section [ s : qkd ] , we suggest a quantum key distribution protocol based on phase compensation and report about its experimental implementation .
section [ s : out ] closes the paper with some concluding remarks .
in our setup ( see fig . [ exp_setup ] ) a two - qubit polarization entangled state is produced by type - i downconversion from a couple of crystals ( beta - barium borate , bbo ) in a non - collinear configuration @xcite .
pairs of correlated photons are generated , distributed on a broad angular and spectral ranges , which are determined by the crystal length @xcite . upon expanding the transverse momentum conservation condition to the first order
, it can be shown that the angular and the spectral degrees of freedom are connected by the relation @xmath0 where @xmath1 are the signal and idler shifts from the central emission angles ( @xmath2 in our case ) , @xmath3 is the signal shift from the central frequency of the downconverted beams , and @xmath4 is a constant depending only on the signal central frequency and angle @xcite . the state at the output of the crystals may be written as @xmath5 \label{eq : state}\end{aligned}\ ] ] where the overall angular distribution @xmath6\,,\ ] ] contains the angular distribution @xmath7 due to the phase - matching conditions and the transmissivity @xmath8\equiv t[\omega(\theta,\theta')]$ ] of an interference filter set on the signal arm .
the ket @xmath9 denotes a single photon state emitted with polarization @xmath10 at angle @xmath11 ( @xmath12 ) along the signal ( idler ) arm , and the integrations ranges from @xmath13 to @xmath14 , @xmath15 being the angular aperture of two slits placed along the paths of the downconverted beams ( see fig .
[ exp_setup ] ) .
experimentally , some care must be taken in order to spatially superimpose the @xmath16 and @xmath17 downconversion beams , so that the angular distribution @xmath7 is actually the same for the two components of the entangled state . from eq .
( [ eq : theta - omega ] ) one sees that for a narrow spectral width ( @xmath18 ) , the angular distribution @xmath19 approaches @xmath20 and maximal entanglement in momentum is thus achieved .
the relative phase in eq .
( [ eq : state ] ) can be written as @xmath21 the first phase term , which can be expanded to the first order as @xmath22 comes from the angle - dependent optical path of vertically polarized photon pairs , generated in the first crystal , which must travel along the second one .
these angular dependent terms are responsible for decoherence of polarization qubit and should be removed in order to obtain an effective source of entangled pairs @xcite . in our apparatus ,
a one dimensional programmable spatial light modulator ( slm ) is placed on the signal and the idler paths ( see fig . [ exp_setup ] ) in order to insert the phase functions @xmath23 , @xmath24 , and @xmath25 .
the first term is used for purification , i.e to remove the angle - dependent phase - shift , @xmath26 , thus realizing a reliable polarization - entanglement source @xcite , whereas the functions @xmath27 and @xmath28 represent the phase - objects we insert on the signal and the idler arm , respectively . notice that spatial light modulators have been recently used in imaging also for a different purpose , i.e that of imposing a set of known random phase distributions @xcite and
implement the so - called computational ghost imaging @xcite , where the intensity detected in the one of the beam is computed offline . given the state in eq .
( [ eq : state ] ) , we can write the probability for the detection of a photon pair within the emission angle range @xmath29 , @xmath30 and with polarization angles @xmath31 ( signal ) and @xmath32 ( idler ) as follows @xmath33 in order to quantify the entanglement of our state we measure the state visibility , which is defined from eq .
( [ eq : c ] ) by placing the slits on the spdc central emission angles , @xmath34 , and setting the aperture to @xmath35mrad , and using the two pairs of angles @xmath36 and @xmath37 , @xmath38 , i.e @xmath39 in fact , once the state has been purified and the phase objects @xmath40 have been inserted , the polarization density matrix reads @xmath41\,{\vert { { \hbox{\small vv}}}\rangle}{\langle { { \hbox{\small hh}}}\vert } \notag \\ & + \varepsilon^*[\phi,\phi^\prime]\,{\vert { { \hbox{\small hh}}}\rangle } { \langle { { \hbox{\small vv}}}\vert}+{\vert { { \hbox{\small vv}}}\rangle}{\langle { { \hbox{\small vv}}}\vert}\big ) \,,\end{aligned}\ ] ] where @xmath42 $ ] is given by @xmath43\}\,.\ ] ] since the angular distribution @xmath19 is symmetric ( see below ) by choosing , without loss of generality , odd phase functions @xmath27 and @xmath28 , we obtain that @xmath44 is real . as a consequence we may write @xmath45 where @xmath46 denotes a bell state and @xmath47 the corresponding mixture . for the state @xmath48
, visibility provides a proper measure of entanglement since the expression in eq .
( [ eq : vis ] ) reduces to @xmath49\equiv\varepsilon$ ] which , in turn , equals the concurrence of the state .
the experimental setup is shown in fig .
[ exp_setup ] : a linearly polarized cw , @xmath50 nm laser diode ( thorlabs lqc@xmath51 - 40p ) pumps a couple of @xmath52 mm thick bbo crystals cut for type - i downconversion . the beam waist is set to @xmath53 m by a telescopic system .
the effective pump power on the generating crystals is of about @xmath54mw .
the @xmath16 and@xmath17 pairs are balanced by a half wave - plate set along the pump path .
a bbo crystal with the proper length and optical axis angle is set on the pump path , and is used to counteract the decoherence effect due to the temporal delay between the two components @xcite .
such crystal introduces a delay time between the horizontal and vertical polarization of the pump which precompensates the delay time between the @xmath17 pair generated by the first crystal and the @xmath16 pair from the second one .
signal and idler beams travel through the slm and are spatially selected by two irises and two slits set at @xmath55 mm .
the low quantum efficiency of our detectors ( @xmath56 ) forces us to couple large angular regions : we set @xmath57 mm which corresponds to @xmath35mrad .
an interference filter if ( fwhm @xmath58 nm ) is set on the signal arm and selects the spectral width , whereas a long - pass filter set on the idler arm ( cut - on wavelength @xmath59 nm ) is used to reduce the background .
photons are focused in two multi - mode fibers ( mmf ) and sent to home - made single - photon counting modules , based on an avalanche photodiode operated in geiger mode with passive quenching . in order to measure the visibility , we insert two polarizers , set at the @xmath36 , corresponding to a maximum in the coincidence rate , and @xmath37 , @xmath38 for the minimum . for the tomographic reconstruction ( see below ) we insert on both paths a quarter - wave plate , a half - wave plate and a polarizer . in our setup
, the purification provided by the slm allows us to generate good polarization - entangled states with visibility up to @xmath60 , which may be further increase by spatially filtering the pump to achieve a gaussian profile .
however , ghost imaging also require spatial entanglement and this can be obtained upon exploiting eq .
( [ eq : theta - omega ] ) , i.e. by narrowing the output spectral range .
we use an interference filter , whose action is denoted by @xmath8 $ ] in eq .
( [ eq : state ] ) , which selects a range of about @xmath58 nm within the overall downconversion spectrum ( @xmath61 nm ) . with @xmath15 = @xmath52mrad and polarizers at @xmath62
( [ profhh ] ) ] , measured within a coincidence time window of @xmath63ns , as a function of the signal and idler slit positions @xmath11 and @xmath12 .
the phase matching central angles correspond to @xmath64 . ] in order to observe the resulting angular correlation we place two slits of aperture @xmath15 = @xmath52mrad , two polarizers at @xmath62 along the downconversion arms and we measure @xmath65 at @xmath66 .
the phase matching ( central ) angles correspond to @xmath67 .
coincidences are taken over an acquisition time of @xmath68s within a coincidence time window of @xmath63ns .
the coincidence counts @xmath69 are reported in fig .
[ profili ] and the experimental results confirm that @xmath19 is approaching @xmath20 for the selected ( narrow ) spectral range . in an analogue way we have also measured @xmath70 , checking experimentally that @xmath71
single phase - object imaging consists in setting @xmath72 and reconstruct the phase function @xmath24 inserted along the signal arm by scanning the coincidences for different emission angles @xmath12 on the idler arm @xcite .
experimentally , we insert the phase function @xmath73 using the slm , with @xmath74rad and @xmath75mrad@xmath76 . in fig . [
f : ghost - im ] we present the results for the ghost imaging : polarizers are set to @xmath77 on the signal and @xmath78 on the idler , the slits aperture are @xmath35mrad for the signal and @xmath79mrad for the idler .
the signal slit is centered on @xmath80 , while the idler slit varies over different values of @xmath81 .
the experimental data , already subtracted of the accidental coincidences ( coincidence window equal to @xmath63ns , acquisition time @xmath82s ) are the red circles , whereas the red solid line is the theoretical prediction as obtained from eq .
( [ eq : c ] ) , i.e. @xmath83\notag\,.\end{aligned}\ ] ] for comparison we also report the measured value of the direct counts ( blue squares , acquisition time @xmath84s ) with the corresponding theoretical prediction ( solid blue line ) , i.e. @xmath85 on the signal beam , with @xmath74rad and @xmath75mrad@xmath76 .
blue squares are the direct counts whereas red circles are the coincidences ( red circles ) .
solid lines are the corresponding theoretical predictions ( blue and red lines respectively ) as a function of the idler slit position . from the coincidence counts observed by scanning the idler beam
, one can reconstruct the phase function along the signal beam . ] as a matter of fact , the insertion of a single phase - object leads to the generation of a set of maximally entangled states at different angles , each one with a different phase term @xmath24 .
the entanglement of the state over a broad angular region is thus reduced to @xmath86 .
once we have reconstructed the signal phase @xmath24 we may further tune entanglement by imposing the phase functions @xmath87 on the idler beam . in this way we nonlocally superimpose two phase object .
we remark that , given the correlation condition @xmath88 ( see fig . [ profili ] ) , the overall phase function inserted by the slm is given by @xmath89 the corresponding visibility values are @xmath90 and @xmath91 , respectively .
the visibility of the entangled state is sligthly lower than the original value @xmath60 since the two beams , though showing high angular correlations , are not delta - correlated . in order to fully characterize the output state , and confirm that visibility is a good figure of merit to discriminate phase functions ,
we have also performed state reconstruction by polarization qubit tomography for the two different output states .
the procedure goes as follows : we measure a suitable set of independent two - qubit projectors @xcite and then reconstruct the density matrix from the experimental distributions using maximum - likelihood reconstruction .
the tomographic measurements are obtained by inserting a quarter - wave plate , a half - wave plate and a polarizer .
, with @xmath74rad and @xmath75mrad@xmath76 ( upper panels ) and of the state obtained with @xmath92 ( lower panels ) .
the corresponding visibilities are given by @xmath93 and @xmath94 , while the measured chsh - bell parameters are @xmath95 and @xmath96 respectively .
, title="fig : " ] , with @xmath74rad and @xmath75mrad@xmath76 ( upper panels ) and of the state obtained with @xmath92 ( lower panels ) .
the corresponding visibilities are given by @xmath93 and @xmath94 , while the measured chsh - bell parameters are @xmath95 and @xmath96 respectively .
, title="fig : " ] , with @xmath74rad and @xmath75mrad@xmath76 ( upper panels ) and of the state obtained with @xmath92 ( lower panels ) .
the corresponding visibilities are given by @xmath93 and @xmath94 , while the measured chsh - bell parameters are @xmath95 and @xmath96 respectively .
, title="fig : " ] , with @xmath74rad and @xmath75mrad@xmath76 ( upper panels ) and of the state obtained with @xmath92 ( lower panels ) .
the corresponding visibilities are given by @xmath93 and @xmath94 , while the measured chsh - bell parameters are @xmath95 and @xmath96 respectively .
, title="fig : " ] in fig .
[ fig : qtomo ] we present the tomographic reconstructions of the density matrix of the two output states : the reconstruction of the state resulting from the overall phase function @xmath97 is reported in the upper panel ( real part on the left , and imaginary part on the right ) . in the lower panel
we show the tomographic reconstruction of the density matrix for the state obtained with @xmath92 ( real part on the left , and imaginary part on the right ) . in order to detect the presence of nonlocal correlations we also measured the bell - chsh parameter @xmath98 where @xmath99 denotes the correlations between measurements performed at polarization angle @xmath100 for the mode @xmath101 .
we found @xmath102 i.e. we have violation of chsh - bell inequality @xcite by more than @xmath103 standard deviations for @xmath97 , whereas no violation of the chsh - bell inequality is found for @xmath92 , for which the measures @xmath104 is less than the threshold @xmath105 by @xmath106 standard deviations .
as a possible application of nonlocal phase compensation , we suggest to exploit the switch between entangled and mixed states for quantum key distribution .
our proposal is based on the fact that alice and bob control the signal and the idler arm respectively of the downconversion output .
they are thus able to insert independently , and in a random sequence , the phase functions @xmath107 and @xmath108 , where @xmath109 . in turn
, this may be used to establish a qkd protocol as the analogue of the random choice of the signal basis or of the measurement basis .
alice then encodes the key ( @xmath110 ) by adding a constant phase ( @xmath111 ) to @xmath107 . upon setting the detection polarizers to @xmath77 and @xmath78
we have that for @xmath112 a highly entangled state is shared , and thus we have a maximum in the coincidence counting rate when @xmath113 and a minimum when @xmath114 . on the other hand , if @xmath115 ( or if an eavesdropper tries to acquire knowledge about the key ) , then alice and bob share a mixed ( or partially mixed ) state , and an intermediate counting rate is detected .
$ ] , @xmath116 $ ] , @xmath117 $ ] , and @xmath118 $ ] respectively the first two configurations correspond to the faithful transmission of the key , whereas the two others cases are the events that alice and bob are going to reject in the sifting stage of the protocol . ] in fig .
[ fig : qkd ] we present the experimental results for the coincidence counting rate in the four possible configurations , which are summarized in table [ tabc ] , and labelled as follows : @xmath119 $ ] denotes the case @xmath120 and @xmath113 , corresponding to the faithful transmission of the symbol `` 1 '' ; analogously @xmath116 $ ] denotes the case @xmath120 and @xmath114 , leading to the faithful transmission of the symbol `` 0 '' .
the events that alice and bob are rejecting in the sifting stage are @xmath118 $ ] and @xmath117 $ ] corresponding to @xmath121 and @xmath111 respectively .
.a qkd protocol based on nonlocal phase compensation : the phase functions @xmath107 and @xmath108 play the role of the random choice of the signal basis , whereas the key ( @xmath110 ) is encoded by adding a constant phase @xmath111 to @xmath107 .
when the phase functions are matched , @xmath122 , we have the transmission of the key symbols . [ cols="^,^,^,^,^",options="header " , ] [ tabc ] notice
that the purification procedure , which removes the angular phase terms @xmath123 , allows us to generate high - quality entanglement states even when broad angular regions are coupled .
indeed , the measurements used for quantum key distribution has been performed with a larger aperture , in order to compensate the low quantum efficiency of photodetectors and to increase the measurement rate .
in conclusion , we have suggested and demonstrated experimentally an entanglement - based scheme to achieve coherent nonlocal compensation of pure phase objects .
our scheme is based on creating two - photon polarization and momentum entangled states where the insertion of a single phase object on one of the beams reduces both the purity of the state and the amount of shared entanglement , and where the original entanglement can be retrieved by adding a suitable phase object on the other beam . in our setup polarization and momentum entangled states are generated by spontaneous parametric downconversion and then purified using a programmable spatial light modulator .
the same device is also used to impose arbitrary space dependent phase functions to the beams , which play the role of arbitrary pure phase objects .
finally , we have suggested a novel quantum key distribution protocol exploiting the effect of nonlocal phase compensation and we have provided its experimental verification .
our results prove experimentally the feasiblity of coherent nonlocal compensation / superposition of pure phase objects and pave the way for further developments , as the reconstruction of the overall phase function imposed to the two beams .
mgap thanks alessandra gatti and konrad banaszek for useful discussions
. 99 a. gatti , e. brambilla , and l. a. lugiato , progr . opt . * 51 * , 251 ( 2008 ) .
r. s. bennink , s. j. bentley , and r. w. boyd , phys .
lett . * 89 * , 113601 ( 2002 ) .
f. ferri , d. magatti , a. gatti , m. bache , e. brambilla , and l.a .
lugiato , phys .
lett . * 94 * , 183602 ( 2005 ) .
o. katz , y. bromberg , and y. silberberg appl .
. lett . * 95 * , 131110 ( 2009 ) .
p. h. souto ribeiro , s. padua , j. c. machado da silva , and g. a. barbosa , phys .
a * 49 * , 4176 ( 1994 ) .
t. b. pittmann , y. h. shih , d. v. strekalov , and s. v. sergienko , phys .
a * 52 * , r3429 ( 1995 ) ; d. v. strekalov , a. v. sergienko , d. n. klyshko , and y. h. shih , phys . rev . lett . * 74 * , 3600 ( 1995 ) .
a. f. abouraddy , b. e. a. saleh , a. v. sergienko , and m. c. teich , phys .
lett . * 87 * , 123602 ( 2001 ) .
y. t. zhang , c. j. he , h. g. li , k. g. wang , chin .
. lett . * 25 * , 2481 ( 2008 ) .
s. gan , s.h .
zhang , t. zhao , j. xiong , x. d. zhang , k. g. wang , appl .
. lett . * 98 * , 111102 ( 2011 ) a. gatti , e. brambilla , and l. a. lugiato , phys .
lett . * 90 * , 133603 ( 2003 ) ; a. gatti , e. brambilla , m. bache , and l. a. lugiato , phys . rev . a * 70 * , 013802 ( 2004 )
. a. f. abouraddy , p. r. stone , a. v. sergienko , b. e. a. saleh , and m. c. teich , phys .
lett . * 93 * , 213903 ( 2004 ) a. m. steinberg , p. g. kwiat , and r. y. chiao , phys . rev
. lett . * 68 * , 2421 ( 1992 ) . c. bonato , a. v. sergienko , b. e. a. saleh , s. bonora , p. villoresi , phys .
* 101 * , 233603 ( 2008 ) .
o. minaeva , c. bonato , b. e. a. saleh , d. s. simon , a. v. sergienko , phys .
lett . * 102 * , 100504 ( 2009 ) .
t. wasak , p. szakowski , w. wasilewski , k. banaszek phys .
a * 82 * , 052120 ( 2010 ) j. d. franson phys .
a * 45 * , 31263132 ( 1992 ) .
s - y . baek , y - w .
cho , y - h .
kim , opt .
* 17 * , 19241 ( 2009 ) . l. hardy , phys .
a * 161 * , 326 ( 1992 ) .
kwiat , e. waks , a. g. white , i. appelbaum , and p.g .
eberhard , phys .
a * 60 * , r773 ( 1999 ) .
m. genovese , phys . rep . *
413 * , 319 ( 2005 ) .
h kim , s. p. kulik , y. shih , phys .
a * 62 * , 011802 ( 2000 ) .
y. nambu , k. usami , y. tsuda , k. matsumoto , k. nakamura , phys .
a * 66 * , 033816 ( 2002 ) . c. cinelli , g. di nepi , f. de martini , m. barbieri and p. mataloni , phys . rev .
a * 70 * , 022321 ( 2004 ) ; m. barbieri , c. cinelli , f. de martini and p. mataloni , las .
* 16 * , 1439 ( 2006 ) .
g. brida , m. genovese , m. v. chekhova , l. a. krivitsky , phys .
a * 77 * , 015805 ( 2008 ) .
r. rangarajan , m. goggin , and p. kwiat , opt
* 17 * , 18920 ( 2009 ) .
s. cialdi , d. brivio , m.g.a .
paris , appl .
lett . * 97 * , 041108 ( 2010 ) .
y. bromberg , o. katz , and y. silberberg , phys .
a * 79 * , 053840 ( 2009 ) .
j. h. shapiro , phys .
a * 78 * , 061802 ( 2008 ) .
p. clemente , v. durn , v. torres - company , e. tajahuerce , and j.lancis , opt , lett .
* 35 * , 2391 ( 2010 ) .
s. cialdi , d. brivio , m. g. a. paris , phys .
a * 81 * , 042322 ( 2010 ) .
s. cialdi , f. castelli , i. boscolo , m. g. a. paris , appl . opt . * 47 * , 1832 ( 2008 ) ; s. cialdi , f. castelli , m. g. a. paris , j. mod . opt . * 56 * , 215 ( 2009 ) .
k. banaszek , g. m. dariano , m. g. a. paris , m. f. sacchi , phys .
a * 61 * 010304(r ) ( 1999 ) .
d. f. v. james , p. g. kwiat , w. j. munro , and a. g. white , phys .
a * 64 * , 052312 ( 2001 ) .
clauser , m.a .
horne , a. shimony , and r.a .
holt , phys .
rev . lett . * 23 * , 880 ( 1969 ) . | we suggest and demonstrate a scheme for coherent nonlocal compensation of pure phase objects based on two - photon polarization and momentum entangled states .
the insertion of a single phase object on one of the beams reduces the purity of the state and the amount of shared entanglement , whereas the original entanglement can be retrieved by adding a suitable phase object on the other beam . in our setup polarization and momentum entangled states
are generated by spontaneous parametric downconversion and then purified using a programmable spatial light modulator , which may be also used to impose arbitrary space dependent phase functions to the beams . as a possible application ,
we suggest and demonstrate a quantum key distribution protocol based on nonlocal phase compensation . |
ever since the discoveries of the first extra - solar planets ( @xcite , @xcite , @xcite ) , exoplanetary science has been one of the hot topics in astronomy in the past two decades .
high - precision instruments and missions such as harps @xcite , corot @xcite , kepler @xcite and the future plato - mission have allowed the number of known exoplanets to grow rapidly .
up to date , the exoplanet encyclopaedia ( _ exoplanet.eu _ ) lists a total of 1822 planets in 1137 planetary systems .
better instruments and enhanced observational techniques are pushing the boundaries of detectable planets down to the super - earth group . in order to achieve this goal , target host stars decrease in mass .
m - dwarfs , and also brown dwarfs @xcite , are suggested as they have smaller radii , masses and are less luminous , presenting opportunities for detecting smaller planets orbiting around them , possibly even within their respective habitable zones .
the solar neighbourhood has been photometrically , spectroscopically and astrometrically studied by the recons team @xcite in order to understand the distribution of stellar types nearby .
their latest finding @xcite indicate that m and later type stars account for 60 - 70% of the stellar population within 10pc of the sun .
the fact that they are so numerous additionally increases the chances of planet detections , making m - dwarfs and brown dwarfs even more desirable survey targets . on the other hand , habitability on planets around these stars
will be limited by their magnetic activity ( see @xcite for details ) . the exoplanet encyclopaedia list
a total of 36 confirmed planets around m - dwarfs with about 2/3 of them with masses under 0.2m@xmath7 .
there are no detections of planets around brown dwarfs so far .
data from kepler suggests that early m - dwarfs have an occurrence rate of , on average , 0.90@xmath8 planets per star with planet parameters in range 0.5 - 4r@xmath9 and p @xmath10 50days @xcite . @xcite
combine radial velocity and adaptive optics direct imaging observations for a sample of 111 m stars .
they report that 6.5 @xmath11 3.0% of the m - dwarfs host a gas giant with mass between 1 - 13m@xmath7 and semi - major axes of less than 20au , corresponding to 0.083 @xmath11 0.019 planets per star in that parameter space .
these results suggest that planets around m dwarfs are abundant , motivating future studies to characterize them in detail ( e.g. @xcite ) .
the evaluation of planet parameters is tightly correlated with the host star s parameters ( e.g. @xcite ) .
therefore our knowledge about extrasolar planets is limited by how well we can characterize the host stars .
the challenge of determining fundamental stellar parameters is not new ( see @xcite ) and not restricted to planetary host stars ( e.g. @xcite and references therein ) .
@xcite generate evolutionary tracks for brown dwarfs and very low mass stars ( vlms ) for different atmospheric metallicities with and without clouds . by comparing observational data to these tracks
, their study demonstrates a variety of plausible stellar radii , and narrowing down this range for a given mass depends on precise estimates of stellar age and metallicity .
@xcite use inverse modelling of directly - imaged data for hr 8799b .
their results indicate that reasonable fits to the data can be obtained for both cloudy and cloud - free atmospheres but with different values for metallicities and element abundances .
such studies indicate the difficulty of inferring precise values for stellar parameters based on atmospheric models .
both , variations in underlying physical assumptions between models and different parameter values , within the same model can lead to a spread in estimates for stellar mass and radii .
it is therefore important to be aware of the limitations of model atmospheres and how they compare to each other .
this paper focuses on the comparison of different model atmosphere families with some focus on the m - dwarf parameter space : effective temperature t@xmath12k , surface gravity spans log(g)@xmath13 ( included young m - dwarfs , log(g)@xmath144.0 , and brown dwarfs , log(g)=5.0 ) , and metallicity [ m / h]@xmath15 ( appendix [ appa ] ) .
not all parameter combinations are available for all model families .
section [ ss : atm ] gives a brief overview of the atmosphere models used in this study . in sect .
[ ss : com ] we explore the similarities and differences in the atmospheric structure of the model families . in sect .
[ ss : phot ] we present the results for the synthetic photometry comparisons .
section [ ss : dis ] contains our discussion .
the following model atmosphere families are included in the comparison study presented in this paper : + all models assume lte , hydrostatic and chemical equilibrium and obey radiative and convective flux conservation .
they model a homogeneous , 1d oxygen - rich atmosphere in plane - parallel geometry .
phoenix models were available in spherical symmetry .
the grid of atlas models utilised spans the following range of stellar parameter : t@xmath6 = 3500 k@xmath14000 k , log(g ) = [email protected] , [ m / h ] = @xmath16 .
all atlas models were calculated with the convection option switched on but with the overshooting option switched off .
the mixing length parameter @xmath17 , ( @xmath18 , @xmath19 local pressure scale height , @xmath20 - local gas pressure , @xmath21 - local gas density , @xmath22 - local gravitational acceleration , where @xmath23 ) , the micro - turbulence velocity v@xmath24 = 2.0 km / s , and solar element abundances from @xcite are used in all atlas models considered here .
the marcs models used span a grid of t@xmath6 = 2500k@xmath14000k , log(g ) = [email protected] , [ m / h ] = @xmath16 . for all marcs models , v@xmath25= 2 km / s , mixing length parameter with @xmath26 and solar element abundances @xcite .
the phoenix models considered are for t@xmath27 = 2500k@xmath14000k , log(g ) = [email protected] , [ m / h ] = 0.0 and @xmath28-element abundance of [ @xmath28/m ] = 0.0 .
values for the mixing length parameter @xmath29 depending on the stellar parameters as depicted in fig . 2 in @xcite .
figures [ tp2500][tp4000 ] provide the detailed information regarding the model atmospheres compared .
the micro - turbulence velocities v@xmath30 km / s according to fig . 3 in @xcite .
the element abundances are solar ( @xcite ) .
the drift - phoenix models are aimed specifically at late - type stars ( m - dwarfs , brown dwarfs ) and giant planet atmospheres as they also model dust cloud formation . the drift module deals with dust treatment , calculating a consistent cloud structure and passing it to the main radiative transfer code ( phoenix ) .
the subset of models used is for the solar metallicity models with 2500k @xmath14 t@xmath31 3000k and 3.0 @xmath14 log(g ) @xmath14 5.5 . mixing length
is set to 2.0 scale heights and micro - turbulence velocity is 2.0 km / s .
solar elements abundances are those from @xcite the different model atmosphere families of models cover different parts of the m - dwarf regime , with atlas barely touching early - type m stars , drift - phoenix covering the late end of this spectral type and marcs and phoenix spanning the entire m - dwarf range .
the different sets of element abundances applied for different model families are summarized in table [ el_abund ] .
all non - solar metalicities are derived from scaled solar values .
more detailed information about the models , e.g. regarding the used opacity sources , are provided in the discussion sect [ ss : diffm ] .
the model atmospheres under investigation do not contain one m - dwarf parameter set that is common to all of them .
therefore , we compare subsets of model families : the atlas@xmath32marcs models for t@xmath6 = 3500k and t@xmath6 = 4000k and varying log(g ) and [ m / h ] values , atlas@xmath32marcs@xmath32phoenix for for t@xmath6 = 3500k and
t@xmath6 = 4000k , [ m / h ] = 0.0 and varying log(g ) , as well as the marcs+drift - phoenix , marcs+phoenix and phoenix+drift - phoenix models for solar metallicity and varying t@xmath6 and log(g ) values .
a total of 141 models were examined .
appendix [ appa ] summarize the parameter values of all models used . | m - dwarfs are emerging in the literature as promising targets for detecting low - mass , earth - like planets . an important step in this process is to determine the stellar parameters of the m - dwarf host star as accurately as possible .
different well - tested stellar model atmosphere simulations from different groups are widely applied to undertake this task .
this paper provides a comparison of different model atmosphere families to allow a better estimate of systematic errors on host - star stellar parameter introduced by the use of one specific model atmosphere family only .
we present a comparison of the atlas9 , marcs , phoenix and drift - phoenix model atmosphere families including the m - dwarf parameter space ( t@xmath0k@xmath14000k , log(g)[email protected] , [ m / h]=@xmath2 ) .
we examine the differences in the ( t@xmath3 , p@xmath4)-structures , in synthetic photometric fluxes and in colour indices .
model atmospheres results for higher log(g ) deviate considerably less between different models families than those for lower log(g ) for all t@xmath5k@xmath14000k examined .
we compiled the broad - band synthetic photometric fluxes for all available model atmospheres ( incl .
m - dwarfs and brown dwarfs ) for the ukirt wfcam zyjhk , 2mass jhks and johnson ubvri filters , and calculated related colour indices .
synthetic colours in the ir wavelengths diverge by no more than 0.15 dex amongst all model families . for all spectral bands considered , model discrepancies in
colour diminish for higher t@xmath6 atmosphere simulations .
we notice differences in synthetic colours between all model families and observed example data ( incl .
kepler 42 and gj1214 ) .
stars : atmosphere models stars : synthetic photometry stars : colour indices |
as one of the richest old open clusters in our galaxy , ngc 188 has been extensively studied .
the breadth of available data , old age , and the abundance of short - period photometrically variable stars and x - ray sources , make ngc 188 a very attractive cluster for both observational and theoretical astronomers .
moreover , the multitude of photometric surveys @xcite , spectroscopic studies @xcite , short - period variable star studies @xcite , including both optical and x - ray photometric monitoring , proper - motion ( pm ) analyses @xcite , and the binary population study by @xcite provide an unprecedented basis from which to study an open cluster .
however , until now , the study of this cluster has lacked a critical component .
this paper presents the first high - precision , comprehensive radial - velocity ( rv ) survey of the cluster , thereby providing the first census of the hard - binary population , measurements of the internal stellar kinematics , and improved membership information .
the recent photometric studies by @xcite and @xcite used ubvri ccd photometry to derive a reddening of @xmath7 , an apparent distance modulus of @xmath8 ( 1.9 kpc ) , and an age of 7 gyr .
the most recent pm study by @xcite ( hereafter , p03 ) suggests that ngc 188 contains @xmath31500 cluster members ( membership probability @xmath5 10% ) down to a limiting magnitude of @xmath2=21 within 17 pc in projection .
we have used the p03 study to form our current cluster sample ( described in section [ sam ] ) .
this rv analysis of ngc 188 is part of the wiyn open cluster study ( wocs ; * ? ? ?
we calculate rv membership probabilities for 848 solar - type stars with @xmath53 radial - velocity measurements ( section [ mem ] ) , finding 349 single cluster members and 124 likely member velocity - variable stars ( section [ var ] ) .
( in the following , we use the term `` single '' to identify stars with no significant rv variation . certainly , many of these stars are also binaries , although generally with lower total mass than the binaries identified in this study .
we assume that the velocity - variables detected in this study are binary- or multiple - star systems . )
combined with available pm data , our precise three - dimensional ( 3d ) kinematic memberships define the most accurate sample of solar - type members of ngc 188 to date .
we use this member sample to construct a cleaned color - magnitude diagram of the cluster ( section [ cmd ] ) from which we identify a blue - straggler ( bs ) population as well as six intriguing stars that lie far from a standard single - star isochrone .
finally , we investigate the spatial distribution of cluster member populations ( section [ spat ] ) and the cluster s velocity dispersion ( section [ disp ] ) .
following papers will provide orbital solutions for most of the velocity variables , likely all of which will be hard binary stars that dynamically power the cluster , as well as detailed comparisons with @xmath9-body models ( e.g. , * ? ? ?
our ongoing wiyn rv survey of ngc 188 began in 1996 .
the following section describes the development of our stellar sample over the course of these eleven years .
our observational history as well as current strategy have been dependent on the available position , photometry and membership measurements .
initially , we composed our stellar sample from the @xcite pm study , which contains photometry and pms ( @xmath10 = 0.5 mas yr@xmath4 ) for 1127 stars covering a 50 x 60 area within the limiting magnitudes of @xmath11 and @xmath12 .
@xcite found 360 cluster members in their sample ( probability @xmath5 70% ) .
these cluster members were prioritized in our observations ( see section [ obs ] ) .
later , a more extensive pm study was performed by p03 , complete to @xmath13 and covering @xmath14 , centered on the j2000.0 coordinates : @xmath15 and @xmath16 .
the pms have a precision down to 0.15 mas yr@xmath4 per coordinate , and the coordinates are accurate to 5 - 10 mas on the international celestial reference system ( icrs ) . of the 7812 objects that were detected ,
p03 found 1490 pm members ( probability @xmath5 10% ) across all magnitudes in the study .
thus , in 2003 , we chose to update our data set to include the additional stars ( both pm members and non - members ) from this study that lie within our magnitude limits , and to use the more accurate coordinates .
our complete stellar sample includes 1498 stars .
the hydra multi - object spectrograph ( mos ) on the wiyn 3.5 m telescope has an effective dynamic range of approximately four magnitudes and a 1 field of view .
fortuitously , the p03 study also covers 1 in spatial extent ; we therefore did not make any spatial cut in our sample .
we have set our limiting magnitude range to 12@[email protected] , covering a large portion of the giant branch through the upper main sequence , including the cluster turnoff and bs stars . at @xmath2 = 16.5 ,
the faint end of our observations extends to @xmath31.5 magnitudes below the cluster turnoff ( e.g. , figure [ cmds ] ) .
our observing setup allows us to consistently determine rvs for stars whose @xmath17 0.4 ; earlier type stars are often rapidly rotating and have fewer absorption lines in our wavelength range .
since our sample includes only 30 stars blueward of this , we also observed them so as to include those that are narrow - lined .
currently , our stellar sample contains stars whose @xmath18 color lies within the range of @xmath19 . after limiting the p03 study by magnitude
, we have a total stellar sample of 1498 stars , comprising a complete sample of stars within our magnitude range and within 0.5 ( 17 pc at a distance of 1.9 kpc ) from the cluster center . the dao data set contains multiple rv measurements of 77 stars within the magnitudes of 10.8@xmath0v@xmath016 covering dates from december 1973 through november 1996 . observations
prior to 1980 were taken by roger griffin and james gunn at the palomar 5 m telescope ( e.g. , * ? ? ?
* ) ; later observations were made by robert mcclure , hugh harris and jim hesser at the dominion astrophysical observatory ( dao ) ( e.g. , * ? ? ?
* ; * ? ? ?
all rv measurements were then converted onto the dao radial - velocity spectrometer ( rvs ) system .
we have used the hydra multi - object spectrograph ( mos ) on the wiyn 3.5 m telescope to conduct our ongoing wocs rv survey of ngc 188 from 1996 to the present .
the telescope combines a large aperture and wide field of view , and provides excellent image quality ( 0.8 arcsecond median seeing ) .
the mos is capable of obtaining simultaneous spectra of up to @xmath380 stars over the 1 field during each configuration ( or _ pointing _ ) .
the hydra fiber positioner executes a fiber setup in @xmath320 minutes with a precision of 0.2@xmath20 .
the isolated fiber - fed bench spectrograph is of conventional design with an all - transmission camera to avoid central obstruction of the filled beam .
we use the blue fibers for our observations , each fiber having a 3.1@xmath20 aperture and an effective spectral window of 300 - 700 nm .
the majority of our observations are centered on 513 nm and extend 25 nm in either direction , covering a rich array of narrow absorption lines and the mg b triplet .
we utilize the x14 filter along with the echelle grating at 11th order , obtaining a resolution of approximately 20000 and a dispersion of 0.13 / pixel on the ccd .
a portion of our observations are centered on 640 nm ( also extending 25 nm in either direction ) .
the detector is a thinned 2048 x 2048 tektronics ccd with 24@xmath21 m pixels and a typical resolution element of @xmath3 2.5 pixels ( @xmath3 19 km s@xmath4 ) . during each pointing
we may utilize @xmath380 fibers to measure simultaneous spectra .
we will generally use @xmath310 fibers to measure the sky , and place the remaining @xmath370 fibers on stars .
therefore , developing a strategy for the placement of these science fibers is critical in order to optimize limited observing time .
this section discusses our chosen strategy for fiber placement as well as our general observing routine .
prior to creating our `` pointing files '' for a given observing run , we perform a detailed prioritization of the stars in our sample based on our available data . in general
, we place our discovered velocity - variable stars at the highest priority , followed by pm members , pm non - members , and finally , those stars that we consider finished . based on monte carlo simulations ,
we require 3 epochs of rv measurements over at least a year to ensure 95% confidence that a star is either constant or variable in velocity ( i.e. , a single star or a binary- or multiple - star system ; * ? ? ?
* ; * ? ? ?
* ) . with three observations in hand
we either derive secure membership probabilities for single stars and place them at the lowest priority ( considering them finished ) , or move binaries to the highest priority in order to expedite orbital solutions . to maximize progress towards orbital solutions , the short - period binaries are the highest priority for observation each night .
longer period binaries are the next priority for 1 - 2 observations per run .
all other fibers are placed on stars for completion of the survey of the entire cluster .
1.5 km s@xmath4 ) , followed by once- , twice- , and un - observed pm members , and finally once- , twice , and un - observed pm non - members . each section is then sorted by distance from the cluster center , to maximize fiber density in the crowded central field . ] during a given observing run , we produce a `` faint master '' and a `` bright master '' which will then be used to make the initial pointing files .
the faint master contains all of the stars from our sample while the bright master contains only stars within the range of 12.5@[email protected] ( 777 object stars ) to be used in the case where light cloud cover would likely prevent us from reliably deriving velocities for the fainter stars . in general
, we use a total of two hours of integration time for the faint pointings split into three 2400-sec integrations . for bright pointings , we take one hour of total integration time split into three 1200-sec integrations . by splitting up our integrations ,
we increase our ability to remove cosmic rays that may appear in our spectra .
calibration is achieved through one 200-sec flat field and two 300-sec thar ( for the 513 nm range ) or cuar ( for the 640 nm range ) emission lamp spectra for each pointing .
currently , we have observed ngc 188 using wiyn over 53 different observing runs totalling 119 separate pointings .
the dao rvs were measured using the dao rvs on the 1.2-meter telescope .
this instrument observes 43 nm of the spectrum centered at 455 nm with a resolution 0.2 , scanning a mask rapidly across the stellar absorption lines .
the properties of the instrument and the observing procedures can be found in @xcite and @xcite .
the ultimate goal of our data reduction routine is to take our ccd image ( the light from the @xmath380 fibers ) , extract the @xmath370 science spectra from this image , and obtain @xmath370 heliocentric rvs .
we describe our data reduction procedure in some detail here as the basic reference for subsequent rv publications from this project . initially , we come away from the telescope with raw fits images of the dispersed light from our science and calibration integrations .
we generally take two thar ( or cuar ) calibration images per pointing , one before the first integration and one after the last integration , in order to account for any potential wavelength shift that may occur during the one or two hours of total integration .
we also take one dome flat per pointing .
all image processing is done within the iraf data reduction environment @xcite .
[ [ bias - subtraction ] ] bias subtraction : + + + + + + + + + + + + + + + + + the wiyn mos ccd contains an overscan strip of 32 x 2048 pixels , containing only the readout bias level . for each row ,
the 32 columns are averaged .
a cubic - spline function is fit to this averaged bias column and is subtracted from each column of the science image .
[ [ spectra - extraction ] ] spectra extraction : + + + + + + + + + + + + + + + + + + + first , we trace the apertures along the image ; dome flats are used to perform this trace as the fibers are well illuminated .
we fit a legendre polynomial function to the position of each spectrum on the chip .
these fits are then saved , to be used later in extracting each science spectrum aperture .
second , we average the flat - field intensity across all fibers , fit a cubic - spline function to this single flat - field spectrum , and , again , save this fit for use in flat fielding the science spectra .
finally , the dome flat is used to correct for fiber - to - fiber throughput variations .
each extracted flat field aperture is averaged to a single number which is then multiplied by the average flat field intensity function so that when the science apertures are flat fielded later , they will be throughput corrected simultaneously . we then determine a pixel to wavelength map using the thar ( or cuar ) calibration spectra , first using the long ( 900-sec ) exposure taken during the day , and then the shorter ( 300-sec ) exposures related to the science images .
we use the long calibration image for our initial map , as the extracted high signal - to - noise spectra yield more precise results . after extracting the spectra from the long calibration image
, we determine a dispersion solution for each aperture by fitting a cubic - spline function to the thar ( or cuar ) emission line wavelengths and pixel centroids .
we then transfer these solutions as initial guesses onto the extracted short calibration spectra associated with our science spectra .
our typical rms residuals around the emission line wavelengths for a fit to a long calibration spectrum vary from @xmath30.006 - 0.009 , and increase slightly for the short calibration exposures . finally , the science spectra are extracted , flat fielded , throughput corrected , and dispersion corrected .
the result is a multi - spectrum fits image containing all science spectra for a given integration .
the preceding steps are performed for each of a set of integrations for a given field using the associated flat fields and calibration images .
[ [ sky - subtraction ] ] sky subtraction : + + + + + + + + + + + + + + + + we generally do not observe during dark time and therefore often obtain some scattered moonlight in our object spectra . in order to avoid the potential confusion due to a second spectrum of different velocity
, we must remove the solar absorption lines from our spectra .
this is achieved through the use of our @xmath310 sky fibers .
we inspect each sky spectrum and then use all acceptable sky spectra from our pointing to create one averaged sky spectrum per integration .
( inclusion of a noisy , low quality or accidental stellar spectrum will degrade our science spectra when sky subtracted . ) during the averaging , the sky spectra are run through an averaged sigma clipping algorithm in order to remove cosmic rays .
finally , this combined sky spectrum is subtracted from each object spectrum , and the procedure is repeated for all integrations . at this point , we generally have three extracted and sky subtracted sets of spectra for each pointing ( i.e. , one set of spectra for each integration of a given pointing ) , which we combine to produce one set of spectra per pointing .
we perform a median filter combination to remove any cosmic ray contamination from our object spectra .
[ [ cross - correlation ] ] cross correlation : + + + + + + + + + + + + + + + + + + we then calculate rvs from these combined object spectra using the iraf task _
fxcor_. this task uses the fourier transforms of two spectra ( generally one science and one template spectrum ) to quickly compute the cross - correlation function ( ccf ) between them , and hence derive their velocity difference .
we fit a gaussian function to the ccf peak and find the centroid , obtaining the velocity offset of the object spectrum from the template .
we allow iraf to correct this velocity for the template offset and the motion of the earth around the sun to return a heliocentric velocity .
these heliocentric velocities are the final product of our data reduction routine . since we are observing solar - type stars , we have chosen to correlate each object spectrum against a high signal - to - noise solar spectrum obtained from a daytime sky observation .
this solar template spectrum was obtained using the same observing setup as used for our science spectra .
currently , the same template is used to reduce all of our wiyn observations . however ,
certainly not all of our stars have a g2v spectral type .
therefore we have investigated for a potential offset in velocity resulting from a template mismatch by binning our measured single member star rvs as a function of @xmath18 color .
there is no evidence for a systematic offset to the level of 0.1 km s@xmath4 .
we are therefore confident that our choice of a g star template has not affected the accuracy of our rv measurements .
information regarding the dao rvs data reduction routine can be found in @xcite and @xcite .
here we present a thorough analysis of our wiyn rv measurement precision and the various sources of dispersion in our measurements . in section [ qual ]
, we briefly discuss our quality control procedures . in section [ prec ]
we derive our measurement precision through the use of the @xmath22 function .
finally , we provide a detailed empirical study of our error budget in section [ erbud ] .
generally the highest signal - to - noise spectra result in the highest ccf peak height values .
we use the peak height of the ccf fit as the measure of the quality of a given measurement in the wiyn data set .
only measurements whose fits lie above a certain cutoff value in ccf peak height are deemed reliable . in order to determine this cutoff value , we first determine the offset of each rv measurement of a star from the mean rv of that star , using only single stars .
we then plot these offsets , for all rv measurements of all single stars , against ccf peak height ( figure [ ccfcut ] ) .
it is clear from figure [ ccfcut ] that above a ccf peak height of @xmath30.4 the rv offsets are quite small while those measurements whose ccf peak heights are lower than 0.4 show a very large scatter .
we therefore adopt a ccf peak height cutoff value of 0.4 , and present here only measurements of this quality .
in addition to our ccf peak height quality criterion , we examine the distribution of rvs for each individual star and visually inspect any measurements that are outliers in the distribution .
occasionally we remove measurements whose ccf , though having a peak height above 0.4 , clearly provides a spurious measurement ( e.g. , inadequate sky subtraction ) . in order to determine the precision of our wiyn rv measurements , we fit a @xmath23 distribution to our empirical distribution of rv standard deviations ( following @xcite ) . the theoretical @xmath23 curve models
the distribution of errors on measurements of single stars .
this function depends on both the precision of all measurements as well as the number of degrees of freedom ( i.e. , one less than the number of measurements ) . for consistency across all stars ,
many of which have only three measurements , here we have used only the first three measurements of each star to calculate the standard deviation of its measurements10% for three measurements ) correction to standard deviations measured using the traditional formula .
therefore we have decided to use the traditional formula to calculate our standard deviation values , and note that we derive the same global precision on our measurements using either method . ] . in order to derive this precision value
, we fit a @xmath23 function with 2 degrees of freedom to the distribution of the rv standard deviations : @xmath24 here , @xmath25 denotes the standard deviation of the rv measurements for a star , and @xmath26 represents the overall precision of our measurements .
the empirical distribution ( figure [ 3all.eoveri ] ) represents the combination of two populations : the single stars ( with lower @xmath25 ) , and the velocity - variable stars ( with higher @xmath25 ) .
the @xmath23 function is designed to fit only the single stars .
for this reason , we chose a maximum @xmath25 for our fit , above which the population is dominated by velocity - variable stars .
we used a standard least - squares determination , varying the value @xmath26 as well as a cutoff value in @xmath25 , to fit the @xmath22 function to our data . a maximum @xmath25 cutoff of 0.7 km s@xmath4 ( shown by the dashed line in figure [ 3all.eoveri ] ) and @xmath26 = 0.4 km s@xmath4 produced the smallest residuals and hence the best fit to the peak at lower @xmath25 .
figure [ 3all.eoveri ] shows the rv standard deviation histogram of all stars with @xmath27 wiyn measurements along with the best fit @xmath22 distribution function .
this analysis yields a nominal precision on our wiyn data of @xmath26 * = 0.4 km s@xmath4*. note that towards higher standard deviations , the data are populated almost entirely by binary stars .
this is reflected in the departure of the data from the theoretical @xmath23 curve .
we can therefore use this plot to facilitate our binary identification routine by assuming that if a given star has a @xmath25 that is higher than a certain value , the star is a velocity variable . by inspection of figure [ 3all.eoveri ]
we note that the theoretical curve drops to very near zero at @xmath28 3 times @xmath26 .
we therefore define our velocity - variable stars as those stars whose rv standard deviations ( by convention , we use _ e _ in place of @xmath25 ) are greater than 4 times our quoted precision ( by convention , we use _ i _ in place of @xmath26 ) , or @xmath29 @xmath28 4 .
we plot this value of @xmath29 = 4 as the dotted line in figure [ 3all.eoveri ] , and we discuss our velocity variables in further detail in section [ var ] .
although we have calculated a single value for our wiyn precision the dispersion of our measurements is affected by photon error , and is thus a function of @xmath2 magnitude .
figure [ v.ccf.eoveri ] shows @xmath22 calculated values of @xmath26 as functions of ccf peak height and @xmath2 magnitude . for fit quality ,
our precision ranges from 0.22 km s@xmath4 for the highest quality rvs and degrades to 0.80 km s@xmath4 towards our ccf peak height cutoff value of 0.4 .
we see a similar range when looking at the precision as a function of @xmath2 magnitude ; the brightest stars are measured to a precision of 0.25 km s@xmath4 while the measurement precision on our faintest stars is reduced to 0.55 km s@xmath4 .
the range in @xmath22-fit values for our precision agrees very well with that derived from our error budget analysis in section [ erbud ] .
we are therefore confident in quoting our overall precision to be @xmath26 = 0.4 km s@xmath4 , which can then be used to identify our binary stars and to evaluate the velocity dispersion of the cluster . in this section
, we present a thorough empirical analysis of the error sources contributing to our measurement precision ; a preliminary report can be found in @xcite .
our internal measurement errors have both systematic and random components that include fiber - to - fiber variations ( @xmath30 ) , photon error ( @xmath31 ) , and pointing - to - pointing variations ( @xmath32 ) , each of which we discuss and quantify here .
sky flats demonstrate the existence of fiber - to - fiber variations of unknown origin . during our eleven years of monitoring the stars in ngc 188 , different configurations of the hydra mos result in the same star being measured through different fibers . thus
fiber - to - fiber variations represent a source of internal error on the set of measurements for a star .
fortunately this error is systematic and can be corrected . during each observing run , we take at least one sky flat as a calibration measure , with all fibers arranged in a circle in the focal plane .
when corrected for the earth s rotational and orbital motions , all fibers should provide radial velocities of 0 km s@xmath4 .
in reality , it has become clear that each fiber has a unique and consistent offset of @xmath33 1 km s@xmath4 .
specifically , for each sky flat we derive a mean velocity across all of the fibers , and then measure the velocity offset from this mean for each fiber .
we take the means of these velocity offsets over @xmath310 years of observing ngc 188 for each of the two observed wavelength ranges , and the results are plotted in figure [ fiber ] .
the offsets for the 513 nm region range from 0.00 km s@xmath4 to 0.70 km s@xmath4 .
the mean of the standard deviations on each mean offset measurement is 0.13 km s@xmath4 .
analysis of the 640 nm range results in similar values .
we assume that the derived offset for a given fiber is independent of the position of the fiber on the focal plane .
our rv measurements are corrected for these offsets on a fiber - by - fiber basis . as described in section [ obs ] , each rv measurement derives from three combined sequential integrations obtained in the same fiber configuration .
since these three integrations are almost contemporaneous and are made through the same fibers for any given star , the standard deviations of the three velocities measured from each of the three integrations must arise from photon error alone .
figure [ photer](a ) shows the distribution of such standard deviations as a function of @xmath2 magnitude , and provides a direct measurement of our photon error .
it is clear that the photon error is a function of @xmath2 magnitude , as expected .
( here we have only included 2400-second integrations . ) indeed , if we bin the data shown in figure [ photer](a ) by intensity ( @xmath34 ) , the trend is well fit by @xmath35 . in figure [ photer](b ) ,
we plot the standard deviations of the rv measurements for each single star as a function of @xmath2 magnitude . for @xmath2 @xmath36 14 ,
the morphologies of figures [ photer](a ) and [ photer](b ) are nearly identical .
for these fainter stars , the dispersion on our measurements is dominated by photon error . however , at @xmath2 @xmath37 14 , the values in figure [ photer](a ) drop below those in figure [ photer](b ) . here , the photon error is lower than our empirical precision .
indeed , at @xmath2 @xmath33 14 , we observe a constant dispersion with an average value of @xmath30.3 km s@xmath4 while our average photon error in figure [ photer](a ) is only @xmath30.2 km s@xmath4 .
we must therefore identify a second component to our random error .
we conjecture that this component is due to systematic offsets in each pointing that are randomly distributed in time .
we label this component the pointing - to - pointing variation : @xmath32 .
thus , we define our internal sources of random error as : @xmath38 because we do not have multiple observations of identical pointing configurations , we need to determine @xmath32 via a statistical approach .
we begin by defining the quantity @xmath39 , the dispersion of the mean rvs of all pointings .
specifically , for each pointing we take the mean rv of the single cluster members , which we call the pointing velocity .
then @xmath39 is the standard deviation of all of these pointing velocities .
note that @xmath39 is not simply @xmath32 , because each pointing velocity derives from a different sample of single cluster members .
the dispersion among pointing velocities is thus a consequence of the pointing - to - pointing variations , the photon errors , and the true motions of the stars .
we digress here briefly to discuss the last .
stars in a gravitationally bound cluster will orbit the center of mass of the system .
the orbital motions around the cluster center - of - mass is typically characterized by a velocity dispersion ( @xmath40 ) .
in addition , orbital motions introduced by a binary companion will offset the individual rv measurements of a given star from the binary system s center - of - mass motion .
when a binary orbital solution is available the true center - of - mass motion is known
. however , there are certainly binaries that we can not detect spectroscopically ( e.g. , long - period binaries ) , and are thus included as single ( constant - velocity ) stars .
these undetected binaries add to the dispersion on our measurements .
we characterize the distribution of undetected binary motions as a gaussian with dispersion @xmath41 .
therefore , we call the combination of the cluster motions and undetected binary motions the combined velocity dispersion : @xmath42 thus , we define the measured pointing variation as : @xmath43}{n } + \sigma_{p - p}^2\ ] ] where @xmath9 corresponds to the average number of single cluster member stars in a given pointing ( within the desired magnitude range ) .
using equation [ ine ] , the apparent dispersion of the rv measurements for a set of stars can be written as : @xmath44 + \sigma_{p - p}^2\ ] ] where @xmath25 is the standard deviation of all single stars mean velocities ( in the appropriate magnitude domain ) .
combining equations [ mp ] and [ scm2 ] allows us to eliminate the bracketed values and recover the pointing - to - pointing dispersion : @xmath45 since the photon error is magnitude dependent , we have chosen to divide our calculation into two magnitude ranges : ( 1 ) 12@xmath0v@xmath3314 and ( 2 ) 15@[email protected] .
we can calculate values for @xmath47 directly from the analysis that lead to figure [ photer](a ) .
next , in domain ( 1 ) we find @xmath39 = 0.33 km s@xmath4 , while in domain ( 2 ) we find @xmath39 = 0.36 km s@xmath4 .
note , it is just fortuitous that both of these values are so close , as the larger photon error at @xmath2 @xmath28 15 is offset by the larger number of stars in this magnitude range . using these values
, we find a @xmath32 to equal 0.26 km s@xmath4 in domain ( 1 ) and 0.33 km s@xmath4 in domain ( 2 ) .
the closeness of these values leads us to suspect that the origin of these small pointing - to - pointing variations is not magnitude dependent .
we proceed to combine our error budgets to compute our internal error in our two magnitude domains .
table [ ervals ] summarizes our results for the error budgets and the internal error as a function of magnitude .
l c c @xmath9 ( stars per pointing ) & 10 & 20 + @xmath25 ( km s@xmath4 ) & 0.70 & 0.76 + @xmath39 ( km s@xmath4 ) & 0.33 & 0.36 + @xmath32 ( km s@xmath4 ) & 0.26 & 0.33 + @xmath47 ( km s@xmath4 ) & 0.20 & 0.51 + + @xmath26 * ( km s@xmath4 ) * & * 0.33 * & * 0.61 * + thus we find our internal error in region ( 1 ) to be @xmath26 = 0.33 km s@xmath4 and in region ( 2 ) to be @xmath26 = 0.61 km s@xmath4 .
these values match very closely to those derived in our @xmath22 analysis presented in section [ prec ] and figure [ v.ccf.eoveri ] .
in our database , we currently have a total of 8252 rvs of 1046 stars from our ngc 188 cluster sample . of those ,
544 rvs of 77 stars were measured at the dao ( or at the palomar 5 m telescope for those prior to 1980 ) ; the remaining data were measured at the wiyn 3.5 m telescope . before combining these two data sets
, we searched for a potential zero - point offset and for differences in precision . in order to check for a potential zero point offset between the two data sets ,
we compared 24 single stars with @xmath53 observations in each data set .
a star is defined as single if the standard deviation of its measurements is less than four times our measurement precision of 0.4 km s@xmath4 ( see sections [ prec ] and [ var ] ) .
we find a mean offset of 0.03 @xmath6 0.1 km s@xmath4 .
we therefore conclude that there is no measurable systematic rv offset between the wiyn and dao data at this level .
we then investigated the precision of each data set , employing a @xmath22-fitting algorithm as described by @xcite and discussed in section [ prec ] . using this method
, we find the precision of the wiyn data set to be 0.4 km s@xmath4 and of the dao data set to be 1.0 km s@xmath4 .
the dao rv errors found here are larger than the typical errors of 0.4 km s@xmath4 normally seen with this instrument @xcite .
we believe these increased errors are due to photon statistics , as the stars in ngc 188 are faint enough to require long exposure times for this relatively small telescope .
we have therefore noted the number of wiyn and dao observations for each star in our master data table .
we have also accounted for this difference in precision when calculating our average rvs and @xmath29 measurements for stars observed at both observatories ( as explained in sections [ mem ] and section [ var ] ) .
the wiyn and dao rv measurements were thus integrated without modification into a single data set .
we note that the combination of the data sets extends our maximum time baseline up to 35 years for some stars .
the basis for our stellar sample is the p03 pm study of ngc 188 , which we take to be complete within our magnitude limits .
we note that 28 of these 1498 stars lack pm measurements .
as mentioned in section [ sam ] , our original cluster sample was composed of the pm members from the @xcite study , and , as such , our observations are most complete within this sample .
we have @xmath53 observations of 99% of the @xcite sample with pm membership probabilities @xmath28 70% . the dinescu et al .
pm members ( p@xmath48 70% ) comprise 27% of their entire sample .
we have also observed 47% of the remainder of the dinescu et al .
sample @xmath53 times , and 65% at least once .
our current stellar sample consists of 1498 stars , including 624 with non - zero pm membership probabilities , taken from the p03 study .
we have @xmath53 observations of 96% of these non - zero pm members , and at least one observation of @xmath2899% of these members .
we have also observed 27% of the remainder of the platais et al .
sample @xmath53 times , and 46% at least once .
we have at least one observation of 69% of our entire stellar sample . in total , we have calculated rv membership probabilities for 848 stars ( each having @xmath5 3 rv measurements ) .
we have plotted our completeness as a function of radius in the left panel of figure [ compfig ] and as a function of @xmath2 magnitude in the right panel of figure [ compfig ] .
the top plots display our completeness in the stars with non - zero p03 pm membership probabilities .
our observations near completeness in both radius and magnitude for this sample .
there is a slight bias towards the cluster center as a result of having prioritized these stars in our observations .
our completeness drops for magnitudes brighter than v=12 as we have not observed these stars at the wiyn 3.5 m ; the data for these stars comes from the dao .
there are only 6 stars with 10.5@xmath0v@xmath3312 and non - zero p03 pm membership probabilities .
additionally our completeness amongst stars observed @xmath53 times begins to drop at @xmath2 @xmath36 16 .
these stars are more difficult to observe , as they require clear skies and large separations from the moon .
the bottom plots display our completeness for our entire stellar sample .
the drop in completeness towards larger radii is due to a similar decrease in frequency of pm members at increasing radius from the cluster center .
the incompleteness in @xmath2 @xmath28 12 within our full stellar sample is again due to a decreasing frequency of pm members among fainter stars .
we have compiled a table of all observed stars within our survey and show the first ten lines in table [ rvtab ] , while the entire table is provided electronically . for each star
, we have included the wocs ( p03 ) i d ( @xmath49 ) , @xcite i d ( @xmath50 ) , @xcite i d ( @xmath51 ) , and @xcite i d ( @xmath52 ) .
we also include the p03 ra , dec , @xmath2 , and @xmath18 .
next we show the number of wiyn ( @xmath53 ) and dao ( @xmath54 ) observations , average rv ( @xmath55 ) , p03 pm membership ( @xmath56 ) , rv membership ( @xmath57 ) , @xmath29 measurement , class ( described in section [ mem ] ) , and a comment .
we use a weighted average to calculate the mean rv and @xmath29 measurements . in the comment field
, we use g for giant stars , bs for blue stragglers , sb1 and sb2 for single- and double - lined spectroscopic binaries , respectively , and rr for rapid rotators .
we define the rapid rotators as those stars whose ccf fwhm is @xmath28 60 km s@xmath4 and are not observed to be an sb2 binary .
( our mean fwhm for non - rr stars is 35 km s@xmath4 . )
we have also included the relevant ids from the photometric and x - ray studies of the cluster ( with prefix `` v '' from @xcite , `` wv '' from @xcite , `` x '' from @xcite , and `` gx '' from @xcite ) , as well as certain short - period variable classifications from @xcite .
we caution the reader that the variability of cases where , for example , @xmath29 @xmath33 5 and only a few observations are available should be considered uncertain .
l l l l c c c c c c c c c c l l 52 & & & 501 & 00:24:48.705 & 85:34:14.36 & 16.370 & 0.661 & 1 & 0 & 12.33 & 31 & & & u & + 91 & & & 857 & 00:26:49.002 & 85:27:19.11 & 16.120 & 0.789 & 5 & 0 & 23.86 & 57 & - & 6.16 & bln & sb1 + 96 & & 1924 & 1158 & 00:27:33.157 & 85:34:59.40 & 13.908 & 0.706 & 13 & 0 & -29.22 & 0 & - & 39.97 & bu & sb1 + 117 & & 1916 & 1379 & 00:28:39.664 & 85:30:37.77 & 14.078 & 1.117 & 9 & 0 & -50.65 & 0 & 0 & 0.48 & sn & + 125 & & 1938 & 973 & 00:27:09.762 & 85:29:30.67 & 15.598 & 0.787 & 3 & 0 & -84.44 & & 0 & 2.74 & sn & + 136 & & 1691 & 1229 & 00:28:19.530 & 85:28:08.38 & 14.856 & 0.928 & 3 & 0 & -94.02 & 0 & 0 & 2.61 & sn & + 142 & & & 1188 & 00:28:13.319 & 85:27:24.98 & 16.000 & 1.030 & 3 & 0 & -12.46 & 0 & 0 & 3.60 & sn & + 149 & & & 1790 & 00:29:39.550 & 85:40:28.12 & 14.156 & 0.214 & 1 & 0 & 5.11 & & & & u & + 160 & & 1929 & 1916 & 00:30:20.571 & 85:38:31.55 & 13.531 & 0.663 & 3 & 0 & -39.95 & 0 & 21 & 3.50 & sn & + 200 & & & 1964 & 00:30:57.379 & 85:31:14.40 & 16.111 & 0.790 & 7 & 0 & -63.76 & 43 & - & 13.17 & bln & sb1 + in order to calculate a given star s rv membership probability , we first fit separate gaussian functions for both the cluster and field - star populations to a histogram of the rv measurements for all single stars .
we then define the membership probability of a given star by the usual equation : @xmath58 where @xmath59 is the cluster - fit value for rv @xmath60 , and @xmath61 is the field - fit value for the same rv @xcite .
we use only single stars in computing the cluster and field fits ; we then re - normalized the gaussians to the full sample size including velocity - variable stars .
figure [ mems](a ) shows the rv histogram as well as the fits to the field and cluster stars used in equation [ memeq ] .
the cluster is best fit by a gaussian centered on -42.43 km s@xmath4 with @xmath10 = 0.76 km s@xmath4 .
the cluster fit indicates a total 485 stars integrated over the gaussian fit .
the field is best fit by a gaussian centered on -35.51 km s@xmath4 with @xmath10 = 38.01 km s@xmath4 and indicates a total of 314 stars . as mentioned above , monte carlo simulations have indicated that we require three observations covering a baseline of at least one year in order to be 95% confident that a star is either constant or variable in velocity ( i.e. , a single star or a binary- or multiple - star system ) .
if this criterion is not met , we classify the star s membership as unknown [ u ] .
we have calculated rv membership probabilities for the 848 stars that do meet this criterion .
for single stars we use the average rv to perform this calculation . for binary stars with orbit solutions we use the center - of - mass velocity @xcite ; for binary stars without orbit solutions ,
we use the average rv .
for single stars and binaries without orbit solutions that have both wiyn and dao rv measurements , we use a weighted average ( weighted by the precision of each observatory ; see section [ comb ] ) to calculate this average rv .
we then round the resulting rv membership probability to the nearest integer value and quote this as the @xmath57 .
upon review of the histogram of membership probabilities ( figure [ mems](b ) ) , we choose to define our rv members as those with a non - zero @xmath57 .
we note that there are only 31 stars with 0% @xmath62 80% .
ngc 188 is out of the plane , and well separated from the field in rv space . as such ,
in the range of rvs that yield @xmath63 0% we would only expect to find a @xmath36% contamination by field stars .
this contamination will arise from field stars whose rvs are similar to those of the cluster members .
we have expanded this portion of figure [ mems](a ) and plot the region in the right side of the figure .
we can securely compute rv membership probabilities for single stars as well as binaries with orbit solutions .
for all stars , we also use the p03 pm membership probabilities as a further constraint on cluster membership , including only those with non - zero @xmath56 as cluster members .
in addition to our rv membership probabilities , we classify single stars as either single members [ sm ] or single non - members [ sn ] , based on the rv membership probability . binary stars with orbit solutions are classified as either binary members [ bm ] or binary non - members [ bn ] , based on the rv membership probability derived using their center - of - mass velocity @xcite . for velocity - variable stars without orbit solutions ,
our rv membership probabilities are less certain ; therefore we simply provide the classification as a distinction .
velocity - variable stars without orbit solutions are classified as binary likely members [ blm ] if @xmath64 0% ( where @xmath55 is the star s average rv ) , binary unknown [ bu ] if the range in rv measurements includes the cluster mean velocity but @xmath65 0% , or binary likely non - member [ bln ] if all rv measurements fall outside of the cluster distribution .
in addition to our stars the have been observed @xmath33 3 times , we also classify certain stars as unknown [ u ] that do not have sufficient measurements to yield a reliable class ( e.g. , rapid rotators with few observations ) .
note that in table [ rvtab ] , we have only included our @xmath57 measurements for the sm , sn , bm , and bn stars , as the rv memberships derived for the remaining classes of stars are uncertain .
table [ memnums ] displays the number of stars within each classification .
we define our sample of cluster members as the sm , bm and blm stars ; in total , we find 473 likely member stars . including only the sm and bm cluster members , we find a mean cluster rv of -42.36 @xmath6 0.04 km s@xmath4 , with a standard deviation about the mean of 0.64 km s@xmath4 .
we find good agreement between our rv membership probabilities and the p03 pm membership probabilities .
there are 452 stars within our magnitude limits for which p03 calculates a @xmath66 80% .
we have secure membership probabilities for 384 of these stars ( meaning either @xmath53 observations on a single star or an orbit solution for a binary star ) . of these , we find 356 ( 92% ) to have a @xmath67 80% and 365 ( 95% ) to have @xmath68 0% .
of the 19 pm member stars for which we find zero rv membership probability , 5 lie far from a single star isochrone . in total
, we find 358 single stars with @xmath67 80% .
of these stars , 289 ( 81% ) have @xmath66 80% and 340 ( 94% ) have @xmath69 0% .
we also find 80 binary stars with orbit solutions to have @xmath67 80% @xcite . of these , 70 ( 88% )
have a @xmath66 80% , and 77 ( 96% ) have a non - zero @xmath56 .
there are 25 single stars and binary stars with orbit solutions for which we find @xmath68 0% that p03 find zero pm membership probability . of these stars , 4
lie far from a single star isochrone . as stated above
, we expect a @xmath36% contamination of field stars within our @xmath280% rv membership cut .
indeed , out of our 453 single stars and binary stars with orbit solutions for which we find @xmath68 0% , we would expect to find @xmath325 field stars .
in order to reduce this contamination , we use both the rv and pm membership information to identify cluster members .
specifically , we have chosen to define our 3d members as those stars that have a non - zero p03 pm membership probability and non - zero rv membership probability .
these 473 stars are used in the subsequent analyses .
rv variable stars are evident from the standard deviations of their measurements .
as described in section [ prec ] and shown in figure [ 3all.eoveri ] , we have chosen a cutoff value of 4 times our precision ( @xmath29 @xmath28 4 ) to distinguish single stars from velocity - variable stars . for stars with both wiyn and dao measurements ,
we take a weighted average of the two @xmath29 measurements resulting from the different precision values ; where @xmath9 is the number of measurements .
the subscript @xmath70 corresponds to wiyn while @xmath71 corresponds to dao .
each respective ( @xmath29)@xmath72 measurement is calculated using the appropriate precision value . ] .
if a star should have multiple observations from one observatory and only one observation from the other , we exclude that single observation and use the other observations with the appropriate precision value in calculating the @xmath29 measurement .
note , we have not included @xmath29 measurements in table [ rvtab ] for rr stars or sb2 binaries , as the uncertainties on these measurements are not well defined . in total , we find 124 velocity - variable cluster members .
we assume that our velocity variables are binary- or multiple - star systems , and seek to derive orbit solutions for all of our detected binaries . to date , we have determined orbit solutions for 93 binaries , 78 of which are probable members ; 16 are double - lined and 62 are single - lined @xcite .
see table [ memnums ] for the number of stars within all of our velocity - variable classes .
we have also examined known photometric variable stars and x - ray sources for membership and rv variability .
table [ phvar ] presents the previously studied photometric - variable stars ( with prefix `` v '' from @xcite and `` wv '' from @xcite ) that are within our magnitude limits and have @xmath53 observations .
table [ txray ] displays similar information for the x - ray sources ( with prefix `` x '' from @xcite and `` gx '' from @xcite ) . for both tables ,
the column headings describe the same measurements as in table [ rvtab ] , with the exception of the first column that provides the respective photometric variable or x - ray i d .
there are a few photometric variable stars that we have observed , but have not included in table [ phvar ] ; each are w uma stars @xcite .
we have observed star 5337 ( v4 ) 5 times at the wiyn 3.5 m and find an average rv of -45.7 km s@xmath4 with a standard deviation of 1.19 km s@xmath4 .
we do not observe this star to be a rapid rotator ; however the correlation function is unusually shaped requiring more detailed analysis .
we require more observations of this star to derive a reliable membership and classification .
we have also observed stars 5361 ( v3 ) , 4989 ( v5 ) , and 5209 ( v13 ) multiple times , but have been unable to derive reliable rv measurements from their spectra , most likely due to their rapid rotation .
l c sm & 349 + sn & 291 + bm & 78 + bn & 15 + blm & 46 + bu & 28 + bln & 35 + u & 204 + l c c c c l l v8 , v372 cep & 5629 & 0 & 0 & - & bn & sb2 , rr secondary , x - ray source ( x26,gx1 ) , rs cvn + v9 & 4736 & 98 & 97 & 0.59 & sm & g + v11 & 4705 & 98 & 98 & - & bm & sb2 , x - ray source ( gx18 ) , g + v12 & 5762 & 66 & 97 & - & bm & sb2 , ea + wv3 & 5379 & 98 & 98 & 14.96 & bm & sb1 , bs + wv4 & 4304 & 98 & 97 & 2.14 & sm & + wv8 & 4750 & 97 & 96 & 2.91 & sm & + wv12 & 4997 & 98 & 85@xmath73 & 0.66 & sm & + wv19 & 6208 & 0 & 0 & 0.35 & sn & + wv24 & 4656 & 97 & 98 & 0.77 & sm & ea + wv26 & 3953 & - & 97 & 4.79 & blm & sb1 + wv28 & 4508 & - & 98 & 30.30 & bu & sb1 , x - ray source ( gx20 ) + wv33 & 5769 & 0 & 0 & 1.90 & sn & + wv34 & 5767 & 98 & 98 & 0.92 & sm & +
l c c c c l l x1 & 1278 & - & 0 & - & bln & rr + x5 & 1141 & 98 & 88 & 0.59 & sm & g + x11 & 483 & - & 0 & - & bu & sb2 ? + x21 & 5258 & - & 92 & - & u & rr + x26 , gx1 & 5629 & 0 & 0 & - & bn & sb2 , rr secondary ,
rs cvn , photometric variable ( v8 ) + x29 & 5027 & 98 & 96 & 1.51 & sm & g , fk comae - like star ( i-1 ; * ? ? ?
* ) + gx18 & 4705 & 98 & 98 & - & bm & sb2 , g , photometric variable ( v11 ) + gx20 & 4508 & - & 98 & 30.30 & bu & sb1 , photometric variable ( wv28 ) + gx27 & 4230 & - & 93 & - & bu & sb1 , rr + gx28 & 4289 & 98 & 98 & 71.16 & bm & sb1 , g +
without any membership selection , the color - magnitude diagram ( cmd ) of ngc 188 , within our magnitude range and spatial coverage , reveals little more than a poorly defined main sequence .
plot ( a ) of figure [ cmds ] shows the locations on a cmd of each of the 1498 stars within our sample of ngc 188 .
plot ( b ) shows only our rv member stars .
the power of our combined rv and pm kinematic membership selection is clearly shown in plot ( c ) , which displays our cleaned cmd plotting only our 473 3d kinematic member stars ( @xmath69 0% and @xmath68 0% ) , and highlighting our 124 velocity - variable member stars .
the main sequence is clearly visible in plot figure [ cmds](c ) along with the main - sequence turnoff at about @xmath74 14.9 and a well - defined giant branch .
we have defined our bs stars as being bluer than the dashed line drawn in figure [ cmds](c ) and having a non - zero pm and rv membership probability .
these stars are both bluer and brighter than the main sequence and main - sequence turnoff .
we have excluded stars near to the turnoff so as to be sure not to include equal - brightness main - sequence binaries .
the blue straggler stars are noted as bs in the comment field of table [ rvtab ] .
we have identified 21 bs stars , 16 of which are velocity variables with @xmath29@xmath28 4 .
two of our bs stars ( 1366 and 8104 ) are fainter than the traditional bs region and are noted below .
all but two of our bs stars were identified by p03 ; star 1947 has @xmath56 = 2% , and star 1366 has @xmath56 = 3% .
interestingly , we note the very large frequency of detected binaries amongst the bs stars ( 16 out of 21 ) .
we will discuss these binaries in detail in our next paper , @xcite .
here we simply note that this large frequency is secure as we have derived reliable orbit solutions for all but three of our bs velocity variables ; two of these remaining blm stars have preliminary orbit solutions and are certainly binary stars .
we find 66 red giant stars , 0.95 . ]
16 of which are velocity variables .
we also note the large scatter amongst the giant branch stars in our cleaned cmd . indeed ,
the ngc 188 giant branch is much broader and less well defined than that of m67 , a similarly aged old open cluster ( 4.5 gyr ) ( e.g. , * ? ? ?
this scatter in the ngc 188 giant branch was previously noted in the photometric studies by @xcite , @xcite , and @xcite , and is confirmed after our careful membership selection .
the origin is still uncertain . in this section
we provide details on the stars of note that we have boxed in figure [ cmds](c ) .
the reader is reminded that we expect @xmath36% contamination from field stars within our sample of stars with @xmath68 0% .
we have attempted to reduce this contamination by excluding stars with @xmath56 = 0%
. however , there may still be field stars within our cluster member sample .
we note that all of these stars are located in the outer regions of the cluster
. this may be an important clue to their formation mechanisms as well as their subsequent dynamical evolution .
however , it may also increase the likelihood that they are field stars . for
most of these stars , the @xmath56 and @xmath57 provide strong evidence for cluster membership .
[ [ section ] ] 3271 : + + + + + + star 3271 is much bluer than its companions on the giant branch .
indeed , this star is too blue to reside on the horizontal branch or the lower agb ( as compared to a 7 gyr solar - metallicity padova isochrone from @xcite ) .
this star is located at a distance of 15.1 arcmin ( 6.4 core radii ) from the cluster center , and has a @xmath56 = 94% and a @xmath57 = 98% .
we do not detect it as a velocity variable , with an @xmath29=0.54 , and 17 rv measurements .
however , this star may indeed be a long - period or low - amplitude binary that lies outside of our detection limits .
no combination of two stars from the main sequence through the giant branch can account for this star s position on the cmd .
we suggest that this star may be a wide red giant - bs binary system .
[ [ section-1 ] ] 1141 : + + + + + + star 1141 has a @xmath56 = 88% , a @xmath57 = 98% , an @xmath29 measurement of 0.59 , and is located at a radius of 17.7 arcmin ( 7.5 core radii ) . this star is redder than its counterparts on the giant branch , and we are unable to reproduce its color by the addition of an undetected binary companion .
this star is particularly intriguing , as it is also detected as an x - ray source ( x5 ) by @xcite .
[ [ section-2 ] ] 3118 : + + + + + + star 3118 is also redder than its counterparts on the giant branch , and has a high @xmath57 at 88% , but only a marginal @xmath56 at 34% .
the system is located at a radius of 18.76 arcmin ( @xmath38 core radii ) from the cluster center .
we detect this star as a sb2 binary and have derived a preliminary orbit with a period of 12 days and a near circular eccentricity @xcite . [ [ and-8104 ] ] 1366 and 8104 : + + + + + + + + + + + + + + + stars 1366 and 8104 are bluer than the main sequence yet fainter than the more `` traditional '' bs population .
star 1366 has a @xmath56 = 23% , a @xmath57 = 96% and an @xmath29 measurement of 1.67 with 3 observations .
this star is located at a radius of 29.6 arcmin ( 12.6 core radii ) from the cluster center .
star 8104 has a @xmath56 = 91% and an @xmath29 measurement of 8.90 with 7 observations .
this system is located at a radius of 14.1 arcmin ( 6.0 core radii ) from the cluster center .
we require more observations to derive a reliable rv membership probability , but we have classified this star as a blm .
bs formation theories that include main - sequence mergers predict bs stars to populate this region of the cmd in addition to the more traditional brighter region ( e.g. , * ? ? ?
* ; * ? ? ?
these stars may in fact be main - sequence merger products that now appear bluer than their main sequence counterparts .
[ [ section-3 ] ] 3259 : + + + + + + star 3259 is found redward of the main sequence and below the subgiant branch .
this star has a @xmath56 = 71% and an @xmath29 measurement of 17.52 with 6 observations .
we require more observations to derive a reliable rv membership probability , but we have classified this star as a blm .
3259 is located at a radius of 15.6 arcmin ( 6.6 core radii ) from the cluster center .
stars found in this region of the cmd are hard to explain through traditional theories of stellar evolution .
no simple photometric combination of cluster members can account for this star s position on the cmd .
though not displayed on figure [ cmds](c ) , star 4989 ( @xmath2=16.10 , @xmath75=0.97 ) is a 95% pm member and is located at a radius of 1.8 arcmin ( 0.8 core radii ) from the cluster center .
this star is identified as a w uma ( v5 ) by @xcite .
we are unable to derive reliable rvs for this star from our seven observations ( the ccf peak heights fall below our 0.4 cutoff ) .
however , a visual inspection of the spectra suggests that star 4989 is likely a rr
. these stars may be analogous to the sub - subgiants discovered in m67 by @xcite .
the currently accepted model predicts that as two - body interactions lead towards equipartition of kinetic energy , we should , in a relaxed cluster , expect the more massive stars and star systems to have fallen to the center of the cluster potential dominated by the less massive stars .
mass segregation has been observed in relaxed clusters of a variety of ages , from the young open clusters pleiades ( 100 myr ; * ? ? ? * ) and preaesepe ( 800 myr ; * ? ? ?
* ) to the old open cluster m67 ( 4.5 gyr ; * ? ? ?
@xcite derived a relaxation time for ngc 188 of @xmath310@xmath76 years , which suggests that ngc 188 may have lived through at least tens of relaxation times .
thus we would expect ngc 188 to be mass segregated .
indeed @xcite found evidence for mass segregation in their photometric study of ngc 188 .
in addition , a study by @xcite found their ngc 188 photometric variable star sample to be centrally concentrated , affirming the presence of mass segregation if their photometric variables are binary stars .
conversely @xcite did not find their bs population to show significantly more central concentration than the giant stars .
if we assume that bs stars are formed through mergers of two ( or more ) stars ( e.g. , * ? ?
* ) , we would expect the bs stars to be more massive than the giants .
hence this finding suggests a lack of mass segregation amongst the bs population , which would appear to disagree with @xcite and @xcite .
therefore , we reanalyze ngc 188 for the presence of mass segregation in light of the precise membership determination presented in this paper .
we compare the spatial distributions of the giant , velocity variable , and bs stars to that of the single stars ( i.e. , stars with constant velocity ) within our ngc 188 sample , and plot the results in figure [ kscum ] .
we first investigate the velocity variables by comparing their spatial distribution to the single stars . a kolmogorov - smirnov ( k - s ) test shows that the velocity variables are derived from a different parent spatial distribution than the single stars with 92% confidence .
presuming that the velocity variables are binary- or multiple - star systems , then the central concentration of the more massive binary star population offers further evidence for the presence of mass segregation in ngc 188 .
next , we investigate the spatial distribution of the ngc 188 bs population by again comparing to the single star population .
a k - s test shows a low distinction of 63% between the two populations .
we therefore do not find the full population of bs stars to be centrally concentrated with respect to the single stars , agreeing with @xcite .
the majority of our bs stars are found in binary systems ; additionally , as stated above , the formation of bs stars is likely the result of mergers of two ( or more ) stars .
hence their combined mass is likely much greater than that of a single star in our sample . as such ,
this lack of central concentration is surprising and may provide a significant insight into the formation mechanisms and subsequent dynamical evolution of the ngc 188 bs stars .
interestingly , in the similarly aged open cluster m67 ( 4.5 gyr ) , @xcite find secure evidence for mass segregation of the bs population .
perhaps importantly , their sample only extended to @xmath35 core radii , whereas our ngc 188 sample extends to @xmath313 core radii .
if we limit our ngc 188 sample to 5 core radii and again compare the spatial distribution of the bs and single star populations , we find 95% distinction between the two populations .
thus , this restricted population of bs stars does show evidence for central concentration .
we also note the seemingly bimodal spatial distribution amongst the ngc 188 bs stars .
we suggest that we may be observing two populations of bs stars , with a centrally concentrated population , within @xmath35 core radii , and a halo population extending to the edge of the cluster .
similar bimodal bs spatial distributions have been observed in some globular clusters ( e.g. , m55 @xcite , 47 tuc @xcite , ngc 6752 @xcite , and m3 @xcite ) .
these bimodal spatial distribution are likely the result of the formation mechanism(s ) and dynamical evolution of the bs populations .
scattering experiments performed by @xcite suggest that bs stars formed through collisions within a population of primarily short - period binaries may be given enough momentum to achieve long radial orbits , spending the majority of their time in the outer cluster regions where the relaxation time is longer
. thus these halo bs stars may not attain significant central concentration .
this result agrees with @xcite who suggest that the bs stars are likely formed in the core and ejected to varying cluster radii depending on the recoil of the dynamical interaction .
bs stars that are only ejected to a few core radii quickly fall back to the cluster center through dynamical friction and mass segregation processes , while those ejected farther may remain as a halo population .
conversely ferraro et al . ( 1997 , 2004 ) suggest that such a bimodal distribution is the result of two bs formation mechanisms , with the outer bs stars formed through mass transfer in primordial binaries , and the inner bs stars formed from stellar interactions which lead to mergers .
we can use equation [ scm2 ] and our detailed analysis of the precision of our measurements to solve for the combined rv dispersion of ngc 188 : @xmath77 . using the average values derived above , the resulting global value of the combined velocity dispersion is @xmath78 = 0.64 @xmath6 0.04 km s@xmath4 .
this value is likely inflated due to the presence of undetected binaries ; we attempt to evaluate this overestimate below .
mathieu ( 1983 , 1985 ) studied the effect of such undetected binaries in the old open cluster m67 through a monte carlo analysis .
this cluster has a similar observed binary frequency and combined velocity dispersion to ngc 188 .
the monte carlo simulations used an @xcite binary distribution .
the results from that study suggest that for a combined velocity dispersion of 0.5 km s@xmath4 and a total binary fraction of 50% , one would expect the combined velocity dispersion to be inflated by 0.23 km s@xmath4 .
given our observed hard binary ( p@xmath33 10@xmath79 ) fraction of roughly 30% , we expect a contribution from undetected binaries to be similar to the values derived by mathieu ( 1983 , 1985 ) .
given the mean velocity dispersion , we can estimate the virial mass of ngc 188 .
we use the equation given by @xcite , in projection : @xmath80 where @xmath40 is the line - of - sight ( radial ) velocity dispersion , and @xmath81 is the projected half - mass radius .
we use the p03 pm members ( i.e. , stars with @xmath69 0% ) to derive @xmath81 = 5.8 @xmath6 0.3 pc .
we estimate @xmath40 = 0.41 @xmath6 0.04 km s@xmath4 , using the correction value derived by mathieu ( 1983 , 1985 ) to recover @xmath40 from our derived @xmath78 given above , and assuming an error equal to that of @xmath78 .
the resulting virial mass is 2300 @xmath6 460 m@xmath1 .
this value agrees well with the total mass one gets by counting the p03 pm members ; if we assume a mass - to - light relationship of @xmath82 and use the proper bolometric correction factors , we find a total mass of 2350 m@xmath1 .
our mass estimate also lies within the 90% confidence range of 230 to 2600 m@xmath1derived by @xcite with an isotropic king model fit @xcite .
additionally , our mass estimate agrees with the total mass of 3800 @xmath6 1600 m@xmath1derived by @xcite from integrating an extrapolated mass function .
we have also looked for a variation in the combined rv dispersion with radial distance from the cluster center , again using only our sm stars ( @xmath68 0% , @xmath69 0% and @xmath29 @xmath33 4 ) . we have chosen to plot the upper limits on the combined rv dispersion without attempting to correct for undetected binaries , and have binned the data in equal steps in log - radius .
we only use the wiyn rv data for this analysis to avoid unnecessary complications from the difference in precision of the two data sets .
the results are shown in figure [ vdisp ] .
the combined rv dispersion ranges from @xmath30.8 km s@xmath4 in the radial bins closest to the cluster center to @xmath30.6 km s@xmath4 in the bins farthest away from the cluster center .
we note that the central rise in combined rv dispersion is likely inflated due to undetected binaries , as the binaries are centrally concentrated in ngc 188 ( see section [ spat ] ) . indeed , applying the correction described above to the central bins would lower these values to approximately that of our farthest bins , where we would expect the least inflation by undetected binaries . moreover , this correction results in an rv dispersion distribution that is indistinguishable from an isothermal distribution .
@xcite also performed a study of the radial dependence of the combined rv dispersion of ngc 188 , though with a smaller sample size than our study .
they found a higher combined dispersion in the inner cluster region of 1.16 ( with a range of 0.88 - 1.81 ) km s@xmath4 , but the values in their remaining radial bins , of 0.51 ( 0.22 - 1.06 ) km s@xmath4 at 5.6 - 13 arcmin and 0.20 ( 0.12 - 0.40 ) km s@xmath4 at 13 - 20 arcmin , are similar to the values we have obtained here .
undoubtedly , the combined dispersion measured by harris & mcclure in the inner bin was inflated due to undetected binaries , as that paper also concluded based on monte carlo models of observations of binaries .
we have presented a detailed discussion of our rv study of the rich old open cluster ngc 188 . as the first paper in a series of wiyn open cluster study papers presenting precise stellar radial - velocity surveys , we have provided an in - depth description of our data reduction process and paid particular care in establishing our measurement precision of 0.4 km s@xmath4 . for the first time , full 3d kinematic membership probabilities
have been used to study the cluster .
we have thus been highly successful in eliminating field star contamination within our sample , finding 473 probable member stars .
importantly , 124 of these are velocity - variable stars , providing an extensive population of dynamically hard binaries to be studied .
we have used our secure member stars to produce a cleaned cmd that displays our population of bs stars , 76% of which are binaries .
we also note the large scatter within the giant branch .
in addition , we have discussed six stars of note that lie far from a standard single - star isochrone .
stars 3271 , 1141 and 3118 are located near the giant branch .
stars 1366 and 8104 are bs stars that lie below the traditional brighter blue straggler region .
star 3259 may be analogous to the sub - subgiants of m67 @xcite .
we have also investigated the spatial distribution of the stellar populations within the cluster .
we find the binary stars in our sample to be central concentrated with 92% confidence , providing further evidence for the presence of mass segregation within ngc 188 .
the bs spatial distribution appears to be bimodal , and we find no statistical evidence that the full bs population is centrally concentrated .
this finding is surprising and may suggest that we are observing two populations of bs stars in ngc 188 , including a centrally concentrated distribution as well as a halo bs population .
further study is necessary to determine whether this bimodal distribution is a sign of two distinct formation mechanisms , or a result of dynamical scattering during the bs formation . our comprehensive error analysis has allowed us to study the rv dispersion of the cluster , and we find an upper limit on our mean combined rv dispersion of 0.64 @xmath6 0.04 km s@xmath4 .
the results from mathieu ( 1983 , 1985 ) suggest that this velocity dispersion may be inflated by as much as 0.23 km s@xmath4 from undetected binaries .
we have used this corrected velocity dispersion to estimate a mass of 2300 @xmath6 460 m@xmath1 in ngc 188 , which agrees well with the value derived by counting the mass in p03 pm members , as well as other ngc 188 mass estimates from the literature .
the combined rv dispersion decreases with increasing cluster radius ; however the dispersion values in the central bins are likely inflated due to undetected binaries . correcting the central bins for undetected binaries
yields a rv dispersion distribution that is indistinguishable from an isothermal distribution .
the wiyn open cluster study will continue its survey of ngc 188 in order to provide membership data on all stars in our complete sample .
we are currently in the process of analyzing our binary population for orbital solutions , and seek to identify all rv variables in the cluster within our observational constraints . with these orbit solutions , we will study the binary distribution in period , eccentricity , and mass ratio , and our identified rv variables will allow us to constrain the cluster binary fraction . as one of the oldest and most widely studied open clusters , ngc 188 holds the key to understanding the long term evolution of binary systems , merger products , blue stragglers , and stellar dynamics in clusters with rich binary populations .
the authors would like to express their gratitude to the staff of the wiyn observatory without whom we would not have been able to acquire these thousands of superb stellar spectra .
many thanks to sydney barnes and sren meibom who both played a significant part in the development of our error budget analysis in section [ erbud ] .
we also want to thank the many undergraduate and graduate students who have obtained spectra over the years at wiyn for this project .
we would also like to acknowledge r. f. griffin and j. e. gunn for contributing their ngc 188 rvs to our project . in turn
, they wish to express their thanks to the palomar observatory for the use of the 5 m telescope .
thanks to murray fletcher for his expertise in developing the dao rvs instrument .
we also thank jim hesser who acquired a portion of the dao ngc 188 data .
finally , we would like to express our gratitude to the referee ted von hippel for his many insightful comments .
this work was funded by the national science foundation grant ast-0406615 and the wisconsin space grant consortium .
abt , h. a. & levy , s. g. 1976 , apjs , 30 , 273 bailyn , c. d. 1995 , , 33 , 133 belloni , t. , verbunt , f. , & mathieu , r. d. 1998 , , 339 , 431 bonatto , c. , bica , e. & santos jr . , j. f. c. 2005 , 433 , 917 dinescu , d. i. , girard , t. m. , van altena , w. f. , yang , t .-
, lee , y .- w . 1996 , , 111 , 1205 efremov , y. n. , kholopov , p. n. , kukarkin , b. v. , & sharov , a. s. 1964 , iau information bull .
stars , no .
76 eggen , o. j. , & sandage , a. 1969 , , 158 , 669 ferraro , f. r. , paltrinieri , b. , fusi pecci , f. , cacciari , c. , dorman , b. , rood , r. t. , buonanno , r. , corsi , c. e. , burgarella , d. , & laget , m. 1997 , , 324 , 915 ferraro , f. r. , beccari , g. , rood , r. t. , bellazzini , m. , sills , a. , & sabbi , e. 2004 , , 603 , 127 fletcher , j. m. , harris , h. c. , mcclure , r. d. , & scarfe , c. d. 1982 , pasp , 94 , 1017 friel , e. d. , janes , k. a. , tavarez , m. , scott , j. , katzanis , r. , lotz , j. , hong , l. , & miller , n. 2002 , , 123 , 2693 geller , a. m. & mathieu , r. d. , 2012 , aj , 144 , 54 geller , a. m. , mathieu , r. d. , harris , h. c. & mcclure r. d. , 2009 , aj , 137 , 3743 girardi , l. , bressan , a. , bertelli , g , & chiosi , c. 2000 , 141 , 371 gondoin , p. 2005
, , 438 , 291 griffin , r. f. & gunn , j. e. 1974 , , 191 , 545 harris , h. c. , & mcclure , r. d. 1985 , in stellar radial velocities , ed .
a. g. davis philip , & d. w. latham , l. davis press , 257 harris , h. c. , & mcclure , r. d. 1985 , 97 , 261 hobbs , l. m. , thornburn , j. a. , & rodriguez - bell , t. 1990 , , 100 , 710h hurley , j. r. , pols , o. r. , aarseth , s. j. , & tout , c. a. 2005 , mnras 363 , 293 janes , k. a. & smith , g. h. 1984 , aj , 89 , 487 kafka , s. , & honeycutt , r. k. 2003 , , 126 276 kaluzny , j. , & shara , m. m. 1987 , , 314 , 585 keeping , e. s. 1962 , introduction to statistical inference , princeton : van nostrand kim , y. c. , demarque , p. , yi , s. k. , & alexander , d. r. 2002 , , 143 , 499 king , i. r. 1966 , aj , 71 , 64 kirillova , t. s. , & pavlovskaya , e. d. 1963 , sva , 7 , 99 lanzoni , b. , dalessandro , e. , perina , s. , ferraro , f. r. , rood , r. t. , & sollima , a 2007 , , 670 , 1065 leonard , p. j. t. , & linnell , a. p. 1992
, aj , 103 , 1928 maveroni , c. , & van t veer , f. 1991 , , 248 , 430 mathieu , r. d. 1983 , ph.d .
thesis california univ . ,
berkeley mathieu , r. d. 1985 , in iau coll .
88 , stellar radial velocities , a. g. d. philip mathieu , r. d. & latham , d. w. 1986 aj , 92 , 1364 mathieu , r. d. 2000 , asp , 198 , 517 mathieu , r. d. , van den berg , m. , torres , g. , latham , d. , verbunt , v. , & stassun , k. 2003 , , 125 , 246 mathieu , r. d. , meibom , s. , & dolan , c. j. 2004 , , 602 , l121 mcclure , r. d. 1974 , , 194 , 355 mcclure , r. d. , & twarog , b. a. 1977 , , 214 , 111 m mcclure , r. d. , flethcer , j. m. , gundmann , w. a. , & richardson , e. h. 1985 , in iau coll .
88 , stellar radial velocities , a. g. d. philip meibom , s. , barnes , s. a. , dolan , c. , & mathieu , r. d. 2001 , aspc , 243 , 711 michaud , g. , richard , o. , richer , j. , & vandenberg , d. a. 2004 , , 606 , 452 platais , i. , kozhurins - platais , v. , mathieu , r. d. , girard , t. m. , & van altena , w. f. 2003 , , 126 , 2992 ( p03 ) randich , s. , sestito , p. , & pallavicini , r. 2003 , , 399 , 122 raboud , d. & mermilliod , j. c. 1998 , , 329 , 101 raboud , d. & mermilliod , j. c. 1998 , , 333 , 897 sabbi , e. , ferraro , f. r. , sills , a. , & rood , r. t. 2004 , 617 , 1296 sandage , a. 1962 , , 135 , 333 sandage , a. , lubin , l. m. , & vandenberg , d. a. 2003 pasp,115,1187 sarajedini , a. , von hippel , t. , kozhurina - platais , v. , & demarque , p. 1999
, , 118 , 2894 sigurdsson , s. , davies , m. b. , & bolte , m. 1994 , apj , 431 , l115 spitzer , l. 1987 , dynamical evolution of globular clusters ( princeton university press ) stetson , p. b. , mcclure , r. d. , & vandenberg , d. a. 2004 , , 116 , 1012 stryker , l. l. 1993 , , 105 , 1081 tody , d. 1993 , `` iraf in the nineties '' in astronomical data analysis software and systems ii , a.s.p .
conference ser . , vol 52 , eds .
hanisch , r.j.v .
brissenden , & j. barnes , 173 .
tonry , j. & davis , m. 1979 , aj , 84 , 1511 twarog , b. a. 1978 , , 220 , 890 twarog , b. a. , ashman , k. m. , & anthony - twarog , b. j. 1997 , aj , 114 , 2556 van t veer , f. 1998 , , 139 , 477 vasilevskis , s. , kelmola , a. , & preston , g. 1958 , aj , 63 , 387 von hippel , t. , & sarajedini , a. 1998 , , 116 , 1789 worthy , g. & jowett , k. j. 2003 , pasp , 115 , 96 zhang , x.b . , deng , l .. tian . , b. , & zhou , x. 2002 , aj , 123 , 1548 zhao , j. l. , tian , k. p. & su , c. g. 1996 , ap&ss , 235 , 93 | we present the results of our ongoing radial - velocity survey of the old ( 7 gyr ) open cluster ngc 188 .
our wiyn 3.5 m data set spans a time baseline of eleven years , a magnitude range of 12@[email protected] ( 1.18 - 0.94 m@xmath1 ) , and a 1 diameter region on the sky . with the addition of a dominion astrophysical observatory ( dao )
data set we extend our bright limit to @xmath2 = 10.8 and , for some stars , extend our time baseline to thirty - five years .
our magnitude limits include solar - mass main - sequence stars , subgiants , giants , and blue stragglers , and our spatial coverage extends radially to 17 pc ( @xmath313 core radii ) .
for the wiyn data we present a detailed description of our data reduction process and a thorough analysis of our measurement precision of 0.4 km s@xmath4 for narrow - lined stars .
we have measured radial velocities for 1046 stars in the direction of ngc 188 , and have calculated radial - velocity membership probabilities for stars with @xmath53 measurements , finding 473 to be likely cluster members .
we detect 124 velocity - variable cluster members , all of which are likely to be dynamically hard binary stars . using our single member stars
, we find an average cluster radial velocity of -42.36 @xmath6 0.04 km s@xmath4 .
we use our precise radial - velocity and proper - motion membership data to greatly reduce field star contamination in our cleaned color - magnitude diagram , from which we identify six stars of note that lie far from a standard single - star isochrone .
we present a detailed study of the spatial distribution of cluster member populations , and find the binaries to be centrally concentrated , providing evidence for the presence of mass segregation in ngc 188 .
we observe the blue stragglers to populate a bimodal spatial distribution that is not centrally concentrated , suggesting that we may be observing two populations of blue stragglers in ngc 188 , including a centrally concentrated distribution as well as a halo population .
finally , we find ngc 188 to have a global radial - velocity dispersion of 0.64 @xmath6 0.04 km s@xmath4 , which may be inflated by up to 0.23 km s@xmath4 from unresolved binaries . when corrected for unresolved binaries , the ngc 188 radial - velocity dispersion has a nearly isothermal radial distribution .
we use this mean corrected velocity dispersion to derive a virial mass of 2300 @xmath6 460 m@xmath1 . |
we are already at a point where we have excellent constraints on the spatial distribution and properties of galaxies in the local universe from the largest surveys ever undertaken , namely the 2-degree field galaxy redshift survey ( 2dfgrs ; * ? ? ?
* ) , and the sloan digital sky survey ( sdss ; * ? ? ?
* ) in the optical , the two - micron all sky survey ( 2mass ; * ? ? ?
* ) in the near - infrared , and the iras redshift survey @xcite selected in the far - infrared .
over the past few years these surveys to study the formation and evolution of galaxies in the universe , and how they trace the large - scale structure of the universe , have been complemented by deeper and narrower surveys at all wavelengths , carefully balancing depth and area to ensure that representative volumes of the universe are observed .
most notably the galaxy and mass assembly ( gama ; * ? ? ?
* ) survey covering @xmath17 square degrees is building on the legacy of the sdss and 2dfgrs spectroscopic surveys to obtain accurate redshift information down to an @xmath18band limit of @xmath19 ( corresponding to @xmath20 @xmath21 ) , with the ukirt infrared deep sky survey large - area survey ( ukidss - las ; * ? ? ? * ) providing important near - infrared data . at the other extreme , the deepest ground - based surveys such as cosmos @xcite and the ukidss ultra deep survey coupled with the subaru - xmm deep survey ( e.g. * ? ? ? * ; * ? ? ? * ) , cover @xmath22 degree@xmath1 fields , relying on photometric redshifts to trace the evolutionary history of galaxies ( e.g. * ? ? ? * ) , coupled with extremely time intensive spectroscopic follow - up(e.g . * ? ? ? * ; * ? ? ? * ) . the visible and infrared survey telescope for astronomy ( vista ; * ? ? ?
* ) is now taking the surveys carried out within ukidss one step further , with its larger field of view and improved sensitivity at the bluer ( @xmath23 ) wavelengths of the near - infrared window .
vista is undertaking several near - infrared public surveys , one of which is the vista deep extragalactic observations ( video ) survey .
the aim of the video survey is to gain a representative view of the universe at @xmath24 , allowing galaxy evolution to be traced over the majority of the history of the universe , from the richest clusters to the field .
the video survey fields are chosen to incorporate current and future multi - wavelength data sets to facilitate the broadest exploitation of the new near - infrared survey data both within eso and across the globe .
the main scientific aims of video are described below . throughout this paper
we use the ab - magnitude system @xcite and adopt the concordance cosmology @xmath25 , @xmath26 and @xmath27kms@xmath28mpc@xmath28 .
how and when were massive galaxies formed ? when did they assemble the bulk of their stellar mass and how ? where does this mass assembly occur ?
these are crucial questions which video is designed to answer .
the advent of deep multi - wavelength surveys has led to huge progress in this field .
there are two ways to measure the build - up of stellar mass : by directly observing star formation , or by measuring the total stellar mass already formed by a given epoch .
the first technique was revolutionised by the lyman break technique ( e.g. * ? ? ?
* ; * ? ? ?
* ) which found large numbers of star - forming galaxies at @xmath29 and galex has stretched this to lower redshifts ( e.g. * ? ? ?
however , these techniques are only sensitive to relatively unobscured systems , and are known to miss much of the luminosity density arising from galaxies at @xmath30(e.g . * ? ? ?
* ; * ? ? ?
herschel space observatory _ is now opening up our view of the universe with both wide- ( e.g. * ? ? ? * ) and deep - field ( e.g. * ? ? ?
* ) surveys , which together probe sufficient volume over a large fraction of cosmic time to determine the full spectral energy distribution of galaxies ( e.g. * ? ? ?
* ) and also the evolution of dust and obscured star formation in the universe ( e.g. * ? ? ? * ; * ? ? ? * ; * ? ? ?
on the other hand , mass estimates of distant galaxies require rest - frame near - infrared wavelength observations to study the parts of galaxy spectral energy distributions that are not dominated by young stars with low mass - to - light ratios .
surveys with ground - based near - infrared observations have allowed substantial recent progress in this area ( e.g. * ? ? ?
? * ; * ? ? ?
* ) , particularly at the bright end of the luminosity function at @xmath31 where both area and moderately deep data are required .
surveys such as the great observatories origins deep survey ( goods ; * ? ? ?
* ) and the cosmic assembly near - infrared deep extragalactic legacy survey ( candels ; * ? ? ?
* ; * ? ? ?
* ) probe to much greater depths but only cover very small sky areas , resulting in studies concentrating on the faint galaxy populations ( e.g. * ? ? ?
* ; * ? ? ?
* ) and galaxies at very high redshift ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
these surveys have led the way in probing galaxy evolution from the earliest times up until the present day , but they are fundamentally limited by the fact that they can not probe scales larger than a few mpc , severely limiting investigations of the environmental dependence of galaxy formation and evolution . the goal of video is to perform an in - depth study of the universe over the redshift range @xmath16 , linking the shallower surveys such as the ukidss - las @xcite , the vista hemisphere survey ( vhs ) and the vista kilo - degree survey ( viking ) with the deeper ultravista survey @xcite .
the depths of the video survey have been chosen to detect galaxies at the knee in the galaxy luminosity function , @xmath32 at @xmath33 , and down to @xmath34 at @xmath35 , thereby enabling us to detect the bulk of the luminosity density arising from galaxies over 90 per cent of the history of the universe and the most massive galaxies at the highest redshifts .
thus we will be able to investigate in exquisite detail which galaxies are in place first , and address the issue of downsizing in the mass function of forming galaxies , where the massive early type galaxies appear to be in place before the less massive galaxies ( e.g. * ? ? ? * ; * ? ? ? * ) .
it is also important to note that the epoch over which video is aimed is a crucial one in the history of the universe , as this is when the bulk of the star formation ( e.g. * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ) and accretion ( e.g. * ? ? ?
* ; * ? ? ?
* ) activity took place .
this is thus the ideal survey with which to investigate the effects that star formation and accretion activity have on galaxy evolution in general .
moreover , the intrinsic rarity of the most luminous agn , starburst and elliptical galaxies means that it is important to survey a large enough area from which the luminosity function and clustering of particular galaxy populations can be constrained , thereby enabling the evolution of bias to be determined , improving on previous area - limited surveys that have measured the clustering of various galaxy populations to only @xmath36degree scales ( e.g. * ? ? ?
* ; * ? ? ?
crucially , video will have sufficient area to carry out these investigations as a function of both redshift and environment .
video will not only be able to detect the galaxies which contribute the bulk of the luminosity density at these redshifts , but the five near - infrared filters , along with ancillary data from optical surveys and the _ spitzer space telescope _
, will provide photometric redshifts for galaxies as faint as @xmath32 all the way out to @xmath33 , and to @xmath37 for the most massive galaxies .
in addition to probing the sub-@xmath32 population up to @xmath38 , video will be uniquely placed to quantify the density and spatial clustering of the most massive galaxies at the earliest epochs ( e.g. * ? ? ?
* ; * ? ? ?
initial work on this topic by @xcite used the combined data from the canada - france - hawaii telescope legacy survey ( cfhtls - d1 ; * ? ? ?
* ) , along with the wircam infrared deep survey ( wirds ; * ? ? ?
* ) , ultravista@xcite and video to show that the bright end of the galaxy luminosity function at @xmath39 follows an exponential decline and that these massive galaxies appear to be both larger and more dusty than their fainter counterparts discovered in deep hst imaging ( e.g. * ? ? ?
* ) . with the full video survey area we will be able carry out the first statistically - significant clustering analysis of massive galaxies towards the epoch of reionization , providing a direct link between the underlying dark matter distribution and galaxy populations and how this evolves up to the highest redshifts .
we note that the video survey will detect a factor of @xmath40 more @xmath41m@xmath42 galaxies than ultravista at @xmath43 , as the limiting factor is area rather than depth .
however , we emphasise that the combination of video and ultravista will provide the premier data set for investigating the clustering properties of the highest redshift ( @xmath44 ) galaxies as a function of mass .
video provides data over the area and depth with which to study the evolution of galaxy clusters from their formation epoch to the present day .
galaxy clusters are essential tracers of cosmic evolution in the universe for two important reasons .
first , clusters are the largest virialized objects whose masses we can measure .
consequently , comparisons of the present - day cluster mass distribution with the distributions at earlier epochs can be used to determine the rate of structure formation , placing constraints on cosmological models ( see e.g. * ? ? ?
* for a review ) .
second , the deep potential wells of clusters also mean that they act as closed astrophysical laboratories that retain their gaseous matter .
therefore clusters possess a wealth of information about the processes associated with galaxy formation such as the efficiency with which baryons are converted into stars and the effects which feedback processes have on galaxy formation . between @xmath45 , clusters of galaxies appear to undergo little in the way of strong evolution , either in their gas phase properties or in the properties of the more massive galaxies within them ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the epoch at @xmath46 is therefore a crucial one for studying their evolution , which must be dramatic in the 4 gyr between @xmath47 ( e.g. * ? ? ?
* ; * ? ? ?
the depth of video is such that it can trace the bright end of the cluster luminosity function to @xmath48 and look in detail at the less luminous cluster galaxies to @xmath49 , while its area should provide a sample of several tens of clusters with masses above @xmath50 m@xmath42 at @xmath51
. once found , such galaxy clusters can be studied in great detail in terms of their galaxy populations ( e.g. * ? ? ?
* ; * ? ? ?
* ) , and agn content ( e.g. * ? ? ?
* ; * ? ? ?
* ) , to trace how galaxies form and evolve in _ closed _ and dense environments
. furthermore , the shear number of clusters will allow the average properties of clusters to be studied in detail as a function of cosmic epoch @xcite .
the sdss has revolutionised studies of quasi - stellar objects ( qsos ) at the highest redshifts ( e.g. * ? ? ?
* ) , providing the first evidence that the epoch of reionization was coming to an end around @xmath52 @xcite .
deeper optical surveys have since been pushing further down the qso luminosity function ( e.g. * ? ? ?
* ) , but pushing to higher redshifts is impossible with optical surveys , regardless of depth , due to the fact that the gunn - peterson trough occupies all optical bands at @xmath53 .
this work is being driven to higher redshift with near - infrared surveys such as the ukidss large area survey , which recently discovered a qso at @xmath54 @xcite and the next step for this work will be with the wide - area vista public surveys such as viking and the vista hemisphere survey , which will probe the bright end of the qso luminosity function at @xmath55 ( e.g. * ? ? ?
however , the shape of the qso luminosity function at these redshifts can only be studied with much deeper near - infrared imaging over a significant survey area .
this is the only direct way to determine the contribution of accreting black holes to the reionization of the universe and constrain the density of black - holes within the first gyr after the big bang , as just using the brightest quasars requires a huge extrapolation to fainter magnitudes . using the current best constraints on the qso luminosity function
@xcite we expect 1 - 2 qsos at @xmath56 in the full video survey ; we will therefore be able to provide constraints on the faint end of the qso luminosity function at @xmath57 , a factor of 10 improvement in the number density of qsos over narrower and deeper surveys such as ultravista and the ukidss - uds . with the video survey data we can also find lower - redshift qsos which are faint and/or reddened .
however , interesting constraints could also be made on the obscured agn population when the video survey data are combined with _ spitzer _
( e.g. swire and servs ; * ? ? ?
* ; * ? ? ?
* ) , x - ray ( e.g. xmm - lss ; * ? ? ?
* ) and _ herschel _ ( e.g. * ? ? ?
* ) observations , utilising near- and mid - infrared selection of agn ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the depth of video is crucial in determining the nature of distant obscured agn .
although the central quasar is obscured in these objects , video will have the sensitivity to detect the host galaxies of such sources up to @xmath58 assuming that their host galaxies are @xmath59 ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the 12 degree@xmath60 covered by video will ensure that the rarest and most luminous agn are found and investigated in terms of their host galaxy properties and comoving space density .
for example , adopting the study of heavily obscured agn from @xcite we expect to find @xmath61 heavily obscured qsos at @xmath62 in the video survey .
this would then allow the study of the range of host galaxy masses and the environmental density of these objects , something that would be severely restricted with cosmos and ultravista , and almost impossible with candels , due to the rarity of these luminous agn .
therefore , video is uniquely placed to probe the cosmic evolution of all agn , both obscured and unobscured and address the question of downsizing in the agn population ( e.g. * ? ? ?
there are several outstanding challenges to progress in type ia supernova
( sn ia ) cosmology @xcite .
the most important include : i ) improved precision in the photometric calibration , ii ) understanding sn ia physics , particularly the sn `` colour laws '' and the role of the progenitor system , iii ) constructing comprehensive low - redshift surveys to anchor cosmological analyses , and iv ) building the first homogeneous samples of sn ia at @xmath51 .
near - infrared observations are an essential tool to exploring these trends further ( e.g. * ? ? ?
as well as verifying the extinction corrections made in the optical , theoretical sn ia models suggest that the intrinsic peak brightness dispersion could be substantially smaller at longer wavelengths @xcite .
observational studies using @xmath4 , @xmath5 photometry at @xmath63 show some evidence for this , indicating that sne ia are equally good standard candles at these wavelengths ( peak dispersion @xmath64mag ) and , have no apparent correlation with optical light curve post - peak decline rate @xcite .
indeed , the combination of optical and near - ir data could provide the strongest constraints on the optical extinction , enabling a substantial improvement in sn ia extinction corrections at all wavelengths .
further , as outlined by @xcite , near - ir light curves offer insight into several key aspects of sn ia physics .
the distinctive light curve morphology is sensitive to both the amount and radial distribution of iron group elements ( radioactive and stable ) synthesised in the explosion .
these in turn reflect the nature of the explosion process and the progenitor star system , including the white dwarf metallicity .
studies at near - ir wavelengths thus offer an exciting opportunity to further understand and improve upon our dust corrections of sne ia and the near - infrared light curves .
the video fields are key fields within the dark energy survey ( des ) supernova survey @xcite and will be able to provide near - infrared data points on the light curves of supernovae discovered in the des supernova survey . furthermore , the spectral energy distribution fitting and photometric redshift accuracy of the supernovae host galaxies , which now play a key role in the cosmological analyses @xcite , will improve substantially when the optical des data are used in conjunction with the deep video near - infrared data , and aid in the determination of host galaxy stellar masses @xcite .
the video survey is being carried out over three of the most widely observed high - galactic - latitude fields , these are two vista tiles ( @xmath65 degree@xmath60 ) in the elais - s1 field , three vista ( @xmath66 degree@xmath60 ) tiles in the xmm - newton large scale structure field , and another three tiles in the extended chandra deep field south , covering a total area of @xmath67 degree@xmath60 .
these fields have a wealth of data from the x - rays through to the radio waveband , and are , along with cosmos / ultravista , the primary fields for observations with future facilities in the southern hemisphere .
figure [ fig : fields ] shows the footprint of the video observations in each of these three fields .
one can see that the individual tiles are distributed in different ways .
the design of these positions was driven by the need to cover the wealth of ancillary data available in these fields , including _ spitzer _ @xcite and _ herschel _ @xcite , in addition to future ground - based optical data from both the vlt survey telescope and the dark energy survey .
further in the future these areas will also be targeted with the deepest radio observations with the square kilometre array precursor telescope meerkat @xcite as part of the deep continuum survey @xcite . ) , xmm - lss ( 4.5 degree@xmath60 ) and the extended chandra deep field south ( 4.5 degree@xmath60 ) . ,
title="fig:",scaledwidth=45.0% ] ) , xmm - lss ( 4.5 degree@xmath60 ) and the extended chandra deep field south ( 4.5 degree@xmath60 ) . ,
title="fig:",scaledwidth=45.0% ] ) , xmm - lss ( 4.5 degree@xmath60 ) and the extended chandra deep field south ( 4.5 degree@xmath60 ) . ,
our aim is to be able to detect a galaxy at or below the break of the galaxy luminosity function over 90 per cent of the age of the universe , which corresponds to @xmath68 .
this ensures that we are sensitive to the bulk of the energy density from galaxies at one of the most important epochs in the universe , where star formation and accretion activity were at a maximum .
we use the @xmath69band luminosity function from cirasuolo et al .
al . 2010 ) where @xmath70 at @xmath71 . assuming a passively evolving stellar population with a high formation redshift
, then we expect a @xmath72 galaxy to have a total magnitude of @xmath73 .
given the seeing constraints ( see section [ sec : observations ] ) and accounting for the apparent size evolution of massive passive galaxies at high redshift ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) we therefore define our 5@xmath74 @xmath6-band limit as @xmath75 in a 2 arcsec diameter aperture .
the survey limits in the shorter near - infrared filters ( @xmath76 ) are then determined assuming an elliptical galaxy template at @xmath33 .
the @xmath77band observations are slightly shallower , and are fixed to that of a flat spectral energy distribution from the @xmath78band to help improve photometric redshift accuracy at redshifts where the 4000 balmer and the lyman-@xmath79 breaks are redshifted between the @xmath80 and @xmath78band filters . [ cols="^,^,^,^,^,^,^,^,^ " , ] photometric redshifts are known to produce ` catastrophically ' poor redshift estimates for some fraction of the sample .
the ` catastrophic ' failure label refers to sources whose predicted redshifts differ from their spectroscopic redshifts by a margin considerably greater than the expected / predicted uncertainties .
the most direct method of determining the accuracy of a photometric redshift procedure is by comparison with available spectroscopic data . in fig .
[ fig : photoz ] we present a comparison for galaxies with counterparts in the vvds @xcite using photometric redshifts determined from the cfhtls visible - wavelength ( @xmath81 ) data alone and with the addition of the @xmath82 data from the video survey . ) . _
( lower panel ) _ the same galaxies but where the photometric redshifts have been determined using the combined cfhtls and @xmath82 video filters .
the stripe of objects along the horizontal at @xmath83 is due to the photometric redshifts fitting to their lowest possible value .
the inset shows the histogram of @xmath84.,title="fig:",scaledwidth=48.0% ] ) .
_ ( lower panel ) _ the same galaxies but where the photometric redshifts have been determined using the combined cfhtls and @xmath82 video filters .
the stripe of objects along the horizontal at @xmath83 is due to the photometric redshifts fitting to their lowest possible value .
the inset shows the histogram of @xmath84.,title="fig:",scaledwidth=48.0% ] versus @xmath85 colour - colour diagram showing the colours of the the vvds spectroscopic sample used in our analysis the photometric - redshift outliers , defined as @xmath86 are denoted by the filled red circles and the galaxies with good photometric redshifts , defined as @xmath87 , are shown as black dots .
this shows that the bulk of the outliers have blue colours compared to the remainder of the galaxy population within the spectroscopic sample .
, scaledwidth=48.0% ] counterparts were identified by cross - matching the video and vvds source positions to within 1 arcsec tolerance .
the vvds catalogue provides quality flags indicating confidence in the reported redshift .
quality flags 2 , 3 and 4 indicate confidence levels of 75 , 95 and 100 per cent respectively whilst objects with a flag of 1 have only indicative redshifts based on the continuum with very few supporting spectral features ( @xmath88 50% confidence ) .
the least secure redshifts are those with quality flag 9 which are based on only a single emission line and assigned a tentative redshift . to evaluate the accuracy of the video photometric redshifts we use only those sources in the vvds sample with quality flags of 4 .
we also remove objects classified as stars by the colour selection criteria described in section [ sec : numcounts ] and sources classified as quasars in the vvds spectroscopy .
following @xcite , we determine the normalized median absolute deviation ( nmad ; * ? ? ?
* ) in @xmath89 , where @xmath90 with @xmath91 and @xmath92 denoting the spectroscopic and photometric redshifts respectively . for the vvds sample
we find the nmad in @xmath89 to be @xmath93 and @xmath94 for the cfhtls and cfhtls+video data sets respectively .
however , the fraction of outliers , defined as sources with @xmath86 reduced from 7.4 per cent for the cfhtls filters alone to 3.8 per cent when the video filters are added , if no prior on the luminosity function is used ( c.f . * ? ? ?
if we use the @xmath15band luminosity function prior , as detailed in @xcite , we find that the fraction of outliers with just the cfht data is significantly reduced to 5.4 per cent , whereas implementing the prior with the video data included reduces the fraction of outliers to 3.3 per cent , although a reduced outlier rate of 2 per cent can be achieved by adopting a liberal @xmath95 cut . therefore , even at @xmath96 where the majority of the spectroscopically determined redshifts lie , the video data reduce the number of catastrophic outliers significantly .
we are able to investigate whether the outliers occupy certain areas of colour - colour space in which the photometric redshifts do not perform very well . fig .
[ fig : outliers ] shows the @xmath97 versus @xmath98 colour - colour diagram for the vvds sample with good redshift quality .
this shows that the galaxies with poor photometric redshifts are generally blue compared to the larger sample with good photometric redshifts .
this is unsurprising as blue galaxies do not exhibit strong spectral breaks at low redshift and are therefore difficult to accurately fit photometric redshifts . inspecting the individual spectra
, we also find a significant fraction with very bright emission lines and although emission lines are incorporated into the photometric redshift templates , they are unable to accurately reproduce all emission line ratios found in galaxies .
indeed , the emission line ratios in some of the galaxies appear to indicate the presence of an agn .
to investigate this further , we restrict the sample to only those galaxies with a best fitting galaxy template , rather than include galaxies which are better fit by an agn template .
we find that the percentage of photometric - redshift outliers in this case to decrease further , to 1.2 and 0.3 per cent for the cfht and cfht+video photometry , respectively .
as noted above , there is a dearth of spectroscopic redshifts in the @xmath16 redshift range which is the niche redshift range for the video survey .
we can expect that the photometric accuracy for galaxies at these redshifts will increase substantially with the near - infrared data provided by the video survey , as the balmer and 4000 breaks are redshifted into the near - infrared filters ( e.g. * ? ? ?
* ; * ? ? ?
* ) . in fig .
[ fig : dndz ] we show the redshift distribution for all sources detected in the video to a limit of @xmath99 .
it is immediately apparent that video provides an excellent data set to study galaxy evolution over cosmic time with the redshift distribution peaking at @xmath100 and extending all the way through to @xmath37 .
the peak observed at @xmath101 indicates the presence of a large structure of galaxies at this redshift , although further work is needed to confirm whether it is indeed real .
also shown in fig .
[ fig : dndz ] is the photometric redshift distribution for sources classified as agn by le phare , taken to be those objects with a lower @xmath95 for a qso template than for a galaxy template .
the redshift distribution of agn tends to extend to higher redshifts than those sources classified by galaxies , as we would expect , given that on average they are brighter than the quiescent galaxy population .
the solid black line is for all sources classified as galaxies from the photometric redshift fitting with le phare .
the red hashed region shows the distribution of sources classified as agn by le phare ( see text for details).,scaledwidth=48.0% ]
in this paper we have given an overview of the scientific goals , the data processing and the technical details of the vista deep extragalactic observations ( video ) survey .
our stacked images from our first major data release in the video - xmm3 field reach @xmath7 depths of @xmath8 , @xmath9 , @xmath10 , @xmath11 and @xmath12 in the ab - mag system in a 2 arcsec diameter aperture , with the expectation that the full depth in the @xmath3 band will be reached in the planned integration times .
this quality of data will eventually cover 12 degree@xmath60 of the best - studied fields in the southern hemisphere , making it a unique survey for investigating the formation and evolution of the rarest and most massive galaxies down to the sub-@xmath32 galaxy population as a function of galaxy environment , over 90 per cent of cosmic time .
for this first major data release , we have emphasized the power of combining the video survey with the cfhtls - d1 optical data in the video - xmm3 field for improving photometric redshifts ( these can be obtained from the video consortium web pagedgb / video ] ) .
the addition of the video bands to optical data results in a considerable decrease in the fraction of catastrophic failures in the photometric redshifts , from 7.4 per cent to 3.8 per cent without an @xmath15band luminosity function prior and from 5.4 to 3.3 per cent when the luminosity function prior is included .
furthermore , the accuracy of the photometric redshifts with the inclusion of the near - infrared video data should increase substantially at @xmath51 due to the ability to bracket strong spectral features , such as the balmer and 4000 breaks , over a broader range of redshifts .
however , we also emphasize the importance of additional deep optical imaging data over the full video survey area .
thus , future observations as part of the dark energy survey and the vlt survey telescope ( vst ) will play a crucial role in the full scientific exploitation of the video data .
the images and catalogues described here will be publicly available from the eso archive and the vista science archive .
access to value - added information , and updates on the progress of the video survey can also be found at the video survey website ( http://star.herts.ac.uk/@xmath88dgb/video ) .
the forthcoming release at the end of june 2012 will also include @xmath2 and @xmath102band data in the elais - s1 northern field , full details of which will accompany the data within the eso archive and the vsa .
the remainder of the video survey data and value - added catalogues will be released on regular 6 - 12 month timescales .
a description of the data products will accompany all releases and be held at the eso and vsa archives , as well as on the video survey website .
this paper is dedicated to the memory of steve rawlings , without whom the video survey may not have gotten off the ground , but also for the never ending enthusiasm , encouragement and occasional drink during 15 years working with mjj .
mjj and dgb acknowledges the support from stfc through the university of hertfordshire extragalactic rolling grant and consolidated grant .
vab and svw acknowedge support of stfc studentships and rjm would like to acknowledge the funding of the royal society via the award of a university research fellowship and the leverhulme trust via the award of a philip leverhulme research prize .
jsd acknowledges the support of the european research council through an adanced grant , and the support of the royal society via a wolfson research merit award .
mk acknowledges the support from the creative research initiative program , no .
2010 - 0000712 , of the national research foundation of korea ( nrfk ) funded by the korea government(mest ) .
mk was partially supported by the eu cost action `` black holes in a violent universe '' .
sjo acknowledges support from the science and technology facilities council [ grant number st / i000976/1 ] .
is acknowledges support from stfc and a leverhulme senior fellowship .
the cfht data used in this paper were based on observations obtained with megaprime / megacam , a joint project of cfht and cea / dapnia , at the canada - france - hawaii telescope ( cfht ) which is operated by the national research council ( nrc ) of canada , the institut national des science de lunivers of the centre national de la recherche scientifique ( cnrs ) of france , and the university of hawaii .
this work is based in part on data products produced at terapix and the canadian astronomy data centre as part of the canada - france - hawaii telescope legacy survey , a collaborative project of nrc and cnrs . | in this paper we describe the first data release of the the visible and infrared survey telescope for astronomy ( vista ) deep extragalactic observations ( video ) survey .
video is a @xmath0 degree@xmath1 survey in the near - infrared @xmath2,@xmath3,@xmath4,@xmath5 and @xmath6 bands , specifically designed to enable the evolution of galaxies and large structures to be traced as a function of both epoch and environment from the present day out to z=4 , and active galactic nuclei ( agn ) and the most massive galaxies up to and into the epoch of reionization . with its depth and area
, video will be able to fully explore the period in the universe where agn and starburst activity were at their peak and the first galaxy clusters were beginning to virialize .
video therefore offers a unique data set with which to investigate the interplay between agn , starbursts and environment , and the role of feedback at a time when it was potentially most crucial .
we provide data over the video - xmm3 tile , which also covers the canada - france - hawaii - telescope legacy survey deep-1 field ( cfhtls - d1 ) .
the released video data reach a @xmath7 ab - magnitude depth of @xmath8 , @xmath9 , @xmath10 , @xmath11 and @xmath12 in 2 arcsec diameter apertures ( the full depth of @xmath13 will be reached within the full integration time in future releases ) .
the data are compared to previous surveys over this field and we find good astrometric agreement with the two - micron all sky survey , and source counts in agreement with the recently released ultravista survey data .
the addition of the video data to the cfhtls - d1 optical data increases the accuracy of photometric redshifts and significantly reduces the fraction of catastrophic outliers over the redshift range @xmath14 from 5.8 to 3.1 per cent in the absence of an @xmath15band luminosity prior .
however , we expect the main improvement in photometric redshifts will come in the redshift range @xmath16 due to the sensitivity to the balmer and 4000 breaks provided by the near - infrared vista filters .
all images and catalogues presented in this paper are publicly available through eso s `` phase 3 '' archive and the vista science archive .
observations : galaxies - general - astronomical data bases : surveys : cosmology : large - scale structure of universe |
it is widely accepted that coalescing binary compact objects , _
i.e_. , those consisting of black holes , neutron stars , or white dwarfs , are important sources of gravitational waves ( gws ) .
ground - based detectors with higher sensitivity are currently under construction , _ e.g_. , advanced - ligo , advanced - virgo , and lcgt , while there are plans for the space - borne interferometers such as _ lisa _ @xcite , _ decigo _ @xcite and _ bbo _ @xcite .
the significant improvement of sensitivity will make it possible to directly detect gws in near future .
_ lisa _ has its best sensitivity at around @xmath6hz and one of its main targets is a merger event of binary massive black holes . on the other hand
, _ decigo / bbo _ has its best sensitivity at around @xmath7hz .
its ultimate goal is to detect primordial gw backgrounds , but it has promising astrophysical sources such as binary stellar - mass black holes .
it is also sensitive to merging signals from binary intermediate - mass black holes .
recently , much attention has been paid to the surrounding environment of the gw sources .
several types of possible electromagnetic ( em ) signatures associated with gw emissions have been proposed , such as afterglows @xcite , precursors @xcite and periodic emission @xcite in the context of a massive black hole coalescence
. those em signatures will be also very useful in localizing the position of a gw source in the sky and in identifying its redshift . by combining the redshift information with a luminosity distance determined by the gw signal
, one can obtain a distance - redshift relation .
this will be one of the best observational probes of the dark energy @xcite
. however , there are not so many works discussing em signals from coalescing binary black holes in the stellar / intermediate mass range @xmath8 .
obviously , no black holes can emit em radiation , unless they interact with their environmental gas .
thus , our interest is naturally led to binary stellar / interstellar - mass black holes in a dense gaseous medium .
@xcite investigated the effect of hydrodynamic interactions between an advection - dominated accretion flow onto a supermassive black hole and inspiraling stars on gw signals .
this effect was shown to be negligible as long as the advection - dominated accretion flow is concerned .
the effects of the interaction with a gaseous disk on the gw phase evolution have been also discussed in the context of an extreme mass ratio inspiral ( emri ) , where a compact object inspirals into a supermassive black hole @xcite . in this case
the situation is more similar to a planet migration .
it has been claimed that the existence of an accretion disk around a supermassive black hole is detectable for a certain parameter range with _
lisa_. in contrast , a binary composed of black holes with nearly equal masses residing in a gaseous environment will form a triple disk system : two accretion disks around the respective black holes and one circumbinary disk surrounding both of them @xcite . then
, the binary black holes and the circumbinary disk should mutually interact , which inevitably affects the orbital evolution of the binary .
the gas surrounding the binary acts on the orbital motion through a drag force ( tidal / resonant interaction ) , leading to the angular momentum transfer from the binary to the gas .
conversely , the angular momentum is carried from the circumbinary disk to the binary by gas accretion . since _
decigo / bbo _ will detect many cycles of gws from the binary inspiral , a rather tiny correction in the gw phase is detectable .
therefore signature of the circumbinary disk can arise in the gw signal even if the circumbinary disk is extremely less massive by many orders of magnitude than the binary . in this paper
, we give an estimate how large the mass accretion rates of the circumbinary disk should be , so that we can detect the effect of the binary - disk interaction on the gw phase evolution at a statistically significant level with _ decigo / bbo_. note that we evaluate the effect of the binary - disk interaction at the first order in a mass accretion rate of the circumbinary disk throughout this paper .
although our current understanding of the binary - disk interaction is poor , we assume here that we knew how the gw phase evolution is modified and hence the standard matched filtering analysis is applicable .
based on our estimate of the mass accretion rate required for detection , we also give a prediction for the number of merger events under detectable influence of the circumbinary disk .
in this section , we first briefly explain the basics about how the phase of gws from a binary evolves as a result of the radiation reaction due to the gw emission .
second , we present the picture that we have in mind about the interaction between the binary and the circumbinary disk , providing a quantitative estimate for the correction due to the binary - disk interaction in the gw phase evolution .
we relate the magnitude of the correction to the mass accretion rate of the circumbinary disk .
then , we evaluate the determination accuracy of the mass accretion rate of the circumbinary disk .
let us first briefly review the evolution of the gw phase of a coalescing binary , neglecting the interaction between the binary and the circumbinary disk .
close enough binaries are expected to evolve their orbits by losing the energy and angular momentum due to the gw emission . here , we assume that binary black holes are in a circular orbit , because the orbital eccentricity decreases rapidly in proportion to @xmath9 @xcite , where @xmath10 is the semi - major axis of the binary . at the leading order of the post - newtonian ( pn )
expansion , the binding energy and orbital angular momentum of a circular binary are , respectively , given by @xmath11 where @xmath12 is the speed of light , @xmath13 is the reduced mass with @xmath14 representing the total mass of binary black holes , @xmath15 is a non - dimensional quantity of @xmath16 , and @xmath17 is an orbital frequency of the binary .
the leading terms in the gw energy and angular momentum fluxes emitted from a binary are given by the well - known quadrupole formulas as @xmath18 with @xmath19 and a dot represents differentiation with respect to time .
for an equal - mass binary , @xmath20 is set to be @xmath21 .
the orbit of the binary gradually decays as a result of the loss of its energy and angular momentum due to the gw emission . equating the time variation of the orbital energy or angular momentum with the loss due to the gw emission , one can derive , to the lowest pn order , @xmath22 , which is translated into the evolution equation for the gw frequency @xmath23 as @xmath24 where we introduced the chirp mass , @xmath25 . integrating this equation with the coalescence time set to @xmath26 ,
the time evolution of the gw frequency is computed as @xmath27\ , \eta_{0.25}^{-3/8 } { \it m}_{10}^{-5/8 } \left(\frac{\it t_{\rm{0}}-t}{1yr}\right)^{-3/8 } \ , , \label{fobs}\end{aligned}\ ] ] where we use @xmath28 and @xmath29 for brevity .
one can translate this frequency evolution into the evolution of @xmath10 , using the relation @xmath30 , that is , @xmath31\ , \eta_{0.25}^{1/4 } { \it m}_{10}^{3/4 } \left(\frac{\it t_{\rm{0 } } - t}{1yr}\right)^{1/4}\ , .
\label{acoales}\end{aligned}\ ] ] as the orbit shrinks with time , the gw frequency gradually increases and finally reaches the frequency at the inner - most stable circular orbit ( isco ) @xmath32\ , { \it m}_{10}^{-1}\ , , \label{fisco}\end{aligned}\ ] ] which is here approximated by the expression in the emri limit irrespective of the mass ratio . the number of the orbital cycles before the coalescence is then calculated as @xmath33 our main interest is in binary black holes within the mass range @xmath34
. when the mass of each black hole is @xmath35 and the observation starts one year before the coalescence , the initial gw frequency and the initial semi - major axis are @xmath36 and @xmath37 , respectively . in the typical case
one can observe @xmath38 cycles of gws .
such a large number of orbital cycles allow us to precisely measure the gw phase correction possibly to the level sensitive to the tiny effect of the circumbinary disk .
the observed waveform of gws from the binary at the leading pn order is given by @xcite @xmath39 where @xmath40 is a factor of at most order unity that depends on the direction and orientation of the binary and @xmath41 is the distance from the earth to the gw source . the gw phase @xmath42 is defined by @xmath43 where @xmath44 is the gw phase at the coalescence . in this subsection , we consider the correction to the phase evolution of gws emitted from an inspiralling binary caused by the interaction with the circumbinary disk .
such binary - disk interaction induces additional extraction of the orbital angular momentum from the binary .
this effect will be tiny compared to the phase evolution driven by the gw emission when the binary separation is close enough to be observable by future gw detectors . even in that case
, however , the correction might be observable because of a large number of cycles of gws in the frequency range of observations .
for simplicity , we assume that the orbital evolution is dominated by the radiation reaction due to the gw emission . in this case it would be natural to assume that the orbit is well approximated by a quasi - circular one .
although the circumbinary disk is not necessarily aligned to the binary orbital plane , here we also assume a simple aligned configuration . the exchange between the orbital angular momentum of the binary , @xmath45 , and the angular momentum of the circumbinary disk is in two ways .
one is the angular momentum transfer through the tidal / resonant interaction @xcite .
the other is through the overflow of the gas from the inner edge of circumbinary disk onto the central binary @xcite .
the transferred gas will form accretion disks around respective black holes @xcite . during this accretion process
, some fraction of the angular momentum of the transferred gas will temporally go to the accretion disks associated with respective black holes . however , since the specific angular momentum of gas at the inner edges of the accretion disks is so tiny , only little fraction of the angular momentum of the accretion disks is transferred to the black hole spins .
thus , most of this angular momentum is transferred outward via the viscous process and is finally added to the orbital angular momentum of the binary @xmath45 via the tidal interaction @xcite .
therefore we can neglect the angular momentum that ends up with the spins of the black holes .
thus , the net angular momentum extracted from the binary orbital one due to interaction with the circumbinary disk , @xmath46 , should be almost identical to the net angular momentum flow inside the circumbinary disk , @xmath47 , which simply follows from the conservation law .
( we take the flow of the angular momentum to be positive when it flows outward . )
now , we roughly estimate the order of magnitude of @xmath47 , without relying on a specific model of the circumbinary disk . for a quasi - stationary disk
, @xmath47 should be independent of the radius @xmath48 , which should be equal to the sum of the flows of angular momentum by the viscous torque @xmath49 and by the mass accretion @xmath50 , @xmath51 except near the inner edge of the disk located at @xmath52 , where the tidal / resonant torque is non - negligible .
note that @xmath53 is negative when the mass is accreting inward . since @xmath54 increases with @xmath48 like @xmath55 for a keplerian disk , @xmath56 and @xmath54 approximately balance with each other except near the inner edge of the circumbinary disk .
if we were discussing an accretion disk around a single central object , the radius of the inner edge of the disk would be much smaller than that of the circumbinary disk . in this case
, @xmath56 should be almost balanced with @xmath54 near @xmath52 . in the present situation , however , the binary exerts the tidal / resonant torque , which truncates the disk at @xmath52 .
then , the surface density of the disk is enhanced near the inner edge by a factor of two or so compared with the case without the tidal / resonant torque ( e.g. , see figure 3 of @xcite ) .
accordingly , the viscous torque is also locally enhanced by about the same ratio .
this implies that @xmath57 where @xmath58 is the keplerian frequency of the circumbinary disk measured at @xmath59 and @xmath60 is set to be positive when the mass flux is inward . in the last approximate equality @xmath59
was approximated by @xmath61 @xcite . since currently we are not able to make the estimate of @xmath62 more precise , we introduce a parameter @xmath63 that represents the efficiency of angular momentum transfer from the binary to the circumbinary disk ( recall that @xmath64 ) .
the angular momentum transfer from a circular equal - mass binary to the circumbinary disk has been recently studied by performing hydrodynamic simulations with @xmath65-prescription @xcite and magneto - hydrodynamics ( mhd ) simulations @xcite . in both simulations
the sound velocity was set by hand instead of solving the energy equation .
in such a numerical setup , one can freely scale the binary mass @xmath66 , the semi - major axis @xmath10 , and the surface density of the circumbinary disk @xmath67 . @xcite
derived the averaged mass accretion rate @xmath68 and the angular momentum flux carried by the circumbinary disk @xmath69 from the simulated result after @xmath70 binary orbital cycles , where @xmath71 is a typical value of @xmath67 introduced for the normalization . on the other hand , @xcite derived @xmath72 and @xmath73 after @xmath74 binary orbital cycles . substituting these values of @xmath75 and @xmath76 into equation ( [ dotjoverj ] )
, we can estimate the efficiency parameter @xmath3 to be @xmath77 for the former simulation and @xmath78 for the latter simulation , independent of @xmath10 , @xmath66 , and @xmath71 .
the scatter of @xmath3 between these two simulations will be mainly attributed to the difference in the strength of shear viscosity .
@xcite assumed that the shakura - sunyaev viscosity parameter @xmath79 to be @xmath80 , whereas the effective value of @xmath79 derived from the mhd simulation by @xcite was @xmath81 .
namely , @xmath3 becomes larger for smaller viscosity .
one can explain this tendency qualitatively as follows : when the mass accretion rate is fixed , the circumbinary disk becomes more massive for smaller viscosity , while a more massive disk gains larger resonant torque from the binary .
the black hole growth due to the mass accretion from the circumbinary disk to each of the black holes via accretion disks can also modify the binary orbital evolution .
we note from equation ( [ dotjoverj ] ) that the growth rate of binary mass @xmath82 is the same order as the rate of the angular momentum transfer @xmath83 , if @xmath3 is the order of unity .
hence , for simplicity , we absorb this mass growth effect into the coefficient @xmath3 in equation ( [ dotjoverj ] ) and neglect the time variation of the binary mass in the following discussion . from the angular momentum conservation @xmath84 we can derive the increasing rate of the gw frequency from equations ( [ binde ] ) , ( [ gwflux ] ) , ( [ dotjoverj ] ) , and ( [ ecl ] ) as @xmath85 . \label{fdot}\end{aligned}\ ] ] the first term in the square brackets of equation ( [ fdot ] ) is the leading contribution from the quadrupole gw radiation given in equation ( [ quadrupoledotf ] ) .
the second term represents the effect of the circumbinary disk , and the ellipsis stands for the higher order pn corrections . recalling that @xmath86
, we note that the correction due to the binary - disk interaction has `` -4pn '' frequency dependence relative to the leading term for a constant @xmath3 . by integrating equation ( [ fdot ] ) with respect to time
, we can express @xmath87 as a function of @xmath88 , which in turn leads to the expression of @xmath88 as a function of @xmath87 : @xmath89 . \label{tgw2}\end{aligned}\ ] ] the gw phase defined by equation ( [ gwp ] ) then becomes @xmath90 . \label{gwphase2}\end{aligned}\ ] ] from the fourier transformation of equation ( [ ht ] ) , we finally find that the sky - averaged gw waveform in the fourier domain under the stationary phase approximation @xcite is given by @xmath91 where the sky - averaged gw phase : @xmath92 . \label{phase}\end{aligned}\ ] ] in this section , we analytically estimate the determination accuracy of the mass accretion rate of the circumbinary disk with _ decigo / bbo_. the determination accuracy of the mass accretion rate is mainly governed by the low frequency side in the observation band , where the effect of the circumbinary disk is the largest .
the lower cutoff frequency is determined by the duration of the observation or the detector sensitivity , @xmath93 where @xmath94 is given by equation ( [ fobs ] ) into which we substitute an appropriate observation period @xmath95 .
we also adopt @xmath96 for the cutoff due to the detector sensitivity of _ decigo / bbo_. then , @xmath97 turns out to be determined by @xmath94 for @xmath98 in the mass range @xmath99 .
the square of the signal - to - noise ratio is defined by @xmath100 where @xmath101 and @xmath102 denotes the effective number of interferometers for _ decigo / bbo _ @xcite and @xmath103 represents a total noise spectral density . as given in equation (
36 ) of @xcite , we adopt the non sky - averaged instrumental noise spectral density for _ decigo / bbo _ as @xmath104\left(\frac{\it f}{1 \mathrm{hz } } \right)^2 + 2.9\times 10^{-49}\,[\rm{hz^{-1 } } ] \nonumber \\ & & + 9.2\times 10^{-52}\,[\rm{hz^{-1}}]\left(\frac{\it f}{1 \mathrm{hz } } \right)^{-4 } .
\label{noise - inst}\end{aligned}\ ] ] this is obtained by multiplying the sky - averaged noise spectrum , which is shown in equation ( 15 ) of @xcite , by 3/20 @xcite .
the third term of the above equation dominates over the other terms at low frequencies so that @xmath105\left({\it f}/1 \mathrm{hz } \right)^{-4}$ ] . for simplicity , we assume in this subsection that @xmath106\left(\frac{f}{1\rm{hz}}\right)^{-4 } \label{noise - neg}\ ] ] by neglecting the white dwarf confusion noise from the total noise spectral density . note that the value of @xmath107 is irrelevant in the following discussion of this subsection . to assess
how accurately we can constrain the mass accretion rate , we extract the phase correction coming from the interaction between the binary and the circumbinary disk in equation ( [ phase ] ) , _
i.e. , _ @xmath108 if there are no degeneracies between the mass accretion rate and the other binary parameters , we can roughly claim that the effect of the circumbinary disk is detectable at the first order in the mass accretion rate as long as the inequality @xmath109 is satisfied .
this inequality is most likely satisfied at @xmath110 because @xmath111 is proportional to a large negative power of @xmath87 .
therefore , we apply @xmath110 to equation ( [ rough ] ) . as a typical source of _ decigo / bbo _ , we obtain @xmath112 for five year observation .
since @xmath113 is found to be proportional to @xmath114 from equations ( [ hf ] ) , ( [ varrho ] ) , and ( [ noise - neg ] ) , it exhibits no dependence on @xmath66 as long as we adopt @xmath115 .
then , the condition ( [ rough ] ) gives an analytic estimate for the determination accuracy of the mass accretion rate , @xmath116 , as @xmath117 in the mass range @xmath0 , where we normalized the determination accuracy by the eddington accretion rate : @xmath118\ , \epsilon_{0.1}^{-1 } { \it m}_{10}\ , , \label{eddrate}\end{aligned}\ ] ] and @xmath119 is the eddington luminosity with @xmath120 and @xmath121 denoting the proton mass and thomson scattering cross section , respectively .
the mass - to - energy conversion efficiency @xmath122 is set to @xmath123 in the following discussion . the relative accuracy in comparison with the eddington rate decreases with the total mass of the binary but increases in proportion to @xmath3 .
this can be understood by the facts that the eddington rate is proportional to @xmath66 and that @xmath124 in equation ( [ sgwpdisk ] ) , which gives @xmath125 .
we will compare this analytic estimate with a more precise numerical one in the succeeding subsection .
we derive the determination accuracy of the mass accretion rate by applying the matched filtering analysis @xcite .
we adopt the phase correction given in equation ( [ phase ] ) as templates , although we do not know the exact form of the correction to the gw phase .
more detailed study for more reliable prediction of the phase correction is left for future work . by neglecting the black hole spins ,
the determination accuracy of the binary parameters @xmath126 can be estimated by @xmath127 , where @xmath128 is the fisher matrix . here
, @xmath129 is the cutoff frequency on the higher frequency side , where the cutoff frequency determined by the detector sensitivity of _ decigo / bbo _ is taken as @xmath130 and @xmath131 is the gw frequency at the isco given in equation ( [ fisco ] ) . for the noise spectrum @xmath103
, we adopt the total noise spectral density ( see equation ( 36 ) of @xcite ) , which includes the confusion noise of wd / wd binaries that masks the instrumental noise below @xmath132hz @xcite .
we also assumed that the ns / ns foreground noise can be cleaned down to the level below the instrumental noise .
figure [ fig : da ] shows to what extent one can constrain the mass accretion rate of coalescing binary black holes from the gw observation with _ decigo / bbo_. the vertical axis shows the standard error of the mass accretion rate normalized by the eddington rate , whereas the horizontal axis shows the total mass of binary black holes .
the mass accretion rate is measurable at a statistically significant level , if it is greater than the determination accuracy @xmath116 .
the solid , dashed , and dotted curves represent @xmath133 for the cases of the observation time of @xmath1341 , 3 , and 5yr , respectively .
the condition @xmath135 is realized in the mass range @xmath0 for one year observation and in the mass range @xmath136 for both three year and five year observations . the determination accuracy is higher for binaries at a closer distance from the earth and is also higher with a longer observation time .
we also note that the dependence of the determination accuracy on the observation time is in good agreement with @xmath137 shown in equation ( [ mdot : low ] ) in the mass range @xmath138 . from figure
[ fig : da ] , we can estimate the mass dependence of the determination accuracy in five year observation as @xmath139 in the mass range @xmath140 . by comparing this estimate with equation ( [ mdot : low ] )
, we find that the analytic estimate tends to give about one order of magnitude higher determination accuracy and that its dependence on the binary mass is slightly different .
these are because we have neglected the contribution from the white dwarf confusion noise in the analytic estimate .
in this section , we estimate the expected number of merger events that have detectable signatures of the binary - disk interaction in gws . the presence of a circumbinary disk around binary black holes with sufficiently large accretion rate will not be ubiquitous .
to achieve such a large accretion rate as shown in equation ( [ mdot : fig ] ) , it will be required that binary black holes are embedded in a dense gaseous environment like a molecular cloud which exists in a star forming region of a galaxy .
its typical length scale of the molecular cloud is @xmath141 and its molecular hydrogen number density is distributed over the range of @xmath142 .
if we consider a binary traveling through a dense molecular cloud , the circumbinary disk will be formed around the binary through the bondi - hoyle - lyttleton ( bhl ) accretion .
the gas within the bhl radius expressed by @xmath143\ , { \it v}_{20}^{-2 } { \it m}_{10}\ , \label{rbhl}\ ] ] will accrete onto the binary , where @xmath144 and @xmath145 with the sound velocity @xmath146 for typical molecular clouds . here
, we also adopted @xmath147 as a fiducial bulk velocity of the binary relative to the ambient medium .
this is indicated from the observations that the radial velocities of some x - ray binaries including groj0422 + 32 in our galaxy are @xmath148 ( see table 1 of @xcite ) .
since the bhl radius is much less than a typical size of a molecular cloud core @xmath149 , the circumbinary disk can be formed even in the core of a molecular cloud .
while the outer edge radius of the circumbinary disk should be less than the bhl radius , its inner edge radius is approximately given by @xmath150 @xcite .
the consistency condition @xmath151 requires @xmath152\ , { \it v}_{20}^{-2}m_{10}$ ] , where @xmath153 can be read from equation ( [ acoales ] ) . at five years before the coalescence
, this condition is easily satisfied .
the rate of the bhl accretion is given by @xcite @xmath154\ , \rho_{-19 } { \it v}_{20}^{-3 } { \it m}_{10}^2\ , , \label{bhlrate}\end{aligned}\ ] ] where @xmath155 with the density of the molecular cloud @xmath156 .
its fiducial value @xmath157 corresponds to a typical density of dense molecular cloud cores . in figure 1
, the dash - dotted line represents this bhl rate normalized by the eddington rate .
the determination accuracy is higher than @xmath158 in the mass range @xmath159 for one year observation and in the almost whole mass range for both three year and five year observations .
it can happen that the mass accretion rate is in the range between @xmath116 and @xmath158 , and then the effect of the circumbinary disk is detectable at a statistically significant level with _
decigo / bbo_. from equations ( [ eddrate ] ) and ( [ bhlrate ] ) , the lowest number density of the molecular cloud for a bhl rate to be measured with the determination accuracy @xmath160 is given by @xmath161\ , \epsilon_{0.1}^{-1 } { \it v}_{20}^{3 } { \it m}_{10}^{-1 } \left(\frac{\delta\dot{\it m}}{\dot{\it m}_{\rm{edd}}}\right ) , \label{nmc}\ ] ] where we substitute @xmath116 into @xmath158 .
following @xcite , the volume fraction of molecular clouds with @xmath162 in our galaxy can be modeled as @xmath163 where @xmath164 is the molecular hydrogen number density , @xmath165 and @xmath166 are the minimum and maximum number densities of typical molecular clouds , @xmath167 is the index of power - law distribution , @xmath168 is the mean surface mass density of molecular clouds , and @xmath169 is the scale - height of the galactic disk . here , we adopted the values of @xmath170 @xcite , @xmath171 , and @xmath172 @xcite . from the condition that @xmath173 should not exceed @xmath174
, we obtain the upper limit of @xmath133 as @xmath175 the merger rate of binary stellar - mass black holes due to the gw emission in our galaxy is given by @xmath176 by `` model a '' of table 4 of @xcite .
the expected number of merger events for the binary black holes within @xmath177 for five year observation is then estimated as @xmath178 where @xmath179 with the number density of galaxies @xmath180 and the redshift @xmath181 is set to @xmath182 corresponding to @xmath183 .
we also estimate @xmath184 in the above equation by substituting equation ( [ mdot : fig ] ) into equation ( [ fmc ] ) . note that @xmath133 satisfies the condition given by equation ( [ mdotmax ] ) .
we have not taken into account the redshift evolution of the merging rate of the binaries per galaxy in the above analysis . when this effect is added , the number of events is expected to be a few times larger .
we conclude that the feasibility of detection of circumbinary disks around binary stellar - mass black holes is marginal in five year observation with @xmath185 .
we emphasize that our analysis so far does not rely on specific models of the circumbinary disk structure .
in this section , we further discuss some astrophysical implications of binary black holes with a circumbinary disk . the binary transfers its orbital energy and angular momentum to the circumbinary disk through the tidal / resonant interaction , which leads to the rapid orbital decay of the binary .
if the binary evolution towards the coalescence is accelerated significantly during its stay in the molecular cloud , the merger rate of binary black holes would be increased in comparison with the previous estimates of @xcite . in this subsection
, we examine this possibility .
let us consider a circular binary whose coalescence timescale due to the gw emission given by @xcite @xmath186\ , { \it m}_{10 } \eta_{0.25}^{-1 } \left(\frac{\it a}{{\it r}_{\rm{g}}}\right)^4 \label{tgw}\end{aligned}\ ] ] exceeds the hubble time .
the binary - disk interaction will be one of the key mechanisms to force such binary to merge within the hubble time .
the orbital decay timescale due to the tidal / resonant interaction corresponds to the inverse transfer rate of the orbital angular momentum , which is given by equation ( [ dotjoverj ] ) as @xmath187\ , { \it j}^{-1 } \epsilon_{0.1 } \eta_{0.25 } \dot{\it m}^{-1}.\end{aligned}\ ] ] this is consistent with equation ( 17 ) of @xcite when @xmath188 is adopted . owing to this tidal / resonant interaction , the orbital evolution
is accelerated even when @xmath189 holds , _
i.e_. , in the range @xmath190 with @xmath191 if the binary stays in the molecular cloud for a sufficiently long term .
since the typical size of a giant molecular cloud is @xmath192 , the crossing time of binary black holes in the molecular cloud is @xmath193\,{\it r}_{\rm{mc,100 } } ( { \it v}_{\rm{bulk}}/20\rm{km\,s^{-1}})^{-1}$ ] , where @xmath194 . during this crossing time
, the binary can keep interacting with the molecular cloud gas .
the condition @xmath195 leads to @xmath196 with the critical accretion rate @xmath197 if @xmath196 , binary black holes can merge within the hubble time , even if the initial separation is too large to coalesce within the hubble time by means of the radiation reaction due to the gw emission alone . in this subsection
, we evaluate the mass of the circumbinary disk with the aid of the assumption that the structure of the circumbinary disk is described by the _ standard disk model _ @xcite .
the circumbinary disk is assumed to be a steady , axisymmetric , geometrically thin , gas - pressure and electron - scattering opacity dominated , keplerian disk . here , we use the following solution of the standard disk @xcite : @xmath198\ , \nonumber \\ & & \times \alpha_{\rm{ss}}^{-4/5 } \epsilon_{0.1}^{-3/5 } m_{10}^{1/5 } \dot{m}^{3/5 } \left(\frac{\it r}{{\it r}_{\rm{g}}}\right)^{-3/5 } , \label{sig } \\
h(r ) & \approx & 1.6\times10^{4}\,[\rm{cm}]\ , \nonumber \\ & & \times \alpha_{\rm{ss}}^{-1/10 } \epsilon_{0.1}^{-1/5 } m_{10}^{9/10 } \dot{m}^{1/5 } \left(\frac{\it r}{{\it r}_{\rm{g}}}\right)^{21/20 } , \label{hr } \\
c_{\rm{s}}(r ) & \approx & 1.7\times10^{8}\,[\rm{cm\,s^{-1}}]\ , \nonumber \\ & & \times \alpha_{\rm{ss}}^{-1/10 } \epsilon_{0.1}^{-1/5 } m_{10}^{-1/10 } \dot{m}^{1/5 } \left(\frac{\it r}{{\it r}_{\rm{g}}}\right)^{-9/20 } , \label{cs}\end{aligned}\ ] ] where @xmath67 , @xmath199 , @xmath200 , @xmath201 , and @xmath202 are the surface density , the scale - height of the disk , the sound velocity measured at the mid - plane temperature of the disk , the gravitational radius , and the normalized mass accretion rate , respectively ( recall that @xmath79 is the shakura - sunyaev viscosity parameter ) . as it is not the whole mass of the circumbinary disk but the mass at around the inner edge of the circumbinary disk that affects the gw phase evolution , we estimate @xmath203 and refers to it as an inner - most disk mass .
it can be evaluated by substituting @xmath204 into the expression of @xmath67 given in equation ( [ sig ] ) .
figure [ fig : dm ] shows the inner - most disk mass of the circumbinary disk that is detectable by gws .
they are evaluated at the beginning of the observation period of @xmath134@xmath205 , @xmath206 , and @xmath207 yr , respectively .
we also adopt @xmath208 .
it shows that the existence of the circumbinary disk can be confirmed , even if the disk mass is less than the binary mass by many orders of magnitude .
this is because we can observe many ( @xmath209 ) cycles of gws .
we can understand the binary mass dependence of @xmath210 in figure [ fig : dm ] in the following way .
the semi - major axis for a fixed value of @xmath95 is proportional to @xmath211 from equation ( [ acoales ] ) and the surface density at the inner edge of the circumbinary disk is proportional to @xmath212 from equation ( [ sig ] ) .
when @xmath60 is given in equation ( [ mdot : low ] ) , we find @xmath213 in the mass range @xmath0 for five year observation .
this is roughly in agreement with the dotted curve in figure [ fig : dm ] .
when @xmath214 of the binary is shorter than the viscous timescale evaluated at the inner edge of the circumbinary disk , the binary is decoupled from the circumbinary disk .
after the decoupling , no binary - disk interaction occurs so that it is impossible to detect the effect of the circumbinary disk by the gw observation .
the viscous timescale measured at the inner edge of the circumbinary disk is given by @xmath215 where we used the relation for the disk viscosity @xmath216 and @xmath217 with equations ( [ hr ] ) and ( [ cs ] ) .
the decoupling radius at which @xmath218 is given by @xmath219 even when @xmath214 is shorter than @xmath220 , the binary continues to gravitationally interact with the circumbinary disk until the binary orbit decays down to @xmath221 . in the mass range of @xmath222 ,
the condition @xmath223 is satisfied even at one year before the coalescence .
namely , the binary continues to gravitationally interact with the circumbinary disk , as we described , at least at the early stage of the gw phase evolution during the observation period .
thus , although we take no account of the decoupling effect in figure [ fig : da ] , this neglect has been justified in the mass range @xmath224 . for the larger mass range ,
the determination accuracy is deteriorated by taking into account the decoupling effect . however , for @xmath225 , _ decigo / bbo _ is completely insensitive to the circumbinary disk when the mass accretion rate is smaller than the eddington rate .
the bhl accretion rate can exceed the eddington rate depending on parameters : @xmath226 , @xmath227 , and @xmath66 .
x - ray observations have detected the bright x - ray sources , such as grs 1915 + 105 in our galaxy , with a luminosity over the eddington one ( cf .
such large luminosities can be explained by a supercritical accretion flow onto stellar - mass black holes .
if the circumbinary disk is in a super - critical state , equation ( [ ad ] ) does not apply .
since the accretion timescale is much shorter than the viscous timescale in the super - critical state , the binary will not be decoupled from the circumbinary disk until the coalescence , although it is still unclear how such a circumbinary disk evolves @xcite .
we have derived the minimum mass accretion rate of a circumbinary disk around binary stellar / intermediate - mass black holes necessary for the detection by future space - borne gw detectors such as _ decigo / bbo_. our main conclusions are summarized as follows : 1 . the circumbinary disk with @xmath228\,{\it j}^{-1 } ( { \it m}/10{\it m}_\odot)^{33/23}$ ] with @xmath2 mass - to - energy conversion efficiency , where @xmath3 represents the efficiency of the angular momentum transfer defined by equation ( [ dotjoverj ] ) , within @xmath177 from the earth is detectable at a statistically significant level with _
decigo / bbo _ by observing for five years before the coalescence .
the determination accuracy of the mass accretion rate is higher at a closer distance from the earth , and is also higher for a longer observation time . 2 .
if binary stellar - mass black holes are residing in sufficiently dense molecular clouds , the influence of the circumbinary disk will be detectable . the number of merger events of such binaries within @xmath177 for five year observation , is approximately estimated to be @xmath229 by assuming typical values for the event rate of the black hole coalescence and that molecular clouds have the number density from @xmath230 to @xmath174 given in equation ( [ nmc ] ) .
this suggests that there might be a possibility to detect the circumbinary disk around binary stellar - mass black holes with _
decigo / bbo_. in the estimate presented in this paper , we still have a lot of uncertainties , _
e.g. , _ the formation rate of binary black holes , the molecular hydrogen number density , the relative velocity to the molecular cloud , the efficiency of angular momentum transfer ( @xmath231 : see equation ( [ dotjoverj ] ) ) , and so on .
therefore , we should understand that there remains a possibility that the event rate is one or two orders of magnitude larger than our estimate . for the idealized situation that we discussed in this paper ,
the correction to the gw phase evolution shows @xmath232pn frequency dependence .
this dependence is the same as the one for a braneworld model discussed in @xcite .
similarly , the cosmic acceleration @xcite and the alternative theories of gravity in which the gravitational constant @xmath233 is not a costant @xcite also give the corrections with @xmath232pn frequency dependence .
furthermore , the acceleration acting on a binary in time - independent external gravitational field gives the correction with the same frequency dependence @xcite .
therefore , once the deviation from the standard template is discovered , discriminating the effect of the circumbinary disk from the others will be a difficult task .
however , for example , the binary - disk interaction rapidly decays after the decoupling of the circumbinary disk as we discussed in section [ sec : decoupling ] .
this is a signature unique to the interaction with the circumbinary disk .
in addition , there can be a method to distinguish the effect of the circumbinary disk from that of modified gravity theory by detecting the gw signals from multiple sources .
this is because the effect of modified gravity theory is universal , whereas that of circumbinary disk is not .
therefore , further detailed studies may provide some ways of distinctions .
in this paper we have assumed that the binary is in a circular orbit .
the orbital eccentricity also gives the frequency dependence of negative pn order ( -19/6pn ) @xcite .
inclusion of the orbital eccentricity as a parameter of the waveform reduces the determination accuracy of @xmath232pn order term by some factors .
however , we expect that we can determine both the orbital eccentricity and the mass accretion rate for the large mass parameter range with errors of @xmath229 for typical binary parameters .
@xcite calculated the determination accuracies of full seventeen binary parameters including an initial orbital eccentricity for binary massive black holes with _
lisa _ , using a bayesian analysis under markov chain monte carlo simulations .
they conclude that the _ lisa _ could distinguish between circular orbit and eccentric orbit with a very small orbital eccentricity @xmath234 . furthermore
, we would like to point out that the effect of the circumbinary disk can be important , even if it is suppressed to the undetectable level for each event .
the primary mission of the space - borne gw antennas such as _ decigo / bbo _ is to measure the primordial gw background radiation . in order to achieve this goal ,
abundant binary gw sources become the foreground noises . to remove them , the expected gw signals for the events identified by the matched filtering
will be subtracted from the data . however , if the theoretical templates contain some unknown elements , the subtraction leaves residuals that remain to contribute to confusion noises @xcite . finally , we briefly mention the relevance of our discussion to the ground - based gw detectors . with these detectors , the distance to which we can detect gws from binary black holes
is reduced , and hence the expected event rate is extremely small . using the third - generation - ground - based gw detector : einstein telescope ( et ) , and assuming
high signal - to - noise ratio corresponding to an accidentally nearby event at @xmath235mpc , the effect of the circumbinary disk will be detectable only for the mass accretion rate @xmath236 times higher than the eddington rate .
even if stellar - mass binary black holes are at rest in a dense molecular cloud core , the expected bhl accretion rate is @xmath237 times as much as the eddington rate .
hence , we would be able to safely conclude that the effect of the circumbinary disk becomes relevant only for space - borne gw detectors such as _ decigo / bbo_. the authors are grateful to the members of the regular meeting on gravity and gravitational waves ( gg seminar ) at department of physics , kyoto university for helpful and continuous discussions .
this work has been supported by the grants - in - aid of the ministry of education , science , culture , and sport and technology ( mext ; 21540304 , 22540243 , 23540271 kh , 22340045 sm , 21244033 , 21111006 , 22111507 tt ) and in part by the grant - in - aid for the global coe program `` the next generation of physics , spun from universality and emergence '' . k.y .
is also supported by the japan society for the promotion of science grant no .
@xmath238 .
agol , e. , & kamionkowski , m. 2002 , , 334 , 553 armitage , p. j. , & natarajan , p. 2002
, , 567 , l9 artymowicz , p. , & lubow , s.h .
1994 , , 421 , 651 artymowicz , p. , & lubow , s.h .
1996 , , 467 , l77 barausse , e. , & rezzolla , l. 2008 , phys .
d , 77 , 104027 belczynski , k. , kalogera , v. , & bulik , t. 2002 , , 572 , 407 berti , e. , buonanno , a. , & will , c.m .
2005 , phys .
d 71 , 084025 bode , t. , haas , r. , bogdanovi , t. , laguna , p. , & shomarker , d. 2010 , , 715 , 1117 bondi , h. , & hoyle , f. 1944 , , 104 , 273 bogdanovi , t. , smith , b.d . , sigurdsson , s. , & eracleous , m. 2008 , , 174 , 455 chang , p. , strubble , l.e .
, menou , k. , & quataret , e. 2010 , , 407 , 2007 cuadra , j. , armitage , p.j . ,
alexander , r.d . , & begelman , m.c .
2009 , , 393 , 1423 cutler , c. , & holz , d.e .
2009 , phys .
d , 80 , 104009 culter , c. , & flanagan e.e .
1994 , phys .
d , 49 , 2658 cutler , c. , & harms , j. 2006 , phys .
d , 73 , 042001 danzmann , k. 1997 , class .
, 14 , 1399 done , c. , gierlinski , m. , & kubota , a. 2007 , a&ar , 15 , 1 harms , j. , christoph , m. , markus , o. , & malte , p. 2008
d , 77 , 123010 hayasaki , k. , mineshige , s. , & ho , c.l .
2008 , , 682 , 1134 hayasaki , k. , mineshige , s. , & sudou , h. 2007 , , 59 , 427 hayasaki , k. 2009 , , 61 , 65 hayasaki , k. , ueda , y. , & isobe , n. 2010 , , 62 , 1351 holz , e.d .
, & hughes , a.s .
2005 , , 629 , 15 kato , s. , fukue , j. , & mineshige , s. _ black - hole accretion disks _ , ( kyoto univ . press .
key , j.s . , &
cornish , n.j .
2011 , phys .
d , 83 , 3001 kocsis , b. , yunes , n. , & loeb , a. 2011 , phys .
d , 84 , 024032 macfadyen , i.a . , & milosavljevi , m. , 2008 , , 672,83 mii , h. , & totani , t. 2005 , , 628 , 873 luc blanchet . 2006 , living rev .
relativity , 9 , 4 milosavljevi , m. , & phinney , e.s .
2005 , , 622 , l93 narayan , r. 2000 , , 536 , 663 nelemans , g.t.m .
, & van den heuvel , e.p.j . 1999 , , 352 , l87 peters , p.c .
1964 , phys .
, 136 , 1224 phinney , e.s . , et al . 2003 , _ big bang observer mission concept study _
( nasa ) farmer , a.j . , & phinney , e.s .
2003 , , 346 , 1197 sanders , d.b . , solomon , p.m. & scoville , n.z .
1984 , , 276 , 182 seto , n. , kawamura , s. , & nakamura , t. 2001 , phys .
lett , 87 , 221103 shi , ji - m . , & krolik , h.j . ,
lubow , s.h . , & hawley , f. j. 2011 , arxiv:1110.4866 shakura , n. i. , & sunyaev , r. a. 1973 , , 24 , 337 schutz , b.f .
1986 , , 323 , 310 tanaka , t. , & menou , k. 2010 , , 714 , 404 yagi , k. , tanahashi , n. , & tanaka , t. 2011 , phys .
d , 83 , 084036 yunes , n. , pretorius , f. , & spergel , d. 2010 , phys .
d , 81 , 064018 yunes , n. , coleman , m. , & thornburg , j. 2011 , phys .
d , 83 , 044030 yunes , n. , kocsis , b. , loeb a. , & haiman , z. 2011 , phys .
lett , 107 , 171103 | when binary black holes are embedded in a gaseous environment , a rotating disk surrounding them , the so - called circumbinary disk , will be formed .
the binary exerts a gravitational torque on the circumbinary disk and thereby the orbital angular momentum is transferred to it , while the angular momentum of the circumbinary disk is transferred to the binary through the mass accretion .
the binary undergoes an orbital decay due to both the gravitational wave emission and the binary - disk interaction .
this causes the phase evolution of the gravitational wave signal .
the precise measurement of the gravitational wave phase thus may provide information regarding the circumbinary disk . in this paper
, we assess the detectability of the signature of the binary - disk interaction using the future space - borne gravitational wave detectors such as _ decigo _ and _ bbo _ by the standard matched filtering analysis .
we find that the effect of the circumbinary disk around binary black holes in the mass range @xmath0 is detectable at a statistically significant level in five year observation , provided that gas accretes onto the binary at a rate greater than @xmath1\,{\it j}^{-1}({\it m}/10{\it m}_\odot)^{33/23}$ ] with @xmath2 mass - to - energy conversion efficiency , where @xmath3 represents the efficiency of the angular momentum transfer from the binary to the circumbinary disk .
we show that @xmath4 coalescence events are expected to occur in sufficiently dense molecular clouds in five year observation .
we also point out that the circumbinary disk is detectable , even if its mass at around the inner edge is by over @xmath5 orders of magnitude less than the binary mass . |
exclusive reactions @xmath5 which may be accessed in @xmath6 collisions have been shown @xcite to have a partonic interpretation in the kinematical region of large virtuality of one photon and of small center of mass energy .
the scattering amplitude factorizes in a long distance dominated object the generalized distribution amplitude ( gda ) and a short distance perturbatively calculable scattering matrix .
a phenomenological analysis of the @xmath7 channel @xcite has shown that precise experimental data could be collected at intense @xmath6 collider experiments such as babar and belle . meanwhile
, first data on the @xmath8 channel at lep have been published @xcite and analyzed @xcite , showing the compatibility of the qcd leading order analysis with experiment at quite modest values of @xmath9 . in this paper , we focus on the comparison of processes @xmath10 and @xmath11 in the context of searching an exotic isospin @xmath2 resonance decaying in two @xmath3 mesons ; such channels have recently been studied at lep by the l3 collaboration @xcite . a related study for photoproduction
@xcite raised the problem of @xmath1 enhancement with respect to @xmath0 at low energies .
one of the solutions of this problem was based on the prediction @xcite and further exploration @xcite of the possible existence of isotensor state , whose interference with the isoscalar state is constructive for neutral mesons and destructive for charged ones .
this option was also independently considered in @xcite .
the crucial property of such an exotic state is the absence of @xmath12 wave function at any momentum resolution . in other words ,
quark - antiquark component is absent both in its non - relativistic description and at the level of the light - cone distribution amplitude .
this is by no means common : for instance , the @xmath13 state which is a quark - gluon hybrid at the non - relativistic level is described by a leading twist quark - antiquark distribution amplitude @xcite .
contrary to that , an isotensor state on the light cone corresponds to the twist @xmath14 or higher and its contribution is thus power suppressed at large @xmath9 .
this is supported by the mentioned l3 data , where the high @xmath9 ratio two of the cross sections of charged and neutral mesons production points out an isoscalar state .
we studied both perturbative and non - perturbative ingredients of qcd factorization for the description of an isotensor state .
namely , we calculated the twist @xmath14 coefficient function and extracted the non - perturbative matrix elements from l3 data . our analysis is compatible with the existence of an isotensor exotic meson with a mass around @xmath15 gev .
the reaction which we study here is @xmath17 , where @xmath3 stands for the triplet @xmath3 mesons ; the initial electron @xmath18 radiates a hard virtual photon with momentum @xmath19 , with @xmath20 quite large .
this means that the scattered electron @xmath21 is tagged . to describe the given reaction , it is useful to consider the sub - process @xmath22 . regarding the other photon momentum @xmath23
, we assume that , firstly , its momentum is almost collinear to the electron momentum @xmath24 and , secondly , that @xmath25 is approximately equal to zero , which is a usual approximation when the second lepton is untagged . in two @xmath3 meson production , we are interested in the channel where the resonance corresponds to the exotic isospin , _ i.e
_ @xmath26 , and usual @xmath27 quantum numbers . the @xmath27 quantum numbers are not essential for our study .
because the isospin @xmath2 has only a projection on the four quark correlators , the study of mesons with the isospin @xmath2 can help to throw light upon the four quark states .
we thus , together with the mentioned reactions , study the following processes : @xmath28 and @xmath29 , where meson @xmath30 possesses isospin @xmath26 .
considering the amplitude of the @xmath31 subprocess , we write @xmath32 | 0 \rangle , \end{aligned}\ ] ] where @xmath33 denotes the quark electromagnetic current @xmath34 with the charge matrix @xmath35 belonging to @xmath36 group .
the photon polarization vectors read @xmath37 for the real and virtual photons , respectively .
the coefficient functions of twist @xmath2 operators to operator product expansion of currents product in ( [ amp01 ] ) were discussed in detail in @xcite , while the contributions of new twist @xmath14 operators are described by coefficient functions calculated long ago in @xcite when considering the problem of twist @xmath14 corrections to deep inelastic scattering .
let us now turn on the flavour or isospin structure of the corresponding amplitudes .
the @xmath38 state with @xmath39 can be projected on both the two and four quark operators , while the state with @xmath26 on the four quark operator only .
indeed , let us start from the consideration of the vacuum to@xmath38 matrix element in ( [ amp01 ] ) @xmath40 where the quark fields are shown with free flavour indices and @xmath41 stands for the corresponding @xmath42-matrix .
the isoscalar and isovector gda s @xmath43 in ( [ cor2q ] ) are well - known , see for instance @xcite .
note that , in ( [ cor2q ] ) , the correspondence between triplets @xmath44 and @xmath45 is given by the standard way .
moreover , for the coefficient function at higher order in the strong coupling constant , the corresponding matrix element gives us @xmath46 [ \bar\psi_{f_2}(z ) \gamma_2 \psi_{g_2}(\xi ) ] |0\rangle.\end{aligned}\ ] ] using the clebsch - gordan decomposition , we obtain @xmath47\ , [ \bar\psi_{f_2 } \psi_{g_2 } ] \biggr)^{i=0,\,i_z=0}&\rightarrow & -\frac{1}{\sqrt{3}}\biggl [ \frac{1}{2 } \tau^0_{f_1 g_1}\tau^0_{f_2 g_2 } + \tau^+_{f_1 g_1}\tau^-_{f_2 g_2 } + \tau^-_{f_1 g_1}\tau^+_{f_2 g_2 } \biggr ] \tilde\phi^{i=0,\,i_z=0}\end{aligned}\ ] ] for the isospin @xmath48 and @xmath49 projection of the four quark operator in ( [ cor4q ] ) , and @xmath50\ , [ \bar\psi_{f_2 } \psi_{g_2 } ] \biggr)^{i=2,\,i_z=0}&\rightarrow & \frac{1}{\sqrt{6}}\biggl [ \tau^0_{f_1 g_1}\tau^0_{f_2 g_2 } - \tau^+_{f_1 g_1}\tau^-_{f_2 g_2 } - \tau^-_{f_1 g_1}\tau^+_{f_2 g_2 } \biggr ] \tilde\phi^{i=2,\,i_z=0}\end{aligned}\ ] ] for the isospin @xmath2 and @xmath49 projection of the four quark operator in ( [ cor4q ] ) .
the four quark gda s @xmath51 can be defined in an analogous manner as the two quark gda s .
hence , one can see that the amplitudes ( [ amp01 ] ) for @xmath1 and @xmath0 productions can be written in the form of the decomposition : @xmath52 where the subscripts @xmath2 and @xmath14 in the amplitudes imply that the given amplitudes are associated with the two and four quark correlators , respectively .
the amplitudes corresponding to @xmath0 production are not independent and can be expressed through the corresponding amplitudes of @xmath1 production .
indeed , one can derive the following relations : @xmath53 the amplitude of two @xmath3 meson production in two photon collision can be also presented through a resonant intermediate state .
the vacuum to @xmath38 matrix element in the _ r.h.s .
_ of ( [ amp01 ] ) can be traded for @xmath54 |0\rangle.\end{aligned}\ ] ] where @xmath55 is the resonance with three possible isospin @xmath56 .
note that , in our case , only isospin @xmath48 and @xmath2 cases are relevant due to the positive @xmath57-parity of the initial and final states .
the matrix element @xmath58 defines the corresponding coupling constant of meson and @xmath59 |0\rangle$ ] is considered up to the second order of strong coupling constant @xmath60 , _
i.e _ this matrix element is written as a sum of two- and four - quark correlators .
previously , the theoretical description of the experimental data collected for the @xmath1 production has been performed in @xcite .
now , the subject of our study is the differential cross section corresponding to both the @xmath1 and @xmath0 productions in the electron positron collision . using the equivalent photon approximation @xcite we find the expression for the corresponding cross section : @xmath61 where the usual weizsacker - williams function @xmath62 is used . in ( [ xsec5 ] )
, the cross section for the subprocess reads @xmath63 where the amplitude @xmath64 is defined by ( [ sumam ] ) . for the case of @xmath1 production ,
the cross section ( [ xsec5 ] ) takes the form : @xmath65 ^ 2 + \biggr .
\nonumber\\ \biggl . & & \frac{\gamma_{r^2 } m_{r^2}}{\beta_2((m^2_{r^2}-w^2)^2+\gamma_{r^2}^2m^2_{r^2 } ) } \biggl[\frac{\alpha_s(q^2)m^2_{r^2}}{q^2}{\bf s}^{i=2,i_3=0}_4\biggr]^2 + \biggr .
\nonumber\\ \biggl . & & 2\sqrt{\frac{\gamma_{r^0}\gamma_{r^2 } m_{r^0 } m_{r^2}}{\beta_0 \beta_2 } } \frac{(m^2_{r^0}-w^2)(m^2_{r^2}-w^2)+(\gamma_{r^0}m_{r^0})(\gamma_{r^2}m_{r^2 } ) } { ( ( m^2_{r^0}-w^2)^2+\gamma_{r^0}^2m^2_{r^0})((m^2_{r^2}-w^2)^2+\gamma_{r^2}^2m^2_{r^2})}\times \biggr . \nonumber\\
& \biggl [ { \bf s}^{i=0,i_3=0}_2 + \frac{\alpha_s(q^2)m^2_{r^0}}{q^2}{\bf s}^{i=0,i_3=0}_4\biggr ] \frac{\alpha_s(q^2)m^2_{r^2}}{q^2}{\bf s}^{i=2,i_3=0}_4 \biggr ) , \nonumber\end{aligned}\ ] ] where @xmath66 stand for the total widths .
the dimensionful structure constants @xmath67 and @xmath68 are related to the nonperturbative vacuum to meson matrix elements .
the @xmath69functions are also defines in the standard ways : @xmath70 and @xmath71 .
the function @xmath72 in ( [ xsec7 ] ) is equal to @xmath73.\end{aligned}\ ] ] the differential cross section corresponding to @xmath0 production can be obtained using ( [ rel1 ] ) , we have @xmath74 ^ 2 + \biggr . \nonumber\\
\biggl . & & \frac{1}{4}\frac{\gamma_{r^2 } m_{r^2}}{\beta_2((m^2_{r^2}-w^2)^2+\gamma_{r^2}^2m^2_{r^2 } ) } \biggl[\frac{\alpha_s(q^2)m^2_{r^2}}{q^2}{\bf s}^{i=2,i_3=0}_4\biggr]^2- \biggr . \nonumber\\
\biggl . & & \sqrt{\frac{\gamma_{r^0}\gamma_{r^2 } m_{r^0 } m_{r^2}}{\beta_0 \beta_2 } } \frac{(m^2_{r^0}-w^2)(m^2_{r^2}-w^2)+(\gamma_{r^0}m_{r^0})(\gamma_{r^2}m_{r^2 } ) } { ( ( m^2_{r^0}-w^2)^2+\gamma_{r^0}^2m^2_{r^0})((m^2_{r^2}-w^2)^2+\gamma_{r^2}^2m^2_{r^2})}\times \biggr . \nonumber\\ \biggl . & & \biggl [
{ \bf s}^{i=0,i_3=0}_2 + \frac{\alpha_s(q^2)m^2_{r^0}}{q^2}{\bf s}^{i=0,i_3=0}_4\biggr ] \frac{\alpha_s(q^2)m^2_{r^2}}{q^2}{\bf s}^{i=2,i_3=0}_4 \biggr ) , \nonumber\end{aligned}\ ] ] note that we have explicitly separated out , in ( [ xsec7 ] ) and ( [ xsec8 ] ) , the running coupling constant @xmath75 which appears in the twist @xmath14 terms . because of we will study the @xmath9 dependence of the corresponding cross sections at rather small values of @xmath9
, we use the shirkov and solovtsov s analytical approach @xcite to determine the running coupling constant in the region of small @xmath9 . detailed discussion on different aspects of using the analytical running coupling constant may be found in @xcite and references therein .
in the previous section we derived the differential cross sections @xmath76 for both the @xmath1 and @xmath0 channels , based on the qcd analysis .
these expressions contain a number of unknown phenomenological parameters , which are intrinsically related to non perturbative quantities encoded in the generalized distribution amplitudes .
one should now make a fit of these phenomenological parameters in order to get a good description of experimental data .
the best values of the parameters can be found by the method of least squares , @xmath77-method , which flows from the maximum likelihood theorem , but we postpone a comprehensive @xmath77-analysis to a forthcoming more detailed paper . here , we implement a naive fitting analysis to get an acceptable agreement with the experimental data .
thus , we have the following set of parameters for fitting : @xmath78 we start with the study of the @xmath79 dependence of the cross sections . for this goal , following the papers @xcite
, we determine the cross section of process @xmath80 normalized by the integrated luminosity function : @xmath81 where the definition of the luminosity function @xmath82 is taken from @xcite .
the value @xmath83 corresponds to the center of each bin , see @xcite .
focussing first on the region of larger @xmath9 we fit the parameters associated with the dominant contribution which comes from the twist @xmath2 term amplitude , which is associated with the non - exotic resonance ( or background ) with isospin @xmath39 .
generally speaking , there are many isoscalar resonances with masses in the region of @xmath84 gev . to include their total effect we introduce a mass and width for an `` effective '' isoscalar resonance .
we then determine the values of the mass and width by fitting the data for the region of larger @xmath9 ( _ i.e. _ , when @xmath9 is in the interval @xmath85 gev@xmath86 ) .
we thus can fix the parameters @xmath68 , @xmath87 and @xmath88 .
good agreement can be achieved with @xmath89 , @xmath90 and @xmath68 within the interval @xmath91 . as can be expected
the width of the effective isoscalar `` resonance '' is fairly large .
it means that we actually deal with a non - resonant background .
next , we fit the @xmath92dependence of the cross section for small values of @xmath9 , _
i.e. _ @xmath93 gev@xmath86 . in this region
all twist contributions may be important .
we find that the experimental data can be described by the following choice of the parameters : @xmath94 , @xmath95 while the parameters @xmath96 and @xmath97 are in the intervals @xmath98 and @xmath99 , respectively . @xmath100{fig1.eps}\ ] ] @xmath100{fig2.eps}\ ] ] further , we include in our analysis the @xmath9 dependence of @xmath1 and @xmath0 production cross sections , _ i.e. _ @xmath101 , which should fix the remaining arbitrariness of the parameters .
we finally find that the best description of both @xmath79 and @xmath9 dependence is reached at @xmath102 note that these rather small values of twist @xmath14 structure constants @xmath103 compared to the twist @xmath2 structure constant @xmath104 indicate that leading twist contribution dominate for the values @xmath9 around or greater than @xmath105 .
this should be compared with what was obtained in a particular renormalon model in @xcite . our theoretical description of the lep experimental data are presented on figs .
[ rho0rho02][rhorho1 ] .
the plots depicted on figs .
[ rho0rho02][rho+rho- ] have the following notations : the short - dashed line corresponds to the contribution coming from the leading twist term of ( [ xsec7 ] ) ; the dash - dotted line to the contribution from the twist @xmath14 term of ( [ xsec7 ] ) ; the middle - dashed line to the contributions from the interference between twist @xmath2 and twist @xmath14 terms of ( [ xsec7 ] ) and ( [ xsec8 ] ) ; the long - dashed line to the contribution from the interference between isoscalar and isotensor terms . finally the solid line corresponds to the sum of all contributions . on fig .
[ rhorho1 ] , we present the lep data and our theoretical curves for both the @xmath1 and @xmath0 production differential cross sections as functions of @xmath9 . the solid line on fig .
[ rhorho1 ] corresponds to the @xmath1 differential cross section while the dashed one to the @xmath0 differential cross section .
@xmath100{fig3.eps}\ ] ] @xmath100{fig4.eps}\ ] ]
the fitting of lep data based on the qcd factorization of the amplitude into a hard subprocess and a generalized distribution amplitude thus allows us to claim evidence of the existence of an isospin @xmath26 exotic meson @xcite with a mass in the vicinity of @xmath4 and a width around @xmath106 .
the contributions of such an exotic meson in the two @xmath3 meson production cross sections ( see , ( [ xsec7 ] ) and ( [ xsec8 ] ) ) are directly associated with some twist @xmath14 terms that we have identified . at large @xmath9 , these twist @xmath14 contributions become negligible and the behaviours of the @xmath1 and @xmath0 cross sections are controlled by the leading twist @xmath2 contributions , see fig .
[ rho0rho02 ] and [ rho+rho-2 ] . figs .
[ rho0rho0 ] and [ rho+rho- ] show the increasing role of higher twist contributions when decreasing @xmath9 .
namely , the interference between twist @xmath2 and @xmath14 amplitudes gives the dominant contributions to @xmath1 production in the lower @xmath9 interval , and is thus responsible of the @xmath79 dependence of the cross section in these kinematics . in particular ,
in this interference term the main contribution arises from the interference between isoscalar and isotensor structures , see the long - dashed lines on fig .
[ rho0rho0 ] and [ rho+rho- ] .
@xmath100{fig55.eps}\ ] ] analysing the @xmath9 dependence , we can see that due to the presence of a twist @xmath14 amplitude and its interference with the leading twist @xmath2 component , the @xmath1 cross section at small @xmath9 is a few times higher than the @xmath0 cross section , see fig .
[ rhorho1 ] . while for the region of large @xmath9 where any higher twist effects are negligible the @xmath1 cross section is less than the @xmath0 cross section by the factor @xmath2 , which is typical from an isosinglet channel ( see also ( [ xsec7 ] ) and ( [ xsec8 ] ) ) . the reaction @xmath107 and its qcd analysis in the framework of ref .
@xcite thus proves its efficiency to reveal facts on hadronic physics which would remain quite difficult to explain in a quantitative way otherwise .
the leading twist dominance is seen to persist down to values of @xmath9 around @xmath105 .
other aspects of qcd may be revealed in different kinematical regimes through the same reaction @xcite .
its detailed experimental analysis at intense electron colliders within the babar and belle experiments is thus extremely promising .
data at higher energies in a future linear collider should also be foreseen .
note that the non - perturbative calculations of the relevant @xmath26 twist @xmath14 matrix elements also deserve special interest .
in particular , one may follow the ideas developed for pion distribution @xcite which allowed to relate higher and lower twists in multicolour qcd .
the generalization for the case of @xmath3 mesons , anticipated by the authors of @xcite , and use of crossing relations between various kinematical domains provided by radon transform technique @xcite may allow to apply these result in the case under consideration . in conclusion ,
let us stress that the l3 data allows to estimate the contribution of higher twist four quark light cone distribution to the production amplitude of vector meson pairs .
our numerical analysis leads to a rather small width for the corresponding resonant state , which is nothing else as an exotic four - quark isotensor meson . at the same time , a more elaborate experimental , theoretical and numerical analysis is required to confirm , with better accuracy , the smallness of the width and the existence of an exotic meson .
we are grateful to n.n .
achasov , a. donnachie , j. field , k. freudenreich , m. kienzle , n. kivel , k.f .
liu , m.v .
polyakov and i. vorobiev for useful discussions and correspondence .
is indebted to theory division of cern and cpht , cole polytechnique , for warm hospitality .
i.v.a . expresses gratitude to theory division of cern and university of geneva for financial support of his visit .
this work has been supported in part by rffi grant 03 - 02 - 16816 .
thanks nato for a grant .
m. diehl , t. gousset , b. pire and o.v .
teryaev , phys.rev .
lett . * 81 * , 1782 ( 1998 ) [ hep - ph/9805380 ] ; + m. diehl , t. gousset and b. pire , nucl .
* b * ( proc.suppl . ) * 82 * , 322 ( 2000 ) [ hep - ph/9907453 ] .
p. achard _ et al .
_ [ l3 collaboration ] , phys .
b * 597 * , 26 ( 2004 ) ; phys .
b * 604 * , 48 ( 2004 ) ; phys .
b * 615 * , 19 ( 2005 ) .
j. l. rosner , phys .
d * 70 * , 034028 ( 2004)[arxiv : hep - ph/0404245 ] .
achasov , s.a .
devyanin and g.n .
shestakov , phys .
b * 108 * , 134 ( 1982 ) and z.phys .
c * 16 * , 55 ( 1982 ) .
achasov , s.a .
devyanin and g.n .
shestakov , z.phys .
c * 27 * , 99 ( 1985 ) ; n. n. achasov and g. n. shestakov , sov .
* 34 * , 471 ( 1991 ) [ usp .
nauk * 161 * , 53 ( 1991 ufnaa,161n6,53 - 108.1991 ) ] . i. v. anikin , b. pire , l. szymanowski , o. v. teryaev and s. wallon , phys .
d * 71 * , 034021 ( 2005 ) [ arxiv : hep - ph/0411407 ] and phys .
d * 70 * , 011501 ( 2004 ) [ arxiv : hep - ph/0401130 ] .
r. l. jaffe and m. soldate , phys .
d * 26 * , 49 ( 1982 ) .
m. diehl , phys .
rept . * 388 * , 41 ( 2003 ) [ arxiv : hep - ph/0307382 ] .
v. m. budnev , i. f. ginzburg , g. v. meledin and v. g. serbo , phys .
rept . * 15 * ( 1974 ) 181 .
d. v. shirkov and i. l. solovtsov , phys .
lett . * 79 * , 1209 ( 1997 ) . j. h. field , nucl .
b * 168 * , 477 ( 1980 ) .
a. p. bakulev , k. passek - kumericki , w. schroers and n. g. stefanis , phys .
d * 70 * , 033014 ( 2004 ) [ erratum - ibid .
d * 70 * , 079906 ( 2004 ) ] [ arxiv : hep - ph/0405062 ] .
s. s. agaev , m. guidal and b. pire , eur .
j. c * 37 * , 457 ( 2004 ) [ arxiv : hep - ph/0403266 ] ; j. r. andersen , phys .
b * 475 * , 141 ( 2000 ) [ arxiv : hep - ph/9909396 ] .
l. maiani , f. piccinini , a. d. polosa and v. riquer , aip conf .
* 756 * ( 2005 ) 321 [ arxiv : hep - ph/0501077 ] .
b. pire , l. szymanowski and s. wallon , arxiv : hep - ph/0501155 and arxiv : hep - ph/0410108 ; + b. pire and l. szymanowski , phys .
d * 71 * ( 2005 ) 111501 .
n. y. lee , p. v. pobylitsa , m. v. polyakov and k. goeke , j. phys .
g * 27 * , l127 ( 2001)[arxiv : hep - ph/0011166 ] . | we present a theoretical estimate for the cross - section of exclusive @xmath0 and @xmath1-meson production in two photon collisions when one of the initial photons is highly virtual .
we focus on the discussion of the twist 4 contributions which are related to the production of an exotic isospin @xmath2 resonance of two @xmath3 mesons .
our analysis shows that the recent experimental data obtained by the l3 collaboration at lep can be understood as a signal for the existence of an exotic isotensor resonance with a mass around @xmath4 . |
time reversal ( tr ) is the process of recording the signal , time - reversing and re - propagating the signal .
when the signal is from a localized source the time reversed field is expected to focus on the source .
time reversal of acoustic waves has led to applications in ultrasound and underwater acoustics including brain therapy , lithotripsy , nondestructive testing and telecommunications @xcite .
an even greater potential holds for the time reversal of electromagnetic waves which is closely related to optical phase conjugation @xcite .
recently , time reversal experiments with optical and micro waves demonstrate phase conjugation of optical near field and the robustness of turbidity suppression by optical phase conjugation @xcite .
motivated by these exciting experiments , we reconsider time reversal from basic principle , discuss various tr schemes and their mutual relations , and make general observations about the focusing properties of near - field and far - field time reversal .
first , we review the principle of perfect time reversal in a closed cavity based on the green second identity and discuss the focusing properties of the perfect tr kernel ( the porter - bojarski kernel ) .
we review the well known fact that in the free space the tr focal spot of a field with source is always diffraction - limited even when all the evanescent components are recorded by time reversal mirror ( section [ sec1 ] and [ sec2 ] ) .
the perfect time reversal is `` perfect '' only for the source - free fields . on the contrary ,
any medium inhomogeneities , no matter how weak , always give rise to a nonvanishing evanescent component in the time reversed field .
next we derive four other time reversal schemes ( monopole , dipole and two mixed modes ) from the porter - bojarski kernel and analyze their behaviors with _ one _ planar trm .
we derive their mutual relationships for the paraxial wave .
for the boundary behavior , we show that for paraxial wave only the monopole tr scheme satisfies the exact boundary condition while for the spherical wave only the md - mode tr scheme satisfies the exact boundary condition .
a main interest here is the near - field focusing property of various trms .
we show that to achieve a subwavelength focusing the near - field time reversal mirror should involve dipole fields in the sense that the trm records not just the field but the normal gradient of the field and/or back - propagate the dipole field in proportion to the phase - conjugate recorded data .
the dipole trm records and re - transmits the dipole field while the mixed mode trm involves only one of the processes .
the monopole trm records and transmits only the monopole field .
the transverse resolution of the dipole trm is linearly proportional to the distance between the point source and trm .
the mixed mode trm has the similar ( linear ) behavior in three dimensions but in two dimensions the transverse resolution behaves as the square - root of the distance between the point source and trm .
the monopole trm is ineffective to focus below wavelength ( section [ sec3 ] ) .
we also point out that tr focal spot with a finite aperture is generally closer to the trm than the source point in the free space or a weakly scattering medium such as consisting of phase scatterers ( section [ sec4 ] ) .
in other words , in the absence of multiple scattering in the medium , tr as focusing or imaging method is biased . in comparison ,
the conventional matched field processor and the phase processor always produce a centered focal spot ( section [ sec4 ] ) .
lastly in section [ sec6 ] , we explain how the random effect in multiple scattering can restore the centeredness of tr focal spot on the source point as well as the stability condition for the robustness of turbidity suppression by phase conjugation @xcite .
consider a real - valed signal in the temporal fourier representation @xmath0 the real - valuedness of @xmath1 implies that @xmath2 .
therefore , time reversal is equivalent to phase conjugation of every frequency component @xmath3 ideally , tr turns a divergent wave into a convergent wave , thus focusing wave energy .
a practical way of phase conjugation within some control domain is to operate tr from the boundary of the control domain as we consider next .
consider a monochromatic scalar wave @xmath4 propagating in a medium characterized by the refractive index @xmath5 which in general varies in space .
such a @xmath4 satisfies the helmholtz equation [ helm ] u+k^2n^2u = f subject to suitable boundary conditions where @xmath6 is the source of compact , localized support .
we assume that the medium is dissipationless , i.e. @xmath7 is real - valued . in time reversal ,
the source @xmath6 emits a wave field @xmath4 which is then recorded at the boundary @xmath8 of the ( topologically open ) domain @xmath9 .
the typical domain can be one of the following three kinds : bounded with a closed boundary , infinite slab with two planar boundaries , or half space with one planar boundary ( figure [ domain ] ) .
consider a bounded domain @xmath9 first .
let both @xmath4 and @xmath10 be recorded at @xmath8 , phase - conjugated and back - propagated into the domain @xmath9 and let @xmath11 be the resulting wave field .
let @xmath12 be the green function of ( [ helm ] ) .
the time reversal principle originally proposed in @xcite is described by [ green - thm ] v(r)_d u^*(r)g(r , r)n d(r ) -_d
u^*(r)n g(r , r)d(r ) . clearly , @xmath11 is a _ source - free _ solution of the helmholtz equation in @xmath9 .
whenever sommerfeld s radiation condition [ rad ] r g(r , r)ik g(r , r ) , for all @xmath13 sufficiently far away from @xmath14 , ( [ green - thm ] ) can be easily extended to the second type of domains ( infinite slabs ) through a limiting procedure . under this condition
there is no difference between type ( i ) ( bounded ) and type ( ii ) ( infinite slab ) domains .
let @xmath15 be the time reversal mirror ( trm ) considered as a part of @xmath8 .
at each point on the boundary @xmath16 we impose either the sound soft boundary condition ( @xmath17 ) or the sound hard boundary condition ( @xmath18 ) . the boundary condition on trm
@xmath15 is flexible except that the presence of trm does not affect the near - field behavior of the green function , see ( [ near2])-([near3 ] ) .
we call either type ( i ) or ( ii ) domain of figure [ domain ] _ closed cavity_. by the second green s identity we have & & _ d u^*(r)g(r , r)n d(r ) -_d u^*(r)n g(r , r)d(r ) + & = & _ d u^*(r)(+k^2n^2(r))g ( r , r ) dr -_d ( + k^2n^2(r ) ) u^*(r ) g(r , r)dr.[green3 ] the time reversal operation described by ( [ green - thm ] ) is perfect for the field _ inside _
@xmath9 when the field is source - free inside @xmath9 ( i.e. @xmath4 satisfies ( [ helm ] ) with @xmath19 and the second integral on the right hand side of ( [ green3 ] ) drops out ) [ out2 ] v(r)= u^*(r),rd while @xmath20 by ( [ green3 ] ) .
hence ( [ green - thm ] ) has been proposed as the basis for time reversal in a closed cavity @xcite . on the other hand ,
if @xmath9 contains sources as in typical applications , then the time reversal described by ( [ green - thm ] ) is diffraction - limited . indeed , [ green4][tr - sinc ] v(r)&= & u^*(r)-_d f^*(r)g(r , r)dr + & = & _ d [ g^*(r , r)-g(r , r ) ] f^*(r)dr + & = & _ d k_pb(r , r)f^*(r)dr,rd with the porter - bojarski kernel [ pb ] k_pb(r , r)=-2 where @xmath21 denotes the imaginary part , c.f .
@xcite . with a real - valued @xmath7
, @xmath22 is an example of a source - free radiation field in @xmath14 for each @xmath13 . in the free space , ( [ pb ] ) is always diffraction - limited ( see section [ sec2 ] ) .
the reason that @xmath23 is because the time symmetry is broken when the field is phase conjugated but not the source .
a time - reversed source is a sink .
suppose now the point source becomes a point sink after emitting the initial wave and suppose the sink can absorb the time - reversed , incoming wave and prevent the re - emission of the outgoing wave with efficiency @xmath24 , @xcite . then ( [ green4 ] ) becomes [ green - sink ] v(r ) & = & g^*(r , r)-(1-)g(r , r_0),rd , which , denoted by @xmath25 , is the tr kernel with a point sink of efficiency @xmath26 .
when @xmath27 then @xmath28 .
as we shall see in section [ sec2 ] , the tr focusing property for @xmath25 is infinitely enhanced over that for ( [ pb ] ) .
in this section we derive approximate tr schemes from ( [ green - thm ] ) using sommerfeld s radiation condition ( [ rad ] ) .
[ sec3 ] for a closed cavity , using the sound soft / hard condition on @xmath29 , we write + & = & _ d [ _ a g^*(r , r)g(r , r)n d(r ) -_a g^*(r , r)n g(r , r)d(r ) ] f^*(r)dr which yields the alternative expression for the porter - bojarski kernel for a closed cavity [ pb2 ] k_pb(r , r)&=&_a g^*(r , r)g(r , r)n d(r ) -_a g^*(r , r)n g(r , r)d(r ) now we pretend @xmath30 and use condition ( [ rad ] ) to motivate two other tr schemes .
we consider all three types of domains depicted in figure [ domain ] .
replacing @xmath31 with @xmath32 or @xmath32 with @xmath33 in ( [ pb2 ] ) leads to , respectively , the monopole tr kernel [ lb ] k_m(r , r)&=&- 2ik_a g^*(r,r)g(r , r)d(r ) and the dipole tr kernel [ dp ] k_d(r , r)&=&-2ik_a n g^*(r , r)ng(r , r ) d(r ) . in addition , we introduce mixed mode tr kernels [ md ] k_md(r , r)&=&_a g^*(r , r)g(r , r)n d(r ) + k_dm(r , r)&=&-_a g^*(r , r)n g(r , r)d(r)[dm ] which are parts of ( [ pb2 ] ) . analogously we define the various tr schemes for a source - free field @xmath4 : [ lb ] v_m(r)&=&- 2ik_a u^*(r)g(r , r)d(r ) + [ dp ] v_d(r)&=&-2ik_a n u^*(r)ng(r , r ) d(r ) + [ md ] v_md(r)&=&_a u^*(r)g(r , r)n d(r ) + v_dm(r)&=&-_a u^*(r)n g(r , r)d(r)[dm ] + v_pb(r)&=&v_md(r)+v_dm(r)[pb ] the various tr schemes can be readily implemented for the acoustic wave . for the dipole trm ,
the transducers record the normal pressure gradient and emit the normal dipole field in proportion to the phase - conjugate recorded data .
for the mixed mode trm , the transducers record the pressure ( resp .
pressure gradient ) and emit the normal dipole ( resp .
monopole ) field in proportion to the phase - conjugate recorded data ( see , for example , @xcite for a similar proposal ) .
the time reversal schemes described by ( [ pb2])-([dm ] ) and ( [ lb])-([pb ] ) are main objects of subsequent analysis .
note that the expressions ( [ pb2])-([dm ] ) , ( [ lb])-([pb ] ) are source - free fields in the domain @xmath9 regardless whether the initial wave is source - free or not . in the case of paraxial wave ,
( [ pb2 ] ) , ( [ lb ] ) , ( [ md ] ) and ( [ dm ] ) are explicitly connected with one another as follows .
consider the half space @xmath34 with a planar trm on the transverse plane @xmath35 .
let @xmath36 be the paraxial green function where @xmath37 are the longitudinal coordinate and @xmath38 are the transverse coordinates ( see appendix a ) .
let @xmath39 be the location of the source .
assume that the trm is away from the medium inhomogeneities i.e. @xmath40 .
then it can be shown that [ 38 ] k_md(z , x , z_0,x_0 ) & = & 12 k_m(z , x , z_0,x_0 ) + i2k_t g^*(z_0,x_0,0,x ) _t g(z , x,0,x)dx + & = & k_dm(z , x , z_0,x_0 ) [ 39]where @xmath41 denotes the transverse gradient w.r.t .
@xmath42 ( appendix a ) .
therefore , [ mono2 ] k_md(r , r_0 ) & = & k_dm(r , r_0)= 12 k_pb(r , r_0 ) .
likewise , we have [ md]v_md(z , x)&=&12 v_m(z , x)+i2k _ t u^*(0,x)_t g(z , x,0,x)dx + & = & v_dm(z , x ) = 12 v_pb(z , x ) . though not fundamental to many time reversal applications , it is sometimes desirable for the tr schemes to satisfy the boundary condition [ bc ] k_x ( r , r_0)=g^*(r , r_0 ) , v_x(r)=u^*(r ) , ra where @xmath43 .
we have already seen that in a closed cavity @xmath44 satisfies the boundary condition but @xmath45 does not . in appendix
[ app : bc ] we show that , surprisingly , _ for the paraxial wave , only the monopole tr scheme satisfies the boundary condition ( [ bc ] ) and for the spherical wave in the free space , only the mixed mode tr scheme @xmath46 , multiplied by @xmath47 , satisfy ( [ bc])_.
in this section , we analyze the near - field focusing property of various tr schemes .
we show that @xmath48 give rise to a subwavelength focal spot of size proportional to the distance from the source to trm while @xmath49 and @xmath45 can hardly achieve subwavelength focusing . in the free space , the focal spot size for the paraxial wave is always comparable to the fresnel length @xmath50 which is not applicable for @xmath51 .
consider the porter - bojarski kernel for the free space green function in three dimensions , [ near ] k_pb & = & 2|r - r| and compare it with the real part of @xmath25 defined by ( [ green - sink ] ) [ near22 ] & = & , > 0 which dominates over the imaginary part for @xmath52 and @xmath53 .
resolution of an imaging system generally refers to either the focal spot size or the minimum resolvable distance between two points or lines ( two - point or -line resolution ) . and these two ideas are related : the focal spot size is a crude estimate of the minimum resolvable distance .
with all the insights into superresolution techniques , it is now well accepted that in the absence of noise and with perfect knowledge of the imaging system there is no limitation to the two - point resolution .
fundamentally , this is because an image with infinite signal - to - noise ratio can convey an arbitrary amount of information @xcite .
since we do not account for the noise explicitly we will focus on the idea of focal spot size as a measure of focusing and image sharpness . according to rayleigh s criterion which essentially defines the spot size as the distance of the first zeros to the maximum point , ( [ near ] ) has a spot size of @xmath54 and ( [ near22 ] ) has a spot size of @xmath55 .
but in reality the focal spot is in some sense `` in the eye of the viewer '' and one readily recognizes the vastly sharper graph of ( [ near22 ] ) than that of ( [ near ] ) .
this contrast is better described by the standard ( houston ) criterion used in astronomy which is to adopt the `` full width at half maximum '' ( fwhm ) @xcite .
the fwhm of ( [ near ] ) is about @xmath54 while the fwhm of ( [ near22 ] ) is zero since the maximum of ( [ near22 ] ) is infinity ( therefore infinitely sharp ) .
indeed , the singularity is the signature of a point source .
the subwavelength , superresolution of ( [ near22 ] ) can be understood from weyl s angular spectrum of plane waves . for the imaging direction @xmath56 , the free space green function in three dimensions has the angular spectrum representation = ik2 [ weyl ] where r&= & + s_3&=&,s_1 ^ 2+s_2 ^ 21 + s_3&=&i , s_1 ^ 2+s_2 ^ 2>1 @xcite .
the integrand in ( [ weyl ] ) with @xmath57 corresponds to the propagating ( homogeneous ) wave and that with @xmath58 corresponds to the evanescent ( inhomogeneous ) wave .
let @xmath59 and @xmath60 denote the homogeneous and the evanescent fields , respectively . since @xmath61 is the transverse wavenumber
, @xmath59 contains the diffraction - limited information about the point source while @xmath60 contains the sub - wavelength information about the point source .
the first important observation is that @xmath60 is always real - valued .
consequently , the tr kernel @xmath45 in the case of @xmath62 contains no evanescent component .
secondly , @xmath59 is bounded and in the @xmath63-neighborhood of the source point the estimate holds : @xmath64|=o(\delta)$ ] .
in other words , in the immediate vicinity of the source point @xmath65 $ ] essentially coincides with @xmath60 .
this means that @xmath65 $ ] contains all the subwavelength information about the point source .
therefore , ( [ tr - sinc ] ) is always diffraction - limited no matter how narrow the support of @xmath6 is and how close the source is to the trm .
this means that in the absence of scattering even when the evanescent waves are recorded , phase - conjugated and back - propagated , the resulting field is still smeared below the scale of wavelength . in general ,
if @xmath66|\gg |\im[g]|$ ] in the near field as is the case for the free space , the latter does not contribute to the leading order effect of the near field tr and hence that tr , which affects only the phase , has a secondary effect on the subwavelength resolution of the near field tr .
since we are concerned here with the asymptotic behavior of near field tr as the point source approaches the trm , we relax the definition of fwhm . instead of `` full width at half maximum '' in the literal sense
we also refer to the ( asymptotic ) resolution as the asymptotic size of a focal spot as the maximum width at the same ( or the minimum width at a slower ) asymptotic behavior of the point - spread function .
for example , in this metric , the focal spot size for ( [ near ] ) is @xmath54 while the truncated version of ( [ near22 ] ) : [ cutoff ] \ { 1 , } , 1 has an asymptotic spot size @xmath67 .
next we shall analyze the near - field focusing property of the various tr schemes .
we assume the near field asymptotic [ near2 ] |g(r , r)| & ~&1|r - r|,|r - r| . for @xmath68 ( [ near2 ] )
is the generally correct asymptotic as the near field singularity is determined solely by the laplacian in the helmholtz equation @xcite .
in addition to ( [ near2 ] ) we further assume [ near3 ] g(r , r ) & ~&r - r|r - r|^3 , |r - r| .
consider the case of a point monopole source located at @xmath69 in close proximity of @xmath15 .
let @xmath70 be the closest point on @xmath15 to @xmath69 and @xmath71 . using ( [ near2 ] )
we have the estimate @xmath72 since the singularity in the integrand of ( [ lb ] ) is [ sing1 ] k|r-r_0||r - r|,|r - r_0| , |r-r_0| .
more explicitly , the integral ( [ lb ] ) can be simplified by replacing the surface integral over @xmath15 by the tangent plane at @xmath73 . without loss of generality
, we may assume the tangent plane to be the @xmath74 plane and @xmath75 .
the leading order transverse profile of @xmath49 around @xmath69 in @xmath76 is given by [ 20 ] kdxdy,1 where the integration is restricted to a bounded area , say the circular disk of radius @xmath77 , around the origin to avoid logarithmic divergence at large @xmath78 .
let @xmath79 .
then ( [ 20 ] ) has the same asymptotic as [ 20 ] kdxdy,1 where the integration is over the punctured disk with holes of radius @xmath63 centered at @xmath80 and @xmath81 since the two holes have a vanishing contribution to ( [ 20 ] ) .
the change of variables from @xmath78 to @xmath82 renders ( [ 20 ] ) into [ 20 ] kdxdy integrated over the punctured disk of radius @xmath83 with shrinking holes of radius @xmath84 centered at @xmath80 and @xmath85 . hence [ 41 - 2 ] k_m(r , r_0)~ \
{ -k,&(0,1 ) + k,&=0 .,1 where the logarithmic divergence for @xmath86 is due to the integration ( [ 20 ] ) on the scale of @xmath87 .
when @xmath63 is so small that @xmath88 is the dominant behavior of @xmath49 near the source point , ( [ 41 - 2 ] ) says that at @xmath89 , @xmath49 is roughly at half the maximum and therefore the transverse resolution is proportional to @xmath90 .
however , this scaling behavior requires extremely small @xmath63 to manifest itself . outside this regime
, the focal spot size would decrease slowly as @xmath63 decreases ( see the comments on figure [ fig:3d ] ) . ; ( right ) 3d longitudinal profiles with fwhm @xmath91 note that the transverse profiles of the @xmath92 and @xmath93 coincide . ,
title="fig:",width=302 ] ; ( right ) 3d longitudinal profiles with fwhm @xmath91 note that the transverse profiles of the @xmath92 and @xmath93 coincide .
, title="fig:",width=302 ] for @xmath94 and @xmath95 , @xmath96 is orthogonal to @xmath15 and thus @xmath97 .
therefore the leading order transverse profile for @xmath98 is given approximately by [ 24 ] 1kdxdy. the integrand in ( [ 24 ] ) has a different singularity than that in ( [ 20 ] ) .
indeed , for @xmath99 , the dominant contribution to ( [ 24 ] ) comes from the disks of radius @xmath63 around @xmath80 and @xmath81 which is easily estimated to be @xmath100.\ ] ] by contrast , @xmath101 .
hence the focal spot size in this case is smaller than @xmath102 asymptotically .
we conclude that the focal spot size for the dipole tr is essentially @xmath67 and that the dipole tr can in principle break the diffraction limit to achieve an arbitrarily fine resolution by bring the trm sufficiently close to the source . by the same analysis as before , the integrals ( [ md ] ) and ( [ dm ] )
can be approximated by , respectively , [ 26 ] & & dxdy + & & -dxdy[27 ] .
set @xmath79 . in the case of ( [ 26 ] )
, the dominant contribution ( @xmath103 ) comes from integration around @xmath81 for @xmath104 while , in the case of ( [ 27 ] ) , the dominant contribution ( @xmath103 ) comes from integration around @xmath80 . in both cases @xmath105 .
hence the focal spot size is smaller than @xmath106 asymptotically .
we conclude that the focal spot size for the mixed mode tr is essentially @xmath67 and that the mixed mode tr can achieve an arbitrarily fine resolution as the trm approaches the point source .
moreover , the fact that the maximum values of the point - spread functions are inversely proportional to the focal spot size is consistent with the cut - off version of ( [ near22 ] ) in the following sense .
truncate the green function ( [ near2 ] ) at the level @xmath107 as in ( [ cutoff ] ) .
then the asymptotic spot size of the resulting function is @xmath67 . in other words ,
the mixed mode tr of near field recovers qualitatively the asymptotic behavior of the green function near the source point . when the superresolution terms from @xmath93 and @xmath92 combine in @xmath45 , they cancel each other resulting in diffraction - limited focal spot . in figure [ fig:3d ]
the normalized transverse and longitudinal profiles are shown and their resolutions calculated for various tr schemes .
the profiles are normalized to be unity at the source location . in the simulation ,
we use @xmath108 , @xmath109 and @xmath110 . before normalization
, the @xmath49 curve has the maximum less than @xmath111 and hence is far from the square - root asymptotic regime discussed in the paragraph following ( [ 41 - 2 ] ) . in the right plot
demonstrates the square - root behavior of @xmath112 in two dimensions . , title="fig:",width=302 ] in the right plot demonstrates the square - root behavior of @xmath112 in two dimensions .
, title="fig:",width=302 ] it is instructive to compare the scaling behavior of resolution in two and three dimensions . in two dimensions , ( [ 24 ] ) becomes [ 24 ] 1kdx. the same calculation shows that @xmath113\ ] ] and @xmath114 .
the @xmath115-resolution asymptotic is given by @xmath116 .
for the mixed mode , ( [ 26 ] ) and ( [ 27 ] ) become [ 26 ] & & 1kdx + & & [ 27 ] -1kdx yielding @xmath117\ ] ] and @xmath118 . hence , the asymptotic resolution for the mixed mode tr is given by @xmath119 .
empirically , the square - root regime for the mixed mode tr in two dimensions sets in much earlier than the monopole tr in three dimensions . here lies the main difference between the two and three dimensional cases .
namely , in two dimensions the transverse resolution for the mixed mode tr is proportional to the _ square - root _ of the distance between the source and trm . for the monopole tr in two dimensions , @xmath49 is uniformly bounded w.r.t . @xmath63 and
the resolution can not be deduced from the asymptotic analysis .
it is likely though that the monopole tr resolution is qualitatively independent of @xmath63 .
this is amply confirmed by our numerical simulations ( figure [ fig:2d - res ] ) . in figure [ fig:2d - res ] the transverse resolutions , denoted by @xmath120 , for the respective tr schemes as a function of the distance between the source and trm are plotted for the range @xmath121 $ ] .
the dipole resolution @xmath122 exhibits the linear behavior in two and three dimensions and the mixed mode resolution @xmath123 exhibits the linear behavior in three dimensions and square - root behavior in two dimensions .
both the dipole and mixed mode resolutions are order of magnitude better than the monopole resolution . in figure [ fig:2d ]
the normalized transverse and longitudinal profiles in two dimensions are shown and fwhm calculated for various tr schemes with the same physical parameters as in figure [ fig:3d ] .
the unnormalized transverse profile for the mixed mode tr has the maximum about 6 .
the dipole resolution is much better than the mixed mode resolution which in turn is much better than the monopole resolution in both the transverse and longitudinal directions . ; ( right ) 2d longitudinal profiles : fwhm@xmath124 note that the transverse profiles for the @xmath92 and @xmath93 coincide .
, title="fig:",width=302 ] ; ( right ) 2d longitudinal profiles : fwhm@xmath124 note that the transverse profiles for the @xmath92 and @xmath93 coincide .
, title="fig:",width=302 ] in summary , in the case of a monopole source the dipole tr produces the best resolution in both transverse and longitudinal directions , roughly in proportion to the distance between the source and trm .
the mixed mode tr can deliver nearly the same performance in three dimensions and roughly the square - root resolution in the transverse direction in two dimensions . between the two mixed mode tr schemes , @xmath92 is still better than @xmath93 in the longitudinal direction .
a simple explanation for the superior resolution of trm involving the dipole field is that a dipole source has a stronger singularity than a monopole source ( cf .
( [ near3 ] ) versus ( [ near2 ] ) )
we have already seen that the perfect tr described by ( [ pb ] ) with @xmath62 is diffraction - limited and contains no evanescent component of the initial outgoing field ( see the remark following ( [ weyl ] ) ) . for the half space ( domain ( iii ) of figure [ domain ] ) , the result is similar : suppose the initial wave field is perfectly phase - conjugated on the trm
( i.e. the boundary condition ( [ bc ] ) is satisfied exactly by , say , the scheme @xmath92 ) .
then in the interior of the half space the homogenous components of the initial wave are perfectly phase - conjugated but the evanescent components are exponentially damped by the factor @xmath125 depending on the distance @xmath56 from the trm to the field point @xmath126 ( appendix [ sec : be ] ) .
on the other hand , for the paraxial wave ( [ para ] ) , perfect phase conjugation on one plane implies perfect phase conjugation in the half space due to the fact that the paraxial wave field is uniquely determined by the direction of propagation and the initial condition on one plane ( appendix [ sec : be ] ) . as a consequence , @xmath127 and @xmath128 are perfect tr schemes in their respective contexts .
in particular , @xmath49 can focus perfectly back on the location of the point source in the paraxial approximation .
the perfect focusing of time reversed paraxial wave is , of course , unphysical because the evanescent waves are not accounted for .
the other aspect that is more impractical than unphysical with the perfect conjugation of the homogeneous component of the spherical wave , is that an infinite aperture ( @xmath129 ) is required for such a result . in this section ,
we show that when @xmath9 is a half space and @xmath15 is bounded , the maximum point of the tr focal spot does _ not _ correspond to the exact source location @xmath69 in general .
in contrast , when @xmath129 in the paraxial regime , since @xmath49 is a perfect tr scheme for the paraxial wave , the focal spot is always centered .
this bias effect tends to be much more pronounced in the longitudinal direction . as the aperture of trm increases , the transverse offset ( bias ) decreases while the longitudinal offset may persist if there is insufficient scattering in the medium , cf .
section [ sec6 ] .
this is why we consider only the transverse resolution in the previous section .
let us recast the problem in more general terms .
let @xmath130 be the phase of @xmath131 and define [ ams ] p_m(r ) & = & |_a g^*(r,r_0 ) e^i(r , r ) w(r , r ) d(r)|^2 where @xmath132 is a non - negative function .
the monopole tr kernel @xmath49 corresponds to @xmath133 .
direct calculation of @xmath134 yields + & = & _ aa |g^*(r,r_0)||g(r,r_0)| w(r_0 , r ) w(r_0 , r ) ( i(r_0,r)-i(r_0,r))d(r)d(r ) + & & + _ aa |g^*(r,r_0)||g(r,r_0)| ( w(r , r ) w(r , r))_r = r_0 d(r)d(r ) .
the first term vanishes due to anti - symmetry with respect to @xmath135 .
consider the case of a planar trm on the plane @xmath35 with the paraxial wave propagating in the @xmath56-direction ( appendix a ) .
suppose the medium inhomogeneities act like a phase object affecting only the phase of the free propagator .
namely , @xmath36 takes the form [ phase - obj ] g(z , x , z,x)=- 14(z - z ) e^ik(z - z ) e^ik|x - x|^22(z - z)e^i(z , x , z,x),x , x^2 . note that for ( [ phase - obj ] ) the expression ( [ ams ] ) is divergent and thus invalid at @xmath95 if @xmath129 . for a finite aperture , @xmath136 and thus @xmath137 points in the negative @xmath56 direction and
consequently @xmath69 tends to be farther away from @xmath15 than the maximum point of @xmath138 . in other words , the focal spot of
the monopole tr tends to be closer to @xmath15 than the source location . even in the case of a closed cavity the offset ( bias )
is reduced but not completely eliminated except for special cases .
now let us change the perspective and think of tr as a means of imaging a point source in a perfectly known medium . in this context
a few widely used @xmath139 s are : w(r , r)&=&|g(r , r)|^-1 , + w(r , r)&=&w(r)/,wl^2(a , d ) , + w(r , r)&=&|g(r , r)|/. for the inverse filter , @xmath140 and thus @xmath134 tend to point away from the tr mirror @xmath15 and hence the source location tends to be closer to @xmath15 than the maximum point of @xmath138 . again , when @xmath141 or in the far field regime this offset can be reduced .
direct substitution shows that @xmath142 for the two latter choices of @xmath143 .
moreover , @xmath69 is the maximum point of @xmath138 for both choices of @xmath143 . in the case of the @xmath143-weighted phase processor
, we clearly have @xmath144 in the case of the conventional matched field processing , it is well known @xcite that [ mfp ] v(r_0 , r)=g(r_0,r)/ is the solution to the optimization problem : maximize the quantity |_a g^*(r,r_0 ) v(r ) d(r)|^2 over @xmath145 subject to the constraint [ const ] @xmath146 is maximized at @xmath95 with the conventional matched field processing . in summary , as an imaging method
both tr and the inverse filter are generally biased ; the former tends to underestimate the range of the source while the latter tends to overestimate it .
the conventional matched field processor is unbiased and produces the optimal signal - to - noise ratio .
the phase processor , which uses only the phase of the green function , is a simple alternative to the conventional matched field processor .
moreover , in the regime described by ( [ phase - obj ] ) , @xmath147 is a function of the distance between the source point and the trm . in such a case
the conventional matched field processor coincides with the phase processor with @xmath148 independent of @xmath13 .
similar conclusions can be drawn for the pure dipole case with [ ams - dipole ] p_d(r ) & = & |_a g^*(r,r_0)n e^i(r , r ) w(r , r ) d(r)|^2 where @xmath149 is the phase of @xmath150 .
@xmath151 can be analyzed as above with @xmath152 replaced by its normal derivative . for @xmath153 ,
consider the paraxial regime for which ( [ 38 ] ) is valid .
each term on the right hand side of ( [ 38 ] ) can be analyzed as before .
the result is that the mixed mode tr s are biased , especially in the longitudinal direction , just like the monopole and dipole tr s .
, and longitudinal ( right ) , at @xmath154 , profiles of the intensity of the @xmath155 green function at wavelength @xmath156 with 1000 particles randomly distributed in @xmath157\times [ 0,5000]$ ] and source ( not shown ) at @xmath158 whose @xmath159 is @xmath156 .
( reproduced from @xcite ) , title="fig:",width=283 ] , and longitudinal ( right ) , at @xmath154 , profiles of the intensity of the @xmath155 green function at wavelength @xmath156 with 1000 particles randomly distributed in @xmath157\times [ 0,5000]$ ] and source ( not shown ) at @xmath158 whose @xmath159 is @xmath156 .
( reproduced from @xcite ) , title="fig:",width=321 ] in this section , we explain how randomness can make the tr focal spot centered at the location of the source point .
consider typical profiles of the green function in a multiple scattering medium as depicted in figure [ green - graph ] .
a main feature of the intensity profiles is the rapid fluctuations in the transverse and longitudinal directions .
these fluctuations upon differentiation and averaging ( integration over @xmath160 ) as in the expression = 12_aa |g(r , r ) g(r , r)|^2_r = r_0 d(r)d(r ) [ rand ] essentially cancel one another .
more precisely , the degrees of scattering and randomness can be measured by the spatial spread of the propagation channel .
if the spatial spread increases to infinity , then the trm acquires an _ effective _ aperture that is proportional to the spatial spread @xcite .
then , as per discussion at the beginning of section [ sec4 ] , the accuracy of approximate tr s improve and the focal spot becomes more centered and sharper .
moreover , counterintuitively , the tr image can avoid statistical fluctuations if the medium is sufficiently random .
let us recall some analytical results from the literature .
consider tr in a random medium occupying a half space with trm on the planar boundary .
when the random channel is the rayleigh fading regime ( i.e. obeying zero - mean gaussian statistics ) with a divergent spatial spread the monopole tr ( [ lb ] ) with trm s aperture much larger than the coherence length @xmath161 will focus on the source point with a spot size @xmath162 @xcite .
the conventional coherence theory suggests that the coherence length is of the order of wavelength @xcite .
however , spatial correlations of much smaller extent has been recently measured in the near field of random media @xcite .
this effect in conjunction with the preceding theory may be the key to understanding the subwavelength focusing observed in the time reversal experiment reported in @xcite .
the result of stability and focusing is a special case of what was originally established for the setting of multiple frequencies and multiple source points @xcite . with straightforward modification of the arguments in @xcite
the same result can be extended to any tr of the types ( [ lb])-([dm ] ) in the paraxial regime , the only difference being in the asymptotic shape of the focal spot .
experiments show consistently similar results on stability and focusing properties @xcite
. in conclusion , random fluctuations in the intensity of green s function is a necessary condition for the desired focusing property of time reversal with a finite aperture .
we have reexamined various time reversal schemes that can be employed from the boundary of a domain ; they involve monopole and/or dipole fields .
we have explored their relationships in the paraxial regime .
for the boundary behavior , we show that for paraxial wave only the monopole tr scheme satisfies the exact boundary condition while for the spherical wave only the md - mode tr scheme satisfies the exact boundary condition .
we have also shown that for the paraxial wave the standard monopole trm produces the perfect result in the entire domain but for the near field the monopole trm can hardly achieve subwavelength focusing . on the other hand ,
the dipole trm is approximate in the paraxial regime but is capable of producing a focal spot size linearly proportional to the distance from trm to the source location , thus focusing on a subwavelength scale when the trm is sufficiently close to the source point .
the mixed mode trm has the similar ( linear ) behavior in three dimensions but the square - root asymptotic in two dimensions .
the monopole trm is , if possible at all , ineffective to focus below wavelength .
we have seen how two mixed mode effects combine in the so called perfect tr scheme for a closed cavity and cancel their respective subwavelength focusing property .
the perfect trm is `` perfect '' only for a source - free initial field .
we have examined an often neglected effect of biased focusing associated with trm of a finite aperture in a weak- or non - scattering medium .
this effect pertains to all five tr schemes discussed in the paper . in a multiply scattering medium ,
the bias is greatly reduced and the focal spot becomes centered at the source location .
the removal of bias by random scattering is attributed to the enlarged effective aperture and the random fluctuations in the _ intensity _ of the green function .
let @xmath163 be the longitudinal coordinate and @xmath164 the transverse coordinates .
let @xmath165 be the green functions for the paraxial wave equation [ para ] i+_t+k2(n^2(z , x)-1)=0 where @xmath166 is the transverse laplacian .
the plus sign represents positive @xmath56 propagating wave and minus sign negative @xmath56 propagating wave .
the full green function @xmath36 in the paraxial regime can be expressed as [ phase2 ] g(z , x , z,x ) & = & ie^ik(z - z)2k g^-_p(z , x , z,x),z > z + g(z , x , z,x ) & = & ie^ik(z-z)2k g^-_p(z , x , z,x),z
< z let @xmath34 , the plane @xmath35 be occupied by the trm and @xmath39 be the location of the source .
assume that the trm is away from the medium inhomogeneities i.e. @xmath167 .
then @xmath92 can be written as [ a1 ] k_md(r , r_0)&=&-ie^ik(z - z_0)4kg_p^-*(0,x , z_0,x_0 ) g^-_p(0 , x , z , x)dx + & & + e^ik(z - z_0)4k^2 g_p^-*(0,x , z_0,x_0 ) z g^-_p(z,x,z , x)|_z=0dx. by the assumption @xmath40 and ( [ para ] ) , we have k_md(r , r_0)&=&-ie^ik(z - z_0)4kg_p^-*(0,x , z_0,x_0 ) g^-_p(0,x , z , x)dx + & & + ie^ik(z - z_0)8k^3 g_p^-*(0,x , z_0,x_0 ) _t g^-_p(0,x , z , x)dx + & = & -ikg^*(0,x , z_0,x_0 ) g(0,x , z , x)dx + & & + i2k _t g^*(0,x , z_0,x_0)_t g(0,x , z , x)dx after integration by parts . the result for @xmath93
can be likewise derived .
let us first focus on the boundary values of ( [ lb])-([dm ] ) .
let @xmath168 be a paraxial wave propagating in the negative @xmath56 direction which can be written as @xmath169 where @xmath170 is the solution of eq .
( [ para ] ) with the minus sign . in the paraxial regime with @xmath167
, we have from ( [ phase2 ] ) and ( [ para ] ) that [ lb ] v_m(0,x)&= & u^*(0,x)g^+_p(0,x,0,x)dx + [ dp ] v_d(0,x)&=&-ik _ t u^*(0,x)g^+_p(0,x,0,x)dx -i2k^3 _ t^2
u^*(0,x ) g^+_p(0,x,0,x)dx + [ md ]
v_md(0,x)&=&12 u^*(0,x)g^+_p(0,x,0,x)dx + 14k^2 _ t u^*(0,x)g^+_p(0,x,0,x)dx + v_dm(0,x)&=&12 u^*(0,x)g^+_p(0,x,0,x)dx + 14k^2 _ t u^*(0,x)g^+_p(0,x,0,x)dx [ dm ] where @xmath4 is a paraxial wave propagating in the negative @xmath56 direction .
setting @xmath171 , we obtain the following boundary conditions [ bc - para1 ] v_m(0,x)&= & u^*(0,x ) + v_d(0,x)&= & -ik _ t u^*(0,x)-i2k^3 _
t^2 u^*(0,x)[bc - para2 ] + v_md(0,x)&= & 12 u^*(0,x)+14k^2 _
t u^*(0,x)[bc - para3 ] + v_dm(0,x)&=&12 u^*(0,x)+14k^2 _
t u^*(0,x ) [ bc - para4 ] + v_pb(0,x)&= & u^*(0,x)+12k^2 _ t u^*(0,x)[bc - para5 ] .
the kernel @xmath172 satisfy the similar boundary properties . in other words
, only the monopole tr gives the correct boundary values in the case of paraxial wave propagating in the half space .
consider the half space @xmath34 with @xmath173 .
assume @xmath62 , the free space green function .
differentiating the weyl representation for @xmath174 with respect to @xmath175 , we find that [ a32 ] zg_0(z , x , z,x).|_z = z=0=12 ( x - x ) . from ( [ a32 ] )
it is immediately clear that @xmath176 however , neither @xmath177 nor @xmath93 satisfies the boundary condition as we show now . in the neighborhood of @xmath15 , the wave field @xmath4 in @xmath9
can be represented in terms of the plane wave spectrum : [ plane - wave ] u(x , y , z ) & = & e^ik(s_1x+s_2y ) e^iks_3z a(s_1,s_2 ) + e^ik(s_1x+s_2y ) e^-iks_3z b(s_1,s_2 ) where @xmath178 is given in ( [ weyl ] ) .
consider for simplicity one - way wave with , say , @xmath179 .
then the phase - conjugated field has the form u^*(x , y , z ) & = & _ s_1 ^ 2+s_2 ^ 21 ds_1ds_2 e^ik(s_1x+s_2y ) e^-iks_3z a^*(-s_1,-s_2 ) + & & + _
s_1 ^ 2+s_2 ^ 2 > 1 ds_1ds_2 e^ik(s_1x+s_2y ) e^iks_3z a^*(-s_1,-s_2)and thus u^*(x , y,0 ) & = & ik e^ik(s_1x+s_2y)s_3 a^*(-s_1,-s_2)[-i_\{s_1 ^ 2+s_2 ^ 21}+i_\{s_1 ^ 2+s_2 ^ 2>1}].[a20 ] where @xmath180 is the indicator function of the set @xmath181 . by plancherel s identity we obtain v_dm(0,x)&= & 2 e^ik(s_1x+s_2y ) a^*(-s_1,-s_2)[-i_\{s_1 ^ 2+s_2 ^ 21}+i_\{s_1 ^ 2+s_2 ^ 2>1 } ] + & & 2u^*(0,x ) . namely , no constant multiple of @xmath177 satisfies the boundary condition .
likewise , one can show that none of the other tr schemes than @xmath182 satisfies the boundary condition , with some adjustable constant factor .
consider the half space @xmath34 with @xmath173 and point source located at @xmath183 , figure [ fig : half - space ] .
we assume that the medium inhomogeneities @xmath159 is away from the trm so that the gap @xmath184 is a non - empty infinite slab .
figure [ fig : half - space ] .
let @xmath185 be the initial wave propagating in the negative @xmath56 direction where @xmath186 satisfies ( [ para ] ) with the minus sign .
if @xmath187 is perfectly conjugated by , say , either the scheme described by @xmath127 or @xmath44 , then the resulting field , denoted by @xmath188 , is given by @xmath189 where @xmath190 satisfies ( [ para ] ) with the plus sign and the initial condition @xmath191 now since @xmath192 satisfies ( [ para ] ) with the plus sign and the initial condition @xmath193 , we have from the uniqueness theorem for the initial value problem of ( [ para ] ) that @xmath194 and consequently @xmath195 the same argument and conclusion apply to the paraxial wave with a point source .
the case with the spherical wave is similar but more subtle since the evanescent wave is involved .
it can be shown that the green function in @xmath196 can be expressed as superposition of plane waves propagating in the negative @xmath56 direction g(r , r_0)= e^ik(s_1x+s_2y ) e^-iks_3z b(s_1,s_2),rd^-\{z < z_0 } [ 62 ] in analogy to ( [ plane - wave ] ) .
let @xmath143 be a source - free field propagating across @xmath35 into @xmath9 and be represented as w(r)= e^ik(s_1x+s_2y ) e^iks_3z a(s_1,s_2),rd^- .
[ 61 ] note the sign in the exponent associated with @xmath56 here is different from that in ( [ 62 ] ) since they propagate in the opposite directions .
suppose @xmath143 is the phase conjugate field of @xmath152 at @xmath35 , i.e. @xmath198 for @xmath35 , equating ( [ 61 ] ) with ( [ 62 ] ) conjugated we obtain the condition a(s_1,s_2)&= & b^*(-s_1,-s_2 ) .
[ evan - pc ] this result is first observed in @xcite .
let @xmath199 and @xmath200 be , respectively , the homogeneous and the evanescent components of @xmath143 , corresponding to , respectively , the integration ( [ 61 ] ) restricted to @xmath57 and @xmath201 .
the homogeneous and the evanescent components @xmath202 of ( [ 62 ] ) are defined analogously . then ( [ evan - pc ] ) lead to , respectively , w_h^*(r)&=&g_h(r),rd^-
+ w_i^*(r)&= & e^ik(s_1x+s_2y ) e^-kz b(s_1,s_2),rd^- . in other words ,
the homogenous components of the initial wave is perfectly phase - conjugated but the evanescent components are exponentially damped by the factor @xmath203 depending on the distance from the trm to the field point .
d. cassereau and m. fink , _ ieee trans .
. control _ * 39 * ( 1992 ) 579 .
clemmow , _ the plane wave spectrum representation of electromagnetic fields _ , pergamon press , oxford , 1966 .
w. a. kuperman and d. r. jackson `` ocean acoustics , matched - field processing and phase conjugation '' in _ imaging of complex media with acoustic and seismic waves _ , pp .
43 - 97 , springer berlin / heidelberg , 2002 . | the concept of time reversal ( tr ) of scalar wave is reexamined from basic principles .
five different time reversal mirrors ( trm ) are introduced and their relations are analyzed .
for the boundary behavior , it is shown that for paraxial wave only the monopole tr scheme satisfies the exact boundary condition while for the spherical wave only the md - mode tr scheme satisfies the exact boundary condition . the asymptotic analysis of the near - field focusing property is presented for two dimensions and three dimensions .
it is shown that to have a subwavelength focal spot the trm should consist of dipole transducers .
the transverse resolution of the dipole trm is linearly proportional to the distance between the point source and trm .
the mixed mode trm has the similar ( linear ) behavior in three dimensions but in two dimensions the transverse resolution behaves as the square - root of the distance between the point source and trm .
the monopole trm is ineffective to focus below wavelength .
contrary to the matched field processing and the phase processor , both of which resemble tr , tr in a weak- or non - scattering medium is usually biased in the longitudinal direction , especially when tr is carried out on a _ single _ plane with a finite aperture .
this is true for all five tr schemes . on the other hand ,
the tr focal spot has been shown repeatedly in the literature , both theoretically and experimentally , to be centered at the source point when the medium is multiply scattering .
a reconciliation of the two seemingly conflicting results is found in the random fluctuations in the intensity of the green function for a multiply scattering medium and the notion of scattering - enlarged effective aperture . |
in two previous papers ( buta & block 2001 , hereafter bb01 ; block et al . 2001 ) , we outlined a new approach to quantifying the observed bar strengths of galaxies . instead of relying on deprojected bar ellipticities
, we used a theoretical equation ( combes & sanders 1981 ) based on the forcing of the bar implied by the near - infrared light distribution .
this method , called the relative bar torque , or @xmath1 , method , was applied to 36 galaxies by bb01 and later to nearly 40 more galaxies by block et al .
the method has also been applied to more than 100 two micron all sky survey ( 2mass ) galaxies by laurikainen , salo , & rautiainen ( 2002 ) and by laurikainen & salo ( 2002 ) , who also refined the method .
most recently , block et al .
( 2002 ) applied the @xmath1 method to more than 150 galaxies in the ohio state university bright galaxy survey ( osubgs , eskridge et al .
2002 ) and used the results to show that normal galaxies may double their mass by accretion in 10@xmath7 years .
a difficulty with the @xmath1 method as applied in these previous studies is that the bar strengths based on the method could be affected by spiral arm torques .
@xmath1 represents the maximum ratio of the tangential force to the mean axisymmetric radial force . if the bar is a typical sb - type bar ( rather than an oval )
, then this maximum will usually be a good approximation to the bar strength and the force ratio map will show a characteristic butterfly pattern ( bb01 ) . however , many barred spirals have pronounced spiral arms that break directly from the ends of the bar , and these arms can affect the bar butterfly pattern and increase the apparent bar strength .
the original intent of bb01 was that @xmath1 should measure bar strength , not a combination of bar and spiral arm strength .
there are good reasons for trying to find a way to separate the effects of the spiral from the bar .
first , if we want to investigate scenarios of bar formation in disk galaxies ( e.g. , sellwood 2000 ) , then we should have measures of bar torques , not spiral plus bar torques .
second , with a separation analysis , we can check theoretical predictions that bars with larger torques drive spirals with higher amplitudes ( elmegreen & elmegreen 1985 and references therein ; also see yuan & kuo 1997 and references therein ) . in this paper , we outline a straightforward method of separating the bars from the spirals using 2.1@xmath3 m near - infrared images .
the method uses fourier techniques in conjunction with a simple assumption of how to extrapolate the bar into the spiral - dominated regions .
the method works effectively even for the strongest bars with the strongest spirals . for the first time
, we can investigate spiral arm torques in galaxies and even define spiral torque classes in the same manner as bb01 defined bar torque classes .
we illustrate the method using a representative example , the sab(rs)bc spiral ngc 6951 .
the image we use is a @xmath8-short , or @xmath9 , image obtained during a run with the 4.2-m william herschel telescope ( wht ) in 2001 .
the image scale is 024 per pixel and the field of view is 4@xmath101 square .
a total of 15 spiral galaxies was observed for investigating bar and spiral torques .
full details of these observations , and analysis of the remaining galaxies , will be provided in block et al .
( 2003 , hereafter paper ii ) where the techniques described in this paper will be applied to examine the relations between bars and spirals .
the @xmath1 method was fully described in bb01 .
the dimensionless parameter @xmath1 can be interpreted as the maximum gravitational bar torque per unit mass per unit square of the circular speed . to derive it , we process a near - infrared image by removing all foreground stars , and then deproject the image using available orientation parameters . for ngc 6951
, we used a mean position angle @xmath11 and axis ratio @xmath12 based on isophotal ellipse fits on the 2.1@xmath3 m image itself .
the values used were @xmath12 = 0.773 and @xmath11 = 143@xmath131 .
these are in good agreement with optical photometric estimates of the same parameters from mrquez & moles ( 1993 ) .
iraf routine imlintran was used for the deprojection . to facilitate our analysis ,
we have rotated the deprojected image such that the bar axis is horizontal . the bar position angle in the raw deprojected image
was measured using ellipse fits , and iraf routine rotate was used for the final rotation .
the deprojected image was then centered within an array of dimension 2@xmath14 , where @xmath15=10 for the wht image , and run through a program which transforms the near - infrared image into a two - dimensional potential ( quillen , frogel , & gonzlez 1994 , hereafter qfg ) . from the 2d potential , planar forces are calculated and then decomposed into radial and tangential components . our main analysis is based on maps of the ratio of the tangential force to the mean axisymmetric radial force , the latter derived from the @xmath0=0 component of the potential .
we assume a constant mass - to - light ratio , but in addition to the force ratios , we also compute a predicted axisymmetric rotation curve in order to evaluate the correctness of this assumption , especially in the spiral arm regions .
we have also made a refinement to our use of the qfg potential method , based on a study by laurikainen & salo ( 2002 ) , who noted that the qfg convolution integral for the vertical dimension included some gravity softening .
we use a revised lookup table from h. salo ( private communication ) for an exponential vertical density profile without softening .
laurikainen & salo ( 2002 ) showed that the relative bar torques of bb01 are too low by about one bar class because of this softening .
the computation of a potential from a near - infrared image requires a value for the vertical scaleheight , which can not be directly measured for ngc 6951 . in bb01 and
block et al .
( 2001 ) , it was assumed that all galaxies had the same vertical exponential scaleheight as our galaxy , @xmath16 = 325pc .
however , this approach required knowledge of the distance to each galaxy , which had to be based on radial velocities . here
we derive @xmath16 by scaling a value from the radial scalelength , @xmath17 .
as shown by de grijs ( 1998 ) , the ratio @xmath17/@xmath16 ( based on mostly @xmath18-band and some @xmath8-band surface photometry ) depends on hubble type , being larger for later types compared to earlier types . for ngc 6951
, we estimated a radial exponential scalelength using an azimuthally - averaged @xmath9 surface brightness profile .
the slope of the outer light profile provides an approximation to a radial scalelength , which we obtained to be @xmath17 = 33 . from a bulge / disk decomposition of an @xmath18-band luminosity profile , mrquez & moles ( 1993 )
obtained a disk effective radius of 42@xmath1976 for ngc 6951 , which corresponds to a radial exponential scalelength of 25@xmath195 .
this is in reasonable agreement with our estimate . for an sbc spiral
, de grijs s analysis shows that @xmath17 = ( 6@xmath62)@xmath16 on average . for a redshift distance of 24.1 mpc ( tully 1988 ) for ngc 6951 , this gives @xmath16 = 640pc , about twice the galactic value .
the average value of @xmath16 in de grijs s sample is 600@xmath6400pc .
we found it necessary to remove some of the very strong star - forming regions from the near - infrared image of ngc 6951 before the potential was calculated .
these objects can cause local maxima or minima in the force ratio maps that may be unreliable if the mass - to - light ratios of these regions differ from the dominant old stellar background .
it is well - known that hii regions and luminous red supergiants impact the 2.2@xmath3 m spectral region ( e.g. , knapen et al .
1995 ) , and can be locally important at the 33% level ( rhoads 1998 ) .
the basic idea of our approach is that the bar is a feature dominated mostly by even fourier terms in a relatively fixed position angle .
we compute the relative fourier intensity amplitudes @xmath20 , where @xmath0 is an integer index , as a function of radius and make the assumption , when necessary , that the relative bar intensity declines past a maximum in the same or a similar manner as it rises to that maximum .
that is , the fourier amplitudes relative to the axisymmetric background have a single maximum at radius @xmath21 and decline smoothly and roughly symmetrically to either side of this radius .
the bar extrapolation involves scaling the sine and cosine amplitudes of the even fourier terms according to the ratio , @xmath22 , of the @xmath0=0 amplitudes at symmetric radii @xmath23 .
this is essential because the average intensity of the background starlight decreases with increasing radius and because we extrapolate @xmath20 , not @xmath24 alone .
as long as the fourier phases are relatively constant , this scaling is reasonable .
our assumption about how to make the extrapolations is based on data from mostly pure bar plus disk systems , and on cases where the bar and spiral are well - enough separated that we can differentiate their distinct contributions in plots of @xmath20 and the @xmath0=2 phase @xmath25 .
ohta , hamabe , & wakamatsu ( 1990 ) presented relative fourier amplitudes for six early - type barred systems that showed three characteristics of the relative fourier amplitudes of bars : ( 1 ) strong bars have significant higher order terms , such that even @xmath0=10 can be important ; ( 2 ) relative amplitudes rise and then decline past a maximum that lies roughly in the middle of the apparent bar ; ( 3 ) the radii of the maxima for higher order terms ( @xmath0 = 4 , 6 , 8 , etc . )
can shift to slightly larger values compared to @xmath0 = 2 .
correct treatment requires that we examine plots of the relative amplitudes of all even fourier terms used , to evaluate especially effects ( 2 ) and ( 3 ) . to illustrate these points
, we use the galaxy ngc 4394 from the ohio state university bright galaxy survey ( osubgs , eskridge et al .
2002 ) . even in blue light ,
the spiral structure of this galaxy is weak , and in the osubgs @xmath26-band image , it is quite subdued .
thus , ngc 4394 can serve as our model case of mostly a bar imbedded in a disk , and we can examine how the relative fourier amplitudes behave with radius .
figure [ n4394 ] shows the even relative fourier amplitudes to @xmath0=20 for ngc 4394 versus radius based on a deprojected version of the osubgs @xmath26-band image where the bulge has been properly treated with a two - dimensional decomposition ( laurikainen et al .
each fourier term shows a well - defined maximum in the bar region .
when the relative fourier intensities are extrapolated past this maximum in the same manner as they rose to that maximum , the extrapolations are seen to be in fairly good agreement with the observed relative intensities .
there is weak spiral structure near and outside the ends of the bar , but it contributes little to the higher order terms .
detailed examination of the radius , @xmath21 , of the peak in each plot shows that it increases to slightly larger values with increasing @xmath0 , as expected from point 3 above .
the idea , according to ohta , hamabe , & wakamatsu ( 1990 ) , is that the narrow ends of the bar occur at the largest radii , and will be most evident as a result in the higher - order terms at those larger radii . finally , the phases of the terms ( not shown ) are relatively constant throughout the bar .
thus , ngc 4394 demonstrates that our assumption of approximate symmetry for the even fourier terms as a function of radius for a bar is a fair one .
the symmetry assumption is less evidently correct for the @xmath0=2 term in ngc 4394 .
instead , this term shows a strong asymmetric decline past its maximum .
the reason for this difference is that the main bar in ngc 4394 is imbedded in a weak extended oval that has no higher order terms than @xmath0=2 .
there is also some weak spiral structure at the ends of the bar . for these reasons ,
the @xmath0=2 term in ngc 4394 would have to be extrapolated as shown in figure [ n4394 ] to isolate the main bar .
the @xmath0=10 term in ngc 4394 has a maximum about 13% of the maximum of the @xmath0=2 term .
thus , it is not really negligible compared to @xmath0=2 , verifying point 2 above . for a very strong bar , such as that in ngc 1530 ,
the @xmath0=10 term is 19% of the maximum relative amplitude of the @xmath0=2 term .
one should never think naively of a bar as an @xmath0=2 structure .
only a broad oval is likely to be a pure @xmath0=2 structure .
once the fourier mappings of the bar are determined , the next step in the procedure is to sum the even ( @xmath272 ) terms of the light distribution out to a specified radius , where we assume the bar goes to zero ; there is also usually an inner radius where the bar is assumed to go to zero .
both of these effects are seen clearly for ngc 4394 in figure [ n4394 ] , where terms @xmath0=4 and 6 approach zero near @xmath28=10@xmath19 and @xmath28=55@xmath19 .
terms @xmath0=8 , 10 , and 12 approach zero near @xmath28=20@xmath19 . for most of the even terms in ngc 4394
, there appears to be some amplitude inside 10@xmath19 due in part in the finite pixel size ( 1@xmath195 ) .
each fourier term is treated on an individual basis to allow the radius of the maximum , @xmath21 , to occur at a different position .
the maximum number of fourier terms we use is based on analysis of ngc 1530 , the most strongly barred galaxy in the wht sample . using the bb01 approach ,
ngc 1530 would have been bar class 7 . to deduce how many fourier terms
might be needed for a bar / spiral separation analysis in this kind of case , we analyzed ngc 1530 by comparing gravitational torques derived from a full resolution image with those derived from fourier - smoothed images which cut the terms at @xmath0=10 and @xmath0=20 .
we found that except for the effects of noise , the fourier - smoothed image with terms to @xmath0=20 adequately represented the bar and that including higher order terms is not essential . cutting
the analysis at @xmath0=10 was not sufficient in that case .
however , for many galaxies , only the lower - order terms would be needed for a separation analysis , and we allow for the option of cutting the number of terms when appropriate .
the method is illustrated for ngc 6951 in figures [ plot6951 ] , [ n6951mplots ] , [ n6951images ] , and [ n6951ratios ] .
ngc 6951 is an ideal test case because it has a well - defined bar , and its spiral structure breaks directly from the ends of this bar .
it is a very typical example . in figure [ plot6951]a
, we show the relative @xmath9 fourier intensity amplitues for the @xmath0=2 , 4 , 6 , 8 , and 10 components .
( figure [ n6951mplots ] shows the relative amplitudes to @xmath0=20 on a larger scale , as for ngc 4394 . ) between radii of 6 and 32 , the bar dominates these amplitudes and we see a smooth rise in most of the terms . beyond 32 , the relative @xmath0=2 amplitude rises to a higher maximum , and the @xmath0=2 phase decreases slightly and then rises ( see figure [ plot6951]b ) .
we assume the bar does not end abruptly , but that in the absence of the spiral , the even relative fourier amplitudes due to the bar decline in the same manner as they rose to the bar maximum , which we took to occur at @xmath29 = 32@xmath195 .
this was chosen because @xmath30 shows a plateau near this radius , and the higher - order terms all show a maximum near or just outside this radius as well .
it corresponds to a position approximately in the apparent middle of the bar .
we require that the bar still be significant in @xmath0 = 2 at its apparent edge , and allow it drop in relative @xmath0=2 amplitude sharply thereafter .
this extrapolates the bar into the spiral zone ( as indicated in figure [ plot6951]a ) .
the impact of the choice of @xmath29 is examined in the next section . for the higher order terms ,
figure [ n6951mplots ] shows clear maxima between 33 and 42 that are attributable mainly to the bar .
these maxima do appear to shift outwards a little with increasing @xmath0 ( especially for @xmath31 6 ) , as expected from point ( 3 ) above .
for these terms , we simply reflect the even amplitudes around the apparent radius @xmath21 where the maxima occur , and scale them according to the mean intensity at each radius outside @xmath21 . for ngc 6951 ,
these extrapolations are shown by the crosses in figure [ n6951mplots ] .
figure [ n6951mplots ] also shows that fourier terms to @xmath0=18 are still detectable in the bar of ngc 6951 . including such terms provides a very good approximation to the total galaxy image . to verify this
, we computed potentials from both the full image at maximum resolution , and a fourier - smoothed image based on the sum of all even and odd terms up to @xmath0=20 .
we found that the fourier - smoothed @xmath0=0 - 20 image gives virtually the same potential as the full image , with differences being mainly attributable to noise and the occasional bright star - forming region .
interestingly , figure [ n6951mplots ] shows that the @xmath0=10 term goes to zero at @xmath28=45 , while our extrapolations make the other terms go to zero at 55 to 60 .
this shows the limitations of our approach in the sense that the bar relative intensity profiles may not be as symmetric as we assume .
none of the terms in ngc 4394 shows a similar disagreement with the other terms , so @xmath0=10 may be unusual in ngc 6951 .
our procedure in general is that if any term shows a significant portion of its decline past the maximum , we will use the decline as measured and extrapolate as little as possible .
this is what is done for @xmath0=10 in figure [ n6951mplots ] .
however , we have tested that our maximum relative torque results for ngc 6951 would be the same even if the @xmath0=10 term were extrapolated as for the other terms .
the impact of cutting all the even - order @xmath32 terms at @xmath28=45@xmath19 is considered in the next section . with extrapolations defined for each fourier term , we computed the images shown in figure [ n6951images ] .
full " image is the original deprojected and rotated image , with the bar horizontal .
this is the full resolution image and includes all noise .
@xmath0=0 - 20 sum " is the fourier - smoothed version of the full image .
it is based on 21 fourier terms , including all even and odd terms .
the bar " image is the one that uses the extrapolations shown in figure [ n6951mplots ] .
it was computed by summing the @xmath272 fourier terms within the limits set for the bar , and has no net flux.@xmath330 terms in a fourier series , when integrated over all azimuths , have no net flux .
the net flux in the bar would be the sum of all of its higher - order terms plus its @xmath0=0 term .
the net flux in the bar is brought back when we add the @xmath0=0 image to the `` bar '' image . ]
the fourth image is the @xmath0=0 fourier image , which shows the axisymmetric part of the @xmath9 light distribution .
the remaining images in figure [ n6951images ] show the separated bar and spiral images .
the image bar+disk " is the sum of the bar " and @xmath0=0 images .
spiral+disk " is the @xmath0=0 - 20 sum " minus the bar " image . with this procedure ,
we place most of the noise and all of the odd fourier terms into the spiral+disk " image .
figure [ n6951ratios ] shows the force ratio maps @xmath34 , where @xmath35 is the tangential force and @xmath36 is the mean axisymmetric radial force , for the full , fourier - smoothed , bar+disk , and spiral+disk images of ngc 6951 . in the full image , one clearly sees the effects of both the spiral and the bar .
a dominant butterfly " pattern is present with extra structure due to the spiral .
the fourier - smoothed image looks very much the same , only with less noise . in the bar+disk ratio map , we see four symmetrically placed
maximum points " which lie near the ends of the bar , as noted in bb01 .
the spiral+disk image shows a less symmetric , rough butterfly pattern as well .
thus , our approach appears to have separated the bar and spiral , and we can see how each component contributes to the total ratio map .
our next step is to use the ratio maps to derive single numbers that characterize the bar strength and the spiral strength . for this purpose
, we derive the maximum value , @xmath37 , of the ratio of the tangential force to the mean radial force as a function of radius .
these curves are interpolated bilinearly from the force ratio maps .
figure [ n6951ratios ] shows that the structure in the force ratio maps alternates by quadrants with two maximum positive values and two maximum negative values at each radius . bb01
derived @xmath38 in each quadrant , and then averaged these four values to get a single number , @xmath1 , called the bar strength " in the galaxy .
here we use a similar procedure , but as a function of radius . in figure
[ n6951ratios ] , the dotted curves show the maxima(minima ) for each quadrant . for the bar+disk , a symmetric pattern for these maxima(minima )
is mapped , as expected , while for the spiral a more complex mapping is found .
because the spiral is more complicated than the bar , we divided the ratio map into two 180@xmath13 sections around a vertical line , and searched for the maxima and minima in each section .
the sharp breaks in the spiral mapping are discontinuities in azimuth but not in radius , and are attributable to structure in the arms . at each radius , we averaged the values of @xmath38 to get the plots shown in figure [ plot6951]c .
the dashed curve shows that the bar has a maximum force ratio at @xmath39=315 ( 3.7 kpc ) , similar to the value chosen from the relative intensity amplitudes ( smallest circle superposed on the full image in figure [ n6951images ] ) .
the spiral maximum in the dotted curve occurs at @xmath40=575 ( 6.7 kpc , largest circle superposed on the full image ) . for comparison , the solid curve in figure [ plot6951]c shows the maxima from the total fourier - smoothed image , while the crosses show the maxima from the full image .
these curves agree well and show again that 21 fourier terms are adequate for the analysis of ngc 6951 .
figure [ plot6951]c shows that @xmath37(bar)=@xmath37(spiral ) at @xmath28@xmath41465 .
this corresponds to the middle circle superposed on the full image in figure [ n6951images ] .
the circle lies just outside the ends of the bar , and passes through the bright spiral arc on the right side of the bar . from these curves we derive @[email protected] and @[email protected] , where the uncertainties include only the internal scatter in maximum values .
the small internal uncertainty in @xmath1 is due to the fact that only even fourier terms were used to define the bar .
the larger internal uncertainty in @xmath2 is due to the fact that all fourier terms , even and odd , were used for the spiral .
additional uncertainties are discussed in the next section .
the maximum total relative gravitational torque is @xmath42=0.340 .
thus , our analysis shows that in this case , ignoring the effects of the spiral would lead us to overestimate the bar strength by about 20% .
nevertheless , the bb01 bar class " remains at 3 in either case .
figure [ plot6951]d shows the predicted , normalized rotation curve for ngc 6951 , based on the @xmath0=0 term of the derived gravitational potential .
the apparent flatness of the predicted curve for radii beyond @xmath28=10@xmath19 is consistent with observed rotation profiles along the photometric major axis obtained by mrquez & moles ( 1993 ) and prez et al .
( 2000 ) , and supports our assumption of a constant mass - to - light ratio , at least for this galaxy .
the maximum rotation velocity in ngc 6951 is @xmath43=230 km s@xmath5 ( mrquez & moles 1993 ) , favoring a maximal disk according to kranz , slyz , & rix ( 2003 ) .
bb01 discussed in detail the uncertainties in estimating gravitational torques from near - infrared images .
one of the principal uncertainties is in the vertical scaleheight @xmath16 . since this parameter can be measured directly only for edge - on galaxies , it has to be assumed . as shown by
de grijs ( 1998 ) , @xmath44 , where @xmath45 is the isothermal scaleheight ) depends on hubble type , but within any type there is a significant scatter . at stage
sbc , individual values of this ratio range from 4 to 8 .
this range must include both a cosmic scatter component as well as fitting and observational uncertainties . in our estimated value of @xmath1 for ngc 6951
, the scatter in the ratio leads to an uncertainty of @xmath60.022 , while for @xmath2 it leads to an uncertainty of @xmath60.020 .
thus , the uncertainty due to vertical scaleheight is at the 8 - 9% level for both @xmath1 and @xmath2 .
this excludes the uncertainty in our derived value of @xmath17 for ngc 6951 , which is sensitive to tke sky subtraction .
this is discussed further below , but note again that some of the uncertainty in this parameter due to observational errors and decomposition uncertainties must be included in the spread in @xmath46 at each type .
related to the issue of scaleheights is the possibility that the bar in ngc 6951 might be thicker than its disk , an effect thought to cause a boxy / peanut - shaped bulge structure in some edge - on disk galaxies ( bureau & freeman 1999 ) .
however , laurikainen & salo ( 2002 ) showed that such a non - constant vertical scaleheight is unimportant in the evaluation of relative bar torques , finding that it affects @xmath1 by less than 5% .
since we can not be sure what the vertical structure of the bar in ngc 6951 is like compared to its disk , we assume this could contribute an uncertainty of 0.05@xmath1 for our estimate of @xmath1 . for bar / spiral separation
, another possible major uncertainty will be the extrapolations of the bar fourier amplitudes
. the largest fourier term in any bar will be the @xmath0=2 term .
we first investigate how the choice of the radius of the @xmath0=2 maximum impacts our results .
for our analysis of ngc 6951 in the previous section , we used 325 for this radius , and obtained , as already noted , @[email protected] and @[email protected] .
if we use 355 instead ( figure [ unc]a and c ) , we obtain @xmath1=0.293 and @xmath2=0.190 , while if we use 295 ( figure [ unc]b and d ) , we obtain @xmath1=0.272 and @xmath2=0.234 . thus , the uncertainties in @xmath1 and @xmath2 due to the extrapolation will be correlated in the sense that if @xmath1 is too high , then @xmath2 will be too low and vice - versa .
the choice of the radius of the @xmath0=2 maximum appears to affect @xmath2 more than @xmath1 for ngc 6951 . a @xmath610% uncertainty in the radius of the bar maximum leads to @xmath64% uncertainty in @xmath1 and @xmath610% uncertainty in @xmath2 .
note that the greatly different @xmath0=2 extrapolations have little or no impact on the radii of the bar and spiral maxima .
both extrapolations still give @xmath39@xmath4131 and @xmath40@xmath4157 .
however , the radius of the crossover point , where @xmath47 , is sensitive to the @xmath0=2 extrapolation , ranging from 52 in figure [ unc]c to 43 in figure [ unc]d .
the second issue connected with the bar extrapolation is the symmetry assumption of the relative amplitudes .
we have noted that the @xmath0=10 term in ngc 6951 seems to violate this assumption and goes to zero at @xmath28=45@xmath19 , while our assumed extrapolations for the other terms go to zero at larger radii . to test the impact of the symmetry assumption
, we use the same extrapolations as before for all even @xmath0@xmath4810 , but cut all @xmath312 at @xmath28=45@xmath19 .
this approximates the asymmetry of the @xmath0=10 term for all even @xmath312 .
cutting the @xmath0=2 term at the same radius appears to be too drastic , however , because it leaves a sharp edge in the separated bar and spiral images .
thus , we have left the extrapolation for @xmath0=2 the same as before for this test .
the cutoff for the higher - order terms changes the derived relative maximum torques to @xmath1 = 0.281 and @xmath2 = 0.220 , amounting to 1% for @xmath1 and 4% for @xmath2 .
this shows that , even if the terms actually do cut off at a radius smaller than implied by our extrapolations , we do not commit a serious error in @xmath1 and @xmath2 if we ignore it .
other uncertainties discussed by bb01 included the flattening of the bulge , bulge deprojection stretch , uncertainties in the orientation parameters , sky subtraction uncertainties , and the constant @xmath49 assumption .
ngc 6951 has a low enough inclination that bulge deprojection stretch has a negligible effect on the derived maximum torques , which occur for both the bar and spiral far outside the bulge - dominated area .
uncertainties due to orientation parameters can be significant ( see table 3 of bb01 ) .
buta et al .
( 2003 ) show that , for a galaxy inclined about 40@xmath13 , the typical uncertainties in relative maximum torques is @xmath60.030 for @xmath65@xmath13 uncertainty in inclination and @xmath64@xmath13 uncertainty in major axis position angle .
the favorable agreement between the observed and predicted rotation curves of ngc 6951 suggests that dark matter is not important in the inner parts of the galaxy , and that our assumption of a constant @xmath49 is probably correct in this case . in general ,
the best way to evaluate the effects of dark matter on relative maximum torque calculations will be to compare predicted near - infrared rotation curves with observed ones , if available .
sky subtraction errors could impact our torque calculations , because the field of view of near - infrared images is usually limited and the sky level can not always be precisely determined . for ngc 6951
, we estimate a sky level uncertainty of @xmath60.2 adu in the intensity scale of the @xmath9 image .
such an uncertainty in the sky level will naturally affect our estimate of @xmath17 and hence also @xmath16 .
we derive an uncertainty of @xmath64@xmath19 in @xmath17 for @xmath60.2 adu uncertainty in the sky , corresponding to an uncertainty in the vertical scaleheight of @xmath678pc if @xmath16=(1/6)@xmath17 .
this combined sky subtraction / vertical scaleheight uncertainty leads to an uncertainty of @xmath60.010 in @xmath1 and @xmath60.011 in @xmath2 .
as might have been expected , the sky subtraction error impacts @xmath2 more than @xmath1 since the arms lie at larger radii , although the effect is only slight .
the change is 3.5% for @xmath1 and 5% for @xmath2 .
thus , allowing for the uncertainties in vertical scaleheight , bar extrapolation , sky subtraction , orientation parameter uncertainties , and the asymmetry in the spiral arms themselves , we derive @[email protected] and @[email protected] for ngc 6951 . finally , there will always be individual cases that require special treatment , particularly if the bar has considerable asymmetry .
in such cases , allowance for odd fourier terms in the bar is needed , and we will describe in paper ii an approach to dealing with them .
we have shown that straightforward fourier techniques can be used to make a reasonable separation of a bar from a spiral .
with such a separation , we can extend the bb01 gravitational bar torque method to spirals and define a spiral strength as well as a bar strength . in the case of ngc 6951 ,
the maximum relative total gravitational torque is 0.34 , while the individual bar and spiral strengths are [email protected] and [email protected] , respectively .
thus , ngc 6951 is bar class 3 and spiral class 2 , following table 1 of bb01 .
the general applicability of our separation method has not been fully evaluated since it has only been used for our sample of 15 wht galaxies and two additional galaxies from other sources . however , as we will show in paper ii , a reasonable separation was possible in each of those cases .
we anticipate that the method will be applicable to most spirals , but that some galaxies will require special treatment .
objects having multiple bars , very weak bars in strong ovals , or very asymmetric bars are examples of such cases .
we consider some of these cases in paper ii .
we rename the maximum relative total gravitational torque parameter as @xmath42 , to remove any ambiguity about what it represents . in cases with strong bars and weaker spirals , the actual bar strength @xmath1
@xmath41 @xmath42 , while in cases with strong spirals and weaker bars , @xmath2 @xmath41 @xmath42 . in general
, a separation analysis will be needed to investigate real bar torques . however , the derivation of @xmath42 alone is useful for some studies ( e.g. , investigations of the impact of gas accretion in galaxy disks ; block et al .
2003 ) , and is the most straightforward parameter to derive .
more details on the practical aspects of deriving @xmath42 for a large number of galaxies is provided by laurikainen et al .
( 2003 ) .
we thank h. salo and e. laurikainen for helpful discussions concerning refinements of the @xmath1 method , and for providing a revised lookup table for the vertical dimension in our analysis and for the deprojected osubgs @xmath26-band image of ngc 4394 .
we also thank an anonymous referee for several helpful comments .
rb acknowledges the support of nsf grant ast-0205143 to the university of alabama and also the anglo american chairman s fund during a visit to the university of the witwatersrand in 2002 november .
the research of dlb is supported by the anglo american chairman s fund , with deep gratitude expressed to ceo m. keeton , special advisor h. rix , and the board of trustees .
dlb is also most appreciative to sasol for funding the sabbatical period of his research .
the william herschel telescope is operated on the island of la palma by the isaac newton group in the spanish observatorio del roque de los muchachos of the instituto de astrofsica de canarias .
funding for the osu bright galaxy survey was provided by grants from the national science foundation ( grants ast-9217716 and ast-9617006 ) , with additional funding from the ohio state university . | we describe a fourier - based method of separating bars from spirals in near - infrared images .
the method takes advantage of the fact that a bar is typically a feature with a relatively fixed position angle , and uses the simple assumption that the relative fourier amplitudes due to the bar decline with radius past a maximum in the same or a similar manner as they rose to that maximum .
with such an assumption , the bar can be extrapolated into the spiral region and removed from an image , leaving just the spiral and the axisymmetric background disk light .
we refer to such a bar - subtracted image as the `` spiral plus disk '' image .
the axisymmetric background ( fourier index @xmath0=0 image ) can then be added back to the bar image to give the `` bar plus disk '' image .
the procedure allows us to estimate the maximum gravitational torque per unit mass per unit square of the circular speed for the bar and spiral forcing separately , parameters which quantitatively define the bar strength @xmath1 and the spiral strength @xmath2 following the recent study of buta & block . for the first time , we are able to measure the torques generated by spiral arms alone , and we can now define spiral torque classes , in the same manner as bar torque classes are delineated .
we outline the complete procedure here using a 2.1@xmath3 m image of ngc 6951 , a prototypical sab(rs)bc spiral having an absolute blue magnitude of @xmath421 and a maximum rotation velocity of 230 km s@xmath5 .
comparison between a rotation curve predicted from the @xmath0=0 near - infrared light distribution and an observed rotation curve suggests that ngc 6951 is maximum disk in its bar and main spiral region , implying that our assumption of a constant mass - to - light ratio in our analysis is probably reliable .
we justify our assumption on how to make the bar extrapolation using an analysis of ngc 4394 , a barred spiral with only weak near - infrared spiral structure , and we justify the number of needed fourier terms using ngc 1530 , one of the most strongly - barred galaxies ( bar class 7 ) known .
we also evaluate the main uncertainties in the technique . allowing for uncertainties in vertical scaleheight , bar extrapolation , sky subtraction , orientation parameters , and the asymmetry in the spiral arms themselves , we estimate @[email protected] and @[email protected] for ngc 6951 . |
* material characterization . *
the suspensions are prepared by mixing mono - dispersed glass beads of density @xmath111 g cm@xmath13 with a newtonian liquid of density @xmath113 and viscosity @xmath41 .
most of the experiments are conducted with a suspension of beads of diameter @xmath75 @xmath76 m ( with a standard deviation of 20 @xmath76 m ) in water ( @xmath114 g cm@xmath13 , @xmath115 pa s ) .
in addition , suspension of larger glass beads of diameter @xmath77 @xmath76 m ( with a standard deviation @xmath116 @xmath76 m ) immersed in water or in a mixture of ucon lubricant fluid ( dow , 75-h-90,000 ) and water ( @xmath117 g cm@xmath13 , @xmath118 pa.s ) are used to explore the influence of bead diameter and fluid viscosity .
care was taken to use two batches of grains with similar internal friction coefficient ( tested by comparing the angle of repose ) , in order to get similar value of the parameter @xmath54 .
the value of the critical volume fraction @xmath94 obtained by the pressure measurements ( change of sign of the peak pore - pressure after impact ) in the case of @xmath119 @xmath76 m is @xmath120 and is thus different from the other suspension ( for which @xmath121 ) , probably due to differences in grain size distribution .
an important parameter of the model is the darcy permeability @xmath122 of the granular suspension , which relates the flow rate across the porous medium to the gradient of fluid pressure @xcite .
the permeability of each suspension were obtained by imposing a downward gravity - driven flow through the grains and by measuring the corresponding flow rate , giving @xmath123 m@xmath124 for the 170 @xmath76 m diameter beads and @xmath125 m@xmath124 for the 500 @xmath76 m diameter beads .
the variation of the permeability with the initial volume fraction of the pile is less than 15 @xmath126 in the range studied and neglected in the model . + * suspension preparation . *
the particles are mixed with the working liquid and placed in a large container ( side length : 18 cm , depth : 27 cm ) . at the beginning of each experiment ,
the granular bed is fluidized by injecting the liquid through a porous plate placed at the bottom of the container . when the tank is almost filled with the liquid , the liquid supply is stopped and the particles are allowed to sediment .
the tank wall is then gently tapped to compact the bed .
the tapping on the side of the container is performed by a pendulum , released from a constant height , to ensure repeatability of the compacting process . before each impact
, the granular bed height @xmath127 is measured at each side of the container with an accuracy of 0.5 mm for a range of 9 to 10 cm depending on the number of taps .
the initial volume fraction is deduced from these measurements with an accuracy of @xmath128 .
the excess liquid is later drained - off so that a thin liquid layer is kept on the surface of the pile ( @xmath8 mm ) .
the exact amount of water left does not affect the experimental results presented in this study . finally , a solid sphere made of glass ( density @xmath129 g cm@xmath13 , diameter @xmath28 or @xmath130 mm ) is let to freely fall on the granular bed , impacting at a speed @xmath19 .
a high - speed video camera records the dynamics ( up to 150,000 frames per seconds ) . + *
pore pressure measurements . * to limit the perturbation induced by a sensor in the region of the suspension affected by impact , a low footprint pore
pressure sensor was designed . a thin
, @xmath131 cm long , l shaped stainless steel tube ( internal diameter : @xmath132 mm , thickness : @xmath133 mm ) , ended by a fine metallic grid preventing the grains from entering into the tube is attached to a water - proof chamber .
the tube and the chamber are filled with the working liquid . a hydrophone (
brel & kjr @xmath134 ) with a flat frequency response in the range 1 hz-15 khz is used to measure the high frequency pressure signal within the chamber .
an additional low frequency differential pressure sensor ( honeywell ) is used to determine the slow relaxation dynamics .
because of the fine grid and of the complex geometry of the pore pressure sensor , a careful calibration of this measuring device had to be conducted through a specific experiment described in extended data fig .
the calibration chamber is divided into three compartments : compartment ( i ) contains the suspension to be impacted , compartment ( ii ) contains a similar suspension but with a different height and finally compartment ( iii ) is a water - filled chamber containing the pore pressure sensor and a bare hydrophone separated from compartment ( i ) by a porous plate allowing transmission of the pressure signal .
the purpose of compartment ( ii ) is to ensure that the thin tubes are fastened just as the tubes in the experiment of the main text . after impact under conditions similar to the main experiment ,
the transfer function of the pore pressure sensor was determined by measuring the signals from the bare hydrophone ( bh ) and the embedded hydrophone ( eh ) of the pore pressure sensor at different grain sizes and fluid viscosities .
+ * refined impact model . * the model presented in the main text can be slightly extended to offer a more accurate description of the dynamics when the indentation is not small and for larger values of @xmath51 .
the proposed modifications lie in the following two arguments : 1 . in the main text model
the radius of contact was approximated by @xmath135 , valid for small @xmath57 .
when the indentation @xmath57 is not small the radius of contact is better described by the following form : @xmath136 with @xmath137 if @xmath138 and @xmath139 otherwise .
2 . in the main text model the parameter @xmath54 , relating the normal stress @xmath140 acting of the ball and the confining granular pressure @xmath141 , was assumed to be constant .
however , this parameter changes with the internal friction angle of the pile @xmath142 , which itself depends on the initial volume fraction @xmath51 @xcite . to account for this dependence of @xmath54 with @xmath51
, we model @xmath54 by the classical expression used in soil mechanics to describe the bearing capacity of a frictional soil ( the so - called load - bearing capacity factor ) @xcite , that is @xmath143 and we relate the internal friction angle @xmath142 with @xmath51 using the dilatancy angle @xcite , so that @xmath144 , where @xmath145 is the internal friction angle at the critical state @xmath146 and @xmath147 as before . with these two modifications ,
the equation for the dynamics of the sphere reads @xmath148,\ ] ] where @xmath149 and @xmath150 and @xmath151 , with @xmath152 this model yields results that depend both on @xmath153 and @xmath82 . the predictions from this model with @xmath154 and @xmath155
are shown in fig .
3 of the main text . +
* extension to compressible interstitial fluids .
* in the model presented in the text , the interstitial fluid is assumed incompressible . when fluid compressibility is taken into account ,
the linearized mass conservation equation for the fluid phase becomes : @xmath156 , where @xmath157 is the compressibility of the fluid .
using as before the mass conservation of the solid phase and taking the divergence of the darcy law ( [ eq : darcy ] ) gives : @xmath158 when the diffusion time scale @xmath159 is short compared to the impact timescale @xmath102 , the compressible term @xmath160 is small compared to the darcy term @xmath161 and one recovers the incompressible darcy - reynolds equation ( 1 - 4 ) for the pore pressure .
by contrast , when @xmath162 , the compressible term is dominant and the pore - pressure is given by : @xmath163 , that is @xmath164 . in fig 4 , the pore
pressure is computed as @xmath165 when the compressible number @xmath166 and as @xmath167 when @xmath168 .
s1 : calibration procedure for the pore pressure measurement .
( a ) calibration chamber ( @xmath169 cm , @xmath170 cm ) and signal @xmath171 ( resp .
@xmath172 ) recorded by the bare hydrophone ( resp . by the embedded hydrophone ) during a typical impact condition with @xmath173 , @xmath174 m s@xmath22 , @xmath175 mm , @xmath119 @xmath76 m , @xmath176 pa s. the embedded hydrophone ( grid and tube ) induces a strong distorsion in both in the amplitude and temporal characteristics of the input signal .
the inverse transfer function of the sensor in the fourier space , @xmath177 , is obtained from the signals @xmath171 and @xmath178 using the following formula : @xmath179 , where @xmath180 $ ] and @xmath181 is a filtering function used to enforce causality . the inverse transfer function
is then averaged over 3 trials performed at the same impact condition and used for all calibrations made in the study .
( b ) red : input signal reconstructed from the signal of the embedded hydrophone ( inset ) using the inverse transfer function @xmath177 determined previously , for a different impact velocity ( @xmath182 m s@xmath22 ) .
the reconstructed signal @xmath183 is obtained from the signal of the embedded hydrophone @xmath172 using the formula : @xmath184 $ ] .
black : input signal @xmath185 measured directly by the bare hydrophone for the same impact .
the transfer function introduces spurious oscillations on long times in the reconstructed signal but correctly reproduces the peak pressure just after impact .
( c ) comparison between the peak pressure measured with the bare hydrophone @xmath186 and the peak pressure of the reconstructed signal @xmath187 for different impact conditions ( black crosses : @xmath174 m s@xmath22 ; blue circles : @xmath182 m s@xmath22 ; red squares : @xmath188 m s@xmath22 ; green diamonds : @xmath189 m s@xmath22 , @xmath190 cm ; purple star : @xmath189 m s@xmath22 , impact in a looser pile ) . in all cases ,
the reconstructed peak pressure is within 20 @xmath126 of the real peak pressure .
[ figsupp1 ] fig .
s2 : penetration dynamics in the dense regime .
( a ) experimental position , speed and acceleration for @xmath191 , @xmath192 , @xmath193 , @xmath194 , @xmath195 , @xmath196 ( glass beads @xmath75 @xmath76 m in water , @xmath28 mm , @xmath72 m s@xmath22 ) .
( b ) same data rescaled using the characteristic time @xmath197 and length @xmath198 together with the model prediction ( dashed line ) .
[ fig : model1 ] fig .
s3 : transient hardening is associated with negative pore
pressure in dense suspensions of cornstarch in water .
( a ) a cylinder is impulsively moved by hand and the pore
pressure is recorded simultaneously .
the pressure sensor is connected to a tube filled with water and ended by a fine semi - permeable paper that prevents cornstarch particles to enter the tube .
( b ) pore pressure versus time ( cornstarch 55 @xmath126 wt in distilled water , the suspension has been vigorously stirred before the experiment to avoid sedimentation effects ) .
the cylinder is put in motion at @xmath199 .
[ fig : scaling ] tab .
s1 : impact parameters and granular / fluid properties used to plot the phase diagram in fig .
4 . for each material , the permeability @xmath122 is computed using carman - kozeni formula @xcite : @xmath200 .
the air compressibility is given by @xmath201 , where @xmath105 is the air pressure and @xmath202 .
[ fig : table ] r. jackson , _ the dynamics of fluidized particles _
( cambridge university press , cambridge , 2000 ) .
androtti b , forterre y , pouliquen o ( 2013 ) _ granular media : between fluid and solid _ _ cambridge university press_. a. schofield and p. wroth , _ critical state soil mechanics _ ( cambridge university press , cambridge , 1968 ) . | a sharp transition between liquefaction and transient solidification is observed during impact on a granular suspension depending on the initial packing fraction .
we demonstrate , via high - speed pressure measurements and a two - phase modeling , that this transition is controlled by a coupling between the granular pile dilatancy and the interstitial fluid pressure generated by the impact .
our results provide a generic mechanism for explaining the wide variety of impact responses in particulate media , from dry quicksand in powders to impact - hardening in shear - thickening suspensions like cornstarch .
1.5 cm impacts on particulate media like granular materials and suspensions present an astonishingly rich phenomenology @xcite . along with its astrophysical @xcite and ballistics applications @xcite ,
impact dynamics is an object of active research to understand the high - speed response of granular matter @xcite . in dry granular media ,
impact by a solid object results in the formation of a corona of granular ejecta and a solid like plastic deformation leading to a permanent crater @xcite . for fine powders in air ,
granular jets and cavity collapse occur during impact @xcite .
subsequent studies showed that the ambient pressure of the interstitial fluid ( air ) is an important element for the observed fluid like behavior @xcite , while for denser packing the impact penetration is much reduced @xcite .
however , the question of the physical mechanisms and control parameters that give rise to such a wide variety of phenomena is still largely open .
recently , studies on shear thickening suspensions ( cornstarch ) showed completely different behaviors . above a critical velocity ,
an impacting object immediately stops @xcite , or in some cases generates cracks @xcite , as if hitting a solid .
this phenomenon has been related to the propagation of dynamic jamming fronts in the bulk @xcite but the mechanism remains unclear and overlooks the role of fluid / grains couplings , which are known to strongly affect the transient behavior of saturated granular materials @xcite .
whether impact - activated solidification relies on such couplings or on the complex rheology of the suspension is a pivotal question for clarifying the physics of shear - thickening fluids a still highly debated topic @xcite .
the objective of this letter is to address these questions and elucidate the role of the interstitial fluid and the initial volume fraction on the diverse impact phenomenology observed in granular materials and dense suspensions during the last decade . to avoid difficulties associated with colloidal interactions between particles ( like in shear - thickening suspensions ) or fluid compressibility ( like in powders in air ) , we study here the impact of a freely - falling rigid sphere on a simple granular suspension @xcite made up of macroscopic , heavy particles ( glass beads in the range 0.11 mm ) immersed in an incompressible liquid ( water , viscous oil ) .
the initial packing fraction of the suspension @xmath0 ( the ratio of the volume of the glass beads to the total volume ) is controlled by first fluidizing the mixture and then compacting the sediment , before removing the excess liquid ( see fig .
1(a ) and supplemental materials for the detailed set - up ) .
for a granular pile of frictional spherical particles , @xmath0 typically takes values between 0.55 and 0.62 @xcite .
the suspension is kept fully saturated so that surface tension effects can be neglected .
remarkably , two very distinct impact regimes are observed depending on the initial packing fraction [ fig .
1(b ) ] . for
initially loose packing ( @xmath1 ) , the ball readily sinks in the suspension , giving rise to a collapsing cavity and a central jet ( see movie 1 in supplemental material ) .
this is the typical behavior of a sphere impacting a liquid pool wherein the dynamics is dominated by fluid inertia @xcite .
such fluid - like behavior is also strikingly similar to that observed when a rigid sphere hits a dry loose powder under atmospheric pressure @xcite .
by contrast , for dense packing ( @xmath2 ) , the ball stops abruptly within a few milliseconds as it hits the surface ( see movie 2 in supplemental material ) , with a huge deceleration of about @xmath3 @xmath4 ( where @xmath5 m s@xmath6 ) [ fig .
1(c ) ] .
this solid - like behavior is strongly reminiscent of the impact activated solidification observed in shear - thickening suspension like cornstarch @xcite .
it also clearly depends on the grain size and viscosity of the interstitial fluid between the particles . using coarser particles in the same fluid tends to suppress the extreme deceleration , which can be restored by increasing the fluid viscosity ( see movie 3 in supplemental material )
the impact dynamics is also very different when the same grains are put in air instead of water . in this case , as reported in previous studies @xcite , grains ejecta followed by the formation of a permanent crater is observed ( see movie 4 in supplemental material ) .
therefore both the initial packing fraction and the interstitial fluid play a key role on the suspension behavior during impact . to explain how such a drastic change in behavior can occur with only a slight change of packing fraction ( @xmath7 ) , we rely on a pore - pressure feedback mechanism : a coupling between the deformation of the granular medium and the pressure of the interstitial fluid between the grains @xcite . as first described by o. reynolds @xcite , when a dense granular packing starts to flow
, it must dilate .
since the medium is saturated with an incompressible liquid , the fluid is sucked in , as evidenced by the bright zone developing beneath the impacting sphere in the dense case [ fig .
1(b ) ] . therefore ,
when the ball hits the pile , the interstitial fluid pressure ( pore pressure ) drops instantaneously , which in turn presses the grains against each other thereby enhancing the friction .
thus , the medium is transiently solidified .
a loosely packed granular bed , on the other hand , tends to compact when it deforms .
therefore , a rise in pore
pressure is produced during impact that can balance the weight of the grains .
this suppresses the contact network , resulting in local or global fluidization .
we proceed a direct verification of this mechanism by developing a high - frequency measurement of the interstitial fluid pressure inside the suspension , just under the impact [ fig .
2(a ) ] ( see fig . s1 and supplemental material for the calibration procedure ) . in the loose case [ fig .
2(b ) , upper panel ] , the pore
pressure shows a sudden positive peak after impact . the measured peak pressure ( @xmath8 kpa ) is larger than the effective weight of the suspension above the sensor , @xmath9 pa ( @xmath10 , @xmath11 cm , and @xmath12 kg m@xmath13 ) , indicating a fluidization of the medium . at longer times
sedimentation proceeds and the pressure slowly relaxes . by contrast , in the dense case [ fig .
2(b ) lower panel ] , the peak pressure is negative and its magnitude ( @xmath14 kpa ) is much higher than the confining pressure due to gravity , indicating that particles are strongly pressed against each other , effectively leading to a solidification of the medium .
interestingly , the perturbation triggered by the impact remains localized near the impact point as shown by the pressure profile along the @xmath15-direction within the medium ( inset of lower panel of fig .
2b ) .
the decay length scale is @xmath16 1 cm and much smaller than the container size , ensuring that the results reported are independent of wall effects .
the transition between a positive and negative peak pressure occurs for a critical packing fraction @xmath17 , which is independent of the projectile diameter @xmath18 and impact speed @xmath19 [ fig .
2(c ) ] .
this value is consistent with previous rheological measurements of the jamming packing fraction using frictional spherical particles @xcite . in the suspension .
( b ) , pore pressure as function of time in the loose case ( upper panel ) and dense case ( lower panel ) ( same experimental conditions as in fig . 1
, @xmath20 cm ) . in the dense case ,
the negative peak pressure intensity decreases with the distance from the impact ( inset ) .
( c ) , peak pore pressure as function of the initial packing fraction @xmath0 for various impact speeds @xmath19 , sphere diameters @xmath18 and sensor positions @xmath15 ( diamonds : @xmath21 m s@xmath22 , @xmath23 mm , @xmath24 cm ; circles : @xmath25 m s@xmath22 , @xmath23 mm , @xmath26 cm ; square : @xmath27
m s@xmath22 , @xmath28 mm , @xmath29 cm ; triangles : @xmath30 m s@xmath22 , @xmath28 mm , @xmath29 cm ) . ]
in order to model the impact dynamics , the coupling between the grain matrix deformation and the interstitial fluid should necessarily be taken into account .
for an assembly of rigid particles , the simplest dilatancy law relating the evolution of the volume fraction @xmath31 and the packing deformation at the onset of plastic flow is given by @xcite @xmath32 where @xmath33 is the absolute shear rate of the granular medium and @xmath34 is the reynolds dilatancy angle @xcite , assumed proportional to @xmath35 , and @xmath36 a constant of order unity .
this deformation of the granular matrix induces in turn an interstitial fluid flow , which , for the low reynolds numbers considered here , is described by the darcy law @xcite @xmath37 where the vector fields @xmath38 and @xmath39 denote the particle and liquid velocity , respectively , @xmath40 is the liquid pore pressure , @xmath41 the fluid viscosity and @xmath42 the permeability of the granular pile . assuming that the liquid and the particles are incompressible : @xmath43 and @xmath44 , and taking the divergence of the darcy law ( [ eq : darcy ] ) gives @xmath45 , in which spatial variations of @xmath31 have been neglected . using the reynolds dilatancy equation ( [ eq : reynolds ] )
, we finally obtain a poisson - like equation for the pore
pressure @xmath46 in which the sign of the source term is imposed only by @xmath35 . therefore , the pore pressure generated by the impact scales as @xmath47 where @xmath48 , @xmath49 is the velocity scale for the particle velocity field , @xmath50 the typical extent up to which deformation is experienced by the granular bed , and @xmath36 a constant of order unity .
this relation predicts that the pore pressure is positive ( fluid - like response ) or negative ( solid - like response ) depending only on the sign of @xmath51 , while its magnitude is controlled by both the grain diameter and fluid viscosity , in agreement with observations ( fig . 2c and movie 3 in the supplementary material ) . the darcy - reynolds model ( [ eq : darcyreynoldsscaling ] ) can also be used to quantitatively infer the penetration dynamics in the dense case ( @xmath52 ) . assuming a frictional rheology for the granular suspension @xcite and neglecting the confining pressure due to gravity in front of the pore
pressure , the contact stress on the impactor is @xmath53 , where @xmath54 is an effective friction coefficient @xcite and @xmath55 is the pore pressure ( [ eq : darcyreynoldsscaling ] ) , in which @xmath56 , where @xmath57 is the penetration depth , and @xmath58 the typical radius of the contact area of the projectile ( consistent with our measurements of the pressure profile , see inset of lower panel of fig .
2b ) . using newton s second law for the impactor @xmath59 , where @xmath60 is the density of the projectile , and assuming small penetration ( @xmath61 )
, the penetration @xmath57 then evolves according to the non - dimensional equation ( after integration with initial conditions @xmath62 and @xmath63 ) : @xmath64 , where @xmath65 is the dimensionless penetration and @xmath66 a dimensionless time given by @xmath67 , the typical stopping time @xmath68 ( the time at which @xmath69 ) and the maximal acceleration @xmath70 plotted against @xmath51 for different impact speeds ( diamonds : @xmath71 m s@xmath22 , disks : @xmath72 m s@xmath22 , squares : @xmath73 m s@xmath22 , with @xmath28 mm ) , impactor size ( circles : @xmath74 m s@xmath22 , @xmath23 mm ) and suspension properties ( diamonds , disks , squares , circles : particles @xmath75 @xmath76 m in water ; crosses : particles @xmath77 @xmath76
m in a viscous fluid @xmath78 pa@xmath79s , stars : particles @xmath77 @xmath76 m in water , with @xmath80 mm and @xmath81 m s@xmath22 ) .
( b ) rescaled data plotted against @xmath82 compared with the predicted scaling laws ( thick dashed line ) and the improved model ( thin dashed lines , see methods ) .
( c ) pore pressure peak measured at different depth and rescaled by the predicted scaling laws : @xmath83 and @xmath84 ( @xmath85 , same legend except for circles , for which : @xmath72 m s@xmath22 and squares , for which : @xmath86 m s@xmath22 ) . ] vs @xmath51 ) showing diverse impact behaviours ( symbols ) for a wide range of grain size ( @xmath87 @xmath76 m ) , impact velocity ( @xmath88 m s@xmath22 ) , fluid viscosity ( @xmath89 pa s ) and fluid compressibility ( @xmath90 pa ) with data from our work ( dense suspensions , hatched area ) and previous studies for glass beads in air @xcite ( orange ) , impacts at ultra - low air pressure mimicking planetary craters @xcite ( purple ) , impacts on fine powders at atmospheric and low air pressure ( green @xcite , red @xcite , dark blue @xcite ) , impact on suspensions of cornstarch and water @xcite ( light blue ) .
filled and open symbols correspond to incompressible and compressible case .
the vertical bars give the range of @xmath91 covered in the corresponding study and the horizontal bars give the uncertainty on @xmath92 ( see table in the supplemental material ) . ]
[ fig : model1 ] this dynamics is compared with experiments at various initial packing fraction @xmath93 for a given projectile and impact velocity in fig .
s2 ( see supplemental material ) . as @xmath0 approaches @xmath94 , the indentation depth increases and the maximal deceleration decreases , while the stopping time increases .
a collapse of all trajectories is obtained when the variables are rescaled according to the model prediction ( see fig .
s2 in supplemental material ) . quantitative agreement is achieved by taking @xmath95 .
such a value is consistent with experimental measurements of @xmath36 and @xmath54 for glass beads @xcite , for which @xmath96 and @xmath97 .
furthermore , the darcy - reynolds model captures the impact dynamics for a wide range of physical parameters : using different suspension mixtures and different impact parameters , the data for the maximal indentation depth @xmath98 , stopping time @xmath99 and maximal deceleration @xmath70 collapse on the prediction when plotted as function of @xmath82 [ fig .
3(a , b ) ] .
finally , we systematically measure the pore
pressure profile below the impact point for different impactor parameters and suspensions [ fig .
3(c ) ] .
an universal exponential decay is obtained when the pore pressure is scaled by the maximal pore pressure given by the darcy - reynolds scaling ( [ eq : darcyreynoldsscaling ] ) and the depth @xmath15 is scaled by @xmath84 , thereby conclusively supporting the model .
these results can be easily extended to suspensions for which the interstitial fluid is not a liquid but a gas , like dry powders at different operating air pressure . as long as the diffusion timescale for gas expulsion @xmath100 , where @xmath101 is the gas compressibility , is small
compared to the impact timescale @xmath102 , the fluid can be assumed incompressible and the darcy - reynolds scaling ( [ eq : darcyreynoldsscaling ] ) of the pore pressure still holds ( see supplemental material ) .
otherwise , the fluid has no time to escape from the pores during impact .
the pore
pressure is then given by a gas state equation @xmath103 , where @xmath104 and @xmath105 is the gas pressure . in fig .
4 , we compare the predicted pore pressure @xmath55 with the typical confining pressure due to gravity @xmath106 for our study and previous impact studies covering a wide range of particulate media and impact conditions , for both incompressible and compressible interstitial fluids . when @xmath107 , the pore
pressure built - up during impact dominates the dynamics , yielding quicksand - like or solid - like response depending on the sign of @xmath51 , in agreement with observations @xcite . by contrast when @xmath108 the interstitial fluid contributes negligibly to forces between grains and so
, one recovers the classical dry granular case phenomenology @xcite .
our study thus provides an unifying picture to explain the main regimes observed during the impact of a solid object onto a mixture of grains and fluid , at least when surface tension effects can be neglected ( saturated suspension ) .
it also provides a mechanism to explain the impact activated solidification observed in more complex shear thickening media like cornstarch @xcite . in these systems , the critical packing fraction @xmath94 is expected to be a decreasing function of the impact velocity @xcite , and not a material constant like in the present study .
thus , at high enough impact velocities , the suspension should become dilatant and solidify due to the pore pressure feedback mechanism .
since the cornstarch particles are about 10 @xmath76 m and the suspension permeability scale as @xmath42 , the darcy - reynolds scaling ( [ eq : darcyreynoldsscaling ] ) predicts a 100-fold increases of the pore
pressure compared to that of the newtonian suspension with glass beads ( @xmath109 @xmath76 m ) .
this is about @xmath110 mpa for typical impact conditions , which is in good agreement with impact stress estimations in cornstarch @xcite [ fig .
4 ] .
we confirm this viewpoint by performing qualitative measurements of pore pressure in cornstarch suspensions . when an object is impulsively - moved in a concentrated suspension of cornstarch in water ,
pore
pressure takes indeed a huge drop ( see fig .
s3 in supplemental material ) .
therefore , it is remarkable that the conjunction of two distinguished mechanisms in granular media , the darcy law and the reynolds dilatancy , could explain this long - standing puzzle : why we can run on cornstarch . 1 j. c. ruiz - surez , rep .
. phys .
* 76 , * 066601 ( 2013 ) .
m. omidvar , m. iskander and s. bless , int . j. impact
. eng . * 66 * , 60 ( 2014 ) .
h. j. melosh , _ impact cratering : a geologic process _
( oxford university press , oxford , 1989 ) .
n. j. wagner and j. f. brady , physics today * 62 , * 27 ( 2009 ) .
a. h. clark , a. j. petersen , l. kondic and r. p. behringer , phys .
rev .
lett .
* 114 , * 144502 ( 2015 ) .
a. m. walsh , k. e. holloway , p. habdas and j. r. de bruyn , phys .
rev .
lett .
* 91 , * 104301 ( 2003 ) .
j. s. uehara , m. a. ambroso , r. p. ojha and d. j. durian , phys .
rev .
lett . * 90 * 194301 ( 2003 ) .
s. yamamoto , k. wada , n. okabe and t. matsui , icarus * 183 , * 215 ( 2006 ) .
s. deboeuf , p. gondret and m. rabaud , phys .
rev .
e * 79 , * 041306 ( 2009 ) .
j. o. marston , e. q. li and s. t. thoroddsen , j. fluid .
mech .
* 704 , * 5 ( 2012 ) .
r. zhao , q. zhang , h. tjugito and x. cheng , proc .
natl acad .
sci .
usa * 112 * , 342 ( 2015 ) .
s. t. thoroddsen and a. q. shen , phys .
fluids * 13 , * 4 ( 2001 ) .
d. lohse , r. rauh , r. bergmann and d. van der meer , nature * 432 , * 689 ( 2004 ) .
j. r. royer _ et al .
_ nature physics * 1 , * 164 ( 2005 ) .
g. caballero , r. bergmann , d. van der meer , a. prosperetti and d. lohse , phys .
rev .
lett .
* 99 , * 018001 ( 2007 ) .
j. r. royer _ et al .
_ phys .
rev .
e * 78 , * 011305 ( 2008 ) .
j. r. royer , b. conyers , e. i. corwin , p. j. eng and h. m. jaeger , europhys . lett .
* 93 , * 28008 ( 2011 ) .
s. r. waitukaitis , h. m. jaeger , nature * 487 , * 205 ( 2012 ) .
m. roch , e. myftiu , m. c. johnston , p. kim and h. a. stone , phys .
rev .
lett .
* 110 , * 148304 ( 2013 ) .
r. m. iverson , reviews of geophysics .
* 35 , * 245 ( 1997 ) .
m. pailha and o. pouliquen , j. fluid mech .
* 633 , * 115 ( 2009 ) .
b. andreotti , y. forterre and o. pouliquen , _ granular media : between fluid and solid _ ( cambridge university press , cambridge , 2013 ) .
n. fernandez _
et al . _
phys .
rev .
lett .
* 111 , * 108301 ( 2013 ) .
r. seto , r. mari , j. f. morris and m. d. morton , phys .
rev .
lett .
* 111 , * 218301 ( 2013 ) .
m. wyart and m. e. cates , phys .
rev .
lett .
* 112 , * 098302 ( 2014 ) .
e. brown and h. m. jaeger , rep .
prog .
phys .
* 77 , * 046602 ( 2014 ) .
a. fall _ et al .
_ phys .
rev .
lett .
* 114 , * 098301 ( 2015 ) .
f. boyer , e. guazzelli and o. pouliquen , phys .
rev .
lett . * 107 , * 188301 ( 2011 ) .
a. m. worthington , _ a study of splashes _ ( longmans green and co. , 1908 ) .
o. reynolds , phil . mag .
series * 20 , * 469 ( 1885 ) .
s. roux and f. radjai , in _ physics of dry granular media _ ( springer , 229236 , 1998 ) .
h. darcy , _ les fontaines publiques de la ville de dijon _ ( victor dalmont , paris , 1856 ) .
r. jackson , _ the dynamics of fluidized particles _
( cambridge university press , cambridge , 2000 ) .
a. schofield and p. wroth , _ critical state soil mechanics _ ( cambridge university press , cambridge , 1968 ) .
a. seguin , y. bertho , f. martinez , j. crassous and p. gondret , phys .
rev .
e * 87 , * 012201 ( 2013 ) . * acknowledgment * the authors thank mathieu leger and sady nol for the design of the experimental set up .
we also thank emil dohlen and students from polytechmarseille for preliminary results .
this work was supported by the french national research agency ( anr ) through the program no .
anr-11-js09 - 005 - 01 and by the labex mec ( anr-10-labx-0092 ) and the a*midex project ( anr-11-idex-0001 - 02 ) funded by the `` investissements davenir '' french government program . +
* author information * the authors declare no competing financial interests .
correspondence and requests for materials should be addressed to y.f .
( [email protected] ) . |
coronal seismology became an efficient new tool that uses standing mhd waves and oscillations as a tool to explore the physical parameters of the solar corona .
there are three basic branches of solutions of the dispersion relation for propagating and standing mhd waves : the slow mode branch ( with acoustic phase speeds ) , the fast mode branch and the alfvn branch ( with alfvnic phase speeds ) .
furthermore , each branch has a symmetric and asymmetric solution , termed the sausage and kink mode ( roberts et al .
all of these mhd oscillation modes have been now detected with imaging observations in recent years : transverse fast kink - mode oscillations with trace ( aschwanden et al . 1999 ; nakariakov et al .
1999 ) ; longitudinal ( slow - magnetoacoustic ) modes with sumer ( wang et al . 2002 ; ofman & wang 2002 ) , and fast sausage - mode oscillations probably with the _ nobeyama radioheliograph _ ( asai et al . 2001 ; melnikov , reznikova , & shibasaki 2002 ) .
the latter type of fast sausage - mode oscillations is the least established one , although numerous non - imaging radio observations exist that have been interpreted earlier in terms of this mode .
however , a recent study by nakariakov et al .
( 2003 ) pointed out that the dispersion relation and oscillation period has been incorrectly applied to the data , because the highly dispersive nature of the phase speed and the long - wavelength cutoff in the wave number has been ignored . here
we revisit the relevant earlier radio observations in the light of these new insights and attempt to derive the correct physical parameters .
the dispersion relation for magnetosonic waves in cylindrical magnetic flux tubes has infinitely many types of long - wavelength solutions in the fast - mode branch ( @xmath3 ) , with the lowest ones called sausage modes ( @xmath4 ) and kink modes ( @xmath5 ) .
the solutions are depicted in fig . 1 , showing that the kink mode solution extends all the way to the long - wavelength limit @xmath6 , while the sausage mode has a cutoff at a phase speed of v@xmath7v@xmath8 , which has no solution for small wave numbers @xmath9 .
the propagation cutoff for the sausage mode occurs at the cutoff wave number @xmath10 ( edwin & roberts 1983 ; roberts et al .
1984 ) , @xmath11^{1/2 } \left({j_{0,s } \over a } \right)\ , \quad s=1,2,3, ... \ ] ] where @xmath12 are the zeros of the bessel function @xmath13 . here
we restrict our attention to the genuine sausage mode with transverse mode number @xmath14 , i.e. @xmath15 . the cutoff frequency @xmath16 at the cutoff is @xmath17 . under coronal conditions ,
the sound speed amounts typically to @xmath18 km s@xmath19 ( with @xmath20 mk ) , which is much smaller than the typical alfvn speed ( @xmath21 km s@xmath19 ) so we have @xmath22 . in this case the tube speed , @xmath23 is similar to the sound speed , @xmath24 , and the expression for the cutoff wave number @xmath10 ( eq . 1 ) simplifies to , @xmath25^{1/2}\ ] ] in the low-@xmath26 corona , the thermal pressure is much smaller than the magnetic pressure , so that we can assume almost identical magnetic field strengths inside and outside of coronal loops , i.e. , @xmath27 , so that the ratio of external to the internal alfvn speed @xmath28 is essentially a density ratio @xmath29 , @xmath30^{1/2}\ ] ] from this expression we see that for typical density ratios inferred in the solar corona , e.g. , @xmath31 ( aschwanden et al . 2003 ) , the cutoff wave number @xmath32 falls into the range of @xmath33 . therefore , we would expect that a long - wavelength sausage mode oscillation is completely suppressed for the slender loops for which kink mode oscillations have been observed , which have wave numbers of @xmath34 ( aschwanden et al .
coronal conditions with @xmath35 were also inferred from fast kink mode oscillations ( nakariakov & ofman 2001 ) .
of course , higher harmonics would have correspondingly shorter wavelengths , i.e. @xmath36 with @xmath37 ) . on the other side ,
the short - wavelength limit is given by the maximum possible loop width ( @xmath38 ) , so the maximum wave number is @xmath39 .
the occurrence of global sausage mode oscillations therefore requires special conditions , @xmath40 , i.e. , very high density contrast @xmath41 and relatively thick loops to satisfy @xmath42 . the high density ratio , i.e. , @xmath43 or @xmath44 , yields the following simple expressions for the cutoff wave number @xmath10 , @xmath45 since the wavelength of the fundamental eigenmode ( with harmonic number @xmath46 in longitudinal direction and mode number @xmath14 in transverse direction ) corresponds to the double loop length , so that the wavenumber relates to the loop length by @xmath47 , the cutoff wave number condition @xmath42 implies a constraint between the loop geometry ratio @xmath48 and the density contrast @xmath49 , @xmath50 the numerical factor @xmath51 applies to the fundamental ( @xmath14 ) sausage mode . since geometric parameters ( such as @xmath52 and @xmath53 )
can be measured easier than densities , we might turn the cutoff condition around and formulate it as a density contrast requirement for a given loop aspect ratio , @xmath54 this clearly indicates that slender loops with a high length - to - width ratio @xmath55 would require extremely high density contrast @xmath41 .
typical active region loops , which have only a moderate density contrast in the order of @xmath56 , would be required to be extremely fat , with width - to - length ratios of @xmath57 .
the density contrast is much higher for flare loops or postflare loops , up to @xmath58 . in this case
, a length - to - width ratio of @xmath59 would be allowed for sausage mode oscillations .
this brings us to the conclusion that global sausage type oscillations are only expected in fat and dense loops , basically only in flare and postflare loops , a prediction that was not appreciated until recently ( nakariakov et al .
the restriction of the sausage mode wave number cutoff is visualized in fig .
2 , where the permitted range of geometric loop aspect ratios @xmath60 is shown as function of the density contrast @xmath61 . since the sausage mode is highly dispersive in the long - wavelength part of the spectrum ( see fig . 1 ) , the phase speed @xmath62 is a strong function of the wavenumber @xmath63 .
the phase speed diagram fig .
1 shows that the phase speed equals to the external alfvn speed at the long - wavelength cutoff , @xmath64 , and tends to approach the internal alfvn speed in the short - wavelength limit , @xmath65 . for coronal conditions ,
the phase speed of both the fast kink and fast sausage mode are bound by the internal and external alfvn velocities , @xmath66 therefore , also the period ( @xmath67 ) of the standing sausage mode is bound by these two limits , @xmath68 at the lower limit we can derive a simple relation for the sausage mode period from the long - wavelength cutoff , where @xmath69 , using eq .
6 , @xmath70 an approximation that is also derived in rosenberg ( 1970 ) and roberts et al .
however , one should be aware that this relation represents only a lower limit that applies at @xmath71 , while for all other valid wavenumbers ( @xmath42 ) the loop length is shorter ( @xmath72 ) and thus the global sausage mode period @xmath73 is also shorter ( nakariakov et al . 2003 ) , given by the actual phase speed @xmath74 defined by the dispersion relation ( fig .
1 ) at a particular value @xmath75 , @xmath76 provided that sufficiently fat and overdense loops exist according to the requirement ( eq . 6 or 7 ) , we expect for a loops radius of @xmath77 km and an alfvn speed of @xmath21 km s@xmath19 a typical sausage mode period of @xmath78 s ( according to eq .
table 1 contains a compilation of radio observations of oscillation events with periods in the range of 0.5 s @xmath79 1 min .
most of these observations have been interpreted in terms of mhd fast sausage mode oscillations , mainly for three reasons : ( 1 ) eigen - modes of standing mhd waves provide a natural mechanism for oscillations with a strictly regular period ; ( 2 ) the period range of @xmath1 s was expected for the mhd fast sausage mode under coronal conditions , while other mhd oscillations modes have significantly longer periods ; and ( 3 ) the mhd fast sausage mode modulates the loop density , magnetic field strength , and related radio emission ( plasma emission or gyrosynchrotron emission ) , while the mhd kink mode does not modulate the loop density in first order .
in addition we might add a forth reason which has not been appreciated previously : these events listed in table 1 were generally reported to occur during flares , which explains the high densities that are expected on theoretical grounds ( due to the sausage mode wave number cutoff criterion @xmath80 , see eq .
7 ) , which does not occur for the mhd kink mode . because the loop density plays a critical role for mhd fast sausage mode oscillations ( eq .
7 ) , we listed in table 1 only observations that have information on the period @xmath81 as well on the electron density @xmath82 of the oscillating loops .
there are only two observations with direct electron density measurements , using soft x - ray emission measures and constraints from the microwave spectrum , which report electron densities of @xmath83 @xmath84 ( asai et al .
2001 ) and @xmath85 @xmath84 ( melnikov et al . 2002 ) .
however , we might also extract information on the electron density from any radio burst that is produced by an emission mechanism near the plasma frequency , since the plasma frequency depends only on the electron density @xmath82 , @xmath86 } \qquad { \rm [ hz ] } \ .\ ] ] plasma emission is the dominant emission mechanism for metric and decimetric radio bursts up to frequencies of @xmath87 ghz , while free - free emission , gyroresonance , and gyrosynchrotron emission dominate at higher frequencies ( e.g. , see dulk 1985 ) . a cursory look at table 1 shows that most of the reported oscillations fall in the frequency range of @xmath88 mhz , so this corresponds to electron densities of @xmath89 @xmath84 in the case of plasma emission . in fig .
3 we display the observed oscillation periods @xmath81 as function of the observed radio frequency @xmath90 .
the observed radio frequency uniquely constrains the density @xmath82 of the oscillating loop segment , for the case of plasma emission . in order to estimate the external density @xmath91
we can use a standard model of the average background corona , for instance the baumbach - allen model , which as function of the normalized height @xmath92 is , @xmath93 \approx { 4\times 10 ^ 8 \over r^9 } \quad [ { \rm cm}^{-3}]\ ] ] where the approximation on the right - hand - side holds for @xmath94 .
this density model @xmath95 is shown in fig .
4 ( top ) . assuming that the fattest possible loops have a width - to - length ratio of @xmath96 , the minimum required density ratio is @xmath97 ( eq . 7 ) for loops oscillating in the mhd fast sausage mode due to the long - wavelength cutoff criterion ( eq . 5 ) .
this cutoff criterion implies the constraint that physical solutions are only possible for heights @xmath98 mm , for oscillations observed in the plasma frequency range of @xmath99 ghz ( grey area in fig.4 top ) .
the two events observed by asai et al .
( 2001 ) and melnikov et al .
( 2002 ) do not fall in this range , because their density is higher and would correspond to plasma frequencies of @xmath100 and 3 ghz . the important conclusion that follows , based on the two assumptions that the baumbach - allen model represents a realistic density model of the background corona and that oscillating loops are not fatter than a quarter of their length , is that physical solutions for the mhd fast sausage mode are only possible for coronal heights of @xmath98 mm in the observed frequency range of @xmath101 ghz , where most of the fast oscillations have been observed ( fig . 3 ) .
next we investigate the constraints introduced by the pulse periods , which are found in the range of @xmath102 s ( fig .
3 ) . denoting the length of the loop segment that participates in the mhd fast sausage mode oscillation with @xmath103
, we obtain an upper limit from the phase speed at the wavenumber cutoff @xmath10 ( eqs . 9 , 11 ) , @xmath104 where the alfvnic speed @xmath105 is given by the background density model @xmath95 ( eq . 13 ) , @xmath106\ , \ ] ] where we used a mean molecular weight of @xmath107 for the h : he=10:1 coronal abundance .
we plot the sausage mode period @xmath73 as function of the wavelength @xmath103 in fig . 4 ( bottom ) , for a mean coronal magnetic field of @xmath108 g ( solid line ) , as well as for a factor of 2 smaller or larger values ( @xmath109 g ) .
we find that the corresponding wavelength of the oscillating loop segments have values in the range of @xmath110 mm @xmath111 10 mm ( gray area in fig
. 4 bottom ) for pulse periods of @xmath112 s , which is much smaller than the full loop length ( @xmath113 mm , see gray area in fig
. 4 top ) required for the fundamental sausage mode . because physical solutions are only possible in heights @xmath114 mm , the oscillating loop segment has to be located near the top of the loop .
thus we conclude that only a small segment of the loop is oscillating in the mhd sausage mode .
such small oscillating loop segments require wavelengths that correspond to higher harmonics @xmath115 of the fast sausage mhd mode , @xmath116 while lower harmonics can not oscillate due to the wave number cutoff .
the situation is illustrated for the case @xmath117 in fig . 5 : for instance , if the cutoff is @xmath118 , only wavelengths with @xmath119 are able to oscillate , which is most likely to occur at the loop top where the density contrast and wavelength match the sausage wavenumber cutoff criterion .
alternatively , oscillations at a higher harmonic could occur along the entire loop , but favorable conditions for detection could be restricted to certain segments only ( e.g. , due to line - of - sight angles or absorption of plasma emission ) . an alternative interpretation of these fast radio pulsations in the decimetric range could be made in terms of propagating fast sausage mode mhd waves , where the wavelength is not prescribed by the loop length , but independently by the driver .
this interpretation has been proposed in roberts et al .
however , if this would be the case , the radio dynamic spectra should show parallel drifting bands of emission for every propagating node .
such decimetric burst types have indeed been observed and are called _ decimetric fiber bursts _
( e.g. , rosenberg 1972 ; bernold 1980 ; slottje 1981 ) , but many of the observations listed in table 1 show narrowband pulsations without parallel drifting bands , and thus are not consistent with ( impulsively generated ) propagating fast sausage mhd waves .
note that the measurements of asai et al .
( 2001 ) and melnikov et al . (
2002 ) have longer periods and higher densities than all decimetric observations listed in table 1 , and thus actually allow for global sausage mode oscillations of the entire loop length , for reasonable magnetic fields in the order of @xmath120 g ( marked with crosses in fig
. 4 bottom ) .
this study leads us to the following conclusions : 1 . only the two imaging observations of fast oscillation events ( @xmath121 s ) by asai et al .
( 2001 ) and melnikov et al .
( 2002 ) are consistent with global mhd fast sausage mode oscillations , satisfying the relation between period and loop lengths as well as the wave number cutoff criterion for the fundamental sausage mode .
the required magnetic field is of the order of @xmath120 g. 2 .
all earlier radio observations of fast oscillation events , which have been reported in the frequency range of @xmath122100 mhz @xmath123 1 ghz and with periods of @xmath124 s , imply for reasonable coronal magnetic field strengths ( @xmath125 g ) that the length of the loop segments participating in the mhd fast sausage mode oscillation is much smaller ( @xmath126 mm ) than the full loop length ( @xmath127 mm ) .
the sausage mode oscillation seems to be confined to segments where the largest density contrast to the background occurs , most likely near the loop top , in a segment corresponding to a higher harmonic node ( see fig . 5 ) .
3 . the wave number cutoff criterion seems to introduce a bimodality of mhd fast mode sausage oscillation phenomena : ( 1 ) partial loop oscillations confined to loop segments ( @xmath128 ) of high harmonic @xmath129 nodes for low - density ( @xmath130 @xmath84 ) loops that are observable at the plasma frequency in the metric / decimetric frequency range ( @xmath101 ghz ) ; and ( 2 ) global loop oscillations ( @xmath131 ) at the fundamental mode ( @xmath46 ) for high - density ( @xmath132 @xmath84 ) loops that are observable in soft x - rays and via gyrosynchrotron emission in microwaves . in this study
we explored the diagnostic of mhd fast sausage mode oscillations .
it is important to notice that the observed period does not provide a diagnostic of the alfvnic travel time across the loop diameter ( @xmath133 ) , as it was commonly believed , but rather provides a lower limit of this quantity ( nakariakov et al .
moreover , the new finding that fast periods are likely to be caused by oscillations of partial loop segments ( at higher harmonic nodes ) , implies that the magnetic field @xmath134 can not be determined by the loop diameter and sausage mode period , unless the harmonic number can be constrained from imaging observations .
this study illustrates specific constraints that are contingent on the sausage wave number cutoff , which have been ignored in previous studies .
[ abrami , a .
1970 , solar phys . ,
11 , 104 . ]
[ abrami , a .
1972 , nature 238/80 , 25 . ]
[ achong , a .
1974 , solar phys . ,
37 , 477 - 482 . ]
[ asai , a . , shimojo , m. , isobe , h. , morimoto , t. , yokoyama , t. , shibasaki , k. , and nakajima , h. 2001 , apj 562 , l103 . ]
[ aschwanden , m.j . 1986 , solar phys . , 104 , 57 - 65 . ]
[ aschwanden , m.j . &
benz , a.o .
1986 , aa , 158 , 102 - 112 . ]
[ aschwanden , m.j .
1987b , _ pulsations of the radio emission of the solar corona .
analysis of observations and theory of the pulsating electron - cyclotron maser _ , phd thesis , eth zurich , pp.1 - 173 . ]
[ aschwanden , m.j . , bastian , t.s . &
gary , d.e .
, 1992c , _ bull.amer.astron.soc . _ 24/2 , 802 . ]
[ aschwanden , m.j . ,
benz , a.o . , & montello , m .
1994a , apj , 431 , 432 - 449 . ]
[ aschwanden , m.j . ,
benz , a.o .
dennis , b.r .
& kundu , m.r .
1994b , apjs , 90 , 631 - 636 . ]
[ aschwanden , m.j . ,
fletcher , l . , schrijver , c . , and alexander , d . , 1999 ,
apj , , 520 , 880 . ]
[ aschwanden , m.j . ,
depontieu , b . ,
schrijver , c.j . , and title , a .
2002 , solar phys . , 206 , 99 . ] [ aschwanden , m.j . ,
nightingale , r.w . ,
andries , j .
, goossens , m . , and van doorsselaere , t .
2003 , apj , subm . ]
[ aurass , h . ,
chernov , g.p . ,
karlicky , m . , kurths , j . , & mann , g .
1987 , solar phys .
, 112 , 347 - 357 . ]
[ bernold , t.e.x .
1980 , aas 42 , 43 - 58 . ]
[ chernov , g.p . &
kurths , j 1990 , sov .
34(5 ) , 516 - 522 . ] [ chernov , g.p . , markeev , a.k . , poquerusse , m . , bougeret , j.l . , klein , k.l .
mann , g . ,
aurass , h . , & aschwanden , m.j .
1998 , aa , 334 , 314 - 324 . ]
[ degroot , t .
1970 , solar phys .
, 14 , 176 . ] [ drge , f .
1967 , zeitschr.astrophys .
66 , 200 . ]
[ dulk , g.a . 1985 ,
astron . astrophys .
23 , 169 . ]
[ edwin , p.m . and roberts , b . 1983 , solar phys . , 88 , 179 . ] [ elgaroy , . 1980 ,
aa , 82 , 308 - 313 . ]
[ gotwols , b.l .
1972 , solar phys . ,
25 , 232 - 236 . ]
[ kai , k .
& takayanagi , a .
1973 , solar phys . ,
29 , 461 - 475 . ]
[ kliem , b . ,
karlicky , m . , & benz , a.o .
2000 , aa , 360 , 715 - 728 . ]
[ kurths , j .
& herzel , h .
1986 , solar phys . , 107 , 39 - 45 . ] [ kurths , j . &
karlicky , m . 1989 , solar phys . , 119 , 399 - 411 . ]
[ kurths , j . ,
benz , a.o . , & aschwanden , m.j .
1991 , aa , 248 , 270 - 276 . ]
[ mclean , d.j , sheridan , k.v . , steward , r.t .
, & wild , j.p . 1971 , nature 234 , 140 - 142 . ] [ mclean , d.j .
& sheridan , k.v . 1973 ,
solar phys . , 32 , 485 - 489 . ] [ melnikov , v.f . ,
reznikova , v.e . , & shibasaki , k .
2002 , in proc . of intern .
_ active processes on the sun and stars _ , ( eds .
zaitsev , v.v . and yasnov , l.v .
, p.225 . ]
[ nakariakov , v.m . ,
ofman , l . ,
deluca , e . ,
roberts , b . , davila , j.m .
1999 , science 285 , 862 . ]
[ nakariakov , v.m .
& ofman , l . 2001 , aa , 372 , l53 . ]
[ nakariakov , v.m . ,
melnikov , v.f . , and reznikova , v.e .
2003 , aa , , subm . ]
[ ofman , l . and wang , t.j .
2002 , apj , 580 , l85 . ]
[ pick , m. & trottet , g. 1978 , solar phys . , 60 , 353 - 359 . ]
[ roberts , b . , edwin , p.m . , and benz , a.o .
1984 , apj , 279 , 857 . ]
[ rosenberg , h .
1970 , aa , 9 , 159 . ]
[ rosenberg , h .
1972 , solar phys . ,
25 , 188 - 196 . ]
[ sastry , ch.v . ,
krishan , v . , & subramanian , k.r .
1981 , j. astrophys .
astron . 2 , 59 - 65 . ]
[ slottje , c. 1981 , _ atlas of fine structures of dynamic spectra of solar type iv - dm and some type ii radio bursts _ ,
dwingeloo , the netherlands . ]
[ tapping , k.f .
1978 , solar phys . ,
59 , 145 - 158 . ]
[ trottet , g. , kerdraon , a. , benz , a.o . , & treumann , r. 1981 , aa , 93 , 129 - 135 . ] [ wang , t.j . ,
solanki , s.k .
, curdt , w . ,
innes , d.e . , and dammasch , i.e .
2002 , apj , 574 , l101 . ]
[ wiehl , h.j . ,
benz , a.o .
, & aschwanden , m.j .
1985 , solar phys . ,
95 , 167 - 179 . ]
[ zaitsev , v.v . ,
stepanov , a.v . , & chernov , g.p .
1984 , solar phys . ,
93 , 363 - 377 . ]
[ zlobec , p. , messerotti , m. , dulk , g.a . , & kucera , t. 1992 , solar phys .
, 141 165 - 180 . ]
drge ( 1967 ) & @xmath135 & 240 , 460 mhz & 0.2 - 1.2 & kiel + rosenberg ( 1970 ) & 1 & 220 - 320 mhz & 1.0 - 3.0 & utrecht + de groot ( 1970 ) & @xmath136 & 250 - 320 mhz & 2.2 - 3.5 & utrecht + abrami ( 1970 , 1972 ) & 3 & 239 mhz & 1.7 - 3.1 & trieste + mclean et al . ( 1971 ) & 1 & 100 - 200 mhz & 2.5 - 2.7 & culgoora + rosenberg ( 1972 ) & 1 & 220 - 320 mhz & 0.7 - 0.8 & utrecht + gotwols ( 1972 ) & 1 & 565 - 1000 mhz & 0.5 & silver spring + kai & takayanagi ( 1973 ) & 1 & 160 mhz & @xmath1371.0 & 17 ( nobeyama ) + mclean & sheridan ( 1973 ) & 1 & 200 - 300 mhz & [email protected] & culgoora + achong ( 1974 ) & 1 & 18 - 28 mhz & 4.0 & kingston + tapping ( 1978 ) & 14 & 140 mhz & 0.06 - 5 & cranleigh + pick & trottet ( 1978 ) & 1 & 169 mhz & 0.37 , 1.7 & 5-7 ( nanay ) + elgaroy ( 1980 ) & 8 & 310 - 340 mhz & 1.1 & oslo + bernold ( 1980 ) & @xmath139 & 100 - 1000 mhz & 0.5 - 5 & zurich + slottje ( 1981 ) & @xmath140 & 160 - 320 mhz & 0.2 - 5.5 & dwingeloo + trottet et al . (
1981 ) & 1 & 140 - 259 mhz & [email protected] & 5 ( nanay ) + sastry et al .
( 1981 ) & 1 & 25 - 35 mhz & 2 - 5 & 30 ( gauribidanur ) + zlobec et al .
( 1992 ) & 1 & 333 mhz & 9.8 - 14.2 & 0.7-1.5 ( vla ) + zaitsev et al . ( 1984 ) & 23 & 45 - 230 mhz & 0.3 - 5 & izmiran + wiehl et al . ( 1985 ) & 1 & 300 - 1000 mhz & 1 - 2 & zurich + aschwanden ( 1986 , 1987b ) & 10,60 & 300 - 1100 mhz & 0.4 - 1.4 & zurich + aschwanden & benz ( 1986 ) & 10 & 237 , 650 mhz & 0.5 - 1.5 & zurich + kurths & herzel ( 1986 ) & 1 & 480 - 800 mhz & 1.0 , 3.5 & tremsdorf + aurass et al .
( 1987 ) & 1 & 234 mhz & 0.25 - 2 & tremsdorf + kurths & karlicky ( 1989 ) & 1 & 234 mhz & 1.3 , 1.5 & tremsdorf + chernov & kurths ( 1990 ) & 10 & 224 - 245 mhz & 0.35 - 1.3 & izmiran + kurths et al .
( 1991 ) & 25 & 234 - 914 mhz & 0.07 - 5.0 & zurich + aschwanden et al .
( 1992c ) & 1 & 1.5 ghz & 8.8 & 0.2-0.9 ( vla , ovro ) + aschwanden et al .
( 1994a , b ) & 1,1 & 300 - 650 mhz & 1.15 , 1.8 & zurich + chernov et al .
( 1998 ) & 1 & 164 - 407 mhz & 0.2 & 5-7(nanay ) + kliem et al .
( 2000 ) & 1 & 600 - 2000 mhz & 0.5 - 3.0 & zurich + asai et al .
( 2001 ) & 1 & 17 ghz & 6.6 & 10 `` ( nobeyama ) + melnikov et al .
( 2002 ) & 1 & 17 , 34 ghz & 8 - 11 , 14 - 17 & 10'',5 " ( nobeyama ) + | a recent study by nakariakov et al . pointed out that the dispersion relation of mhd sausage mode oscillations has been incorrectly applied to coronal loops , neglecting the highly dispersive nature of the phase speed and the long - wavelength cutoff of the wave number . in the light of these new insights
we revisit previous observations that have been interpreted in terms of mhd sausage mode oscillations in coronal loops and come to the following conclusions : ( 1 ) fast sausage mhd mode oscillations require such a high electron density imposed by the wave number cutoff that they can only occur in flare loops ; ( 2 ) in the previously reported radio observations ( @xmath0 mhz to 1 ghz ) with periods of @xmath1 s , the fast sausage mhd mode oscillation is likely to be confined to a small segment ( corresponding to a high harmonic node ) near the apex of the loop , rather than involving a global oscillation over the entire loop length . the recent microwave and soft x - ray observations of fast periods ( @xmath2 s ) by asai et al . and melnikov et al .
, however , are consistent with fast sausage mhd oscillations at the fundamental harmonic . # 1#2 * figure # 1 : * _ _ # 2 _ _ # 1 epsf |
neutrino flavor oscillations have been observed in many solar , reactor and accelerator experiments ( see the recent reviews in refs .
@xcite ) , in agreement with the currently standard paradigm of three - neutrino ( @xmath2 ) mixing .
the global fits of neutrino oscillation data in the framework of @xmath2 mixing @xcite give us rather precise information on the values of the elements of the three mixing angles which parameterize the neutrino mixing matrix and on the values of the two independent neutrino squared - mass differences , the smaller `` solar '' squared - mass difference @xmath4 and the larger `` atmospheric '' squared - mass difference @xmath5 .
however , the standard @xmath2 mixing paradigm has been challenged by indications in favor of short - baseline oscillations generated by a new larger squared - mass difference @xmath6 : the reactor antineutrino anomaly @xcite , which is a deficit of the rate of @xmath7 observed in several short - baseline reactor neutrino experiments in comparison with that expected from the latest calculation of the reactor neutrino fluxes @xcite ; the gallium neutrino anomaly @xcite , consisting in a short - baseline disappearance of @xmath8 measured in the gallium radioactive source experiments gallex @xcite and sage @xcite ; the signal of short - baseline @xmath9 oscillations observed in the lsnd experiment @xcite .
the simplest extension of @xmath2 mixing which can describe these short - baseline oscillations taking into account other constraints is the 3 + 1 mixing scheme @xcite , in which there is an additional massive neutrino at the ev scale and the masses of the three standard neutrinos are much smaller .
since from the lep measurement of the invisible width of the @xmath10 boson we know that there are only three active neutrinos ( see ref .
@xcite ) , in the flavor basis the additional massive neutrino corresponds to a sterile neutrino @xcite , which does not have standard weak interactions . a fundamental questions that remains open
is : are neutrinos dirac or majorana particles ?
this question can not be investigated in neutrino oscillation experiments , where the total lepton number is conserved and there is no difference between dirac neutrinos with a conserved total lepton number and truly neutral majorana neutrinos , which do not have a conserved total lepton number . the most promising process which can reveal the majorana nature of neutrinos is neutrinoless double - beta decay , in which the total lepton number changes by two units ( see the recent review in ref .
@xcite ) . from the present knowledge of the neutrino squared - mass differences and mixing angles it is possible to predict the possible range of values for the effective majorana mass @xmath0 in neutrinoless double - beta decay as a function of the absolute scale of neutrino masses
( see ref .
@xcite ) , which is still unknown , up to the upper bound of about 2 ev at 95% c.l . established by the mainz @xcite and troitsk @xcite tritium @xmath1-decay experiments .
the introduction of a sterile neutrino at the ev mass scale can change dramatically the prediction for the possible range of values for the effective majorana mass in neutrinoless double - beta decay @xcite . in this paper
we present accurate predictions for @xmath0 taking into account the results of the global fit of solar , atmospheric and long - baseline reactor and accelerator neutrino oscillation data presented in ref .
@xcite and the results of an update @xcite of the global fit of short - baseline neutrino oscillation data presented in ref .
we are particularly interested to determine accurately the conditions for which @xmath11 , which may be probed experimentally in the near future ( see refs .
@xcite ) , and the conditions for which there can be a cancellation of the different mass contributions to @xmath0 , which leads to an unfortunate uncertainty for the possibility of ever observing neutrinoless double - beta decay ( unless it is induced by new interactions and/or the exchange of new particles ; see refs .
@xcite ) .
the plan of this paper is as follows . in section [ sec:3nu ]
we discuss the predictions for @xmath0 in the standard @xmath2 framework , taking into account the two possible normal and inverted mass orderings . in section [ sec:3p1 ]
we discuss how these predictions are modified in the 3 + 1 mixing framework . in section [ sec : conclusions ] we draw our conclusions .
in the standard three - neutrino ( @xmath2 ) mixing framework , the effective majorana mass in neutrinoless double - beta decay is given by @xmath12 where @xmath13 is the partial contribution of the massive majorana neutrino @xmath14 with mass @xmath15 .
the elements @xmath16 of the mixing matrix , which quantify the mixing of the electron neutrino with the three massive neutrinos , can have unknown complex phases , which generate the two complex phases @xmath17 and @xmath18 in eq .
( [ mbb3nu ] ) .
since the values of these phases is completely unknown , all predictions of the value of @xmath0 must take into account all the possible range of these phases , from 0 to @xmath19 .
we use the results of the global fit of solar , atmospheric and long - baseline reactor and accelerator neutrino oscillation data presented in ref .
@xcite , which are given in terms of the mixing angles @xmath20 , @xmath21 that determine the absolute values of the first row of the mixing matrix @xmath22 in the standard parameterization : @xmath23 the results for the neutrino squared - mass differences are expressed in terms of the solar and atmospheric squared mass differences , which are defined by @xmath24 where @xmath25 . given this assignment of the squared mass differences , it is currently unknown if the ordering of the neutrino masses is normal ( no ) , such that @xmath26 or inverted ( io ) , such that @xmath27 .
we discuss these two cases separately in the following subsections . ,
@xmath28 and @xmath29 allowed intervals of the three partial mass contributions to @xmath0 in eq .
( [ mbb3nu ] ) as functions of the lightest mass @xmath30 in the case of @xmath2 mixing with normal ordering . ]
( see eq .
( [ mbbpm ] ) ) as a function of @xmath30 for the best - fit values of the partial mass contributions in the case of @xmath2 and 3 + 1 mixing with normal ordering of the three lightest neutrinos . ] as a function of the lightest neutrino mass @xmath30 in the case of @xmath2 mixing with normal ordering .
the signs in the legend indicate the signs of @xmath31 for the four possible cases in which cp is conserved . the intermediate yellow region is allowed only in the case of cp violation .
the 90% upper limit is explained in the main text . ] as a function of effective electron neutrino mass @xmath32 in eq .
( [ mb ] ) in the case of @xmath2 mixing with normal ordering .
the legend is explained in the caption of fig .
[ 3nunombbvsmin ] .
the limits are explained in the main text . ] as a function of sum of the neutrino masses @xmath33 in eq .
( [ sum ] ) in the case of @xmath2 mixing with normal ordering .
the legend is explained in the caption of fig .
[ 3nunombbvsmin ] .
the limits are explained in the main text . ]
.ranges of @xmath30 , @xmath32 and @xmath33 for which there can be a complete cancellation of the three partial mass contributions to @xmath0 for the best - fit values ( b.f . ) of the oscillation parameters and at @xmath34 , @xmath28 and @xmath29 in the case of @xmath2 mixing with normal ordering . [ cols="^,^,^,^,^",options="header " , ] [ tab:3p1-inv ] allowed regions in the @xmath32@xmath0 plane in the cases of @xmath2 and 3 + 1 mixing with inverted ordering of the three lightest neutrinos . ] allowed regions in the @xmath33@xmath0 plane in the cases of @xmath2 and 3 + 1 mixing with inverted ordering of the three lightest neutrinos . ] the best - fit value and the @xmath34 , @xmath28 and @xmath29 allowed intervals of the partial mass contributions to @xmath0 in the case of 3 + 1 mixing with inverted ordering of the three lightest neutrinos are shown in fig .
[ 3p1iopartial ] .
one can see that there can be a total cancellation between the partial contribution @xmath35 and the dominant @xmath36 in the case of @xmath2 mixing ( see fig .
[ 3nuiopartial ] ) for @xmath37 .
indeed , fig .
[ ratiombbvsmin ] shows that the value of @xmath38 ( see eq .
( [ mbbpm ] ) ) corresponding to the best - fit values of the partial mass contributions is negative for @xmath39 .
figure [ 3p1iombbvsmin ] depicts the allowed regions in in the @xmath40@xmath0 plane . comparing fig .
[ 3p1iombbvsmin ] with fig .
[ 3nuiombbvsmin ] one can see that the predictions for @xmath0 are completely different in the @xmath2 and 3 + 1 cases if there is an inverted ordering of the three lightest neutrinos , in agreement with the discussions in refs .
the ranges of values of @xmath40 for which there can be a complete cancellation of @xmath0 are given in tab .
[ tab:3p1-inv ] .
figures [ 3p1iombbvsmb ] and [ 3p1iombbvssum ] show the allowed regions in the @xmath32@xmath0 and @xmath33@xmath0 planes . figures [ compiombbvsmb ] and [ compiombbvssum ] show the comparison of the @xmath29 allowed regions in the same planes in the cases of @xmath2 and 3 + 1 mixing with inverted ordering of the three lightest neutrinos .
if the inverted ordering will be established by oscillation experiments ( see refs .
@xcite ) , it will be possible to exclude @xmath2 mixing in favor of 3 + 1 by restricting @xmath32 and @xmath0 or @xmath33 and @xmath0 in the corresponding large region at small @xmath0 allowed only in the 3 + 1 case .
we have presented accurate calculations of the effective majorana mass @xmath0 in neutrinoless double-@xmath1 decay in the standard case of @xmath2 mixing and in the case of 3 + 1 neutrino mixing indicated by the reactor , gallium and lsnd anomalies ( see refs .
we have taken into account the uncertainties of the standard @xmath2 mixing parameters obtained in the global fit of solar , atmospheric and long - baseline reactor and accelerator neutrino oscillation data presented in ref .
@xcite and the uncertainties on the additional mixing parameters in the 3 + 1 case obtained from an update @xcite of the global fit of short - baseline neutrino oscillation data presented in ref .
@xcite .
we have shown that the predictions for @xmath0 in the cases of @xmath2 and 3 + 1 mixing are quite different , in agreement with the previous discussions in refs .
our paper improves these discussions by taking into account the uncertainties of all the mixing parameters and presenting all the results at @xmath34 , @xmath28 and @xmath29 .
we have presented accurate comparisons of the allowed regions in the planes @xmath32@xmath0 and @xmath33@xmath0 of measurable quantities , taking into account the two possibilities of normal and inverted ordering of the three light lightest neutrinos .
we have shown that future measurements of these quantities may distinguish the @xmath2 and 3 + 1 cases if the mass ordering is determined by oscillation experiments ( see refs .
@xcite ) .
e. z. thanks the support of funding grants 2013/02518 - 7 and 2014/23980 - 3 , so paulo research foundation ( fapesp ) .
the work of c. giunti is partially supported by the research grant _ theoretical astroparticle physics _
number 2012cppyp7 under the program prin 2012 funded by the ministero dellistruzione , universit e della ricerca ( miur ) . | we present accurate predictions of the effective majorana mass @xmath0 in neutrinoless double-@xmath1 decay in the standard case of @xmath2 mixing and in the case of 3 + 1 neutrino mixing indicated by the reactor , gallium and lsnd anomalies .
we have taken into account the uncertainties of the neutrino mixing parameters determined by oscillation experiments .
it is shown that the predictions for @xmath0 in the cases of @xmath2 and 3 + 1 mixing are quite different , in agreement with previous discussions in the literature , and that future measurements of neutrinoless double-@xmath1 decay and of the effective light neutrino mass in @xmath1 decay or the total mass of the three lightest neutrinos in cosmological experiments may distinguish the @xmath2 and 3 + 1 cases if the mass ordering is determined by oscillation experiments .
we also present a relatively simple method to determine the minimum value of @xmath0 in the general case of @xmath3-neutrino mixing .
published in : * c. giunti , e.m .
zavanin , jhep 1507 ( 2015 ) 171 * |
in the last several years , the basic model of particle acceleration above a pulsar polar cap ( pc ) has been undergoing significant revision .
sturrock ( 1971 ) , ruderman & sutherland ( 1975 ) and arons & scharlemann ( 1979 ) originally proposed that particles are accelerated by an induced electric field , producing curvature radiation ( cr ) photons which create electron - positron pairs in the strong magnetic field .
the pairs short out the electric field above a pair - formation front ( pff ) , self - limiting the acceleration . in the process of screening or shorting - out the electric field ,
some fraction of the positrons is accelerated back toward the pcs and heats the surface of the neutron star ( ns ) , producing a potentially observable x - ray emission component ( arons 1981 ) .
two recent developments have introduced important changes to this picture .
first was the finding of muslimov & tsygan ( 1992 ) that the effect of inertial - frame dragging near the ns surface greatly increases the induced electric field @xmath2 above the pc in space - charge limited flow models ( arons & scharlemann 1979 ) .
the second was the realization that inverse - compton radiation of the primary particles can produce pairs which could potentially screen the electric field ( zhang et al .
1997 , harding & muslimov 1998 ; hereafter hm98 ) .
we have investigated the electric field screening and pc heating in the revised space - charge limited flow ( sclf ) model . in the first paper ( harding & muslimov 2001 ; paper i ) , we have presented our results for screening and pc heating by cr pffs , as in arons ( 1981 ) , but using the frame - dragging electric field of muslimov & tsygan ( 1992 ) .
paper i outlined a self - consistent calculation of the pff height , returning positron flux and the screening scale length , where the @xmath2 and the primary flux are adjusted for the change in charge density caused by the returning positrons .
a main assumption of this calculation was that the screening scale is small compared to the height of the pff , the location where the first pairs are produced .
this turns out to be generally true for screening by cr - produced pairs because the pair cascade multiplicity grows rapidly over small distances due to the strong dependence of cr photon energy on particle energy .
thus , the existence of a cr pff always results in pc heating and full screening of @xmath2 .
we found that the pc heating luminosity , as a fraction of the spin - down luminosity , increases with pulsar characteristic age , @xmath3 , and should be detectable for pulsars with @xmath4 yr .
the most significant heating occurs for pulsars near the death line for cr pair production , which is @xmath5 yr for normal period pulsars and @xmath6 yr for millisecond pulsars .
our predicted x - ray luminosity due to pc heating is about a factor of ten higher than the x - ray luminosity predicted by arons ( 1981 ) , due to the increase in accelerating voltage drop resulting from the inclusion of inertial - frame dragging effects .
we also predicted that older pulsars should have higher pc surface temperatures from heating . in this second paper ( paper ii ) , we present results of our investigation of electric field screening and pc heating by inverse - compton scattering ( ics ) radiation pffs .
we investigate pc heating by ics produced pairs in all pulsars , including those which produce cr pairs .
for the older pulsars that do not produce pairs through cr , pc heating by positrons from ics cascades is especially important .
our treatment of the ics pffs follows closely that of hm98 , where cyclotron - resonant and non - resonant scattering are considered as separate radiation components .
hm98 found that the ics pffs are located much closer to the ns surface than are the cr pffs , because primaries with lorentz factor of only @xmath7 can produce pairs via ics whereas lorentz factors of at least @xmath8 are required for production of pairs via cr .
they also found that the ics pffs of positrons returning from the upper pff occur at a significant distance above the surface , so it is possible that ics pairs fronts are unstable if screening of @xmath2 at the lower pff occurs .
however , hm98 did not determine whether the ics pairs were capable of screening either at the upper or lower pffs . this paper will attempt to answer those questions with a calculation similar to that of paper i. in section ii
, we describe the acceleration model for this calculation .
we use a solution for @xmath2 from poisson s equation which imposes an upper boundary condition requiring that @xmath9 at the location of screening ( as in hm98 ) , which differs from that of paper i which used a solution with no upper boundary .
the solution with an upper boundary is more appropriate for screening at ics pffs which form close to the surface , where the screening scale is generally comparable to the pff height .
this solution is required here because pairs from ics photons may screen @xmath2 close to the ns surface . in this case
the value of @xmath2 is suppressed due to proximity of the screening to the surface .
section ii also presents a calculation of ics pff height and the location of a new ics pair death line .
we find that virtually all known pulsars can produce pairs by either resonant ics ( rics ) or non - resonant ics ( nrics ) in a pure dipole field .
hibschmann & arons ( 2001 ) ( hereafter ha01 ) have recently taken a different approach to determining pulsar death - lines including ics produced pairs .
they define the pff as the location where the pair multiplicity achieves that required for complete screening of the @xmath2 , whereas we define the pff as the location where pair production begins ( i.e. where the first pairs are produced ) . in the case of cr , this distinction is minor since the screening scale is small compared to the pff height , but in the case of ics the distinction is very important since the screening scale is comparable to the pff height .
ha01 also assume that the pair multiplicity required for screening is determined by the difference between the actual charge and the goldreich - julian charge at a distance from the ns surface that is roughly equal to the pc radius , whereas in our calculation the charge density required for screening is determined by the charge deficit at the location of creation of pairs which can be much smaller . with their more restrictive definition of the pff ,
their death lines differ significantly from ours . in section iii
, we give self - consistent solutions for the fraction of positrons returning to the pc and for the screening scale .
we also explore the question of whether positrons returning from the upper pff can screen @xmath2 near the lower pff close to the ns surface .
the subject of pulsar pair death lines , i.e. which nss are capable of pair production , will be discussed in section iv . in section
v , we present our calculations of pc heating luminosity due to returning positrons , giving both numerical and analytic estimates .
summary and conclusions , as well as a comparison of results from both papers i and ii will be given in section vi .
in this paper we exploit the same model for charged particle acceleration that is described in paper i. namely , in the acceleration region we use the appropriate solution for the electric field and potential presented in paper i. for example , for most ics pair fronts , where the screening occurs over the length scale smaller than the pc size , we use our formulae ( a7)-(a10 ) of paper i , whereas in the screening region ( see 3.2 for details ) for all cases we model the electric field by equation ( 26 ) of paper i. for our analytic estimates we shall use the same simplified expressions for the accelerating electric field ( cf .
[ 34 ] , [ 35 ] of paper i ) , e_6 = b_12p _ 0.15 \ { ll 1.3 z p^-1/2 & , + 510 ^ -3 p^-1 & , .
[ epar ] where the upper expression corresponds to the unsaturated regime ( rising part of the accelerating field ) , and the lower - to the saturated regime ( nearly constant accelerating field ) . equation ( [ epar ] ) assumes that @xmath10 and @xmath11 ; @xmath12 esu , @xmath13 , @xmath14 g , and @xmath15 is the altitude in units of stellar radius , where @xmath16 is the magnetic colatitude in units of the pc half - angle , @xmath17 is the angle between the magnetic and spin axes , @xmath18 is the component of the electric field parallel to the magnetic field , @xmath19 is the surface value of the magnetic field strength , @xmath20 is the dimensionless general relativistic parameter originating from the frame - dragging effect and accounting for the stellar compactness and moment of inertia .
note that throughout this paper in all practical formulae we assume that @xmath21 is dimensionless value of pulsar spin period measured in seconds .
also , in all our analytic estimates based on the above formula we shall discriminate between unsaturated and saturated regimes of acceleration , with the unsaturated / saturated regime occuring in the case where the characteristic altitude of pair formation is smaller / larger than the pc radius .
the formal criteria corresponding to these regimes , e.g. in b - p diagram , depend on the radiation mechanism for pair producing photons and can be derived from the conditions @xmath22 and @xmath23 for unsaturated and saturated regimes , respectively , where @xmath24 is the characteristic altitude of screening scaled by the pc radius ( see eqs .
[ [ zeta*cr]]-[[zeta*nr ] ] below ) .
these criteria translate into p p _ _ unsaturated regime _ and p > ~p _ _ saturated regime _ , where @xmath25 is defined as _ curvature radiation _
p_^(cr ) = 0.1 b_12 ^ 4/9 . before we present @xmath25 for the ics case , note that for the physical conditions we discuss in this paper it is convenient ( especially in our analytic calculations ) to treat the ics photons as generated via two different regimes of scattering : resonant ( r ) and non - resonant extreme klein - nishina scattering ( nr ) . in reality , the spectrum of ics photons from an electron of lorentz factor @xmath0 , a subset of which produce pairs , may be produced by scatterings in both the r and nr regimes .
however , the first pairs which mark the pff location will come from one of the two regimes .
thus , in our analytic calculations , to avoid unnecessary complexity , we shall differentiate between r and nr regimes .
now let us present the expressions for @xmath25 corresponding to these regimes _ resonant ics _
p_^(r ) = 0.1 b_12 ^ 6/7 , _ non - resonant ics _
p_^(nr ) = 0.4 b_12 ^ 4/7 . in this section
we shall explore the altitudes of pffs produced by the ics photons . to estimate the altitude of the pff above the stellar surface we can use , as in our previous papers ( see hm98 , eq .
[ 1 ] ; and paper i , eq .
[ 37 ] ) , the following expression s_0 = min [ s_a ( _ min ) + s_p ( _ min ) ] , [ s0 ] where @xmath26 is the acceleration length that is required for an electron to produce a photon of energy @xmath27 , and @xmath28 is the photon pair - attenuation length .
_ resonant ics _ in the r regime of ics
the characteristic energy of a scattered photon is ~2 b , [ ] where @xmath29 ( @xmath30 ) is the local value of the magnetic field strength in units of critical field strength , @xmath31 g. in this paper in our analytic estimates we assume that @xmath32 , which is justified for all cases except for the case of millisecond pulsars . by substituting @xmath33 into the expression for @xmath34 ( see hm98 or paper i for details ) we get s_p^(r ) = c_^(r)/ , where @xmath35 .
as far as the acceleration length @xmath36 is concerned , it is independent of the radiation mechanism ( provided that the radiation losses are negligible during acceleration ) , and the formulae of paper i ( see eqs . [ 38 ] ) are still applicable .
after substituting expressions for @xmath36 and @xmath37 into ( [ s0 ] ) we find that @xmath38 is minimized at _
min^(r ) = 10 ^ 6 \ { ll 0.9 p^-1/6b_12 ^ -1 & if pp_^(r ) , + 0.2 p^-3/4b_12 ^ -1/2 & if p > ~p_^(r ) . .
[ gamr ] now we can evaluate expression ( [ s0 ] ) at @xmath39 to calculate the dimensionless altitude ( scaled by a ns radius ) of the pff z_0^(r ) s_0^(r)(_min^(r))/r = 10 ^-2 \ {
ll 7 p^2/3b_12 ^ -1 & if pp_^(r ) , + 17 p^5/4 b_12 ^ -3/2 & if p > ~p_^(r ) . .
[ z0r ] _ non - resonant ics _ in the nr regime
we can use the extreme klein - nishina formula , and write for the energy of the scattered photon ~. the corresponding photon pair - attenuation length is s_p^(nr ) = c_^(nr)/ , where @xmath40 .
the expressions for @xmath41 and @xmath42 then read _
min^(nr ) = 10 ^ 5 \ { ll p^-1/6b_12 ^ -1/3 & if pp_^(nr ) , + 0.6 p^-3/4 & if p > ~p_^(nr ) , .
[ gamnr ] and z_0^(nr ) = 10 ^-2 \ { ll 3 ( p / b_12)^2/3 & if pp_^(nr ) , + 4 p^5/4 / b_12 & if p > ~p_^(nr ) . .
[ z0nr ] the unsaturated regime of formula ( [ gamnr ] ) agrees within a factor of @xmath43 2 with our numerical calculations for the ms and middle - aged pulsars .
the location of the pff for the different pair - producing radiation processes is more accurately determined using numerical calculations of the minimum height at which the first pairs are produced .
here we incorporate full spatial dependence ( in @xmath44 and @xmath45 ) of quantities such as the magnetic field and radius of curvature in computing the altitude of the pff .
however , as we will discuss in section 3.2 , we use a one - dimensional model for the pair dynamics to treat the screening of the electric field about the pff .
the numerical calculation of the altitude of the pff follows closely that of hm98 , but with one notable exception . as discussed above , we use the solutions for @xmath2 with an upper boundary to compute the ics pffs ( as in hm98 ) for normal pulsars , whereas we use the solutions for @xmath2 with no upper boundary ( but which saturates at infinity ) to compute the cr pffs ( as in paper i ) and the ics pffs for ms pulsars
otherwise , we have used the same expressions for the energy of pair producing photons : equation ( 29 ) of hm98 for cr photons and equation ( 43 ) of hm98 for ics photons .
we have also made the assumption ( as in hm98 ) that the primary electrons travel along the magnetic axis in computing the ics energy loss and scattering rate .
unlike in the analytic estimates presented , we do not separate the rics from the nrics photons but take a weighted average for the ics
photon energy : ^(ics ) = ^(r ) ^(r ) + ^(nr ) ^(nr ) .
calculations of @xmath46 as a function of the colatitude @xmath16 for four different sets of pulsar parameters are shown in figure 5 .
comparison of figures 5a and 5b confirms that the value of @xmath46 is independent of pc temperature , since rics photons have average energies of @xmath47 and nrics occurs primarily in the extreme klein - nishina limit , where the average scattered photons energy is @xmath48 . with numerical calculation of the pff altitude , we are able to determine the pulsar parameter space in which the formation of a pff by the different photon processes is possible ( i.e. where @xmath49 ) .
figure 1 shows the cr and ics pff parameter space as a function of pulsar period @xmath21 and surface field strength @xmath19 , as determined by the dipole formula @xmath50 .
we have also plotted observed pulsars from the atnf catalog ( http://www.atnf.csiro.au/ pulsar/ ) , which includes pulsars from the parkes multibeam survey ( manchester et al .
2001 ) having measured @xmath51 .
for the @xmath19 and @xmath21 values above the lines in fig . 1 for the different processes , we were able to find numerical solutions to equation ( [ s0 ] ) .
pulsars below the lines are therefore not capable of producing pairs from that process .
we find that the majority of pulsars can not produce pairs through cr .
this result has been well known for some time ( arons & scharlemann 1979 , arons 1998 , zhang et al .
we find that virtually all known pulsars , with the exception of only a few ms pulsars , are capable of producing pairs via ics .
for the majority of pulsars , those with @xmath52 , the pairs are produced from nrics photons .
we have plotted ics pff boundaries for three different temperatures and , as expected , there is only a small dependence on pc temperature .
the slight variation due to pc temperature occurs for higher field pulsars where scattering is not in the extreme klein - nishina regime and the photon energies become somewhat temperature dependent .
the failure of the lowest - field ms pulsars to produce ics pairs in our calculation may be due to the failure of our @xmath2 solution , obtained in the small - angle approximation , to accurately model the accelerating field in the ms pulsars which have large pcs with @xmath53 ; or to our use of a canonical ns model in the case of ms pulsars which may have undergone significant mass accretion in low - mass binary systems .
we are currently exploring these effects to correct this shortcoming .
also , we shall examine the possibility of photon - photon pair formation in ms pulsars .
we will discuss the broader implication of pair death lines in section 4 .
to estimate the maximum fractional density of positrons returning from the upper pffs we can use equation ( 33 ) of paper i and the expressions for @xmath46 derived in 2.2 .
we must note , however , that in paper i ( where we investigated the cr case alone ) in our analytic calculations of the fraction of returning positrons , we justifiably assumed that the screening occurs within the length scale much less than @xmath46 .
this is not the case for ics , for which the screening scale is determined by the scale of growth of energy of pair - producing photons . for the ics photons the energy of pair - producing photons @xmath54 , so that the scale of growth of @xmath0 governs the screening scale @xmath55 , i.e. _ s ~_z = z_0 \ { ll 0.5 z_0 & if pp_^(r , nr ) , + z_0 & if p > ~p_^(r , nr ) . .
[ deltas ] thus , in the case of ics the screening effectively occurs over a more extended region .
this means that even within the screening region the primary electrons keep accelerating and generating pair - producing photons .
since the pair - production rate per primary electron in the case of the ics photons is not as high as in the case of cr photons , the assumption that the altitude of the screening is @xmath46 would result in a significant underestimation of the returning positron fraction at the ics - controlled pair - fronts . to come up with better estimates we should add @xmath55 to the calculated values of @xmath46 , or , according to formula ( [ deltas ] ) , multiply @xmath46 by a factor 1.5 and 2 , for the unsaturated and saturated regimes , respectively .
evaluating expressions for @xmath56 ( here @xmath57 is the charge density of returning positrons , and @xmath58 is the goldreich - julian charge density ; see also eq . [ 33 ] of paper i ) at @xmath59 , we get _ resonant ics _ ( _ + ) ^(r ) = 10 ^ -2 \ { ll 2 p^2/3/b_12 & if pp_^(r ) , + 5 p^5/4 / b_12 ^ 3/2 & if p > ~p_^(r ) , .
[ frac_r ] and _ non - resonant ics _ ( _ + ) ^(nr ) = 10 ^ -3 \ { ll 7 ( p / b_12)^2/3 & if pp_^(nr ) , + 9 p^5/4 /b_12 & if p > ~p_^(nr ) . .
[ frac_nr ] in their paper ha01 assume that the fraction of returning positrons ( see 5.3 ) is of order of @xmath60 , where @xmath61 cm ) is the pc radius , and @xmath62 is the ns radius ( @xmath63 ) .
in contrast , in our calculations the fractional density of returning positrons is proportional to the dimensionless screening altitude ( see paper i ) , and can be roughly estimated as @xmath64 z_0 $ ] .
thus , the ratio of our value for the density of returning positrons to that of ha01 is of order of @xmath65 , which for the observed pulsar parameters may range from @xmath66 to @xmath67 .
more specifically , for short - period pulsars our calculated fractions of returning positrons are up to a factor of @xmath68 less than those of ha01 , while for relatively long - period pulsars they are larger by about the same factor .
besides this quantitative difference , there is a significant difference in the dependence of our fractional densities on @xmath69 and @xmath21 .
note that in reality , the screening scale and thus also @xmath70 is dependent on the pc temperature , as will be shown in the numerical calculations .
we can not model this dependence analytically since we have assumed that the @xmath55 is a constant multiple of @xmath46 for all pulsars .
our numerical calculation of the returning positron fraction and scale length of the ics pair screening follows that described in paper i for cr pair screening .
as detailed in that paper , we first compute the pair source function in energy and altitude above the pff and then compute the dynamical response of the pairs to the @xmath2 above the pff by solving the continuity and energy equations of the pairs to obtain the charge density .
the @xmath2 above the pff is parameterized as an exponential with scale height @xmath55 , which is readjusted at each iteration to equal a multiple of the height at which the computed charge density from the pairs equals the difference between the primary beam charge density and goldreich - julian charge density . in computing the source function of pairs from ics photons ,
we consider a hybrid spectrum of rics and nrics photons from the primary particle , of lorentz factor @xmath48 , at a given altitude of its acceleration . in the ics photon spectrum ,
photons with energies @xmath71 are produced primarily by rics .
since the rics spectrum from a single electron cuts off sharply for photon energies @xmath72 , photons above this energy , up to a maximum energy of @xmath73 will be produced by nrics .
we therefore describe the ics photon spectrum for @xmath71 by a rics spectrum based on an expression given by dermer ( 1990 ) . assuming delta - function distributions of electrons of lorentz factor @xmath48 , and thermal photons of energy @xmath74 and incident angles @xmath75 in the lab frame , dermer s ( 1990 ) equation ( 10 ) for the distribution of scattered photons with energy @xmath76 , integrated over scattered angles , @xmath77 , becomes [ n^r ] n^r(_s)dtd_s = c_t _ b2(_+ - _ - ) 1 ^ 2 ^ 3n_0_0
^ 2[j_0 + j_1 + 12(j_2-j_1 ) ] , where @xmath78 , @xmath79 , & \nonumber \\
u_- = \gamma\epsilon_b^{-1}\ , { \rm min}[2\epsilon_s , \epsilon_0(1-\beta\mu_- ) ] , & \end{aligned}\ ] ] and @xmath80 , \end{aligned}\ ] ] i_1(u ) = u - ( u+1)^-1 - 2(u+1 ) , i_2(u ) = u + + ( a^-1 + a)^-1 , where @xmath81 is the thompson cross section , @xmath82 is the electron velocity in units of the velocity of light , @xmath83 is the cyclotron energy in units of @xmath84 , @xmath85 , @xmath86 is the fine - structure constant and @xmath87 is the density of thermal photons of energy @xmath88 .
each scattered photon is assumed to be emitted along the direction of the particle momentum ( i.e. along the local magnetic field ) .
although the distribution of scattered photon angles depends on energy ( see fig .
3 of dermer 1990 ) , the photons at pair - producting energies scattered by relativistic electrons are mostly emitted along the field direction .
since we will be treating scattering of both upward - moving electrons and downward - moving positrons , we have @xmath89 in equation ( [ u ] ) , where _ c = _ c = h is the largest scattering angle at height @xmath90 above the ns surface , and @xmath91 is the radius of the hot pc .
the ics photon spectrum for @xmath92 is produced by nrics .
we use a spectrum for relativistic ( but non - magnetic ) scattering which is based on an expression of jones ( 1968 ) , derived originally to apply to scattering of an isotropic distribution of photons by a relativistic electron , which we have modified to apply to a semi - isotropic distribution of thermal photons [ n_nr ] n^nr(_s)dtd_s = n_0r_0 ^ 2 c_^2_0 , where q = e_s_0(1-e_s ) , e_s = _ s , _ 0 = 2_0 _ . here
@xmath93 is the classical electron radius . for upward - moving electrons , @xmath94 and for downward - moving positrons , @xmath95 .
our method for calculating the pair source function from ics photons is similar to that used in paper i for cr photons .
however , unlike in the case of cr , higher generations of pairs from synchrotron radiation of first - generation pairs are important in the screening , since the attenuation length for the sychrotron photons is less than the ics screening scale . as in paper
i , the contribution to the pair source function from first generation pairs is computed by dividing the ics spectrum radiated by the primary particle at each step along its path into discrete energy intervals .
a representative photon from each ics energy bin is propagated through the local field to determine whether it escapes or produces a pair , in which case the location of the pair is recorded ( for details of such a calculation , see paper i and harding et al .
the pairs for each ics spectral energy interval of width @xmath96 are then weighted in the source function by the number " of rics or nrics photons , @xmath97 , represented by the test photon in that energy bin , estimated by integrating the photon spectrum over the bin n^r , nr(_s ) = sc__s-_s2 ^_s+_s2n^r , nr(_s)dtd_s d_s , where @xmath98 is the size of the particle path length .
the contribution of higher generations of pairs to the source spectrum is computed by simulating a pair cascade in the method described by baring & harding ( 2001 ) , except that we do not include the possibility of photon splitting in the present calculation . by limiting the surface magnetic field we consider to @xmath99 , we can safely neglect splitting . as described in baring & harding ( 2001 ) ,
the created pairs from the first generation make transitions between landau states to radiate synchrotron / cyclotron photons until they reach the ground landau state .
each synchrotron / cyclotron photon is individually traced through the local magnetic field until it either escapes or creates a pair , which radiates another generation of photons . a recursive routine
is called upon the radiation of each photon , so that a large number of pair generations may be simulated .
one notable difference was introduced in the present cascade simulation in order to treat the synchrotron radiation from ms pulsars .
the low surface fields of the ms pulsars , combined with the photon energies @xmath100 required to produce pairs , give very large landau states for the created pairs , making it impossible to treat the synchrotron / cyclotron photon emission discretely .
for local fields @xmath101 , we treated the synchrotron radiation from the pairs by the same method as for the ics radiation from the primaries , i.e. as a continuous spectrum . to describe the total spectrum radiated by a particle as it decays from a high initial landau state ( i.e. large initial pitch angle ) to the ground state ( zero pitch angle ) we use the calculated synchrotron spectrum of tademaru ( 1973 )
n_s()d= 12(b_0)^-1/2 ^ -3/2d , where @xmath102 is the initial pitch angle of the particle and @xmath103 is the local magnetic field strength .
surprisingly , for the ms pulsars we are required to limit the upper end of the synchrotron spectrum at an energy equal to the particle lorentz factor in order to avoid violating energy conservation .
this is a well - known problem in high magnetic fields ( e.g. harding & preece 1987 ) , but in this case the very high lorentz factors of the pairs give a synchrotron critical energy @xmath104 . imposing the energy conservation condition limits the number of pair generations .
the discrete qed treatment of the synchrotron emission in high magnetic fields ( described above ) is able to treat the effect of pair production in the ground landau state , which also severely limits the number of pair generations in fields @xmath105 ( baring & harding 2001 ) . since the ics photon spectrum has a much weaker dependence on the particle lorentz factor than the cr photon spectrum , the ics pair source function grows much more slowly with height above the pff . in the case of nrics ,
the photon emission rate decreases as @xmath106 , while the rics emission rate first increases sharply at low @xmath48 as the soft photons come into resonance in the particle rest frame , and then decreases as @xmath107 , whereas in the case of cr the photon emission rate increases as @xmath48 . even though the characteristic ics photon energies are increasing with @xmath48 , the stronger decrease in photon number with height as the particle accelerates , combined with the decrease in soft photon density and angle with altitude , makes screening by pairs produced by ics photons much more difficult and in some cases impossible
. in nearly all cases ,
the screening scale lengths are comparable to the particle acceleration length , requiring different treatment of the self - consistent screening calculation from what was used in paper i for cr screening . the onset of pair production and possible screening close to the ns surface couples the entire @xmath2 solution to the screening scale .
we therefore must correct the accelerating @xmath2 as well as the @xmath2 in the screening region above the pff on every iteration of the screening scale length .
this is done by adjusting the upper boundary of the @xmath2 solution ( eqs .
[ a7 ] - [ a10 ] of paper i ) to @xmath108 , so that the electric potential distribution below the pair front is accurate .
we still assume the exponential form with scale height @xmath55 ( eq . [ 26 ] of paper i ) for @xmath2 above the pff .
but unlike our calculation of cr pair screening , the location of the pff will change and adjust on every interation of @xmath55 . the exponentional function @xmath109 $ ] describing the @xmath2 above the pff is also renormalized at the pff at each iteration so that the value of @xmath110 is equal to the maximum value of the @xmath2 solution between @xmath111 and @xmath112 .
we also must take into account the increase in the charge deficit in the screening region , which will also cause ics screening to be more difficult .
figure 2 shows an example of an ics pair source function , integrated over energy , so that the growth of pairs as a function of distance @xmath113 above the pff is displayed .
the first generation pairs ( those produced by primary electrons ) is almost flat before beginning to decrease . including pairs from higher cascade generations produces an initially sharp increase in number of pairs vs. height , but the pair number quickly levels off and also eventually decreases .
the ics pair source function is a sharp contrast to the cr pair source function shown in figure 1 of paper i , which increases nearly exponentially with height above the pff .
this raises the interesting question of whether ics pairs are capable of screening @xmath2 to all altitudes , even if they are able to initially screen the local @xmath2 .
figure 3 shows an example of a self - consistent solution for the charge density @xmath114 due to polarization of pairs and @xmath115 in the screening region .
the charge density initially increases faster than the charge deficit , which is increasing approximately as @xmath116 . in this case ,
screening of @xmath2 is achieved locally , until the pair source function begins to decline with @xmath116 , causing the charge density due to pairs to decrease . as the charge density due to pairs decreases
while the charge deficit is increasing , screening can not be maintained and the @xmath2 will again begin to grow .
we can not model this effect in the present calculation , which would require the solution of poisson s equation throughout the screening region .
however , it is clear that ics screening can not short - out @xmath2 and produce a limit to particle accleration at low altitudes .
ics screening , even if locally complete , will only slow the particle acceleration until the ics pair density falls off and beyond this point the particles will resume their acceleration until they can produce cr pairs , which will then screen the @xmath2 at higher altitudes . for many pulsars
, however , ics can not even achieve locally complete screening .
figure 4 shows computed boundaries of ics local screening for different assumed pc temperatures , @xmath117 k , as a function of pulsar surface magnetic field and period .
pulsars above and to the left of these boundaries can achieve local screening of @xmath2 at a colatitude of @xmath118 , while pulsars below and to the right of the boundaries can not . although the pff boundaries were nearly independent of @xmath119 , the screening boundaries are very sensitive to the temperature of the soft photons . for low @xmath119 ,
only very high - field pulsars , those whose pairs source functions are dominated by rics , can screen with ics photons . as @xmath119 increases , the boundaries move down , but are limited in their movement to the right as they approach the ics pff boundaries ( figure 1 ) .
there is also some dependence of the screening boundaries on @xmath16 , in that the boundary for a given @xmath119 will be higher at a lower value of @xmath16 .
this means that screening is more difficult in the inner parts of the pc , so that some pulsars may have local ics screening only in the outer part of their pc .
even though complete ics screening is not achieved for all pulsars , there will still be positrons which turn around and return to heat the pc .
in fact this heating can be substantial since in the absence of complete screening which shuts of the acceleration , most of the positrons from ics pairs will return to the pc .
figure 5 shows self - consistent calculations of the returning positron fraction , as well as the pff altitude @xmath46 which we have discussed in section ( 2.2 ) , as a function of @xmath120 . figures 5a and 5b , which have the same @xmath21 and @xmath69 values but temperatures @xmath121 and @xmath122 respectively , illustrate that although there is very little difference in the pff altitude with a large increase in pc temperature there is a large increase in returning positron fraction .
the dependence of @xmath123 on @xmath119 is not reflected in the analytic estimates ( eqs . [ [ frac_r ] ] , [ [ frac_nr ] ] ) which are based only on a constant multiple of the charge deficit at the pff altitude , proportional only to @xmath46 .
the size of @xmath123 is actually dependent on the number of pairs produced above @xmath46 which is not dependent on the charge deficit if screening is not achieved .
comparison of figures 5a and 5c shows that pulsars with larger periods have relatively larger @xmath46 and also larger fractions of returning positrons .
this is due to the fact that longer period pulsars have more difficulty producing pairs due to the smaller pc size and therefore larger radii of curvature of the last open field line . as illustrated in figure 5d ,
ms pulsars have very large @xmath46 and large fractions of returning positrons , but smaller than expected from the analytic solutions , because they do not produce enough pairs to screen the field .
figure 6 shows the dependence of returning positron fraction on pc temperature @xmath119 and surface field strength .
the steep dependence of @xmath123 on @xmath119 at low temperatures @xmath124 is due to the increase in numbers of high energy ics photons that produce pairs . for @xmath125 , in the case of normal pulsars
there is local screening at these field strengths , which limits the number of returning positrons to be proportional to @xmath126 rather than to @xmath119 .
the screening therefore produces a saturation in the pc heating , which will be seen in the calculation of pc heating luminosities to be presented in section v. as discussed in hm98 , the positrons returning from the upper pff will be accelerated through the same voltage drop as the primary electrons and thus will reach the same high energies as the primaries that produced them .
they will therefore initiate pair cascades as they approach the ns surface .
hm98 speculated that the pairs from these downward moving cascades may screen the @xmath2 above the surface since they found that the pffs of the positrons are located at significant altitudes above the surface , due to asymmetry in the ics between upward moving and downward moving particles .
now that we can calculate the fraction of positrons returning from the upper ics pff , we can investigate the question of screening at the lower pff .
we have simulated cascades below the pffs of test positrons returning from the upper pff .
the source functions of pairs from the downward cascades and the charge density are computed in the same way as at the upper pff , and the screening scale height is determined self - consistently .
but there are several critical differences between screening at upper and lower pffs .
first , the flux of returning positrons from the ics pff is small compared to the primary flux and the charge density must be weighted with the value @xmath123 .
second , the value of @xmath2 is very small near the ns surface , making it harder to turn around the electrons from the pairs .
third , the charge deficit is much smaller than at the upper pff and decreasing toward the stellar surface , being proportional to the altitude of the returning positron pair front .
this guarantees that there will be a point above the ns surface where @xmath114 will equal the charge deficit .
we have computed pair source functions from cascades of downward - moving positrons and the resulting charge densities below the lower pff for a range of pulsar parameters .
we find that for nearly all pulsars , the charge densities do not reach values high enough to screen @xmath2 significantly above the ns surface . for pulsars having surface fields @xmath127 and periods @xmath128 , pair multiplicities from the near - surface cascades are not high enough to balance the low fraction of returning positrons , primarily because most pairs are produced in the ground landau state in the high near - surface fields . for pulsars having lower surface fields and longer periods , screening can occur for some cases just above the surface at the highest pc temperatures .
these pulsars are near the ics screening boundary ( figure 4 ) .
therefore , according to our calculations it seems that ics pffs will be stable except possibly near the screening boundaries .
before we discuss the pulsar death line , it is instructive to calculate the luminosity of the primary electron beam at the pff
. the luminosity of the primary beam as a function of @xmath15 ( @xmath129 ) can be written as ( see eq . [ 73 ] of mh97 , and cf .
[ 61 ] of paper i ) [ lprim_gen ] l_prim = c _ s(z ) ( z , ) ( z,)ds , where @xmath130 ( @xmath131 is the gravitational radius of a ns of radius r ) , @xmath132 is the charge density of the electron beam , @xmath133 is the electrostatic potential , @xmath134 , and @xmath135 is an element of solid angle in the pc region ( see also paper i for details ) . here the integration is over the area of a sphere cut by the polar flux tube at altitude @xmath15 . similar to our derivation of formulae ( 63)-(65 ) of paper
i , we can write for the luminosity of the primary beam l_prim = f_prim e_rot , [ lprim ] where @xmath136 is the pulsar spin - down luminosity ( @xmath137 , where @xmath138 is the surface value of the magnetic field strength corrected for the general - relativistic red shift ; see eq .
[ 66 ] of paper i for details ) , and @xmath139 is the efficiency of converting spin - down luminosity into the luminosity of the primary beam .
thus , by comparing the above two equations , we can get the following expression for @xmath139 f_prim = 610 ^ -3 p^-1/2 \ { ll & if pp_^(cr , r , nr ) , + 0.6 & if p > ~p_^(cr , r , nr ) , .
[ fprim ] where @xmath140 is the altitude scaled by the pc radius , @xmath141 . in the derivation of formula ( [ fprim ] ) we assumed @xmath142 and @xmath143 .
now we should evaluate expression ( [ fprim ] ) at the altitude beyond the pff where the electric field is screened , or at @xmath144 .
for the cr case @xmath145 , while for the ics pair fronts ( for both r and nr regimes ; see also eq . [
[ deltas ] ] ) _ \ { ll 1.5 _ 0 & if pp_^(r , nr ) , + 2 _ 0 & if p > ~p_^(r , nr ) . . after substituting @xmath146 we get the following explicit expressions for @xmath24 _ curvature radiation _ _
^(cr ) = \ { ll 30 p^9/7/b_12 ^ 4/7 = 2 b_12 ^ 5/7_6 ^ 9/14 & if pp_^(cr ) , + 430 p^9/4 / b_12 = 4 b_12 ^ 5/4_6 ^ 9/8 & if p > ~p_^(cr ) , . [ zeta*cr ] _ resonant ics _ _
^(r ) = \ { ll 10 p^7/6/b_12 = b_12 ^ 1/6_6 ^ 7/12 & if pp_^(r ) , + 30 p^7/4 /b_12 ^ 3/2 = b_12 ^ 1/4_6 ^ 7/8 & if p > ~p_^(r ) ,
. [ zeta*r ] _ non - resonant ics _ _
^(nr ) = \ { ll 3 p^7/6/b_12 ^ 2/3 = 0.3 b_12 ^ 1/2_6 ^ 7/12 & if pp_^(nr ) , + 6 p^7/4 /b_12 = 0.2 b_12 ^ 3/4_6 ^ 7/8 & if p > ~p_^(nr ) , .
[ zeta*nr ] where @xmath147 . inserting expressions ( [ zeta*cr ] ) - ( [ zeta*nr ] ) into formulae ( [ lprim ] ) , ( [ fprim ] )
we get _ curvature radiation _
l_prim^(cr ) = 10 ^ 32
erg / s \ { ll 0.3 p^-27/14b_12 ^ 6/7 = 20 b_12 ^ -15/14_6 ^ -27/28 & if pp_^(cr ) , + 0.1 p^-9/4b_12 = 10 b_12 ^ -5/4_6 ^ -9/8 & if p > ~p_^(cr ) , . or l_prim^(cr ) = 10 ^ 16 ( erg / s)^1/2 e_rot^1/2 \ { ll p^1/14b_12 ^ -1/7 & if pp_^(cr ) , + 0.3 p^-1/4 & if p > ~p_^(cr ) , .
f_prim^(cr ) = 0.01 \ { ll 7 b_12 ^ 13/14_6 ^ 29/28 & if pp_^(cr ) , + 5 b_12 ^ 3/4_6 ^ 7/8 & if p > ~p_^(cr ) , . _ resonant ics _
l_prim^(r ) = 10 ^ 31 erg / s \ { ll 0.3 p^-13/6=30 b_12 ^ -13/6_6 ^ -13/12 & if pp_^(r ) , + 0.05 p^-11/4b_12 ^ 1/2=20 b_12 ^ -9/4_6 ^ -11/8 & if p >
~p_^(r ) , . or l_prim^(r ) = 10 ^ 15 ( erg / s)^1/2 e_rot^1/2 \ { ll p^-1/6b_12 ^ -1 & if pp_^(r ) , + 0.2 p^-3/4b_12 ^ -1/2 & if p > ~p_^(r ) , .
f_prim^(r ) = 10 ^ -2 \ { ll b_12 ^ -1/6_6 ^ 11/12 & if pp_^(r ) , + 0.8 b_12 ^ -1/4_6 ^ 5/8 & if p > ~p_^(r ) , . _
non - resonant ics _
l_prim^(nr ) = 10 ^ 30 erg / s \ { ll 0.3 b_12 ^ 2/3p^-13/6=30 b_12 ^ -3/2_6 ^ -13/12 & if pp_^(nr ) , + 0.1 b_12p^-11/4=30 b_12 ^ -7/4_6 ^ -11/8 & if p > ~p_^(nr ) , . or
= 10 ^ 15 ( erg / s)^1/2 e_rot^1/2 \ { ll 0.1 b_12 ^ -1/3p^-1/6 & if pp_^(nr ) , + 0.05 p^-3/4 & if p > ~p_^(nr ) , .
f_prim^(nr ) = 10 ^ -3 \ { ll b_12 ^ 1/2_6 ^ 11/12 & if pp_^(nr ) , + 2 b_12 ^ 1/4_6 ^ 5/8 & if p > ~p_^(nr ) . .
the above formulae are just different representations of the luminosity that a primary ( electron ) beam achieves by the moment the @xmath2 is screened , for each of the radiation processes we discuss .
note that the eight known egret @xmath48-ray pulsars fit an empirical relation @xmath148 .
this is reproduced by the above expression for the luminosity of the primary beam at the cr pff if the primary electrons are nearly 100% efficient at conversion of their energy to high - energy @xmath48-rays ( see zhang & harding 2000 ) , but not by the luminosity at the ics pair fronts .
but we have argued that the primary particle acceleration is not stopped at the ics pff , so that the potential drop at the cr pff determines the final acceleration energy and thus the @xmath48-ray luminosity of the pulsar . in pulsar physics
the term death line " was originally introduced for radio pulsars and refers to a line separating the domain ( normally in the @xmath21@xmath51 diagram ) favouring pair formation from the domain where pair formation does not occur .
most pc models for radio pulsars imply therefore that radio emission turns off ( see sturrock 1971 ; ruderman & sutherland 1975 ; and chen & ruderman 1993 ) if the potential drop required for pair production exceeds the maximum which can be achieved over the pc .
this concept of death line implicitly assumes that the pair - formation is a necessary condition for pulsar radio emission , and that pulsars become radio quiet after crossing the death lines during their evolution .
some authors argue that this condition may not be sufficient , though .
for example , ha01 define a pulsar s death as the condition when pair - production is too weak to generate the pair cascade multiplicity required to screen the accelerating electric field .
we believe that pair - formation is vital for pulsar operation , but we suggest that the pairs required for radio emission may not necessarily screen or shut off the accelerating electric field ( see also paper i ) .
in addition , we adopt a canonical definition of a pulsar s death as an essentially pairless regime of operation . in general , to write a formal criterion specifying the death line one needs to know the characteristic voltage drop in the pulsar acceleration region . for the acceleration model we employ in our study this voltage drop
can be calculated as _
acc = 10 ^ 10 v b_12 p^-5/2 \ { ll 3 & if pp_^(cr , r , nr ) , + 2 & if p > ~p_^(cr , r , nr ) . .
[ phi_acc ] the corresponding calculated lorentz - factor of a primary electron accelerating from the pc is _
acc = e _ acc = 10 ^ 4 b_12 p^-5/2 \ { ll 6 & if pp_^(cr , r , nr ) , + 4 & if p > ~p_^(cr , r , nr ) . . [ g_acc ] thus , our criterion for the pulsar death line separating radio - active from radio - quiet pulsars translates into _ acc _ min .
[ deathline ] however , the above expression for @xmath149 , for fixed values of @xmath69 and @xmath21 , still depends on a dimensionless acceleration altitude , @xmath140 , and can not be used therefore in criterion ( [ deathline ] ) .
it is important to note , that in our numerical calculations , at each altitude , we check the instant value of a particle s lorentz - factor against the above criterion .
if it meets this criterion , we pick the corresponding bulk pulsar parameters such as @xmath69 and @xmath21 and store them as those of the death line . generally , to make our analytic derivation of the death line self - consistent , we need an additional independent equation that relates @xmath140 with e.g. @xmath69 and @xmath21 .
obviously , the formal substitution of values of @xmath150 calculated in 2.2 into @xmath149 in criterion ( [ deathline ] ) would lead to a trivial identity , simply because in our derivation of @xmath24 we have already used the corresponding values of @xmath39 .
instead , we suggest using equations ( [ fprim ] ) and ( [ g_acc ] ) to eliminate @xmath140 to get _ acc 610 ^ 7 f_prim b_12 p^-2 .
[ gamma_acc ] the advantage of this formula for @xmath149 is that it does not discriminate between unsaturated and saturated regimes of acceleration of primary electrons and is factorized by the efficiency of the pulsar accelerator , @xmath139 , which may be generally regarded as a free pulsar parameter , characterizing the efficiency of primary acceleration above the pulsar pc .
now , let us use formula ( [ gamma_acc ] ) in criterion ( [ deathline ] ) to get explicit expressions describing the death lines for different underlying radiation mechanisms for pair - producing photons .
note that for this purpose in criterion ( [ deathline ] ) we should evaluate @xmath151 at @xmath152 , where @xmath153 is the minimum pulsar efficiency allowing pair formation .
this completes our formal definition of death lines and makes it physically sensible : for the fixed values of @xmath69 and @xmath21 , to specify the onset of pair formation , one needs to know the minimum energetics a primary electron beam should have to enable pair formation , i.e. one needs to know @xmath153 .
we must note that , within the framework of this approach , the analytic death lines derived in previous studies implicitly assume some fixed value for @xmath153 , the same for all pulsars and for all relevant pair - formation mechanisms , and that those values of @xmath153 are not necessarily meaningful . on the contrary ,
our analytic death lines are explicitly determined by a pulsar bulk parameter @xmath153 .
the resultant analytic death lines ( or rather parameter spaces with allowed pair formation ) in the @xmath21@xmath51 diagram read _ curvature radiation _ \
{ ll 218 p - 1.75f_prim^min - 16 & if pp_^(cr ) , + 52 p - 2f_prim^min - 17 & if p > ~p_^(cr ) , .
[ dl_cr ] _ resonant ics _ \
{ ll 56 p - f_prim^min - 17.4 & if pp_^(r ) , + 23 p - 1.33f_prim^min - 18.9 & if p > ~p_^(r ) , .
[ dl_r ] _ non - resonant ics _ \
{ ll 74 p - 1.5f_prim^min - 19.8 & if pp_^(nr ) , + 32 p - 2f_prim^min - 21.6 & if p > ~p_^(nr ) . .
[ dl_nr ] the fitting of observational data with the above expressions requires specification of the pair - producing efficiency ( @xmath153 ) of a pulsar pc accelerator . in this paper
we calculate pulsar death lines numerically , and by comparing the analytic death lines with our numerical death lines we can estimate the values of @xmath153 ( see next section ) .
it is of fundamental importance to know if this value of @xmath153 ( i.e. @xmath139 calculated along a pair death line ) is the same for all pulsars , and how it depends on the radiation process responsible for the pair formation . we find that the analytic death lines defined by ( [ dl_cr])-([dl_nr ] ) with the appropriately chosen value of @xmath153 , different for each radiation mechanism , fit satisfactory the observational data and are in a good agreement with our numerically calculated death lines .
this means that our introduction of the parameter @xmath153 and the assumption that this parameter only weakly depends on pulsar @xmath69 and @xmath21 values , for a given mechanism of generation of pair - producing photons , may be quite justified for the analytic tackling of pulsar death lines .
the analytic expressions above for the pair death lines differ significantly from those derived by zhang et al .
the pair death lines of zhang et al .
( 2000 ) were derived from the condition ( in our notation ) that @xmath154 , which implicitly assumes that @xmath155 , whereas we allow @xmath156 to be a free parameter .
as we shall see in the next section , comparison with numerical calculations implies @xmath157 .
also , zhang et al . ( 2000 ) computed the altitude of the pff and thus the value of @xmath158 from the condition @xmath159 , that photon attenuation length is equal to the length required for the primary electron to produce one ics photon , since the acceleration length @xmath160 .
this introduced a dependence of the pff altitude on pc temperature , which they took to be the self - consistent temperature from pc heating .
by contrast , our analytic expression for the pff altitude ( eq . [ [ s0 ] ] ) does not depend on pc temperature .
finally , zhang et al . ( 2000 ) included pairs from only rics and did not include pairs from nrics in their ics death lines .
the discussion and results of our numerical calculation of the cr and ics pair death lines were presented in figure 1 . since our numerical solution for the location of the pffs
are computed by iteration , we can unambiguously determine the pair death lines without having to define the efficiency , @xmath153 , as was needed in the analytic expressions above .
we must note that in our derivation of the above analytic death - line conditions we did not require the value of accelerating potential drop to be maximum at the pff . in other words , we did not impose any _
ad hoc _ constraint on the acceleration altitude , thus making the latter consistent with the accelerating potential drop .
it is remarkable that for each of the radiation mechanisms the parameter @xmath153 ( minimum efficiency of acceleration of primary beam required to set pair formation ) remains nearly constant along the numerical death line . by comparing our numerically computed death lines with those given by ( [ dl_cr])-([dl_nr ] ) we find that @xmath161 for cr , 0.003 - 0.02 for rics and @xmath162 for nrics pair fronts .
these efficiencies are still small compared to the maximum efficiency we esimate in the last paragraph of this section .
another important finding is that production of pairs in all observed pulsars by cr photons would require @xmath163 indicating the difficulty in interpreting the observational data in terms of a cr mechanism alone . on the contrary ,
ics - based death lines imply much less consumption of pulsar spin - down power for pair creation .
this result and our finding that the ics - generated pairs tend to only partially screen the accelerating electric field may suggest the occurrence of `` nested '' pair formation regions : the lower - altitude pairs produced by the ics photons and the higher - altitude pairs produced by the cr photons generated by particles accelerating through the region of the ics pair formation .
this fact implies that if a pulsar is below the cr death line in figure 1 , and if any ics screening is inefficient , then the primary beam acceleration is not limited by pair production and will be producing ample cr high - energy photons .
the upper limit for the pulsar @xmath0-ray luminosity in this case can be estimated by using equation ( [ lprim_gen ] ) with the integrand evaluated at the maximum value of the potential given by formula ( 13 ) of hm98 ( see also eq . [ a4 ] of hm98 ) : l _ , max = c _ s ( ) ( ) ( , ) _ maxds ( ) , [ lgamma ] where @xmath164\kappa \phi _ 0 $ ] , @xmath165 , and @xmath166 , where @xmath167 is the spherical cross - sectional area of the polar flux tube at radial distance r ( @xmath168 ) .
note that in formula ( [ lgamma ] ) the integral @xmath169 represents the total electron current flowing from the pc and is a constant .
thus , integrating over @xmath120 and @xmath170 , we arrive at l _ , max 1.5 ( 1- ) e_rot 0.6 e_rot .
[ lgamma_max ] here we used @xmath142 and @xmath143 .
we predict , therefore , that pulsars tend to be efficient @xmath0-ray sources if they are to the right of their cr death lines in b - p diagram .
note that expression ( [ lgamma_max ] ) can also be used to set the possible upper limit for @xmath20 , which is around 0.3 .
now we would like to demonstrate one more remarkable property of a pulsar pc accelerator .
let us take the expression for the electric potential as a function of @xmath140 ( see eq .
[ [ phi_acc ] ] ) and evaluate it at the screening altitude ( at @xmath171 ) . after substituting the corresponding expressions for @xmath24 ( see eqs . [
[ zeta*cr]]-[[zeta*nr ] ] ) into equation ( [ phi_acc ] ) we get the following self - consistent formulae for the critical voltage drop ( characteristic voltage drop at the screening altitude ) _ curvature radiation _ _ acc ( _ ) = 10 ^13 v \ { ll 2 ( _ 6/p)^1/14 & if pp_^(cr ) , + p^-1/4 & if p > ~p_^(cr ) . .
_ resonant ics _
_ acc ( _ ) = 10 ^11 v \ { ll 3 p^-7/6_6 ^ 1/2 & if pp_^(r ) , + 2 p^-5/4_6 ^ 1/4 & if p > ~p_^(r ) , .
_ non - resonant ics _
_ acc ( _ ) = 10 ^11 v \ { ll 2 p^-1/2_6 ^ 1/6 & if pp_^(nr ) , + p^-3/4 & if p > ~p_^(nr ) . .
these formulae show very weak dependence of critical voltage drop on pulsar parameters @xmath69 and @xmath21 , especially for the cr case , as also shown by hm98 .
this fact simply indicates that the critical voltage drop needed for the ignition of a pair formation is mainly determined by the pair - formation microphysics itself . as a result , as we move from the lower left to the upper right corner of the @xmath21@xmath51 diagram , the variation ( by orders of magnitude ) in the altitude of the pff effectively compensates the corresponding change in voltage drop due to its scaling with pulsar parameters @xmath69 and @xmath21 , thus maintaining it near its critical value .
note that in similar expressions derived by ha01 ( see their eqs .
[ 73]-[75 ] ) the characteristic voltage drop varies by @xmath43 four orders of magnitude over the whole range of pulsar spin periods @xmath172 s.
to estimate the efficiency of pc heating by returning positrons , we can use formulae ( 64 ) , ( 65 ) of paper i again evaluated at @xmath173 .
thus , setting @xmath142 and @xmath143 , we get for the heating efficiency , @xmath174 , _ resonant ics _ [ f+r ] f_+^(r ) = 10 ^-5 p^-1/2_6 ^ 3/2 \ { ll 2.5 & if pp_^(r ) , + 1.5 & if p > ~p_^(r ) , . _
non - resonant ics _ [ f+nr ]
f_+^(nr ) = 10 ^-5 p^1/2_6 \ { ll 1.0 & if pp_^(nr ) , + 0.6 & if p > ~p_^(nr ) . .
here we shall summarize the explicit expressions for the estimated pc luminosities and surface temperatures due to the heating by positrons returning from the pair fronts set by the cr and ics photons .
our estimates of the pc temperature are based on a standard formula @xmath175 , where @xmath176 is the stefan - boltzmann constant .
this formula implies therefore that the pc is heated homogeneously , and that the heated area is confined by the pc radius . _ curvature radiation _
l_+^(cr ) = 10 ^31 erg / s p^-1/2 \ { ll 0.4 p^-5/14_6 ^ -1/7 & if pp_^(cr ) , + 1.0 & if p > ~p_^(cr ) , .
[ l+cr ] t_pc^(cr ) = 10 ^6 k p^-1/4 \ { ll p^-5/56_6 ^ -1/28 & if pp_^(cr ) , + 2 &
if p > ~p_^(cr ) , . [ tcr ] _ resonant ics _
l_+^(r ) = 10 ^28 erg / s p^-5/2 _ 6 ^ 1/2 \ { ll 0.8 & if pp_^(r ) , + 0.5 & if p > ~p_^(r ) , .
[ l+r ] t_pc^(r ) 810 ^3 k p^-9/4 _ 6 ^ 1/8 , [ tr ] _ non - resonant ics _ [ l+nr ]
l_+^(nr ) = 10 ^28 erg / s p^-3/2
\ { ll 0.3 & if pp_^(nr ) , + 0.2 & if p > ~p_^(nr ) , .
t_pc^(nr ) 410 ^4 k p^-5/4 .
[ tnr ] we caution that the above formulae are upper limits to the pc heating luminosity , in that they assume locally complete screening . as we have shown in section 3
, such screening does not occur in many cases , most notably for older pulsars .
thus as will be shown by the numerical calculation presented in the next section , eqs .
( [ l+nr ] ) and ( [ tnr ] ) do not accurately predict @xmath177 and @xmath178 for ms pulsars , but give values that are much too high .
let us compare the above temperature estimates with those of ha01
. at @xmath179 and @xmath180 s our estimate of @xmath181 is roughly the same as the corresponding estimate presented by ha01 ( see their eqs .
[ 58 ] ) . for the ics cases
( see their eqs . [ 56 ] , [ 57 ] and [ 59 ] , [ 60 ] ) their estimates are systematically bigger , by a factor of 3 to 12 .
however , for the ms pulsars we predict higher temperatures . also , the effect that the ics contribution to the pc temperatures dominates over the cr contribution for ms pulsars is more pronounced in our formulae . in addition , for long - period pulsars our formulae predict more drastic decline in the pc heating temperatures with a pulsar period .
we also comment that both the analytic estimates and numerical calculations of @xmath177 depend on the value of @xmath182 , related to the compactness and moment of inertia of the ns .
we numerically evaluate the luminosity of pc heating due to ics pffs in a similar way to that given by equation ( 62 ) of paper i. the fraction of returning positrons times the potential drop between the pff and the surface is integrated in @xmath16 across the pc .
however , since in the case of ics pffs the potential varies significantly across the width of the pff , we also integrate the product of the potential and the returning positron fraction at each altitude between @xmath46 and @xmath108 : l_+ = 2 c s_pc _ 0 ^ 3 f(1 ) _ 0 ^ 1
_ z_0^z_0 + _ s _ + ( z - z_0 , ) ( z,)dzd= _ 0 ^ 1 l_+()d [ l_+ii ] where @xmath183 is the area of the pc .
figure 7 shows the dependence of @xmath184 on @xmath16 , which reflects the distribution of heating across the pc .
most of the pc heating due to the ics pair front occurs at small @xmath16 , near the magnetic pole for normal pulsars , but in the outer part of the pc for ms pulsars .
the total positron heating luminosity , scaled with the spin - down luminosity , as a function of characteristic pulsar age , @xmath3 is shown in figure 8 for two different values of pc temperature . the numerically computed @xmath185 increases with @xmath186 , in agreement with the analytic estimate ( eqs .
[ [ f+r ] ] and [ [ f+nr ] ] ) , as long as there is locally complete screening . when complete screening is no longer achieved , at a @xmath186 value that depends on @xmath21 and @xmath119 , the heating rate saturates because the number of pairs produced is not sufficient to return a positron flux proportional to @xmath46 .
even at very high pc temperatures , the ms pulsars have reached saturation of the pc heating rate because they are beyond the ics screening boundary .
the heating luminosity is larger for a lower value of @xmath119 before saturation because the pff is at higher altitude and the potential drop is higher , giving the positrons more energy before they reach the ns surface .
figure 9 displays the same calculations of the pc heating luminosity , not scaled to the spin - down luminosity , to more easily compare with observations . in these figures
we have also plotted for comparison the calculations of pc heating luminosity from the cr pffs presented in paper i. the high @xmath186 end of the cr heating rate curves marks the cr pff pair death line for that period . where heating from positrons returning from the cr pff is present
, it is several orders of magnitude larger than the heating from the ics pff , so that in these pulsars heating from the ics pairs makes a negligible contribution to the total pc heating rate . for normal pulsars ,
heating from ics pair fronts is not high enough to be detectable at present , even in the absence of cr pair heating .
however , emission from ics pairs provides the only source of pc heating for the known ms pulsars , which have ages @xmath187 yr , since they can not produce cr pairs .
this emission may be detectable if the pc temperatures are above @xmath188 k. the relatively high ics pair heating rates in the ms pulsars is due primarily to the higher voltage drop necessary to make pairs . even though the fraction of returning positrons is not large , the positrons gain more energy before hitting the ns surface .
the analytic estimates for @xmath177 are in reasonably good agreement with our numerical calculations ( within a factor of 10 ) in the regime where locally complete screening is occurring . where compete screening is not achieved
, the analytic estimates of @xmath177 and also of @xmath178 given in section 5.1 will greatly overestimate the true values .
this is especially true in the case of the ms pulsars .
we caution that equations ( [ l+r ] ) - ( [ tnr ] ) can not be used for pulsars that are beyond the screening boundaries of figure 4 . in figure 10
, we show the predicted luminosities from ics pc heating , @xmath177 as a function of pc surface temperature ( solid lines ) for two different pulsar periods and a surface field of @xmath189 g , so that the corresponding characteristic ages would be @xmath190 yr for p = 2 ms and @xmath191 yr for p = 5 ms .
there is a definite dependence @xmath192 seen in these curves , which can not be modeled analytically since these cases do not have full screening .
also plotted ( dashed lines ) are the luminosities , @xmath193 , emitted by a blackbody at pc radiating temperature @xmath194 and different heated surface areas @xmath195 .
the intersections of the curves roughly indicate values of temperature where self - sustained heating , ( i.e. where the surface emission at a given temperature supplies just enough returning positrons from ics pairs to maintain the pc at that temperature ) is possible for heated pcs of a particular area . since the standard pc area is @xmath196 for p = 2 ms and @xmath197 for p = 5 ms , self - sustained pc heating emission requires heated and radiating areas much smaller than the entire pc area .
the intersection points of the blackbody and @xmath177 curves in figure 10 are not entirely self - consistent self - sustained heating models since we assume a radiating area of @xmath198 for our calculations .
the fully self - consistent models would have radiating areas somewhat larger at a given temperature , but still smaller than @xmath198 .
of the six millisecond radio pulsars which have been detected as pulsed x - ray sources , most have narrow pulses and power - law spectra indicating that their emission is dominated by non - thermal radiation processes . however , two pulsars , psr j0437 - 4715 ( p = 5.75 ms , @xmath199 yr ) and psr j2124 - 3358 ( p = 4.93 ms , @xmath200 yr ) , have possible thermal emission components which would imply that some surface heating is taking place .
the emission from psr j0437 - 4715 has a dominant power - law component but a two - component model is required for an acceptable fit .
a two - component power - law plus blackbody fit to combined euve and rosat data ( halpern et al .
1996 ) give a temperature of @xmath201 k , luminosity @xmath202 and emitting area @xmath203 for the thermal component .
recent chandra observations of psr j0437 - 4715 ( zavlin et al . 2001 ) confirm that at least one thermal component plus a power law is needed to fit the spectrum .
their preferred model consists of a two - temperature thermal blackbody , with a hotter pc core of @xmath204 k and @xmath205 km and a cooler pc rim of @xmath206 k and @xmath207 km . a blackbody fit to asca emission from psr j2124 - 3358 ( sakurai et al .
2001 ) yields a temperature @xmath208 k , luminosity @xmath209 and pc emitting area @xmath210 .
comparing our results in figure 10 with the observed values of @xmath194 , @xmath211 and @xmath195 for psr j0437 - 4715 and psr j2124 - 3358 , we see that the predicted luminosities , areas and temperatures for self - sustained pc heating are in a range comparable to observed values .
thus ics pair fronts could be a plausible source of pc heating for some ms pulsars .
recent chandra high - resolution x - ray observations of the galactic globular cluster 47 tuc ( grindlay et al .
2001a ) have detected all 15 known ms radio pulsars in the cluster as well as a number of other suspected ms pulsars .
it is thought that the x - ray source population of 47 tuc is dominated by ms pulsars having soft spectra and luminosities @xmath212 .
grindlay et al .
( 2001b ) have found a dependence @xmath213 for the ms pulsars in the cluster , not too different from what we have found for pc heating from ics pair fronts ( cf .
figure 9b ) .
we have explored production of electron - positron pairs by photons produced through ics of thermal x - rays by accelerated electrons above a pulsar pc . since the accelerating primary electrons can produce pairs from ics photons at much lower energies than are required to produce pairs from cr photons
, it is very important to investigate the consequences of ics pair fronts for @xmath2 screening and for pc heating .
we have defined pair death " lines in @xmath214 space as the boundary of pair production for pulsars , having dipole magnetic fields of surface strength @xmath215 g. operationally , the existence of a pair front is determined by a finite solution to equation ( [ s0 ] ) for @xmath38 , the altitude of the onset of pair creation .
although we are able to give analytic formulae for the altitude of the pffs for different radiation processes , the location of the pair death lines must be determined numerically .
the existence of a pair front requires much less than one pair per primary electron ( but still many pairs from the whole pc beam ) , since the very first pairs are created in the declining high - energy tail of the radiated spectrum , and also much less than the number of pairs per primary required for screening of @xmath2 .
we find that virtually all known radio pulsars are capable of producing pair fronts with ics photons .
a smaller number , less than half , are able to also produce pairs via cr .
if the acceleration model we use is correct and pulsars have dipole fields , then this result implies that relatively few pairs are required for coherent radio emission .
self - consistent calculations show that ics pair fronts produce lower fluxes of returning positrons and lower pc heating luminosities than cr pair fronts .
this is due to the higher efficiency of the ics process in producing pairs at lower altitudes where both the charge deficit required to screen ( and thus the returning positron flux ) and the accelerating voltage drop are smaller . for pulsars with surface magnetic fields in the
normal " range of @xmath216 g , ics heating luminosities are several orders of magnitude lower than cr heating luminosities .
however , for ms pulsars having surface fields in the range @xmath217 g , production of any pairs requires such high photon energies that ics pair fronts occur at higher altitudes , where acceleration voltage drops are high enough to produce significantly more pc heating .
since most ms pulsars can not produce pairs through cr , ics pair fronts provide the only means of external pc heating .
we find that for surface temperatures @xmath218 k , ics heating luminosities are in the range of detection .
we find that ics pairs are able to screen the _ local _ @xmath2 in some pulsars having a high enough pc temperature , but that this local screening will not produce a complete screening of the accelerating field at all altitudes and thus will not stop the acceleration of the primary beam .
this is because the number of ics pairs grows slowly , on a scale length comparable to the altitude of the pff , and then declines while the charge deficit which maintains @xmath2 continues to increase with altitude .
even if the ics pair production is vigorous enough to achieve local screening of @xmath2 , it eventually can not produce the charge density ( from returning positrons ) to keep up with the increasing charge deficit ( produced by the combination of flaring field lines and inertial frame - dragging ) .
the primary particles may slow their acceleration briefly , due to the local screening , but will resume acceleration once the ics pair production declines . at higher altitudes , those pulsars to the left of the cr pair death line will reach the lorentz factors ( @xmath219 ) required to produce cr pairs .
in contrast to ics pair fronts , the growth of pairs above the cr pff is rapid and robust due to the sensitivity of cr photon energy and emission rate on particle lorentz factor , producing complete screening of @xmath2 in a very short distance ( cf .
paper i ) . in pulsars
to the right of the cr pair death line , there is not complete screening and the primary particles will continue accelerating to high altitude with their lorentz factor being possibly limited by cr reaction .
we have also investigated the proposal by hm98 that pairs produced as the returning positrons are accelerated toward the ns may be able to screen the @xmath2 above the surface . using our calculated values of the returning positron flux in cases where
local screening has been achieved at the ics pair fronts , we find that in most cases screening does not occur at a significant distance above the ns surface to cause disruption of a steady state or formation of pair fronts . in the cases where screening does occur near pffs significantly above the surface , the pulsars are near the ics screening boundary .
the resulting instability would then not move the start of the acceleration to higher altitudes , as hm98 had envisioned , but would probably weaken or disrupt the screening at the upper ics pair front , resulting in a decrease in returning positron flux which would weaken the screening at the lower pff , etc .
pc heating by cr pair fronts will dominate for pulsars to the left of the cr pair death line , while ics pair fronts will supply the pc heating for pulsars to the right of the cr pair death line .
while we have given analytic expressions for the fraction of returning positrons and pc heating luminosities from ics pair fronts , these are only good estimates above the ics screening boundaries of figure 4 . below the ics screening boundaries , where local screening is not achieved for
that pc temperature , the numerical values of returning positron fraction and heating luminosity fall well below the analytic estimates .
this will be true for ms pulsars , nearly all of which are below the ics screening boundary for a pc temperature of even @xmath220 k. our results are dependent on a number of assumptions inherent in our calculations .
first , we have assumed that the pattern of thermal x - ray emission is from a heated pc and is a pure , isotropic blackbody . according to recent calculations ( zavlin et al .
1995 ) of radiation transfer in magnetized ns atmospheres , the thermal emission is not pure blackbody or isotropic , but a somewhat cooler blackbody consisting of pencil and fan beam components .
both effects of full surface ( cooler ) emission and pencil beaming would tend to decrease ics radiation and thus ics pair production efficiency .
however , a large fan beam component would tend to increase ics efficiency .
second , we have used a hybrid scheme to describe the ics radiation spectrum in which the rics has been treated as classical magnetic thompson scattering and nr ( or continuum ) ics has been treated as relativistic but non - magnetic . in reality , both rics and nrics should be treated as a single process with one cross section . while the magnetic qed scattering cross section has been studied for some time ( e.g. herold 1979 , daugherty & harding 1986 ) , simple expressions in limited cases are only beginning to become available ( e.g. gonthier et al . 2000 ) .
our present treatment is probably accurate for magnetic fields @xmath221 which includes most of the radio pulsars .
the location of a pulsar relative to the pair death lines may be important not only to its radio and thermal x - ray emission characteristics , but also to its high - energy emission properties . as we have argued in this paper , ics pair fronts will not limit the acceleration voltage drop in pulsars but that acceleration will continue until it is limited by a cr pair front .
the voltage drop at the cr pair front ( together with the size of the pc current ) is therefore expected to determine the high - energy emission luminosity . in section 4 , we have noted that the acceleration voltage drop of pulsars that produce cr pair fronts is remarkably insensitive to pulsar parameters , leading to the prediction that high - energy luminosity , @xmath222 , should be simply proportional to pc current ( which is proportional to @xmath223 ) , which seems to be borne out by observations ( e.g. thompson 2000 ) .
however , pulsars that do not produce cr pair fronts do not have such a limit on acceleration voltage drop and should depart from the @xmath224 dependence , and approach a @xmath225 dependence . indeed
, such a departure must occur if they are not to exceed 100% efficiency in converting rotational energy loss to high - energy emission .
we predict that this change in @xmath222 dependence should occur along the cr pair death line .
none of the pulsars which have detected @xmath48-ray emission are to the right of the cr pair death line although some , such as geminga , are close .
the large area gamma - ray telescope ( glast ) will have the sensitivity to detect @xmath48-ray emission from significant numbers of radio pulsars beyond the cr pair death line and so should be able to test this prediction .
we thank josh grindlay , jon arons , george pavlov and bing zhang for comments and discussion .
we also thank the referee for valuable comments .
this work was supported by the nasa astrophysics theory program .
arons , j. 1981 , apj , 248 , 1099 .
arons , j. 1983 , apj , 266 , 215 .
arons , j. 1998 , in neutron stars and pulsars , ed . n. shibazaki et al .
( tokyo : univ .
academy press ) , p. 339 .
arons , j. , & scharlemann , e. t. 1979 , apj , 231 , 854 . baring , m. , & harding , a. k. 2001 , apj , 547 , 929 .
becker , w. , & trumper , j. 1997 , 326 , 682 .
chen , k. , & ruderman , m. 1993 , apj , 402 , 264 .
daugherty , j. k. , & harding , a. k. 1986 , apj , 309 , 362 .
dermer , c. 1990 , apj , 360 , 197 .
gonthier , p. l. , harding , a. k. , baring , m. g. , costello , r. m. , & mercer , c. l. 2000 , apj , 540 , 907 .
grindlay , j. e. , heinke , c. , edmonds , p. d. , & murray , s. s. 2001a , science , 292 , 2290 .
grindlay , j. e. , camilo , f. , heinke , c. , edmonds , p. d. , cohn , h. , & lugger , p. 2001b
, apj , submitted .
halpern , j. p. , martin , c. & marshall , h. l. 1996 , apj , 462 , 908 .
harding , a. k. , & preece , r. d. 1987 , apj , 319 , 939 .
harding , a. k. , baring , m. g. , & gonthier , p. l. 1997 , apj , 476 , 246 .
harding , a. k. , & muslimov , a. g. 1998 , apj , 508 , 328 ( hm98 ) .
harding , a. k. , & muslimov , a. g. 2001 , apj , 556 , 987 ( paper i ) .
herold , h. 1979 , phys .
d , 19 , 2868 .
hibschman , j. a. , & arons , j. 2001 , apj , in press ( ha01 ) jones , f. 1968 , phys .
, 167 , 1159 .
manchester , r. n. et al .
2001 , mnras , in press .
muslimov , a. g. , & tsygan , a. i. 1992 , mnras , 255 , 61 ( mt92 ) .
muslimov , a. g. , & harding , a. k. 1997 , apj , 485 , 735 ( mh97 ) .
ruderman , m. , & sutherland , p. g. 1975 , apj , 196 , 51 .
sakurai , i. et al .
2001 , pasj , 53 , 535 .
sturrock , p. 1971 , apj , 164 , 529 .
tademaru , e. 1973 , apj , 183 , 625 .
taylor , j. h. , manchester , r. n. & lyne , a. g. 1993 , apjs , 88 , 529 .
thompson , d. j. 2000 , in high energy gamma - ray astronomy , ed .
f. a. aharonian & h. j. volk ( aip : new york ) , p. 103 .
zavlin , v. e. , pavlov , g.g . ,
shibanov , yu .
a. , & ventura , j. 1995 , a & a , 297 , 441 .
zavlin , v. e. et al .
2001 , apj , submitted .
zhang , b. , qiao , g. j. , lin , w. p. , & han , j. l. 1997 , apj , 478 , 313 .
zhang , b. , & harding , a. k. 2000 , apj , 532 , 1150 .
zhang , b. , harding , a. k. , & muslimov , a. g. 2000 , apj , 531 , l135 . | we investigate the production of electron - positron pairs by inverse compton scattered ( ics ) photons above a pulsar polar cap ( pc ) and calculate surface heating by returning positrons .
this paper is a continuation of our self - consistent treatment of acceleration , pair dynamics and electric field screening above pulsar pcs .
we calculate the altitude of the inverse compton pair formation fronts , the flux of returning positrons and present the heating efficiencies and x - ray luminosities .
we revise pulsar death lines implying cessation of pair formation , and present them in surface magnetic field - period space .
we find that virtually all known radio pulsars are capable of producing pairs by resonant and non - resonant ics photons radiated by particles accelerated above the pc in a pure star - centered dipole field , so that our ics pair death line coincides with empirical radio pulsar death .
our calculations show that ics pairs are able to screen the accelerating electric field only for high pc surface temperatures and magnetic fields .
we argue that such screening at ics pair fronts occurs locally , slowing but not turning off acceleration of particles until screening can occur at a curvature radiation ( cr ) pair front at higher altitude . in the case where no screening occurs above the pc surface
, we anticipate that the pulsar @xmath0-ray luminosity will be a substantial fraction of its spin - down luminosity .
the x - ray luminosity resulting from pc heating by ics pair fronts is significantly lower than the pc heating luminosity from cr pair fronts , which dominates for most pulsars .
pc heating from ics pair fronts is highest in millisecond pulsars , which can not produce cr pairs , and may account for observed thermal x - ray components in the spectra of these old pulsars .
= cmr5 = cmbx5 = cmr6 = cmbx6 = cmr7 = cmbx7 # 1@xmath1 # 1#2 |
the principal aim of the @xmath10 physics experimental programs is to measure the angles ( denoted by @xmath11 , @xmath12 and @xmath1 ) of the triangle representing the unitarity relation : @xmath13 , where @xmath14 is the cabibbo - kobayashi - maskawa ( ckm ) matrix .
the idea is to overdetermine the angles of this triangle and thus test the ckm paradigm of cp violation .
methods to determine @xmath1 ( @xmath15 arg @xmath16 including the electroweak penguin ( ewp ) diagram contribution have been suggested @xcite . to ( over)determine @xmath1 to test the ckm theory of cp violation , we measure @xmath1 using different techniques .
it is thus useful to have new methods to determine @xmath1 . with this motivation , in this letter
, we give two new methods to determine @xmath1 using time _ integrated _
rates for the decays @xmath2 , @xmath8 ( method 1 ) and @xmath17 ( method 2 ) . as in all the other methods mentioned above ,
flavor @xmath0 symmetry is used in both the methods .
we will write the decay amplitudes in terms of flavor @xmath0 invariant amplitudes @xcite .
these are denoted by @xmath18 , @xmath19 , @xmath20 , @xmath21 and @xmath22 and correspond to the @xmath23 linearly independent ways of forming flavor @xmath0 singlets from the initial meson @xmath24 , the two final state mesons belonging to the flavor @xmath0 octet and the effective weak hamiltonian which transforms as a @xmath25 . @xmath26 and @xmath27 denote the parts of these amplitudes generated by tree level and penguin operators respectively .
these invariant amplitudes include soft final state rescattering effects .
some of the methods to determine @xmath1 @xcite , including the ones which use the decays @xmath28 @xcite and the decay @xmath29 @xcite neglect rescattering effects .
in particular , these methods assume that the decay amplitude @xmath30 has no weak phase @xmath31 from the tree level operators . in the language of the @xmath0 invariant amplitudes ,
this is equivalent to assuming that the annihilation amplitudes of @xmath24 is contracted directly with the hamiltonian . ]
@xmath32 are suppressed by @xmath33 and a combination of the @xmath0 invariant amplitudes , @xmath34 is zero ( both of which are valid in the absence of significant rescattering effects ) .
rescattering effects can enhance the annihilation contributions _ and _ lead to significant @xmath34 @xcite . in this letter , we neglect annihilation contributions but _ do not _ assume any relation between @xmath35 , @xmath36 and @xmath37 or the other @xmath0 invariant amplitudes .
thus , rescattering effcts are partly included .
the decay amplitudes for @xmath38 can be written as @xcite @xmath39 here , @xmath40 ( @xmath41 and @xmath42 ) and @xmath43 , @xmath44 denote the penguin amplitudes due to quark @xmath45 running in the loop . using the unitarity of the ckm matrix , _
@xmath46 , we get @xmath47 where @xmath48 and @xmath49 .
a simliar notation is used for @xmath50 and @xmath51 .
henceforth , we will write the decay amplitudes using this notation . the decay amplitudes for @xmath52 are @xcite @xmath53 using eqns.([b0pik ] ) and ( [ b+pik ] ) , we get an expression for @xmath54 , the amplitude for @xmath55 , @xmath38 with final state @xmath56 , @xmath57 the decay amplitude for @xmath58 is @xmath59 the qcd penguin diagram ( which is @xmath60 ) does not contribute to this decay since this decay has the transition @xmath61 .
so , @xmath62 is the ewp contribution .
neubert , rosner @xcite showed that @xmath63 where @xmath64 is the ratio of wilson coefficients ( wc s ) of the ewp operators ( with quark @xmath45 running in the loop ) and the tree level operators in the effective hamiltonian so that @xmath65.\ ] ] the top quark ewp diagram with @xmath66 exchange is enhanced by @xmath67 and so @xmath68 giving @xmath69,\ ] ] where @xmath70 . since @xmath71 and @xmath72 , we get latexmath:[\[- \sqrt{2 } { \cal a } ( b^+ \rightarrow \pi^+ \pi^0 ) \approx 8 \ ; c_{15}^t and setting the strong phase of @xmath37 to zero , _
i.e. , _ the ewp contribution is @xmath74 and can thus be neglected in the decay amplitude @xmath75 .
thus , @xmath76 can be determined directly from the decay rate @xmath58 .
similarly , the expression for @xmath54 ( eqn.([a3/2 ] ) ) simplifies to @xmath77 so that @xmath78 where @xmath79 is given by @xmath80 , _
i.e. , _ in this case , due to the ckm factors , the ewp contribution @xmath81 _ is _ important .
thus , knowing @xmath76 from the decay rate @xmath58 and @xmath1 , we can determine @xmath54 and conversely @xmath1 can be determined if the ( magnitude ) of @xmath54 is known using some other method and the decay rate @xmath7 is measured . in using this relation , it is crucial that the parameter @xmath79 is calculable . in the analysis up to now , annihilation contributions are included .
the decay amplitudes for @xmath2 , @xmath55 decays to @xmath82 and @xmath83 can be written as @xcite @xmath84 @xmath85 where @xmath86 , @xmath87 and @xmath88 are the amplitudes which contribute only to @xmath10 meson decays to a final state involving @xmath89 @xcite .
, @xmath90 is similar to @xmath91 , @xmath92 ( eqn.([not ] ) ) . ]
@xmath87 and @xmath88 are annihilation amplitudes .
we assume that the mass eigenstates @xmath93 and @xmath94 are given by the the canonical mixing : @xmath95 this mixing is consistent with the present data @xcite .
then , the decay amplitudes for @xmath55 , @xmath2 decays to @xmath93 and @xmath96 are @xmath97 @xmath98 from eqns.([b+pik ] ) , ( [ b+keta81 ] ) and ( [ etamix ] ) , we get the relation @xcite @xmath99 = 1 [ figbketa ] as mentioned earlier , the magnitude of @xmath54 is known as a function of @xmath1 in terms of the @xmath7 decay rate ( eqn.([a3/2nr ] ) ) .
this , for a given @xmath1 , we can construct the two triangles formed by @xmath38 , @xmath54 and @xmath52 , @xmath54 ( corresponding to eqns.([a3/2 ] ) and @xmath100 s @xmath101 and @xmath102 respectively in fig.[figbketa ] ) and the quadrangle formed by @xmath52 , @xmath103 , @xmath104 corresponding to eqn.([b+pikketa ] ) ( @xmath105 of fig.[figbketa ] ) .
: @xmath100 s @xmath102 and @xmath101 can be on the same side of the common base @xmath106 and similarly in the quadrangle @xmath105 the vertices @xmath107 and @xmath108 can be on the same side of the diagonal @xmath109 . ]
thus , we know the phases ( in the convention where the phase of @xmath76 is zero ) of the decay amplitudes @xmath52 , @xmath110 , @xmath104 and @xmath111 as a function of @xmath1 form this construction ( the magnitudes are , of course , known from the measurement of the decay rates ) . we have included the annihilation contributions up to now .
if we _ neglect _
the annihilation amplitudes , @xmath112 , @xmath87 and @xmath88 , using eqns.([b0pik ] ) , ( [ b+keta ] ) and ( [ b0keta ] ) , we get the relations : @xmath113 @xmath114 thus , we can predict the decay amplitudes @xmath115 , @xmath116 as a function of @xmath1 since , as mentioned above , we know the magnitudes and phases ( the latter from fig.[figbketa ] ) of all the other decay amplitudes in eqns.([b+b0etakrel1 ] ) and ( [ b+b0etakrel2 ] ) .
in fact , the decay amplitude @xmath115 is shown as @xmath117 in fig.[figbketa ] .
once the decay rate @xmath115 or @xmath116 is measured , @xmath1 can be determined .
thus , @xmath1 can be determined ( up to a four - fold discrete ambiguity ) by the measurement of the decay rates for @xmath118 modes
@xmath119 , @xmath120 , @xmath103 , @xmath104 , @xmath121 , @xmath122 ( or @xmath123 ) .
this method is based on the method of gronau _ et al .
_ @xcite . in @xcite ,
the annihilation amplitudes are neglected _ and _ the relation @xmath124 is assumed . in other words , rescattering effects
are neglected so that the amplitude for the decay @xmath125 has no weak phase @xmath31 .
we neglect the annihilation amplitudes , but do not assume any relation between the @xmath126 s .
thus , we include partly the rescattering effects .
the decay amplitudes for @xmath127 , @xmath89 are @xcite @xmath128 @xmath129 with the canonical mixing ( eqn.[etamix ] ) ) , we get @xmath130 @xmath131 .
\label{bseta2}\end{aligned}\ ] ] neglecting the annihilation ampltudes , @xmath112 , @xmath87 and @xmath88 , from eqns.([b0pik ] ) , ( [ b+pik ] ) , ( [ bseta ] ) and ( [ bseta2 ] ) we get the relations @xcite : @xmath132
@xmath133 = 1.25 [ figbspieta ] from eqns.([bsb0b+rel ] ) and ( [ etaetaprime ] ) we see that the decay amplitudes @xmath38 , @xmath52 form the sides and @xmath134 ( or @xmath96 ) the diagonal of a quadrangle shown in fig.[figbspieta ] .
thus , the measurement of these @xmath23 decay rates fixes this quadrangle s @xmath135 and @xmath136 can be on the same side of the common base @xmath137 in fig.[figbspieta ] . ]
of which the other diagonal is @xmath54 ( see eqns.([a3/2 ] ) and fig.[figbspieta ] ) @xcite .
knowing the magnitude of @xmath54 from fig.[figbspieta ] and the decay rate @xmath7 , we can determine @xmath1 from eqn.([a3/2nr ] ) .
thus , @xmath1 can be determined ( up to a two - fold discrete ambiguity ) by the measurement of the decay rates for @xmath138 modes
@xmath139 , @xmath120 , @xmath121 , and @xmath140 ( or @xmath141 ) . in the method of @xcite , measurement of the rates for the cp - conjugate processes of all the above modes
is also required to determine @xmath1 .
we comment on the accessibility of the various decay modes used in the two methods .
the @xmath2 and @xmath55 decay modes should be accessible at the @xmath142 machines whereas the @xmath4 decay mode will only be accessible at a hadron machine . since the qcd penguin does not contribute to this @xmath143 decay , the decay rate is expected to be small .
the measurements of the decay rates to cp eigenstate final states : @xmath4 , @xmath144 and @xmath145 require external tagging .
as mentioned earlier , we have used flavor @xmath0 symmetry in both the methods . in the factorization approximation ,
@xmath0 breaking in the tree level amplitudes can be incorporated by factors of @xmath146 ( see , for example , @xcite ) .
for example , @xmath147 .
however , since some of the strong penguin operators are @xmath148 , in the penguin amplitudes , the @xmath0 breaking effects are difficult to estimate , but the breaking will still be less than @xmath149 . in method 2 , we use the decay mode @xmath150 which does not have the qcd penguin contribution , but does have the ewp contribution .
the ewp operators @xmath151 have very small wc s whereas the ewp operators with significant wc s , @xmath152 , are fierz - equivalent to the tree level operators @xmath153 @xcite .
so in the factorization approximation , the corrections due to @xmath0 breaking in relating the penguin amplitudes for @xmath150 to the ones for @xmath2 , @xmath154 are given by factors of @xmath155 .
we have also assumed that the @xmath0 breaking in the strong phases is small .
a possible justification is that at the energies of the final state particles @xmath156 , the phase shifts are not expected to be sensitive to the @xmath0 breaking given by , say , @xmath157 ( which is much smaller than the final state momenta ) .
however , it is hard to quantify this effect .
both the methods can be used with cp - conjugates of all the decay modes as well .
we have neglected annihilation amplitudes : @xmath112 , @xmath87 and @xmath88 .
the validity of this assumption can be checked by comparing the decay rates @xmath134 and @xmath158 these two decay amplitudes differ only in the annihilation contribution ( see eqn.([bseta2 ] ) ) . in the absence of significant annihilation contribution , the decay rate for @xmath158 should be twice that for @xmath134 . in summary ,
we have discussed two new methods ( based on flavor @xmath0 symmetry ) to determine the weak phase @xmath1 using the decays @xmath2 , @xmath159 ( @xmath94 ) and @xmath134 @xmath160 , respectively .
these methods partly take into account rescattering effects .
m. gronau , o.f .
hernndez , d. london , j.l .
rosner , phys .
* d52 * , 6374 ( 1995 ) .
deshpande , x - g .
he , phys .
75 * , 3064 ( 1995 ) .
m. gronau , j.l .
rosner , phys .
rev . * d53 * , 2516 ( 1996 ) .
m. neubert , j.l .
rosner , phys .
lett . * 81 * , 5076 ( 1998 ) .
buras , r. fleischer , hep - ph/9810260 ; m. gronau , d. pirjol , hep - ph/9811335 ; k. agashe , n.g .
deshpande , hep - ph/9812278 , to be published in phys .
* b*. see , for example , a.s .
dighe , phys .
rev . * d54 * , 2067 ( 1996 ) .
see , for example , m. neubert , phys .
b424 * , 152 ( 1998 ) ; m. gronau , j.l .
rosner , phys .
rev . * d58 * , 113005 ( 1998 ) .
m. neubert , j.l .
rosner , phys .
b441 * , 403 ( 1998 ) .
gilman , r. kauffman , phys . rev .
* d36 * , 2761 ( 1987 ) ; erratum - ibid.*d37 * , 3348 ( 1988 ) . | we suggest two methods ( based on flavor @xmath0 symmetry ) to determine the ckm angle @xmath1 using the decays @xmath2 , @xmath3 and @xmath4 , respectively . rescattering effects
are partly included
we neglect annihilation amplitudes , but do not assume any other relation between the @xmath0 invariant amplitudes .
we use the fact that the amplitude ( including the electroweak penguin contribution ) for @xmath2 , @xmath5 with final state @xmath6 is known as a function of @xmath1 from the decay rate @xmath7 .
epsf oits-671 + .05 in * determining the weak phase @xmath1 using the decays @xmath2 , @xmath8 and @xmath9 * .15 in k. agashe and n.g .
deshpande .1 in _ institute of theoretical science + university of oregon + eugene or 97403 - 5203 _ .05 in |
the relationship between anomalous charge transport and magnetism in the vicinity of metal - insulator transition is a longstanding subject in the physics of strongly correlated electron systems .
the ru - based metal caruo@xmath2 is considered to have a paramagnetic ground state with itinerant characteristics of ru 4d electrons @xcite . in caruo@xmath2
, magnetization is found to be significantly enhanced at low temperatures @xcite , while no long - range magnetic order is observed in microscopic measurements such as neutron scattering and nmr @xcite .
in addition , electrical resistivity @xmath12 shows an unusually large magnitude at high temperatures , corresponding to a very small mean - free pass of electron conduction comparable to lattice constants @xcite .
these features imply that caruo@xmath2 involves instabilities for both magnetism and charge transport , which are enhanced by strong electronic correlation between itinerant ru 4d electrons . in the mixed compound caru@xmath0mn@xmath1o@xmath2 @xcite
, it is revealed that a ferromagnetic ( fm ) order evolves above @xmath3 , and becomes stable in the intermediate @xmath7 range .
both spontaneous magnetization @xmath13 and curie temperature @xmath14 show maximum values at @xmath15 ( @xmath16 and @xmath17 ) , followed by the reductions of them with further increasing @xmath7 .
furthermore , an antiferromagnetic ( afm ) spin arrangement with the g - type structure is observed in the fm ordered phase .
finally , at @xmath18 the fm order vanishes , and only the g - type afm order appears below @xmath19 @xcite . by doping mn , the feature of @xmath12 also changes from metallic to less conductive above @xmath20 , but the magnitude of @xmath12 still remains in small values in the wide @xmath7 range , that is , smaller than @xmath21 cm even at low temperatures for @xmath22 . at @xmath18 , in contrast , @xmath12 dramatically increases with decreasing temperature , whose magnitude at low temperatures exceeds @xmath23 @xmath24 @xcite .
these properties indicate that doping mn into caruo@xmath2 modifies the itinerant electronic states due to strong correlations between ru 4d and mn 3d electrons , and it significantly affects the magnetic and transport properties . to clarify the variation of the d electronic states
, we have performed the low - temperature specific - heat measurements for caru@xmath0mn@xmath1o@xmath2 .
polycrystalline samples of caru@xmath0mn@xmath1o@xmath2 for @xmath25 were prepared by means of the conventional solid - state method . the mixture of appropriate amounts of caco@xmath2 , ruo@xmath26 and mno were first calcined at 800 @xmath27c for 24 hours .
they were shaped into pellets after careful mixing , and then sintered at 1200 @xmath27c for 48 hours . by means of powder x - ray diffraction technique
we checked that all the samples have a distorted perovskite structure ( the gafeo@xmath2-type orthorhombic structure ) @xcite without any extrinsic phase .
specific heat @xmath28 was measured between 3 k and 275 k with a thermal - relaxation method using both a commercial system ( ppms : quantum design ) and a hand - made equipment , where we used plate - shaped samples with the mass of about 4 mg and 40 mg , respectively .
the thermal - relaxation curves for all the measurements were well definitive , and the @xmath28 data obtained from both equipments were consistent within the experimental accuracy .
ac - susceptibility was measured in the temperature range of 5 - 300 k to estimate the fm and afm transition temperatures .
frequency and amplitude of the applied ac field were 180 hz and @xmath29 , respectively .
mn@xmath1o@xmath2 . in ( a ) , closed and open arrows indicate the onsets of fm and afm orders , respectively , estimated from the peak position of the ac susceptibility .
the types of magnetic order are determined from the previous reports on magnetization and neutron scattering measurements @xcite .
the inset of ( a ) shows an enlargement of @xmath28 around the afm transition temperatures for @xmath30 and 1 . ,
scaledwidth=75.0% ] figure 1(a ) shows temperature variations of the specific heat @xmath31 for caru@xmath0mn@xmath1o@xmath2 .
clear jumps ascribed to the afm transition are observed at 122.7 k for @xmath18 and 125.1 k for @xmath30 ( see inset of fig .
1(a ) ) , but no distinct anomaly is seen in the @xmath31 curves for @xmath22 .
this contrasts with temperature dependence of the real part of ac susceptibility @xmath32 for @xmath33 , where the fm and afm transitions are indicated by the appearance of peaks ( fig .
the transition temperatures estimated from @xmath32 are consistent with those determined by the dc magnetization and neutron scattering measurements @xcite .
the peaks seen in @xmath32 are somewhat broaden probably due to the distribution of the transition temperatures in the sample generated by doping .
we consider this effect to be one of the reasons for the reduction of jumps in @xmath31 . on the other hand ,
it is interesting that similar reduction of anomaly is observed in the @xmath31 data for mn - doped srruo@xmath2 @xcite .
srruo@xmath2 is a ferromagnet with itinerant characteristics of ru 4d electrons , and @xmath31 exhibits a jump at @xmath34 . in the mixed compound srru@xmath0mn@xmath1o@xmath2 , the anomaly at @xmath14 in @xmath31 is rapidly suppressed in the fm metallic phase for @xmath35 , but a shoulder - like anomaly again appears in @xmath31 at the afm transition temperatures in the strongly insulating region for @xmath36 . as for caru@xmath0mn@xmath1o@xmath2 ,
the electrical resistivity is much reduced for @xmath22 .
the disappearance of anomaly in @xmath31 may thus be related to the small entropy change associated with the magnetic transitions in the small - resistivity region , as well as the broadening of the onset due to the inhomogeneity . ,
( b ) electronic specific - heat coefficient @xmath6 and lattice contribution on specific heat @xmath37 , and ( c ) @xmath14 and @xmath38 estimated from the @xmath32 data for caru@xmath0mn@xmath1o@xmath2 . the definition of parameters @xmath6 and @xmath37 is described in the text .
the solid lines in ( c ) are guides to the eye @xcite . , scaledwidth=75.0% ] displayed in fig .
2(a ) is low - temperature specific heat divided by temperature @xmath4 plotted as a function of @xmath5 .
@xmath4 for entire @xmath7 range are roughly proportional to @xmath5 .
the finite @xmath4 values for @xmath39 except at @xmath18 indicate that the fermi - liquid excitation as well as phonon are mainly responsible for @xmath31 .
however , @xmath4 in the small and intermediate @xmath7 range shows a very weak deviation from the @xmath5 function with a downward curvature below @xmath40 ( @xmath41 ) . it may be ascribed to the spin fluctuation generated in the magnetically ordered phase .
the @xmath4 data are fitted with a function of @xmath42 to derive the electronic specific heat coefficient @xmath6 . in fig .
2(b ) , we show @xmath7 variations of @xmath6 and @xmath37 obtained from the best fit in the temperature range between 3 k and 10 k. for comparison , we also plot the @xmath43 phase diagram obtained from the @xmath32 data in fig .
2(c ) . the @xmath6 value for @xmath8 is estimated to be 84 mj / k@xmath9 mol , which is consistent with that reported previously @xcite .
the magnitude of @xmath6 does not change in the paramagnetic phase for @xmath10 , but linearly decreases with increasing @xmath7 in the magnetically ordered region ( @xmath44 ) . it is highly reduced and becomes @xmath45 or less for @xmath46 , where @xmath14 decreases through the maximum at @xmath15 . on the other hand , @xmath37 is estimated to be 0.19 mj / k@xmath47 mol for @xmath8 that yields the debye temperature of @xmath48 .
it is reduced in mn - rich region and finally becomes 0.085 mj / k@xmath47 mol ( @xmath49 ) at @xmath18 .
this feature seems to be consistent with the volume contraction in mn - rich concentrations found by x - ray diffraction measurements @xcite . in caru@xmath0mn@xmath1o@xmath2
, it is proposed that a double - exchange mechanism between mn ions plays a crucial role in the evolution of the fm order in mn - rich region , and in lower @xmath7 range an interplay of double - exchange and superexchange interactions between d electrons located at ru and mn ions may yield the formation of inhomogeneous fm and afm orders @xcite . on the other hand , the present @xmath28 investigation revealed that @xmath6 shows relatively large values in small and intermediate @xmath7 ranges , while it is strongly reduced in mn - rich region .
in particular , @xmath6 at @xmath50 is estimated to be 38 mj / k@xmath9 mol , which is almost the same magnitude as that for the ferromagnetic metal srruo@xmath2 ( 33 mj / k@xmath9 mol ) .
the large @xmath6 values indicate that d electrons have itinerant characteristics in this @xmath7 range .
furthermore , we observed that @xmath6 is nearly constant in the paramagnetic region , and it is then reduced when the fm order occurs above @xmath3 .
such a variation of @xmath6 can not be simply attributed to an increase in the fraction of d electrons localized at mn ions in the sample , but strongly suggests that the itinerant d electrons originating from both ru and mn are tightly coupled with the magnetic orders .
this is supported by our recent magnetization and neutron - diffraction measurements for @xmath50 , where the magnetization shows typical features on fluctuation expected from itinerant electron spins , and the neutron - diffraction experiments revealed the occurrence of the long - range magnetic orders @xcite .
we performed @xmath28 measurements for caru@xmath0mn@xmath1o@xmath2 ( @xmath25 ) .
low - temperature @xmath6 value is estimated to be 84 mj / k@xmath9 mol for pure caruo@xmath2 , and it is unchanged by doping mn in the paramagnetic region for @xmath10 . in the magnetically ordered phase , @xmath6 is linearly reduced with increasing @xmath7 for @xmath51 , and then approaches @xmath52 mj / k@xmath9 mol or less for @xmath53 .
the relatively large magnitude of @xmath6 in the small and intermediate mn concentrations suggests that d electrons have itinerant characteristics contrary to localized ones expected in mn - rich concentrations , and they play important roles in the magnetic orderings .
we thank y. nishihara and a. yamaji for fruitful discussion .
this work was carried out by the joint research in issp , the university of tokyo , and partly supported by a grant - in - aid for scientific research on innovative areas heavy electrons " ( no.23102703 ) from the ministry of education , culture , sports , science and technology of japan .
99 cao g , mccall s , shepard m , crow j e and guertin r p , 1997 _ phys .
b _ * 56 * 321 kiyama t , yoshimura k , kosuge k , michor h and hilscher g , 1998 _ j. phys .
jpn . _ * 67 * 307 kiyama t , yoshimura k , kosuge k , mitamura h and goto t , 1999 _ j. phys .
jpn . _ * 68 * 3372 ferner i , nowik i , bradaric i and gospodinov m , 2000 _ phys . rev .
b _ * 62 * 11332 martinez j l , prieto c , rodriguez - carvajal j , de andres a , vallet - regi m and gonzalez - calbet j m , 1995 _ j. magn .
magn mater .
_ * 140 - 144 * 179 - 180 mukuda h , ishida k , kitaoka y , asayama k , kanno r and takano m , 1999 _ phys . rev .
b _ * 60 * 17 yoshimura k , imai t , kiyama t , thurber k r , hunt a w and kosuge k , 1999 _ phys .
lett . _ * 83 * 4397 cao g , korneta o , chikara s , delong l e and schlottmann p , 2008 _ solid state commun . _ * 148 * 305 sugiyama t and tsuda n , 1999 _ j. phys
jpn . _ * 68 * 1306 shames a i , rozenberg e , martin c , maignan a , raveau b , andr g and gorodetsky g , 2004 _ phys . rev .
b _ * 70 * 134433 markovich v , auslender m , fita i , puzniak r , martin c , wisniewski a , maignan a , raveau b and gorodetsky g , 2006 _ phys
b _ * 73 * 014416 taniguchi t , mizusaki s , okada n , nagata y , lai s h , lan m d , hiraoka n , itou m , sakurai y , ozawa t c , noro y and samata h , 2008 _ phys .
b _ * 77 * 014406 mizusaki s , taniguchi t , okada n , nagata y , hiraoka n , itou m , sakurai y , noro y , ozawa t c and samata h , 2009 _ j. phys . : condens . matter _ * 21 * 276003 kawanaka h , yokoyama m , noguchi a , bando h and nishihara y , 2009 _ j. phys . :
condens . matter _
* 21 * 296002 wollan e o and koehler w c , 1955 _ phys .
_ * 100 * 545 neumeier j j and cohn j l , 2000 _ phys .
b _ * 61 * 14319 yamaji a , yokoyama m , nishihara n , narumi y and kindo k , 2010 _ j. phys . :
_ * 200 * 012232 .
kikugawa n , balicas l and mackenzie a p , 2009 _ j. phys .
jpn . _ * 78 * 014701 | low - temperature specific heat of caru@xmath0mn@xmath1o@xmath2 was measured to clarify the role of d electrons in ferromagnetic and antiferromagnetic orders observed above @xmath3 . specific heat divided by temperature @xmath4
is found to roughly follow a @xmath5 function , and relatively large magnitudes of electronic specific heat coefficient @xmath6 were obtained in wide @xmath7 range . in particular
, @xmath6 is unchanged from the value at @xmath8 ( 84 mj / k@xmath9 mol ) in the paramagnetic state for @xmath10 , but linearly reduced with increasing @xmath7 above @xmath11 .
these features of @xmath6 strongly suggest that itinerant d electrons are tightly coupled with the evolution of magnetic orders in small and intermediate mn concentrations . |
the main result of this paper ( theorem [ main ] ) says that the special fiber of a galois branched cover of the open @xmath0-adic disc is determined by characteristic zero fibers .
the motivation for such a theorem comes from the _ global lifting problem for galois covers of curves _ : if @xmath1 is a finite group , @xmath2 is an algebraically closed field of characteristic @xmath3 , and @xmath4 is a finite @xmath1-galois branched cover of smooth projective @xmath2-curves , does there exist a lifting of @xmath5 to a @xmath1-galois branched cover of smooth projective @xmath6-curves , where @xmath6 is a discrete valuation ring of mixed characteristic with residue field @xmath2 ?
this problem has been much studied ; see e.g. @xcite , @xcite , @xcite , @xcite , @xcite , @xcite . as a brief survey of the subject , we mention the following results : * ( grothendieck , @xcite ) if @xmath5 is at most tamely ramified , then a lifting exists over @xmath6 , where @xmath6 is any complete dvr with residue field @xmath2 . moreover , the lifting is unique once we fix a lifting of @xmath7 and the branch locus of @xmath5 .
* not all wildly ramified covers are liftable : there exist curves of genus @xmath8 in positive characteristic whose automorphism groups are too large to be automorphism groups of genus @xmath9 curves in characteristic zero , by the hurwitz genus bounds
. if @xmath10 is such a curve , it is then clear that @xmath11 is not liftable to mixed characteristic .
* there are examples of non - liftable @xmath0-elementary abelian covers .
see @xcite for an example with @xmath12 for @xmath13 .
* oort , sekiguchi , and suwa showed in @xcite that cyclic covers lift if @xmath0 divides @xmath14 at most once .
their method was global in nature , involving a group scheme degeneration of kummer theory to artin - schreier theory .
* using local methods ( see below ) , green and matignon proved in @xcite that cyclic covers lift if @xmath0 divides @xmath14 at most twice . * pagot has shown in @xcite that klein - four covers always lift .
* @xmath15-covers are liftable by @xcite , where the method of differential data is used .
* green has shown in @xcite that the @xmath16-cyclic covers that occur as automorphisms of lubin - tate formal groups are liftable .
the guiding conjecture in the subject was provided by f. oort ( @xcite , i.7 ) who suggested the cyclic covers always lift . in section 5
, we derive an arithmetic reformulation of a stronger form of this conjecture , which specifies the ring @xmath6 over which the lifting should occur : if @xmath4 is cyclic of order @xmath17 with @xmath18 , then @xmath5 lifts over @xmath19 $ ] , where @xmath20 denotes the witt vectors of @xmath2 .
a major breakthrough in the subject came when green and matignon discovered that the obstructions to lifting are not global in nature , but rather local . indeed , using either rigid patching @xcite or deformation theory @xcite , the global lifting problem reduces to a local one , due to the _
( @xcite section iii , @xcite corollaire 3.3.5 ) _ let @xmath21 be a ramification point for the @xmath1-galois cover @xmath4 , and consider the @xmath22-galois cover @xmath23 obtained by completion , where @xmath22 is the inertia subgroup at @xmath24 .
suppose that for each such ramified @xmath21 , the map @xmath25 can be lifted to a @xmath22-galois cover of open @xmath0-adic discs , @xmath26 , where @xmath27)$ ]
. then the local liftings , @xmath28 , can be patched together to yield a lifting of @xmath5 to a @xmath1-galois cover of smooth projective @xmath6-curves . since @xmath29 $ ] , we are led to consider the following _ local lifting problem for galois covers of curves _ : given a finite @xmath1-galois extension of power series rings @xmath30|k[[z]]$ ] , does there exist a lifting to a @xmath1-galois extension @xmath31|r[[z]]$ ] , where @xmath6 is a mixed characteristic dvr with residue field @xmath2 ? one could also consider the ( weaker ) _ birational local lifting problem for galois covers of curves _ : given a finite @xmath1-galois extension of laurent series fields @xmath32 , does there exist a lifting to a @xmath1-galois extension of normal rings @xmath33 $ ] ? by a lifting in this case ,
we mean that @xmath34 is an integral domain ( where @xmath35 is a uniformizer for @xmath6 ) , and the fields @xmath36 and @xmath37 are isomorphic as @xmath1-galois extensions of @xmath38 . in particular , in the birational version of the local problem , we do not require @xmath39 to be smooth : in terms of geometry , this corresponds to allowing the curve @xmath10 to acquire singularities . as mentioned above
, the local lifting problem does not always have a positive solution . on the other hand ,
garuti has shown in @xcite that the birational local lifting problem _ does _ always have a positive solution , so the birational problem is indeed weaker than the local lifting problem .
it is clearly advantageous to work with laurent series fields rather than power series rings , however , and the following criterion for good reduction ensures that we may do so without sacrificing the smoothness of our liftings . _
( @xcite section 5 , @xcite 3.4 ) _ let @xmath40 be a normal integral local ring , which is also a finite @xmath41$]-module .
assume moreover that @xmath34 is reduced and @xmath42 is separable .
let @xmath43 be the integral closure of @xmath44 , and define @xmath45
. also , setting @xmath46 , denote by @xmath47 the degree of the different of @xmath48\otimes k)$ ] , and by @xmath49 the degree of the different of @xmath42 .
then @xmath50 , and if @xmath51 , then @xmath52 $ ] . using this criterion , we obtain the suppose that @xmath30|k[[z]]$ ] is a @xmath1-galois extension of power series rings
. then a @xmath1-galois extension of normal integral local rings , @xmath33 $ ] , is a lifting of @xmath30|k[[z]]$ ] if and only if it is a birational lifting of @xmath32 , and @xmath53 .
hence , the local lifting problem can be reformulated as follows : given a @xmath1-galois extension @xmath32 , does there exist a @xmath1-galois birational lifting @xmath33 $ ] which preserves the different , i.e. such that @xmath53 ?
it is this last formulation of the local lifting problem that provides our motivation for studying galois covers of the open @xmath0-adic disc .
in particular , given such a @xmath1-galois branched cover @xmath54 , we are interested in determining geometric and arithmetic properties of the special fiber @xmath55 ( such as irreducibility , separability , and the degree of the different @xmath49 ) from the corresponding properties of the generic fiber @xmath56 and its specializations at various points @xmath57 . our main result ( theorem [ main ] ) provides precisely such a characterization of the special fiber in terms of characteristic zero data .
roughly speaking , our result says that the special fiber of a galois cover of the open @xmath0-adic disc `` wants '' to be the field of norms of the characteristic zero fibers , and the degree to which this fails is the phenomenon of inseparability .
our work can be regarded as a concrete investigation of the class field theory of the open @xmath0-adic disc , and our main result suggests that the local lifting problem would be answered by a grunwald - wang type theorem for the open disc , with control over the generic different . in section 2 of this paper
we review the basic structure of the open @xmath0-adic disc , and then in section 3 we describe the theory of the field of norms due to fontaine and wintenberger , which plays a major role in our main result .
section 4 contains the proof of our main theorem characterizing the special fiber of a galois branched cover of the open @xmath0-adic disc in terms of the characteristic zero fibers of the cover .
an arithmetic reformulation of the oort conjecture is deduced in section 5 , together with a new proof of this conjecture in the @xmath0-cyclic case .
let @xmath58 be a complete discretely valued field with residue characteristic @xmath3 .
we make the following notational conventions : * @xmath59 denotes the valuation ring of @xmath58 ; * @xmath60 denotes the maximal ideal of @xmath59 ; * @xmath61 denotes the residue field of @xmath59 ; * @xmath62 denotes the normalized discrete valuation on @xmath58 , so that @xmath63 ; * @xmath64 is the absolute value on @xmath58 induced by @xmath62 , normalized so that @xmath65 . *
if @xmath66 is the completion of an algebraic extension of @xmath58 , then we also denote by @xmath62 ( resp .
@xmath64 ) the unique prolongation of @xmath62 ( resp .
@xmath64 ) to @xmath66 .
let @xmath58 be a complete discretely valued @xmath0-adic field , with valuation ring @xmath67 .
then the _ open @xmath0-adic disc _ ( over @xmath58 ) is defined to be @xmath68\otimes_r k)$ ] , and its smooth integral model is denoted by @xmath27)$ ] .
the key result for understanding the structure of the open @xmath0-adic disc is the _
( @xcite vii.3.8 , prop .
6)_[wpt ] suppose that @xmath69 $ ] has a nonzero _ reduced series _
@xmath70 $ ] , of valuation @xmath71 .
then @xmath72 can be written uniquely as @xmath73 where all @xmath74 and @xmath75 is a unit in @xmath41 $ ] .
the degree @xmath76 is called the _ weierstrass degree _ of @xmath72 .
polynomials @xmath77 with coefficients from @xmath78 as in the proposition are called _ distinguished polynomials _ , and we see that the ring @xmath41 $ ] has the following properties : it is a 2-dimensional regular local ring with maximal ideal @xmath79 , where @xmath35 is a uniformizer for @xmath6 . moreover , if @xmath80 is a height 1 prime of @xmath41 $ ] , then either @xmath81 , or @xmath82 for some irreducible distinguished polynomial in @xmath83 $ ] .
it follows that @xmath41\otimes k$ ] is a dedekind domain whose maximal ideals are in one - to - one correspondence with the irreducible distinguished polynomials over @xmath6 .
finally , the geometric points of @xmath84 can be described as : @xmath85 \ \textrm{irred .
distinguished}\right\}\\ & = & \left\{\alpha\in \overline{k } \ | \
|\alpha|_k<1 \right\},\end{aligned}\ ] ] which explains the name of @xmath84 . as a consequence of the weierstrass preparation theorem
, we see that an arbitrary nonzero power series @xmath69 $ ] can be written in the form @xmath86 , where @xmath87 , the polynomial @xmath88 is distinguished of degree @xmath89 , and @xmath75 is a unit . in the course of our investigation , we will often have to work with the ring @xmath41_{\mathfrak{m}}$ ] , the local ring of @xmath90 at the generic point of the special fiber @xmath91)\subset d$ ] .
the previous remarks imply that an arbitrary nonzero element @xmath92_{\mathfrak{m}}$ ] has the form @xmath93 where the @xmath94 are distinguished polynomials .
in particular , the denominator @xmath95 will be relatively prime to almost all height one primes of @xmath41 $ ] , so if @xmath96 is one of these primes , we will have @xmath92_\mathcal{p}$ ] , and it will make sense to look at the image of @xmath97 in @xmath41_\mathcal{p}/\mathcal{p}\cong k(\alpha)$ ] , where @xmath98 is a root of @xmath99 in @xmath100 . when we have chosen a particular root @xmath98 , we will refer to the image of @xmath97 in @xmath101 as the _ specialization _ of @xmath102 at the point @xmath103 , and denote it by @xmath104 .
more generally , if @xmath105 is a polynomial with coefficients in @xmath41_\mathfrak{m}$ ] , then we can apply the previous reasoning to each of the finitely many coefficients @xmath106 .
we conclude that for almost all points @xmath103 , we can specialize to obtain the polynomial @xmath107.\ ] ] in what follows , we will refer to this argument ( which allows us to specialize polynomials almost everywhere ) as the _
weierstrass argument_. suppose that @xmath108 is a sequence of points corresponding to a sequence @xmath109 with each @xmath110 being a uniformizer for the discrete valuation field @xmath111 .
moreover , suppose that @xmath112 as @xmath113 , so that the points @xmath114 are approaching the boundary of @xmath84 .
equivalently , we are assuming that the ramification index @xmath115 goes to @xmath116 with @xmath117 .
given @xmath118_\mathfrak{m}$ ] , we can consider the specialization of @xmath102 at @xmath119 for @xmath120 ( by the weierstrass argument ) .
we find that @xmath121 note that for any @xmath122 we have @xmath123 .
hence , if @xmath124 is the weierstrass degree of @xmath94 , then for @xmath120 we have @xmath125 for all @xmath122 , since @xmath126 .
it follows that @xmath127 , so @xmath128 thus , we see that the normalized valuations of the specializations @xmath129 are bounded below by @xmath130 .
moreover , if @xmath131 ( i.e. if @xmath132 ) , then @xmath133 as @xmath134 . finally , if @xmath135 , then @xmath136 for @xmath120 , even if @xmath137 .
the preceding remarks can also be applied to the finitely many coefficients of a polynomial @xmath138_{\mathfrak{m}}$ ] as in ( [ spoly ] ) .
we thereby obtain a uniform lower bound on the normalized valuations of the coefficients of @xmath139 , independently of @xmath117 .
moreover , it is easy to check whether these specialized coefficients are integral , and whether their valuations remain bounded as @xmath134
. we will refer to this argument ( which yields information on the valuations of specializations ) as the _ ramification argument _ in the sequel .
in this section we briefly recall the definitions and important properties of the upper and lower ramification filtrations .
see @xcite , chapter iv for a complete treatment .
let @xmath58 be a complete discretely valued field , and @xmath140 a finite galois extension .
then we define a function @xmath141 by @xmath142 the function @xmath143 just defined differs slightly from the function @xmath144 defined by serre in @xcite : @xmath145 for @xmath146 .
we prefer to use the function @xmath143 in order to match the notations of @xcite .
let @xmath140 be a finite galois extension with group @xmath1 .
then for an integer @xmath147 , @xmath148 is called the _ @xmath149th ramification subgroup in the lower numbering_.
we extend the indexing to the set of real numbers @xmath150 by setting @xmath151 for @xmath152 the normal subgroups @xmath153 form a decreasing and separated filtration of @xmath1 , with @xmath154 and @xmath155 the inertia subgroup of @xmath1 .
the _ herbrand function of @xmath140 _ , @xmath156 , is given by @xmath157 the function @xmath158 is an increasing , continuous , piecewise - linear bijection , and hence has an increasing , continuous , and piecewise - linear inverse @xmath159 . using the inverse function @xmath159 , we define a new ramification filtration on the galois group @xmath1 . for a real number @xmath160 ,
define @xmath161 the decreasing and separated filtration @xmath162 is called _ the ramification filtration in the upper numbering_. note that @xmath163 and @xmath164 for @xmath165 . the importance of the upper ramification filtration stems from the fact that it behaves well under the formation of quotient groups : [ upperram ] let @xmath166 be a normal subgroup of @xmath167 , with fixed field @xmath168 .
then the upper ramification filtration on @xmath169 is induced by the upper ramification filtration of @xmath1 : @xmath170 now suppose that @xmath140 is an infinite galois extension .
then proposition [ upperram ] allows us to define an upper ramification filtration on the profinite group @xmath146 : for a real number @xmath160 , define @xmath171 where the limit is over all finite galois subextensions @xmath172 of @xmath140 . the groups @xmath162 form a decreasing , exhaustive , and separated filtration of @xmath1 by closed normal subgroups , called the _ upper ramification filtration_. we say that a real number @xmath173 is a _ jump _ for the upper ramification filtration if @xmath174 for all @xmath175
in particular , if @xmath176 is a separable closure of @xmath58 , then the absolute galois group @xmath177 is equipped with its upper ramification filtration @xmath178 .
the field of norms construction applies to a certain type of field extension , which we now describe .
the basic reference for this material is @xcite .
let @xmath58 be a complete discrete valuation field with perfect residue field @xmath61 of characteristic @xmath3 , and @xmath179 a fixed separable closure .
then an extension @xmath140 contained in @xmath176 is called _ arithmetically profinite ( apf ) _ if for all @xmath180 , the group @xmath181 is open in @xmath182 .
if we set @xmath183 , then this definition means simply that @xmath184 is a finite extension of @xmath58 for all @xmath180 .
since the upper ramification filtration is separated , it follows that @xmath185 , which implies that @xmath186 .
a concrete example of an infinite apf extension is @xmath187 , and in this case @xmath188 .
the apf property allows us to define an ( inverse ) herbrand function : indeed , if @xmath140 is a ( possibly infinite ) apf extension , then we set @xmath189 and define @xmath190 then @xmath159 is increasing , continuous , and piecewise - linear , with inverse @xmath158 which is also increasing , continuous , and piecewise - linear .
of course , when @xmath140 is finite , @xmath158 coincides with our previous definition of the herbrand function .
an important quantity attached to an apf extension @xmath140 is @xmath191 in terms of the ramification subextensions @xmath192 , the quantity @xmath193 is the supremum of the indices @xmath194 such that @xmath195 . in the case
where @xmath140 is galois with group @xmath196 , we have @xmath197 , and @xmath193 is the first jump in the upper ramification filtration on @xmath1 . note that @xmath198 if and only if @xmath140 is totally ramified , and @xmath199 if and only if @xmath140 is totally wildly ramified . because the inverse herbrand function @xmath159 and the quantity @xmath193 will be essential for our later work , we include here the [ apfprops ] let @xmath200 and @xmath201 be two extensions of @xmath58 contained in @xmath179 with @xmath202
. then 1 . if @xmath203 is finite , then @xmath204 is apf if and only if @xmath205 is apf ; 2 .
if @xmath205 is finite , then @xmath204 is apf if and only if @xmath203 is apf ; 3 . if @xmath204 is apf then @xmath203 is apf ; 4 . if @xmath204 is apf , then @xmath206 , and if in addition @xmath203 is finite , then @xmath207
. parts i ) and ii ) of this proposition say that the apf property is insensitive to finite extensions of the top or bottom , while part iii ) says that the apf property is inherited by subextensions .
part iv ) says that the quantity @xmath208 can only increase in subextensions or under a finite extension of the base @xmath203 .
in the latter case , we get a lower bound on the increase of @xmath208 in terms of the inverse herbrand function @xmath209 . given an infinite apf extension @xmath140 ,
let @xmath210 denote the set of finite subextensions of @xmath140 , partially ordered by inclusion .
the key technical fact about the extension @xmath140 is the following property of the quantity @xmath208 , which generalizes part iv ) of the previous proposition : [ apfinfty ] the numbers @xmath211 for @xmath212 tend to @xmath116 with respect to the directed set @xmath210 .
having discussed some general properties of infinite apf extensions , we are now ready to describe the field of norms construction , following @xmath213 : given an infinite apf extension @xmath140 , set @xmath214{}}e^*,\ ] ] the transition maps being given by the norm @xmath215 for @xmath216 . then define @xmath217
thus , a nonzero element @xmath98 of @xmath218 is given by a norm - compatible sequence @xmath219 .
we wish to endow this set with an additive structure in such a way that @xmath218 becomes a field , called the _ field of norms _ of @xmath140 .
this is accomplished by the following if @xmath220 , then for all @xmath212 , the elements @xmath221 converge ( with respect to the directed set @xmath222 ) to an element @xmath223 .
moreover , @xmath224 is an element of @xmath218 . with this definition of addition ,
the set @xmath218 becomes a field , with multiplicative group @xmath225 .
moreover , there is a natural discrete valuation on @xmath218 . indeed ,
if @xmath226 denotes the maximal unramified subextension of @xmath140 ( which is finite over k by apf ) , then @xmath227 does not depend on @xmath228 .
in fact ( @xcite , thorme 2.1.3 ( ii ) ) , @xmath218 is a complete discrete valuation field with residue field isomorphic to @xmath229 ( which is a finite extension of @xmath61 ) .
the isomorphism of residue fields @xmath230 comes about as follows .
for @xmath231 , let @xmath232\in l^0 $ ] denote the teichmller lifting . that is , @xmath233:k_l\rightarrow l^0 $ ] is the unique multiplicative section of the canonical map @xmath234 .
note that @xmath235 is of @xmath0-power degree for all @xmath236 , so @xmath237}\in k_l$ ] for all such @xmath238 , since @xmath229 is perfect .
the element @xmath239}])_{e\in\mathcal{e}_{l|l^1}}$ ] is clearly a coherent system of norms , hence ( by cofinality ) defines an element @xmath240 .
the map @xmath241 is a field embedding which induces the isomorphism @xmath242 mentioned above .
the following result will be used several times in the proof of our main result , theorem [ main ] . before stating it
, we make a for any subfield @xmath243 , define @xmath244 [ approxapf ] let @xmath140 be an infinite apf extension and @xmath245 be any finite extension of @xmath246 contained in @xmath66
. then 1 . for any @xmath247
, there exists @xmath248 such that @xmath249 2 . for any @xmath250
, we have @xmath251 the construction just described , which produces a complete discrete valuation field of characteristic @xmath252 from an infinite apf extension @xmath140 is actually functorial in @xmath66 .
precisely , @xmath253 can be viewed as a functor from the category of infinite apf extensions of @xmath58 contained in @xmath179 ( where the morphisms are @xmath58-embeddings of finite degree ) to the category of complete discretely valued fields of characteristic @xmath0 ( where the morphisms are separable embeddings of finite degree ) . moreover
, this functor preserves galois extensions and galois groups .
fixing an infinite apf extension @xmath140 , the functorial nature of @xmath253 allows us to define a field of norms for _ any _ separable algebraic extension @xmath254 .
namely , given such an @xmath200 , define the directed set @xmath255<\infty\}$ ] , and note that @xmath256{}}l'.\ ] ] then we define the field of norms @xmath257{}}x_k(l').\ ] ] with this definition , we can consider @xmath258 as a functor from the category of separable algebraic extensions of @xmath66 to the category of separable algebraic extensions of @xmath218 .
[ cats ] the field of norms functor @xmath258 is an equivalence of categories . in particular
, @xmath259 is a separable closure of @xmath218 , and we have an isomorphism @xmath260 .
since @xmath218 is a complete discrete valuation field with residue field @xmath229 , it follows that any choice of uniformizer @xmath261 for @xmath218 yields an isomorphism @xmath262 , defined by sending @xmath263 to @xmath264 . via this isomorphism
, an element @xmath265 corresponds to a power series @xmath266 $ ] .
the following lemma describes the relationship between @xmath267 and the coherent system of norms @xmath268 in terms of the chosen uniformizer @xmath261 .
first we need to introduce some notation .
given a power series @xmath269,\ ] ] define for each @xmath236 a new power series @xmath270}}]z^i= \sum_{i=0}^\infty ( f_{l|k}(a_i))_e z^i\in r_e[[z]].\ ] ] [ specialcong ] for all @xmath271 , we have @xmath272 for all @xmath236 . by definition of the isomorphism @xmath273 ,
if @xmath274 , then @xmath275 now by proposition [ approxapf ] , for any @xmath236 we have @xmath276 thus we see that @xmath277 the congruence of the previous lemma can be replaced by an equality if one is willing to restrict attention to lubin - tate extensions of local fields . this is a theorem of coleman ( @xcite , theorem a ) , to which we now turn .
a special class of infinite apf extensions are the lubin - tate extensions , which we now briefly recall ( see @xcite and @xcite chapter v ) .
let @xmath166 be a finite extension of @xmath278 , and @xmath279 a lubin - tate formal group associated to a uniformizer @xmath35 of @xmath166 .
then @xmath279 is a formal @xmath280-module , and interpreting the group @xmath279 in @xmath281 makes this ideal into an @xmath280-module ( here @xmath281 is the maximal ideal of the valuation ring of @xmath282 ) .
let @xmath283 be the @xmath284-torsion of this @xmath280-module .
then @xmath285 for all @xmath117 , and in particular @xmath286 is finite .
now define @xmath287 , which is an infinite totally ramified abelian extension of @xmath166 , the _ lubin - tate extension of @xmath166 associated to @xmath35_. let @xmath58 be a complete unramified extension of @xmath166 with frobenius element @xmath288 , and define @xmath289 . then @xmath140 is an infinite abelian apf extension , with ramification subfields @xmath290 ( @xcite , corollary v.5.6 ) .
moreover , we have @xmath291=q^{m-1}(q-1)$ ] , where @xmath292
. we will refer to extensions of this type as lubin - tate extensions , despite the fact that they are really the compositum of a lubin - tate extension with an unramified extension .
now fix a primitive element @xmath293 for the tate module @xmath294 that is , @xmath295 is a generator for @xmath286 as an @xmath280-module , and if @xmath296_{\gamma}$ ] denotes the endomorphism of @xmath279 corresponding to @xmath35 , then @xmath296_{\gamma}(\omega_{m+1})=\omega_m$ ] for all @xmath297 .
note that the frobenius @xmath298 acts coefficient - wise on the ring @xmath299[\frac{1}{z}]$ ] .
[ co ] for all @xmath300 , there exists a unique @xmath301[\frac{1}{z}]^*$ ] such that for all @xmath297 , @xmath302 coleman s theorem has the following consequence in our situation ( we omit the proof , since we will not make use of this result in what follows ) : [ colemanlifting ] let @xmath140 be a lubin - tate extension as described above , and suppose that @xmath261 is a uniformizer for @xmath218 which is also a primitive element for the tate module @xmath303 ( this is always possible if @xmath304 ) . as described previously , @xmath264 determines an isomorphism @xmath305\rightarrow r_{x_k(l)}$ ] by sending @xmath263 to @xmath264 .
consider a power series @xmath306 $ ] corresponding to @xmath307 under the isomorphism above , where @xmath308 .
then @xmath309 where @xmath310 $ ] is the coleman power series for @xmath98 coming from theorem [ co ] .
hence , a choice of uniformizer ( that is also a primitive element ) defines a lifting of the multiplicative monoid @xmath311}\mathbb{f}_{q^l}[[z]])-\{0\}$ ] to @xmath299^*$ ] . of course
, if @xmath312 is finite then the first monoid above is simply @xmath305-\{0\}$ ] .
note that the multiplicativity of this lifting is guaranteed by the uniqueness in coleman s theorem .
we extend this to a lifting @xmath313}\mathbb{f}_{q^l}[[z]])\rightarrow r_k[[z]]$ ] by setting @xmath314 , and we note that this _ coleman lifting _ provides an alternative to the more obvious teichmller lifting ( coefficient - wise ) of power series .
the use of @xmath10 allows for some simplification in the proof of theorem [ main ] in the case @xmath304 and @xmath264 a primitive element .
however , theorem [ main ] holds for @xmath315 and also for an arbitrary choice of uniformizer @xmath264 , and the proof in the general case given below has recourse to the usual teichmller lifting , @xmath316 . given a totally ramified infinite apf extension @xmath140 ,
we have seen how any choice of a uniformizer @xmath317 determines an isomorphism @xmath318 defined by sending @xmath263 to @xmath264 ( here we set @xmath319 ) .
we would now like to explicitly describe a connection between the field of norms @xmath218 and the open @xmath0-adic disc @xmath68\otimes k)$ ] that will underly the rest of our investigation ( here @xmath320 ) .
namely , the special fiber of the smooth integral model @xmath321)$ ] is @xmath91)$ ] , with generic point @xmath322 . via the isomorphism above coming from the choice of uniformizer @xmath264
, we can thus identify @xmath323 with @xmath324 . on the other hand ,
each component @xmath325 of @xmath264 is a uniformizer in @xmath238 , and in particular has absolute value @xmath326 .
hence , each @xmath325 corresponds to a point @xmath327 with residue field @xmath238 . in terms of the dedekind domain @xmath41\otimes k$ ] , the point @xmath328 corresponds to the maximal ideal @xmath329 generated by the minimal polynomial of @xmath325 over @xmath6 .
thus , the uniformizer @xmath264 defines a net of points @xmath330 which approaches the boundary . in summary
, we have the following picture : @xmath331\otimes k\\ \qquad\qquad \ z \qquad\qquad @ > > > \
\qquad\qquad\pi=(\pi_e)_e \qquad \leftrightsquigarrow \qquad \quad \{x^e\}_e \end{cd}$ ]
let @xmath140 be a lubin - tate extension as described in section [ coleman ] , with residue field @xmath319 .
hence , there exists a @xmath0-adic local field @xmath166 such that @xmath312 is unramified and @xmath332 , where @xmath333 is an honest lubin - tate extension , associated to a formal group @xmath279 . as usual , we let @xmath334 , and we recall that @xmath335=\#(k_h)^{m-1}=q^{m-1}.$ ] choose a uniformizer @xmath317 , which yields the identification @xmath336 as well as the net of points @xmath330 as described in the last section . consider a @xmath1-galois regular branched cover @xmath337 , with @xmath338 normal .
we consider this cover to be a family over @xmath339 , and we introduce the following notations : * @xmath55 denotes the special fiber of the cover , obtained by taking the fibered product with @xmath340 ; * @xmath56 denotes the generic fiber , obtained by taking the fibered product with @xmath341 ; * for each @xmath212 , we denote by @xmath342 the fiber of @xmath343 at @xmath327 ; * if @xmath344 is an affine scheme , then @xmath345 denotes the total ring of fractions of @xmath344 , obtained from the ring of global sections , @xmath346 , by inverting all non - zero - divisors ; * if the special fiber @xmath347 is reduced , then @xmath348 is a product of @xmath349 copies of a field @xmath350 , which is a finite normal extension of @xmath351 ; * since only finitely many of the points @xmath328 are ramified in the cover @xmath56 , for @xmath238 large the fiber @xmath342 is reduced and we have an isomorphism @xmath352 , where @xmath353 is a finite galois extension of fields ; * @xmath354 denotes the degree of the different of @xmath353 ; * @xmath355 denotes the compositum of the fields @xmath66 and @xmath356 in @xmath179 . [ main ] let @xmath337 be a @xmath1-galois regular branched cover of the open @xmath0-adic disc , with @xmath338 normal and @xmath347 reduced
. then 1 .
if @xmath55 is generically separable , then there exists a cofinal set @xmath357 such that for @xmath358 large , we have @xmath359 and @xmath360 as subfields of @xmath361 . moreover , for these @xmath238 , the functor @xmath258 induces an isomorphism @xmath362 which respects the ramification filtrations .
in particular , if @xmath49 is the degree of the different of @xmath363 , then @xmath364 .
if @xmath347 is irreducible , then @xmath55 is generically inseparable if and only if @xmath365 .
3 . if @xmath1 is abelian , then @xmath366 for @xmath238 large , independently of any separability assumption . in particular , @xmath347 is irreducible if @xmath342 is irreducible for @xmath238 large .
if @xmath2 is a finite field , say @xmath367 , then we can take the cofinal set @xmath368 in i ) of the theorem to be @xmath369 .
that is , in the case of a finite residue field , @xmath368 is independent of the particular cover @xmath337 .
the knowledgeable reader will note that much of our proof of i ) is inspired by the proof in @xcite of the essential surjectivity statement in proposition [ cats ] .
the main difficulty is to spread the construction of @xmath213 over the open @xmath0-adic disc .
let @xmath370 , so that @xmath33 $ ] is a @xmath1-galois extension of normal rings ( here @xmath67 ) .
we are assuming that @xmath34 is reduced , where @xmath35 is a uniformizer of @xmath6 .
moreover , we have @xmath371 for @xmath238 large ( here @xmath329 is the maximal ideal of @xmath41\otimes k$ ] corresponding to @xmath328 ) .
i ) : suppose that @xmath55 is generically separable , which means that the field extension @xmath372 is separable , hence galois . by the primitive element theorem
, there exists @xmath373 such that @xmath374 $ ] . moreover , we can choose @xmath375 to be integral over @xmath376 $ ] , say with minimal polynomial @xmath377[t]$ ] .
further , since @xmath38 is infinite , we can choose @xmath349 different primitive elements @xmath378 such that the corresponding minimal polynomials @xmath379[t]$ ] are distinct .
even more , by krasner s lemma , we may assume that each @xmath380[t]$ ] , so that in fact @xmath381 $ ] for some @xmath382 .
having fixed this @xmath383 , we take @xmath384 . setting @xmath385
, the chinese remainder theorem implies that we have an isomorphism @xmath386/(f(t))\cong\prod_{j=1}^{n_s}k((z))[\chi_j]= \prod_{j=1}^{n_s}\mathcal{k}\cong f(y_k)\ ] ] let @xmath387 be the element of @xmath388 corresponding to @xmath389 under this isomorphism , and choose a lifting , @xmath390 , of @xmath387 to @xmath391 .
denote the minimal polynomial of @xmath390 over @xmath41_{(\varpi)}$ ] by @xmath392 , so that @xmath393
. then @xmath392 has the form @xmath394_{(\varpi)}[t].\ ] ] now by the weierstrass preparation theorem , the coefficients of @xmath392 have the form @xmath395 where @xmath69 $ ] and the denominator is a distinguished polynomial .
moreover , because @xmath396[t]$ ] , it follows that each @xmath397 $ ] , which implies that either @xmath398 in @xmath41 $ ] ( in which case @xmath399 ) , or the weierstrass degree of @xmath72 is greater than @xmath400 ( the degree of the denominator ) .
now by the weierstrass argument described in section [ warg ] , for @xmath238 large we can specialize the polynomial @xmath392 at the point @xmath401 to obtain the polynomial @xmath402 $ ] .
for @xmath238 large , the specialized polynomial @xmath403 lies in @xmath404 $ ] , where @xmath405 is the valuation ring of @xmath238 .
this follows immediately from the previous remarks and the ramification argument applied to @xmath406 in the notation of section [ rarg ] . since @xmath377[t]=r_{x_k(l)}[t]$ ]
is separable , we have @xmath407 in @xmath218 . since @xmath408 , we know by proposition [ apfinfty ] that @xmath409 , so there exists @xmath410 such that for @xmath411 we have @xmath412 now @xmath413[t].\ ] ] under our fixed identification of @xmath38 with @xmath218 , each coefficient @xmath414 corresponds to a coherent system of norms @xmath415 .
hence , we can write @xmath416.\ ] ] now let @xmath417 $ ] be the polynomial obtained from @xmath418 by selecting the @xmath238th component from each coefficient : @xmath419.\ ] ] [ disc ] for @xmath358 large , we have @xmath420 .
let @xmath421[t]$ ] be the teichmller lifting of @xmath418 : @xmath422 then @xmath1 and @xmath423 both reduce mod @xmath35 to @xmath5 , hence @xmath424 for some @xmath425_{(\varpi)}[t]$ ] .
specializing at @xmath401 for @xmath358 now yields the equation @xmath426 for some @xmath427 $ ] . indeed , by lemma [ specialcong ] , we have @xmath428 but @xmath429=q^{lt_e}=(q^l)^{t_e}$ ] for @xmath358 .
the operation of raising to the @xmath430th power on @xmath431 is the identity , and since the coefficients of @xmath414 lie in @xmath431 , it follows that @xmath432 .
hence @xmath433 , from which equation ( [ specialeq ] ) follows immediately .
consider @xmath434 now by proposition [ approxapf ] , @xmath435 .
moreover , for @xmath238 large we have @xmath436 indeed , the ramification argument ( section [ rarg ] ) applied to @xmath437 shows that the valuations of the coefficients of @xmath438 go to infinity as @xmath238 gets large , and we know that @xmath439 .
it follows that @xmath440 for @xmath358 large , which implies by @xmath441 that @xmath442 for @xmath358 large .
now let @xmath443 be a root of @xmath403 in @xmath444 , and define @xmath445 .
then @xmath446 is apf by proposition [ apfprops ] , and we set @xmath447 .
moreover , note that for @xmath238 large we have @xmath448 and thus @xmath449 ( recall that @xmath450 ) .
this follows from the fact that there exists @xmath451 $ ] such that @xmath452 ( for example , take @xmath9 to be the product of the denominators of the coefficients @xmath106 of @xmath453_{(\varpi)}[t]$ ] ) .
then the conductor of the subring @xmath454\otimes k)_g[\xi]\subset(\mathcal{a}\otimes k)_g$ ] defines a closed subset of @xmath343 , and if @xmath328 lies outside the image of this set in @xmath84 , then the splitting of @xmath392 mod @xmath329 determines the fiber @xmath342 ( see @xcite , prop .
i.8.3 ) . in particular , @xmath455 is galois over @xmath66 . at this point
we introduce the following lemma from @xmath213 , which it is easy to check holds in our situation , since it is a direct consequence of lemma [ disc ] : [ lindis ] for @xmath456 large , the extensions @xmath457 and @xmath458 are linearly disjoint .
moreover , we have @xmath459 since for @xmath460 large , @xmath457 is totally wildly ramified , it follows from this lemma that @xmath461 so @xmath462 is totally wildly ramified .
hence , proposition [ approxapf ] says that there exists @xmath463 such that @xmath464 our immediate goal is to prove the following lemma about the polynomial @xmath465 $ ] .
[ limit ] @xmath466 .
note that @xmath467 moreover , as mentioned above , @xmath462 is totally ramified , hence @xmath468 denote by @xmath469 $ ] the polynomial obtained by replacing each coefficient of @xmath470\subset x_k(\tilde{l}_e)[t]$ ] by its component in @xmath471 .
then by the linear disjointness of @xmath457 and @xmath458 , it follows that @xmath472 and we have @xmath473 by proposition [ approxapf ] , we have @xmath474 on the other hand , by the ramification argument ( again applied to @xmath437 ) and equation [ specialeq ] @xmath475 where @xmath476 is the maximal degree of the denominators in the coefficients of @xmath451_\mathfrak{m}[t]$ ] .
thus , we see that @xmath477 since @xmath478 . together with the fact that @xmath479 , this implies that @xmath480 thus we have shown that @xmath481 but @xmath482 for @xmath238 large , and since @xmath476 is a constant , we also have @xmath483 .
it follows that @xmath484 so that @xmath466 as claimed . replacing the net @xmath485 by a subnet
, we may assume that it converges to a root @xmath486 of @xmath5 .
but then @xmath486 is conjugate to one of the roots @xmath487 from the beginning of this proof , and since @xmath372 is galois , we have that @xmath488 .
moreover , by krasner s lemma , @xmath489 for @xmath358 large .
this implies that @xmath490 for @xmath358 large , and i claim that this inclusion is actually an equality .
for this we need a simple preliminary lemma .
note that if @xmath491 , then @xmath492 and we have by definition @xmath493 [ galnorm ] given @xmath491 , suppose that @xmath494 is such that @xmath495 using proposition [ approxapf ] , choose an element @xmath496 such that @xmath497 then @xmath498 this follows from a straightforward computation using proposition [ approxapf ] .
we wish to apply this lemma with @xmath499 and @xmath500 , so we compute @xmath501 for @xmath238 large by lemma [ disc ] . since @xmath502 , it follows that @xmath443 satisfies the hypothesis of lemma [ galnorm ] for @xmath238 large , and we conclude that @xmath503 .
this immediately implies that @xmath504 , because if the inclusion were proper , then there would exist @xmath505 in @xmath506 such that @xmath507 , which is a contradiction since @xmath508 .
thus , in order to show that @xmath509 , we just need to show that @xmath510 . but the net @xmath511 converges to @xmath486 , and ( ) shows that the krasner radii @xmath512 for some constant @xmath10 independent of @xmath238 . hence for @xmath238 sufficiently large so that @xmath513 , krasner s lemma tells us that @xmath510 as required .
thus , we have shown that @xmath514 for @xmath515 large .
it now follows from the fundamental equality that @xmath516 : @xmath517}=\frac{\deg{f}}{[l_e : l ] } = \frac{\deg{f}}{[e':e]}=n_e.\ ] ] it remains to prove the statement about the galois groups . by the general theory of the field of norms
, we have @xmath518 moreover , since @xmath519 , and @xmath457 and @xmath353 are linearly disjoint , it follows that @xmath520 thus , we just need to show that the ramification filtrations are preserved under these isomorphisms . first note that for all @xmath521 sufficiently large , we have @xmath522 , since by the preceding proof we have that @xmath523 and @xmath258 is an equivalence of categories . denote this common field by @xmath524 .
[ rambehavior ] for @xmath525 and @xmath238 large , we have @xmath526 .
since the lower ramification filtration is determined by the function @xmath149 , it follows that the isomorphism @xmath527 induced by @xmath258 preserves the ramification filtrations .
since the degree of the different depends only on the ramification filtration , it follows that @xmath364 for @xmath238 large .
this completes the proof of i ) .
+ ii ) : note that by part i ) , if @xmath365 , then the special fiber must be generically inseparable , without any irreducibility hypothesis . now suppose that @xmath347 is irreducible and @xmath55 is generically inseparable .
let @xmath528 be the first ramification group at the unique prime of @xmath40 lying over @xmath529 .
taking @xmath528-invariants , we obtain the tower @xmath530 .
since @xmath528 is a nontrivial @xmath0-group , it has a @xmath0-cyclic quotient .
hence @xmath531 has a @xmath0-cyclic subcover @xmath532 , and we have the tower @xmath533 .
now consider the associated tower of special fibers @xmath534 which corresponds ( by considering the generic points ) to a chain of field extensions @xmath535 . here
the extension @xmath536 is purely inseparable of degree @xmath0 , defined by @xmath537 .
note that there is no extension of constants in this tower because the cover @xmath337 was assumed to be regular .
the extension @xmath538 is separable and totally ramified , so the minimal polynomial of @xmath539 over @xmath38 is eisenstein : @xmath540[t],\ ] ] where @xmath541 is a unit .
it follows that @xmath542 is the minimal polynomial of @xmath543 over @xmath38 .
now let @xmath390 be a lifting of @xmath543 to the localized ring of global sections @xmath544 .
then @xmath390 is integral over @xmath391 and we let @xmath545 $ ] be its minimal polynomial . since @xmath546 and @xmath390 is a lifting of @xmath543 , it follows that the degree of @xmath547 is also @xmath548 , and @xmath549 modulo @xmath35 . using the teichmller lifting @xmath550\rightarrow r[[z]]$ ]
we find that : @xmath551 for some polynomial @xmath552_{(\varpi)}[t]$ ] of degree at most @xmath553 in @xmath554 .
setting @xmath401 , we get the specialized polynomial @xmath555,\ ] ] which for @xmath238 sufficiently large is eisenstein by the ramification argument applied to @xmath556 . letting @xmath557 denote the image of @xmath390 in @xmath558 , it follows by degree considerations that @xmath559 is irreducible for @xmath238 large and @xmath557 is a uniformizer for the field @xmath558 .
we obtain the chain of field extensions @xmath560 , and can compute the different as follows : @xmath561 but @xmath562 and the latter quantity goes to @xmath116 with @xmath238 . by multiplicativity of the different in towers
we conclude that @xmath563 so @xmath564 goes to @xmath116 with @xmath238 as claimed .
+ iii ) : we now assume that @xmath1 is abelian , but make no separability assumption on the special fiber @xmath55 .
since @xmath1 is abelian , the decomposition groups at the @xmath349 primes of @xmath391 lying over @xmath565)$ ] all coincide .
call this decomposition group @xmath566 . taking @xmath566-invariants ,
we observe that @xmath567 is a @xmath568-galois regular branched cover with totally split special fiber : @xmath569 in particular , there is no further splitting in the special fiber @xmath570 .
hence , we can apply part i ) to the cover @xmath567 , and a review of the beginning of the proof of i ) shows that we may take @xmath571 .
we conclude that @xmath572 for @xmath238 large ( here @xmath573 is the number of components of @xmath574 ) .
since @xmath575 is surjective , it follows that @xmath576 as claimed .
this completes the proof of part iii ) , and hence of theorem [ main ] .
using theorem [ main ] , we deduce a local lifting criterion for finite abelian extensions of @xmath577 . for this , consider a lubin - tate extension @xmath140 as described in section [ coleman ] , with @xmath578 for some finite extension @xmath579 . then choose a uniformizer @xmath317 , which defines an isomorphism @xmath580 as well as a net of points @xmath330 ( see section [ disc ] )
. we will be interested in the cofinal sequence of ramification subfields @xmath581 , and will use the simplified notation @xmath582 , etc .
[ arithcrit ] suppose that @xmath1 is a finite abelian group , and let @xmath254 be a @xmath1-galois extension , corresponding to the @xmath1-galois extension @xmath583 via the field of norms functor .
suppose that @xmath337 is a @xmath1-galois regular branched cover with @xmath338 normal and @xmath347 reduced . then @xmath337 is a smooth lifting of @xmath583 if and only if there exists @xmath382 such that for @xmath120 and @xmath584 , we have @xmath585 as @xmath1-galois extensions of @xmath66 , and @xmath586 .
first suppose that @xmath337 is a smooth lifting of @xmath583 .
then @xmath55 is generically separable , so by the proof of part i ) of theorem [ main ] there exists @xmath382 such that for @xmath120 and @xmath584 , we have @xmath587 and @xmath588 . since @xmath253 is an equivalence of categories ,
we conclude that @xmath589 for these values of @xmath117 . moreover , by the local criterion for good reduction ( see introduction ) , we have @xmath53 , which implies that @xmath586 for @xmath120 and @xmath584 , as claimed .
now suppose that there exists @xmath382 so that @xmath585 and @xmath586 for @xmath120 and @xmath584
. then by part iii ) of theorem [ main ] , @xmath347 is irreducible , and then by part ii ) , @xmath55 is generically separable .
hence we may apply part i ) to conclude that there exists @xmath590 such that @xmath591 and @xmath592 for @xmath120 and @xmath593 .
but the two arithmetic progressions @xmath594 and @xmath595 have a common subsequence .
it follows that @xmath596 and @xmath53 , so @xmath337 is a birational lifting of @xmath583 which preserves the different . by the local criterion for good reduction
, it follows that @xmath337 is actually a smooth lifting .
in particular , we obtain an arithmetic reformulation of the ring specific oort conjecture concerning the liftability of cyclic covers over @xmath597 .
set @xmath598 , and let @xmath599 .
then @xmath140 is lubin - tate for @xmath600 and @xmath601 .
moreover , if @xmath10 is a finite cyclic group , define @xmath602\subset r_l$ ] .
then we have the following arithmetic form of the ring specific oort conjecture from the introduction : it can be shown by standard techniques of model theory that the ring specific oort conjecture over @xmath597 implies the ring specific oort conjecture over @xmath2 , where @xmath2 is an arbitrary algebraically closed field of characteristic @xmath0 .
we conclude this paper by giving a direct proof of the arithmetic form of the ring specific oort conjecture for @xmath0-cyclic covers over @xmath597 .
we begin by recalling some facts about @xmath0-cyclic extensions for which we omit the straightforward proofs . recall that @xmath598 , and @xmath599 .
[ minpoly ] suppose that @xmath606 is a uniformizer for @xmath218 , so that @xmath607 .
let @xmath608 be the minimal polynomial of @xmath609 over @xmath610 .
then for any @xmath611 , there exists @xmath612 so that if @xmath613 , the coefficients of @xmath614 $ ] satisfy @xmath615 for @xmath616 .
we will also need a result about the stability of the ramification filtration under base change by subfields of an infinite apf extension @xmath140 .
this result is actually an immediate consequence of lemma [ rambehavior ] , which we used in the proof of theorem [ main ] .
[ stable ] suppose that @xmath617 is a finite @xmath1-galois extension , obtained by base change to @xmath66 from a @xmath1-galois extension @xmath618 .
let @xmath619 be an element of the galois group , and for @xmath620 define @xmath621 .
then there exists @xmath622 such that for @xmath623 we have @xmath624 .
consequently , the canonical isomorphism @xmath625 preserves the ramification filtrations on these groups . in particular , the conductor of @xmath626 is equal to the conductor of @xmath627 for @xmath623 .
[ kummer ] let @xmath254 be a @xmath0-cyclic field extension
. then there exists @xmath628 and a principal unit @xmath629 such that @xmath630 , where @xmath631 .
moreover , @xmath632 may be chosen so that the conductor of @xmath618 is stable under base change by @xmath633 ( lemma [ stable ] ) . modifying @xmath194 by a @xmath0-power in @xmath634 , we may assume that @xmath635 where @xmath636 is a unit in @xmath634 , @xmath637 , and @xmath638 is the artin conductor of @xmath618 let @xmath254 be a @xmath0-cyclic extension , and choose @xmath632 as described in lemma [ kummer ] , so that @xmath639 , where @xmath640 now set @xmath641 and increase @xmath632 if necessary so that @xmath642 .
by proposition [ approxapf ] , there exists a unit @xmath643 such that @xmath644 .
hence we can write @xmath645 where @xmath646 is a unit of index at least @xmath201 .
now @xmath647 , so by hensel s lemma @xmath648 has a @xmath649th root in @xmath218 .
then replace the uniformizer @xmath606 by @xmath650 . again calling this uniformizer @xmath264
, we find that @xmath651 where @xmath652 is a unit of index at least @xmath201 , say @xmath653 where the @xmath654 are teichmller representatives .
now set @xmath655^\times$ ] and consider the extension of normal rings @xmath656 $ ] defined generically by the kummer equation @xmath657 by @xcite , proposition 1.4 , the special fiber of this extension , @xmath658 , is reduced .
hence , we can apply theorem [ main ] to the @xmath0-cyclic cover @xmath659 . setting @xmath660 yields the integral equation @xmath661 the right hand side has @xmath638 zeros , yielding @xmath638 ramified points , each with ramification index @xmath0 .
it follows that the degree of the generic different is @xmath662 . specializing the equation at @xmath663 yields @xmath664 and
so all of the specializations @xmath665 are field extensions with conductor @xmath638 , hence degree of different @xmath666 .
theorem [ main ] says the special fiber is separable and irreducible , and @xmath51 .
hence , the lifting is smooth , and it remains to verify that @xmath667 for @xmath120 .
but at level @xmath632 we have @xmath668 , hence @xmath669 . for @xmath670 ,
the extension @xmath665 is defined by adjoining a @xmath0th root of @xmath671 , so by kummer theory it suffices to show that this element differs from @xmath672 by a @xmath0th power in @xmath66 .
the following lemma thus completes the proof of the @xmath0-cyclic case of the arithmetic form of the ring specific oort conjecture over @xmath597 .
in fact , i claim that @xmath675 is a @xmath0th power in @xmath676 . for simplicity
, we will assume that @xmath304 in the computation that follows ( the case @xmath315 is similiar ) .
moreover , since any principal unit in @xmath676 of index greater than @xmath677 is a @xmath0th power in @xmath676 , we will work in the quotient ring @xmath678 to simplify the computation .
by lemma [ minpoly ] we have @xmath679 and we may assume @xmath680 for each @xmath149 .
then since @xmath681 has index at least @xmath682 in @xmath610 , we have the following congruence in @xmath683 : @xmath684 multiplying by a @xmath0th power and remembering that @xmath685 yields @xmath686 finally , we compute @xmath687 indeed , @xmath688 for some @xmath689 , and so @xmath690 hence , the quantity @xmath691 is a @xmath0th power in @xmath676 as required . | motivated by the local lifting problem for galois covers of curves , this paper investigates galois branched covers of the open @xmath0-adic disc . using the field of norms functor of fontaine and wintenberger ,
we show that the special fiber of a galois cover is determined by arithmetic and geometric properties of the generic fiber and its characteristic zero specializations . as applications ,
we derive a characteristic zero reformulation of the abelian local lifting problem , and give a new proof of the liftability of @xmath0-cyclic covers . |
in particle physics , the neutrino plays a remarkable role of a `` tiny '' particle .
indeed , the scale of neutrino mass @xmath0 is much lower than that of the charged fermions ( @xmath1 , @xmath2 ) .
interaction of neutrinos with matter is extremely weak as compared to that in the case of other known elementary fermions , and it can be mediated via the weak or electromagnetic channel . in this context ,
neutrino electromagnetic properties are of particular interest , for they open a door to `` new physics '' beyond the standard model ( sm ) @xcite . in spite of appreciable efforts in searches for electromagnetic properties of neutrino , up to
now there is no experimental evidence favoring their nonvanishing electromagnetic characteristics .
however , the recent development of our knowledge of neutrino mixing and oscillations , supported by the discovery of flavor conversions of neutrinos from different sources , makes quite plausible the assumption that neutrinos have `` nonzero '' electromagnetic properties .
the latter include , in particular , the electric charge , the charge radius , the anapole moment , and the dipole electric and magnetic moments .
the neutrino magnetic moments ( nmm ) expected in the sm are very small and proportional to the neutrino masses are used throughout unless otherwise stated . ]
@xcite : @xmath3 with @xmath4 being the electron bohr magneton , and @xmath5 is the electron mass . thus any larger value of @xmath6 can arise only from physics beyond the sm ( a recent review of this subject can be found in @xcite ) .
current direct experimental searches @xcite for a magnetic moment of the electron ( anti)neutrinos from reactors have lowered the upper limit on @xmath6 down to @xmath7 @xcite .
these ultra low background experiments use germanium crystal detectors exposed to the neutrino flux from a reactor and measure the energy @xmath8 deposited by the neutrino scattering in the detector .
the sensitivity of such a search to nmm crucially depends on lowering the threshold for the energy transfer @xmath8 , due to the enhancement of the magnetic scattering relative to the standard electroweak one at low @xmath8 .
the paper is organized as follows . in section [ el - m_prop ]
, we discuss electromagnetic characteristics that one may expect in the cases of dirac and majorana neutrinos .
specific theoretical aspects of searches for nmm are considered in section [ moment ] .
section [ conclusions ] summarizes this work .
in general the matrix element of the electromagnetic current @xmath9 can be considered between different neutrino initial @xmath10 and final @xmath11 states of different masses , @xmath12 and @xmath13 : @xmath14 in the most general case consistent with lorentz and electromagnetic gauge invariance , the vertex function is defined as ( see ref .
@xcite and references therein ) @xmath15(q^2\gamma_\mu-\gamma_\mu\!\!\not{q})+f_m(q^2)_{ij}i\sigma_{\mu\nu}q^\nu+f_e(q^2)_{ij}i\sigma_{\mu\nu}q^\nu\gamma_5,\ ] ] where @xmath16 , @xmath17 , @xmath18 , and @xmath19 are respectively the charge , anapole , dipole magnetic , and dipole electric neutrino form factors , which are matrices in the space of neutrino mass eigenstates @xcite .
let us briefly discuss the diagonal case @xmath20 .
the hermiticity of the electromagnetic current and the assumption of its invariance under discrete symmetries transformations put certain constraints on the neutrino form factors , which are in general different for the dirac and majorana cases . in the case of dirac neutrinos
, the assumption of cp invariance combined with the hermiticity of the electromagnetic current @xmath9 implies that the electric dipole form factor vanishes , @xmath21 . at zero momentum transfer only @xmath22 and @xmath23 , which are called the electric charge and the magnetic moment , respectively ,
contribute to the hamiltonian @xmath24 , which describes the neutrino interaction with the external electromagnetic field @xmath25 .
hermiticity also implies that @xmath26 , @xmath27 , and @xmath28 are real .
in contrast , in the case of majorana neutrinos , regardless of whether cp invariance is violated or not , the charge , dipole magnetic and electric moments vanish , @xmath29 , so that only the anapole moment can be non - vanishing among the electromagnetic moments .
note that it is possible to prove @xcite that the existence of a non - vanishing magnetic moment for a majorana neutrino would bring about a clear evidence for cpt violation . in the off - diagonal case
@xmath30 the hermiticity by itself does not imply restrictions on the form factors of dirac neutrinos .
it is possible to show @xcite that if the assumption of cp invariance is added , the form factors @xmath26 , @xmath28 , @xmath31 , and @xmath27 should have the same complex phase . for the majorana neutrino ,
if cp invariance holds , there could be either a transition magnetic or a transition electric moment . finally , as in the diagonal case , the anapole form factor of a majorana neutrino can be nonzero .
the neutrino dipole magnetic and electric form factors ( and the corresponding magnetic and electric dipole moments ) are theoretically the most well studied among the form factors .
they also attract a notable attention from experimentalists , although the nmm value ( [ mu_nu ] ) predicted in the sm is many orders of magnitude smaller than the present experimental limits achievable in terrestrial experiments .
the most sensitive and established method for the experimental investigation of the nmm is provided by direct laboratory measurements of electron ( anti)neutrino - electron scattering at low energies in solar , accelerator , and reactor experiments .
a detailed description of various experiments can be found in @xcite .
the cross section for electron ( anti)neutrino scattering on a free electron can be written @xcite ( see also @xcite ) as a sum of the sm and nmm contributions , @xmath32 where @xmath33 is the incident neutrino energy and @xmath8 is the energy transfer .
the sm contribution is constant in @xmath8 at @xmath34 : @xmath35 \approx 10^{-47 } { \rm cm^2/kev}. \label{sm}\ ] ] in contrast , the nmm contribution @xmath36 exhibits a @xmath37 enhancement at low energy transfer .
note that the nmm contribution to the cross section changes the helicity of the neutrino , contrary to the sm contribution and to the possible contribution from the neutrino charge radius .
therefore , for relativistic neutrino energies the interference between @xmath38 and @xmath39 is a negligible effect in the total cross section ( [ cr_sec ] ) .
the current experiments with reactor ( anti)neutrinos have reached threshold values of @xmath8 as low as few kev and are likely to further improve the sensitivity to low energy deposition in the detector . at low energies however one can expect a modification of the free - electron formulas ( [ sm ] ) and ( [ fe ] ) due to the binding of electrons in the germanium atoms , where e.g. the energy of the @xmath40 line , 9.89kev , indicates that at least some of the atomic binding energies are comparable to the already relevant to the experiment values of @xmath8 . in the case
@xmath41 , which is relevant to the experiments with reactor ( anti)neutrinos , it can be shown @xcite that the sm and nmm contributions to the neutrino scattering on atomic electrons are @xmath42 @xmath43 where @xmath44 and @xmath45 are the free - electron results given by ( [ sm ] ) and ( [ fe ] ) , respectively . a key quantity that determines cross sections ( [ sm_bind ] ) and ( [ fe_bind ] ) is the so - called dynamical structure factor @xmath46 , which is a function of the energy and momentum transfer values , @xmath8 and @xmath47 . for a free electron ,
one has in a nonrelativistic limit @xmath48 , which upon substitution in ( [ sm_bind ] ) and ( [ fe_bind ] ) immediately yields the free - electron formulas ( [ sm ] ) and ( [ fe ] ) .
recently it was claimed @xcite that the atomic binding effects must result in a significant enhancement of the nmm contribution .
however , that early claim was later disproved @xcite , thus also disproving the upper bound on the @xmath6 value , @xmath49 , obtained in @xcite .
it was demonstrated @xcite by means of analytical and numerical calculations that the atomic binding effects are adequately described by the so - called stepping approximation introduced in @xcite from interpretation of numerical data . according to the stepping approach ,
the sm and nmm contributions are simply given by @xmath50 where the @xmath51 sum runs over all occupied atomic sublevels , with @xmath52 and @xmath53 being their occupations and binding energies .
the following important conclusions can be drawn from the stepping approximation ( [ step ] ) .
firstly , the atomic effects reduce the sm and nmm contributions compared to their free - electron values .
secondly , the ratio between the sm and nmm contributions is not affected by the atomic binding effects .
the above theoretical findings strongly support the upper limit @xmath54 recently reported by the gemma collaboration @xcite .
this bound obtained in terrestrial experiments with reactor ( anti)neutrinos is only by an order of magnitude weaker than the most stringent astrophysical constraint @xmath55 @xcite .
a general and model - independent upper bound on the dirac nmm , that can be generated by an effective theory beyond the sm , is @xmath56 @xcite ( the limit in the majorana case is much weaker ) .
thus , the searches for nmm are close to the territory where new physics can reveal itself
. one of the authors ( a.i.s . ) is thankful to lothar oberauer , georg raffelt and robert wagner for the kind invitation to participate in the 12th international conference on topics in astroparticle and underground physics .
the work was supported by rfbr grant no .
11 - 02 - 01509-a .
99 studenikin a i 2009 _ nucl .
b ( proc . suppl . ) _ * 188 * 220 fujikawa k and shrock r 1980 _ phys
lett . _ * 45 * 963 wong h t et al 2007 _ phys . rev
. d _ * 75 * 012001 ( _ preprint _ hep - ex/0605006 ) . beda a g et al 2007 _ phys .
nucl . _ * 70 * 1873 ( _ preprint _ hep - ex/0705.4576 ) beda a g et al 2010 _ phys . part .
lett . _ * 7 * 406 ( _ preprint _ hep - ex/0906.1926 ) beda a g et al 2010 _ preprint _ hep - ex/1005.2736 shrock r e 1982 _ nucl
b _ * 206 * 359 kayser b 1982 _ phys
* 26 * 1662 kayser b 1984 _ phys .
d _ * 30 * 1023 nieves j f 1982 _ phys .
* 26 * 3152 wong h t and li h b 2005 _ mod .
* 1103 vogel p and engel j 1989 _ phys rev d _ * 39 * 3378 voloshin m b 2010 _ phys . rev .
lett . _ * 105 * 201801 ( _ preprint _ hep - ph/1008.2171 ) kouzakov k a , studenikin a i and voloshin m b 2011 _ phys .
rev d _ * 83 * 113001 ( _ preprint _ hep - ph/1101.4878 ) kouzakov k a , studenikin a i and voloshin m b 2011 _ jetp lett . _ * 93 * 623 ( _ preprint _ hep - ph/1105.5543 ) wong h t , li h b and lin s t 2010 _ phys
* 105 * 061801 ( _ preprint _ hep - ph/1001.2074 ) kouzakov k a and studenikin a i 2011 _ phys .
b _ * 696 * 252 ( _ preprint _ hep - ph/1101.4878 ) kouzakov k a and studenikin a i 2011 _ nucl .
b ( proc . suppl . )
_ * 217 * 353 ( _ preprint _ hep - ph/1108.2872 ) kopeikin v i , mikaelyan l a , sinev v v and fayans s a 1997 _ phys . at .
nucl . _ * 60 * 1859 raffelt g 1990 _ phys .
lett . _ * 64 * 2856 bell n et al 2005 _ phys .
lett . _ * 95 * 151802 | we give a brief outline of possible neutrino electromagnetic characteristics , which can indicate new physics beyond the standard model .
special emphasis is put on recent theoretical development in searches for neutrino magnetic moments . |
the recognition that supernovae , massive star winds , and other impulsive energy inputs are important heat sources for the interstellar medium ( ism ) in galaxies , and that they generate the warm and hot phases , was a turning point for the theory of the ism in galaxies ( e.g. , mckee & ostriker 1977 ) .
studies of giant expanding gas shells in our galaxy ( e.g. , heiles 1984 ) , and other nearby disk galaxies showed that the energies required to produce these structures are much greater than those produced by a single supernova or the winds of its stellar progenitor ( see reviews of brinks 1990 , van der hulst 1996 ) .
however , young clusters of hot stars and their multiple supernova do produce sufficient energy to make these supershells ( e.g. , mac low , mccray & norman 1989 , norman & ikeuchi 1989 , tenorio - tagle , rozyczka , & bodenheimer 1990 , and tomisaka 1992 ) . with this realization the theory of superbubbles , breakout , chimneys and large - scale outflows developed quickly , and joined the older theory of galactic fountains ( see shapiro & field 1976 , bregman 1980 , cox 1981 , shapiro & benjamin 1991 , and schulman , bregman , & roberts 1994 ) in contributing to our understanding of turbulent ism heating and the disk - halo connection . at about the same time , observational discoveries on the nature of several components of the ism led to a improved understanding of its overall phase structure ( see the reviews in the book of van der hulst 1997 ) .
one important component is the warm neutral medium , wnm , or the lockman layer in our galaxy .
this component consists mostly of small hi clouds , and diffuse ( cirrus ) material distributed in a substantially thicker disk than the cold component ( e.g. , dickey & lockman 1990 , malhotra 1995 , and haynes & broeils 1997 ) .
a second component is the warm ionized medium ( wim ) , or diffuse ionized medium ( dim ) , or the reynolds layer in our galaxy , which is observed in h@xmath1 and other emission lines ( e.g. , reynolds 1996 , rand 1997a ) .
these two media are continuous and overlapping , though the wim extends to greater scale heights above the disk than the wnm .
both components have been studied in detail in our galaxy ( see the review of dickey & lockman 1990 , and the recent hi study of malhotra 1995 ) , and are probably common constituents of late - type galaxies ( see van der hulst 1997 ) .
the line emission of the wim makes it the easier component to study in other galaxies ; it is observed in : 1 ) dwarf irregular galaxies ( e.g. , hunter & gallagher 1997 ) , 2 ) edge - on disk galaxies , especially the vigorously star - forming object ngc 891 ( howk & savage 1997 , rand 1996 , 1997a , b , 1998 , swaters , sancisi & van der hulst 1997 ) , 3 ) large - scale outflows from nuclear starburst galaxies ( e.g. , lehnert & heckman 1996 ) , and 4 ) nearby disk galaxies at arbitrary inclinations ( wang , heckman , & lehnert 1997 ) .
these studies also confirm the association of the various warm - to - hot components ( henceforth , collectively whm ) with star formation regions , superbubbles , and large - scale outflows .
this association , in turn , suggests connections with the still hotter ism components observed in x - rays and radio continuum .
the question of how the extensive whm is heated ( and how the wim is ionized ) in galaxies that are not experiencing extensive starbursts is not entirely answered .
strong impulsive energy sources are quite localized , while the whm is not , in either our galaxy or others ( rand 1996 , wang , et al .
uv photofluxes can transfer energy over long distances , and cosmic rays and magnetic fields can also contribute to the pressure support .
however , a variety of evidence indicates that other sources are needed to heat the whm over most scales .
the evidence includes the following : 1 ) fountain and superbubble models , with a good deal of mechanical energy injection , are very successful in accounting for the characteristics of the hot halo gas ( shapiro & field 1976 , bregman 1980 , li & ikeuchi 1992 , mckee 1993 ) , 2 ) an extra source of support beyond that associated with random motions of typical clouds is needed to support the galactic hi ( wnm ) layer ( malhotra 1995 ) , 3 ) the `` disturbed '' wim in external galaxies seems to require additional energy source ( rand 1997a , b , 1998 , wang et al .
1997 ) , 4 ) observations of faraday rotation in our galaxy ( see summary in minter & spangler 1996 ) demonstrate the existence of turbulence on small scales in the diffuse ionized gas , and models suggest that turbulent heating is important on those scales ( minter & spangler 1997 , minter & balser 1997 ) .
turbulence , generated by the impulsive sources and propagated by ( magneto)acoustic waves and mass flows may be the missing ingredient on the intermediate scales , as well as the large and small scales .
the idea that interstellar cloud structure is turbulent is decades old ( see larson 1979 , scalo 1987 and references therein ) .
hydrodynamic models of local regions of the ism with heating and cooling sources included clearly illustrate the development of turbulence ( e.g. , rosen & bregman 1995 , passot , vazquez - semadeni , & pouquet 1995 ) .
chappell & scalo ( 1997 ) argue that clouds themselves are multifractal manifestations of the interstellar turbulence .
elmegreen ( 1997 ) agrees that interstellar clouds have a fractal structure , and further proposes that the `` holes and gaps '' that make up the intercloud medium are the result of turbulent heating rather than `` clearing '' by supernova explosions .
norman and ferrara ( 1996 ) calculate that the turbulent energy injection into the interstellar gas is characterized by a very broad band spectrum , and that the general turbulent pressure may exceed the thermal pressure by 1 - 2 orders of magnitude .
thus , there is increasing evidence that _ turbulence supplies a large part of the heating needed to maintain the continuous range of phases in the ism , and much of the pressure support .
_ this principle is the central assumption on which the models described below are based .
heating and cooling processes in the interstellar gas generally have characteristic length scales of less than a kiloparsec or so .
these local thermal processes co - exist with global regularities in the structure of gas - rich disks , like the nearly universal surface density profiles of the cold gas and stellar components .
the profile forms of these components are often described as a negative exponential functions of radius in the disk , though the gas surface density is also well described as having a @xmath2 form ( e.g. , struck - marcell 1991 ) .
kennicutt s ( 1989 , 1990 , 1998a ) influential studies of star formation ( henceforth sf ) in galaxy disks showed that the neutral gas surface density varies with radius in such a way that it is always nearly equal to the radially dependent threshold density for gravitational instability .
these results revived earlier suggestions that local gravitational instabilities are needed to assemble the massive clouds where star clusters are formed .
the threshold surface density also varies as @xmath2 if the circular velocity is a constant , independent of radius , as observed in many disks .
it is widely believed that the disk gas surface density profile is a result of initial conditions , and the disk formation process .
the reader is referred to the papers of steinmetz & m@xmath3ller ( 1995 ) , dalcanton , spergel , & summers ( 1997 ) , and mo , mao , & white ( 1998 ) for recent discussions of disk formation and early evolution . however , as noted above , thermal and turbulent pressure forces , as well as gravity and centrifugal force , are important in the whm , and this makes it less likely that initial profiles are `` frozen out '' in these media .
the effects of turbulent pressure on the large - scale structure of the whm , have not yet received much attention . in the context of the secular evolution of the cold gas ,
struck - marcell ( 1991 ) pointed out that a @xmath2 surface density profile was required to maintain the conditions of hydrostatic equilibrium in the disk with minimal transport by shear viscosity ( see below ) .
the same arguments apply to the whm , and more strongly , because of its shorter sound - crossing timescale .
this leads us to the hypothesis that _ in any disk where the heating by sf activity is sufficient to maintain and cycle a large fraction of the gas through the whm , turbulent hydrodynamic forces will regulate the gas surface density to the @xmath2 profile . _
furthermore , the recent work of martin ( 1998 and references therein ) suggests that substantial cycling rates are quite plausible .
this hypothesis provides a second pillar on which the models below are built .
dopita ( 1985 ) described one of the first models of global self - regulation by star formation .
his model was based on the assumption of equipartition between turbulent and thermal pressures . in the models derived below
we will consider cases where the turbulent pressure exceeds the thermal pressure in the whm phases , as suggested by norman and ferrara .
this is analogous to the situation in the cold cloud ensemble , where cloud random velocities are supersonic .
we will , however , assume pressure balance between different thermal phases . many mechanisms of self - regulation based on the intrinsic stability properties of local star - cloud systems in disks have been suggested .
in particular , models with cloud buildup by collisional agglomeration and disruption by sf activity have been popular , since they often yield ( schmidt - type ) sfrs with power - law dependences on local gas density ( e.g. , scalo & struck - marcell 1984 , struck - marcell & scalo 1987 , dopita 1990 , dopita & ryder 1994 ) .
paravano has extensively investigated how star formation can be regulated by the thermal conversion processes that operate between warm diffuse phases and the small cool clouds ( 1988 , 1989 , also see franco & shore 1984 , diaz - miller , franco , & shore 1998 , and bertoldi & mckee 1997 for related models ) . with reasonable approximations
he finds that these processes also produce a schmidt law sfr .
silk ( 1997 , and references therein ) has argued for a somewhat different star formation law , that includes a dependence on local shear , and has described self - regulation and a derivation of the tully - fisher relation in a model based on this law .
kennicutt ( 1998a , b ) finds that the global properties of star formation in galaxies are consistent with both the schmidt and silk / wyse phenomenological laws .
there are several difficulties with many of these approaches to self - regulation .
the first is that most require some arbitrary phenomenological assumptions , i.e. , about sf or cloud collision rates , equipartition , or constraints on the ambient pressure .
the second is that most are local in the sense that they do not include the hydrodynamic couplings to adjacent regions . on the other hand ,
the so - called chemodynamical models are based on a large - scale hydrodynamical treatment .
for example , samland , hensler , & theis ( 1997 and references therein ) have recently presented two - dimensional hydrodynamical models with three stellar components and two discrete gaseous phases .
inevitably , there are many uncertain parametrizations in the couplings between the components , which limits the predictive power of the models .
moreover , they do not include the effects of turbulent stresses in the intercloud medium , which we feel are essential to an understanding of the large - scale structure .
the primary role of these stresses distinguish the models below from most previous ones .
a good conceptual model of gas dynamics and star formation in galaxy disks requires a quasi - static solution of multiphase hydrodynamic equations that include the important thermohydrodynamic forces of self - regulation .
ideally , the model should be simple enough to allow a clear understanding of both the structure of individual disks and the universal relations between disks . as a step towards this goal
we here introduce a model of a two - component ism described by cylindrically symmetric hydrodynamic equations , with turbulent stress terms .
the two components are a cold isothermal ( cloud ) phase and a mean whm described by a locally adiabatic equation of state ( see below ) .
this two - phase model is a minimal description of the multiphase interstellar gas , but it is sufficiently complex to capture many interesting behaviors , as we will see below . we will describe a simple analytic version of the model here , but the model is readily generalizable to a continuous range of phases , albeit with much increased complexity in the equations .
we plan to develop that generalization in a later paper . in this paper
we do not consider the effects of magnetic forces or cosmic rays separately from the thermal or turbulent pressures .
as part of the definition of a `` quasi - static '' disk , we will assume that the mass exchanges between phases balance locally .
then the time - independent mass continuity equations are of the form , @xmath4 where @xmath5 is the phase component surface density and @xmath6 is the component radial velocity , and with i = c , w for the two phases . for convenience , we will frequently use the approximation that @xmath7 , where @xmath8 is the component mass density , and @xmath9 is the component scale height in the direction perpendicular to the disk .
we assume hydrostatic equilibrium in the vertical direction ( i.e. , perpendicular to the disk ) . when the disk self - gravity dominates the vertical potential gradient , then the solutions to the component hydrostatic equations give component scale heights that increase slowly with radius ( as @xmath10 , see malhotra 1995 for a discussion of the application of this approximation to the milky way ) .
the zero radial flow solution , @xmath11 , to equation ( 1 ) is often assumed .
however , if @xmath12 , then there is a more general family of solutions in which @xmath13 .
especially interesting are the solutions with @xmath14 , describing opposed radial flows in the two components , but with zero net radial mass transport .
this allows the slow inflow of high density cold ( or cooling ) gas to replace gas heated into the warm phases by sf , which are on average flowing outward .
the flow of individual mass elements might consist of little more than a circulation within a local fountain ( like a transient convective cell ) , with the ensemble of fountains making up the global inflow / outflow solution .
figure 1 provides a schematic view . because the scale - height ratio @xmath15 is small , inflow dominates in the midplane of the disk , sandwiched between warm outflows above and below .
the existence of two discrete radial velocities is a consequence of the assumption of two distinct phases , and the model can be generalized to a continuous range of phases as a function of distance from the midplane with continuously stratified radial velocities .
the fact that the @xmath2 profile is a good approximation to the observed cold gas distributions in large parts of many late - type galaxy disks ( struck - marcell 1991 ) provides empirical support for the constant radial velocity inflow@xmath0outflow solutions as opposed to more complicated radius - dependent flows . but
why should this profile be universal ?
several decades of numerical studies suggest that when the cold gas density significantly exceeds the gravitational instability threshold in all or part of a disk , then the result is the rapid development of instabilities that generate readily observable clumps .
( an example of the clumping instability in disk formation is presented in noguchi 1998 . )
the massive clumps would generate strong sf and heating , which would regulate the cold gas to lower densities by the processes described above .
thus , we do not expect to find densities much in excess of the threshold , which scales as @xmath2 in a flat rotation curve disk .
hydrodynamical stability arguments also lead to a preference for the @xmath2 surface density profile , as described by struck - marcell ( 1991 ) .
for example , if the mean random velocity of the cold cloud ensemble is a constant independent of radius ( i.e. , the ensemble is isothermal ) , then only the @xmath2 profile yields a constant net pressure force between adjacent annuli in the disk .
any deviation from that profile generates a pressure gradient proportional to the deviation , which would allow the nonlinear amplification of disturbances .
this is not consistent with a hydrodynamic steady state .
as noted above , this argument is even stronger when applied to the whm .
a qualitative , but very general argument , is based on the observation that there is no obvious characteristic scale ( other than the scale of the rising rotation curve in the center ) over the range extending from a few kiloparsec to tens of kiloparsecs in late - type disks .
this suggests that the important forces within the disk all have essentially the same radial or distance scaling .
if not , we would expect there to be observational signatures associated with changes in the dominant force . to make this point more definite ,
we write the radial momentum equation for the cold gas component as follows , @xmath16 where @xmath17 , @xmath18 are a gravitational scale mass and radius , respectively , and @xmath19 is the azimuthal velocity in the cold component .
the terms on the right - hand - side represent the pressure gradient , gravitational , the centrifugal accelerations . in equation
( 2 ) the gravitational and centrifugal accelerations both have a @xmath2 scaling .
there is no power - law solution for @xmath20 that yields the desired scaling for the advection term .
however , when @xmath20 is constant this term vanishes .
moreover , the isothermal equation of state also yields the correct scaling in the pressure gradient acceleration for any power - law density profile , though we generally expect this term to be negligible in the cold gas .
thus , the adopted density and velocity profiles make up the only self - similar family of steady solutions to equations ( 1 ) and ( 2 ) .
the remaining momentum equations do not require any changes in the density and velocity scalings .
the azimuthal momentum equations are described in the next section , the radial momentum equation for the whm is , @xmath21 this equation is very similar to the previous one , and in both cases the left - hand - side is zero for the constant radial velocity models .
one difference is that we expect the pressure gradient term , which contains both thermal and turbulent pressures , to be significant in this component , in contrast to the cold component .
the additional last term derives from an effective turbulent shear viscosity in the azimuthal direction .
this term must be included because the change in specific angular momentum due to `` viscous '' forces changes the effective centrifugal force from what we would expect in the absence of `` viscosity '' .
thus , in the warm gas , the non - negligible pressure gradient , and the steady input of angular momentum allow gravity to be balanced with less centrifugal force , while maintaining a constant velocity outflow . there is a corresponding rate of decrease of angular momentum in the cool gas , but assuming that the total mass and angular momentum of the cold gas are greater than the whm , we have assumed that this term is negligible in equation ( 2 ) .
we will discuss the viscous shear terms further in the next section , but we note here that the term in equation ( 3 ) contains the quantity , @xmath22 , which we define as @xmath23 , and @xmath24 is the turbulent velocity dispersion in the whm . in this simple model
we will assume that the viscous coefficients @xmath25 in the two phases , the nearly circular azimuthal velocities @xmath26 and the velocity dispersions are all constant . in the case of the velocities
this assumption is in accord with observation , see van der hulst ( 1997 ) .
the pressure gradient is due to the random motions of the cold gas elements , and not cloud internal pressures .
many processes may contribute to the viscosity in the interstellar gas , including : viscous shear within the cloud ensemble , drag against the diffuse components , enhancements of these by bars and spiral waves , turbulent and magnetohydrodynamic couplings , etc .
at the same time , turbulent pressure and turbulent angular momentum transfers to the whm can provide support against gravity , as well as driving the radial flow . as mentioned above
, the support against gravity implies that the circular velocity of this medium can be less than that of the cold gas .
if the two ( or multiple ) phases have substantially different rotation speeds , @xmath22 , then we expect turbulent shear viscosity between vertical layers to be the dominant viscous term .
cold gas heated by sf will be mixed into the whm via chimneys , fountains , and bubble shells , and will add angular momentum to the whm . at the same time , cooling lumps or filaments of the whm , with lower specific angular momentum than the cold gas , will rain onto the midplane , decreasing the angular momentum of the cool clouds that sweep them up .
the net result will be somewhat like the friction between two thin disks forced together at different speeds . with the assumption that this is the only significant viscous term we can write the steady azimuthal momentum equation as , @xmath27 where i = c , w as usual , and this equation is valid for both components , though the radial velocities have different signs . the right hand term can be understood as dimensionally similar to a navier - stokes kinematic viscosity term of the form - @xmath28 , with a viscosity coefficient of order @xmath29 .
in this case , the `` mean free path '' @xmath30 is of order the size of a typical chimney , fountain , or the turbulent zone around an sf region , which we assume to be roughly constant across the disk . the vertical second derivative we approximate as of order @xmath31 , where as noted above , the whm scale height @xmath32 .
the coefficient @xmath33 is assumed to be of order unity .
although the viscous term in the above equation may look somewhat unusual , in fact , it is a variant of the usual @xmath1-viscosity for supersonic turbulence , @xmath34 , with scales appropriate to this problem . as noted above
, we can understand the last term in equation ( 3 ) as a momentum source term equivalent to a reduction in the centrifugal acceleration due to azimuthal viscosity .
we assume that the radial velocities are small , and so , radial viscous accelerations are negligible .
the centrifugal reduction can be treated as an acceleration , which we estimate as , @xmath35 where the first equality is derived on the assumption that this acceleration must have a magnitude sufficient to reduce the azimuthal velocity to zero in a time @xmath36 . the second equality results from substituting the previous equation .
it is an additional simplifying assumption that the coefficient on the right - hand - side of this equation equals that of the previous equation , though we expect them to be of the same order .
this term is usually neglected in viscous disk studies because @xmath6 is small , but it is important here .
we will also see later that @xmath37 , so @xmath38 , which justifies neglecting the term in equation ( 2 ) . because of the low densities and high temperatures that characterize most of the whm , we generally expect radiative cooling timescales to be long , so in our simple model an adiabatic equation of state is more appropriate than an isothermal one .
this raises a potential problem with the scaling of the pressure gradient acceleration .
however , shock heating plays an important role in determining the temperature structure of the whm , and the frequency and intensity of the shocks varies with radius . since shocks generate entropy , this suggests that the mean specific entropy of the whm varies with radius , and so , the adiabat of the warm gas must also vary with radius . in this simple model
we write the equation of state for the warm gas as , @xmath39 where @xmath40 is the thermal pressure , @xmath41 , and the adiabatic constant k varies with radius .
( in reality , it probably also varies with vertical height z. ) in particular , we obtain the desired self - similar form for the pressure gradient term with a radially dependent adiabatic constant of the form , @xmath42 . if the warm gas is flowing outward , this increase in specific entropy would be the result of the cumulative effects of shocks ( see appendix a ) .
it is interesting that _ while the warm phase is assumed to be locally adiabatic in this simple model , the variation of pressure ( and observables that depend on thermal temperature ) with radius is the same as a globally isothermal gas . _
( note that in terms of surface density and surface pressure we have @xmath43 , with @xmath44 , and @xmath45 , and again , @xmath46 . ) at this point , the structure of a minimal model is nearly completely defined by the empirically constrained ( @xmath2 ) surface density profiles , the adopted equations of state for thermal pressures , the continuity equations , the radial and azimuthal momentum equations , and the condition of pressure balance between phases . using the ideal gas law to relate ( total ) pressure @xmath47 to the temperature @xmath48
, we have the following equation for the inter - phase pressure balance , @xmath49 where the * r * is the gas constant , @xmath50 is a correction factor for the turbulent contribution to the total pressure in the whm , and the right - hand - side represents the pressure in random cold cloud motions ( probably also turbulence dominated ) . for simplicity and consistency ,
we assume that @xmath51 is a constant , and so , turbulent and thermal pressures scale with radius in the same way .
in this model the turbulent pressure gradient can provide some support in the whm against gravity , so the circular velocity of this medium can be smaller than that of the cold gas .
recently , swaters , sancisi , & van der hulst ( 1997 ) discovered that the rotation velocities of hi gas located in a plane parallel to , but above , the midplane of the edge - on galaxy ngc 891 are less than those in the midplane by @xmath52 . over most of the disk
the velocity difference was about @xmath53 .
the highest values come from a couple of points within a radius of about 6.0 kpc . in the case of these points ,
the gas above the midplane may have originated in an outflow from smaller radii , and thus , considerably down the solid body part of the rotation curve .
if so , the large velocity offset may be due to conservation of the small angular momenta in this gas .
swaters et al .
provide several possible explanations for the general effect , including local outflows and the angular momentum effect just mentioned .
they also mention the possibility of `` asymmetric drift '' , though they are skeptical that velocity dispersions as large as suggested by the empirical formula for asymmetric drift in our galaxy are achievable .
however , since galactic asymmetric drift is a phenomenon of collisionless stars on orbits with large epicycles , or quite flattened elliptical orbits , we do not believe it is relevant unless elements of the whm can become isolated and collisionless for large parts of their orbits . instead
, we believe that the general velocity offset could be the result of substantial turbulence in this vigorously star - forming object ( see rand 1996 , 1997a , b , 1998 , and howk & savage 1997 )
. we will provide a simple numerical example below of how the model can account for this phenomenon in this section .
in fact , we view this phenomenon , together with the vertical radial velocity gradient , as essential predictions of the model .
thus , detailed rotation studies , like that of swaters et al . , in other nearby ,
edge - on disks with known `` dig '' components like ngc 3079 , ngc 4631 , ngc 5775 , ngc 4302 ( rand 1996 and references therein ) , and ngc 55 ( ferguson , wyse , & gallagher 1996 , hoopes , walterbos , greenwalt 1996 ) should provide a strong test of the model .
we can now describe how to scale the model , and provide some concrete examples . for brevity
, we will not consider boundary conditions here , but simply assume an infinite radius solution .
we begin by initializing several observable quantities , and also the constant radial mass transfer rate @xmath54 , which is not generally observed .
however , this quantity is related to the global sfr , since sf activity is responsible for the viscous and pressure forces .
numerical simulations should eventually yield a relation between them .
martin s ( 1998 ) studies of dwarf galaxies found the intriguing result that `` shells lift gas out of the disk at rates comparable to , or even greater than , the current galactic star formation rates . ''
this finding lends much credibility to the idea that both radial mass transfer and phase exchanges occur in late - type disks at interesting rates .
the remaining quantities we initialize are @xmath55 , and @xmath56 . in principle , these are all evaluated at a particular radius @xmath18 .
in fact , the last three are constrained to be constants , independent of radius .
the parameters determining the overall gravitational potential ( essentially @xmath19 ) can be determined from observations of stellar kinematics .
estimates for the values of @xmath57 ( or @xmath58 ) and @xmath59 can be derived from hi observations . like the radial mass flux , the values of @xmath59 for the cold clouds and @xmath24 of the whm turbulence will be determined by the sfr .
thus , we expect that for a universal ( e.g. , halo ) gravitational potential , _ this family of model star - forming disks is primarily two - dimensional , with the two dominant parameters being the cool gas density at some point and the sfr .
_ the values of the mean whm turbulence parameters @xmath24 and @xmath56 are the most difficult to evaluate .
however , @xmath56 primarily enters the equations in the combination @xmath60 , and so , whenever @xmath61 , turbulent wave pressure dominates in the whm and the exact value of @xmath56 does not greatly affect the other quantities .
moreover , @xmath56 is not independent of @xmath24 ( both are functions of the sfr ) .
the variable @xmath24 , or a dimensionless combination like @xmath62 , can be viewed as a scaling parameter of the model , like the mach number in shock hydrodynamics .
different mean whm values , appropriate for galaxies with different sfrs , are obtained by changing the value of this parameter , as we will demonstrate .
given our initial parameter values , we can now derive the scaling equations for the remaining variables .
first , from the condition of vertical hydrostatic balance , we get the cool gas scale height , @xmath63 ^{1/2},\ ] ] where @xmath64 is the local star density .
( the stars are assumed to have a larger scale height than the cool gas , so @xmath64 is essentially constant . )
next , we use the continuity equation to derive @xmath20 , the cool gas inflow , and from equation ( 2 ) we derive the value of @xmath19 @xmath65 this determines the last of the cool gas parameters , and we proceed to the remaining whm quantities . using equation ( 7 ) and the definition of @xmath51 , we note that , @xmath66 this equation can be solved directly for @xmath48 , but it also allows us to solve for the ratio of component scale heights
. the scale height ratio equals the square root of the ratio of effective temperatures ( see expression for @xmath67 above ) , which includes both a thermal and turbulent part in the case of the whm .
these parts are related by the previous equation , and so , we can write , @xmath68 using the temperature expression above in the pressure equation ( eq .
( 7 ) ) , we get the related result for the component density ratio , @xmath69 the previous two equations can now be combined to give the component surface density ratio , @xmath70 finally , we return to the radial momentum equation ( eq .
3 ) to determine @xmath71 .
the left - hand side of this equation is zero , assuming constant mean radial velocities , and for the pressure gradient term the radial dependence of the pressure in this model ( @xmath72 ) allows us to write , @xmath73 then substituting from equation ( 4 ) for the @xmath1 term , multiplying by r , and dividing by @xmath74 , the radial momentum equation can be written , @xmath75 } = 0.\ ] ] where we have also used the approximation that @xmath76 this is a bondi - parker , accretion / wind equation , and is quadratic in the variable @xmath77 .
it is a central result of the model .
to evaluate that equation numerically , we use equation ( 13 ) for @xmath78 , and the continuity equation , with a given mass flux , for @xmath20 . as an example , let us assign the following representative values : @xmath79 @xmath80 @xmath81 @xmath82 @xmath83 @xmath84 a mass flow of @xmath85 and a stellar density of @xmath86 ( used for computing scale heights ) .
then we derive values of @xmath87 @xmath88 @xmath89 @xmath90 @xmath91 @xmath92 @xmath93 and @xmath94 note that the radial velocities are very small compared to the azimuthal velocities , and also less than the turbulent velocity dispersions .
thus , we expect radial velocities to be quite difficult to observe .
if instead of the above value for @xmath24 , we substitute a higher value of @xmath95 , we obtain values of @xmath96 , and @xmath97 .
the thicker whm layer and the larger azimuthal velocity difference in this case are both in accord with the observations of ngc 891 .
the surface density of the whm relative to that of the cool gas increases linearly with @xmath24 , consistent with the idea that the strong whm emission in ngc 891 is the result of strong turbulence driven by the vigorous sf . on the other hand , we noted above that swaters et al .
( 1997 ) did not see such high velocity dispersions in their hi observations of ngc 891 .
however , these authors note that in the more nearly face - on galaxy ngc 6946 , vertical velocities of up to @xmath98 were detected by kamphuis & sancisi ( 1993 ) .
these , and related observations , can be readily understood if the turbulent motions of hi gas in the thick disk and halo are primarily vertical .
this situation is very natural if the most of the high dispersion hi gas is either entrained in local fountains and outflows , or is in the form of `` high velocity clouds '' consisting of cooled halo material falling back onto the disk ( e.g. , benjamin 1999 , benjamin & danly 1997 ) . the impressive recent study of thilker , braun & walterbos ( 1998 ) on large hi shells in ngc 2403 also provides input on this question .
these authors find mean in - plane shell expansion velocities of @xmath99 , and individual cases extending up to @xmath100 .
we expect that much of the turbulent energy has already been vented in these large bubbles , and that the shells are observed in a deceleration stage .
recent optical observations also provide evidence for the vigorous turbulence required by our model .
wang , heckman & lehnert s ( 1997 ) spectroscopic study of the dim in half a dozen nearby disks led them to suggest the existence of two components .
the first is a quiescent dim with low ionization states and line widths of @xmath101 , versus the high ionization state , disturbed dim with line widths of @xmath102 .
they suggest that the former is photoionized by diffuse o star radiation , while the latter is mechanically heated by supernovae and winds . in sum ,
turbulence with values of @xmath103 or greater seems in accord with recent observations , and seems to yield very reasonable model values for actively star - forming disks .
in constructing the model above , we have implicitly assumed that the sf law above threshold is constrained by a self - regulated heating and cooling balance . in the simplest case
, we assume that all the important heating processes are directly proportional to the sfr ( e.g. , o star photoheating and turbulent wave heating ) .
the important cooling processes in the whm include : 1 ) adiabatic cooling of high pressure gas elements , 2 ) radiative cooling in the mean whm with a rate proportional to the whm density squared , and 3 ) turbulent shock dissipation in the whm , which depends on the sfr and gas density squared .
the last two cooling rates generally scale with mean whm gas density squared , and thus , require a similar density dependence in the sfr .
that is , a schmidt law on average , albeit in density , rather than surface density .
however , the processes involved are sufficiently complex that deviations from an m=2 density power would not be surprising .
thus , we write the following schematic equation for the heating and cooling balance for regions above the sf threshold , @xmath104 the left - hand - side of this equation represents the schmidt law heating .
the right - hand - side is a schematic collisional dissipation term , with `` cross section '' @xmath105 and `` sound speed '' c. for example , this term could represent radiative cooling in the whm via collisional excitation , with the temperature dependence contained within the factor @xmath106 .
however , if we make the approximation that @xmath107 , and assume that @xmath108 is universal , then the equation provides a scaling for the net dissipation cross section @xmath105 in terms of @xmath24 and @xmath109 .
( see appendix a for further discussion of the constrained sfr . )
there is a substantial literature on individual interstellar heating and cooling terms on many scales ( e.g. , see the recent discussions of norman & ferrara 1996 , and ferriere 1998 ) .
nonetheless , we still have a long way to go to fully understand the broad - band heating and cooling terms in the ism .
any more specific formulation of the balance equation would probably require the introduction of insecure parametrizations with .
it would also require additional physics , beyond that included in the simple model considered here .
we will not pursue these topics in this paper , but merely point out that they may be easier to study in the context of the well - defined global structure provided by the model .
equations ( 8)-(15 ) show that the model is a simple similarity solution to the hydrodynamic equations . in the limit of small radial velocities and small turbulence ( and thus little whm component )
the model must be essentially the same as the self - similar , viscous , mestel disks studied recently by bertin ( 1997 ) and mineshige & umemura ( 1996 ) .
( our model also has some similarities to the one - component convective model of waxman ( 1978 ) . )
the self - similar structure of the model helps to understand the universal properties of gas - rich galaxy disks , like the tully - fisher relation between maximum circular velocities and the total luminosities of disk galaxies ( e.g. , courteau 1997 and references therein ) , and universal gas density profiles ( see e.g. , the extensive hi study of broeils & rhee 1997 ) . eisenstein and loeb ( 1996 ) point out that the small dispersion in the observational tully - fisher relation suggests the operation of a `` strong feedback process '' that `` regularize(s ) sf and gas dynamics , '' like the model presented here . on the other hand , mo et al .
( 1998 ) believe that the small tully - fisher scatter could in fact come out of early galaxy formation processes .
even so , eisenstein and loeb are probably also correct if a large fraction of galaxies experience merger events subsequent to their formation . that is , while we expect galaxy collisions and mergers to disrupt the `` universal '' disk structure , turbulent self - regulation will re - establish it .
some of these questions are answered by the new numerical hydrodynamical models of galaxy formation in several cosmologies by elizondo et al .
these models included multiple gas phases and supernova feedback .
the authors found that the tully - fisher scatter of their model galaxies was within the acceptable both with and without the feedback effects included .
however , only the feedback models reproduced the correct slope of the tully - fisher relation , and the slope was quite sensitive to the feedback amplitude .
another regularity , freeman s law , states that high surface brightness galaxies all have about the same central surface brightness , or that disk galaxies have a maximum central surface brightness ( see courteau 1996 and references therein ) .
this too would seem to be a natural consequence of large - scale sf regulation , albeit in the central regions of the disk . in a number of nearby starburst galaxies the gas surface density continues to follow a power - law as far into the center as it can be resolved ( struck - marcell 1991 , young et al .
in many cases this may be the result of gas inflow driven by a bar component or other disturbance .
in many other late - type disks the gradient in the surface density flattens to a value of about 10 solar masses per square parsec in the central , rising rotation curve region ( though often with a central , molecular gas spike , see young et al .
we speculate that this latter case represents a normal quiescent state .
disks of both profile types are observed to have a comparable value of @xmath110 at the radius where the rotation curve flattens ( broeils & rhee 1997 ) .
we conjecture that freeman s law may be the result of the fact that most stars form in the centers of bright , late - type disks at a rate appropriate to this gas density , and over comparable timescales .
the following list summarizes the properties and some probable consequences of the models presented in this paper .
\1 ) a @xmath2 surface density profile is assumed in both thermal phases ( on the basis of the least dissipation principle and other arguments , see section 2.2 ) .
it is also assumed that vertical scale heights are determined by local self - gravity , and so , increase slowly with radius ( as @xmath111 ) .
this implies that mean volume densities scale as , @xmath112 .
\2 ) thus , the ratio of gas phase densities are constant across the star - forming region of the galaxy disk .
the model predicts that the value of this ratio depends on the amount of turbulence , and specifically , on the parameter @xmath113 .
\3 ) we assume that the circular velocities of each gas phase are constants independent of radius .
the equations of state and the assumption that all momentum equation terms have the same radial scaling implies that the remaining velocities , @xmath114 , are also constant with radius .
\4 ) in general , the model allows all of the velocities in each phase - azimuthal , radial , and dispersive - to have different ( non - zero ) values .
the ngc 891 effect of different rotational velocities as a function of height above the disk ( swaters et al .
1997 ) is predicted to be generic in turbulent disks .
\5 ) the model predicts a hierarchy in velocity magnitudes in each phase , i.e. , azimuthal velocities @xmath115 velocity dispersions @xmath116 radial velocities .
the low value of the latter will make them difficult to observe .
this hierarchy is the result of the similarity equations and the simple wind / accretion equation ( eqs .
( 8) - ( 15 ) ) .
\6 ) however , even such low velocity radial flows are consistent with mass fluxes comparable to typical sfrs in late - type disks . in the absence of radial flows , we would expect gas consumption at smaller disk radii , to modify the gas density profile .
radial replenishment can prevent this and effectively distribute the consumption across the disk .
the radial flow may also draw on reservoirs of gas in the non - star - forming outer disk , further increasing the global consumption time .
evolutionary effects will be considered in a later paper .
\7 ) similarly , galactic abundance gradients will be smoothed by the large - scale radial flows .
the simple closed - box model of chemical evolution within isolated disk annuli is not appropriate in the context of these radial circulation models .
however , the quantitative effects of radial flow are complicated by the fact that the flows are slow . e.g. , with a radial flow velocity of order @xmath117 , the timescale for a gas element to cross a disk of radius @xmath118 is a few billion years . moreover ,
motion of a gas element will generally be partly advective and partly diffusive in this turbulent environment .
thus , the typical smoothing time may be only a little less than the typical disk age .
\8 ) the model requires a balance between heating and cooling .
heating is primarily the result of sf activity , and most cooling terms depend on the second power of the gas density ( assuming the constant phase balances of the model ) .
thus , a schmidt law sfr is the natural result of the thermohydrodynamical balance .
\9 ) if the schmidt law ( or a related parametrization , such as that of silk and wyse , see silk 1997 ) is in fact a consequence of global hydrodynamic self - regulation , then there are some immediate corollaries .
perhaps , the most important is that transient , burst modes of sf are possible when disturbances take galaxies far from the regulated state .
thus , sf phenomenologies may be very different in highly disturbed galaxies ( e.g. , struck - marcell & scalo 1987 ) , or during galaxy formation .
on the other hand , the schmidt parametrization may be marginally valid in environments where , @xmath119 , with @xmath120 defined as an appropriate `` age '' .
for example , the schmidt law may work in waves in both grand design and collisional galaxies if the wave crossing time is longer than @xmath121 .
the same argument may be valid in the centers of major merger remnants .
\10 ) global regularities in star - forming disks , such as the tully - fisher relations and the freeman law , may also be the result of global self - regulation , of the type inherent in the present models .
they may also be the result of formation processes , including turbulent self - regulation during formation .
continuing self - regulation is important for maintaining the global regularities , and restoring them following a disturbance .
\11 ) the model equations can readily be generalized beyond the two - phase version described here to include a continuous range of phases .
more sophisticated treatments of viscosity , turbulence , heating and cooling processes can be included , much as detailed nuclear rates and opacities are included in stellar evolution models . in sum ,
the hydrodynamic similarity model presented above is an attempt to bring together the essential thermohydrodynamical processes needed for a coherent conceptual picture of actively star - forming galaxy disks as self - regulated , multiphase , `` dissipative structures . ''
the basic hydrodynamic structure of each phase is much like that of an isothermal polytrope , but these are not quasi - static , equilibrium states .
the model assumes that there are turbulent flows on many scales , and the whm is more accurately viewed as a set of locally adiabatic states , with specific entropy gradients .
the disk structure described by this model , with gas profiles regulated in accord with minimal dissipation and transport , and sf simultaneously sustained and moderated by slow , radial flows , probably describes most late - type disk galaxies .
acknowledgements : we are very grateful to julia k. burzon for creating figure 1 .
we also want to thank p. n. appleton , r. benjamin , v. charmandaris , b. elmegreen , l. sage , j. scalo , j. m. van der hulst and especially an anonymous referee and our editor s. shore for providing helpful ( regulating ! ) feedback .
we are also grateful for partial page charge support from an american astronomical society small grant .
the positive radial entropy gradients in the whm component of the model presented above may seem unphysical , especially since this gas is expanding in an average outward flow .
however , there is an obvious entropy source in the nonlinear acoustic waves that partially support this flow . at the same time , radiative cooling provides an obvious entropy sink , yet when this gas experiences significant cooling it is generally transformed into the cool component .
thus , on average , the entropy of gas that stays in the whm either increases , or is balanced by adiabatic expansion .
specifically , consider a non - cooling element of the whm moving outward in the mean flow above the midplane of the disk .
suppose for simplicity , that its specific entropy is significantly increased only when it passes directly over a young star cluster , assuming such star clusters are the primary source of shock turbulence .
then , the rate at which the entropy of that element is increased will be proportional to the number of young clusters it passes over per unit time .
( note that because of the reduced azimuthal velocity of the whm , the element will pursue a spiral trajectory as viewed in a reference frame comoving with the midplane gas . ) in this equation , @xmath124 is the usual sfr ( mass of stars produced per unit area per unit time ) , @xmath125 is the mean cluster mass , and @xmath126 is the mean lifetime of the massive stars in the cluster .
the final equality assumes a schmidt law of the form , @xmath127 , and a surface density profile of the form @xmath128 .
the constant @xmath129 is the combination of all constants in the previous equality .
we further assume that the gas element expands in the azimuthal direction by an amount proportional to r as it moves outward .
this is just the expansion that is required to maintain the assumed surface density profile .
it also guarantees that the whm element covers a constant fraction of each thin annulus it crosses .
thus , the gas element crosses a constant fraction of the clusters @xmath130 in each thin annulus , and it is reasonable to assume that the rate of shock hits and entropy increase it experiences is proportional to this annular cluster fraction .
thus , the net entropy increase in traveling from radius @xmath131 to radius r is where the second equality makes use of equation ( 6 ) , in section 2.3 , ( and the subscript v emphasizes that these are volume quantities ) .
assuming that this equals the preceding equation to within an additive constant , we have , as noted in section 2.3 the quasi - steady model presented above requires that @xmath136 .
one factor contained within the constant @xmath133 is the magnitude of the sfr , e.g. , the sfr at a particular radius .
thus , the self - regulating feedback processes can adjust the sfr in such a way that @xmath137 is driven towards unity , giving the desired entropy gradient .
therefore , the fact that the entropy gradient is a power - law follows from the @xmath138 spatial dependence of the schmidt law sfr , while the value of the power depends on the magnitude of the sfr . in other words , the entropy gradient required for a hydrodynamic steady state can be achieved by feedback adjustments to the amplitude and spatial dependence of the sfr .
now let us consider the effects of these self - regulating processes from a slightly different point of view . according to equations ( 6 ) and ( 7 ) the mean whm temperature scales as , @xmath139 . for a globally adiabatic gas , k is constant , and @xmath140 ( for constant @xmath51 ) .
while , as described above , for a locally adiabatic gas , with a radial entropy gradient such that @xmath141 , we have @xmath48 constant .
if the latter alternative does not obtain , then the variation of scale height with radius will be different than the @xmath142 form assumed above .
qualitatively , if the scale height increases less rapidly , then warm gas remains closer to the midplane and is denser , so cooling rates are increased .
if more of this gas goes into the cool phase , we expect that the sfr will increase ( relative to the quasi - steady model ) , driving increased turbulence and heating , and increasing the scale height .
conversely , if the scale height increases more rapidly with radius , the sfr will be less than in the steady model , eventually diminishing pressure support , and reducing the scale height . more generally , because of the temperature dependence of the cooling rates , significant temperature gradients would likely result in pressure imbalances , which would lead to time - dependent convection ( as well as thermal conduction ) .
that is , temperature gradient states do not generally satisfy the steady state equations above .
these considerations are qualitative , and have loopholes , but they do show why global states with modest entropy gradients would be preferred .
they also provide some insight into how closely the schmidt law sfr is connected to such states . | supernova explosions , and winds and energetic photon fluxes from young star clusters drive outflows and supersonic turbulence in the interstellar medium in galaxy disks , and provide broad spectrum heating which generates a wide range of thermal phases in the gas .
star formation , the source of the energy inputs , is itself regulated by heating and phase exchanges in the gas .
however , thermohydrodynamic self - regulation can not be a strictly local process in the interstellar gas , since galaxy disks also have a nearly universal structure on large scales .
we propose that turbulent heating , wave pressure and gas exchanges between different regions of disks play a dominant role in determining the preferred , quasi - equilibrium , self - similar states of gas disks on large - scales . in this paper
we present simple families of analytic , thermohydrodynamic models for these global states , which include terms for turbulent pressure and reynolds stresses . in these model disks
star formation rates , phase balances , and hydrodynamic forces are all tightly coupled and balanced .
the models have stratified radial flows , with the cold gas slowly flowing inward in the midplane of the disk , and with the warm@xmath0hot phases that surround the midplane flowing outward .
the models suggest a number of results that are in accord with observation , as well as some novel predictions , including the following .
1 ) the large - scale gas density and thermal phase distributions in galaxy disks can be explained as the result of turbulent heating and spatial couplings .
2 ) the turbulent pressures and stresses that drive radial outflows in the warm gas above and below the disk midplane also allow a reduced circular velocity there .
this effect was observed by swaters , sancisi & van der hulst in ngc 891 , a particularly turbulent edge - on disk .
the models predict that the effect should be universal in such disks .
3 ) since dissipative processes generally depend on the square of the gas density , the heating and cooling balance in these models requires a star formation rate like that of the schmidt law .
conversely , they suggest that the schmidt law is the natural result of global thermohydrodynamical balance , and may not obtain in disks far from equilibrium
. = cmr9 |
we consider a quantum system consisting of @xmath0 isolated qubits with the hilbert space @xmath1 .
we assume that states are normalized to one and neglect a global phase . in this way
the space of pure states is isomorphic to the complex projective space @xmath2 . in the following , by @xmath3\in\mathbb{p}(\mathcal{h})$ ]
we will denote a state corresponding to a vector @xmath4 each qubit is located in a different laboratory and the available operations are restricted to the local unitaries described by the group @xmath5 .
two states are called locally unitary equivalent ( lu equivalent ) if and only if they can be connected by the action of @xmath6 , that is , belong to the same @xmath6-orbit
. the problem of local unitary equivalence of states can be in principle solved by finding the set of @xmath6-invariant polynomials , i.e. polynomials that are constant on @xmath6-orbits ( see @xcite for another approach ) .
when the number of qubits is large this , however , becomes hard as the number of polynomials grows exponentially with the number of constituents of the system .
the problem of lu equivalence for bipartite and three - qubit pure states was recently studied form the symplecto - geometric perspective @xcite ( see also @xcite ) .
in particular , the connection with the symplectic reduction was established .
the current paper can be seen as a generalization of these ideas to an arbitrary number of qubits . among the @xmath6-invariant polynomials
there are @xmath0 polynomials @xmath7)\right)\right\ } _ { l=1}^{l}$ ] , where @xmath8)$ ] are the reduced one - qubit density matrices .
consequently , for two lu equivalent states @xmath9\in\mathbb{p}(\mathcal{h})$ ] the spectra of the corresponding reduced one - qubit density matrices are the same . if we denote by @xmath10 the map which assigns to @xmath3\in\mathbb{p}(\mathcal{h})$ ] the shifted spectra of its reduced one - qubit density matrices , i.e. @xmath11)=\{\mathrm{diag}(-\lambda_1,\lambda_1),\ldots,\mathrm{diag}(-\lambda_l,\lambda_l)\}$ ] , where @xmath12 and @xmath13 is the increasingly ordered spectrum of @xmath14)$ ] , then the states satisfying the above necessary condition form a fiber of @xmath10 .
fibers of @xmath10 are connected collections of @xmath6-orbits @xcite . moreover , the image , @xmath15 , is a convex polytope @xcite .
the polynomials @xmath7)\right)\right\ } _
{ l=1}^{l}$ ] restricted to a fiber of @xmath10 are constant functions .
therefore , typically , for two states @xmath16 $ ] and @xmath17 $ ] with @xmath18)=\psi([\phi_{2}])$ ] , where @xmath19 denote the collection of spectra of one - qubit density matrices , some additional @xmath6-invariant polynomials are needed to decide the lu equivalence .
the number of these polynomials is given by the dimension of the _ reduced space _
@xmath20 ( see @xcite ) .
interestingly , @xmath21 may not be the same for every @xmath22 , that is , some collections of spectra of reduced one - qubit density matrices require less additional polynomials to solve lu equivalence problem than others . in particular , if the fiber @xmath23 contains exactly one @xmath6-orbit , i.e. @xmath24 , no additional information is needed and any two states @xmath9\in\psi^{-1}\left(\alpha\right)$ ] are lu equivalent . in this paper
we find the formula for the dimension of the reduced space @xmath25 , for any @xmath22 and for an arbitrary number @xmath0 of qubits .
our result is obtained in two steps .
first , we consider the points @xmath26 which belong to the interior of the polytope @xmath15 . in this case
the map @xmath10 is _ regular _ and the calculation is rather straightforward .
the dimension of @xmath27 does not depend on @xmath28 .
moreover , @xmath29 is equal to the cardinality of the spanning set of @xmath6-invariant polynomials . for points
@xmath30 which belong to the boundary of @xmath15 the problem requires more advanced methods and turns out to be more interesting . in particular , for a large part of the boundary of @xmath15 we have @xmath31 .
we also observe that for @xmath30 corresponding to the @xmath32)\right\ } _ { l=1}^{l}$ ] such that @xmath33 matrices are maximally mixed @xmath34 .
in this section we briefly discuss the considered problem and present the main results of the paper on the 4 qubits example .
recently , the problem of finding @xmath21 was considered for three qubits @xcite . in particular
it was shown that for points in the interior of the polytope @xmath15 , @xmath35 , whereas for points on the boundary @xmath24 .
the uniform behaviour of @xmath21 on the boundary of @xmath15 in case of three qubits is , as already indicated in ref.@xcite , a low dimensional phenomenon . as we explain in section [ sec : the - reduced - spaces ] , for an arbitrary number of @xmath0 qubits the boundary consists of three parts characterized by a different behaviour of @xmath21 .
the first part is the polytope @xmath36 that corresponds to a system with one qubit less , that is , @xmath37 .
the second corresponds to changing one of the non - trivial inequalities ( [ nier1 ] ) into an equality .
the third represents situations when @xmath33 one - qubit density matrices are maximally mixed . the clear distinction between these parts of the boundary
can be seen already in the four qubits case .
the four - qubit polytope @xmath15 is a @xmath38-dimensional convex polytope spanned by 12 vertices ( see appendix for the proof and the list of vertcies ) . the dimension of the reduced space in the interior of @xmath15 is @xmath39 ( see formula ( [ eq : fiber - interior-1 ] ) ) . in figure
[ fig : three - kinds - of ] the above mentioned three different parts of the boundary are shown . in particular , in figure [ fig : three - kinds - of](a ) we see that for @xmath40-dimensional face of @xmath15 corresponding to three qubits , @xmath35 in the interior and @xmath24 on the boundary which agrees with results of ref . @xcite . on the other hand , inside the @xmath40-dimensional face shown in figure [ fig : three - kinds - of](b )
corresponding to one of @xmath41 being maximally mixed we have @xmath42 .
the boundary of this face contains : @xmath43-dimensional faces corresponding to two of @xmath41 being maximally mixed - @xmath44 , @xmath45-dimensional faces - three of @xmath41 are maximally mixed and @xmath46 , and finally , the vertex denoted by @xmath47 when all one - particle reduced density matrices are maximally mixed - @xmath48 .
therefore , as mentioned in the introduction , @xmath34 . finally , in figure [ fig : three - kinds - of](c ) we see the @xmath40-dimensional face of @xmath15 with @xmath24 .
three parts of the boundary of @xmath15 .
the numbers denote @xmath21 .
if the number is missing , @xmath24 .
the vertices are defined in appendix.,width=529 ] in the next sections we show how to calculate @xmath21 for any point of @xmath15 for an arbitrary number of qubits .
we start with a rigorous statement of the problem and the solution of its easy part . for a detailed description of symplecto - geometric methods in quantum information theory
see for example @xcite .
let @xmath1 be the @xmath0-qubit hilbert space and denote by @xmath2 the corresponding complex projective space .
it is known that @xmath2 is a symplectic manifold with the fubiny - study symplectic form @xmath50 .
the action of @xmath5 on @xmath2 is symplectic , i.e. it preserves @xmath50 .
consequently , there is the _ momentum map _ for this action . in the considered setting this map
is given by @xmath51 @xmath52)=\left\ { \rho_{1}([\phi])-\frac{1}{2}i,\ldots,\rho_{l}([\phi])-\frac{1}{2}i\right\ } , \ ] ] where @xmath53 is the lie algebra of @xmath6 , @xmath32)\right\ } _ { l=1}^{l}$ ] are the reduced one - qubit density matrices , @xmath54 is @xmath55 identity matrix and @xmath56 .
the map @xmath57 is equivariant , i.e. for any @xmath58 @xmath59)=\left\ { \rho_{1}([g\phi])-\frac{1}{2}i,\ldots,\rho_{l}([g\phi])-\frac{1}{2}i\right\ } = \\
= \left\ { g\left(\rho_{1}([\phi])-\frac{1}{2}i\right)g^{\ast},\ldots , g\left(\rho_{l}([\phi])-\frac{1}{2}i\right)g^{\ast}\right\ } = g\mu([\phi])g^{\ast}=\mathrm{ad}_{g}\mu([\phi]),\end{gathered}\ ] ] where @xmath60 is the hermitian conjugate of @xmath61 and @xmath62 , for any @xmath63 .
the equivariance of @xmath57 implies that @xmath6-orbits in @xmath2 are mapped onto adjoint orbits in @xmath64 .
consequently , the necessary condition for two states @xmath9\in\mathcal{h}$ ] to be on the same @xmath6-orbit in @xmath2 is @xmath65)=\mu(k.[\phi_{2}])$ ] . on the other hand , adjoint orbits in @xmath64
are determined by the spectra of matrices @xmath32)\right\ } _ { l=1}^{l}$ ] .
the characteristic polynomial @xmath66 for @xmath8)$ ] reads @xmath67)-\frac{1}{2}i\right)\nu+\left(\left(\mathrm{tr}\left(\rho_{l}([\phi])-\frac{1}{2}i\right)\right)^2-\mathrm{tr}\left(\rho_{l}([\phi])-\frac{1}{2}i\right)^{2}\right).\label{eq : ch - poly}\end{gathered}\ ] ] using @xmath68)\right)=1 $ ] , equation ( [ eq : ch - poly ] ) reduces to @xmath69)\right)-\frac{1}{2}.\label{eq : ch - poly-1}\end{gathered}\ ] ] one can thus say that the necessary condition for two states @xmath9\in\mathbb{p}(\mathcal{h})$ ] to be on the same @xmath6-orbit in @xmath2 is that @xmath70)\right)=\mathrm{tr}\left(\rho_{l}^{2}([\phi_{2}])\right).\end{gathered}\ ] ] let @xmath71 be the positive weyl chamber in @xmath64 , i.e. @xmath72 we define the map @xmath73 to be @xmath74):=\mu(k.[\phi])\cap i\mathfrak{t}_{+}=\left\ { \tilde{\rho}_{1}([\phi])-\frac{1}{2}i,\ldots,\tilde{\rho}_{l}([\phi])-\frac{1}{2}i\right\ } , \end{gathered}\ ] ] where each @xmath75)$ ] is a diagonal @xmath55 matrix whose diagonal elements are given by the increasingly ordered spectrum @xmath76\right)=\{p_{l},\,1-p_{l}\}$ ] of @xmath8)$ ] , that is , @xmath77 .
the shifted spectrum , i.e. the spectrum of @xmath8)-\frac{1}{2}i$ ] is given by @xmath78 , where @xmath79 under these assumptions the image @xmath15 is known to be a convex polytope , defined by the following set of inequalities @xcite @xmath80 note that the polytope @xmath15 gives 1 - 1 parametrization of @xmath81 , that is , of all available adjoint @xmath6-orbits .
moreover , for each point @xmath22 the set @xmath82 is a connected @xmath6-invariant stratified symplectic space @xcite .
we will next briefly describe the structure of this space @xcite .
note that isotropy groups of points belonging to a fixed @xmath6-orbit are conjugated .
first , one can decompose @xmath2 according to the conjugacy classes of isotropy groups , i.e. into sets of points , whose isotropies are conjugated .
this decomposition divides @xmath2 into sets of orbits characterized by the same isotropy type .
these sets are in general non - connected , thus performing the decomposition further into connected components , one obtains the stratification @xmath83 there exists an unique highest dimensional stratum @xmath84 , which is open and dense .
moreover , it can be shown that for each @xmath22 the set @xmath85 is smooth and that the decomposition @xmath86 gives the stratification of @xmath87 .
let @xmath88 be the isotropy subgroup of @xmath19 with respect to the adjoint action , i.e. @xmath89 . because @xmath87 is invariant to the action of @xmath88 , one has that the _ reduced space _ @xmath90 .
the stratification given by ( [ eq : strat ] ) induces the stratification of @xmath49 in a natural way .
what is more , every @xmath49 also has an unique highest dimensional stratum @xmath91 . if @xmath92 then @xmath93 therefore it is clear that the dimension of the highest dimensional stratum in @xmath49 , or more precisely @xmath94 is the number of @xmath6-invariant polynomials needed to decide lu equivalence of states satisfying @xmath11)=\alpha$ ] .
the problem of finding @xmath21 is essentially different for points @xmath19 in the interior and on the boundary of @xmath15 .
using ( [ nier1 ] ) the boundary of @xmath15 can be divided into three parts * case 1 * : : @xmath33 of @xmath95s are equal to @xmath96 . * case 2 * : : at least one of the inequalities @xmath97 is an equality .
* case 3 * : : @xmath33 of @xmath95s are equal to @xmath98 . by the permutation symmetry of inequalities ( [ nier1 ] ) ,
it is enough to consider one example for each of these cases . in the remaining two paragraphs of this section we find @xmath99 for the interior and the boundary points satisfying the condition of the first case .
for the points @xmath19 in the interior of the polytope @xmath15 the calculation of @xmath21 turns out to be rather straightforward . to begin ,
let us denote by @xmath100 the union of @xmath6-orbits of maximal dimension in @xmath2 .
the principal isotropy theorem @xcite implies that this set is connected , open and dense . in order to calculate the maximal dimension of @xmath6-orbits in @xmath2 we use the following fact ( see ref .
@xcite ) [ dim - orbit ] [ dim - orbit]assume that @xmath101 .
then a generic @xmath6-orbit has dimension of the group @xmath6 . for @xmath0-qubits , @xmath102 ,
one easily checks that the polytope @xmath15 given by inequalities ( [ nier1 ] ) , is @xmath0-dimensional . on the other hand , for the weyl chamber defined by ( [ eq : chamber ] ) , we have @xmath103 . therefore , by proposition [ dim - orbit ] the @xmath6-orbits belonging to @xmath100 have dimension of the group @xmath5 , i.e @xmath104 .
moreover , @xmath105 contains the interior of the polytope @xmath15 ( see ref.@xcite ) and for @xmath19 in the interior of the polytope @xmath15 the set @xmath106 is the highest dimensional stratum of @xmath82 ( see ref.@xcite ) . hence @xmath107 note that already in the case of three qubits we have @xmath35 , that is , one needs @xmath108 , @xmath6-invariant polynomials to decide lu equivalence of states whose spectra of reduced density matrices are the same and belong to the interior of @xmath15 .
the exponential growth of @xmath21 , which for large @xmath0 is of order @xmath109 , can be seen as a usual statement that the number of @xmath6-invariant polynomials needed to distinguish between generic @xmath6-orbits grows exponentially with the number of particles .
the case 1 is once again straightforward . to see this note that when the first @xmath33 of @xmath95 s are equal to @xmath96 inequalities ( [ nier1 ] ) reduce to the inequalities for the polytope of @xmath110 qubits .
moreover , a generic state belonging to @xmath82 is of the form @xmath111 where @xmath112 is a @xmath33-qubit separable state and @xmath113 is an arbitrary state of @xmath110 qubits .
therefore , using ( [ eq : fiber - interior-1 ] ) , we get @xmath114 in the next two sections we find @xmath21 for cases 2 and 3 .
in this section we show that @xmath24 for @xmath19 s satisfying assumptions of case 2 .
assume that one of the inequalities @xmath97 is an equality , e.g. @xmath115 using the fact that states mapped by @xmath10 onto @xmath19 are @xmath6-orbits through states mapped by @xmath57 onto @xmath19 , i.e. @xmath116 , one can , if more convenient , use map @xmath57 instead of @xmath10 to calculate @xmath21 .
it turns out that this is the case for the considered @xmath19 s , as we have the following : [ prop1 ] let @xmath117 be such that @xmath118)$ ] satisfies ( [ eq : condition1 - 1 - 1 ] ) .
let @xmath119 $ ] be a vector perpendicular to the plane given by ( [ eq : condition1 - 1 - 1 ] ) .
then @xmath120 is an eigenvector of @xmath121 , where @xmath122 . in order to prove this
, we will use the fact characterizing the image of the differential of the momentum map @xcite : [ prop2 ] the image of @xmath123}:t_{[\phi]}m\rightarrow i\mathfrak{k}$ ] is equal to the annihilator of @xmath124}$ ] , the lie algebra of the isotropy subgroup @xmath125}\subset k$ ] .
( proposition [ prop1 ] ) the lie algebra of @xmath6 is a real vector space equipped with the inner product given by @xmath126 .
the matrices @xmath127 @xmath128 @xmath129 where @xmath130 are pauli matrices , form an orthogonal basis of @xmath53 .
the matrix representing the collection @xmath131)-\frac{1}{2}i,\ldots,\rho_{l}([\phi])-\frac{1}{2}i\right\ } $ ] is of the form : @xmath132)-\frac{1}{2}i)\otimes i\otimes\ldots\otimes i+\ldots+i\otimes\ldots\otimes i\otimes(\rho_{l}([\phi])-\frac{1}{2}i).\end{gathered}\ ] ] note that @xmath123}$ ] is the map that transports vectors tangent to @xmath2 at the point @xmath3 $ ] to the tangent space to @xmath53 , @xmath133)}\mathfrak{k}\cong\mathfrak{k}$ ] .
assume that @xmath134)$ ] belongs to @xmath15 and that spectra of the corresponding matrices @xmath135)\}_{i=1}^{l}$ ] are nondegenerate .
in this case the lie algebra @xmath136 of the isotropy subgroup @xmath137 is given by the diagonal matrices in @xmath53 , the set of which we denote by @xmath138 .
on the other hand by the equivariance of the momentum map ( [ equvariance ] ) we have @xmath124}\subset\mathfrak{k}_{\alpha}=\mathfrak{t}$ ] .
in other words , states @xmath3 $ ] mapped by @xmath57 to @xmath22 with non - degenerate spectra of @xmath135)\}_{i=1}^{l}$ ] can have isotropy given by at most @xmath138 . as the off - diagonal matrices in the image of @xmath123}$ ]
are orthogonal to @xmath138 , by proposition [ prop2 ] , in order to find @xmath124}$ ] we need to find matrices in @xmath138 orthogonal to ( annihilated by ) diagonal matrices from @xmath123}$ ] .
note that since @xmath139 and @xmath140 , for @xmath19 inside @xmath15 the diagonal matrices in @xmath133)}\mathfrak{k}$ ] , for @xmath3\in\mathbb{p}(\mathcal{h})^{max}$ ] , span the space @xmath138 and hence the isotropy @xmath124}=0 $ ] . on the other hand , for @xmath19 satisfying ( [ eq : condition1 - 1 - 1 ] ) the space @xmath133)}\mathfrak{k}\cap\mathfrak{t}\neq\mathfrak{t}$ ] . in order to find matrices in @xmath138 orthogonal to @xmath133)}\mathfrak{k}\cap\mathfrak{t}$ ]
note that any element of @xmath138 can be written as @xmath141 .
the inner product of two matrices of this type reads @xmath142 i.e. is equal to the standard inner product of vectors @xmath143 $ ] and @xmath144 $ ] in @xmath145 .
the vectors @xmath146 corresponding to the diagonal matrices from the image of @xmath123}$ ] are tangent to @xmath15 at the point @xmath118)$ ] satisfying ( [ eq : condition1 - 1 - 1 ] ) . therefore , if @xmath119 $ ] is a vector perpendicular to the plane given by ( [ eq : condition1 - 1 - 1 ] ) then the corresponding operator @xmath121 , where @xmath122 is the element of the lie algebra of the isotropy group @xmath124}$ ] .
consequently @xmath147 , for some @xmath148 .
the vector @xmath149 $ ] is perpendicular to the plane given by ( [ eq : condition1 - 1 - 1 ] ) .
the corresponding operator @xmath150 reads @xmath151 where @xmath152 , @xmath153 . by proposition [ prop1 ]
we need to consider eigenspaces of @xmath150 .
we have the following : [ the - spectrum ] the matrix @xmath150 , defined by ( [ matrix - x ] ) , is a diagonal @xmath154 matrix .
the eigenvalues of @xmath150 are the integers chosen from @xmath155 to @xmath0 with the step @xmath43 , that is @xmath156 .
the multiplicity of eigenspace @xmath157 is @xmath158 . the matrices @xmath159 are diagonal and their spectra are @xmath160 .
consequently , the matrix @xmath150 is also diagonal and its eigenvalues are sums of eigenvalues of @xmath159 s .
one can easily verify that the eigenvalues of @xmath150 belong to the set @xmath156 . to see this note that the eigenvalue @xmath161 arises as a sum of eigenvalues of @xmath162 , where @xmath33 out of @xmath0 eigenvalues of @xmath159 s are positive ( @xmath163 ) and @xmath110 negative ( @xmath164 ) .
therefore the multiplicity of @xmath157 is @xmath158 . as a direct consequence of proposition [ the - spectrum ]
we need to consider @xmath165 eigenspaces of @xmath150 . in the following
we describe the structure of these spaces and show that only one of them , that is , @xmath166 contains states @xmath3 $ ] for which @xmath118)$ ] consists of diagonal matrices whose diagonal elements satisfy ( [ eq : condition1 - 1 - 1 ] ) .
the result is obtained in two steps .
first in proposition [ states - form ] we determine @xmath157 and show that condition ( [ eq : condition1 - 1 - 1 ] ) is not satisfied for @xmath167 .
then in proposition [ k=1 ] we prove that for @xmath166 condition ( [ eq : condition1 - 1 - 1 ] ) is satisfied .
denote by @xmath168 the subspace of @xmath169 spanned by separable states of @xmath0 qubits such that @xmath33 out of @xmath0 qubits are in the ground state @xmath170 and the remaining @xmath110 qubits are in the excited state @xmath171 , for example , @xmath172 .
assume that @xmath173 if @xmath174 .
[ states - form]states which belong to eigenspace @xmath157 are of the form @xmath175 , where @xmath176 and @xmath177 .
the reduced one - qubit density matrices for any @xmath178 are diagonal . for any @xmath178 condition ( [ eq : condition1 - 1 - 1 ] )
is equivalent to @xmath179 .
we first determine vectors spanning eigenspace @xmath157 of @xmath150 .
the eigenvalue @xmath161 arises as the sum of eigenvalues of @xmath162 with @xmath33 out of @xmath0 eigenvalues equal to @xmath180 and @xmath110 equal to @xmath181 .
note , however , that matrix @xmath152 , whereas @xmath153 .
therefore , the eigenvalue @xmath161 corresponds to separable states with either the first qubit in the ground state @xmath170 and the remaining @xmath182 qubits in a state from @xmath183or with the first qubit in the excited state @xmath171 and the remaining @xmath182 qubits in a state from @xmath184 .
thus the generic state belonging to @xmath157 can be written as @xmath185 where @xmath186 and @xmath187 are the separable states spanning @xmath183 and @xmath184 respectively .
it is straightforward to see that the reduced one - qubit density matrices of @xmath120 are diagonal .
the first one is of the form : @xmath188)-\frac{1}{2}i=\left(\begin{array}{cc } \sum_{l=1}^{{l-1 \choose k}}|a_{l}|^{2}-\frac{1}{2 } & 0\\ 0 & \sum_{l=1}^{{l-1 \choose k-1}}|b_{l}|^{2}-\frac{1}{2 } \end{array}\right),\ ] ] and hence @xmath189 we now show that ( [ eq : condition1 - 1 - 1 ] ) is equivalent to @xmath179 . as matrices @xmath8)$ ] are diagonal , the @xmath190)\right)_{11}$ ] entry of each @xmath191 , which is equal to @xmath192 , is the sum of @xmath193 and @xmath194 coefficients corresponding to vectors with the @xmath195-th qubit in the excited state @xmath171 .
consequently , in the sum @xmath196)\right)_{11}$ ] , each @xmath193 coefficient occurs @xmath110 times and each @xmath194 coefficient @xmath197 times .
therefore , @xmath198 using ( [ eq : lambda1 ] ) @xmath199 hence equation ( [ eq : condition1 - 1 - 1 ] ) reads @xmath200 using fact [ states - form ] one easily finds that for normalized state , i.e. when @xmath201 condition ( [ eq : condition1 - 1 - 1 ] ) can be satisfied only when @xmath202 .
the following proposition ensures that indeed this is the case .
[ k=1]the reduced one - qubit density matrices of states @xmath203 are diagonal and satisfy condition ( [ eq : condition1 - 1 - 1 ] ) .
the eigenspace @xmath166 is @xmath0-dimensional .
any vector @xmath203 can be written as @xmath204 that is , @xmath120 is a linear combination of separable state where all qubits are in the excited state and the states for which the first and one additional qubits are in the ground state ( while other are in the excited state ) .
assume that @xmath120 is normalized , i.e. @xmath205 .
it is straightforward to calculate @xmath206)-\frac{1}{2}i=\left(\begin{array}{cc } 0 & -\frac{1}{2}+\sum_{k\neq l}^{l}|c_{k}|^{2 } \end{array}\right),\,\,\ , i\in\{2,\ldots , l\},\\ \rho_{1}([\phi])-\frac{1}{2}i=\left(\begin{array}{cc } -\frac{1}{2}+\sum_{k=2}^{l}|c_{k}|^{2 } & 0\\ 0 & |c_{1}|^{2}-\frac{1}{2 } \end{array}\right).\end{gathered}\ ] ] note that we can assume that @xmath207 .
this means that for all @xmath208 we have@xmath190)-\frac{1}{2}i\right)_{11}\leq0 $ ] as required .
it is also easy to see that condition ( [ eq : condition1 - 1 - 1 ] ) is equivalent to @xmath209 and is satisfied . by proposition [ k=1 ] , states mapped by @xmath10 onto @xmath19 s satisfying ( [ eq : condition1 - 1 - 1 ] ) belong to @xmath6-orbits through @xmath203 . on the other hand states @xmath203
are @xmath210-equivalent to @xmath0-qubit @xmath211 state , where @xmath212 is complexification of @xmath5 and @xmath213={\mbox{$| 01\ldots1 \rangle$}}+{\mbox{$| 101\ldots1 \rangle$}}+\ldots+{\mbox{$| 1\ldots10 \rangle$}}.\end{gathered}\ ] ] this can be easily seen by changing @xmath214 on the first qubit of ( [ eq : almost w ] ) .
it was shown in ref.@xcite that the variety @xmath215}$ ] is _ spherical _
, i.e. reduced spaces stemming from the restriction @xmath216}}$ ] are zero - dimensional . therefore : let @xmath22 be such that at least one of the inequalities @xmath97 is equality
. then @xmath24 .
as we showed in section [ sec : the - reduced - spaces ] , for points @xmath19 in the interior of @xmath15 the dimension of the reduced space is @xmath217 in the following we show that for @xmath218 with matrices @xmath219 @xmath220 this means that the dimension @xmath21 drops by @xmath43 every time one of the reduced density matrices becomes maximally mixed .
the argument for this is based on the existence of _ stable states _ which we discuss first . in the following
we assume that @xmath221 is any compact semisimple group acting in the symplectic way on the complex projective space @xmath2 , where @xmath222 can be , for example , the hilbert space of @xmath0 qubits .
we will denote by @xmath223 the corresponding momentum map and assume that @xmath224 .
let @xmath225 be the complexification of @xmath221 .
following ref .
@xcite we denote @xmath226\in\mathbb{p}(\mathcal{h}):\,\overline{\tilde{g}.[\phi]}\cap\tilde{\mu}^{-1}(0)\neq\emptyset\}.\end{gathered}\ ] ] it is known that the set @xmath227 is an open dense subset of @xmath2 .
moreover the set @xmath228 is also an open dense subset of @xmath2 ( see ref.@xcite ) .
we will use the following terminology , typical for geometric invariant theory : 1 .
@xmath3 $ ] is unstable iff @xmath3\notin x(\tilde{\mu})$ ] , 2 .
@xmath3 $ ] is semistable iff @xmath3\in x(\tilde{\mu})$ ] . among semistable states we distinguish the class of stable states . by definition ,
a semistable state @xmath3 $ ] is stable if and only if @xmath229)=0 $ ] and @xmath230=\mathrm{dim}\tilde{k}$ ] .
note that since for @xmath3\in\tilde{\mu}^{-1}(0)$ ] one has @xmath231=2\mathrm{dim}\tilde{k}.[\phi]$ ] ( see ref.@xcite ) , the condition @xmath230=\mathrm{dim}\tilde{k}$ ] can be phrased as @xmath232=\mathrm{dim}\tilde{g}$ ] .
remarkably , the existence of a stable state implies that almost all semistable states are stable , in particular almost all states in @xmath233 are stable @xcite . note that since @xmath234 is open and dense in @xmath2 and a generic @xmath235-orbit in @xmath234 has dimension
@xmath236 we get @xmath237 one of the central results in the geometric invariant theory reads @xcite @xmath238 hence , under the assumption of stable states existence and using ( [ eq : relation ] ) and ( [ eq : categorical ] ) we get @xmath239 as we will see formula ( [ eq : dim - categorical ] ) plays a major role in showing ( [ eq : case3 ] ) .
let @xmath240 , where @xmath241 and @xmath242 .
we first consider the natural action of @xmath243 on the first @xmath33 qubits in @xmath2 , where @xmath1 .
the momentum map @xmath244 for this action gives the first @xmath33 reduced density matrices .
therefore , @xmath245 consists of all states with the first @xmath33 reduced density matrices maximally mixed , but no assumption is made on the spectra of the remaining @xmath246 matrices . in the following
we assume that there exists a stable state for @xmath243-action on @xmath2 ( see lemma [ lemma - stable ] for proof ) . under this assumption and using formula ( [ eq : dim - categorical ] ) the dimension of @xmath247 is @xmath248 recall that @xmath249 is a stratified symplectic space and we consider the highest dimensional stratum which is a symplectic manifold .
removing @xmath243 freedom does not affect @xmath250 action , i.e. the actions of @xmath251 and @xmath252 commute .
therefore , we can consider action of @xmath250 on the highest dimensional stratum of @xmath249 .
the momentum map @xmath253 for @xmath250 action on @xmath249 gives the remaining @xmath110 reduced density matrices .
moreover , using inequalities ( [ nier1 ] ) with @xmath254 it is straightforward to see that the image of the corresponding map @xmath255 is @xmath110 dimensional polytope . using fact [ dim - orbit ] and formula ( [ eq : fiber - interior-1 ] ) , for a point inside of this polytope , e.g. when @xmath256 , the dimension of @xmath255-fiber is @xmath257 but the @xmath255-fiber is exactly the reduced space we look for , i.e. the one which corresponds to @xmath254 and @xmath256 . therefore ,
as promised @xmath258 in order to complete the above reasoning we now show that an appropriate stable state indeed exists .
[ lemma - stable]let @xmath241 , @xmath259 .
if @xmath260 then the @xmath0-qubit state @xmath261=\left(|0\ldots0\rangle+|1\ldots1\rangle\right)+\left(|110\ldots0\rangle+|001\ldots1\rangle\right)+\left(|1010\ldots0\rangle+|0101\ldots1\rangle\right)+\\ + \ldots+\left(|10\ldots01\rangle+|01\ldots10\rangle\right),\end{gathered}\ ] ] is @xmath243-stable . for @xmath262 @xmath261=\alpha\left(|0000\rangle+|1111\rangle\right)+\left(|1100\rangle+|0011\rangle\right)+\left(|1010\rangle+|0101\rangle\right)+\\
+ \left(|1001\rangle+|0110\rangle\right),\ \alpha\in\mathbb{r}\setminus\{1,-3\},\end{gathered}\ ] ] is @xmath243-stable .
we need to show that the first @xmath33 reduced density matrices of @xmath3 $ ] are maximally mixed and that @xmath263=\mathrm{dim}k_{1}$ ] .
note first that if the state @xmath3 $ ] is stable with respect to @xmath5 action it is also stable with respect to @xmath264 action .
therefore , we will show that @xmath3 $ ] is @xmath6-stable .
the state @xmath3 $ ] consists of @xmath260 pairs of separable states . in each pair
the second vector is the first vector with the swap @xmath265 performed on every qubit .
the first pair is ghz state . in the remaining pairs
the first vector is such that the first and one additional qubits are in the excited state @xmath171 while the remaining @xmath266 qubits are in the ground state @xmath170 .
the construction ensures that @xmath118)=0 $ ] .
what is left is to calculate the dimension @xmath267 $ ] , where @xmath268 .
this is equivalent to calculating the dimension of the tangent space @xmath269}g.[\phi]$ ] which is generated by the action of lie algebra @xmath270 on @xmath3 $ ] .
more precisely let @xmath271 be the basis of @xmath272 .
let @xmath273 our goal is to show that vectors @xmath274 are linearly independent and orthogonal to @xmath120 .
denote by @xmath275 a separable state whose @xmath195-th qubit is in the excited state @xmath171 and remaining qubits are in the ground state @xmath170 ( e.g. a 5 - qubit state @xmath276 and a 6 - qubit state @xmath277 ) .
it is straightforward to verify that @xmath278 moreover , the action of @xmath279 gives vectors that can be obtained from the above set of vectors by preforming the swap operation on every qubit .
the vectors obtained by the action of @xmath280 are linearly independent , because each vector from this group contains a unique separable state of the form @xmath275 ( an analogous argument can be applied to the set of vectors obtained by the action of @xmath279 ) .
we next examine the linear independence of vectors from groups @xmath281 and @xmath282 .
first , note that @xmath283 are linearly independent from all vectors @xmath284 , because they consist of separable states of the form @xmath285 that do not appear in the vectors @xmath284 ( it is not true in the 4 - qubit case , but we will return to this problem later ) .
the last thing to show is the linear independence of the vector @xmath286 from the vectors @xmath287 .
note that this vector is orthogonal to @xmath288 .
assume that @xmath289 can be expressed as the linear combination of the remaining vectors @xmath290 we will show that this leads to a contradiction . to this end , let us calculate the sum @xmath291 using ( [ eq : linear ] ) we get @xmath292 on the other hand , @xmath293 thus , comparing the coefficients by @xmath294 and @xmath295 in equations ( [ eq : sum1 ] ) and ( [ eq : sum2 ] ) , we get @xmath296 this is a contradiction , because the first equation ( [ eq : cond1 ] ) is equivalent to @xmath297 , which can not be satistied by @xmath298 for all @xmath195 ( which is implied by the second equation ( [ eq : cond1 ] ) ) .
clearly for any @xmath299 we also have @xmath300 .
we are left with vectors @xmath301 .
it is straightforward to see that @xmath302 .
the matrix of coefficients for @xmath303 is given by @xmath304 in the basis @xmath305 . by direct calculation one
checks that @xmath306
. therefore the dimension of @xmath307 $ ] is equal to the dimension of @xmath308 , which is @xmath309 and @xmath120 is @xmath6-stable . in the case of 4 qubits
we need to consider a slightly different state @xmath261=\alpha\left(|0000\rangle+|1111\rangle\right)+\left(|1100\rangle+|0011\rangle\right)+\left(|1010\rangle+|0101\rangle\right)+\\ + \left(|1001\rangle+|0110\rangle\right),\ \alpha\in\mathbb{r}\setminus\{1,-3\}.\end{gathered}\ ] ] it can be shown by similar calculation that this state is @xmath6 - stable .
given spectra of one - qubit reduced density matrices , we found the formula for the minimal number of polynomials needed to decide lu equivalence of @xmath0-qubit pure states .
as we showed this number is the same for spectra belonging to the interior of the polytope @xmath15 .
this is not the case on the boundary where the behaviour of @xmath99 is not uniform . in particular , for a large part of the boundary of @xmath15 we have @xmath31 .
we also observed that for @xmath30 corresponding to the @xmath32)\right\ } _ { l=1}^{l}$ ] such that @xmath33 matrices are maximally mixed @xmath34 .
the methods used in this paper can be in principle applied to @xmath0-particle systems with an arbitrary finite - dimensional one - particle hilbert spaces . the argument for points in the interior of @xmath15
can be used _
mutatis mutandis _ in this case .
we note , however , that inequalities describing the polytope @xmath15 are much more complicated when @xmath310 ( see ref .
@xcite ) and therefore the problem for boundary points is of higher computational complexity .
nevertheless , one should expect that , similarly to the qubit case , there is a large part of the boundary characterized by @xmath311 .
we would like to thank marek ku for many discussions on the geometric aspects of quantum correlations and his continuous encouragement .
this work is supported by polish ministry of science and higher education iuventus plus grant no .
ip2011048471 .
99 bredon , g. e. , introduction to compact transformation groups , * 46 * , academic press , 1972 .
christandl , m. , doran , b. , kousidis , s. , walter , m. , eigenvalue distributions of reduced density matrices .
arxiv:1204.0741 , 2012 .
grnbaum , b. kaibel , volker ; klee , victor ; ziegler , gnter m. , eds .
, convex polytopes ( 2nd ed . ) , new york & london , 2003 .
guillemin , v. , sternberg , s. , convexity properties of the moment mapping , * 67 * , 491513 , 1982 .
heinzner , p. , huckleberry , a. , khlerian potentials and convexity properties of the moment map , * 126 * , 6584 , 1996 .
higuchi , a. , sudbery , a. , and szulc , j. , one - qubit reduced states of a pure many - qubit state : polygon inequalities , * 90 * , 107902 , 2003 .
kirwan , f. c. , convexity properties of the moment mapping , * 77 * , 547552 , 1984 kirwan , f. c. , cohomology of quotients in symplectic and algebraic geometry , * 31 * , princeton , nj : princeton university press , 1982 .
klyachko , a. quantum marginal problem and representation of the symmetry group , kraus , b. , local unitary equivalence of multipartite pure states , * 104*(2 ) , 020504 , 2010 .
kraus , b. , local unitary equivalence and entanglement of multipartite pure states , arxiv 1005.5295 , 2010 .
meinrenken , e. , woodward , c. , moduli spaces of flat connections on 2-manifolds , cobordism , and witten s volume formulas , , progr .
math . * 172 * , 271295 , 1999 .
molladavoudi , s. on the symplectic reduced space of three - qubit pure states , arxiv 1005.5295 , 2012 .
mumford , d. stability of projective varieties , , 1977 .
sawicki , a. , huckleberry , a. , ku , m. , symplectic geometry of entanglement , * 305 * , 441468 , 2011 .
sawicki , a. , ku , m. , geometry of the local equivalence of states , * 44 , * 495301 , 2011 .
sawicki , a. , walter , m. , ku , m. when is a pure state of three qubits determined by its single - partite reduced density matrices ?
, * 46 , * 055304 , 2013 .
walter , m. , doran , b. , gross , d. , christandl , m. , entanglement polytopes , arxiv:1208.0365 , 2012 .
in order to find vertices of the polytope @xmath15 , it is more convenient to view the inequalities describing the polytope in terms of the minimal eigenvalues of the reduced one - qubit density matrices ( remembering that the shifted spectra are given by @xmath312 ) . as proven in @xcite these inequalities
read : where each @xmath314 denotes the minimum eigenvalue of the reduced one - qubit density matrix describing @xmath315-th qubit , @xmath316 .
the general algorithm of finding the vertices of the polytope given by a set of inequalities is to choose the set of @xmath0 of the inequalities , write them as a set of equations and check whether there exists a unique solution @xcite .
if obtained solution satisfies the remaining @xmath317 inequalities then it defines a vertex of @xmath15 .
the following fact describes the structure of vertices of @xmath15 [ vertices]vertices of the polytope @xmath15 in the case of @xmath0 qubits are given by equations @xmath318 such that the number of indecies @xmath315 for which @xmath319 belongs to @xmath320 , that is , @xmath321 . first note that each vertex of this form can be obtained by picking @xmath322 equations from and @xmath323 linearly independent equations from .
one easily checks that in each case @xmath317 remaining inequalities are trivially satisfied .
thus , points described by conditions are indeed vertices of @xmath15 .
a single exceptional case occurs when @xmath324 , e.g. @xmath325 .
then by ( [ eq : ntrivineq ] ) we get @xmath326 , which is a contradiction
. we will now show that these are all solutions .
we achieve our goal by considering all remaining possibilities of choosing @xmath0 out of @xmath104 inequalities .
before we proceed let us introduce some useful notation .
any choice of @xmath0 out of @xmath104 inequalities ( and turning them into equalities ) is uniquely given when the following three auxiliary sets are specified : @xmath327 @xmath328 @xmath329 we have @xmath330 , where @xmath331 denotes the number of elements of the finite set @xmath54 .
we have already covered cases when @xmath332 .
what remains to be checked are three possibilities : @xmath333 or @xmath334 and the case when each @xmath335 is non - empty .
firstly , let us consider the case when @xmath334 .
if additionally @xmath333 , one can check by direct calculations that the set of @xmath0 equations from @xmath336 gives @xmath337 for all @xmath315 .
next , if the set @xmath338 is non - empty , i.e. we choose some number of @xmath314 s equal to zero and combine these conditions with the equations from @xmath336 , our problem either reduces to a problem analogous to the previous case with @xmath333 , or there exists such @xmath339 that reads @xmath340 with @xmath337 .
because all @xmath314 s are positive or equal to zero , we obtain in both cases that @xmath341 for all @xmath342 .
one of the last things to check is the case when the only empty set is @xmath338 , i.e. @xmath333 , @xmath343 and @xmath344 .
if , in addition @xmath345 , there exists such @xmath339 that reads @xmath346 .
this implies that @xmath347 , which is a contradiction .
now , if @xmath348 from the same equation we get that for all @xmath349 @xmath341 and @xmath350 . this is a contradiction , because we assumed that @xmath348 .
this argument remains also true in the case when all sets are non - empty . by the above fact
, the number of the vertices of the polytope @xmath15 for @xmath0 qubits is the number of ways to place @xmath33 out of @xmath0 @xmath314 s equal to @xmath96 and the remaining @xmath314s equal to zero on @xmath0 places : @xmath351 moreover , to find the @xmath182 dimensional faces of the polytope , one has to change one of the inequalities ( [ eq : ntrivineq ] ) , ( [ eq : trivineq1 ] ) , ( [ eq : trivineq2 ] ) into equality and find vertices that satisfy this condition ( the minimal number of vertices sufficient to span such face is @xmath0 ) . in this way , for @xmath352 qubits ,
one obtains that there are @xmath104 such faces .
the 3-dimensional faces can be divided into three groups obtained by : ( 1 ) changing one inequality from ( [ eq : ntrivineq ] ) into equality : the face is spanned by @xmath353 and three @xmath354 vertices , ( 2 ) choosing one @xmath355 : the face is spanned by three @xmath354 vertices , three @xmath356 vertices and @xmath357 , ( 3 ) choosing one @xmath358 : the face is spanned by @xmath353 , three @xmath354 vertices and one @xmath356 vertex . | given @xmath0-qubit states with the fixed spectra of reduced one - qubit density matrices , we find a formula for the minimal number of invariant polynomials needed for solving local unitary ( lu ) equivalence problem , that is , problem of deciding if two states can be connected by local unitary operations .
interestingly , this number is not the same for every collection of the spectra .
some spectra require less polynomials to solve lu equivalence problem than others .
the result is obtained using geometric methods , i.e. by calculating the dimensions of _ reduced spaces _ , stemming from the symplectic reduction procedure . |
as well known , the @xmath0 weak decay may give us abundant information about ckm elements , so that it stands as a complementary field to the meson decays .
these processes are also good probes for the factorization hypothesis which has been extensively explored for dealing with hadronic transitions @xcite .
recently many semi - leptonic and non - leptonic decays of @xmath0 are observed and measured @xcite , moreover the lhcb is expected to accumulate a large data - sample of b - hadrons to offer a unique opportunity for studying @xmath0 , thus we would like to investigate the @xmath0 weak decay more systematically . as for the @xmath0
decays the key is how to evaluate the form factors which parameterize the hadronic matrix elements .
there are many approaches advocated to this aspect @xcite . in our previous paper
@xcite we studied @xmath0 to @xmath4 weak decay in the light - front quark model @xcite and the results seem to be quite reasonable .
the light - front quark model is a relativistic quark model based on the light - front qcd @xcite .
the basic ingredient is the hadron light - front wave function which is explicitly lorentz - invariant .
the hadron spin is constructed using the melosh rotation .
the light - front approach has been widely applied to calculate various decay constants and form factors for the meson cases @xcite . in our earlier work
, we adopted the diquark picture for baryons @xcite which especially is well explored and proved to be a good approximation for such processes where the diquarks are not broken during the transition .
indeed , it has been known for a long time that two quarks in a color - antitriplet state attract each other and may form a correlated diquark @xcite . the diquark picture of baryons is considered to be appropriate for low momentum - transfer processes @xcite . concretely , under the diquark approximation , @xmath0 and @xmath4 are of the one - heavy - quark - one - light - diquark(ud ) structure which is analogous to the meson case . in this paper
we will apply these method to @xmath0 decaying into light hadrons such as proton or @xmath2 which is made of three light quarks .
these hadrons may also be regarded to possess quark - diquark structure @xcite .
some authors @xcite calculated the form factors of @xmath0 decaying into light baryons and the corresponding decay rates .
@xcite explored @xmath5 by using the method of pqcd and they concluded the perturbative analysis is reliable only for @xmath6 . in ref .
@xcite the branching ratio of @xmath7 in pqcd was evaluated ( @xmath8)@xcite , instead , cheng used the nonrelativistic quark model to obtain this branching ratio as @xmath9 which is lower than the experimental value ( @xmath10 ) . in a recent study ,
the authors of @xcite used the light - cone sum rules to calculate the @xmath11 transition form factors . in this work
, we study the form factors of @xmath12 and @xmath13 in the light - front model with the diquark picture , and then we calculate the rates of @xmath14 , @xmath15 , as well as several other non - leptonic decays of @xmath0 .
when @xmath16 decays into light baryons , the energy of the light baryon in the @xmath0 rest frame is @xmath17 which is much larger than its mass @xmath18 and the hadronic scale @xmath19 .
one important feature of this region is that the light hadrons move nearly along the light cone .
it is argued in @xcite that the active quark created from b quark by weak interaction carries most of the energy of the final light baryon . under the large energy limit ( leet @xcite ) and heavy quark limit ( hqet @xcite )
we can obtain the relations between@xmath20 and @xmath21 , which may help to achieve the orders of @xmath20 .
we write up these relations in section ii , and then derive the form factors ( @xmath22 and @xmath23 ) of @xmath24 and @xmath25 in section iii .
we carry out the numerical computations in section iv .
finally , section v is devoted to discussions by which we will draw our conclusion .
the form factors for the weak transition @xmath26 where @xmath27 represents a light baryon ( refers to @xmath28 in this study ) , are defined in the standard way as @xmath29 u_{\lambda_{b}}(p , s_z ) \nonumber \\ & & -\bar u_{h}(p',s'_z)\left[\gamma_{\mu } g_{1}(q^{2 } ) + i\sigma_{\mu \nu } \frac { q^{\nu}}{m_{\lambda_{b}}}g_{2}(q^{2})+ \frac{q_{\mu}}{m_{\lambda_{b}}}g_{3}(q^{2 } ) \right]\gamma_{5 } u_{\lambda_{b}}(p , s_z),\end{aligned}\ ] ] where @xmath30 , @xmath31 and @xmath32 denote heavy quark and light quark , @xmath27 stands as the light baryon , respectively .
the above formulation is the most general expression with only constraints of enforcing the lorentz invariance and parity conservation for strong interaction .
there are six form factors @xmath33 ( i=1,2,3 ) in total for the vector and axial vector current @xmath34 where the light - quark @xmath35 denotes @xmath36 for @xmath1 and @xmath37 for @xmath2 .
all the information about the strong interaction is involved in those form factors . since @xmath38
, we will be able to write @xmath39 as @xmath40 and similarly for @xmath41 in the following formulations .
another parametrization in terms of the four - velocities is widely used and is found to be convenient for the heavy - to - heavy transitions , such as @xmath42 .
but for the heavy - to - light transitions at the large recoil region where the energy of final light baryon @xmath27 is much larger than its mass , it is more convenient to use following formulation .
analogous to heavy quark symmetry in heavy - to - heavy case , there is a large energy symmetry relations for the heavy - to - light at large energy recoil @xcite . for the heavy - to - light baryon transition ,
the symmetry has not been searched up until present . in this subsection
, we explore the large - energy symmetry and show that they lead to a simplification of the form factors : the six form factors are reduced to three independent ones .
let us introduce the velocity @xmath43 of initial @xmath0 and a light front unit vector @xmath44 by v= , n= , where @xmath45 is the energy of @xmath27 .
using these vectors , the amplitude of the weak transition @xmath26 is parameterized by @xmath46 u_{\lambda_{b}}(v , s_z)- \nonumber \\ & & \bar u_{h}(n , s'_z)\left[g_1(e)\gamma_{\mu } + g_2(e)v_{\mu}+g_3(e)n_{\mu}\right ] \gamma_{5 } u_{\lambda_{b}}(v , s_z ) .
\end{aligned}\ ] ] up to leading order in @xmath47 the relation between the two parametrization schemes is @xmath48 where @xmath49 is the mass of @xmath0 .
we have neglected the mass of final light baryon compared to @xmath49 . under the large energy limit
, the light energetic quark @xmath35 is described by the two - component spinor @xmath50 where @xmath51 is another light front unit vector and the heavy quark is replaced by @xmath52 .
the weak current @xmath53 in the full qcd is matched onto the current @xmath54 in the effective theory at tree level .
for an arbitrary matrix @xmath55 , @xmath56 has only three independent dirac structures .
one convenient choice is discussed in @xcite : @xmath57 , @xmath58 and @xmath59 .
thus , we have |q^b&=&|_^ h_v+n^| h_v , + @xmath60 .
the three independent form factors are defined by h(p,s_z)|h_v_b(p , s_z ) & = & |u_h(n , s_z)u__b(v , s_z)_0(e ) , + h(p,s_z)|_5 h_v_b(p , s_z ) & = & |u_h(n , s_z)_5 u__b(v , s_z)_5(e ) , + h(p,s_z)| _ h_v_b(p , s_z ) & = & |u_h(n , s_z)_u__b(v , s_z)_(e ) .
then , we find & & f_1=g_1=_(e ) ; f_2=g_2=0 ; + & & f_3=_0(e)-_(e ) ; g_3=_(e)-_5(e ) . from the above equation
, we obtain the relations among the form factors : f_1+f_2=g_1-g_2 ; f_2=-f_3 ; g_2=-g_3
. this is one major result in this work .
the @xmath61 and @xmath62 are not independent , but related to @xmath63 and @xmath23 . in the diquark picture ,
the heavy baryon @xmath64 is composed of one heavy quark @xmath65 and a light diquark [ ud ] . in order to form a color singlet hadron
, the diquark [ ud ] is in a color anti - triplet . because @xmath64 is at the ground state ,
the diqaurk is a @xmath66 scalar ( @xmath67 , @xmath68 ) and the orbital angular momentum between the diquark and the heavy quark is also zero , i.e. @xmath69
. however the situation is complicated for light baryon even thought it is in the ground state .
the diquark in light baryon may be a @xmath66 scalar or a @xmath70 vector .
fortunately the diquark is a spectator in the concerned transition and its spin is not affected so that only the scalar diquark can transit into the final baryon and one only needs to consider the scalar diquark structure of the light baryon . in the light - front approach , the heavy baryon @xmath71 composed of only scalar diquark with total momentum @xmath72 and spin
@xmath73 can be written as @xmath74(p_2)\ra,\end{aligned}\ ] ] and the light baryon ( total momentum @xmath72 , spin @xmath75 , composed of @xmath66 scalar diquark and orbital angular momentum @xmath76 ) has the similar form , @xmath77(p_2)\rangle,\end{aligned}\ ] ] where @xmath31 , @xmath78 $ ] represent heavy quark and diquark respectively and @xmath79 denotes the helicity , where @xmath80 and @xmath81 are the color and flavor indices , @xmath82 , @xmath83 are the on - mass - shell light - front momenta defined by @xmath84 and @xmath85(p_2)\rangle= b^{\dagger}_{\lambda_1}(p_1)a^{\dagger}(p_2)| 0\ra,\non\\ & & [ a(p ' ) , a^{\dagger}(p)]=2(2\pi)^3\delta^3(\tilde{p}'-\tilde{p } ) , \nonumber\\ & & \{d_{\lambda'}(p'),d_{\lambda}^{\dagger}(p)\}= 2(2\pi)^3\delta^3(\tilde{p}'-\tilde{p})\delta_{\lambda'\lambda}.\end{aligned}\ ] ] the coefficient @xmath86 is a normalized color factor and @xmath87 is a normalized flavor coefficient , & & c_f^bcc_f^bc q^(p_1,_1)[q_b^q_c^](p_2)| q^(p_1,_1)[q_b^q_c^](p_2 ) + & & = 2 ^ 2(2)^6 ^ 3(_1-_1)^3 ( _ 2-_2)__1_1 , + & & c_f^abcc_f^abc
q_a^(p_1,_1)[q_b^q_c^](p_2)| q_a^(p_1,_1)[q_b^q_c^](p_2 ) + & & = 2 ^ 2(2)^6 ^ 3(_1-_1)^3 ( _ 2-_2)__1_1 . in order to describe the motion of the constituents
, one needs to introduce intrinsic variables @xmath88 with @xmath89 through @xmath90 where @xmath91 s are the light - front momentum fractions satisfying @xmath92 .
the variables @xmath88 are independent of the total momentum of the hadron and thus are lorentz - invariant .
the invariant mass square @xmath93 is defined as @xmath94 the invariant mass @xmath95 is in general different from the hadron mass @xmath96 which satisfies the physical mass - shell condition @xmath97 .
this is due to the fact that in the baryon , heavy quark and diquark can not be on their mass shells simultaneously .
we define the internal momenta as k_i=(k_i^-,k_i^+,k_i)=(e_i - k_iz , e_i+k_iz , k_i)= ( , x_im_0,k_i ) .
it is easy to obtain @xmath98 where @xmath99 denotes the energy of the i - th constituent .
the momenta @xmath100 and @xmath101 constitute a momentum vector @xmath102 and correspond to the components in the transverse and @xmath103 directions , respectively . in the momentum space , the function @xmath104 appearing in eq .
( [ eq : lfbaryon ] ) is expressed as @xmath105 where @xmath106 is the light - front wave function which describes the momentum distribution of the constituents in the bound state with @xmath107 ; and @xmath108 is the corresponding clebsch - gordan coefficient with total spin of the scalar diquark @xmath109 ; @xmath110 is the well - known melosh transformation matrix element which transforms the the conventional spin states in the instant form into the light - front helicity eigenstates , _
1|^_m(x_1,k_1,m_1 ) & = & , where @xmath111 denotes a dirac spinor in the light - front ( instant ) form and @xmath112 is a unit vector in the @xmath103 direction .
in practice , it is more convenient to use the covariant form for the melosh transform matrix @xcite @xmath113 where @xmath114 for the scalar diquark .
if the diquark is a vector which is usually supposed to be the case for the @xmath115 baryon , the melosh transform matrix should be modified ( since it is irrelevant to our present work , we omit the corresponding expressions ) .
the baryon state is normalized as ( p,s,s_z)|(p , s , s_z)=2(2)^3p^+ ^3(-)_ss_s_zs_z , the same for @xmath116 . thus , the light - front wave function obeys the constraint |(x , k_)|^2=1 . in principle
, the wave functions can be obtained by solving the light - front bound state equations .
however , it is too hard to calculate them based on the first principle , so that instead , we would like to adopt a phenomenological function , and obviously , a gaussian form is most preferable , ( x , k_)=n ( ) . with n=4()^3/4 ,
= . where @xmath117 determines the confinement scale .
the phenomenological parameters in the light - front quark model are quark masses and the hadron wave function parameter @xmath117 which should be prior determined before numerical computations can be carried out and we will do the job in the later subsections . equipped with the light - front quark model description of @xmath119 and @xmath120 , we can calculate the weak transition matrix elements @xmath121 where @xmath122 is a flavor - spin factor of i ( initial particle ) decaying into f ( final particle ) . following @xcite ,
the flavor - spin functions of @xmath0 , proton and @xmath2 take the forms in the diquark picture @xmath123 } , \nonumber\\ & & \chi_s^{p}=us_{[u , d]},\,\,\ , \chi_v^{p}=[uv_{[u , d]}-\sqrt{2}dv_{[u , u ] } ] /\sqrt{3 }
\nonumber\\ & & \chi_s^{\lambda}=[us_{[d , s]}-ds_{[u , s]}-2ss_{[u , d]}]/\sqrt{6},\,\,\,\,\ , \chi_v^{\lambda}=[uv_{[d , s]}-dv_{[u , s]}]/\sqrt{2 } \end{aligned}\ ] ] where s and v denote scalar and axial vector diquark .
we can get @xmath124 , which are consistent with @xcite , and & & |=_0_0==1 , + & & m_1=m_b , m_1=m_q , m_2=m_[ud ] . with @xmath72 and @xmath125 denoting the momenta of initial and final baryons , @xmath126 are the momenta of @xmath65 and @xmath127 quarks , respectively . because the diquark is a scalar , one does not need to deal with the spinors which make computations more complex . in this framework , at each effective vertex , only the three - momentum rather than the four - momentum is conserved , hence @xmath128 and @xmath129 . from @xmath129
, we have x=x , k_=k_+x_2q_. with @xmath130 , @xmath131 .
thus , eq . ( [ s2 ] )
is rewritten as @xmath132 following @xcite , we get the the final expressions for the @xmath133 weak transition form factors @xmath134 } { \sqrt{\left[\left(m_1+x_1m_0\right)^2+k_{2\perp}^2\right ] \left[\left(m'_1+x_1m'_0\right)^2+k_{2\perp}^{'2}\right ] } } , \non \\
g_1(q^2)&=&n_{if}\int{\frac{dxd^2k_{\perp}}{2(2\pi)^3 } } \frac{\phi_{h}(x',k'_{\perp } ) \phi_{\lambda_q}(x , k_{\perp})[-k_{2\perp } \cdot k'_{2\perp}+(x_1m_0+m_1)(x'_1m'_0+m'_1 ) ] } { \sqrt{\left[\left(m_1+x_1m_0\right)^2+k_{2\perp}^2\right ] \left[\left(m'_1+x_1m'_0\right)^2+k_{2\perp}^{'2}\right ] } } , \non \\
\frac{f_2(q^2)}{m_{\lambda_q}}&=&\frac{n_{if}}{q^i_{\perp } } \int{\frac{dxd^2k_{\perp}}{2(2\pi)^3 } } \frac{\phi_{h}(x',k'_{\perp } ) \phi_{\lambda_q}(x , k_{\perp } ) [ ( m_1+x_1m_0)k_{1\perp}^{\prime i}-(m'_1+x'_1m'_0)k_{1\perp}^i ] } { \sqrt{\left[\left(m_1+x_1m_0\right)^2+k_{2\perp}^2\right ] \left[\left(m'_1+x_1m'_0\right)^2+k_{2\perp}^{'2}\right ] } } , \non \\
\frac{g_2(q^2)}{m_{\lambda_q}}&=&\frac{n_{if}}{q^i_{\perp } } \int{\frac{dxd^2k_{\perp}}{2(2\pi)^3 } } \frac{\phi_{h}(x',k'_{\perp } ) \phi_{\lambda_q}(x , k_{\perp } ) [ ( m_1+x_1m_0)k_{1\perp}^{\prime i}+ ( m'_1+x'_1m'_0 ) { k}_{1\perp}^i]}{\sqrt{\left[\left(m_1+x_1m_0\right)^2 + k_{2\perp}^2\right]\left[\left(m'_1+x_1m'_0\right)^2 + k_{2\perp}^{'2}\right]}}.\non\\\end{aligned}\ ] ] it is noted that the form factors @xmath61 and @xmath62 can not be extracted in our method because we have imposed the condition @xmath135 .
the fact that the calculated @xmath63 and @xmath23 at @xmath136 are small compared to @xmath137 and @xmath138 and the large energy limit relations @xmath139 and @xmath140 show that using the large energy limit relations for @xmath61 and @xmath62 does not produce substantial theoretical errors .
in this section , we obtain formulations for the rates of semi - leptonic and non - leptonic processes . in this work ,
we concern only the exclusive decay modes .
generally the polarization effects may be important for testifying different theoretical models , so that we would pay more attention to the physical consequences brought up by them .
the transition amplitude of @xmath24 contains several independent helicity components . according to the definitions of the form factors for @xmath12 given in eq .
( [ s1 ] ) , the helicity amplitudes @xmath143 are related to these form factors through the following expressions @xcite h^v_,0&= & ( ( + ) f_1-f_2 ) , + h^v_,1&=&(-f_1 + f_2 ) , + h^a_,0&= & ( ( -)g_1+g_2 ) , + h^a_,1&=&(-g_1- g_2 ) , where @xmath144 .
the helicities of the @xmath145-boson @xmath146 can be either @xmath147 or @xmath148 , corresponding to the longitudinal and transverse polarizations .
following the definitions in literature , we decompose the decay width into a sum of the longitudinal and transverse parts according to the helicity states of the virtual w - boson .
the differential decay rate of @xmath142 is = + , and the longitudinally ( l ) and transversely ( t ) polarized rates are respectively @xcite & = & , + & = & . where @xmath149 is the momentum of the proton in the rest frame of @xmath0 .
the relations between @xmath150 and @xmath151 can be found in @xcite . integrating over the solid angle
, we obtain the decay rate as = _
1^_maxd , where the upper bound of the integration @xmath152 corresponds to the maximal recoil . in order to compare our results with those in the literatures
, we use the variable @xmath153 in the expression for the differential decay rate .
the polarization of the cascade decay @xmath154 is expressed by various asymmetry parameters @xcite . among them , the integrated longitudinal and transverse asymmetries are defined by a_l&= & , + a_t&=&. the ratio of the longitudinal to transverse decay rates @xmath155 is defined by r== , and the longitudinal proton polarization asymmetry @xmath156 is given as p_l&= & + & = & . from the theoretical aspects , the non - leptonic decays are much more complicated than the semi - leptonic ones because of the strong interaction .
generally , the present theoretical framework is based on the factorization assumption , where the hadronic matrix element is factorized into a product of two matrix elements of single currents .
one can be written as a decay constant while the other is expressed in terms of a few form factors according to the lorentz structure of the current . for the weak decays of mesons
, such factorization approach is verified to work very well for the color - allowed processes and the non - factorizable contributions are negligible . for the non - leptonic
decays @xmath158 , the effective interaction at the quark level is @xmath159 .
the relevant hamiltonian is & & h_w = v_ubv_q_1q_2^*(c_1o_1+c_2o_2 ) , + & & o_1=(|u b)_v - a ( |q_2q_1)_v - a , o_2=(|q_2b)_v - a ( |u q_1)_v - a . where @xmath160 denotes the short - distance wilson coefficient
, @xmath161 is the ckm matrix elements , @xmath162 stands for @xmath36 and @xmath163 for @xmath164 in the context . then one needs to evaluate the hadronic matrix elements pm | h_w | _
b= v_ubv_q_1q_2^*_i=1,2c_i pm | o_i | _ b. under the factorization approximation , the hadronic matrix element is reduced to pm |
o_i | _ b = p | j_|_b m | j^ | 0 . where @xmath165 is the @xmath166 weak current .
the first factor @xmath167 is parameterized by six form factors as done in eq . ( [ s1 ] ) .
the second factor defines the decay constants as follows [ p1 ] p(p)|a_|0&=&f_pp _ , + s(p)|v_|0&=&f_sp _ , + v(p,)|v_|0&=&f_vm_v^ * _ , + a(p,)|a_|0&=&f_vm_a^ * _ , where @xmath168 denotes a pseudoscalar ( vector ) meson , and @xmath169 denotes a scalar ( axial - vector ) meson . in the definitions , we omit a factor @xmath170 for the pseudoscalar meson decay constant . in general , the transition amplitude of @xmath171 can be written as [ p2 ] m(_bp p)&=&| u_p(a+b_5)u__b , + m(_bp v)&=&| u_p^*u__b , where @xmath172 is the polarization vector of the final vector or axial - vector mesons . including the effective wilson coefficient @xmath173
, the decay amplitudes under the factorization approximation are @xcite [ p3 ] a&=&f_p(-)f_1(m^2 ) , + b&=&f_p(+)g_1(m^2 ) , + a_1&=&-f_vm , + a_2&=&-2f_vm , + b_1&=&f_vm , + b_2&=&2f_vm , where @xmath174 and @xmath96 is the @xmath175 mass .
replacing @xmath72 , @xmath176 by @xmath177 and @xmath178 in the above expressions , one can easily obtain similar expressions for scalar and axial - vector mesons .
the decay rates of @xmath179 and up - down asymmetries are @xcite @xmath180 , \non\\ \alpha&=&-\frac{2\kappa{\rm re}(a^*b)}{|a|^2+\kappa^2|b|^2 } , \end{aligned}\ ] ] where @xmath181 is the proton momentum in the rest frame of @xmath0 and @xmath182 . for @xmath183 decays , the decay rates and up - down asymmetries are [ p5 ] & = & , + & = & , where @xmath45 is the energy of the vector ( axial vector ) meson , and @xmath184 theses decays proceed only via the internal w - emission . with the factorization assumption ,
the amplitude is a ( _ bm)= v_qbv_qs^*a_2 m| in general , we can use the same formula ( eqs.([p1])-([p6 ] ) ) to obtain the decay rates and up - down asymmetries of @xmath186 . note that
: ( 1 ) at this time @xmath79 is replaced by @xmath187 , ( 2 ) when q and @xmath188 are u and @xmath189 respectively , the final meson may be @xmath190 , @xmath191 or @xmath192 .
for the decay constants of @xmath190 , @xmath191 and @xmath192 , we have ^0||u__5u|0&=&f^u_^0p _ , + ||u__5u|0&=&f^u_p _ , where@xmath193 , @xmath194 and @xmath195 can be get form @xcite .
in this section we perform the numerical computations of the form factors for @xmath196 and @xmath197 , then using them we estimate the rates of @xmath198 , @xmath199 and @xmath200 where @xmath96 stands as various mesons . in our calculation , the quark masses of @xmath201 and @xmath202 are taken from @xcite ; @xmath203 is set to be 0.3 gev ; the mass of diquark [ ud ] , parameters @xmath204}$ ] , @xmath205}$ ] and @xmath206}$ ] are chosen from @xcite . the baryon masses @xmath207 gev , @xmath208 gev , @xmath209 gev come from @xcite .
the input parameters are collected in table [ t3 ] .
[ t3 ] .input parameters in lfqm ( in units of gev ) . [ cols="^,^,^,^,^,^,^",options="header " , ] from table [ tab : t8 ] we can find that there are some differences among the predictions by various theoretical approaches . in our calculation , the @xmath210 is nearly equal , however @xmath211 is bigger than @xmath212 in @xcite .
in this work , we carefully investigate the processes where a heavy baryon decays into a light baryon plus a lepton pair ( semi - leptonic decay ) or a meson ( non - leptonic decay ) in terms of the light - front - quark model(lfqm ) .
besides the regular input parameters such as the quark masses and well measured decay constants of various mesons , there is only one free parameter to be determined , that is @xmath117 in the light front wavefunction . in our earlier work @xcite , by fitting the data of the semi - leptonic decays @xmath213 , we obtained the values of @xmath214}$ ] .
similarly , we fix the values @xmath206}$ ] for proton and @xmath205}$ ] for @xmath2 .
our numerical results are shown in corresponding tables and some measurable quantities such as the up - down asymmetries are also evaluated
. a clear comparison of our prediction on the decay rate of @xmath215 with the results predicted by other models and as well as the experimental data is also explicitly presented .
one can notice that our result for @xmath3 is @xmath216 which is in good agreement with the data .
the success is not too surprising even though the model we adopt is much simplified .
definitely this value obtained in this work is closer to the cental value of measurement than the previous evaluations , but since there is a large uncertainty in the data , one still can not justify which model is more preferable than others because within two standard deviations , all the numerical results achieved with all the approaches listed in the table are consistent with data . the asymmetry parameter which may be important for determining the applicability of the adopted model ,
is estimated as @xmath217 , which is generally consistent with that obtained in other models and approaches .
of course the details , especially the branching ratios will be further tested by the more accurate experiments in the future .
besides the semi - leptonic decays , we also estimate the branching ratios of several non - leptonic decay modes which are listed in table viii .
recently the cdf collaboration@xcite has measured the branching ratios of @xmath218 and @xmath219 as @xmath220 and @xmath221 .
our prediction on @xmath222=(@xmath223 ) is consistent with the measurement of the cdf within one standard deviation , but for @xmath224 our value is @xmath225 , one order smaller than the data of the cdf collaboration . following the literature , in our calculation , we employ the factorization scheme where the emitted pseudoscalar meson ( @xmath175 or k ) is factorized out and described by the common - accepted form factor @xmath226 where @xmath227 and @xmath228 are the corresponding axial current , decay constant of meson m and its four - momentum respectively .
it is noticed that in the case @xmath218 , at the vertex @xmath229 the cabibbo - kabayashi - maskawa entry is approximately @xmath230 whereas for the case @xmath219 , the ckm entry is @xmath231 , thus comparing with @xmath218 , the amplitude of the process @xmath232 is suppressed by a factor @xmath233 .
thus besides a small difference between the final phase spaces of the two reactions , one can roughly estimate that @xmath234 , and this estimate is consistent with our numerical results .
therefore the smallness of @xmath224 seems reasonable .
however the data of cdf show completely different results that @xmath235 is anomalously larger than @xmath236 .
in fact , in our calculations on the non - leptonic decays , we only consider the contributions from the tree diagrams and neglect the penguin - loop effects .
for the mode of @xmath218 , the penguin contribution can be safely neglected compared to the tree level . however , for the mode of @xmath219 , the tree level contribution is suppressed by the cabibbo - kabayashi - maskawa entry @xmath237 while for the penguin diagram , the main contribution comes from the loop where top quark is the intermediate fermion . in the case
, the ckm entry would be @xmath238 which is almost two orders larger than @xmath237 .
thus even though there is a loop suppression of order @xmath239 , it is compensated by the much larger ckm entry .
this situation was discussed in @xcite where the authors used the pqcd method to carry out the calculations .
in fact , we make a rough estimation of the contribution from the top - penguin , and the result is almost five times larger than the contribution from the tree diagram given above
. however , from another aspect , when the penguin diagram is taken into account , the factorization is dubious .
that is why we do not include the loop contributions in this present work , but will make a detailed discussion in our coming paper . actually , even including the penguin contribution , the theoretically estimated branching ratio of @xmath224 is still below the data and obviously smaller than that of @xmath236 .
if this measurement is valid and approved by further experiments , it would be a new anomaly which may hint an unknown mechanism which dominates the transition or new physics beyond the standard model and @xmath235 . ] and it is also consistent with the result of @xcite .
the good agreement of our results on the semileptonic decays of @xmath0 to light baryon and several non - leptonic decay modes with data indicates the following points .
first , the diqaurk picture : as we know , two quarks in a color - anti - triplet attract each other and constitute a cooper - pair - like subject of spin 0 or 1 . however , until now , many theorists still doubt the justifiability of the diquark picture .
it is true that even though the diquark structure was raised almost as early as the birth of the quark model , its validity or reasonability of application is still in sharp dispute .
in fact , it should be rigorously testified by experiments .
we have argued that for some processes , the diquark picture may be more applicable than in the others .
actually , in our case , we can convince ourselves that the picture should apply . as aforementioned , diquark is only a spectator in the transitions which we concern in this work , therefore its inner structure may not affect the numerical results much .
secondly , the produced baryon is very relativistic , i.e. very close to the light - cone , generally the details of the inner structure of the spectator diquark may not be important , this interpretation is somehow similar to the parton picture which was conceived out by feynman and bjorken long time ago .
namely at very high energy collisions , the interaction among partons can be ignored at the leading order , thus in our case the interaction between the quark which undergoes a transition , and the spectator diquark should be weak and negligible .
third is that the small effects caused by the inner structure of the diquark may be partly included in the parameter @xmath117 of the light - front wavefuction .
the agreement with data indicates that the diquark picture and the light - front quark model indeed apply in the analysis of the heavy baryon transiting into a light one .
moreover , since we employ the factorization scheme to deal with the non - leptonic decays , we find that to some modes , it works well , but to some modes where loop contributions may dominate or just are comparable to the tree contributions , the scenario encounters serious challenges@xcite
. we further investigate the measurable polarization asymmetries . because the information on the polarization asymmetries may be more sensitive to the model adopted in the theoretical calculations than the decay width , accurate measurements
would discriminate various models and indicate how to improve the details of the models .
moreover , we also predict the rates and asymmetries of several similar modes of @xmath0 non - leptonic decays in the same model , and the results are listed in table iiiv of last section .
the numbers will be tested in the future .
fortunately , the high luminosity at lhcb can provide large database on @xmath0 and moreover , with great improvements of experimental facility and detection technique , we expect that more and more accurate measurements will be carried out in the near future and theorists will be able to further testify , improve , or even negate our present models .
indeed , the baryons are much more complicated than mesons , but careful studies on the processes where baryons are involved would be very beneficial for getting better insight into the hadron structure and underlying principles , especially the non - perturbative qcd effects including the factorization and plausibility of the diquark picture
. the lhcb will be an ideal place to do the job .
the work is supported by the national natural science foundation of china and the special grant for the ph.d program of the chinese education ministry , one of us ( ke ) would like to thank the tianjin university for financial support . c. albertus , e. hernandez and j. nieves , phys .
d * 71 * , 014012 ( 2005 ) ; h. shih , s. lee and h. li , phys .
d * 61 * , 114002 ( 2000 ) ; m. huang , h. jin , j. k@xmath242rner and c. liu , phys .
b * 629 * , 27 ( 2005 ) ; m. pervin , w. roberts and s. capstick , phys . rev .
c * 72 * , 035201 ( 2005 ) ; d. ebert , r. faustov and v. galkin , phys .
d * 73 * , 094002 ( 2006 ) ; f. cardarelli and s. simula , phys .
d * 60 * , 074018 ( 1999 ) .
m. terentev , sov .
* 24 * , 106 ( 1976 ) ; v. berestetsky and m. terentev , @xmath243 .
* 24 * , 547 ( 1976 ) ; * 25 * , 347 ( 1977 ) ; p. chung , f. coester , and w. polyzou , phys . lett . *
b 205 * , 545 ( 1988 ) . c. hwang and z. wei , j. phys .
g * 34 * , 687 ( 2007 ) ; c. d. lu , w. wang and z. t. wei , phys . rev .
d * 76 * , 014013 ( 2007 ) [ arxiv : hep - ph/0701265 ] ; z. t. wei , h. w. ke and x. f. yang , phys .
d * 80 * , 015022 ( 2009 ) [ arxiv:0905.3069 [ hep - ph ] ] ; h. w. ke , x. q. li and z. t. wei , arxiv:0907.5465 [ hep - ph ] .
h. dosch , m. jamin and b. stech , z. phys .
c * 42 * , 167 ( 1989 ) .
p. kroll , b. quadder and w. schweiger , nucl .
* b 316*,373 ( 1989 ) ; p. ball , h.g .
dosch , z. phys . * c 51 * , 445 1991 ; j. k@xmath242rner and p. kroll , phys . lett .
* b 293 * , 201 1992 ; r. jakob , p. kroll , m. schurmann and w. schweiger , z. phys .
* a 347 * , 109 ( 1993 ) ; j. bolz , p. kroll and j. k@xmath242rner , z. phys . * a 350 * , 145 ( 1994 ) ; j. k@xmath242rner and p. kroll , z. phys .
* c 57 * , 383 ( 1993 ) .
d. ebert , r. faustov and v. galkin , phys .
d 73 * , 094002 ( 2006 ) ; x. guo and t. muta , phys . rev .
* d 54 * , 4629 ( 1996 ) ; x. guo , a. thomas and a. williams , _ ibid _ * d 59 * , 116007 ( 1999 ) ; y. yu , h. ke , y. ding , x. guo , h. jin , x. li , p. shen and g. wang , commun . theor . phys . * 46 * , 1031 ( 2006 ) ; y. yu , h. ke , y. ding , x. guo , h. jin , x. li , p. shen and g. wang , arxiv : hep - ph/0611160 .
j. k@xmath245rner and m. kr@xmath246mer , phys .
b * 275 * , 495 ( 1992 ) ; p. bialas , j. k@xmath245rner , m. kr@xmath246mer , and k. zalewski , z. phys .
c * 57 * , 115 ( 1993 ) ; j. k@xmath242rner , m. kr@xmath246mer and d. pirjol , prog . part .
* 33 * , 787 ( 1994 ) .
d. becirevic and a. b. kaidalov , phys . lett .
b * 478 * , 417 ( 2000 ) [ arxiv : hep - ph/9904490 ] .
p. ball and r. zwicky , phys .
d * 71 * , 014015 ( 2005 ) [ arxiv : hep - ph/0406232 ] ; _ ibid _ * d71 * , 014029 ( 2005 ) [ arxiv : hep - ph/0412079 ] . | in this work we calculate the branching ratios of semi - leptonic and non - leptonic decays of @xmath0 into light baryons ( @xmath1 and @xmath2 ) , as well as the measurable asymmetries which appear in the processes , in the light front quark model ( lfqm ) . in the calculation
, we adopt the diquark picture and discuss the justifiability of applying the picture in our case .
our result on the branching ratio of @xmath3 is in good agreement with data .
more predictions are made in the same model and the results will be tested in the future experiments which will be conducted at lhcb and even ilc . |
here we shall demonstrate how to derive quantum pigeonhole paradoxes from ghz paradoxes obtained from qudit graph states corresponding to ghz graphs @xcite . a weighted graph is defined by a set @xmath111 of @xmath112 vertices and a set of edges specified by the adjacency matrix @xmath113 whose entry @xmath114 denotes the weight of the edge connecting two vertices @xmath31 .
we consider here only undirected graph without self loop so that the adjacent matrix is symmetric and has zero diagonal entries .
a ghz graph @xcite is defined to be a weighted graph @xmath115 satisfying @xmath116 and @xmath117 as a result @xmath118 must be even and @xmath119 with @xmath120 .
ghz graphs exist for all values of @xmath121 and even values of @xmath118 with some examples shown in fig.3 .
consider a system of @xmath112 qudits labeled with @xmath122 .
for each qudit @xmath123 we denote by @xmath124 its computational basis and by @xmath125 its generalized bit and phase shifts for which it holds @xmath126 and the commutation rule @xmath127 .
we denote @xmath128 the graph state @xmath129 for a given graph @xmath115 is defined by the @xmath4 common eigenstate of @xmath112 commuting unitary observables @xmath130 , i.e. , @xmath131 for all @xmath123 . for a ghz graph it holds @xmath132 based on which we can construct a ghz paradox as well as a state - independent parity proof of ks theorem as shown in table ii . , respectively .
thick black and red edges have weight @xmath133 and @xmath134 , respectively .
all possible ghz graphs on 3 and 4 vertices are shown in ( a ) and ( b ) , where @xmath135 , @xmath136 , and @xmath137 with @xmath138 . ] all unitary observables @xmath140 labeled with rows @xmath141 and columns @xmath142 in table ii have spectrum @xmath143 and observables in the same column or row commute with each other .
quantum mechanically , it is clear that we have an identity @xmath144 in noncontextual models @xmath145 observables @xmath140 assume predetermined noncontextual values @xmath146 and the above product with observables replaced by their realistic values , since @xmath147 , is equal to @xmath148 , a contradiction .
a quantum pigeonhole paradox about the trouble of putting @xmath112 pigeons into @xmath118 holes can be demonstrated as follows .
let the system of @xmath112 qudits be prepared in the common eigenstate @xmath149 of @xmath150 corresponding to eigenvalues @xmath151 and be post - selected into the common eigenstate @xmath152 of @xmath153 with outcomes @xmath154 .
outcomes @xmath155 and @xmath156 are not possible unless @xmath157 because otherwise the pre- and post - selected states will be orthogonal . given successful pre- and post - selection ,
if we had measured the observable @xmath158 in the intermediate stage we would obtain outcome @xmath159 for each @xmath123 because otherwise @xmath160 will vanish , where we have denoted by @xmath161 the projection to the eigenspace of @xmath162 corresponding to eigenvalue @xmath60 .
however if these values @xmath163 are measured in the context @xmath164 , whose outcomes are denoted by @xmath165 , or a possible configuration of @xmath112 pigeons in @xmath118 pigeonholes , then we should have @xmath166 , a contradiction .
in other words , the possible answers to @xmath112 questions about how @xmath112 pigeons are distributed in @xmath118 pigeonholes , namely , @xmath158 with @xmath123 , provided by measurement context @xmath167 are incompatible with every configuration of directly putting @xmath112 pigeons into @xmath118 levels , namely , the measurement context @xmath164 .
to demonstrate a paradox , the pre - selected state is not necessarily an entangled state and can also be chosen to be a product state .
for preparation we measure observables @xmath164 with outcomes denoted by @xmath168 , respectively , and for post - selection we still measure observables @xmath153 with outcomes @xmath169 . thus the pre - selected state @xmath98 is the common eigenstate of @xmath164 corresponding to eigenvalue @xmath170 while the post - selected state @xmath152 is the common eigenstate of @xmath153 corresponding to eigenvalue @xmath155 .
all outcomes are possible . in the intermediate stage , if we had measured observable @xmath171 then we would obtain outcome @xmath172 where @xmath173 since otherwise @xmath174 would vanish .
the dilemma lies in the fact that on the one hand the outcome for measuring observable @xmath175 should be @xmath176 , since @xmath177 , from the above arguments and on the other hand @xmath152 is an eigenstate of @xmath175 corresponding to eigenvalue @xmath178 . | a kind of paradoxical effects has been demonstrated that the pigeonhole principle , i.e. , if three pigeons are put in two pigeonholes then at least two pigeons must stay in the same hole , fails in certain quantum mechanical scenario .
here we shall show how to associate a proof of kochen - specker theorem with a quantum pigeonhole effect and vise versa , e.g. , from state - independent proofs of kochen - specker theorem some kind of state - independent quantum pigeonhole effects can be demonstrated . in particular , a state - independent version of the quantum cheshire cat , which can be rendered as a kind of quantum pigeonhole effect about the trouble of putting two pigeons in two pigeonholes , arises from peres - mermin s magic square proof of contextuality .
quantum theory confronts us with a reality that is radically different from a classical one because of its contextuality . roughly speaking
, quantum contextuality refers to the property that any realistic theory trying to complete quantum mechanics so as to avoid indeterministic measurement outcomes has to be contextual , i.e. , the predetermined outcomes of measuring an observable may depend on which set of compatible observables that might be measured along side .
this is a seminal no - go theorem proved by kochen and specker ( ks ) @xcite and independently by bell @xcite .
bell s nonlocality as revealed by , e.g. , the violation of bell s inequality @xcite or arguments without inequality @xcite , is a special form of contextuality enforced by space - like separation . in different scenarios
there are different proofs of kochen - specker theorem , which can be state - dependent @xcite , state - independent and deterministic @xcite , and statistical yet state - independent @xcite .
recently a kind of quantum pigeonhole effect @xcite was demonstrated that the pigeonhole principle , which states that if three pigeons are to be put into two pigeonholes then two of the pigeons must stay in the same hole , does not hold in some pre- and post - selection scenarios .
this effect is even promoted to be a principle to explore the nature of the quantum correlations .
consider a system of three qubits representing three pigeons in two pigeonholes , i.e. , two eigenstates @xmath0 of @xmath1 , where three pauli matrices and identity matrix are denoted simply by @xmath2 .
initially the system is prepared in the state @xmath3 , which is the common @xmath4 eigenstate of commuting observables @xmath5 . at the final stage , @xmath6 component of each qubit
is measured @xmath7 and only those outcomes with three + 1 are kept , i.e. , the common @xmath4 eigenstate @xmath8 is post - selected . in the intermediate stage between the preparation and post - selection , one asks what would happen if we had tested whether each pair of two qubits is in the same state or not .
this is equivalent to measure the observable @xmath9 on each pair of qubits @xmath10 and @xmath11 since the outcome @xmath4 or @xmath12 , which corresponds to the projectors @xmath13 , indicates that two qubits are in the same or different states , respectively , with respect to the the computational basis .
it was then argued that the detectors corresponding to @xmath14 would never fire because of the identities @xmath15 as a result the detector corresponding to @xmath16 would always fire for each pair of qubits @xmath17 , meaning that each pair of qubits would have stayed in different states , a violation to the pigeonhole principle .
the quantum pigeonhole effect described above is obviously a kind of pre- and post - selection paradox and , as pointed out by leifer and spekkens @xcite , any such kind of paradox is associated with a proof of quantum contextuality . in any noncontextual realistic model , according to kochen and specker @xcite , there is a so called ks value assignment of @xmath18 to all the rays , i.e. , rank-1 projections , in the relevant hilbert space satisfying the following three rules .
the _ noncontextuality rule _ states that a ray is assigned to value 0 or 1 regardless of which complete orthonormal bases it belongs to , the _ orthogonality rule _ states that two orthogonal rays can not be assigned to value 1 simultaneously , and the _ completeness rule _ states that within a complete basis orthonormal basis there is at least one ray that is assigned to value 1 .
a finite set of rays having no ks value assignment is a proof of the ks theorem , i.e. , a demonstration of quantum contextuality , and the proof is state - dependent if some states are assigned to value 1 _ a priori_. the absence of ks value assignments is sufficient for the nonexistence of a noncontexual model but not necessary .
there are also state - independent proofs admitting ks value assignments @xcite . in the above demonstration of quantum
pigeonhole effect , by denoting @xmath19 and @xmath20 , the set of all the relevant 34 pure states @xmath21 where @xmath22 denotes one of possible cyclic permutations of @xmath23 and @xmath24 , provides a state - dependent proof of ks theorem if @xmath25 and @xmath26 are assigned to value 1 .
in fact for any possible choice of @xmath22 and @xmath27 , both two rays @xmath28 have to be assigned to value 0 because of eq.([cd ] ) and the orthogonality rule . as a result
, either @xmath29 or @xmath30 has to be assigned to value 1 for any given @xmath27 and @xmath31 , according to the completeness rule . or equivalently , all three rank-2 projections @xmath32 must be assigned to value 1 . due to the pigeonhole principle , in each one of the eight computational bases @xmath33 at least two qubits are in the same states and therefore is orthogonal at least to one of the three subspaces with projections @xmath32 .
that is to say all eight computational bases have to be assigned to value 0 , which contradicts the completeness rule .
all three rules of ks value assignment are also assumed implicitly or explicitly in the demonstration of the quantum pigeonhole effect .
the orthogonality and completeness rules , which can be enforced by aharonov - bergmann - lebowitz rule @xcite for pre- and post - selection , have been used in the arguments against certain outcomes of the intermediate measurements .
the noncontextuality has also been assumed implicitly in the intermediate measurement of the observable @xmath34 for each pair of qubits , which appears in two different measurement contexts in order to extract a contradiction .
the first measurement context is given by @xmath35 which has been used to show that the detector @xmath36 would never fire , i.e. , @xmath34 would take value @xmath12 indication qubit @xmath10 and @xmath11 are in different states .
the second measurement context is given by @xmath37 whose eigenstates represent all possible configurations of three pigeons in two pigeonholes .
there is a contradiction only if the value of @xmath34 obtained in the first context is used in the second context .
it is clear that no paradox will be present if the assumption of noncontextuality is dropped .
the counterfactual measurements might yield contextual values . for other possible outcomes of the post - selection measurement , similar quantum pigeonhole effects can still be demonstrated @xcite . here
we shall remove further the dependency of the initial state so that we have a state - independent version of the quantum pigeonhole effect . on a system of three qubits
we at first measure the complete set @xmath5 with outcomes @xmath38 being arbitrary , and at the final stage we measure the complete set @xmath7 with outcomes @xmath39 being arbitrary . by the first measurement the system is prepared in the state @xmath40 and by the second measurement the system is post - selected into the state @xmath41 .
let us now examine what would happen if we had measured @xmath34 at the intermediate stage between the preparation and post - selection .
denoting @xmath42 we have @xmath43 for arbitrary @xmath17 .
that is , the detector @xmath16 ( or @xmath36 ) would never fire if @xmath44 ( or 1 ) so that qubits @xmath10 and @xmath11 would have stayed in the same ( or different ) states , respectively . since @xmath45 there is an even number of pairs of qubits such that @xmath44 , i.e. , among three qubits there is an even number of pairs of qubits that would have stayed in the same state .
this is impossible due to the fact that to put three pigeons into two pigeonholes the number of pairs of pigeons staying in the different holes must be even .
the above state - independent version of the quantum pigeonhole paradox also gives rise to a state - independent proof of kochen - specker theorem due to waegell and aravind @xcite .
this proof includes 48 rays defined by eigenstates of 6 maximal sets of mutually commuting observables @xmath46 with @xmath47 .
there is an elegant parity proof given by the waegell - aravind s triangle configuration as shown in fig.1 .
quantum mechanically , the product of three observables connected by each thin ( blue ) straight line is 1 while the product of three observables connected by each thick ( red ) line is @xmath12
. thus the product of all those line products of three observables equals @xmath12 . in a noncontextual realistic theory , where all observables have some predetermined values @xmath48 independent of the contexts , the same product yields value 1 since each observable appears twice in the product .
any proof of kochen - specker theorem via logical contradiction can give rise to the demonstration of some sort of quantum pigeon effect .
this is because in each ks proof there is at least a basis , one of which can be taken without loss of generality to be the computational basis .
the logical contradiction can be pushed back to the impossibility of the ks value assignment to this computational basis .
we note that the computational bases represent all possible configurations of pigeons in holes and satisfy obviously the orthogonality and completeness rules , i.e. , one and only one configuration is realized .
thus the contradiction as seen in a ks proof can thus be regarded as a violation of some kind of generalized pigeonhole principle .
for example the 3-box paradox @xcite , which is associated with clifton s state - dependent proof of ks theorem @xcite , roughly shows that a single pigeon , when put into three pigeonholes , can stay simultaneously in two holes .
now let us see some more examples .
as the first example , some kind of robust quantum pigeonhole effect can be demonstrated by using peres - mermin s magic square for three qubits @xcite as shown in table i ( left ) .
let the system be prepared in any state @xmath25 ( may even be mixed ) in the common @xmath4 eigenspace of three observables @xmath49 in the fist row , which is spanned by @xmath50 .
at the final stage we make a post - selection to any state @xmath26 in the common + 1 eigenspace of three observables @xmath51 in the third row , which is spanned by @xmath52 . at the immediate stage
we ask what would happen if we had measured three observables @xmath53 in the second row , testing whether each pair of qubits was in the same state or not . in this
general setting eq.([cd ] ) still hold so that each pair of qubits would have stayed in different states , a violation to the pigeonhole principle .
the dependency of the pre- and post - selection can be easily removed and the generalization of peres - mermin s magic square as well as the quantum pigeonhole effect to the case of an odd number of qubits is straightforward .
@xmath54 as the second example , a kind of paradoxical effect can be demonstrated in the case of two pigeons in two pigeonholes , which turns out to be the quantum cheshire cat @xcite , a curious situation of a grin without a cat encountered by alice only in wonderland and demonstrated recently in a neutron experiment @xcite .
consider a spin half particle , e.g. , neutron , in an interferometer with two paths , where the cat is represented by the particle and its grin by its spin .
this is effectively a two - qubit system with the first qubit representing the spatial degree of freedom , i.e. , the path , while the second qubit representing its spin .
let the system be prepared initially in the state @xmath55 and post - selected to @xmath56 at the final stage . because @xmath57 the particle would take path @xmath58 if we had observed its path , where we have denoted by @xmath59 the projection to the eigenspace of observable @xmath60 corresponding to eigenvalue @xmath61 with @xmath62 . because @xmath63 if we had observed its spin components along @xmath64 direction the answer would be spin up @xmath65 . because @xmath66 we see that the path and the spin are anti - correlated , i.e.
, if the spin is up @xmath65 then the path @xmath67 should be taken while if the spin is down @xmath68 then the path @xmath58 should be taken . to sum up ,
given the pre and post - selection , the particle would surely travel along path @xmath69 and its spin would be up along @xmath64 direction while the path and spin would be anti - correlated , i.e. , its spin would be measured along a different path @xmath70 from what is taken by the particle , i.e. , @xmath69 , which is exactly our quantum cheshire cat .
this demonstration of quantum cheshire cat , being a kind of pre- and post - selection paradox , also gives rise to a state - dependent proof of kochen - specker theorem including the eigenstates of @xmath71 , @xmath72 and @xmath73 in addition to @xmath25 and @xmath26 . by considering suitable pre- and post - selection measurements the dependency of the quantum cheshire cat on the preparation and post - selection
can be removed . for the preparation we measure observables @xmath74 while for the post - selection we measure observables @xmath75 .
thus initially the system is prepared in the state @xmath76 which is the common eigenstate of @xmath77 and @xmath78 corresponding to eigenvalues @xmath79 and @xmath80 , respectively , with @xmath81 , and post - selected into the state @xmath82 with @xmath83 .
the following three observations constitute the quantum cheshire cat .
firstly , if we had observed which path were taken by the particle we would find the particle in the path @xmath84 with @xmath85 since @xmath86 vanishes otherwise . secondly , if we had observed the spin along @xmath64 direction we would find the spin in the state @xmath87 with @xmath88 since @xmath89 vanishes otherwise .
lastly , if we had observed the correlation @xmath90 then the detector @xmath91 would never fire if @xmath92 because in this case @xmath93 vanishes .
this means that the spin state @xmath87 , as revealed by the measurement @xmath94 , has to be correlated with the path @xmath95 which is different from the path @xmath96 taken by the particle as revealed by the measurement @xmath97 .
this state - independent version of the quantum cheshire cat turns out to be exactly the parity proof by peres - mermin s magic square @xcite as shown in table i ( right ) .
three commuting observables in three columns are taken as the post - selection , intermediate , and preparation measurements respectively .
if we take three sets of row observables instead then we obtain a paradoxical effect that it is impossible to put two pigeons in two holes in a pre- and post - selection scenario .
as the last example , let us see how to derive pigeonhole effects from greenberger - horne - zeilinger ( ghz ) paradoxes @xcite .
consider mermin s magic pentagram proof of ks theorem for three qubits as shown in fig.2 .
let the system be prepared in the ghz state @xmath98 , i.e. , one of the common eigenstates of three commuting observables @xmath99 , with @xmath22 being one of three cyclic permutations of @xmath23 , corresponding to eigenvalues @xmath100 respectively .
as the post - selection we measure observables @xmath5 with outcomes denoted by @xmath101 , i.e. , the post - selection state is @xmath102 .
we note that outcomes must satisfy @xmath103 where @xmath104 and @xmath105 since otherwise we will have orthogonal pre- and post - selected states @xmath106 for each successful pre- and post - selection , if we had measured observables @xmath34 for @xmath17 then we would register outcome @xmath107 because otherwise @xmath108 would vanish . since @xmath109 indicates
whether two qubits are in the same state ( computational basis ) or not , from @xmath110 it follows that there is an odd number of qubit pairs in different state , which is impossible because if we put three pigeons in two holes the number of pairs of pigeons in different state should be even .
this is a state - independent version of an early proposal of the pigeonhole effect @xcite using entangled states .
ghz paradoxes for multi ( even ) levels systems and multi particles can be systematically constructed from the so called ghz graph @xcite and corresponding graph states .
each paradox will give rise to a similar quantum pigeonhole effect as above ( see appendix ) .
to summarize , we have established an intimate relation between the quantum pigeonhole effects and proofs of ks theorem . on the one hand
this relation enables us to understand those paradoxical effects in terms of contextuality , since the assumption of noncontextuality is indispensable in all these paradoxical effects . on the other hand
this relation helps to obtain from known proofs of ks theorem some other interesting effects , such as state - independent versions and the impossibility of putting two pigeons in more than two pigeonholes .
most interestingly , the latter effect is related to the quantum cheshire cat , for which we also find a state - independent version with the help of peres - mermin s magic square . from the experimental point of view
, the removal of the dependency on the post - selection states enhances the success probability by 300% .
lastly , it will be interesting to find out what kinds of paradoxical effects correspond to those ks proofs without logical contradictions .
we are indebted to m. howard for pointing out a mistake in previous version and l. vaidman for providing a helpful reference @xcite .
this work is funded by the singapore ministry of education ( partly through the academic research fund tier 3 moe2012-t3 - 1 - 009 ) and the national research foundation , singapore ( grant no .
wbs : r-710 - 000 - 008 - 271 ) .
99 s. kochen and e.p .
specker , j. math . mech . * 17 * , 59 ( 1967 ) .
j.s . bell , rev .
mod .
phys .
* 38 * , 447 ( 1966 ) .
j.s . bell , physics ( n.y . ) * 1 * , 195 ( 1964 ) .
j.f .
clauser , m.a .
horne , a. shimony , and r.a .
holt , phys .
rev .
lett .
* 23 * , 880 ( 1969 ) .
d. m. greenberger , m.a .
horne , a. shimony , and a. zeilinger , am . j. phys .
58 , 1131 ( 1990 ) .
n. d. mermin , phys .
rev .
lett . * 65 * , 3373 ( 1990 ) .
w. tang , s. yu , and c.h .
oh , phys . rev .
lett . * 110 * , 100403 ( 2013 ) .
l. hardy , phys . rev
. lett . * 68 * , 2981 ( 1992 ) ; l. hardy , phys .
rev .
lett . * 71 * , 1665 ( 1993 ) .
r. clifton , am .
j. phys .
* 61 * , 443 ( 1993 ) .
a.a .
klyachko , m.a .
can , s. binicioglu , and a.s .
shumovsky , phys .
rev .
lett .
* 101 * , 020403 ( 2008 ) .
a. cabello , j.m .
estebaranz , g. garca - alcaine , phys .
lett .
a * 212 * , 183 ( 1996 ) .
a. peres , j. phys .
a * 24 * , l175 ( 1991 ) .
s. yu and c.h .
oh , phys .
rev .
lett . * 108 * , 030402 ( 2012 ) ; s. yu and c.h .
oh , arxiv : 1112.5513 . y. aharonov , f. colombo , s. popescu , i. sabadini , d.c .
struppa , and j. tollaksen , arxiv : 1407.3194 y. aharonov , s. nussinov , s. popescu , and l. vaidman , phys .
rev .
a * 87 * , 014105 ( 2012 ) .
m. s. leifer and r. w. spekkens , phys .
rev .
lett . * 95 * , 200405 ( 2005 ) .
y. aharonov , p.g .
bergmann , and j.l .
lebowitz , phys .
rev . * 134 * , b1410 ( 1964 ) .
m. waegell and p.k .
aravind , j. phys . a : math
. theor . * 45 * , 405301 ( 2012 ) .
y. aharonov and l. vaidman , j. phys .
a : math . theor . * 24 * , 2315 ( 1991 ) .
y. aharonov , s. popescu , d. rohrlich , and p. skrzypczyk , n. j. phys .
* 15 * , 113015 ( 2013 ) .
t. denkmayr , etal , nature .
comm . * 5 * , 4492 ( 2014 ) .
a. peres , phys .
lett .
a * 151 * , 107 ( 1990 ) ; n.d .
mermin , phys .
rev .
lett . *
65 * , 3373 ( 1990 ) . |
we consider the iterative solution of large - scale discrete ill - posed problems @xmath1 where the norm @xmath2 is the 2-norm of a vector or matrix , and the matrix @xmath3 is extremely ill conditioned with its singular values decaying gradually to zero without a noticeable gap .
this kind of problem arises in many science and engineering areas , such as signal processing and image restoration , typically when discretizing fredholm integral equations of the first - kind @xcite .
in particular , the right - hand side @xmath4 is affected by noise , caused by measurement or discretization errors , i.e. , @xmath5 where @xmath6 represents the gaussian white noise vector and @xmath7 denotes the noise - free right - hand side , and it is supposed that @xmath8 . because of the presence of noise @xmath9 in @xmath4 and the ill - conditioning of @xmath3 , the naive solution @xmath10 of is meaningless and far from the true solution @xmath11 , where the superscript @xmath12 denotes the moore - penrose generalized inverse of a matrix .
therefore , it is necessary to use regularization to determine a best possible approximation to @xmath11 @xcite .
the solution of can be analyzed by the svd of @xmath3 : @xmath13 where @xmath14 and @xmath15 are orthogonal matrices , and the entries of the diagonal matrix @xmath16 are the singular values of @xmath3 , which are assumed to be simple throughout the paper and labelled in decreasing order @xmath17 . with , we obtain @xmath18 throughout the paper , we assume that @xmath19 satisfies the discrete picard condition : on average , the coefficients @xmath20 decay faster than the singular values . to be definitive , for simplicity we assume that these coefficients satisfy a widely used model in the literature , e.g. , ( * ? ?
* and 153 ) and @xcite : @xmath21 let @xmath22 be the transition point such that @xmath23 which can be written as @xmath24 , the solution of the modified problem that replaces @xmath3 by its best rank @xmath0 approximation @xmath25 in ( [ eq1 ] ) , where @xmath26 , @xmath27 and @xmath28 .
remarkably , @xmath29 is the minimum - norm least squares solution of the perturbed problem that replaces @xmath3 in by its best rank @xmath0 approximation @xmath30 , and the best possible tsvd solution of by the tsvd method is @xmath31 @xcite .
a number of approaches have been proposed for determining @xmath22 , such as discrepancy principle , discrete l - curve and generalized cross validation ; see , e.g. , @xcite for comparisons of the classical and new ones . in our numerical experiments
, we use the l - curve criterion in the tsvd method and hybrid lsqr .
the tsvd method has been widely studied ; see , e.g. , @xcite . for a small and moderate , the tsvd method has been used as a general - purpose reliable and efficient numerical method for solving . as a result
, we will take the tsvd solution @xmath31 as a standard reference when assessing the regularizing effects of iterative solvers and accuracy of iterates under consideration in this paper . as well known , it is generally not feasible to compute svd when is large . in this case ,
one typically projects onto a sequence of low dimensional krylov subspaces and gets a sequence of iterative solutions @xcite .
the conjugate gradient ( cg ) method has been used when @xmath3 is symmetric definite @xcite .
as a cg - type method applied to the semidefinite linear system @xmath32 or the normal equations system @xmath33 , the cgls algorithm has been studied ; see @xcite and the references therein .
the lsqr algorithm @xcite , which is mathematically equivalent to cgls , has attracted great attention , and is known to have regularizing effects and exhibits semi - convergence ( see @xcite , @xcite , and also @xcite ) : the iterates tend to be better and better approximations to the exact solution @xmath34 and their norms increase slowly and the residual norms decrease . in later stages , however , the noise @xmath9 starts to deteriorate the iterates , so that they will start to diverge from @xmath34 and instead converge to the naive solution @xmath35 , while their norms increase considerably and the residual norms stabilize .
such phenomenon is due to the fact that a projected problem inherits the ill - conditioning of .
that is , as the iterations proceed , the noise progressively enters the solution subspace , so that a small singular value of the projected problem appears and the regularized solution is deteriorated .
as far as an iterative solver for solving is concerned , a central problem is whether or not a pure iterative solver has already obtained a best possible regularized solution at semi - convergence , namely whether or not the regularized solution at semi - convergence is at least as accurate as @xmath31 .
as it appears , for krylov subspace based iterative solvers , their regularizing effects critically rely on how well the underlying @xmath0-dimensional krylov subspace captures the @xmath0-dimensional dominant right singular subspace of @xmath3 . the richer information the krylov subspace contains on the @xmath0-dimensional dominant right singular subspace , the less possible a small ritz value of the resulting projected problem appears and thus the better regularizing effects the solver has .
to precisely describe the regularizing effects of an iterative solver , we introduce the term of _ full _ or _ partial _ regularization : if the iterative solver itself computes a best possible regularized solution at semi - convergence , it is said to have the full regularization ; in this case , no additional regularization is needed . here , as defined in the abstract , a best possible regularized solution means that it is at least as accurate as the best regularized solution obtained by the truncated singular value decomposition ( tsvd ) method .
otherwise , it is said to have the partial regularization ; in this case , in order to compute a best possible regularized solution , its hybrid variant , e.g. , a hybrid lsqr , is needed that combines the solver with additional regularization @xcite , which aims to remove the effects of small ritz values , and expand the krylov subspace until it captures all the dominant svd components needed and the method obtains a best possible regularized solution .
the study of the regularizing effects of lsqr and cgls has been receiving intensive attention for years ; see @xcite and the references therein .
however , there has yet been no definitive result or assertion on their full or partial regularization .
to proceed , we need the following definition of the degree of ill - posedness , which follows hofmann s book @xcite and has been commonly used in the literature , e.g. , @xcite : if there exists a positive real number @xmath36 such that the singular values satisfy @xmath37 , then the problem is termed as mildly or moderately ill - posed if @xmath38 or @xmath39 ; if @xmath40 with @xmath41 considerably , @xmath42 , then the problem is termed severely ill - posed .
it is clear that the singular values @xmath43 of a severely ill - posed problem decay exponentially at the same rate @xmath44 , while those of a moderately or mildly ill - posed problem decay more and more slowly at the decreasing rate @xmath45 approaching one with increasing @xmath46 , which , for the same @xmath46 , is smaller for the moderately ill - posed problem than it for the mildly ill - posed problem .
other minimum - residual methods have also gained attention for solving . for problems with @xmath3 symmetric , minres and its preferred variant mr - ii
are alternatives and have been shown to have regularizing effects @xcite .
when @xmath3 is nonsymmetric and multiplication with @xmath47 is difficult or impractical to compute , gmres and its preferred variant rrgmres are candidates @xcite .
the hybrid approach based on the arnoldi process was first introduced in @xcite , and has been studied in @xcite .
recently , gazzola _
et al_. @xcite have studied more methods based on the lanczos bidiagonalization , the arnoldi process and the nonsymmetric lanczos process for the severely ill - posed problem .
they have described a general framework of the hybrid methods and present krylov - tikhonov methods with different parameter choice strategies employed . in this paper , we focus on lsqr .
we derive bounds for the 2-norm distance between the underlying @xmath0-dimensional krylov subspace and the @xmath0-dimensional right singular space .
there has been no rigorous and quantitative result on the distance before .
the results indicate that the @xmath0-dimensional krylov subspace captures the @xmath0-dimensional dominant right singular space better for severely and moderately ill - posed problems than for mildly ill - posed problems . as a result
, lsqr has better regularizing effects for the first two kinds of problems than for the third kind . by the bounds and the analysis on them , we draw a definitive conclusion that lsqr generally has only the partial regularization for mildly ill - posed problems , so that a hybrid lsqr with additional explicit regularization is needed to compute a best possible regularized solution .
we also use the bounds to derive an estimate for the accuracy of the rank @xmath0 approximation , generated by lanczos bidiagonalization , to @xmath3 , which is closely related to the regularization of lsqr .
our results help to further understand the regularization of lsqr , though they appear less sharp .
in addition , we derive a bound on the diagonal entries of the bidiagonal matrices generated by the lanczos bidigonalization process , showing how fast they decay .
numerical experiments confirm our theory that lsqr has only the partial regularization for mildly ill - posed problems and a hybrid lsqr is needed to compute best possible regularized solutions .
strikingly , the experiments demonstrate that lsqr has the full regularization for severely and moderately ill - posed problems .
our theory gives a partial support for the observed general phenomena . throughout the paper ,
all the computation is assumed in exact arithmetic .
since cgls is mathematically equivalent to lsqr , all the assertions on lsqr apply to cgls .
this paper is organized as follows . in section [ sectionmain ]
, we describe the lsqr algorithm , and then present our theoretical results on lsqr with a detailed analysis . in section [ sectionexp ] ,
we report numerical experiments to justify the partial regularization of lsqr for mildly ill - posed problems .
we also report some definitive and general phenomena observed .
finally , we conclude the paper in section [ sectioncon ] . throughout the paper
, we denote by @xmath48 the @xmath0-dimensional krylov subspace generated by the matrix @xmath49 and the vector @xmath50 , by @xmath51 the frobenius norm of a matrix , and by @xmath52 the identity matrix with order clear from the context .
lsqr for solving is based on the lanczos bidiagonalization process , which starts with @xmath53 and , at step ( iteration ) @xmath0 , computes two orthonormal bases @xmath54 and @xmath55 of the krylov subspaces @xmath56 and @xmath57 , respectively .
define the matrices @xmath58 and @xmath59 .
then the @xmath0-step lanczos bidiagonalization can be written in the matrix form @xmath60 where @xmath61 denotes the @xmath62-th canonical basis vector of @xmath63 and the quantities @xmath64 and @xmath65 denote the diagonal and subdiagonal elements of the @xmath66 lower bidiagonal matrix @xmath67 , respectively . at iteration @xmath0
, lsqr computes the solution @xmath68 with @xmath69 note that @xmath70 .
we get @xmath71 as stated in the introduction , lsqr exhibits semi - convergence at some iteration : the iterates @xmath72 become better approximations to @xmath34 until some iteration @xmath0 , and the noise will dominate the @xmath72 after that iteration .
the iteration number @xmath0 plays the role of the regularization parameter .
however , semi - convergence does not necessarily mean that lsqr finds a best possible regularized solution as @xmath67 may become ill - conditioned before @xmath73 but @xmath72 does not yet contain all the needed @xmath22 dominant svd components of @xmath3 . in this case , in order to get a best possible regularized solution
, one has to use a hybrid lsqr method , as described in the introduction .
the significance of is that the lsqr iterates can be interpreted as the minimum - norm least squares solutions of the perturbed problems that replace @xmath3 in by its rank @xmath0 approximations @xmath74 , whose nonzero singular values are just those of @xmath67 .
if the singular values of @xmath67 approximate the @xmath0 large singular values of @xmath3 in natural order for @xmath75 , then lsqr must have the full regularization , and the regularized solution @xmath76 is best possible and is as comparably accurate as the best possible regularized solution @xmath31 by the tsvd method .
hansen s analysis @xcite shows that the lsqr iterates have the filtered svd expansions : @xmath77 where @xmath78 , and @xmath79 are the singular values of @xmath67 . in our context , if we have @xmath80 for some @xmath73 , the factors @xmath81 , @xmath82 are not small , meaning that @xmath72 is already deteriorated and becomes a poorer regularized solution , namely , lsqr surely does not have full regularization . as a matter of fact , in terms of the best possible solution @xmath31
, it is easily justified that the full regularization of lsqr is equivalent to requiring that the singular values of @xmath67 approximate the @xmath0 largest singular values of @xmath3 in natural order for @xmath75 , so it is impossible to have @xmath80 for @xmath73 .
the regularizing effects of lsqr critically depend on what @xmath83 mainly contains and provides .
note that the eigenpairs of @xmath84 are the squares of singular values and right singular vectors of @xmath3 , and the tridiagonal matrix @xmath85 is the projected matrix of @xmath84 onto the subspace @xmath83 , which is obtained by applying the symmetric lanczos tridiagonalization process to @xmath84 starting with @xmath86 @xcite .
we have a general claim deduced from @xcite and exploited widely in @xcite : the more information the subspace @xmath83 contains on the @xmath0 dominant right singular vectors , the more possible and accurate the @xmath0 ritz values approximate the @xmath0 largest singular values of @xmath3 ; on the other hand , the less information it contains on the other @xmath87 right singular vectors , the less accurate a small ritz value is if it appears . for our problem , since the small singular values of @xmath3 are clustered and close to zero , it is expected that a small ritz value will show up as @xmath0 grows large , and it starts to appear more late when @xmath83 contains less information on the other @xmath87 right singular vectors . in this sense , we say that lsqr has better regularizing effects since @xmath72 contains more dominant svd components . using the definition of canonical angles @xmath88 between the two subspaces @xmath89 and @xmath90 of the same dimension @xcite , we have the following theorem , which shows how well the subspace @xmath56 , on which lsqr and cgls work , captures the @xmath0-dimensional dominant right singular space . [ thm2 ]
let the svd of @xmath3 be , and assume that its singular values are distinct and satisfy @xmath40 with @xmath41 .
let @xmath91 be the subspace spanned by the columns of @xmath92 , and @xmath93 .
then @xmath94 with the @xmath95 matrix @xmath96 to be defined by and @xmath97 _ proof_. let @xmath98 consist of the first @xmath99 columns of @xmath100 defined in .
we see @xmath101 is spanned by the columns of the @xmath102 matrix @xmath103 with @xmath104 partition the matrices @xmath105 and @xmath106 as follows : @xmath107 where @xmath108 .
since @xmath109 is a vandermonde matrix with @xmath43 distinct for @xmath110 , it is nonsingular .
thus , by the svd of @xmath3 , we have @xmath111 with @xmath112 define @xmath113 . then @xmath114 and the columns of @xmath115 form an orthonormal basis of @xmath116 .
write @xmath117 . by definition
, we obtain @xmath118 which proves and indicates that @xmath119 is monotonically increasing with respect to @xmath120 .
we next estimate @xmath120 .
we have @xmath121 so we need to estimate @xmath122 .
it is easily justified that the @xmath123-th column of @xmath124 consists of the coefficients of the lagrange polynomial @xmath125 that interpolates the elements of the @xmath123-th canonical basis vector @xmath126 at the abscissas @xmath127 .
consequently , the @xmath123-th column of @xmath128 is @xmath129 from which we obtain @xmath130 since @xmath131 is monotonic for @xmath132 , it is bounded by @xmath133 .
furthermore , let @xmath134 .
then for @xmath135 and @xmath41 we have @xmath136 by absorbing those higher order terms into @xmath137 .
note that in the above numerator we have @xmath138 and @xmath139 it is then easily seen that their product is @xmath140 on the other hand , by definition , the denominator @xmath141 in is exactly one for @xmath142 , and it is strictly bigger than one for @xmath143 .
therefore , for any @xmath0 , we have @xmath144 . from this and
it follows that @xmath145 therefore , for @xmath146 and @xmath41 considerably , from we have @xmath147 * remark 2.1 * we point out that should not be sharp . as we have seen from the proof , the factor @xmath148 seems intrinsic and unavoidable , but the factor @xmath149 in is an overestimate and
can certainly be reduced .
is an overestimate since @xmath133 for @xmath123 not near to @xmath0 is considerably smaller than @xmath150 , but we replace all them by their maximum @xmath151 .
in fact , our derivation clearly illustrates that the smaller @xmath123 is , the smaller @xmath133 than @xmath152 .
recall the discrete picard condition .
then @xmath153 we observe that @xmath154 almost remains constant for @xmath73 . for @xmath155 , note that all the @xmath156 almost remain the same .
thus , we have @xmath157 , meaning that @xmath116 does not capture @xmath158 as well as it does for @xmath73 .
* remark 2.2 * the theorem can be extended to moderately ill - posed problems with the singular values @xmath159 considerably and @xmath0 not big since , in a similar manner to the proof of theorem [ thm2 ] , we can obtain by the first order taylor expansion @xmath160 which , unlike @xmath161 for severely ill - posed problems , depends on @xmath0 and increases slowly with @xmath0 for @xmath39 considerably .
however , for mildly ill - posed problems , from above we have @xmath162 considerably for @xmath163 .
* remark 2.3 * a combination of and and the above analysis indicate that @xmath116 captures @xmath158 better for severely ill - posed problems than for moderately ill - posed problems .
there are two reasons for this .
the first is that the factors @xmath164 are basically fixed constants for severely ill - posed problems as @xmath0 increases , and they are smaller than the counterparts for moderately ill - posed problems unless the degree @xmath36 of its ill - posedness is far bigger than one and @xmath0 small .
the second is that the factor @xmath165 is smaller for severely ill - posed problems than the factor @xmath151 for moderately ill - posed problems for the same @xmath0 .
* remark 2.4 * the situation is fundamentally different for mildly ill - posed problems : firstly , we always have @xmath162 substantially for @xmath166 and any @xmath0 , which is considerably bigger than @xmath165 for moderately ill - posed problems for the same @xmath0 .
secondly , @xmath167 defined by is closer to one than that for moderately ill - posed problems for @xmath75 .
thirdly , for the same noise level @xmath168 and @xmath169 , we see from the discrete picard condition and the definition of @xmath22 that @xmath22 is bigger for a mildly ill - posed problem than that for a moderately ill - posed problem . all of them show that @xmath116 captures @xmath158 _ considerably better _ for severely and moderately ill - posed problems than for mildly ill - posed problems for @xmath75 . in other words , our results illustrate that @xmath116 contains more information on the other @xmath87 right singular vectors for mildly ill - posed problems , compared with severely and moderately ill - posed problems .
the bigger @xmath0 , the more it contains .
therefore , @xmath116 captures @xmath158 more effectively for severely and moderately ill - posed problems than mildly ill - posed problems .
that is , @xmath116 contains more information on the other @xmath87 right singular vectors for mildly ill - posed problems , making the appearance of a small ritz value more possible before @xmath73 and lsqr has better regularizing effects for the first two kinds of problems than for the third kind .
note that lsqr , at most , has the full regularization , i.e. , there is no ritz value smaller than @xmath170 for @xmath73 , for severely and moderately ill - posed problems .
our analysis indicates that lsqr generally has only the partial regularization for mildly ill - posed problem and a hybrid lsqr should be used . *
remark 2.5 * relation and @xmath167 indicate that @xmath116 captures @xmath158 better for severely ill - posed problems than for moderately ill - posed problems .
there are two reasons for this .
first , the all the @xmath164 are basically a fixed constant @xmath44 for severely ill - posed problems , which is smaller than those ratios for moderately ill - posed problems unless @xmath36 is rather big and @xmath0 small .
second , the quantities @xmath171 for severely ill - posed problems are smaller than the corresponding @xmath165 for moderately ill - posed problems .
let us investigate more and get insight into the regularization of lsqr .
define @xmath172 which measures the quality of the rank @xmath0 approximation @xmath173 to @xmath3 .
based on , we can derive the following estimate for @xmath174 .
[ thm3 ] assume that is severely or moderately ill posed .
then @xmath175 _ proof_. let @xmath25 be the best rank @xmath0 approximation to @xmath3 with respect to the 2-norm , where @xmath26 , @xmath27 and @xmath28 .
since @xmath173 is of rank @xmath0 , the lower bound in is trivial by noting that @xmath176 .
we now prove the upper bound . from ,
we obtain @xmath177 it is easily known that @xmath178 with @xmath179 having orthonormal columns .
then by the definition of @xmath119 we obtain @xmath180 numerically , it has been extensively observed in the literature that the @xmath174 decay as fast as @xmath181 and , more precisely , @xmath182 for severely ill - posed problems ; see , e.g. , @xcite .
they mean that the @xmath173 are very good rank @xmath0 approximations to @xmath3 .
recall that the tsvd method generates the best regularized solution @xmath183 . as a result , if @xmath184 , the lsqr iterate @xmath185 is reasonably close to the tsvd solution @xmath31 for @xmath170 is reasonably small .
this means that lsqr has the full regularization and does not need any additional regularization to improve @xmath76 . as our experiments will indicate in detail , these observed phenomena are of generality for both severely and moderately ill - posed problems and thus should have strong theoretical supports .
compared to the observations , our appears to be a considerrable overestimate .
we next present some results on @xmath186 appearing in . if @xmath187 , the lanczos bidiagonalization process terminates , and we have found @xmath0 exact singular triples of @xmath3 @xcite . in our context , since @xmath3 has only simple singular values and @xmath4 has components in all the left singular vectors , early termination is impossible in exact arithmetic , but small @xmath186 is possible .
we aim to investigate how fast @xmath186 decays .
we first give a refinement of a result in @xcite .
[ thm1 ] let @xmath188 be the svd of @xmath67 , where @xmath189 and @xmath190 are orthogonal , and @xmath191 , and define @xmath192 and @xmath193 .
then @xmath194 _ proof_. from and @xmath195 , we obtain @xmath196 so holds . from
, we get @xmath197 note that @xmath198 .
then we get @xmath199 we remark that it is an inequality other than the equality in a result of @xcite similar to . in combination with the previous results and remarks , this theorem shows that once @xmath186 becomes small for not big @xmath0 , the @xmath0 singular values of @xmath67 may approximate the large singular values of @xmath3 , and it is more possible that no small one appears for severely ill - posed problems and moderately ill - posed problems . as our final result , we establish an intimate and interesting relationship between @xmath186 and @xmath174 , showing how fast @xmath186 decays .
[ thm4 ] it holds that @xmath200 _ proof_. with the notations as in theorem [ thm1 ] , we have @xmath201 .
so , by , we have @xmath202 note that @xmath203 .
therefore , from we obtain @xmath204 the theorem indicates that @xmath186 decays at least as fast as @xmath174 , which , in turn , means that @xmath186 may decrease in the same rate as @xmath181 , as observed in @xcite for severely ill - posed problems .
in this section , we report numerical experiments to illustrate the the regularizing effects of lsqr .
we will demonstrate that lsqr has the full regularization for severely and moderately ill - posed problems , stronger phenomena than our theory proves , but it only has the partial regularization for mildly ill - posed problems , in accordance with our theory , for which a hybrid lsqr is needed to compute best possible regularized solutions .
we choose several ill - posed examples from hansen s regularization toolbox @xcite .
all the problems arise from the discretization of the first kind fredholm integral equation @xmath205 for each problem we use the codes of @xcite to generate a @xmath206 matrix @xmath3 , true solution @xmath34 and noise - free right - hand @xmath19 . in order to simulate the noisy data
, we generate the gaussian noise vector @xmath9 whose entries are normally distributed with mean zero . defining the noise level @xmath207
, we use @xmath208 , respectively , in the test examples . to simulate exact arithmetic ,
the full reorthogonalization is used during the lanczos bidiagonalization process .
we remind that , as far as ill - posed problem is concerned , our primary goal consists in justifying the regularizing effects of iterative solvers , which are _ unaffected by sizes _ of ill - posed problems and only depends on the degree of ill - posedness . therefore , for this purpose , as extensively done in the literature ( see , e.g. , @xcite and the references therein )
, it is enough to test not very large problems .
indeed , for @xmath99 large , say , 1,0000 and more , we have observed completely the same behavior as that for @xmath99 not large , e.g. , @xmath209 used in this paper . a reason for using @xmath99 not large is because such choice makes it practical to fully justify the regularization effects of lsqr by comparing it with the tsvd method , which suits only for small and/or medium sized problems for computational efficiency .
all the computations are carried out in matlab 7.8 with the machine precision @xmath210 under the microsoft windows 7 64-bit system .
we consider the following two severely ill - posed problems @xcite .
* example 1 * this problem shaw arises from one - dimensional image restoration , and can be obtained by discretizing the first kind fredholm integral equation with @xmath211 $ ] as both integration and domain intervals .
the kernel @xmath212 and the solution @xmath213 are given by @xmath214 @xmath215 * example 2 * this problem wing has a discontinuous solution and is obtained by discretizing the first kind fredholm integral equation with @xmath216 $ ] as both integration and domain intervals .
the kernel @xmath212 , the solution @xmath213 and the right - hand side @xmath217 are given by @xmath218 @xmath219 these two problems are severely ill - posed , whose singular values @xmath40 with @xmath220 for shaw and @xmath221 for wing , respectively . in figure
[ fig : res ] , we display the curves of the sequences @xmath174 and @xmath186 with @xmath222 , respectively .
they illustrate that the quantities @xmath174 decrease as fast as @xmath181 and both of them level off at the level of @xmath223 for @xmath0 no more than 20 , and after that these quantities are purely round - offs and are reliable no more .
moreover , the curves of quantities @xmath186 always lie below those of @xmath174 , which coincides with theorem [ thm4 ] .
we can see that the decaying curves with different noise levels are almost the same .
furthermore , we observe that @xmath182 for severely ill - posed problems , indicating that the @xmath173 are very good rank @xmath0 approximations to @xmath3 with the approximate accuracy @xmath181 and that @xmath67 does not become ill - conditioned before @xmath73 . as a result ,
the regularized solutions @xmath72 become better approximations to @xmath34 until iteration @xmath22 , and they are deteriorated after that iteration . at iteration @xmath22 , @xmath76 only captures the @xmath22 dominant svd components of @xmath3 and suppress the other @xmath224 svd components , so that it is a best possible regularized solution . as a result ,
the pure lsqr has the full regularization for severely ill - posed problems
. we will give a more direct justification on these assertions in section 3.3 . in figure [ fig2 ] , we plot the relative errors @xmath225 with different noise levels for these two problems . obviously
, lsqr exhibits semi - convergence phenomenon .
moreover , for smaller noise level , we get better regularized solutions at the cost of more iterations , as expected .
, @xmath181 and @xmath186 for the problem shaw with @xmath226 ( left ) and @xmath227 ( right ) ; ( c)-(d ) : plots of decaying behavior of the sequences @xmath174 and @xmath181 for the problem wing with @xmath227 ( left ) and @xmath228 ( right).,width=264,height=188 ] ( a ) , @xmath181 and @xmath186 for the problem shaw with @xmath226 ( left ) and @xmath227 ( right ) ; ( c)-(d ) : plots of decaying behavior of the sequences @xmath174 and @xmath181 for the problem wing with @xmath227 ( left ) and @xmath228 ( right).,width=264,height=188 ] ( b ) , @xmath181 and @xmath186 for the problem shaw with @xmath226 ( left ) and @xmath227 ( right ) ; ( c)-(d ) : plots of decaying behavior of the sequences @xmath174 and @xmath181 for the problem wing with @xmath227 ( left ) and @xmath228 ( right).,width=264,height=188 ] ( c ) , @xmath181 and @xmath186 for the problem shaw with @xmath226 ( left ) and @xmath227 ( right ) ; ( c)-(d ) : plots of decaying behavior of the sequences @xmath174 and @xmath181 for the problem wing with @xmath227 ( left ) and @xmath228 ( right).,width=264,height=188 ] ( d ) with respect to @xmath222 for the problems shaw ( left ) and wing ( right).,width=264,height=188 ] ( a ) with respect to @xmath222 for the problems shaw ( left ) and wing ( right).,width=264,height=188 ] ( b ) we now consider the following two moderately ill - posed problems @xcite .
* example 3 * this problem heat arises from the inverse heat equation , and can be obtained by discretizing volterra integral equation of the first kind , a class of equations that is moderately ill - posed , with @xmath216 $ ] as integration interval .
the kernel @xmath229 with @xmath230 * example 4 * this problem is the famous phillips test problem .
it can be obtained by discretizing the first kind fredholm integral equation with @xmath231 $ ] as both integration and domain intervals .
the kernel @xmath212 , the solution @xmath213 and the right - hand side @xmath217 are given by @xmath232 @xmath233 @xmath234 , @xmath181 and @xmath186 for the problem heat with @xmath226 ( left ) and @xmath227 ( right ) ; ( c)-(d ) : plots of decaying behavior of the sequences @xmath174 and @xmath181 for the problem phillips with @xmath227 ( left ) and @xmath228 ( right).,width=264,height=188 ] ( a ) , @xmath181 and @xmath186 for the problem heat with @xmath226 ( left ) and @xmath227 ( right ) ; ( c)-(d ) : plots of decaying behavior of the sequences @xmath174 and @xmath181 for the problem phillips with @xmath227 ( left ) and @xmath228 ( right).,width=264,height=188 ] ( b ) , @xmath181 and @xmath186 for the problem heat with @xmath226 ( left ) and @xmath227 ( right ) ; ( c)-(d ) : plots of decaying behavior of the sequences @xmath174 and @xmath181 for the problem phillips with @xmath227 ( left ) and @xmath228 ( right).,width=264,height=188 ] ( c ) , @xmath181 and @xmath186 for the problem heat with @xmath226 ( left ) and @xmath227 ( right ) ; ( c)-(d ) : plots of decaying behavior of the sequences @xmath174 and @xmath181 for the problem phillips with @xmath227 ( left ) and @xmath228 ( right).,width=264,height=188 ] ( d ) from figure [ fig3 ] , we see that @xmath174 decreases as fast as @xmath181 , and @xmath186 decays as fast as @xmath174 .
however , slightly different from severely ill - posed problems , we can observe that the @xmath174 may not be so close to the @xmath181 , as reflected by the thick rope formed by three lines . by comparing the behavior of @xmath174 for severely and moderately ill - posed problem ,
we come to the conclusion that the @xmath0-step lanczos bidiagonalization may generate more accurate rank @xmath0 approximation for severely ill - posed problems than for moderately ill - posed problems , namely , the rank @xmath0 approximation @xmath173 may be more accurate for severely ill - posed problems than for moderately ill - posed problems .
nonetheless , we have seen that , for the test moderately ill - posed problems , all the @xmath174 are still excellent approximations to the @xmath181 , so that lsqr still has the full regularization . in figure [ fig4 ]
, we depict the relative errors of @xmath72 , and observe analogous phenomena to those for severely ill - posed problems .
a distinction is that now lsqr needs more iterations for moderately ill - posed problems with the same noise level . with respect to @xmath222 for the problems heat ( left ) and phillips ( right).,width=264,height=188 ] ( a ) with respect to @xmath222 for the problems heat ( left ) and phillips ( right).,width=264,height=188 ] ( b ) for the previous four severely and moderately ill - posed problems , we now compare the regularizing effects of the pure lsqr and the hybrid lsqr with the additional tsvd regularization used within projected problems .
we show that lsqr has the full regularization and no additional regularization is needed , which is based on the observation that at semi - convergence the regularized solution by lsqr is as accurate as that obtained by the hybrid lsqr for each problem . in the sequel , we only report the results for the noise level @xmath227 .
results for other @xmath235 are analogous and thus omitted .
figures [ fig5 ] ( a)-(b ) and figures [ fig6 ] ( a)-(b ) indicate that the relative errors of approximate solutions obtained by the two methods reach the same minimum level , and the hybrid lsqr simply stabilizes the regularized solutions with the minimum error .
this means that the pure lsqr itself has already found a best possible regularized solution at semi - convergence and no additional regularization is needed .
so it has the full regularization .
our task is to determine such @xmath0 , which is the iteration where @xmath236 starts to increase dramatically while its residual norm remains almost unchanged .
the l - curve criterion fits nicely into this task . in these examples ,
we also choose @xmath237 for the pure lsqr .
figure [ fig5 ] ( c ) and figures [ fig6 ] ( c)-(d ) show that the regularized solutions are generally very good approximations to the true solutions .
however , we should point out that for the problem wing with a discontinuous solution , the large relative error indicates that the regularized solution is a poor approximation to the true solution , as depicted in figure [ fig5 ] ( d ) .
such phenomenon is due to the fact that the regularization of lsqr and its hybrid variants is unsuitable for the ill - posed problems with discontinuous solutions . for such kind of problems ,
more reasonable regularization is total variation regularization , which takes the form @xmath238 with @xmath239 some @xmath240 matrix and @xmath241 the 1-norm @xcite . with respect to lsqr and lsqr with additional tsvd regularization for @xmath227 ;
( c)-(d ) : the regularized solutions @xmath242 for the pure lsqr for the problems shaw ( left ) and wing ( right).,width=264,height=188 ] ( a ) with respect to lsqr and lsqr with additional tsvd regularization for @xmath227 ; ( c)-(d ) : the regularized solutions @xmath242 for the pure lsqr for the problems shaw ( left ) and wing ( right).,width=264,height=188 ] ( b ) with respect to lsqr and lsqr with additional tsvd regularization for @xmath227 ; ( c)-(d ) : the regularized solutions @xmath242 for the pure lsqr for the problems shaw ( left ) and wing ( right).,width=264,height=188 ] ( c ) with respect to lsqr and lsqr with additional tsvd regularization for @xmath227 ; ( c)-(d ) : the regularized solutions @xmath242 for the pure lsqr for the problems shaw ( left ) and wing ( right).,width=264,height=188 ] ( d ) obtained by the pure lsqr and lsqr with the additional tsvd regularization for @xmath227 ; ( c)-(d ) : the regularized solutions @xmath242 for the pure lsqr for the problems heat ( left ) and phillips ( right).,width=264,height=188 ] ( a ) obtained by the pure lsqr and lsqr with the additional tsvd regularization for @xmath227 ; ( c)-(d ) : the regularized solutions @xmath242 for the pure lsqr for the problems heat ( left ) and phillips ( right).,width=264,height=188 ] ( b ) obtained by the pure lsqr and lsqr with the additional tsvd regularization for @xmath227 ; ( c)-(d ) : the regularized solutions @xmath242 for the pure lsqr for the problems heat ( left ) and phillips ( right).,width=264,height=188 ] ( c ) obtained by the pure lsqr and lsqr with the additional tsvd regularization for @xmath227 ; ( c)-(d ) : the regularized solutions @xmath242 for the pure lsqr for the problems heat ( left ) and phillips ( right).,width=264,height=188 ] ( d ) in what follows , we compare the regularizing effects of the pure lsqr and hybrid lsqr for mildly ill - posed problems , showing that lsqr has only the partial regularization and a hybrid lsqr should be used for this kind of problem to improve the regularized solution by lsqr at semi - convergence .
* example 5 * the problem deriv2 is mildly ill - posed , which is obtained by discretizing the first kind fredholm integral equation with @xmath216 $ ] as both integration and domain intervals .
the kernel @xmath212 is green s function for the second derivative : @xmath243 and the solution @xmath213 and the right - hand side @xmath217 are given by @xmath244 figure [ fig7 ] ( a ) shows that the relative errors of approximate solutions by the hybrid lsqr reach a considerably smaller minimum level than those by the pure lsqr , a clear indication that lsqr has the partial regularization .
as we have seen , the hybrid lsqr expands the krylov subspace until it contains enough dominant svd components and , meanwhile , additional regularization effectively dampen the svd components corresponding to small singular values .
for instance , the semi - convergence of the pure lsqr occurs at iteration @xmath245 , but it is not enough . as the hybrid lsqr shows , we need a larger six dimensional krylov subspace @xmath246 to construct a best possible regularized solution .
we also choose @xmath237 for the pure lsqr and the hybrid lsqr .
figure [ fig7 ] ( b ) indicates that the regularized solution obtained by the hybrid lsqr is a considerably better approximation to @xmath34 than that by the pure lsqr , especially in the non - smooth middle part of @xmath34 . and the regularized solution @xmath242 with respect to lsqr and lsqr with the additional tsvd regularization for the problem deriv2 and @xmath227.,width=264,height=188 ] ( a ) and the regularized solution @xmath242 with respect to lsqr and lsqr with the additional tsvd regularization for the problem deriv2 and @xmath227.,width=264,height=188 ] ( b )
for large - scale discrete ill - posed problems , lsqr and cgls are commonly used methods .
these methods have regularizing effects and exhibit semi - convergence .
however , if a small ritz value appears before the methods capture all the needed dominant svd components , the methods have only the partial regularization and must be equipped with additional regularization so that best possible regularized solutions can be found . otherwise , lsqr has the full regularization and can compute best possible regularized solutions without additional regularization needed .
we have proved that the underlying @xmath0-dimensional krylov subspace captures the @xmath0 dimensional dominant right singular space better for severely and moderately ill - posed problems than for mildly ill - posed problems .
this makes lsqr have better regularization for the first two kinds of problems than for the third kind .
furthermore , we have shown that lsqr generally has only the partial regularization for mildly ill - posed problems .
numerical experiments have demonstrated that lsqr has the full regularization for severely and moderately ill - posed problems , stronger than our theory predicts , and it has the partial regularization for mildly moderately ill - posed problems , compatible with our assertion .
together with the observations @xcite , it appears that the excellent performances of lsqr on severely and moderately ill - posed problems generally hold . as for future work , it is more appealing to derive an accurate estimate for @xmath120 other than @xmath247 , as it plays a crucial role in analyzing the accuracy @xmath174 of the rank @xmath0 approximation , generated by lanczos bidiagonalization , to @xmath3 .
accurate bounds for @xmath174 are the core of completely understanding the regularizing effects of lsqr , but our bound for @xmath174 is conservative and is expected to be improved on substantially . since cgls is mathematically equivalent to lsqr , our results apply to cgls as well .
our current work has helped to better understand the regularization of lsqr and cgls .
but for a complete understanding of the intrinsic regularizing effects of lsqr and cgls , we still have a long way to go , and more research is needed .
we thank the three referees very much for their valuable suggestions and comments , which made us improve the presentation of the paper .
, _ ill - conditioning of the truncated singular value decomposition , tikhonov regularization and their applications to numerical partial differential equations_. numer .
linear algebra appl . , 18 ( 2011 ) , pp . | lsqr , a lanczos bidiagonalization based krylov subspace iterative method , and its mathematically equivalent cgls applied to normal equations system , are commonly used for large - scale discrete ill - posed problems .
it is well known that lsqr and cgls have regularizing effects , where the number of iterations plays the role of the regularization parameter .
however , it has long been unknown whether the regularizing effects are good enough to find best possible regularized solutions . here
a best possible regularized solution means that it is at least as accurate as the best regularized solution obtained by the truncated singular value decomposition ( tsvd ) method . in this paper
, we establish bounds for the distance between the @xmath0-dimensional krylov subspace and the @xmath0-dimensional dominant right singular space .
they show that the krylov subspace captures the dominant right singular space better for severely and moderately ill - posed problems than for mildly ill - posed problems .
our general conclusions are that lsqr has better regularizing effects for the first two kinds of problems than for the third kind , and a hybrid lsqr with additional regularization is generally needed for mildly ill - posed problems . exploiting the established bounds
, we derive an estimate for the accuracy of the rank @xmath0 approximation generated by lanczos bidiagonalization .
numerical experiments illustrate that the regularizing effects of lsqr are good enough to compute best possible regularized solutions for severely and moderately ill - posed problems , stronger than our theory predicts , but they are not for mildly ill - posed problems and additional regularization is needed .
ill - posed problem , regularization , lanczos bidiagonalization , lsqr , cgls , hybrid 65f22 , 65j20 , 15a18 |
atomically thin @xmath0he film adsorbed on graphite is an ideal model system for studying strong correlation effects in two - dimensional ( 2d ) fermions . in this system
, the correlation can be tuned in a wide range by changing @xmath0he areal density ( @xmath2 ) .
intensive studies have been done especially in the density region around the 4/7 commensurate phase in the second layer @xcite .
grafoil , an exfoliated graphite , is an adsorption substrate commonly used in these kinds of experiments because of its large surface area ( @xmath3 20 m@xmath4/g ) and moderate thermal conductivity .
however , grafoil consists of a lot of microcrystallites ( platelets ) with mosaic angle distributions .
thus , the experimental results sometimes suffer from the heterogeneity effects of substrate .
for example , the non - exponential decay of nmr spin - echo signals recently observed near the 4/7 phase @xcite and the linear larmor frequency - dependence of transverse spin relaxation rate ( @xmath5 ) @xcite are believed to be due to such heterogeneity effects .
therefore , it is important to obtain distributions of the platelet size and the mosaic angle spread accurately enough for each batch if possible .
so far , there exist only a few experimental determinations of the platelet distributions , and it is desired to cross - check the results by different methods . in this paper , we demonstrate that the mosaic angle distribution can be determined with good accuracy by analyzing continuous - wave ( cw ) nmr spectra and consequently that the distribution claimed by the previous neutron scattering experiments @xcite is not correct .
the distribution of the platelet size in grafoil estimated from scanning tunneling microscopy ( stm ) measurements @xcite is widely spread below 100 nm . the mosaic angle distribution of platelets was determined by the neutron scattering experiments @xcite as follows . for 2d crystals , the intensity diffracted at an angle @xmath6 by the bragg reflection with the plane @xmath7 is represented as @xmath8 here , @xmath9 and @xmath10 are a normalization constant , the multiplicity of the @xmath11 reflection and the molecular form factor , respectively .
@xmath12 is the debye - waller factor , and @xmath13 where @xmath14 and @xmath15 .
@xmath16 and @xmath17 are the wavelength , 2d plane spacing for the @xmath18-reflection and a parameter defining the average size of the diffracting arrays , respectively . using the orientational distribution function of grafoil platelets @xmath19 , the instrumental resolution factor @xmath20 and @xmath21 , the angle dependence of the scattering intensity @xmath22 is expressed as @xmath23 thus @xmath24 can be determined by comparing measured diffraction peaks with ones calculated through eq.([eq : nscat_int ] ) .
_ @xcite assumed @xmath24 as the sum of gaussian and some constant he adsorbed on grafoil for two different field directions : ( a ) @xmath25 and ( b ) @xmath26 .
the points are experimental data for the @xmath27 21.5 nm@xmath28 sample at temperatures indicated ( from ref .
the solid and dashed lines are calculated spectra with eq .
( [ eq : dist_nmr ] ) and eq . ( [ eq : dist_neu ] ) , respectively . ] @xmath29 and obtained @xmath30 from their own neutron diffraction data for ar monolayer adsorbed on grafoil . as is shown in fig .
[ fig : distribution ] , this has a large isotropic powder - like component @xmath31 in addition to the angle dependent term with @xmath32 .
however , this is hard to believe since it is too isotropic judging from the existing ample experimental properties of thin films adsorbed on grafoil and of the substrate itself , e.g. , the thermal conductivity @xcite , that all show the significant anisotropy .
in this section , we introduce a quite different method to determine the mosaic angle distribution of grafoil platelets based on an analysis of cw - nmr spectra of @xmath0he thin films adsorbed on this substrate . here
we use the nmr data for the high density second - layer @xmath0he at very low temperatures , where the nuclear spins are ferromagnetically aligned along the direction of applied magnetic field @xmath33 , measured by schiffer _
the nmr frequency shift due to the demagnetization effect is expressed as @xmath34 , where @xmath35 is the maximum positive value of frequency shift , @xmath36 is the degree of spin polarization and @xmath37 is the angle between @xmath33 and the vector normal to the average plane of grafoil @xmath38 @xcite .
since the inclination of each platelet about the average plane has two degrees of freedom , it is designated by two angular variables , @xmath39 and @xmath40 , in the spherical coordinate with @xmath38 being the positive direction of @xmath41-axis .
then , the nmr line shape about the field corresponding to the larmor frequency @xmath42 is represented as he adsorbed on grafoil . the dashed , solid and dash - dotted lines are calculated spectra with @xmath43 deg , respectively .
@xmath44 is fixed at 0.045 . otherwise , the same as fig .
[ fig : cw ] . ]
he adsorbed on grafoil .
the dashed , solid and dash - dotted lines are the calculated spectra with @xmath45 , respectively .
@xmath46 is fixed at 27.5 deg . otherwise , the same as fig .
[ fig : cw ] . ]
@xmath47 here @xmath48 , and @xmath49 is the intrinsic line width .
thus , @xmath24 can be determined by fitting the measured nmr spectra to eq.([eq : cw ] ) .
we note that similar analyses have been briefly reported by previous workers @xcite for the @xmath25 direction .
we fitted the nmr data at @xmath50 = 0.28 mk for @xmath27 21.5 nm@xmath28 taken by schiffer _
_ @xcite assuming the gaussian distribution ( eq .
( [ eq : dist_fn ] ) ) , and obtained @xmath51 the mosaic angle distribution determined by this method is shown in fig .
[ fig : distribution_new ] , which is much more anisotropic than fig .
[ fig : distribution ] .
the fitted results are shown by the solid lines ( red ) in fig .
[ fig : cw ] .
the fitting quality is remarkably good both for ( a ) @xmath25 and ( b ) @xmath26 .
in addition to this , the small fitted value for @xmath44 indicates the appropriateness of the gaussian functional form for @xmath24 .
note that there are small bumps near @xmath42 in the spectra for both @xmath33 directions which are not reproduced by the fittings in fig .
[ fig : cw ] .
these are attributable to magnetization of the paramagnetic first - layer @xmath0he @xcite . on the other hand , the dashed lines ( blue ) in fig .
[ fig : cw ] are spectra calculated from eq .
( [ eq : dist_neu ] ) , the distribution claimed by taub _
they look nearly the same for both directions of @xmath33 because of the large isotropic component @xmath31 and do not explain the experimental data at all .
we noticed that , in ref .
@xcite , they represented the platelet inclination only with @xmath39 without considering the @xmath40 variation .
we believe that is why their estimation of the mosaic angle distribution is unrealistically isotropic . note that the fitting for the @xmath25 direction is very sensitive to the presence of the isotropic component @xmath31 , while that for the @xmath26 is not .
let us estimate the sensitivity of cw - nmr spectra in determining the mosaic angle distribution in this method . in figs .
[ fig : cw_param_delta ] and [ fig : cw_param_h ] , we plotted the spectra calculated with slightly different @xmath46 and @xmath44 values from those of the best fitting by @xmath52 deg and @xmath53 , respectively .
the spectra are sensitive enough to discriminate such small variations of the distributions .
the constraint that the spectra for both @xmath25 and @xmath26 directions should be fitted simultaneously with the same parameters enhances the sensitivity .
in this paper , we proposed a useful method to estimate the mosaic angle distribution of grafoil platelets from cw - nmr spectra of highly polarized monolayer @xmath0he adsorbed on the substrate . by analyzing the nmr data taken by the stanford group @xcite
, we have proved that grafoil used by them consists of two components .
one is isotropic like powder ( about 5 % ) and the other is distributed in the gaussian function of inclination @xmath39 with a standard deviation @xmath54 deg .
these values are much more anisotropic ( or 2d like ) than ones determined from the previous analysis by taub _
@xcite who used the neutron diffraction data of ar monolayer adsorbed on grafoil .
we have found that their integration of the angular distribution is inappropriate .
the results shown in this paper will be valuable to elucidate intrinsic 2d properties from various experimental data for monolayer systems adsorbed on grafoil substrate .
it is desirable to reanalyze their data with the mosaic angle distribution of eqs .
( [ eq : dist_fn ] ) and ( [ eq : dist_nmr ] ) by taking account of details of the experimental setup .
we thank useful discussions with bill mullin .
this work was financially supported by grant - in - aid for scientific research on priority areas ( no .
17071002 ) from mext , japan . s.t . acknowledges support from the jsps research fellowship . | we described details of a method to estimate with good accuracy the mosaic angle distributions of microcrystallites ( platelets ) in exfoliated graphite like grafoil which is commonly used as an adsorption substrate for helium thin films .
the method is based on analysis of resonance field shifts in continuous - wave ( cw ) nmr spectra of @xmath0he ferromagnetic monolayers making use of the large nuclear polarization of the adsorbate itself .
the mosaic angle distribution of a grafoil substrate analyzed in this way can be well fitted to a gaussian form with a @xmath1 deg spread .
this distribution is quite different from the previous estimation based on neutron scattering data which showed an unrealistically large isotropic powder - like component . |
combinatorics of coloured dyadic intervals refers to a set of techniques created for the study of operators defined through their action on the haar system .
we refer to the treatment of averaging projections by p.w .
jones @xcite , the proof of the vector - valued @xmath0 theorem by t figiel @xcite , the use of the stripe operators in j. lee , p.f.x .
mller , s. mller @xcite , and the study of rearrangement operators on @xmath1 spaces , p.f.x .
mller @xcite , k. smela @xcite , a. kamont , p.f.x .
mller @xcite . here
we study a very natural colouring problem on dyadic trees .
we start out with a coloured collection of dyadic intervals @xmath2 , where we assume that the colours are distributed homogeneously over @xmath2 . given any collection @xmath3 containing @xmath2 we ask if there exists an equally homogeneous colouring of @xmath3 that preserves the colours of @xmath2 ( consistent colouring of @xmath3 ) .
the nature of this problem depends very much on what we agree to call a homogeneous distribution of colours .
our choice of homogeneity is very restrictive , and consequently in working on the problem of consistent colouring we encountered delicate combinatorial questions .
let @xmath4 denote the collection of dyadic intervals in the unit interval @xmath5 $ ] , and let @xmath6 we consider a large collection @xmath7 .
we assume that the intervals in @xmath2 are painted with @xmath8 distinct colours , giving rise to a decomposition @xmath9 it is intuitively clear what it means that the colours @xmath10 are homogeneously distributed among the intervals of @xmath2 .
for instance , we would demand that there exists @xmath11 so that @xmath12 where @xmath13 denotes the cardinality of the collection @xmath14 .
a much stronger measure of homogeneity arises when we ask for ( [ z.1 ] ) to hold over the prespecified collection of testing intervals @xmath15 specifically , if @xmath16 , we would demand that there exists @xmath17 so that @xmath18 where @xmath19 we use an additional rule to express homogeneity with respect to testing intervals that satisfy @xmath20 .
the necessity of such a rule arises from the fact that the cardinalities @xmath21 take values in @xmath22 , hence if @xmath23 , then ( [ z.2 ] ) has to fail
. thus , there are two regimes high cardinality and low cardinality of @xmath24 , and the transistion arises at @xmath25 .
the following definition contains the homogeneity conditions for both regimes , and it addresses the discrete nature of our gauge functions @xmath26 [ c.d1 ] let @xmath27 , and fix @xmath28 , @xmath29 .
let @xmath30 be some decomposition of @xmath2 .
this decomposition is called @xmath31-homogeneous colouring of @xmath32 if for each @xmath33 , @xmath34 one of the following holds : * * _ either _ * @xmath35 , and then @xmath36 * * _ or else _
* @xmath37 , and then @xmath38 remark .
+ + + + + + + we remark that for each ( uncoloured ) @xmath7 , @xmath39 and @xmath40 there is always a @xmath31-homogeneous colouring that can be obtained as follows : enumerate the intervals in @xmath2 from left to right , and simply put @xmath41 later , we refer to such a colouring as a _ colouring modulo @xmath8_. the use of this colouring rule applied to intervals of equal length
appeard in a context similar to ours in @xcite , see also @xcite and @xcite .
[ [ the - problem - of - consistent - colouring . ] ] the problem of consistent colouring .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the problem we treat in this paper is the following .
we are given two disjoint collections @xmath42 .
assume that the collection @xmath2 is coloured , that is , it is given an @xmath31-homogeneous colouring @xmath9 the collection @xmath43 consists of uncoloured intervals .
we would like to colour the intervals in @xmath43 with the same colours @xmath44 , that is to decompose @xmath43 as @xmath45 in such a way that the union @xmath46 has an @xmath31-homogeneous colouring given by @xmath47 that is , we want to obtain an @xmath31-homogeneous colouring of @xmath48 keeping the pre - existing @xmath31-homogeneous colouring of @xmath2 .
we refer to this question as to _ the problem of finding a colouring of @xmath3 consistent with existing colouring of @xmath2 .
_ our treatment of this problem is as follows : 1 . we isolate a condition on @xmath49 and @xmath32 ( previsibility ; see definition [ c.d2 ] ) implying that the problem of consistent colouring for @xmath50 has a solution .
see theorem [ c.t1 ] .
we give examples where the problem of consistent colouring for @xmath50 has just one solution .
moreover , we give examples ( of @xmath32 , its decomposition @xmath51 and @xmath49 ) for which the problem of consistent colouring for @xmath50 does not have a solution .
see proposition [ e.s1 ] .
3 . in section [ game ]
we reformulate the problem of consistent colouring as a two - person game . our results
theorem [ c.t1 ] and proposition [ e.s1 ] translate into winning strategies for the respective players . for the appearence of succesive colourings of dyadic intervals in the context of averaging projections
see @xcite .
in @xcite we constructed supporting trees for some rearrangement operators and thereby proved their boundedness on vector valued @xmath52 spaces .
initially , our approach to defining the supporting trees was by inclusion - exclusion principles and consistent colourings as studied in the present paper .
in the following we isolate a criterion which guarantees the existence of consistent colouring . to formulate this criterion , we use a dyadic interval @xmath33 together with its immediate dyadic successors @xmath53 , i.e. intervals @xmath54 such that @xmath55 and @xmath56 . the problem of consistent colouring lead us to the following condition : we are given disjoint collections @xmath42 and @xmath28 .
we say that the pair @xmath57 is _
@xmath8-previsible _ if with @xmath46 , the conditions @xmath58 imply @xmath59 to facilitate precise reference in the course of our argument below , we encode this notion of @xmath8-previsibility in the following equivalent definition .
[ c.d2 ] let @xmath42 , @xmath60 .
let @xmath28 .
the pair of collections @xmath61 is called @xmath8-previsible if for every @xmath33 with @xmath62 and its dyadic succesors @xmath63 , the following holds : @xmath64 now , we have the following theorem [ c.t1 ] which gives a sufficient condition for existence of consistent colourings .
[ c.t1 ] fix @xmath28 and @xmath65 , @xmath29 .
let @xmath7 , and let @xmath66 be a fixed @xmath31-homogeneous colouring of @xmath2 .
let @xmath67 be such that the pair @xmath61 is @xmath8-previsible .
then there is a colouring @xmath68 of @xmath43 such that @xmath69 is an @xmath31-homogeneous colouring of @xmath70 .
[ [ remark.-1 ] ] remark .
+ + + + + + + ones first attempt to prove theorem [ c.t1 ] by an inductive argument would be the following : find first @xmath31-homogeneous colouring of @xmath71 for @xmath72 , where @xmath73 .
then carry over these colourings inductively and backwards in time to larger collections @xmath74
, @xmath75 with @xmath76 as follows : assume that for a dyadic interval @xmath77 with successors @xmath63 , the separate @xmath31-homogeneous colourings of @xmath78 and @xmath79 are fixed . then check that the union of these colourings gives an @xmath31-homogeneous colouring of the union @xmath80 .
if this procedure would work , at stage @xmath81 , we would have produced an @xmath31-homogeneous colouring of @xmath71 for each @xmath82 with @xmath83 .
however , such a deterministic approach can not work , as the following example shows .
[ [ example . ] ] example .
+ + + + + + + + take @xmath84 , @xmath85 , and fix a dyadic interval @xmath77 with successors @xmath63 .
assume that collections @xmath86 are such that @xmath87 and @xmath88 .
take an @xmath31-homogeneous colouring of @xmath24 such that @xmath89 @xmath90 note that then we have @xmath91 next , we can choose colourings of @xmath92 and @xmath93 such that @xmath94 @xmath95 then , knowing the colouring of @xmath96 and @xmath97 for @xmath98 with @xmath46 we have @xmath99 so we have separate @xmath31-homogeneous colourings of @xmath78 and @xmath79 . however ,
by taking the union of these colourings we get a colouring of @xmath74 such that latexmath:[\[| { { \cal h}}_i \cap l| = 2 \quad\hbox { for } \quad 1 \leq
i \leq 2r \quad\hbox { and } \quad which is not @xmath31-homogeneous .
note however that the above example does not contradict the assertion of theorem [ c.t1 ] .
in fact given @xmath2 and its colouring as above
it is quite easy to obtain colouring of @xmath43 so that the conclusion of theorem [ c.t1 ] holds . for @xmath98
, we put @xmath101 taking the union with the colouring of @xmath96 , we find that @xmath102 , so we have an @xmath31-homogeneous colourings of @xmath71 , @xmath98 .
finally , @xmath103 so we have an @xmath104-homogeneous colouring of @xmath74 . in response to these examples
, we introduced a stopping time argument running backwards in time that produces the @xmath31-homogeneous colouring of theorem [ c.t1 ] . at stage @xmath81 of our inductive argument
, we will produce consistent @xmath31-homogeneous colourings of collections @xmath71 for @xmath82 with @xmath105 provided that @xmath106 satisfies @xmath107 [ [ proof - of - theorem - c.t1 . ] ] proof of theorem [ c.t1 ] .
+ + + + + + + + + + + + + + + + + + + + + + + + we are going to define colouring of @xmath43 by an inductive argument .
let @xmath108 be such that @xmath109 .
let us observe that if @xmath110 , then @xmath111 .
thus , if the homogeneity conditions respectively are satisfied for @xmath33 with @xmath112 , then they are satisfied for each @xmath33 with @xmath34 . therefore ,
in our procedure of colouring @xmath43 we consider only @xmath113 with @xmath114 .
the inductive argument is used to prove the following statement at each stage @xmath81 , @xmath115 : * inductive hypothesis at the stage @xmath81 : * let @xmath116 if @xmath117 or @xmath118 , then intervals in @xmath97 are still uncoloured .
if @xmath119 and @xmath120 , then all intervals in @xmath97 are coloured , and the colouring of @xmath71 is @xmath31-homogeneous ; as @xmath119 , this means that @xmath121 and @xmath122 * i. the start of the induction .
* let @xmath123
. then either @xmath124 or @xmath125 .
if @xmath124 or @xmath126 , then intervals in @xmath127 are left uncoloured . * i.2 . * if @xmath125 and @xmath128 , then also @xmath129 , which implies that @xmath130 for each @xmath131 .
in such case it is possible to colour intervals in @xmath127 so that @xmath132 for each @xmath133 , @xmath131 . *
ii . the inductive step .
* let @xmath134 .
the inductive assumption states that there is a colouring at stage @xmath135 . we need to prove that there is a colouring at stage @xmath136 . for this take @xmath137 . then @xmath55 with @xmath138 , @xmath139 .
each interval in @xmath24 or @xmath127 is included in @xmath140 or @xmath141 , so we have @xmath142 or @xmath143 .
if @xmath144 then also @xmath145 .
if @xmath143 , then also @xmath146 and @xmath147 . in both cases , by the induction hypothesis , intervals in both @xmath92 and @xmath93 are uncoloured . if @xmath148 , then leave intervals in @xmath127 still uncoloured .
if @xmath149 , then @xmath150 $ ] , and the induction ends .
this means that @xmath151 or @xmath152 .
if @xmath153 , then it is enough to assign elements of @xmath43 to colours different from colours of elements of @xmath2 .
if @xmath152 , then it is enough to colour @xmath3 by modulo @xmath8 method , see ( [ mod.d ] ) .
* ii.2 . * @xmath154 and @xmath155 .
it follows that at least one of collections @xmath156 must be nonempty .
now we separate next two subcases : * ii.2.a .
* @xmath154 , @xmath155 and both @xmath157 , @xmath158 .
* ii.2.b . *
@xmath154 , @xmath155 , @xmath159 , but @xmath158 .
( the case @xmath157 and @xmath160 is symmetric to this one , and there is no need to treat it separately . )
we first treat the case ii.2.a .
then we treat the case ii.2.b .
both these cases have their own subcases .
the case ii.2.b is treated by reducing its subcases to appropriate subcases of ii.2.a
. * ii.2.a.1 . * @xmath161 and @xmath162 .
recall that @xmath157 and @xmath158 . then by induction hypothesis
all intervals in @xmath163 and in @xmath164 are already coloured , i.e. all intervals in @xmath127 are coloured . moreover , by ( [ ( b3 ) ] ) , for each @xmath165 @xmath166 of course , we have also @xmath167 .
* ii.2.a.2 . * @xmath168 and @xmath169 .
then by induction hypothesis all intervals in @xmath92 and in @xmath93 are uncoloured , but the intervals in @xmath24 carry their colours .
now , we need to colour all intervals in @xmath170 . to simplify notation , let @xmath171 we have @xmath172 first consider the case @xmath173 .
then @xmath174 for simplicity , assume that intervals in @xmath175 have colours @xmath176 , and intervals in @xmath177 have colours @xmath178 .
now , we colour intervals in @xmath127 .
first , colour intervals in @xmath92 using colours @xmath179 , and then , if necessary ( i.e. @xmath180 ) , continuing with @xmath181 colours from @xmath182 ; since @xmath183 , in this way we assign colours to all intervals in @xmath92 .
next , we assign colours to intervals in @xmath93 .
if @xmath184 , then assign first colours @xmath185 , then continue with colours @xmath186 , and then if necessary with colours @xmath187 . if @xmath188 , then just choose @xmath189 different colours from @xmath190 and @xmath179 .
with such colouring of intervals in @xmath92 and in @xmath93 we find that both @xmath191 this implies that for each @xmath192 or @xmath193 we have @xmath194 and @xmath195 .
moreover , we get @xmath196 , which implies @xmath197 it remains to consider the case @xmath198
. then the homogeneity assumption on the decomposition of @xmath2 for @xmath53 and for @xmath77 implies @xmath199 for simplicity , assume that intervals in @xmath175 have colours @xmath200 and intervals in @xmath177 have colours @xmath201 and @xmath202 ( note that @xmath203 , since by assumption @xmath204 ) . to colour intervals in @xmath92
choose @xmath205 colours from @xmath201 .
to colour intervals in @xmath93 choose @xmath189 colours from @xmath206 .
this is possible since @xmath207 and @xmath208 .
observe that in this way we get @xmath209 therefore , for each @xmath192 or @xmath210 we have @xmath211 , while for @xmath77 we have @xmath197 * ii.2.a.3 .
* @xmath168 and @xmath212 .
recall that @xmath213 and @xmath158 , by the defining condition of the case ii.2.a .
then by induction hypothesis all intervals in @xmath92 are uncoloured , but the intervals in @xmath175 carry their colours .
since the pair @xmath57 is @xmath8-previsible , we have @xmath214 .
therefore , @xmath215 , and by condition of the @xmath31-homogeneity for @xmath2 , we get @xmath216 and @xmath217 , @xmath131 , satisfy ( [ ( b3 ) ] ) .
if @xmath218 as well , then all intervals in @xmath74 come from @xmath24 , and there is nothing to do . let @xmath219 .
we need to colour @xmath205 intervals in @xmath92 . to simplify notation , let @xmath220 .
note that @xmath221 .
let @xmath222 let @xmath223 be an ordering of @xmath224 such that @xmath225 , there are more colours in @xmath224 than intervals in @xmath226 .
now attach the colours @xmath227 , bijectively , to intervals in @xmath226
. then @xmath228 .
consequently , @xmath229 for each @xmath192 .
the colouring of @xmath230 is @xmath31-homogeneous by the assumption .
it remains to check that @xmath231 satisfy ( [ ( b3 ) ] ) . since @xmath214 and @xmath232 we have @xmath233 consequently , since the colouring of @xmath2 is @xmath31-homogeneous , we have @xmath234 and @xmath235 @xmath236 moreover , @xmath237 @xmath238 @xmath239
let @xmath240 be such that @xmath241
. then @xmath242 or @xmath243 .
if @xmath242 , then @xmath244 by ( [ ( c3.a ) ] ) , and ( [ ( b3 ) ] ) is satisfied for @xmath77 and @xmath3 because of ( [ ( c2 ) ] ) and inequality @xmath245 . if @xmath243 , then we have either @xmath246 or @xmath247 . the ordering defined by ( [ ( c4 ) ] )
implies that @xmath248 in the first case and @xmath249 in the latter case . if @xmath249 , then ( [ ( b3 ) ] ) is satisfied for @xmath77 and @xmath3 because of ( [ ( c1 ) ] ) , ( [ ( c3.c ) ] ) and inequality @xmath250 .
if @xmath248 and @xmath251 then we need to check inequality @xmath252 for @xmath253 inequality ( [ ( c5 ) ] ) is satisfied because of ( [ ( c1 ) ] ) combined with ( [ ( c3.a ) ] ) and ( [ ( c3.b ) ] ) . for @xmath254 inequality ( [ ( c5 ) ] ) is satisfied because of ( [ ( c2 ) ] ) and ( [ ( c3.b ) ] ) .
when @xmath251 , then ( [ ( c3.b ) ] ) and ( [ ( c3.c ) ] ) combined with the ordering ( [ ( c4 ) ] ) imply @xmath255 therefore @xmath256 for all @xmath257 , so inequality ( [ ( c5 ) ] ) is satisfied , even with @xmath258 on the left - hand - side , for @xmath259 .
* ii.2.a.4 . *
@xmath260 and @xmath169 .
this case is analogous to ii.2.a.3 .
next , we proceed with the case ii.2.b . * ii.2.b.1 . * @xmath261 and @xmath162 .
recall that by the condition defining case ii.2.b @xmath262 and @xmath263 .
then by the induction hypothesis all intervals in @xmath93 are already coloured , but intervals in @xmath92 are uncoloured .
we need to colour intervals in @xmath92 .
it is enough to colour them modulo @xmath8 , see ( [ mod.d ] ) . after this
, we get a colouring of @xmath74 such that both @xmath140 and @xmath141 satisfy ( [ hom2 ] ) . to check that @xmath77 satisfies ( [ hom2 ] ) as well ,
we proceed as in case ii.2.a.1 .
* ii.2.b.2 . *
@xmath264 and @xmath265 .
induction hypothesis states that intervals in @xmath92 and in @xmath93 are uncoloured .
now we proceed as in case ii.2.a.2 .
* ii.2.b.3 .
* @xmath168 and @xmath232 .
recall that @xmath266 .
therefore , by the previsibility assumption , @xmath267 . since @xmath268 ,
then by the induction hypothesis intervals in @xmath92 are uncoloured , and we need to colour them , in case @xmath269 .
this is done as in case ii.2.a.3 .
* ii.2.b.4 . * @xmath260 and @xmath265 .
recall that @xmath270 . in this case
the induction hypothesis says that intervals in both @xmath163 and @xmath164 are uncoloured .
we need to colour them all .
first , we colour intervals in @xmath93 by giving each of them a different colour which was not used to colour @xmath177 .
this is possible since @xmath265 .
then we colour intervas in @xmath92 by modulo @xmath8 method as in ( [ mod.d ] ) , but starting with colours which have not been used to colour intervals in @xmath79 . in this way
we get a modulo @xmath8 colouring of @xmath74 , which is @xmath271-homogeneous , hence also @xmath31-homogeneous .
this completes the proof of theorem [ c.t1 ] .
here we analyze the role of the previsibility assumption in theorem [ c.t1 ] . throughout this section
we take @xmath272 , @xmath273 , and @xmath274 with @xmath275 and @xmath276 we will define a sequence of collections @xmath277 , of size @xmath278 .
the initial collection @xmath279 is of size @xmath8 , hence
up to permutation it has a unique @xmath31-homogeneous colouring .
then we will check that for @xmath280 , there is a unique @xmath31-homogeneous colouring of @xmath281 keeping the previously determined @xmath31-homogeneous colouring of @xmath282 .
finally , we will see that there is no @xmath31-homogeneous colouring of @xmath283 keeping the previously determined @xmath31-homogeneous colouring of @xmath284 . in our example
below , the parameter @xmath8 determines the size of the initial collection @xmath279 , while the parameter @xmath65 determines the number of steps needed to arrive to a problem of consistent colouring without solution . to define the sequence of collections in question ,
take a chain of dyadic intervals @xmath285 then @xmath286 , and let @xmath287 be the dyadic brother of @xmath288 in @xmath289 , @xmath290 .
thus @xmath291 .
now , take two sets of intervals from @xmath292 : @xmath293 @xmath294 consider the following sequence of collections : @xmath295 [ e.s1 ] the sequence of collection @xmath281 , @xmath296 defined by ( [ w.k ] ) is increasing and it has the following properties : * stage @xmath297 .
there exists exactly one up to permutation
@xmath31-homogeneous colouring of @xmath279 as @xmath298 * stage @xmath240 , @xmath280 .
let @xmath299 be the @xmath31-homogeneous colouring of @xmath282 , obtained at stage @xmath300 .
then there exists exactly one @xmath31-homogeneous colouring of @xmath281 as @xmath301 such that @xmath302 * stage @xmath303 .
let @xmath304 be the @xmath31-homogeneous colouring of @xmath284 , obtained at stage @xmath305 .
there does not exist an @xmath31-homogeneous colouring of @xmath283 as @xmath306 such that @xmath307 [ [ proof .- verification - of - a . ] ] proof .
verification of ( a ) .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + consider possible colourings of @xmath279 .
take @xmath308 as a testing interval .
observe that @xmath309 , so if we want to have @xmath31-homogeneity , we must have and therefore @xmath310 for each @xmath311 . without loss of generality we can assume that @xmath312 has colour 1 , and each @xmath313 has colour @xmath314 , @xmath315 .
therefore for @xmath279 and each testing interval @xmath316 we have @xmath317 for each @xmath131 . [ [ the - basic - observation . ] ] the basic observation .
+ + + + + + + + + + + + + + + + + + + + + + our example is based on iterating systematically the following basic observation .
let @xmath318 .
assume that @xmath281 has an @xmath31-homogeneous decomposition as @xmath319 so that @xmath320 then necessarily @xmath321 [ [ verification - of- . ] ] verification of .
+ + + + + + + + + + + + + + + + + we know already that @xmath312 has to have colour @xmath322 .
to check the claim for @xmath323 , @xmath324 we consider the pair of collections @xmath325 : @xmath326 and testing interval @xmath327 .
elements of @xmath279 included in @xmath327 are @xmath328 .
in addition , @xmath329 , while @xmath330 . therefore we have @xmath331 therefore , of the @xmath31-homogeneity condition for @xmath281 implies that @xmath323 is of colour 1 .
[ [ verificaton - of - b . ] ] verificaton of ( b ) .
+ + + + + + + + + + + + + + + + + + + recall that @xmath296 @xmath332 moreover , by , the only possible @xmath333-homogeneous decomposition of @xmath281 is @xmath334 let s check that for @xmath335 , the above decomposition of @xmath281 is indeed @xmath333-homogeneous .
we present the detailed proof for @xmath336 , since the cases @xmath337 are fully analogous .
first , take as a testing interval @xmath338 , @xmath339 .
then elements of @xmath284 included in @xmath338 are @xmath328 and @xmath340 .
therefore @xmath341 and @xmath342 therefore @xmath343 next take as a testing interval @xmath344
. then elements of @xmath284 included in @xmath345 are @xmath328 , so @xmath346 , @xmath347 therefore @xmath344 also satisfies of the @xmath333-homogeneity condition for @xmath284 .
consequently , @xmath348 also satisfy these conditions .
finally , take as a testing interval @xmath349 , @xmath350 .
the only element of @xmath284 included in @xmath349 is @xmath351 , so @xmath352 , and more precisely @xmath353 thus , @xmath349 ( and consequently , each testing interval included in @xmath349 ) satisfies of the @xmath333-homogeneity condition for @xmath284 .
[ [ verification - of - c . ] ] verification of ( c ) .
+ + + + + + + + + + + + + + + + + + + + consider @xmath284 and @xmath354 . recall that @xmath355 take @xmath308 as a testing interval .
all intervals from @xmath283 are included in @xmath308 , and the colouring yields @xmath356 for @xmath283 and @xmath308 we have to consider of the @xmath333-homogeneity condition .
but the above formulae mean that for @xmath283 and testing interval @xmath308 , the condition is satisfied with @xmath357 , but not with @xmath274 .
[ [ remark.-2 ] ] remark .
+ + + + + + + for @xmath335 , the pair of collections @xmath358 , where @xmath359 is not @xmath8-previsible .
nevertheless , the colouring problem has a solution for @xmath360 .
the examples of proposition [ e.s1 ] grew out of the couterexamples to classical martingale inequalities , see @xcite , or @xcite .
the problem of consistent colourings gives rise to the following two - person game .
the game is played by two players with collections of coloured dyadic intervals in @xmath361 for a fixed @xmath362 .
it starts by fixing @xmath363 @xmath364 and a subcollection @xmath365 with with an @xmath31-homogeneous colouring @xmath366 according to definition [ c.d1 ] .
the rules of the game are as folows : 1 . in the first stage ,
player a chooses a collection @xmath367 and @xmath368 .
player b determines an @xmath31-homogeneous colouring of @xmath369 that preserves the colours of @xmath370 .
2 . at stage
@xmath303 , player a chooses @xmath371 and @xmath372 .
player b determines an @xmath31-homogeneous colouring of @xmath373 preserving the colours of @xmath374 .
the game stops at stage @xmath303 if either @xmath375 , and then player b is the winner , or else if there does not exist an @xmath31-homogeneous colouring of @xmath373 that preserves the colours of @xmath374 .
in the second case , player a is the winner .
the results of this paper are able to predict the outcome of the game as follows .
if we do not pose any constraints on the choice of the collections @xmath281 , then the example in section [ c.s3 ] and proposition [ e.s1 ] describes a realization of our game where player a has a strategy of winning . however ,
if we restrict the moves of player a by imposing that @xmath376 is @xmath8-previsible , then with the aid of theorem [ c.t1 ] and its proof , player b has always a winning strategy . in case
the moves of player a are restricted by @xmath8-previsibility , we modify the stopping rule accordingly : player b is the winner at stage @xmath303 if there does not exist @xmath377 so that @xmath378 and the pair @xmath379 is @xmath8-previsible .
t. figiel , _ singular integral operators : a martingale approach_. in : geometry of banach spaces ( strobl , 1989 ) , 95-110 , london math .
lecture note ser .
, 158 , cambridge univ . press , cambridge , 1990 .
j. lee , p. f. x. mller , s. mller , _ compensated compactness , separately convex functions and interpolatory estimates between riesz transforms and haar projections_. comm .
partial differential equations 36 ( 2011 ) , no .
4 , 547-601 .
j. lindenstrauss , l. tzafriri , classical banach spaces .
ii . function spaces .
ergebnisse der mathematik und ihrer grenzgebiete [ results in mathematics and related areas ] , 97 .
springer - verlag , berlin - new york , 1979 p. f. x. mller , isomorphisms between @xmath380 spaces .
instytut matematyczny polskiej akademii nauk .
monografie matematyczne ( new series ) [ mathematics institute of the polish academy of sciences .
mathematical monographs ( new series ) ] , 66 .
birkhuser verlag , basel , 2005 .
e. m. stein , topics in harmonic analysis related to the littlewood - paley theory .
annals of mathematics studies , no .
63 princeton university press , princeton , n.j . ; university of tokyo press , tokyo 1970 . | we study the problem of consistent and homogeneous colourings for increasing families of dyadic intervals .
we determine when this problem can be solved and when not .
msc 2010 : 05a18 , 91a05 key words : coloured dyadic intervals , consistent and homogeneous colouring , 2-person games |
the large area telescope ( lat ) on board the satellite ( atwood et al .
2009 ) detected 1451 sources in the @xmath2ray band above 100 mev with a significance @xmath64.5@xmath3 during its first 11 months survey ( abdo et al .
2010 , a10 hereafter ) .
this is the 1fgl catalogue and contains both galactic and extragalactic @xmath2ray sources . in a recent paper abdo et al .
( 2010a , a10a hereafter ) classified 831 out of 1451 1fgl sources as blazars .
these form the 1lac catalogue . considering the 796 sources in the 1lac associated with a single counterpart
, there are 37 sources generically classified as agn , 314 bl lac and 285 flat spectrum radio quasars ( fsrq ) , classified according to their optical spectrum .
the remaining 160 sources are candidate blazars but of unknown " class because of lacking an optical spectrum or , if available , because of its poor quality for the optical classification . however , for these sources a counterpart could still be found in one of the radio catalogues adopted for the source identification by a10a ( crates haeley et al . 2007 ; cgrabs healey et al .
2008 ; the bzcat massaro et al
thirty five sources in the 1lac sample have more than one associated counterpart . the 1fgl sample is equally populated in the northern ( 713 sources ) and southern sky ( 738 sources ) .
however , only 50% of the 1fgl sources in the south sky are classified as blazars in the 1lac catalogue with respect to @xmath770% of the 1fgl northern sources which are classified as blazars in the 1lac catalogue .
the recently published australia telescope 20 ghz survey ( at20 g , murphy at al .
2010 ) represents the largest catalogue of high frequency radio sources detected with the australia telescope compact array ( atca ) in a survey conducted from 2004 to 2008 and covering the whole sky south of declination 0@xmath8 .
it contains 5890 radio sources with flux at 20 ghz exceeding 40 mjy and it is complete at a level of 91% above 100 mjy beam@xmath9 . for several sources the fluxes at 8 and 5 ghz are also measured .
the at20 g survey , being conducted at relatively high radio frequencies ( 20 ghz ) , most likely detects the radio emission from the compact cores of agn .
several sources in this survey have flat or inverted radio spectra between 5 and 20 ghz .
we correlate the at20 g radio catalogue with the 1fgl in search for possible counterparts .
this allows us to study the radio versus @xmath2ray properties of these associations .
a possible correlation between the radio and the @xmath2ray emission in radio loud agn can shed light on the physical link between the emission processes in these two energy bands .
the @xmath10 spectra of both fsrq and bl lac objects show two peaks which are widely interpreted as due to the synchrotron and inverse compton emission , respectively .
the spectral energy distribution ( sed ) of blazars form a sequence ( so called blazar sequence " , e.g. fossati et al .
1998 ) with the most powerful sources ( fsrqs ) having the synchrotron and inverse compton peaks in the ir and mev energy range respectively , while the less powerful bl lacs have the synchrotron and inverse compton peak in the optical / uv ( or even x ray ) and gev tev energy range respectively .
the blazar sequence was built by dividing blazars into bins of radio luminosity , thought to be a proxy for the bolometric one , and establishes a link between the radio and the @xmath2ray emission . before
, no conclusive claim could be made on the existence of a possible correlation between radio and @xmath2ray properties for egret sources ( e.g. mucke et al .
taylor et al .
( 2007 ) , by studying the radio properties of egret detected blazars with the vlba imaging and polarimetry survey ( vips extending down to 85 mjy ) found that at low radio fluxes the radio flux density does not directly correlate with the @xmath2ray flux . further investigation of this correlation was possible with the source list of the first three months survey ( lbas , abdo et al .
2009 ; giroletti et al . 2010 ) : while a correlation between the radio flux at 8.4 ghz and the peak flux ( above 100 mev ) appears when considering bl lacs and fsrq together , the statistical evidence is only marginal for fsrqs ( chance correlation probability of 8% ) , being more robust only for bl lacs ( 0.5% ) .
such correlation analysis should consider the different redshift range spanned by these two populations ( e.g. mucke et al .
abdo et al . (
2009 ) found a separation in the @xmath2ray spectral index versus radio luminosity among fsrqs , bl lacs and misaligned agn which suggests that more luminous radio sources have softer @xmath2ray spectra .
recent studies of the radio@xmath2 flux correlation in the lbas sources ( kovalev et al .
2009 ; 2009a ) was conducted using the mojave sample of extragalactic sources ( with a flux limit of 1.5 jy at 15 ghz ) .
kovalev et al .
( 2009 ) find that the parsec scale radio emission and the @xmath2ray flux are strongly related in bright @xmath2ray objects , suggesting that selects the brightest objects from a flux density limited sample of radio loud sources . in this paper
we search for the possible counterparts of the 1fgl sample in the at20 g survey ( 1 ) . by intersecting the association results with the 1lac sample we study the radio vs @xmath2ray properties of these associations ( 2 ) .
at20 g correlation also reveals 8 new associations which are unclassified sources in the 1fgl sample . for two of these
there are observations which , combined with the radio and data , allow to build the broad band sed .
one of these two sources is classified here as a new candidate fsrq ( 3 ) .
we also present , as a first result of an on going study of the properties of the 46 unknown sources of the 1lac sample , the sed of the brightest radio source of the uis class that we associate with one 1fgl source .
we classify this object as a bl lac due to the similarity of its sed to that of oj 287 ( 4 ) .
we discuss our findings in 5 .
one method used to find counterparts ( abdo et al .
2009 , 2010 , 2010a ) relies on positional coincidence ( sutherland & saunders 1992 ) and it is based on bayes statistics . to each association between a source and a candidate counterpart it is assigned a posterior probability based on the counterparts space density around the source . only those associations with a posterior probability larger than a chosen value are considered as likely associations . considering only high latitude sources ( @xmath11 ) the 1lac sample ( abdo et al .
2010 ) contains 709 agn counterparts ( i.e. also multiple counterparts within the source error ellipse ) associated to 671 1fgl sources at high galactic latitudes with posterior probabilities @xmath12% . among these ,
663 have posterior probabilities @xmath680% and 599 are defined as `` clean '' associations if ( i ) they have one single associations ; ( ii ) the posterior probability is larger than 80% and ( iii ) are not flagged as problematic ( e.g. marginal detection or data problems ) in the 1fgl sample . the method adopted by a10a to build the 1lac sample is implemented in
the ` gtsrcid ` tool distributed with the ` lat sciencetools - v9r15p2 ` software .
the association method searches for single radio sources falling within the confidence ellipse of the sources .
we applied this method in cross correlating the at20 g survey catalogue with the 1fgl one . similarly to the 1lac sample we accept only single associations ( i.e. only one radio counterpart falls within the the source error ellipse ) and with a posterior probability @xmath680% . to compute the posterior probability we have to assign the prior probability to the at20 g sample .
this represents the confidence that the at20 g catalogue contains the real counterparts of the sources .
we compute the prior probability of the at20 g catalogue following the procedure described in appendix c of a10 .
the prior probability represents the probability of finding a number of false identification from the correlation of the 1fgl and at20 g real catalogues which is equal to the average number of associations found by the cross correlation of the at20 g sample with randomly generated catalogues of sources . to this aim we created a set of 100 fake catalogues by displacing the sources between 2 and 10 degrees from their real positions . for sources near the galactic plane a smaller displacement in galactic latitude
was adopted ( eq .
c3 of appendix 2 of a10 ) .
we then computed the number of false associations between the at20 g and 1fgl as a function of the prior probability assigned to the at20 g catalogue ( dashed line in fig .
[ fg1 ] ) and the average number of associations between the at20 g and 100 1fgl random catalogues for each value of the prior probability ( solid line in fig .
[ fg1 ] ) .
the intersection between the two curves in fig .
[ fg1 ] gives the value of the prior probability adopted for the at20 g sample .
-1.4 cm -0.5 cm from the cross correlation of the at20 g and the 1fgl catalogue we found 230 associations for the 738 1fgl sources distributed in the southern sky , i.e in the same sky portion covered by the at20 g survey . among these associations
222 sources are already classified as blazars in the 1lac sample : 10 are generically classified as agn , 54 are bl lac , 112 are fsrq and 46 are of unknown type
. there are 8 remaining sources which are newly found associations , i.e. they are not present in the 1lac sample . tab .
3 lists the properties of the 222 1fgl at20 g associations which are classified as blazars in the 1lac sample . in tab .
2 we report the 8 new associations . for each association
we give its posterior probability , the separation between the 1fgl source centroid and the position of the at20 g associated source , the 1fgl and at20 g names , the name of the agn that was associated to the 1fgl source in the 1lac sample and its optical classification ( agn , bll , fsrq or unknown ) and if the 1lac association was classified as clean " or not .
we note that in the generic class of agn there are several types of sources : starburst galaxies ( ngc 253 abdo et al . , 2010b )
, starburst / seyfert 2 ( ngc 4945 lenc & tingay , 2009 ) a low excitation fri radio galaxy ( pks 062535 gliozzi et al . , 2008 ) a high excitation fri radio galaxy ( cen a ) , a narrow line seyfert 1 ( pks 2004447 abdo et al . , 2009b ) .
in the following plots these sources are labelled agn " to distinguish them from the class of bl lacs and fsrq .
we do not enter in the details of the properties of these sub classes of agns because it is out of the scope of the present paper .
[ fg2 ] shows the angular separation between the 1fgl and the at20 g sources for the 230 associations found as a function of the detection significance ( left panel ) and of the posterior probability of the association ( right panel ) .
the angular separation between the 1fgl sources and the associated at20 g counterparts decreases as a function of the detection significance .
this behaviour is common among the different blazar types .
the number of associations concentrates to low significance values where also the association separation is larger . from fig .
[ fg2 ] it appears also that , although our association methods selected only those with a posterior probability @xmath680% , most of the associations found have a probability distribution ( right panel of fig .
[ fg2 ] ) that is skewed towards high probabilities ( @xmath13% ) suggesting that most of sources with relatively low detection significance are found in associations with a high posterior probability .
in fig . [ fg3 ] we compare the distribution of the 20 ghz fluxes of all the 5890 sources of the at20 g sample with the distribution of the radio flux of those sources associated with one source of the 1fgl catalogue .
the 1fgl sources select the radio counterparts with the largest fluxes within the flux limited at20 g survey .
the kolmogorov smirnov ( ks )
test gives a probability of 1.9% that the two distributions are similar .
selecting the associations that are blazars of known type we compare the distributions of radio fluxes of bl lacs and fsrqs . in the lbas sample ,
these were found ( abdo et al . 2009 ) to be different ( considering the 8.4 ghz fluxes ) . using our sample and the 20 ghz flux
we also find that these two populations have different radio fluxes with fsrq being on average brighter than bl lacs ( the chance probability that the two distribution are consistent being only 1.5@xmath14 ) . the radio vs @xmath2ray properties of the first 3 months catalogue ( abdo et al .
2009 ; giroletti et al . 2010 ) showed that there is a possible correlation between the 8.4 ghz radio flux and the @xmath2ray flux .
however , this correlation was found to be statistically significant ( chance probability of 0.05% ) only for the population of bl lacs , while fsrq showed only a marginal correlation ( with a chance probability of 8% ) . without separating the source types , kovalev ( 2009 ; 2009a )
finds that the parsec scale radio emission and the @xmath2ray flux are strongly related in the bright @xmath2ray objects of the lbas sample .
similar results are found with the preliminary one year sample by giroletti et al .
( 2010 ) . through our large sample of associations ( 230 sources ) we can extend this correlation analysis in particular relying on the lower limiting flux ( 40 mjy ) at the higher frequency ( i.e. 20 ghz ) of the at20 g survey .
[ fg4 ] shows that there is indeed a strong correlation between the radio flux at 20 ghz and the @xmath2ray flux ( above 100 mev ) for the associations found from the cross correlation of the at20 g survey with the 1fgl/1lac catalogue ( 1 ) . .
correlation between the @xmath2ray flux and the 20 ghz radio flux density for all 230 associations and for different sub samples .
the spearman correlation coefficient @xmath15 and the chance correlation probability @xmath16 are given .
the correlation is fitted with a function @xmath17 and the slope is given in column 4 . [ cols="<,^,^,^",options="header " , ] | we cross correlate the 11 months survey catalogue ( 1fgl ) with the 20 ghz australia telescope compact array radio survey catalogue ( at20 g ) composed by 5890 sources at declination @xmath00@xmath1 . among the 738 sources distributed in the southern sky we find 230 highly probable candidate counterparts in the at20 g survey . of these
, 222 are already classified as blazars ( 176 of known type and 46 of unknown optical class ) in the 1year lat agn catalogue ( 1lac ) and 8 are new associations . by studying the @xmath2ray and radio properties of these associations we find a strong correlation between the @xmath2ray flux ( above 100 mev ) and the 20 ghz flux density .
this correlation is more than 3@xmath3 statistically significant both for the population of bl lacs and of fsrq considered separately .
we also find that the radio counterparts associated to the sources have on average flat radio spectra between 5 and 20 ghz and that @xmath2ray sources are not preferentially associated with ultra inverted spectrum " radio sources . for 2 of the 8 new associations we build the broad band spectral energy distribution combining , and radio observations .
one of these two sources is identified with the high redshift fsrq swift j1656.33302 ( @xmath4 ) and we classify the other source as a candidate new fsrq .
we also study the brightest radio source of the 46 associations without an optical classification and classify it as a new bl lac candidate
twin " of the prototypical bl lac oj 287 if its redshift is somewhat larger , @xmath50.4 .
bl lacertae objects : general quasars : general radiation mechanisms : non thermal gamma - rays : theory x - rays : general radio continuum : general |
.the standard genetic code . assignment of the 64 possible codons to amino acids or stop signals , with polar requirement of the amino acids indicated in brackets . [ cols= " < , < , < , < " , ] [ tab : tab7 ] in table [ tab : tab7 ] we have listed the possible ways to fill a single box that are compatible with the considered trna wobble rules .
let @xmath2 enumerate the possible trna patterns as listed in the rightmost column of table [ tab : tab7 ] .
we write @xmath3 , @xmath4 , @xmath5 for the number of amino acids , stop codons and unassigned codons present in pattern @xmath6
. + + * problem .
* we now consider the problem of filling 16 boxes ( 64 codons in total ) using 20 different amino acids , @xmath7 stop codons and @xmath8 unassigned codons .
it is useful to solve a slightly more general problem : the number of ways to fill @xmath9 boxes using * @xmath10 amino acids , * each of the first @xmath11 amino acids at least once , * exactly @xmath7 stop codons , and * exactly @xmath8 unassigned codons .
the original problem is obtained by setting @xmath12 and @xmath13 . + + * recurrence . * we denote the number of such fillings by @xmath14 and compute their values by the recurrence @xmath15 with basis @xmath16 * rationale . *
the reasoning behind is the following .
we fill box number @xmath9 first , and worry about the remaining boxes later .
we iterate over the possible trna patterns with variable @xmath17 .
to realise pattern @xmath18 we need @xmath19 amino acids , @xmath20 stop codons and @xmath21 unassigned codons .
there is only one way to choose stop codons and unassigned codons , but we can obtain the amino acids from two sources . we can take some from the @xmath11 still - to - use amino acids that we have to use at least once , and we must take the others from the @xmath22 free amino acids that can be used as desired .
we consider all possible ways to realise the choice : we first iterate over the number of amino acids that we take from the still - to - use pool with variable @xmath23 . selecting @xmath23 out of @xmath11 still - to - use amino acids can be done in @xmath24 ways .
similarly , taking the remaining @xmath25 amino acids from @xmath22 free amino acids can be done in @xmath26 ways .
all these @xmath19 chosen amino acids are different , and so there are @xmath27 ways to instantiate the pattern using them .
now we still have to fill the remaining @xmath28 boxes , using the remaining @xmath29 still - to - use amino acids at least once , while using exactly @xmath30 stop codons and leaving @xmath31 codons unassigned .
+ + * implementation .
* the value @xmath32 can be efficiently evaluated by dynamic programming .
this is achieved by storing all intermediate values of @xmath33 that are computed in memory , and recalling them when they are needed instead of reevaluating @xmath33 .
this way , @xmath14 can be evaluated in time and space @xmath34 .
note that a single call to @xmath14 computes @xmath35 for many @xmath36 , @xmath37 , @xmath38 and @xmath39
. + + * sampling . *
the above dynamic programming implementation has the advantage that it allows uniform sampling over the space of all codes .
we first sample a number uniformly between @xmath40 and @xmath14 .
then we use the recurrence in reverse to determine which code this number corresponds to .
this is done as follows .
say the number sampled was @xmath41 .
we then incrementally evaluate the sum of .
once the partial sum up to @xmath17 surpasses @xmath41 , we know that pattern @xmath18 was used in code number @xmath41 . similarly we decode which amino acids are used and in which order they are placed . by explicitly keeping track of the set of still - to - use amino acids we can retrieve the entire code recursively .
we gratefully acknowledge steven de rooij for taking part in the effort to determine the global minimum .
furthermore we gratefully acknowledge the large improvements of the manuscript by comments from paul g. higgs and two anonymous referees .
novozhilov as , wolf yi , koonin ev , evolution of the genetic code : partial optimization of a random code for robustness to translation error in a rugged fitness landscape , _ biology direct _ 2 , 24 ( 2007 ) .
takai k , classification of the possible pairs between the first anticodon and the third codon positions based on a simple model assuming two geometries with which the pairing effectively potentiates the decoding complex , _ journal of theoretical .
biology _ 242 , 564580 ( 2006 ) .
massey se , searching of code space for an error - minimized genetic code via codon capture leads to failure , or requires at least 20 improving codon reasignments via the ambiguous intermediate mechanism , _
_ 70 , 106115 ( 2010 ) . | the genetic code has been shown to be very error robust compared to randomly selected codes , but to be significantly less error robust than a certain code found by a heuristic algorithm .
we formulate this optimisation problem as a quadratic assignment problem and thus verify that the code found by the heuristic is the global optimum .
we also argue that it is strongly misleading to compare the genetic code only with codes sampled from the fixed block model , because the real code space is orders of magnitude larger .
we thus enlarge the space from which random codes can be sampled from approximately @xmath0 codes to approximately @xmath1 codes .
we do this by leaving the fixed block model , and using the wobble rules to formulate the characteristics acceptable for a genetic code . by relaxing more
constraints three larger spaces are also constructed . using a modified error function ,
the genetic code is found to be more error robust compared to a background of randomly generated codes with increasing space size .
we point out that these results do not necessarily imply that the code was optimized during evolution for error minimization , but that other mechanisms could explain this error robustness .
genetic code , error robustness , origin of life . |
the measurement of hadrons produced in relativistic heavy - ion collisions is a well established tool in the study of the hot and dense matter created in the collisions .
the phenix experiment at the relativistic heavy ion collider ( rhic ) has carried out systematic measurement of hadrons in @xmath1 , @xmath2au , cu@xmath3cu and au@xmath3au collisions at @xmath4 = 200 gev . when compared to existing measurements in @xmath1 and @xmath2au , measurements in heavy - ion collisions
suggest that particle production at high @xmath8 is affected by jet quenching , which is considered to be an effect of extremely dense matter created by the collisions @xcite .
high @xmath8 suppression of @xmath5 and @xmath6 was measured in cu@xmath3cu and au@xmath3au @xcite and the nuclear modification factors ( @xmath7 ) of these mesons were found to be consistent with each other in @xmath8 and centrality .
a comparison with theoretical models was first done for @xmath5 suppression in @xcite , with the result that the suppression increases proportional to the number of participating nucleons as @xmath9 .
this result is consistent with existing energy loss models such as the parton quenching model ( pqm ) @xcite .
lcccccc data set & trigger & sampled events & @xmath10 & threshold & decay channel & reference + 2003 @xmath2au & ert & 5.5b & 2.74 nb@xmath11 & 2.4 gev & @xmath12 & prc75 @xcite + + & & & & 2.4 gev & @xmath13 & prc75 @xcite + + 2004 au@xmath3au & mb & 1.5b & 241 @xmath14b@xmath11 & n / a & @xmath15 & this work + + 2005 @xmath1 & ert & 85b & 3.78 pb@xmath11 & 0.4 gev & @xmath16 & prd83 @xcite + & & & & 1.4 gev & @xmath12 & prd83 @xcite + & & & & 1.4 gev & @xmath13 & prd83 @xcite + + 2005 cu@xmath3cu & mb & 8.6b & 3.06 pb@xmath11 & n / a & @xmath13 & this work + & ert & & & 3.4 gev & @xmath13 & this work + + 2007 au@xmath3au & mb & 5.1b & 813 @xmath14b@xmath11 & n / a & @xmath15 & this work + + 2008 @xmath2au & ert & 160b & 80 nb@xmath11 & 0.6/0.8 gev & @xmath17 & this work + & & & & 2.4 gev & @xmath12 & this work + & & & & 2.4 gev & @xmath13 & this work + the @xmath0 meson comprises light valence quarks similar to the @xmath5 and @xmath6 , but has a larger mass ( 782 mev ) and a spin ( 1 ) .
these differences make the omega measurement an additional probe to a systematic study to understand mechanisms of parton energy loss and hadron production in the collisions .
the @xmath8 dependence of the particle production ratio ( @xmath18 ) and the nuclear modification factors ( @xmath7 ) should add information about the parton energy loss mechanism .
furthermore , using multiple decay channels : a leptonic channel @xmath19 ( with branching ratio [email protected]@xmath2110@xmath22 ) and two hadronic decay channels @xmath12 ( br=([email protected])@xmath2110@xmath23 ) and @xmath13 ( br=(8.90 + 0.27 - 0.23)@xmath2110@xmath23 ) @xcite extends the @xmath8 range by using the hadronic channels at high @xmath8 and the leptonic channel at low @xmath8 .
baseline measurements of the @xmath0 have been performed for @xmath1 via the leptonic channel @xcite , and for the @xmath1 and @xmath2au in the hadronic channel @xcite .
the @xmath24 ratio was found to be independent of transverse momentum and equal to @xmath25 in @xmath1 and @xmath26 in @xmath2au collisions for @xmath27 gev/@xmath28 @xcite .
this article presents the first measurements of @xmath0 meson production in cu@xmath3cu and au@xmath3au collisions at phenix via the @xmath29 channel .
these measurements permit the study of @xmath0 suppression at high @xmath8 .
this paper also presents measurements of @xmath0 in @xmath2au collisions with significantly reduced uncertainties in the hadronic channel and a first measurement in the dielectronic channel .
the phenix experiment is designed specifically to measure electromagnetic probes such as electrons , muons , and photons @xcite .
the detectors of the phenix experiment can be grouped into three categories : inner detectors close to the beam pipe , two central arms with pseudorapidity coverage of @xmath200.35 , each covering 90 degrees in azimuthal angle , and two muon detectors , which have 2@xmath30 azimuthal and pseudorapidity coverage of + ( 1.22.2 ) for the south muon arm and -(1.22.4 ) for the north muon arm .
the central arms are used to measure the @xmath0 mesons at midrapidity .
the inner detectors are used for triggering , measurement of the @xmath31-coordinate of the collision vertex and centrality of the interactions with beam - beam counters ( bbc ) and zero degree calorimeters ( zdc ) .
the central arms are capable of measuring a variety of particles by using drift chambers ( dc ) and pad chambers ( pc ) for tracking and momentum measurement of charged particles , ring imaging erenkov detectors ( rich ) for the separation of electrons up to the @xmath30 erenkov threshold at 4 gev/@xmath28 , and an electromagnetic calorimeter ( emcal ) for the measurement of spatial positions and energies of photons and electrons .
the emcal comprises six sectors of lead scintillator calorimeter ( pbsc ) and two sectors of lead glass calorimeter ( pbgl ) .
additional details of the phenix experimental setup and performance of the detector subsystems can be found elsewhere @xcite .
we used data samples collected in 2004 , 2005 , 2007 , and 2008 as summarized in table [ table : summarydata ] .
the data were taken using a minimum bias trigger ( mb ) and the emc - rich - trigger ( ert ) , which is described below .
the 2003 @xmath2au data were published in @xcite and are included here for comparison .
the 2005 @xmath1 data were published in @xcite and are used as the baseline for @xmath7 in @xmath2au , cu@xmath3cu and au@xmath3au .
two au@xmath3au data samples were taken in 2004 and 2007 .
the mb trigger required a coincidence between the north and south bbc @xcite . in the au@xmath3au data sample taken in 2004 , additional coincidence between the zdc and bbc
was required . to enhance the statistics at high @xmath8 , the ert trigger was used in @xmath1 , @xmath2au and cu@xmath3cu runs which required : 1 ) the event to satisfy the mb trigger conditions ; 2 ) the presence of at least one high-@xmath8 electron or photon candidate in the event . for electron candidates the ert trigger required a minimum energy deposit of 0.4 ( 0.6 and 0.8 ) gev/@xmath28 in a tile of @xmath32 emcal towers matched to a hit in the rich in @xmath1 ( @xmath2au ) collisions . for the photon candidates the ert trigger required a minimum energy deposit of 1.4 , 2.4 and 3.4 gev/@xmath28 in a tile of @xmath33 emcal towers in @xmath1 , @xmath2au , and cu@xmath3cu collisions , respectively . in the @xmath2au and the cu@xmath3cu analysis ,
the mb data set was used to measure @xmath0 production up to 4 gev/@xmath28 in @xmath2au and 6 gev/@xmath28 in cu@xmath3cu ; the ert sample was used at higher @xmath8 .
the ert trigger efficiencies measured for single photons and electrons and calculated for @xmath0 mesons is described in section [ sec : ana : rec ] .
in this section , we describe the event selection and data analysis for reconstructing the leptonic ( @xmath17 ) and hadronic ( @xmath12 and @xmath15 ) decay channels of the @xmath0 .
corrections applied to the raw data to calculate the @xmath0 meson invariant yields and systematic uncertainties related to the measurements are also presented . in the run-4 phenix configuration ,
the correlation of the charge deposited in the bbcs with energy deposited in the zdcs provides a determination of the centrality of the collisions . in the other runs ,
the centralities were only determined by using bbc .
a glauber monte carlo @xcite with the bbc and zdc responses was used to estimate the number of binary nucleon - nucleon collisions ( @xmath34 ) and the number of participating collisions ( @xmath35 ) for each centrality bin @xcite .
events are selected with a reconstructed @xmath31-vertex within 30 cm of the center of the interaction region .
charged tracks were required to have momenta in the range of @xmath36 ( 7.0 ) gev/@xmath28 for the @xmath17 analysis in @xmath1 ( @xmath2au ) @xcite and @xmath37 gev/@xmath28 for the @xmath38 decay channel @xcite . charged particles with @xmath39 gev/@xmath28 have a large bending angle in the axial magnetic field of the phenix central magnet @xcite and most of them do not pass through the entire tracking system .
electrons and positrons are identified mainly by the erenkov photons emitted in the rich by requiring at least two photomultipliers hit in the rich cells matched to the track @xcite .
also , matching of the energy measured for the charged track in the emcal with the momentum measured in the tracking system , @xmath40 , helps to further improve @xmath41 separation .
together the rich and emcal provide an @xmath41 rejection factor of about 1:10@xmath42 .
photon identification is performed by the shower shape criteria in the emcal @xcite , and the energy of the selected @xmath43 clusters is above 0.2 gev .
the leptonic analysis is done only in @xmath1 and @xmath2au . in case of @xmath17 , all electrons and positrons reconstructed in each event
are combined into pairs , resulting in signal peaks which sit on top of a combinatorial background in the invariant mass distribution .
the uncorrelated part of the background is estimated with an event - mixing technique , which combines tracks from different events with similar event centrality and @xmath31-coordinate of the collision vertex .
details of the event mixing procedure are presented in @xcite .
invariant mass of @xmath44 pairs detected by the phenix central arms in @xmath1 collisions ( left ) and minimum - bias @xmath2au collisions ( right ) at @xmath4=200 gev and integrated over @xmath8 .
uncorrelated combinatorial background is subtracted as described in the text .
the spectrum is fit to the @xmath0 and @xmath45 resonances where the masses and widths are set to the pdg values ; the breit - wigner resonance shape is convolved with a gaussian to account for detector mass resolution estimated from simulation and then corrected for the radiative tail .
the @xmath46 contribution is shown as the dotted line with an assumption that the yield is the same as that of the @xmath0 .
the residual continuum component is estimated by a polynomial fit as shown by the dashed line .
, title="fig : " ] invariant mass of @xmath44 pairs detected by the phenix central arms in @xmath1 collisions ( left ) and minimum - bias @xmath2au collisions ( right ) at @xmath4=200 gev and integrated over @xmath8 .
uncorrelated combinatorial background is subtracted as described in the text .
the spectrum is fit to the @xmath0 and @xmath45 resonances where the masses and widths are set to the pdg values ; the breit - wigner resonance shape is convolved with a gaussian to account for detector mass resolution estimated from simulation and then corrected for the radiative tail .
the @xmath46 contribution is shown as the dotted line with an assumption that the yield is the same as that of the @xmath0 .
the residual continuum component is estimated by a polynomial fit as shown by the dashed line .
, title="fig : " ] figure [ fig : massspectra_ee_pp ] shows invariant mass spectra of @xmath44 pairs in @xmath1 and @xmath2au collisions at @xmath4 = 200 gev after subtraction of combinatorial background as described above .
the solid lines show the global fits which include : ( 1 ) contributions from @xmath0 , @xmath46 and @xmath45 mesons approximated with breit - wigner functions convolved with gaussian distributions to account for the detector mass resolution ; masses and widths of the @xmath0 , @xmath46 and @xmath45 are fixed to the pdg values ; the @xmath46 component is calculated assuming that @xmath0 and @xmath46 have the same yield and vacuum branching ratios ; ( 2 ) other correlated residual background , which is dominated by a contribution from jets , is approximated by a second order polynomial function .
the detector resolution , which is determined from simulations , is found to be dependent on mass and momentum and varies from 6 mev/@xmath47 to 18 mev/@xmath47 .
the @xmath0 yield is determined by counting bin contents in a 3 @xmath48 width ( derived from the fitting ) and subtracting the polynomial background . an associated systematic uncertainty from the raw yield extraction
is calculated by varying the background normalization , fitting functions , range and counting methods .
the estimated value is 415% in @xmath1 @xcite and 815% in @xmath2au collisions . in the @xmath12 and @xmath13 channels ,
the first analysis step is to reconstruct @xmath5 mesons by combining pairs of photons reconstructed in an event .
then the mass and width of the @xmath5 peak in the invariant mass distribution of photon pairs are parametrized as a function of transverse momentum .
the 1 @xmath48 width of the @xmath5 peak varies from 13 mev/@xmath47 to 9 mev/@xmath47 as @xmath8 increases from 1 gev/@xmath28 to 4 gev/@xmath28 and is determined by the emcal energy resolution .
a pair of photons is selected as a @xmath5 candidate if its invariant mass is within 2 @xmath48 of the reconstructed @xmath5 mass . in cu@xmath3cu and au@xmath3au , an additional asymmetry cut for @xmath5 candidates
is used to reduce combinatorial background , @xmath49 .
selected @xmath5 candidates , which include true @xmath5s and combinatorial background are combined either with the third photon with energy @xmath50 1.0 gev/@xmath28 for the @xmath13 or with a pair of opposite - sign charged tracks for the @xmath12 decay . in the @xmath1 and @xmath2au analysis ,
the @xmath0 meson raw yields are extracted by fitting the @xmath8 slices of the invariant mass distribution with a combination of a gaussian for the signal and a second order polynomial for the background .
the width and mass of the reconstructed @xmath0 mesons were found to be in good agreement with values expected from simulation .
details of these analyses are described in @xcite . in the cu@xmath3cu and au@xmath3au analysis
, only the @xmath15 channel was analyzed due to high combinatorial background in the @xmath12 channel .
the uncorrelated combinatorial background was estimated using an event mixing technique where the third photon in the @xmath29 decay was taken from the different events with a similar centrality and @xmath31-vertex .
for every @xmath8 bin the calculated background was normalized to match the integral of the foreground at an invariant mass @xmath51 gev/@xmath47 and then subtracted .
an example of the invariant mass distribution and normalized background distributions is shown in fig .
[ fig : massspectra_auau](a ) with the invariant mass distribution after subtraction shown in fig .
[ fig : massspectra_auau](b ) .
the resulting invariant mass distribution contains residual background from correlated particles : the background contributions are from @xmath52 decays , and @xmath5 and @xmath6 , where one of the photons from @xmath5(@xmath6)@xmath53 decay creates a fake @xmath5 candidate for the @xmath15 reconstruction .
the @xmath13 peak is further enhanced by a mixed background subtraction .
finally , raw yields of @xmath0 are extracted by fitting the spectra with a combination of a gaussian and a polynomial .
the width of the gaussian is limited @xmath20 1 mev/@xmath47 in the fit to the data as determined from simulation .
the @xmath0 yield is calculated as an integral of the gaussian .
( a ) invariant mass and scaled mixed background distributions for @xmath29 decay at @xmath54 gev/@xmath28 in au@xmath3au collisions ( b ) invariant mass distribution after subtraction of scaled background . ] systematic uncertainties associated with the raw yield extraction are evaluated using different fitting functions and ranges , different counting methods and kinematic cuts , varying the emc resolution in simulation , and applying different limits for the width of @xmath0 peaks in fits to data .
the estimated value is 1335% in cu@xmath3cu and 2035% in au@xmath3au collisions .
the reconstruction efficiency of the @xmath0 is determined using a geant simulation of the phenix detector tuned to reproduce the performance of the detector subsystems .
the @xmath0 mesons are generated and decayed into corresponding decay channels , and reconstructed with the same analysis chain as the real data .
the generated @xmath0 spectra were weighted to match the measured particle spectra .
it was verified that the simulated positions and widths of the reconstructed particle peaks are consistent with the values measured in real data .
typical ert trigger efficiency . left : trigger efficiencies for single electrons ( 0.6 gev threshold ) and photons ( 3.4 gev threshold ) .
right : trigger efficiencies for @xmath17 and @xmath13 using corresponding triggered electrons / photons . ] typical reconstruction efficiencies for @xmath17 and @xmath55 .
the curve for @xmath55 includes the embedding efficiency in au@xmath3au collisions : solid , dotted , and dot - dashed lines , are respectively for 6092% , 2060% , and 020% centrality . ]
the reconstruction efficiency is divided into three components : @xmath56 , @xmath57 , and @xmath58 .
the efficiency @xmath56 is the reconstruction efficiency for minimum bias events in a low occupancy environment , like in @xmath1 and @xmath2au collisions .
this efficiency accounts for the limited geometrical acceptance , resolution and efficiencies of the detector subsystems as well as for analysis cuts . when a selective ert trigger is used , an additional trigger efficiency factor , @xmath57 , is applied .
this factor measures the efficiency of the ert trigger logic . for higher multiplicity collisions
, one needs to account for the loss of efficiency from increased detector occupancy : this is measured through the embedding efficiency @xmath58 .
a measured raw yield then needs to be corrected for the total efficiency @xmath59 , depending on the collision , centrality , and trigger involved .
the ert data sample was used to measure dielectron and hadronic decay channels of the @xmath0 at high @xmath8 in @xmath1 , @xmath2au and cu@xmath3cu .
the threshold settings for ert are described in section [ sec : exp ] .
the single particle ert efficiency was measured by dividing the energy spectra of gamma clusters or electrons that fired the ert trigger by the energy spectra of all clusters or electrons in the minimum bias data sample .
a typical example of the ert trigger efficiencies for single electrons and single photons as a function of cluster energy is shown on the left in fig .
[ fig : erteff ] .
the level of saturation of trigger efficiency curves is below 100% because of inactive areas of the ert and the rich detectors .
the ert efficiencies for the @xmath0 meson in both the leptonic and hadronic decay modes were evaluated with the help of a monte - carlo simulation .
for all fully reconstructed @xmath0 mesons , the calculated single photon or electron ert efficiency curves were used to calculate the probability that one of the particles in the final state fires the ert trigger .
corresponding trigger efficiencies for @xmath17 and @xmath13 are shown on the right in fig .
[ fig : erteff ] .
more detailed descriptions are presented in @xcite .
figure [ fig : totaleff ] shows typical reconstruction efficiencies @xmath56 for @xmath17 and @xmath15 . in the case of cu@xmath3cu and au@xmath3au collisions ,
an additional efficiency correction @xmath58 due to cluster overlap in high multiplicity environment must be applied . in most central au@xmath3au events ,
the emcal typically detects more than 300 clusters corresponding to a detector occupancy of @xmath6010% . to estimate the corresponding loss in efficiency ,
the simulated @xmath0 decays are embedded into real a@xmath3a events and analyzed .
the merging effect results in @xmath6040% loss of reconstruction efficiency in 020% central au@xmath3au collisions , @xmath6015% loss in 020% central cu@xmath3cu collisions and is almost negligible in peripheral collisions .
the reconstruction efficiencies derived for au@xmath3au collisions at different centralities are shown in fig .
[ fig : totaleff ] . finally , in each bin we apply also
a correction factor @xcite to replace the average value of the yield in the analyzed @xmath8 bin by the value of the yield in the middle of the bin . in @xmath1 and minimum bias @xmath2au collisions ,
the invariant yield is related to the invariant cross section as : @xmath61 where @xmath62 and @xmath63 are the total inelastic cross section , 42.2 mb and 2260 mb respectively . for a given centrality bin
the invariant yields as a function of @xmath8 ( invariant transverse momentum ) are determined from : @xmath64 where @xmath65 is the number of events for a given centrality bin , @xmath66 is the raw yield of @xmath0 for each @xmath8 and centrality bin , @xmath67 , @xmath68 and @xmath69 are , as previously defined , reconstruction efficiency , embedding efficiency and trigger efficiency , respectively .
the trigger efficiency is applied only for the analyses using the ert data set .
br is the decay branching ratio from @xcite , ( 89.2@xmath20 0.7@xmath70 ) for @xmath12 , ( 8.90@xmath20 0.27@xmath70 ) for @xmath55 and ( 7.16@xmath20 0.12@xmath71 ) for @xmath17 .
in addition to uncertainties related to the raw yield extraction described in the corresponding analysis sections , other sources of the uncertainties should also be taken into account .
uncertainties of the ert trigger efficiency and acceptance corrections were estimated by varying the analysis cuts , energy and momentum scales of the emcal and dc by @xmath601% @xcite .
uncertainties of detector response ( mainly from the rich for electron analysis and from the emcal for hadronic analysis ) are estimated by changing particle identification criteria in the analysis .
a summary of assigned systematic uncertainties is listed in table [ table : summaryerror1 ] for @xmath17 in @xmath1 and @xmath2au and in table [ table : summaryerror2 ] for @xmath13 in cu@xmath3cu and au@xmath3au . those are classified into three types : type a is @xmath8-uncorrelated , type b is @xmath8-correlated and type c is the overall normalization uncertainty .
total uncertainties for @xmath17 are 1624% in @xmath1 @xcite and 1926% in @xmath2au .
the total uncertainties for @xmath13 are 1537% in cu@xmath3cu and 2137% in au@xmath3au .
uncertainties for @xmath72 analysis are 720% in @xmath1 and 1015% in @xmath2au , as described in @xcite .
figure [ fig : ptspectra_pp_daupt ] presents the invariant transverse momentum spectra measured for the @xmath0 in @xmath1 and @xmath2au at @xmath73=200 gev .
previously published results are shown with open markers @xcite .
results for different decay channels and data samples agree within uncertainties in the overlap region .
the dashed curves in fig .
[ fig : ptspectra_pp_daupt ] are fixed on @xmath1 results at @xmath74 2 gev/@xmath28 and then scaled by the number of binary nucleon - nucleon collisions ( @xmath34 ) estimated using glauber monte - carlo simulation @xcite for @xmath2au results .
.summary of assigned systematic uncertainties of @xmath15 in cu@xmath3cu and au@xmath3au analysis . [
cols="<,^,^",options="header " , ] invariant transverse momentum spectra measured for the @xmath0 meson in cu@xmath3cu and au@xmath3au collisions at @xmath4=200 gev are shown in fig .
[ fig : ptspectra_auaupt ] .
measurements were performed only in the @xmath13 channel .
results are presented for three centrality bins : 020% , 2060% , 6092% ( 6094% in cu@xmath3cu ) and minimum bias collisions .
the dashed lines represent @xmath34 scaled fits to @xmath1 results , where @xmath34 values were taken from @xcite .
the results show that in peripheral heavy ion collisions @xmath0 production generally follows binary scaling , while in midcentral and central collisions , production of @xmath0 mesons is suppressed at high @xmath8 .
such behavior is similar to one previously observed for other light mesons @xcite and can be attributed to medium induced effects .
measurement of @xmath0 production can be used to study the relative production of vector and pseudoscalar mesons consisting of the same valence quarks , i.e. @xmath18 ratio as a function of transverse momentum . in calculating the @xmath18 ratio , the same methodology from @xcite for the @xmath75/@xmath76 and @xmath5
the charged pion results , @xmath77 , were used to extend neutral pion measurements at the lower limit of the @xmath8 range from 1 to 0.2 gev/@xmath28 . to produce the average pion spectrum in @xmath1 @xcite and @xmath2au collisions @xcite ,
we simultaneously fit @xmath77 and @xmath5 spectra with the modified hagedorn function @xcite . inclusion of the charged pion spectrum in the fit has a small effect in the 12 gev/@xmath28 overlap region , smaller than 5% compared to fitting neutral pions alone .
the resulting fitted pion distributions are used to calculate @xmath18 ratios for @xmath1 and @xmath2au .
uncertainties for the fit values are evaluated by taking into account statistical and systematic uncertainties of the experimental points as described in @xcite .
figure [ fig : omegapi0ratio ] presents the @xmath18 ratio measured in @xmath1 collisions at @xmath73=200 gev as a function of transverse momentum .
open markers show our previous measurements of the @xmath18 ratio @xcite .
one can see good agreement between previous results and this measurement . for completeness we also present similar measurements performed in lower energy experiments : @xmath78be at @xmath79 gev ( e706 @xcite ) , @xmath1 at @xmath73=62 gev ( isr @xcite ) .
please note that the branching ratio for the @xmath13 decay was set equal to ( [email protected])% , which is 6% different from the latest pdg value of ( [email protected])% . within measurement uncertainties
the @xmath18 ratio in hadronic interactions is energy independent at high @xmath8 .
measured @xmath18 ratio as a function of @xmath8 in @xmath1 collisions at @xmath73=200 gev .
( dashed line ) fit of a constant value to data points at @xmath74 2 gev/@xmath28 .
the fit result is @xmath80 .
( gray box ) the overall error of the fitting .
( solid line ) the pythia prediction @xcite for @xmath1 at @xmath73=200 gev . previously published phenix results ( prc75 ) @xcite and other lower energy experiments at @xmath81 gev ( e706 ) @xcite and @xmath82 gev ( isr ) @xcite
are shown for comparison . ]
a linear fit to the ratio at @xmath74 2 gev/@xmath28 gives a value of the linear coefficient consistent with zero within less then one standard deviation ( @xmath83 ) indicating no significant @xmath8 dependence of the ratio at @xmath74 2 gev/@xmath28 .
a fit to a constant gives a value of the ratio equal to @xmath80 consistent with our previous measurement of @xmath25 @xcite .
the pythia prediction of the @xmath0/@xmath30 ratio , shown in fig .
[ fig : omegapi0ratio ] with a solid line , lies above the measured ratio .
the @xmath18 ratios measured in minimum bias @xmath2au , cu@xmath3cu , and au@xmath3au collisions at @xmath4=200 gev are presented in fig .
[ fig : omegapi0ratioa ] . as in the case of @xmath1 collisions
there is no indication that the ratios depend on transverse momentum for @xmath74 2 gev/@xmath28 .
fits to a constant for @xmath74 2 gev/@xmath28 give the following values of the @xmath18 ratio : @xmath84 in @xmath2au , @xmath85 in cu@xmath3cu and @xmath86 in mb au@xmath3au collisions . within the uncertainties the @xmath18 ratios measured in different collision systems for @xmath74 2 gev/@xmath28 are in agreement
this agrees with previous measurements in @xmath2au @xcite within the uncertainties .
the ratios in various collision systems imply similar suppression factors and @xmath8 dependences within the uncertainties for the @xmath0 and @xmath30 production in nucleus - nucleus collisions at high @xmath8 .
@xmath7 for the @xmath0 meson integrated over the range @xmath74 7 gev/@xmath28 as a function of the number participating nucleons ( @xmath35 ) .
results for @xmath5 ( prl101 ) @xcite and @xmath6 ( prc82 ) @xcite are shown for comparison .
the dashed line shows the fitted fractional energy loss function , @xmath87 . ] to quantify medium induced effects on high @xmath8 particle production the nuclear modification factor is defined as : @xmath88 where @xmath89 is the differential yield per event in nucleus - nucleus collisions , @xmath90 is the number of binary nucleon - nucleon collisions averaged over the impact parameter range of the corresponding centrality bin calculated by glauber monte - carlo simulation @xcite , and @xmath62 and @xmath91 are the total and differential cross sections for inelastic @xmath1 collisions , respectively . in the absence of medium - induced effects , the yield of high-@xmath8 particles is expected to scale with @xmath92 , resulting in @xmath93=1 at high-@xmath8 .
figure [ fig : omegardau ] presents @xmath94 measured for the @xmath0 in minimum bias , most central and peripheral @xmath2au collisions at @xmath4=200 gev .
good agreement is observed between different decay modes , and between new and previously published phenix @xmath0 results @xcite shown with open markers . for comparison we also present @xmath5 results published in @xcite in the figure .
in peripheral collisions the measured values of @xmath94 are consistent with unity over the whole @xmath8 range of measurements . in most central collisions a modest cronin - like enhancement is observed in a range of @xmath8 from 2 gev/@xmath28 to 6 gev/@xmath28 and suppression of @xmath0 production at @xmath95 gev/@xmath28 .
a similar enhancement at 26 gev/@xmath28 was previously observed for neutral and charged pions @xcite and @xmath45 mesons @xcite .
suppression of @xmath0 production at higher @xmath8 is in agreement with @xmath5 results @xcite .
similarity of the observed effects for the mesons with very different masses suggests that the collective nuclear effects occur at the partonic level @xcite .
nuclear modification factors measured in cu@xmath3cu and au@xmath3au collisions at @xmath4=200 gev as a function of @xmath8 are shown in fig .
[ fig : omegaraa ] . results are presented for minimum bias , most central ( 020% ) , midcentral ( 2060% ) and peripheral ( 6094% in cu@xmath3cu ; 6092% in au@xmath3au ) collisions .
the nuclear modification factors do not depend on @xmath8 for @xmath74 6 gev/@xmath28 at all centralities . for @xmath96 suppression of @xmath0 production
begins to be observed , with suppression increasing as @xmath35 increases .
figure [ fig : npart ] shows @xmath7 values integrated for @xmath74 7 gev/@xmath28 as a function the number of participants . for @xmath0 mesons we present four centrality bins in @xmath2au , and three centrality bins in cu@xmath3cu and au@xmath3au . for comparison the average values of @xmath7 for @xmath5 @xcite and @xmath6 mesons @xcite mesons for @xmath74 7 gev/@xmath28 are also plotted . to see whether the @xmath0 follows the suppression pattern of @xmath5 and @xmath6 ,
the integrated @xmath7 vs @xmath35 dependence is fit to a fractional energy loss function @xmath97 @xcite .
the parameter @xmath98 , which is an exponent of the power law fit to the @xmath0 @xmath8 spectrum measured in @xmath1 for @xmath745 gev/@xmath28 @xcite , was fixed to 8 .
the fitting gives @xmath99ndf less than three and parameters @xmath100 = ( [email protected])@xmath2110@xmath101 and @xmath102 = 0.55 @xmath20 0.01 . as in @xcite we find the parameter @xmath102 consistent with predictions of the glv @xcite and pqm @xcite models ( @xmath103 ) .
therefore , we can conclude that @xmath0 production has a similar suppression pattern as @xmath5 and @xmath6 which supports the scenario that the energy loss takes place at the parton level in the hot and dense medium formed in the collisions .
we measured production of the @xmath0 meson via both leptonic and hadronic decay channels in @xmath1 , @xmath2au , cu@xmath3cu and au@xmath3au at @xmath4=200 gev .
the invariant transverse momentum spectra show good agreement in different decay channels in @xmath1 and @xmath2au .
the @xmath94 shows a moderate cronin like enhancement at intermediate @xmath8 26 gev/@xmath28 and suppression for @xmath95 gev/@xmath28 in most central @xmath2au collisions .
the measurement of the nuclear modification factor for the @xmath0 meson in cu@xmath3cu and au@xmath3au collisions show that @xmath0 production has a similar suppression pattern as the @xmath5 and @xmath6 within model agreement , thus supporting the scenario that the energy loss takes place at the partonic level in the hot and dense medium formed in the collisions .
we thank the staff of the collider - accelerator and physics departments at brookhaven national laboratory and the staff of the other phenix participating institutions for their vital contributions .
we acknowledge support from the office of nuclear physics in the office of science of the department of energy , the national science foundation , abilene christian university research council , research foundation of suny , and dean of the college of arts and sciences , vanderbilt university ( u.s.a ) , ministry of education , culture , sports , science , and technology and the japan society for the promotion of science ( japan ) , conselho nacional de desenvolvimento cientfico e tecnolgico and fundao de amparo pesquisa do estado de so paulo ( brazil ) , natural science foundation of china ( p. r. china ) , ministry of education , youth and sports ( czech republic ) , centre national de la recherche scientifique , commissariat lnergie atomique , and institut national de physique nuclaire et de physique des particules ( france ) , ministry of industry , science and tekhnologies , bundesministerium fr bildung und forschung , deutscher akademischer austausch dienst , and alexander von humboldt stiftung ( germany ) , hungarian national science fund , otka ( hungary ) , department of atomic energy and department of science and technology ( india ) , israel science foundation ( israel ) , national research foundation and wcu program of the ministry education science and technology ( korea ) , ministry of education and science , russian academy of sciences , federal agency of atomic energy ( russia ) , vr and the wallenberg foundation ( sweden ) , the u.s .
civilian research and development foundation for the independent states of the former soviet union , the us - hungarian fulbright foundation for educational exchange , and the us - israel binational science foundation . | the phenix experiment at the relativistic heavy ion collider has measured @xmath0 meson production via leptonic and hadronic decay channels in @xmath1 , @xmath2au , cu@xmath3cu , and au@xmath3au collisions at @xmath4=200 gev .
the invariant transverse momentum spectra measured in different decay modes give consistent results .
measurements in the hadronic decay channel in cu@xmath3cu and au@xmath3au collisions show that @xmath0 production has a suppression pattern at high transverse momentum , similar to that of @xmath5 and @xmath6 in central collisions , but no suppression is observed in peripheral collisions .
the nuclear modification factors , @xmath7 , are consistent in cu@xmath3cu and au@xmath3au collisions at similar numbers of participant nucleons . |
since the recent discovery of superconductivity at elevated temperature in f - substituted lafeaso , @xcite a large amount of experimental and theoretical effort was concentrated on studies of magnetic , superconducting and normal state properties of fe - based superconductors and related materials .
@xcite of several families of fe - based superconductors discovered by now , the 122 , aefe@xmath2as@xmath2 ( ae = alkaline earth and eu ) family , appears to be the most studied one @xcite due to the reasonable ease of growing large , high quality , single crystals , the availability of multiple substitution sites and the simplicity of the crystal structure . in the 122 system ,
superconductivity , with a @xmath6 k , was first reported on k - doping of bafe@xmath2as@xmath2 .
@xcite subsequently the complete ( ba@xmath7k@xmath8)fe@xmath2as@xmath2 solid solution series was studied @xcite in detail .
kfe@xmath2as@xmath2 stands out among the members of the ( ba@xmath7k@xmath8)fe@xmath2as@xmath2 series , and the fe - based superconductors in general , as a unique material .
it is a stoichiometric end - member of the ( ba@xmath7k@xmath8)fe@xmath2as@xmath2 series , and a rare example of a stoichiometric fe - based superconductor , albeit with a rather low @xmath9 k. @xcite the in - plane resistivity has a metallic behavior with a remarkable residual resistivity ratio , exceeding 1000 in the best single crystals .
@xcite the reported fermi surface of kfe@xmath2as@xmath2 differs from that of the optimally doped ( ba@xmath7k@xmath8)fe@xmath2as@xmath2 having three hole pockets , two centered at the @xmath10 point in the brillouin zone , and one around the @xmath11 point @xcite with no electron pockets .
existence of quantum criticality , and nodal or @xmath12-wave superconductivity was suggested and discussed in number of publications .
@xcite given the unusual properties of this material , it is of importance to have a broad set of data on its physical properties , in particular , related to its superconducting state . in this work
we study the response of kfe@xmath2as@xmath2 s superconducting transition temperature to external pressure , both hydrostatic , via direct measurements , and uniaxial , that is inferred by using the ehrenfest relation for the second order phase transition .
hydrostatic pressure effects historically have been studied for many superconductors , including some fe - based materials .
@xcite uniaxial pressure effects are rarely measured directly , @xcite due to significant technical difficulties , however evaluations of the uniaxial pressure derivatives , @xmath13 , ( @xmath14 ) by combining the thermal expansion and heat capacity data , have been recently made for several bafe@xmath2as@xmath2 - based materials @xcite yielding in - plane and @xmath0-axis uniaxial pressure derivatives of the opposite signs and distinct evolution of these derivatives with doping .
kfe@xmath2as@xmath2 single crystals were grown by using kas flux method .
@xcite briefly , k ( ingot ) , as ( lump ) , and fe ( powder ) were weighed at an atomic ratio of k : fe : as = 5 : 2 : 6 , and loaded into an alumina crucible .
a sealing technique with liquid tin melt was developed to suppress the evaporation of k and as elements .
@xcite all the steps were performed in a glove box under argon gas atmosphere .
the sealed fused silica ampule was heated up to 920@xmath15 c , and then slowly cooled down to 620@xmath15 c at a rate of 1@xmath15 c / h .
thin , plate - like crystals , with dimensions up to @xmath16 mm@xmath17 , can be easily mechanically separated from the kas flux .. the further details of the crystal growth will be reported elsewhere .
@xcite the crystals are plate - like with the @xmath0-axis perpendicular to the plate .
they are soft , micaceous and moderately air - sensitive .
low - field dc magnetization under pressure , was measured in a quantum design magnetic property measurement system , mpms-5 , squid magnetometer using a a commercial , hmd , be - cu piston - cylinder pressure cell . @xcite daphne oil 7373 was used as a pressure medium and superconducting pb as a low - temperature pressure gauge .
thermal expansion data were obtained using a capacitive dilatometer constructed of ofhc copper ; a detailed description of the dilatometer is presented elsewhere .
@xcite the dilatometer was mounted in a quantum design physical property measurement system , ppms-14 , instrument and was operated over a temperature range of 1.8 - 305 k. due to the morphology of the crystals , the dilation was measured only along the @xmath0-axis .
the heat capacity was measured using a hybrid adiabatic relaxation technique of the heat capacity option in a quantum design , ppms-9 instrument .
the low temperature heat capacity of kfe@xmath2as@xmath2 is plotted as @xmath18 vs ( t ) in fig .
the jump in specific heat associated with the superconducting transition is rather sharp . from the isoentropic
construct , shown in fig .
[ f2 ] , @xmath19 k , @xmath20 mj / mol k@xmath21 . from the linear fit of the @xmath18 vs @xmath22 above the superconducting transition ( fig .
[ f2 ] , inset ) , the sommerfeld coefficient , @xmath23 mj / mol k@xmath21 , and the debye temperature , @xmath24 k can be estimated .
these results are comparable to the literature data .
@xcite low field ( @xmath25 oe ) zero - field - cooled ( zfc ) temperature - dependent magnetization data taken under pressures up to @xmath26 kbar are shown in fig .
the signal associated with the superconducting transition in kfe@xmath2as@xmath2 is sharp , under pressure the transition shifts to lower temperatures without any significant broadening .
the pressure dependence of @xmath3 ( defined as an onset of transition in magnetization , see fig .
[ f2 ] ) is shown in the inset to fig .
[ f1 ] . a linear fit of @xmath27 results in the value of the pressure derivative , @xmath28 k / kbar , so that in a simple , linear , approximation superconductivity in kfe@xmath2as@xmath2 can be suppressed by a pressure of @xmath29 kbar .
it appears that the measured @xmath27 dependence has a slight upward curvature .
indeed , one can also fit @xmath30 with a second order polynomial ( shown as a dashed line in the inset to fig .
. then the linear coefficient corresponding to the initial @xmath31 is @xmath32 k / kbar ( the quadratic coefficient equals to @xmath33 k / kbar@xmath21 ) .
very similar results are obtained if a different criterion ( e.g. maximum of @xmath34 ) is used .
temperature dependent @xmath0 - axis dilation , normalized to the value at 1.8 k ( @xmath35 ) , is shown in fig .
the overall behavior is monotonic with some flattening at low temperatures .
the relative change in the @xmath0 - axis from 1.8 to 300 k is similar to that measured in pure aefe@xmath2as@xmath2 ( ae = ba , sr , ca ) @xcite and co - doped bafe@xmath2as@xmath36 @xcite , with the value of the thermal expansion coefficient , @xmath37 , at room temperature being under @xmath38 k@xmath39 . both , the @xmath0-axis dilation and the thermal expansion coefficient show a clear anomaly at the superconducting transition .
the jump in the thermal expansion coefficient at @xmath3 ( using the criterion graphically similar to that for @xmath40 in fig .
[ f2 ] ) can be estimated as @xmath41 k@xmath39 .
the shape and sharpness of the features in the thermal expansion coefficient and the heat capacity at @xmath3 are very similar ( fig .
[ f3 ] , inset ) , that rationalizes use of the same criterion for both measurements .
the value of the hydrostatic pressure derivative , @xmath42 k / kbar , measured for kfe@xmath2as@xmath2 is rather large .
it is close to , but somewhat smaller than @xmath43 k / kbar reported for stoichiometric , polycrystalline , rbfe@xmath2as@xmath2 .
@xcite it has to be noted that for slightly overdoped ( ba@xmath44k@xmath45)fe@xmath2as@xmath2 rather large , negative , pressure derivatives , @xmath46 k / kbar and @xmath47 k / kbar were reported as well .
@xcite the evolution of the @xmath31 values with k - doping in the ( ba@xmath7k@xmath8)fe@xmath2as@xmath2 series , in particular in the overdoped region , will be useful for further understanding of superconductivity in these materials .
the uniaxial , @xmath0 - axis , pressure derivative of @xmath3 can be inferred using the ehrenfest relation @xcite for the second order phase transition : @xmath48 , where @xmath49 is the molar volume ( @xmath50 m@xmath17 for kafe@xmath2as@xmath2 , using the lattice parameters @xcite at 1.7 k ) , @xmath51 is the jump in the thermal expansion coefficient at the phase transition , and @xmath52 is the corresponding jump in the heat capacity .
using the experimental data above , for kfe@xmath2as@xmath2 we found out that the @xmath0 - axis uniaxial pressure derivative is positive , @xmath5 k / kbar .
since the hydrostatic pressure derivative in this case can be written as @xmath53 , we can infer that the uniaxial , @xmath54-axis ( or in - plane ) pressure derivative is negative , @xmath55 k / kbar .
so it appears that kfe@xmath2as@xmath2 is equally sensitive to uniaxial pressure applied in the @xmath56 - plane and along the @xmath0 - axis , however @xmath3 increases when the pressure is applied along the @xmath0-axis and decreases when it is applied in the @xmath56 - plane .
uniaxial pressure derivatives of @xmath3 in members of the 122 family were also reported to have different signs for @xmath56-plane and @xmath0-axis pressure .
we are not aware of such data in the ( ba@xmath7k@xmath8)fe@xmath2as@xmath2 series , but for co - doped ( electron doping ) and p - doped ( isoelectronic substitution ) bafe@xmath2as@xmath2 samples in the overdoped region of the superconducting dome negative values of @xmath57 and positive values of @xmath58 were reported .
@xcite if we consider kfe@xmath2as@xmath2 as an extreme of the overdoped ( ba@xmath7k@xmath8)fe@xmath2as@xmath2 , it is clearly different , in this respect , from the other ( studied so far ) overdoped fe - based superconductors from the 122 family .
it has to be mentioned that the anisotropy and different signs of the uniaxial pressure derivatives of @xmath3 , as inferred from thermal expansion and heat capacity data using the ehrenfest relation , can shed some light on the differences observed in the values of the critical pressure and position of the superconducting dome that vary in publications describing the pressure - temperature phase diagrams in the aefe@xmath2as@xmath2 materials .
@xcite in many cases of the different anvils - based pressure cells an additional , axial , component of pressure is present . with the plate - like aefe@xmath2as@xmath2 samples in a `` convenient geometry '' ,
this means that the @xmath0-axis pressure in such cells is slightly higher than the pressure in the @xmath56 - plane .
then , for the samples with @xmath59 the measured @xmath31 is smaller than the truly hydrostatic value , and vise versa , if @xmath60 , larger than hydrostatic value of @xmath31 is observed in an experiment .
another clear difference between this material and other fe - based , 122 superconductors is that the jump in heat capacity at @xmath3 clearly deviates from the trend suggested in ref . and expanded in ref .
, the so - called bnc scaling ( fig . [ f4 ] ) . at the same time , our data yield @xmath61 that is significantly smaller than the 1.43 value expected for bcs superconductors . these observations are consistent with previous publications @xcite and suggest that superconductivity in kfe@xmath2as@xmath2 neither is a conventional bcs , nor should it be considered as just significantly overdoped fe - based superconductor from the 122 family .
+ to summarize , the superconducting transition temperature of kfe@xmath2as@xmath2 decreases under hydrostatic pressure rather fast , with @xmath62 k / kbar .
the uniaxial , @xmath0 - axis , pressure derivative inferred from the thermal expansion and heat capacity measurements , is positive , that differs from the existing data for overdoped fe - based 122 superconductors .
the jump in heat capacity at @xmath3 for kfe@xmath2as@xmath2 deviates significantly from the empirical trend , @xmath63 ( and its recent modifications ) , observed in a number of fe - based superconductors .
taken together , these results suggest that the superconductivity in kfe@xmath2as@xmath2 could be different from just an extreme overdoped case in the 122 family . detailed , comprehensive studies of the ( ba@xmath7k@xmath8)fe@xmath2as@xmath2 series with @xmath64 approaching 1 , once homogeneous single crystalline samples for these intermediate concentrations are available , might shed a light on physics of this interesting material .
work at the ames laboratory was supported by the us department of energy , basic energy sciences , division of materials sciences and engineering under contract no .
de - ac02 - 07ch11358 .
s.l.b . acknowledges partial support from the state of iowa through iowa state university .
s. avci , o. chmaissem , d. y. chung , s. rosenkranz , e. a. goremychkin , j. p. castellan , i. s. todorov , j. a. schlueter , h. claus , a. daoud - aladine , d. d. khalyavin , m. g. kanatzidis , and r. osborn , phys .
b * 85 * , 184507 ( 2012 ) . k. hashimoto , a. serafin , s. tonegawa , r. katsumata , r. okazaki , t. saito , h. fukazawa , y. kohori , k. kihou , c. h. lee , a. iyo , h. eisaki , h. ikeda , y. matsuda , a. carrington , and t. shibauchi , phys
b * 82 * 014526 ( 2010 ) .
taichi terashima , motoi kimata , nobuyuki kurita , hidetaka satsukawa , atsushi harada , kaori hazama , motoharu imai , akira sato , kunihiro kihou , chul - ho lee , hijiri kito , hiroshi eisaki , akira iyo , taku saito , hideto fukazawa , yoh kohori , hisatomo harima , and shinya uji , phys .
* 104 * , 259701 ( 2010 ) .
reid , m. a. tanatar , a. juneau - fecteau , r. t. gordon , s. ren de cotret , n. doiron - leyraud , t. saito , h. fukazawa , y. kohori , k. kihou , c. h. lee , a. iyo , h. eisaki , r. prozorov , and louis taillefer , phys .
lett . * 109 * , 087001 ( 2012 ) .
a. e. bhmer , p. burger , f. hardy , t. wolf , p. schweiss , r. fromknecht , h. v. lhneysen , c. meingast , h. k. mak , r. lortz , s. kasahara , t. terashima , t. shibauchi , and y. matsuda , phys .
b * 86 * , 094521 ( 2012 ) .
g. m. schmiedeshoff , a. w. lounsbury , d. j. luna , s. j. tracy , a. j. schramm , s. w. tozer , v. f. correa , s. t. hannahs , t. p. murphy , e. c. palm , a. h. lacerda , s. l. budko , p. c. canfield , j. l. smith , j. c. lashley , and j. c. cooley , rev .
instrum . * 77 * , 123907 ( 2006 ) .
as@xmath2 under pressure .
the @xmath3 criterion used in the paper is shown for the lowest pressure curve .
the signal from the pb manometer is also shown .
inset - pressure dependence of @xmath3 of kfe@xmath2as@xmath2 .
lines : linear ( solid ) and second order polynomial ( dashed ) fit of @xmath27 . ,
width=453 ] -axis dilation of kfe@xmath2as@xmath2 .
the data are normalized to @xmath66 value at 1.8 k. upper left inset : low temperature @xmath0-axis dilation and thermal expansion coefficient , @xmath67 , with the anomalies at the superconducting transition .
lower right inset : low temperature thermal expansion coefficient , @xmath67 , and heat capacity , @xmath18 , both showing jump at the superconducting transition .
, width=453 ] vs @xmath3 for the kfe@xmath2as@xmath2 sample , plotted together with literature data for various feas - based superconducting materials .
updated plot @xcite is used to show the literature data .
the line corresponds to @xmath63.,width=453 ] | we present heat capacity , @xmath0-axis thermal expansion and pressure dependent , low field , temperature dependent magnetization for pressures up to @xmath1 kbar , data for kfe@xmath2as@xmath2 single crystals .
@xmath3 decreases under pressure with @xmath4 k / kbar .
the inferred uniaxial , @xmath0-axis , pressure derivative is positive , @xmath5 k / kbar .
the data are analyzed in comparison with those for overdoped fe - based superconductors .
arguments are presented that superconductivity in kfe@xmath2as@xmath2 may be different from the other overdoped , fe - based materials in the 122 family . |
the _ tunnel number _ @xmath4 of a knot ( or link ) @xmath5 in @xmath6 is the minimal number of arcs that , when attached to the knot , gives a graph whose complement is the interior of a handlebody .
it was once naively hoped that this knot invariant might be additive under connected sum of knots , but morimoto ( @xcite ) has found counterexamples in which tunnel number _ degenerates _ under connected sum .
that is , there are knots @xmath7 and @xmath8 for which @xmath9 .
the degree of degeneration possible was soon shown by kobayashi ( @xcite ) to be arbitrarily high .
that is , given @xmath10 there exist knots @xmath7 and @xmath8 for which @xmath11 .
although kobayashi s examples show that there is no uniform bound on the degeneration number , it is still natural to ask if there is a uniform bound less than @xmath12 on the _ degeneration ratio _
@xmath13 more generally , for knot summands @xmath14 we can define the degeneration ratio @xmath15 kobayashi s methods show that the degeneration ratio for each of his examples is at least @xmath16 and morimoto s specific example has degeneration ratio @xmath17 .
if we also consider links , morimoto has shown @xcite that the connected sum of a knot @xmath5 and a @xmath1-component link @xmath18 has tunnel number one exactly if @xmath5 is a @xmath1-bridge knot and @xmath18 is the hopf link .
such examples have degeneration ratio @xmath19 . in the other direction
, kowng showed in @xcite the more general claim that if an irreducible orientable compact @xmath0-manifold @xmath20 contains a collection of tori @xmath21 , then @xmath20 has a generalized heegaard splitting of genus at most @xmath22 for which each component of @xmath21 is contained in a thin level ( see definition [ general ] ) .
when applied to @xmath23 , this has the corollary that @xmath24 so kowng showed that , in general , @xmath25 here we improve this aspect of kowng s results , using a somewhat different approach growing out of the investigations in @xcite , @xcite .
the main theorem shows how to cut a strongly irreducible generalized heegaard splitting of @xmath20 along a family of essential annuli and create generalized heegaard splittings for the resulting manifolds .
analysis of the result shows a bit more than the following : for @xmath26 prime , @xmath27 ; if none of the @xmath26 is @xmath1-bridge , then @xmath28 .
@xmath1-bridge knots seem to play a special role in the theory of tunnel numbers .
morimoto s original example of degeneration used a @xmath1-bridge summand , and kobayashi s ( non - prime ) examples each have for one of their summands a sum of @xmath1-bridge knots .
so it is intriguing that here too the bounds on degeneration are better if there are no @xmath1-bridge knots among the summands .
although we focus on cutting a manifold apart by annuli , the same sort of analysis extends to families of tori , in a way we only briefly describe .
after some preliminaries and a simplified but indicative special case , the outline of the proof is as follows : * step 1 ( section [ sec : annuli ] ) : * we examine annuli in , and on the boundary of , compression bodies .
the goal is to develop criteria that allow us to cut open compression bodies along annuli and glue compression bodies together along annuli so that at the end , the result is still a collection of compression bodies . * step 2 ( section [ sec : break ] ) : * we prove that a strongly irreducible generalized heegaard splitting can be cut along annuli and augmented in such a way as to yield generalized heegaard splittings for the complementary components . in this construction
a bound on the degeneration of the `` index '' is given by the number of `` dipping annuli '' .
* step 3 ( section [ sec : des ] ) : * we show that the generalized heegaard splittings of the complementary components obtained in this way can be destabilized under certain conditions .
* step 4 ( sections [ sec : generalcount ] and [ sec : specialcount ] ) : * we find bounds on the number of dipping annuli .
section [ sec : generalcount ] is quite general , whereas section [ sec : specialcount ] achieves better bounds in the context of a knot complement , when we can assume any surface with meridinal boundary has an even number of boundary components .
* step 5 ( section [ sec : tun ] ) : * we apply the derived inequalities to the study of tunnel numbers .
the paper concludes with an appendix by andrew casson .
he constructs an example which demonstrates that the bound on the number of dipping annuli given in section 7 is , in some sense , best possible .
for standard definitions concerning knots , see @xcite or @xcite and for those concerning @xmath0-manifolds , see @xcite or @xcite .
all manifolds will be orientable . for @xmath29 a properly embedded submanifold of @xmath20 , we denote an open regular neighborhood of @xmath29 in @xmath20 by @xmath30 .
let @xmath5 be a knot in @xmath6 .
denote the , @xmath31 , by @xmath32 .
[ rem : dec ] let @xmath33 be the sum of two knots . then the decomposing sphere gives rise to a decomposing annulus @xmath34 properly embedded in @xmath32 such that @xmath35 .
if @xmath36 , then we may assume that the decomposing spheres are nested , so that @xmath37 .
[ defn : tunnel system ] a for a knot @xmath5 is a collection of disjoint arcs @xmath38 properly embedded in @xmath32 such that @xmath39 is a handlebody .
the , denoted by @xmath4 , is the least number of arcs required in a tunnel system for @xmath5 .
[ defn : cb ] a is a @xmath0-manifold @xmath40 obtained from a connected closed orientable surface @xmath41 by attaching @xmath1-handles to @xmath42 and capping off any resulting @xmath1-sphere boundary components .
we denote @xmath43 by @xmath44 and @xmath45 by @xmath46 .
dually , a compression body is a connected orientable @xmath0-manifold obtained from a ( not necessarily connected ) closed orientable surface @xmath47 by attaching @xmath12-handles .
define the of @xmath40 by @xmath48 . a for a compression body @xmath40 is a set of disks @xmath49 properly embedded in @xmath40 with @xmath50 for @xmath51 , @xmath52 such that the result of cutting @xmath40 along @xmath53 is homeomorphic to @xmath47 .
[ defn : heegaard splitting ] a of a @xmath0-manifold @xmath20 is a decomposition @xmath54 in which @xmath55 , @xmath40 are compression bodies such that @xmath56 and @xmath57 .
we call @xmath41 the or .
[ defn : wr ] a heegaard splitting is ( resp . ) if there are essential disks @xmath58 and @xmath59 , such that @xmath60 ( resp .
@xmath61 ) . a heegaard splitting which is not ( weakly ) reducible is .
[ general ] a heegaard splitting of a compact orientable @xmath0-manifold @xmath20 is a structure @xmath62 .
each of the @xmath63 and @xmath64 is a union of compression bodies , @xmath65 , ( i.e. , @xmath66 is a union of heegaard splittings of a submanifold of @xmath20 ) and @xmath67 .
we say that a generalized heegaard splitting is if each heegaard splitting @xmath68 is strongly irreducible and each @xmath69 is incompressible in @xmath20 .
we will denote @xmath70 by @xmath71 and @xmath72 by @xmath73 .
the surfaces in @xmath71 are called the and the surfaces in @xmath73 the .
let @xmath54 be an irreducible heegaard splitting .
we may think of @xmath20 as being obtained from @xmath74 by attaching all @xmath12-handles in @xmath55 followed by all @xmath1-handles in @xmath40 , followed , perhaps , by @xmath0-handles .
an of @xmath54 is a rearrangement of the order in which the @xmath12-handles of @xmath55 and the @xmath1-handles of @xmath40 are attached .
this rearrangement yields a generalized heegaard splitting . for convenience
, we will occasionally denote @xmath75 by @xmath76 . the main theorem in @xcite together with the calculation ( * ? ? ?
* lemma 2 ) implies the following : [ lem : ind ] suppose @xmath20 is an irreducible compact @xmath0-manifold .
then @xmath20 possesses a genus @xmath77 heegaard splitting if and only if @xmath20 has a strongly irreducible generalized heegaard splitting @xmath78 such that @xmath79 .
one implication comes from untelescoping a heegaard splitting of genus @xmath77 , the other from thinking of a generalized heegaard splitting as an untelescoping of some heegaard splitting .
the latter process is sometimes called the _ amalgamation _ of the generalized splitting into a standard splitting . a strongly irreducible heegaard splitting
can be isotoped so that its splitting surface , @xmath41 intersects an incompressible surface , @xmath80 , only in curves essential in both @xmath41 and @xmath80 .
this is a deep fact and is proven , for instance , in ( * ? ? ?
* lemma 6 ) .
this fact , together with the fact that incompressible surfaces can be isotoped to meet only in essential curves , establishes the following : [ lem : essential ] let @xmath80 be a properly embedded incompressible surface in an irreducible @xmath0-manifold @xmath20 and let @xmath81 be a strongly irreducible generalized heegaard splitting of @xmath20 .
then @xmath82 can be isotoped to intersect @xmath80 only in curves that are essential in both @xmath80 and @xmath83 .
we will first point out a special simplified case in which an upper bound to the degeneration ratio is fairly easily found , and the ideas that are used are indicative of ideas that will be important in the general case .
the special case is this : [ thm : indic]let @xmath7 and @xmath8 be prime knots and assume that @xmath84 possesses a minimal genus heegaard splitting that is strongly irreducible . then @xmath85 .
hence @xmath86 . *
proof : * let @xmath54 be the strongly irreducible heegaard splitting , with @xmath87 and @xmath40 a handlebody . the hypothesis implies that @xmath88 . let @xmath89 denote the annulus in @xmath90 that , when completed by a pair of meridian disks , constitutes a decomposing sphere for @xmath91 .
following lemma [ lem : essential ] we can isotope @xmath41 and @xmath34 until their intersection consists of circles essential in @xmath34 and in @xmath41 .
isotope them further to reduce , as much as possible , the number of such components of @xmath92 .
let @xmath93 and @xmath94 be the essential annuli @xmath95 in @xmath55 and @xmath40 respectively .
a sequence of @xmath96-compressions of @xmath97 could turn each annulus into a disk ; similarly a sequence of @xmath96-compressions of all but the spanning annulus of @xmath98 could turn each annulus into a disk .
if these sets of @xmath96-compressions could be done simultaneously , the result would be that @xmath92 would become a single circle of intersection , inessential in @xmath34 and bounding an essential disk in @xmath40 .
the disk would divide @xmath40 into two handlebodies , and it would follow that @xmath99 so there would be no degeneration ( in fact , a surplus ! ) in general , the boundary compressions of @xmath100 and @xmath101 can not be done simultaneously , since the arcs to which they @xmath96-compress in @xmath41 may intersect .
this problem can be avoided if @xmath41 were first stabilized by attaching a tube parallel to a spanning arc of each annulus in @xmath100 .
this would allow @xmath96-compressions of @xmath100 onto arcs that run along the tubes , without affecting the @xmath96-compressions of @xmath101 to the original @xmath41 .
so we see immediately that @xmath102 , the number of tubes used , is an upper bound on degeneration .
the first goal is then to get a bound on @xmath102 .
since any surface in @xmath6 is separating , any component of @xmath103 has an even number of boundary components , since it can be completed to become a closed surface in @xmath6 by attaching meridian disks of either @xmath7 or @xmath8 .
assume , for initial simplification , that no component of @xmath104 is an annulus .
let @xmath105 and notice then that each component @xmath106 of @xmath107 has at most @xmath108 boundary components , since each component has non - trivial even euler characteristic .
it follows that @xmath109 , so the number of essential curves @xmath110 .
this implies that @xmath111 . combining this argument with the previous one
, we get @xmath112 or @xmath113 an inequality better than required .
now examine the annulus components in @xmath104 .
note first of all that any such annulus component @xmath114 is necessarily @xmath96-parallel in the component @xmath115 ( say ) of @xmath116 in which it lies , since @xmath7 is prime . since the number of ( essential ) circles
@xmath117 has been minimized , @xmath114 is not parallel to a subannulus of @xmath34 .
so @xmath114 must be parallel to the annulus component of @xmath118 that lies in @xmath115 , an annulus on @xmath119 which we denote by @xmath120 .
it follows that there can not be two annulus components of @xmath104 that are adjacent in @xmath41 , since one would be parallel to each of the two annulus components of @xmath118 and so the union of the two annuli in @xmath41 would provide a way to
spin a collar of @xmath121 in a way that would reduce the number of components of @xmath122 .
since no two annuli in @xmath104 are adjacent in @xmath41 , the total number of these annuli can be no larger than the number of circles @xmath123 that we calculated above in the absence of annuli .
that is , there are no more than @xmath124 annuli among the components of @xmath104 .
hence , even allowing annuli components in @xmath104 , @xmath125 so @xmath126 and we get @xmath127 or @xmath128 but we can do better .
the region lying between the annuli @xmath114 and @xmath120 in @xmath115 is homeomorphic to @xmath129 , i. e. a solid torus , and it is known how a strongly irreducible splitting surface like @xmath41 can intersect a solid torus ( cf @xcite ) .
the upshot is that all components of @xmath104 lying between @xmath114 and @xmath120 in @xmath115 are annuli parallel to @xmath114 , except possibly one component ( an _ exceptional _ component ) which consists of a pair of annuli parallel to @xmath114 but then tubed together by a vertical tube .
( see figure 1 . )
an exceptional component can be ignored since , for example , the vertical tube could be slid across @xmath34 into @xmath130 .
notice that , by choosing @xmath114 outermost ( i. e. furthest from @xmath131 ) , all annuli components of @xmath132 lie between @xmath114 and @xmath120 . by the analogous argument in @xmath130 no component of @xmath133 can be an annulus , since an innermost one would be adjacent in @xmath41 to an innermost one of @xmath134 , forming a @xmath96-parallel torus component of @xmath41 , a contradiction .
( this argument is complicated only a little by the possible presence of an exceptional component . ) and @xmath120 ] so we may suppose that all @xmath136 annulus components of @xmath104 lie in @xmath115 , lie between and are parallel to the annuli @xmath114 and @xmath120 .
then at least @xmath137 pairs of these annuli each cut off a parallel pair of annuli in @xmath100 . for each of these pairs
, only one stabilization is required in the process described above , since the annuli are parallel .
( imagine placing the added tube so that it lies in the collar between the annuli in @xmath55 ) .
so if @xmath104 contains @xmath136 annuli there are two new effects : first , the count of @xmath138 is changed to @xmath139 , so the number of essential curves @xmath140 .
this implies that @xmath141 on the other hand , the second effect is that the number of stabilizations required is at most @xmath142 this implies that @xmath143 or @xmath144 as required .
a properly imbedded annulus @xmath145 is _ essential _ if it is incompressible and not @xmath96-parallel . in this section
we study finite sets of disjoint essential annuli in a compression body @xmath40 .
[ lem : br ] if @xmath34 is an essential annulus in a compression body @xmath40 , then either one component of @xmath121 lies on each of @xmath146 and @xmath147 ( i. e. @xmath34 is _ spanning _ ) or @xmath148 . in the latter case ,
@xmath34 is boundary compressible .
* proof : * see ( * ? ? ?
* lemma 9 ) .
suppose is a properly imbedded collection of essential annuli in a compression body @xmath40 so , in particular , the boundary of any non - spanning annulus @xmath34 lies in @xmath146 .
suppose @xmath149 is a @xmath96-compressing disk for the non - spanning annulus @xmath34 , i. e. the interior of @xmath149 is disjoint from @xmath150 , and @xmath151 is the union of a spanning arc of @xmath34 and an arc in @xmath146 .
the arc components of @xmath152 divide @xmath149 into subdisks .
this naturally gives rise to a tree in @xmath149 , in which a vertex is chosen inside each subdisk and two vertices are connected if they abut the same annulus .
( we can ignore closed components of intersection . indeed , since @xmath150 is incompressible , closed components of @xmath152 can be removed by an isotopy with support disjoint from the arcs of intersection . )
it will be useful to extend this tree by attaching , to the vertex corresponding to the subdisk that abuts , an edge that crosses .
the other end of is called the root of the resulting tree ; the other valence one edges are called the leaves of .
each leaf corresponds to a disk cut off by an outermost arc of @xmath153 in @xmath149 .
( see figure 2 . ) . ]
recall that there is a natural order on the set of finite sets of integers ( see ( * ? ? ?
* definition 4.3 ) ) .
one set of integers is compared to another by arranging each in descending ( or at least never - ascending ) order and then comparing them lexicographically .
this ordering has the property that if a subset of a set of integers is replaced by a subset of lower order then the resulting set is of lower order .
the of the tree is the set of distances ( measured in edges traversed ) from the root of to the leaves in .
the complexity of a @xmath96-compressing disk for @xmath154 is the complexity of the associated tree .
a @xmath96-compressing disk for @xmath34 is a @xmath96-compressing disk of minimal complexity ( using the above order on sets of integers ) .
if an annulus @xmath34 has minimal disks abutting it on both sides , we say that @xmath34 is .
[ lem : height ] if @xmath155 is a minimal @xmath96-compressing disk for @xmath156 then no component of @xmath157 is an inessential arc in @xmath150 . in particular , @xmath155 is disjoint from all spanning components of @xmath150 .
furthermore , if @xmath158 is another annulus in @xmath150 and @xmath158 intersects @xmath155 , then @xmath158 is not ambivalent and any arc component of @xmath159 cuts off from @xmath155 a subdisk @xmath160 that is also a minimal disk .
* proof : * immediate , since otherwise replacing @xmath161 by a minimal disk ( on the other side of @xmath158 if possible ) would decrease the complexity of @xmath155 .
the side opposite to the side on which a minimal disk abuts a non - spanning annulus @xmath162 is called the of @xmath34 . if @xmath34 is ambivalent , arbitrarily choose a side to call the root side .
[ defn : indep ] let @xmath163 be a collection of essential non - spanning annuli in the boundary of a compression body @xmath40
. then @xmath150 is if for each @xmath164 there is a meridian disk @xmath155 for @xmath40 such that @xmath165 consists precisely of a single spanning arc of @xmath34 .
an easy outermost arc argument shows that we can take the collection of meridians that arise in definition [ defn : indep ] to be disjoint .
[ lem : glue ] suppose @xmath55 and @xmath40 are compression bodies , and a collection of essential non - spanning annuli @xmath163 is independently longitudinal .
then if @xmath55 is attached to @xmath40 by identifying @xmath150 to any collection of annuli in @xmath166 , the result is a compression body . *
proof : * @xmath150 becomes a collection of boundary compressible annuli in @xmath167 . in this collection each annulus has a boundary compressing disk of complexity @xmath168 ( i. e. a disk with interior disjoint from ) . performing these boundary compressions
turn @xmath150 into a collection of disks .
so an alternate and equivalent construction for @xmath167 would be to first compress @xmath40 along the meridians for @xmath163 and then attach the resulting compression body @xmath169 to @xmath166 along the disks in @xmath170 that are the remnants of @xmath150 .
this alternate construction clearly gives a compression body .
[ lem : cbs ] let @xmath80 be a ( not necessarily connected ) incompressible surface in a compression body @xmath40 such that @xmath171 . then the result of cutting @xmath40 along @xmath80 is a collection of compression bodies .
* proof : * see for instance ( * ? ? ?
* lemma 2 ) .
[ lem : push ] suppose @xmath172 is a set of essential annuli in the compression body @xmath40 .
let @xmath173 be the subcollection of non - spanning annuli
. cut out from @xmath40 a collection of tunnels , one parallel to a spanning arc of each annulus @xmath34 in @xmath174 and lying on the root side of @xmath34 .
call the resulting compression body @xmath169 .
then the closure @xmath175 of any component of @xmath176 is a compression body on which @xmath177 is a collection of independently longitudinal annuli .
* proof : * let @xmath34 be any annulus in @xmath178 and let @xmath179 denote the component of @xmath180 from which tunnels were removed to get @xmath175 .
if , in @xmath40 , @xmath179 lay on the root side of @xmath34 then in constructing @xmath175 a tunnel has been removed that runs parallel to a spanning arc of @xmath34 .
the disk that defines the parallelism is a meridian for @xmath175 that intersects @xmath34 precisely in a spanning arc , as required . if , on the other hand , @xmath179 is not on the root side of @xmath34 then consider a minimal @xmath96-compressing disk @xmath155 for @xmath34 in @xmath40 .
the component @xmath181 of @xmath182 that abuts @xmath34 lies in @xmath179 and abuts any other annulus in @xmath150 , if at all , only on its root side ( see lemma [ lem : height ] ) .
in particular , for each such annulus , the tunnels that are removed to create @xmath175 can be positioned so that the component @xmath183 of @xmath184 that abuts @xmath34 runs over the tunnels instead of across the other annuli .
see figure 3 .
@xmath183 is then a meridian for @xmath175 that intersects @xmath150 only in a single spanning arc of @xmath34 , as required . as a meridian of @xmath175 . ]
a tunnel system for a knot corresponds to a heegaard splitting of the knot complement .
a heegaard splitting , in turn , can be untelescoped to produce a strongly irreducible generalized heegaard splitting , and vice versa .
when we consider degeneration of tunnel number , we shall be concerned with constructing generalized heegaard splittings for each @xmath185 from a strongly irreducible generalized heegaard splitting for @xmath186 .
the more general context for this section is this : let @xmath20 be a compact orientable @xmath0-manifold and @xmath150 a properly imbedded collection of essential annuli in @xmath20 .
suppose @xmath187 is a strongly irreducible generalized heegaard splitting of @xmath20 isotoped so that @xmath188 consists only of curves essential in both @xmath83 and @xmath150 , and such that this number is minimal .
[ defn : dipping ] an annulus component ( or its closure ) @xmath34 of @xmath189 is called a annulus if , for some @xmath190 , @xmath191 and @xmath192 .
[ thm : bif ] let @xmath193 be the components into which @xmath20 is divided by the family of annuli @xmath150 .
then for each @xmath194 there is a generalized heegaard splitting @xmath195 of @xmath196 such that @xmath197 where @xmath198 is the number of dipping annuli among the components of @xmath189 .
* proof : * the central problem in discerning heegaard splittings in the @xmath196 is that cutting @xmath63 or @xmath64 along @xmath150 does not necessarily create compression bodies .
for example , cutting @xmath83 along @xmath150 does not even produce closed surfaces . we endeavor to remedy this fact by longitudinally attaching a solid torus ( which we could view as a collar of @xmath199 ) to each of the annuli in @xmath200 and imbedding in each torus certain annuli with longitudinal boundary .
these annuli , when attached to surfaces @xmath201 ( suitably stabilized ) and @xmath202 will be shown to yield generalized heegaard splittings of the @xmath196 . the first step will be to describe how the new annuli are to be imbedded in the solid torus collar @xmath203 of each @xmath204 .
in @xmath20 itself , it s natural to define the " distance between two of the surfaces in @xmath205 as the smallest number of compression bodies one needs to pass through to get from a point in one surface to a point in the other .
so , for example , the distance from @xmath206 to @xmath69 is @xmath207 and from @xmath76 to @xmath208 the distance is @xmath209 .
@xmath188 is a collection of parallel essential curves in the annulus @xmath34 .
let @xmath210 be a spanning arc of @xmath34 that meets each component of @xmath188 exactly once .
parameterize by @xmath211 .
that is , choose a homeomorphism @xmath212 $ ] .
let @xmath213 $ ] be a continuous extension of the function that assigns to each point in @xmath214 its distance from @xmath76 .
we may as well take @xmath215 to be as simple as possible .
for example , on a segment of @xmath216 that runs between @xmath69 and @xmath217 , say , define @xmath215 to monotonically run from @xmath207 to @xmath218 . on a segment of @xmath216 that lies in @xmath64 and has both ends on @xmath208 ,
define @xmath215 so that it has a single maximum .
let @xmath219 \times [ 0 , 2m]$ ] be the corresponding imbedding of @xmath216 in the first quadrant of @xmath220 , with the endpoints of @xmath216 on the @xmath221-axis .
informally , @xmath222 identifies with the graph of its distance from @xmath76 .
let @xmath149 be the disk that lies below @xmath223 in @xmath224 \times [ 0 , 2 m ] \subset { \mathbb r}^2 $ ] .
each line @xmath225 ( respectively @xmath226 ) intersects @xmath227 at points at which @xmath69 ( respectively @xmath208 ) intersects . in view of this correspondence , let @xmath228 denote the intersection of the line @xmath225 with @xmath149 , @xmath229 denote the intersection of the line @xmath226 with @xmath149 , @xmath230 and @xmath231
. see figure 4 .
consider a component @xmath232 of @xmath233 .
@xmath232 is a polygon with an even number of sides .
the sides lie alternately in @xmath223 and @xmath234 . .
] if @xmath232 is a bigon , one of its sides is a subarc of @xmath223 and one of its sides is a component of @xmath235 , since an incompressible annulus with both ends on @xmath236 in a compression body @xmath237 is @xmath96-parallel .
moreover @xmath215 has a maximum on the corresponding subarc of @xmath216 , since @xmath149 lies below the graph .
that is , the annulus on which the corresponding arc of lies is in some @xmath64 .
if @xmath232 is a quadrilateral , two opposite sides are subarcs of @xmath223 , one side is a component of @xmath238 and one a component of @xmath235 .
if @xmath232 has @xmath2 sides with @xmath239 , then @xmath240 sides will be subarcs of @xmath223 , one side will be a component of @xmath238 and all other sides will be components of @xmath235
. thus there will be @xmath241 sides that are subarcs of @xmath223 and that connect two sides that are components of @xmath242 ; these correspond to spanning arcs of dipping annuli in some @xmath63 .
let @xmath243 be the collection of subarcs of @xmath210 that , in the boundary of some @xmath232 , connect two sides that are components of @xmath235 .
then there is a correspondence between the components of @xmath243 and the collection of dipping annuli that lie in @xmath34 .
now set @xmath244 .
@xmath203 will be attached to each of the two copies of @xmath34 in @xmath245 by the obvious identification @xmath246 of @xmath247 with @xmath248 .
the surfaces @xmath249 can then be extended into @xmath203 by just attaching the collection of annuli @xmath250 .
this operation will not necessarily create compression bodies , but it will do so if we first modify @xmath251 as described below .
stabilize @xmath251 by attaching tubes , one running parallel to a spanning arc on each dipping annulus and lying on the root side of the annulus in the compression body @xmath63 in which it lies .
denote the resulting generalized heegaard splitting of @xmath20 , now of genus @xmath198 higher than originally , by @xmath252 denote @xmath253 by @xmath254 .
it follows from lemma [ lem : push ] that each component of the complement in @xmath255 of the dipping annuli is a compression body on whose boundary the collection of incident dipping annuli are independently longitudinal .
* claim : * attaching each @xmath256 to @xmath196 and capping off the surfaces @xmath257 by the annuli @xmath258 yields a generalized heegaard splitting of @xmath196 . for any @xmath259 , cutting along non - spanning components of @xmath260 ( or @xmath261 ) yields compression bodies by lemma [ lem : cbs ] .
cutting along spanning annuli yields @xmath262 for some compact orientable surface @xmath263 .
when @xmath203 is attached to @xmath196 , then for @xmath232 a region as above , manifolds of the form @xmath264 are attached to @xmath265 .
* case 1 : * @xmath232 is a bigon .
note that a bigon corresponds precisely to a non - spanning annulus component of @xmath266 .
then @xmath264 is a solid torus that is attached to @xmath265 along a longitudinal annulus of @xmath264 .
this does not change the homeomorphism type of @xmath265 .
* case 2 : * @xmath232 is a quadrilateral .
attaching @xmath264 to @xmath265 yields @xmath267 , where @xmath268 is the compact surface obtained by connecting two boundary components of @xmath263 by an annulus .
* case 3 : * @xmath232 is an @xmath2-gon with @xmath239 . here attaching @xmath264 to @xmath269
has the same effect as attaching a @xmath270 , but in addition , attachments are also made along dipping annuli in the corresponding @xmath63 .
@xmath254 has been constructed so that each dipping annulus is independently longitudinal in the component of @xmath271 on which it lies , so the result is still of the form @xmath267 , essentially by lemma [ lem : glue ] . since all components of @xmath272 are eventually connected by annuli in this process , the result is a union @xmath273 ( or @xmath274 ) of compression bodies .
theorem [ thm : bif ] will eventually be applied to the family of annuli @xmath275 that decompose @xmath276 into @xmath277 certain annuli in the tori @xmath203 , now described , will be useful in the next section .
[ defn : plumb ] let @xmath278 be the disk described in the proof of theorem [ thm : bif ] .
let @xmath279 be the intersection of a vertical line @xmath280 with @xmath149 .
then the annulus @xmath281 is called a annulus in @xmath244 .
see figure 4 .
let @xmath282 denote the manifold obtained from @xmath196 by attaching the solid tori @xmath283 to @xmath196 .
note that each plumbline annulus in @xmath284 will intersect each splitting surface @xmath285 in at most one component .
if we replace the collection of annuli in theorem [ thm : bif ] by a collection of essential (= incompressible and not @xmath96-parallel ) tori , then an analogous proof still applies .
the spanning arc @xmath210 must be replaced by an appropriate essential curve on @xmath286 .
more specifically , note that the components of @xmath287 are all parallel .
the curve that replaces @xmath210 must be an essential curve on @xmath286 that intersects each component of @xmath287 exactly once .
the graph of this curve in @xmath288 , constructed in analogy to @xmath223 in the proof of theorem [ thm : bif ] , will cut out an annulus .
this annulus replaces @xmath149 in the construction .
see figure 5 .
this yields the following result .
-axis becomes @xmath289 . ]
[ thm : torus ] let @xmath20 be an orientable @xmath0-manifold containing a family of essential tori @xmath290 .
suppose @xmath291 is a strongly irreducible generalized heegaard splitting of @xmath20 isotoped so that @xmath292 consists only of curves essential in both @xmath83 and @xmath21 , and such that this number is minimal .
let @xmath193 be the components into which @xmath20 is divided by @xmath21 .
then for each @xmath194 there is a generalized heegaard splitting @xmath293 of @xmath196 such that @xmath294 where @xmath198 is the number of dipping annuli in @xmath295 .
the generalized heegaard splitting of each @xmath196 constructed in theorem [ thm : bif ] may not be strongly irreducible and , even if it is , it might still be simplified .
recall a heegaard splitting @xmath54 is if there are properly imbedded disks @xmath296 and @xmath297 such that @xmath298 .
[ rem : stab ] in this case , cutting @xmath55 along @xmath58 ( or @xmath59 ) yields another heegaard splitting of lower genus . a strongly irreducible splitting may amalgamate to a stabilized or reducible splitting , so ( to account for this ) the definition of a stabilized generalized heegaard splitting is necessarily a bit more complicated . [
defn : genstab ] a generalized heegaard splitting @xmath299 is if there are disks @xmath58 and @xmath59 such that @xmath58 is properly embedded in @xmath300 , for some @xmath259 , and @xmath59 is properly embedded in @xmath301 .
furthermore , @xmath298 and each component of @xmath302 is either a disk or an annulus spanning the compression body in which it lies .
if a generalized heegaard splitting is stabilized ( i. e. satisfies definition [ defn : genstab ] ) the associated amalgamated heegaard splitting is stabilized .
it is easy to see that we may assume the annuli components of @xmath302 are essential .
it furthermore follows from @xcite that if the generalized splitting is strongly irreducible , then we may assume each annulus component of @xmath303 or @xmath304 is a spanning annulus in a component of @xmath63 or @xmath64 that is a trivial compression body .
note that if a generalized splitting is stabilized we can create a generalized heegaard splitting of lower genus ( i. e. for which @xmath305 is reduced ) by amalgamating , reducing the genus as in remark [ rem : stab ] , and then untelescoping again .
here is an ad hoc criterion , useful in the present context , for showing that a given generalized splitting is stabilized .
[ defn : tubed ] a surface @xmath306 is a if it can be described as follows : start with a properly imbedded @xmath12-manifold @xmath307 , and a collection @xmath308 of @xmath309 disjoint arcs in the interior of @xmath149 such that @xmath310 . then @xmath311 is a union of tori and annuli .
create @xmath312 by attaching unnested tubes to @xmath313 along the arcs @xmath314 .
viewed dually , @xmath313 is obtained from @xmath312 by simultaneously compressing @xmath309 disks . for @xmath315 a component of @xmath312 , each component of @xmath313 that results from compressing @xmath315 is said to _
come from _
@xmath315 .
[ lem : adhoc ] suppose that a generalized heegaard splitting @xmath316 of a compact @xmath0-manifold @xmath20 intersects an essential solid torus @xmath317 so that the surface @xmath318 can be described as a tubed product with @xmath309 tubes as in definition [ defn : tubed ] .
suppose further that , for each @xmath319 , at most one arc component of @xmath320 comes from @xmath321
. then the splitting is stabilized at least @xmath309 times .
* proof : * the proof is by induction on @xmath322 . if @xmath323 then there is nothing on which to attach tubes , so @xmath324 and there is nothing to prove .
we may as well also assume that no component of @xmath308 is parallel to a subarc of @xmath325 ( that is , no disk component of @xmath326 has boundary the union of a component of @xmath308 and a subarc of @xmath325 ) . for
the corresponding tube is clearly a stabilization , so it can be removed without affecting the truth of the lemma .
suppose @xmath327 consists entirely of arcs .
since each @xmath69 is incompressible in both @xmath328 and @xmath329 , tubes can only have been attached to components of @xmath313 coming from the thick surfaces @xmath251 . by assumption
, there is at most one component of @xmath313 coming from any given @xmath208 and distinct @xmath208 s are separated by components of @xmath71 .
it follows that at least one component of @xmath308 is parallel to a subsegment of @xmath330 , a contradiction .
see figure 6 .
we proceed next to the case where there are closed components of @xmath331 .
each closed component @xmath332 abuts two components of @xmath333 ; call them the _ inside _ and _ outside _ regions neighboring @xmath334 , depending on whether or not @xmath334 separates the region from @xmath335 .
if the outside neighboring region of each closed component is an annulus , and the annulus spans some @xmath63 or @xmath64 then it follows , much as when there are no closed components , that each tube attached is a stabilization .
so we can focus on a closed component @xmath336 which is innermost among those whose outside neighboring component is not a spanning annulus .
it follows that the components of @xmath333 lying inside consist exactly of spanning annuli in the compression bodies , together with a single compressing disk for some @xmath208
. then all tubes coming from arcs in @xmath308 lying inside in @xmath337 are stabilizations .
this means that all components of @xmath83 lying inside the solid torus @xmath338 bounded by @xmath339 can be removed , and both @xmath309 and the number of stablizations will be reduced by the number of tubes lying therein .
indeed , @xmath339 itself can be removed , if it comes from @xmath340 .
so , by induction , we are left with the case in which @xmath334 comes from some @xmath208 , cuts off from @xmath337 a compressing disk in @xmath63 or @xmath64 , say , @xmath63 , and the outside neighbor of @xmath334 is not a spanning annulus of @xmath64 . by judicious choice of curves satisfying these properties
we may further assume that no arc in @xmath308 has both ends on .
( for such an arc is either parallel to a subarc of , and so violates our assumption above , or cuts off from the outside neighboring region another family of circles among which a substitute @xmath334 can be found .
see figure 7 . )
if no tube is attached to in the outside neighboring region , the component @xmath80 of @xmath64 containing that region will be a product , and so @xmath341 can be removed from @xmath83 and still leave a generalized heegaard splitting .
so we may as well assume that a tube is attached along an arc in the outside neighboring region of , and one end of lies on @xmath334 and the other end on another component @xmath342 of @xmath327 .
necessarily @xmath342 comes from @xmath208 , but possibly @xmath342 is an arc component of @xmath327 .
whether @xmath342 is an arc or is closed , the complement of the tube in the annulus @xmath343 is a disk which , together with a meridian disk @xmath344 of @xmath338 form a stabilizing pair for the splitting . compressing @xmath208 along @xmath344 then leaves a heegaard splitting , still intersecting @xmath83 in a tubed product , but with both @xmath309 and the number of stabilizations reduced by one .
since also @xmath322 is reduced by one , the result follows by induction .
we now turn to the setting of theorem [ thm : bif ] in which @xmath20 has a strongly irreducible generalized heegaard splitting and will henceforth assume that the family of annuli @xmath150 , separating @xmath20 into components @xmath193 , is _
complete_. that is , if @xmath345 is a properly imbedded incompressible annulus disjoint from , then @xmath158 is @xmath96-parallel in the component @xmath196 in which it lies .
for example , if @xmath346 , then the knot summation defines a collection @xmath150 of @xmath347 essential annuli in @xmath20 and this is a complete collection if and only if each knot is prime .
indeed , an annulus @xmath158 in a knot complement @xmath348 , with @xmath349 a pair of meridian disks , can be extended to become a decomposing sphere in @xmath6 by capping off @xmath349 with meridian disks .
a resulting summand is trivial if and only if @xmath158 is @xmath96-parallel in @xmath348 .
note that the condition that @xmath150 is complete is weaker than the assumption that each @xmath196 is acylindrical , since it says nothing about incompressible annuli in @xmath196 whose boundaries cross the curves @xmath350 . [ defn : wide ] suppose that the surface @xmath83 has been isotoped to intersect @xmath150 only in curves essential in both @xmath83 and @xmath150 and so that the number of components of @xmath351 is minimal subject to this condition .
then an annular component @xmath352 of @xmath353 is called a annulus if it is adjacent to ( that is , shares a boundary component with ) a dipping annulus in @xmath189 . a component @xmath338 of @xmath354 ( resp @xmath355 ) is if , for an annulus @xmath114 , there is an embedding @xmath356 and an open disk @xmath149 in @xmath114 so that @xmath338 is the image of @xmath357 , with @xmath358 .
exceptional components were first described near the end of the proof of theorem [ thm : indic ] .
see figure 1 .
note that @xmath359 is a @xmath360-punctured sphere and the closure of @xmath361 is a compressing disk for @xmath362 in @xmath63 ( resp @xmath64 ) . for @xmath338 an exceptional component of @xmath354 ( resp @xmath363 ) , the @xmath360-punctured sphere @xmath364 will be called an component of @xmath365 and the annuli obtained by compressing @xmath366 into @xmath63 ( resp @xmath64 ) will be called of @xmath208 .
[ lem : dest ] let @xmath20 be a compact @xmath0-manifold with strongly irreducible generalized heegaard splitting @xmath367 .
let @xmath193 be the components into which @xmath20 is divided by the complete collection of annuli @xmath150 .
suppose there are @xmath136 wide annuli among the components of @xmath353 and @xmath368 exceptional components .
suppose further that each heegaard splitting @xmath369 of @xmath196 in the construction in theorem [ thm : bif ] can be destabilized @xmath370 times
. then @xmath371 . *
proof : * let @xmath372 be an annulus disjoint from @xmath373 whose ends are essential ( hence core curves ) in components @xmath374 and @xmath375 of . here
we will allow @xmath376 but not if @xmath114 is parallel to a subannulus of @xmath376 . unless @xmath377 contains no annuli or exceptional components ,
such an annulus can be found , e. g. parallel to an annulus component of @xmath377 ( or a virtual annulus ) .
since @xmath150 is complete , @xmath114 is boundary parallel in @xmath196 and the annulus @xmath378 to which it is parallel contains as collars of its ends subannuli @xmath379 and @xmath380 with @xmath381 . denote the annulus @xmath382 by @xmath131 .
it is known ( see for example @xcite ) how a strongly irreducible heegaard splitting can intersect a solid torus under the conditions here , so we know that each @xmath208 intersects the solid torus @xmath286 lying between @xmath114 and @xmath383 in @xmath196 in a collection of @xmath96-parallel annuli , plus possibly a component in which two such annuli are tubed together by a @xmath96-parallel tube , i. e. an exceptional component .
since the number of curves in @xmath384 has been minimized , none of the annuli of @xmath385 ( and none of the virtual annuli ) has both its ends on the same component of @xmath386 .
thus each annulus component of @xmath385 ( or virtual annulus ) is an annulus much like @xmath114 .
so with no loss of generality , we may as well assume that @xmath114 is outermost , i. e. no other annulus component of @xmath249 ( or virtual annulus ) cuts off a solid torus containing @xmath114 .
also , @xmath286 can be parameterized as @xmath387 so that the annuli of @xmath385 plus the virtual annuli are just the product of a collection of proper arcs in @xmath388 with @xmath289 .
now attach to @xmath196 the solid tori described in theorem [ thm : bif ] to get @xmath389 .
assume , for initial simplicity , that neither end of @xmath114 abuts a dipping annulus on its root side .
extend @xmath114 by attaching to the ends of @xmath114 at @xmath374 and @xmath375 the plumbline annuli in the tori @xmath390 and @xmath391 described in definition [ defn : plumb ] . call the resulting annulus @xmath392 and denote by @xmath393 the solid torus that @xmath392 cuts off from @xmath389 . by construction , the heegaard splitting @xmath394 intersects @xmath393 in a tubed product , where the tubes are attached in the manner described in theorem [ thm : bif ] or , in the case of exceptional components , by the compressions that create the virtual annuli . by lemma [ lem : adhoc ] it suffices to show that the number of tubes not coming from exceptional components in this tubed product is at least half as large as the number of wide annuli among the components of @xmath395 .
the argument is little changed by assuming , as we henceforth do , that there are no exceptional components . under the homeomorphism @xmath396 ,
any component @xmath397 of @xmath398 is the product of a @xmath399-gon with @xmath289 .
the sides of the @xmath399-gon become annuli components lying in alternately @xmath400 and in @xmath189 . at most one of the sides in each @xmath399-gon
lies in @xmath401 .
if @xmath397 lies in some @xmath63 then either @xmath402 or @xmath403 of its sides are dipping annuli , depending on whether or not @xmath397 abuts @xmath401 .
when @xmath397 abuts @xmath401 then the number of wide annuli is correspondingly @xmath404 if @xmath405 and @xmath406 if @xmath407 .
when @xmath397 does not abut @xmath401 it is simply @xmath403 .
it similarly follows from the construction in theorem [ thm : bif ] that the number of tubes added in @xmath397 to create @xmath408 is @xmath402 if @xmath397 abuts @xmath401 and is @xmath403 or @xmath406 if it does not .
the required inequality then follows from the inequalities @xmath409 when @xmath407 and @xmath397 abuts @xmath401 and @xmath410 when @xmath411 and @xmath397 does not abut @xmath412 .
the previous argument can be extended to include the case in which one or both ends of @xmath114 abut a dipping annulus @xmath34 on its root side ( when the exact construction would need to handle a tube running through @xmath114 ; cf .
figure 8) : just slide the end of @xmath114 out ( away from @xmath286 ) just beyond the end of @xmath34 ( so the end of @xmath114 cuts off a small collar of a boundary component of @xmath413 ) before attaching plumbline annuli in @xmath414 .
the argument then goes through as above .
in this section we use cut and paste techniques to bound the number of dipping annuli in a strongly irreducible generalized heegaard splitting .
[ def : parallel ] suppose @xmath415 is a properly embedded collection of annuli .
then a component of @xmath416 that is a solid torus whose boundary intersects @xmath150 in a single longitudinal annulus is a .
put another way , an annulus that is @xmath96-parallel cuts off a parallelism , unless there are other components of @xmath150 lying between it and @xmath119 .
[ lem : disk ] suppose that @xmath417 is a collection of essential annuli in a compression body @xmath40 and suppose @xmath418 ( i. e. @xmath40 is not a product ) .
then there is an essential disk @xmath149 in @xmath40 such that @xmath149 is disjoint from @xmath150 .
moreover , if not all components of @xmath121 are spanning , then @xmath149 may be chosen so that at least one side of @xmath149 in @xmath419 lies on a parallelism of @xmath150 in @xmath419 .
* proof : * let @xmath420 be a set of defining disks for @xmath40 .
we argue by induction on @xmath421 .
if @xmath422 then all non - spanning components of @xmath150 are @xmath96-parallel in the product @xmath423 .
one of them cuts off a parallelism @xmath338 which , because @xmath150 is essential in @xmath40 , is adjacent to some of the disks in @xmath424 .
then a disk in the boundary annulus @xmath425 that contains all such disks corresponds , back in @xmath40 , to the required disk @xmath149 .
if @xmath426 , consider @xmath427 . since the annuli are incompressible , we can easily remove any circle of intersection .
if there is an arc in @xmath428 that is inessential in @xmath429 , then let @xmath210 be an outermost such arc in @xmath429 , and we may cut the disk @xmath149 in @xmath424 containing @xmath210 along @xmath210 and paste on two copies of the disk cut off by @xmath210 in @xmath429 to obtain a new disk @xmath430 . replacing @xmath149 by @xmath430 in @xmath424
produces a new set of defining disks @xmath431 with @xmath432 .
the result follows by induction . on the other hand ,
if all arcs in @xmath433 are essential in @xmath429 , let @xmath434 be an arc in @xmath433 that is outermost in @xmath424 .
let @xmath34 be the annulus in @xmath429 that gives rise to @xmath434 . cutting and pasting @xmath34 along @xmath434 and the outermost disk cut off in @xmath424 yields a disk @xmath430 disjoint from @xmath429 .
since @xmath34 is essential , it follows that @xmath430 is essential , and is as required .
[ lem : generalcount ] suppose there is a collection @xmath435 of essential annuli in a compression body @xmath40 and @xmath436 is the subcollection of non - spanning annuli .
let @xmath437 consist of those components which are incident to @xmath438 .
let @xmath439 , let @xmath440 and let @xmath441 .
let @xmath442 denote the number of annulus components of @xmath443 .
then @xmath150 has at most @xmath444 components .
* proof : * with no loss of generality we can ignore spanning annuli and assume @xmath445 .
let @xmath446 ; it suffices to show that @xmath447 .
the proof is by induction on @xmath448 . when @xmath449 there are no annuli , and there is nothing to prove .
when @xmath418 it follows from lemma [ lem : disk ] that there is a @xmath96-reducing disk @xmath149 for @xmath40 that is disjoint from @xmath150 .
the result of cutting @xmath40 along @xmath149 is either one or two compression bodies @xmath169 with @xmath450 . in particular
, @xmath451 is lower in ( each component of ) @xmath169 , so we can assume the lemma is true in @xmath169 .
remove all inessential annuli from @xmath452 and call the result @xmath174 .
define @xmath453 . by induction
, it suffices to show that @xmath454 .
the first step , cutting @xmath40 along @xmath149 , decreases @xmath455 by @xmath360 and raises @xmath442 by the number of inessential annuli that result , no more than @xmath1 .
so this step decreases @xmath456 by at least two .
it may create , however , one or two components of @xmath457 that are parallelisms .
next examine what happens when an inessential annulus is removed : @xmath458 goes up by two .
also @xmath442 drops by at least one and will drop by two if @xmath34 is adjacent to an annulus component of @xmath459 .
the latter will happen , for example , if the inessential annulus that s removed is parallel to another one .
thus this step either reduces the number of components of @xmath457 that are parallelisms and simultaneously increases @xmath456 by at most one , or leaves both @xmath456 and the number of parallelisms unchanged .
continue the process until all parallelisms ( which , at the beginning , are no more than two ) are eliminated .
the result is to increase @xmath456 by at most two . combining both steps ,
@xmath456 has not increased .
return now to the original context : @xmath20 is a compact orientable @xmath0-manifold and @xmath150 a complete properly imbedded collection of essential annuli in @xmath20 .
suppose @xmath460 is a strongly irreducible generalized heegaard splitting of @xmath20 isotoped so that @xmath461 consists only of curves essential in both @xmath83 and @xmath150 , and such that this number is minimal .
let @xmath193 be the components into which @xmath20 is divided by the family of annuli @xmath150 .
we apply lemma [ lem : generalcount ] to the annuli @xmath462 .
[ thm : generalcount ] for @xmath20 and @xmath150 as described above , let @xmath198 be the number of dipping annuli in @xmath463 , and @xmath136 be the number of wide annuli in @xmath353 . then @xmath464 * proof : * lemma [ lem : generalcount ] says that for each @xmath465 the number of dipping annuli in @xmath462 is at most @xmath466 , where @xmath442 ( defined in lemma [ lem : generalcount ] ) counts the number of annuli components of @xmath365 with a property that assures that they are adjacent to dipping annuli and therefore wide . that is ,
when lemma [ lem : generalcount ] is applied to @xmath63 the inequality remains true if @xmath442 is replaced by the number of wide annuli in @xmath208 .
summing over all @xmath63 we get @xmath467 as required . in the appendix ,
andrew casson constructs an example that the inequality of theorem [ thm : generalcount ] is in some sense best possible .
that is , given any @xmath468 there is an example for which @xmath469 and @xmath470 [ cor : general ] let @xmath20 and @xmath150 be as described above .
then there is a heegaard splitting @xmath471 for each @xmath196 so that @xmath472 * proof : * let @xmath198 denote the number of dipping annuli in @xmath473 , and let @xmath136 denote the number of wide annuli in @xmath474 .
the construction in theorem [ thm : bif ] yields generalized heegaard splittings for @xmath348 such that @xmath475 .
now substitute for @xmath198 from theorem [ thm : generalcount ] to get @xmath476 according to lemma [ lem : dest ] the induced heegaard splittings for the @xmath348 can be destabilized at least @xmath477 times , yielding heegaard splittings @xmath478 for each @xmath348 with @xmath479 .
it follows that @xmath472 [ cor : generaltunnel ] if @xmath480 are prime knots knots then @xmath481 * proof : * by lemma [ lem : ind ] there is a generalized heegaard splitting @xmath482 for which @xmath483 .
let @xmath150 be the @xmath347 annuli that decompose @xmath484 into the complements of the constituent prime knots .
we may assume that @xmath83 and @xmath150 have been chosen so that @xmath351 consists only of essential curves and then also minimizes the number of such intersections .
the family @xmath150 is complete since each @xmath26 is prime
. apply corollary [ cor : general ] , substitute from lemma [ lem : ind ] , and divide by two to get @xmath485 as required . following theorem [ thm : torus ] ,
the same sort of argument can be applied to essential tori in @xmath20 . for @xmath20 a knot complement ,
the application is to satellite knots .
@xcite let @xmath486 be a knot in a @xmath0-sphere @xmath6 and @xmath55 an unknotted solid torus in @xmath6 with @xmath487 .
assume that @xmath486 is not contained in a @xmath0-ball of @xmath55 .
a homeomorphism @xmath488 onto a tubular neighborhood @xmath489 of a nontrivial knot @xmath490 which maps a meridian of @xmath491 onto a longitude of @xmath492 maps @xmath486 onto a knot @xmath493 .
the knot @xmath5 is called a of @xmath492 , and @xmath492 is called its .
the pair @xmath494 is called a of @xmath5 .
[ thm : sat ] let @xmath5 be a satellite knot , then @xmath495 . * proof : * we merely sketch the proof .
let @xmath496 be a generalized heeggard splitting for @xmath32 for which @xmath497 .
so it intersects @xmath83 only in curves essential in both , and in a minimal number of these curves .
attach copies of @xmath499 to @xmath500 and to @xmath489 and complete the surfaces @xmath501 and @xmath502 to give heegaard splittings of @xmath503 and @xmath504 respectively , following theorem [ thm : torus ] .
the latter can be made a heegaard splitting of @xmath505 by just filling in a solid torus along @xmath498 .
destabilizations can be found for these heeegaard splittings just as in lemma [ lem : dest ] .
we can improve the count of dipping annuli in section [ sec : generalcount ] if we add the one further assumption that each component of @xmath353 has an even number of boundary components .
this condition will certainly be satisfied when the annuli are the decomposing annuli from a knot summation . more generally , [ def : parityclass ] for @xmath506 a collection of circles , the @xmath507 is the class that is non - trivial on each fundamental class of a component of @xmath508 .
informally , the parity class just counts the parity of the number of components , even or odd .
[ lem : parity ] suppose the properly embedded essential annuli @xmath415 have the property that the parity class of @xmath509 is the restriction of a class in @xmath510 .
as usual , let @xmath511 be the closed complementary components of @xmath150 in @xmath20 , so each annulus in @xmath150 becomes two annuli in @xmath512 .
if @xmath513 is an orientable , properly embedded surface in any @xmath196 , then @xmath514 is even .
* proof : * we may as well assume @xmath41 is connected .
let @xmath515 $ ] denote the non - trivial class of @xmath516 , let be the parity class of @xmath509 and let @xmath517 be a class such that @xmath518 . then , after homotoping @xmath519 to @xmath121 we see that the evaluation @xmath520 = [ \delta({\mbox{$\alpha$ } } ) , [ s ] ] = [ \delta(i^*(\tilde{{\mbox{$\alpha$ } } } ) ) , [ s ] ] = 0 $ ] by the exactness of the cohomology sequence for the pair @xmath521 .
[ def : paritycond ] a collection of annuli @xmath522 as in lemma [ lem : parity ] satisfies the .
this section will be a repeat of section [ sec : generalcount ] in the special case in which the annuli @xmath150 satisfy the parity condition in @xmath20 , so every component of @xmath353 has an even number of boundary components .
the more delicate analysis will require a new notion : [ def : special ] suppose that @xmath523 is a collection of essential annuli in a compression body @xmath40 .
a component @xmath338 of @xmath524 is called if there is a planar surface @xmath80 with boundary components @xmath525 and a homeomorphism @xmath526 .
see figure 9 .
the @xmath527 of the special component @xmath338 is defined to be @xmath528 . .
] for example , a boundary parallel annulus cuts off from @xmath40 a parallelism ( see definition [ def : parallel ] ) .
this is a special component of index @xmath1 , for in this case @xmath80 is an annulus and @xmath529 .
more generally , a special component is obtained from a collection of @xmath530 such examples by connecting them with @xmath531 @xmath12-handles .
an exceptional component ( see definition [ defn : wide ] ) is special in the compression body in which it lies .
its index is @xmath12 .
any @xmath96-reducing disk @xmath532 for @xmath40 that lies in a special component @xmath338 divides @xmath338 into two special components in the boundary - reduced manifold @xmath169 , and the sum of their indices is @xmath533 .
we now embark on improving the bound of lemma [ lem : generalcount ] under the parity assumption : every component of @xmath534 has an even number of boundary components .
the argument is in three stages , incorporated in the following lemmas that culminate in lemma [ lem : supercount ] : [ lem : subcount ] suppose there is a collection @xmath435 of essential annuli in a compression body @xmath40 and @xmath535 is the subcollection of non - spanning annuli .
let @xmath536 consist of those components which are incident to @xmath147 .
let @xmath439 and suppose every component of @xmath41 has an even number of boundary components .
let @xmath440 and let @xmath441 .
let @xmath537 be the number of annuli in that are adjacent to @xmath538 ( counted twice if @xmath538 abuts the annulus on both sides ) , @xmath2 be the number of components of that are not adjacent to @xmath538 , and @xmath403 be the number of spanning annuli .
let @xmath539 be the number of special components of @xmath540 ( see definition [ def : special ] )
. then @xmath541 * proof : * if a spanning component of @xmath435 is removed , the only effect is to lower @xmath403 by one and to raise @xmath542 $ ] by one .
so we may as well assume that there are no spanning annuli , so @xmath445 and @xmath543 .
the proof will be by induction on @xmath544 .
when @xmath545 then @xmath40 is a product , and there are no annuli .
in particular , @xmath546 so the inequality follows from @xmath547 adding @xmath12-handles on @xmath146 only increases @xmath548 and has no other effect , so the lemma follows from the case @xmath545 whenever @xmath549 .
so we henceforth assume @xmath550 . for the collection
@xmath523 define @xmath551 following lemma [ lem : disk ] , there is an essential disk @xmath149 in @xmath40 that is disjoint from @xmath150 and cuts off a parallelism from the component of @xmath180 in which it lies .
the result of cutting @xmath40 along @xmath149 is either one or two compression bodies @xmath169 with @xmath450 . in particular , @xmath451 is lower in ( each component of ) @xmath169 , so we can assume the lemma is true in @xmath169 . recall that @xmath149 has been chosen so that at least one of the adjacent components of @xmath552 is a parallelism of @xmath150 in @xmath169 . remove all inessential annuli from @xmath452 and call the result @xmath174 .
abbreviate @xmath553 to simply @xmath554 and set @xmath555 .
we will show that @xmath556 .
since it follows from the inductive hypothesis that @xmath557 we will then be able to conclude that @xmath558 as required .
the process by which @xmath559 can be calculated from @xmath554 consists of two steps : first @xmath96-reduce @xmath40 along @xmath149 , then remove all resulting inessential annuli .
we examine the effect of each move on @xmath554 ( in the interim extending the definition of @xmath554 also to the case in which @xmath150 may contain boundary parallel annuli ) .
there are a number of possible cases that arise in each step .
we will examine each in turn and determine the effect on each of the constituents of @xmath554 , denoting , for example , that @xmath560 goes up by at most two with the notation @xmath561 .
* step a : @xmath96-reducing @xmath40 along @xmath149 . * 1 .
_ @xmath149 is contained in a special component .
_ then @xmath562 , latexmath:[$ @xmath554 is unchanged .
@xmath149 is contained in a non - special component not in @xmath538 .
_ then @xmath562 , @xmath564 , @xmath565 so @xmath554 is unchanged .
@xmath149 lies in @xmath538 .
_ then @xmath562 and @xmath566 .
the annulus abutting the parallelism created either abuts @xmath538 on the other side ( in which case @xmath567 ) or does not ( in which case @xmath567 and @xmath568 ) . hence in any case
@xmath554 does not increase .
* step b : inessential annuli @xmath34 removed .
* there are various cases , depending on what sort of component @xmath569 of @xmath180 lies on the other side of the annulus @xmath34 ( we will say that @xmath34 lies _ on _ the component @xmath569 ) and whether or not the two boundary components of @xmath34 lie on the same or different components of @xmath570 ( we will say that @xmath34 is respectively non - separating or separating ) . for example , it may be that @xmath569 becomes special when @xmath34 is removed . in this case
@xmath34 is necessarily separating , since , by definition of `` special '' , @xmath571 is planar . 1 . _
@xmath34 is separating , does nt lie on @xmath538 , and is nt on a component that becomes special .
_ then @xmath572 , @xmath573 , @xmath574 ( the last since the product component cut off by @xmath34 is special by definition ) .
hence @xmath554 drops by @xmath12 .
@xmath34 lies on a component that becomes special .
_ then @xmath572 , @xmath573 so @xmath554 is unchanged .
@xmath34 is non - separating and does nt lie on @xmath538 .
_ then @xmath575 , @xmath573 , @xmath576 .
hence @xmath554 is unchanged .
@xmath34 is separating and lies on @xmath538 .
_ then @xmath577 , @xmath578 , @xmath579 , @xmath576 .
hence @xmath554 is unchanged .
@xmath34 is non - separating and lies on @xmath538 . _
then @xmath577 , @xmath580 , @xmath579 , @xmath581 hence @xmath554 is unchanged .
thus we conclude that @xmath582 as required .
[ lem : count ] with the hypotheses and notation of lemma [ lem : subcount ] , let also @xmath442 denote the number of annuli components of @xmath443 , as in lemma [ lem : generalcount ] .
then @xmath150 has at most @xmath583 components .
* proof : * by the special assumption that any orientable properly embedded surface in @xmath196 has an even number of @xmath96-components , each component of @xmath443 is either an annulus or has euler characteristic no greater than @xmath584 .
hence we have @xmath585 on the other hand , @xmath586 so @xmath587 adding inequality [ eq : s+ ] , equality [ eq : s- ] , and the equality @xmath588 we get @xmath589 so from lemma [ lem : subcount ] we get @xmath590 as required .
[ lem : supercount ] with the hypotheses and notation of lemma [ lem : count ] , let also @xmath591 be the sum of the indices of the special components of @xmath592 ( see definition [ def : special ] ) .
then @xmath150 has at most @xmath593 components .
* proof : * let @xmath594 ; it suffices to show that @xmath595 .
the proof is by induction on @xmath451 .
if there are no special components ( e. g. when @xmath545 ) , the result follows immediately from lemma [ lem : count ] . otherwise , let @xmath149 be a @xmath96-reducing disk for @xmath40 that lies in a special component @xmath338 of @xmath592
. we will show that @xmath96-reduction of @xmath40 along @xmath149 and the removal of any resulting inessential annuli can not increase @xmath456 in the compression body ( or compression bodies ) that arise .
the first step , cutting @xmath40 along @xmath149 , decreases @xmath596 by @xmath0 , raises @xmath591 by @xmath0 and raises @xmath442 by the number @xmath597 or @xmath1 of parallelisms in the two components @xmath598 .
thus @xmath599 is unchanged .
now examine the result of removing an inessential annulus @xmath162 that cuts off a parallelism , an annulus such as is created when @xmath40 is cut along @xmath149 .
let @xmath600 be the component of @xmath180 that abuts @xmath34 on the opposite side from @xmath338 .
@xmath600 may become special when @xmath34 is removed .
there are several cases : 1 . _
@xmath34 was not adjacent to an annulus component of @xmath601 _ then even if @xmath600 becomes special , @xmath602 so @xmath603 , @xmath604 , @xmath605 ( since the index of a parallelism is @xmath1 ) , @xmath606 . hence @xmath607 .
@xmath34 is adjacent to an annulus component of @xmath601 , @xmath600 is not special or @xmath602 .
_ then @xmath603 , @xmath608 , @xmath605 , @xmath606 .
hence @xmath609 .
@xmath34 is adjacent to an annulus component of @xmath601 , @xmath600 is special and @xmath610 . _
then @xmath603 , @xmath608 , @xmath611 , @xmath606 .
hence @xmath599 is unchanged .
@xmath34 is adjacent to an annulus component of @xmath601 , @xmath600 is special and @xmath612 . _ then @xmath600 becomes a parallelism and @xmath613 , @xmath608 , both @xmath591 and @xmath614 unchanged .
hence @xmath599 is unchanged .
so @xmath599 never increases and may decrease .
continue removing annuli cutting off parallelisms until @xmath615 , as it was before the disk @xmath149 was removed . in the end , @xmath456 also will not have increased .
return now to the original context : @xmath20 is a compact orientable @xmath0-manifold and @xmath150 is a complete properly imbedded collection of essential annuli in @xmath20 satisfying the parity condition ( see definition [ def : paritycond ] ) .
suppose @xmath460 is a strongly irreducible generalized heegaard splitting of @xmath20 isotoped so that @xmath461 consists only of curves essential in both @xmath83 and @xmath150 , and such that this number is minimal .
let @xmath193 be the components into which @xmath20 is divided by the family of annuli @xmath150 .
we will soon apply lemma [ lem : supercount ] to the annuli @xmath462 .
notice that the index of a special component of @xmath355 is non - positive unless the component @xmath338 is of index @xmath12 , when @xmath616 is a @xmath360-punctured sphere .
since @xmath617 visibly compresses in @xmath338 , it follows from strong irreducibility that @xmath617 also compresses in @xmath64 .
it then follows from a standard outermost arc argument , that @xmath618 compresses in @xmath619 .
the result is two annuli .
if the two annuli are parallel , this implies that @xmath338 is exceptional ( see definition [ defn : wide ] ) .
if they are not , then the union of the two annuli , together with the annuli @xmath620 , cuts off a @xmath1-bridge knot or link complement from @xmath416 , necessarily one of the @xmath196 since @xmath150 is complete . a component of @xmath196 which is a @xmath1-bridge knot or link complement containing , as above
, an index @xmath12 special component of some @xmath355 is called an @xmath1-bridge complement .
the following theorem now improves theorem [ thm : generalcount ] , when @xmath415 satisfies the parity condition : [ thm : count ] for @xmath20 and @xmath150 as described above , let @xmath198 be the number of dipping annuli in @xmath463 , @xmath136 be the number of wide annuli in @xmath353 , and @xmath621 be the number of exceptional components cut off by @xmath150 from all the @xmath63 ( see definitions [ defn : dipping ] , [ defn : wide ] ) .
let @xmath622 be the number of interlaced @xmath1-bridge complements among the @xmath196 . then @xmath623 * proof : * we will apply lemma [ lem : supercount ] to the annuli @xmath462 .
suppose , as a first simplification , that each special component of each @xmath355 has non - positive index , so in particular @xmath624 .
then lemma [ lem : supercount ] says that the number of dipping annuli in @xmath462 is at most @xmath625 , where @xmath442 counts the number of annuli components of @xmath365 with a property ( defined in lemma [ lem : count ] ) that assures that they are adjacent to dipping annuli and therefore wide .
that is , when lemma [ lem : supercount ] is applied to @xmath63 the inequality remains true if @xmath442 is replaced by the number of wide annuli in @xmath208 .
summing over all @xmath63 we get @xmath626 as required .
now examine what happens if @xmath355 has a special component @xmath338 of positive index .
then , as explained above , @xmath338 is of index @xmath12 and so is either an exceptional component or part of an interlaced @xmath1-bridge complement . in either case , the positive index contributed by @xmath338 to @xmath591 in lemma [ lem : supercount ] is incorporated into @xmath627 .
[ lem : partial ] for @xmath20 and @xmath150 as described above , let @xmath628 be the manifold obtained from @xmath20 by replacing each interlaced @xmath1-bridge complement @xmath196 with an unknot or unlink complement ( matching longitudes )
. then @xmath628 has a ( perhaps weakly reducible ) generalized heegaard splitting @xmath629 so that each @xmath630 and @xmath631 .
* proof : * one way of constructing @xmath628 would be , on each index @xmath12 special component @xmath632 that comes from an interlaced @xmath1-bridge complement , reglue @xmath63 to @xmath64 differently along @xmath633 . depending on how the @xmath96-reducing disk @xmath634 for @xmath64 separates pairs of boundary components of the @xmath360-punctured sphere @xmath617 , this regluing can be done so that the @xmath96-reducing disk @xmath635 of @xmath338 intersects @xmath636 in either two points or none . in the former case ,
the replacement is by an unknot complement and @xmath338 becomes exceptional ( see definition [ defn : wide ] ) . in the latter ,
the replacement is by the complement of an unlink of two components , and @xmath628 is reducible .
[ cor : special ] let @xmath20 and @xmath150 be as described above , with @xmath415 a complete collection of annuli satisfying the parity condition .
order the summands so that the interlaced @xmath1-bridge complements among the @xmath196 , if any , are @xmath637 .
then there is a heegaard splitting @xmath471 for each @xmath196 so that @xmath638 * proof : * we may assume that the surfaces @xmath83 from the generalized heegaard splitting and the annuli @xmath150 have been isotoped so that @xmath351 consists only of essential curves and a minimal number of them .
let @xmath198 denote the number of dipping annuli in @xmath639 , @xmath136 denote the number of wide annuli in @xmath353 , @xmath368 denote the total number of exceptional components in the @xmath355 and the @xmath354 , and @xmath621 denote only the number that lie in the @xmath640 .
the construction in theorem [ thm : bif ] yields generalized heegaard splittings for @xmath348 such that @xmath475 .
now substitute for @xmath198 from theorem [ thm : count ] to get @xmath641 according to lemma [ lem : dest ] the induced heegaard splittings for the @xmath348 can be destabilized at least @xmath642 times , yielding heegaard splittings @xmath478 for each @xmath348 with @xmath643 it follows that @xmath644 as required .
in this section we apply the results above to tunnel numbers of composite knots .
the results can also be formulated for composite links , though the statements are sometimes a bit more cumbersome .
[ thm : sum ] let @xmath645 be prime knots , and suppose @xmath646 are the @xmath1-bridge knots among the @xmath26 . then there is some @xmath647 such that @xmath648 and @xmath649 is no smaller than any of : * @xmath650 * @xmath651 * @xmath652 * proof : * by lemma [ lem : ind ] there is a generalized heegaard splitting @xmath482 for which @xmath483 . let @xmath150 be the @xmath347 annuli that decompose @xmath484 into the complements of the constituent prime knots . clearly
the non - trivial element of @xmath653 restricts to the parity class on meridians , so the parity condition holds for @xmath654 .
let @xmath655 be the number of components of the @xmath348 that are not interlaced @xmath1-bridge complements .
the family @xmath150 is complete since each @xmath26 is prime .
it then follows from corollary [ cor : special ] that @xmath638 since @xmath1-bridge knots have tunnel number one , we further know that , for @xmath656 , hence @xmath657 , that @xmath658 .
hence another way to write the inequality is @xmath659 or , substituting from lemma [ lem : ind ] and dividing by two , @xmath660 this gives the first inequality .
the second inequality follows from the fact that each @xmath63 contributes at least @xmath1 to @xmath661 and the number @xmath662 of interlaced @xmath1-bridge knot complements can be no bigger than @xmath537 .
the last inequality follows from lemma [ lem : partial ] . here
@xmath628 is the manifold obtained by replacing each @xmath663 with the complement of the unknot .
hence @xmath664 .
it has a generalized heegaard splitting whose constituent compression bodies are the same as that of the untelescoped minimal genus heegaard splitting for @xmath20 itself , so when the splitting of @xmath628 is amalgamated , the result is a heegaard splitting of genus no higher than that of @xmath20 .
let @xmath645 be prime knots .
1 . @xmath665 .
2 . if none of the @xmath26 are @xmath1-bridge knots then @xmath666 .
3 . @xmath667 .
* proof : * the first statement of course is just corollary [ cor : generaltunnel ] but here it can be seen ( somewhat intriguingly ) to follow also from the first and second inequalities of theorem [ thm : sum ] and the fact that , for any @xmath668 , @xmath669 .
the second statement follows immediately from theorem [ thm : sum ] by setting @xmath670 .
the last is immediate if both or neither of the @xmath26 are @xmath1-bridge .
if exactly @xmath8 is @xmath1-bridge , the result follows from the last inequality of theorem [ thm : sum ] .
andrew casson [ example ] for any number @xmath671 , there is a 3-manifold @xmath672 where @xmath673 and @xmath674 are compression bodies , and an essential annulus @xmath34 properly embedded in @xmath20 , such that the following conditions are satisfied . 1 .
@xmath675 is an incompressible torus and @xmath676 .
2 . @xmath677 , and @xmath678 is strongly irreducible .
@xmath679 is incompressible in @xmath20 .
the components of @xmath680 , @xmath681 are essential sub - annuli of @xmath34 .
if @xmath673 has @xmath682 2-handles , and @xmath680 contains @xmath683 components which are disjoint from @xmath684 , then @xmath685 . 1
. @xmath690 has @xmath2 2-handles and @xmath693 is connected .
@xmath694 is strongly irreducible .
@xmath692 consists of @xmath695 essential , non - parallel annuli disjoint from @xmath693 together with @xmath696 essential , non - parallel annuli meeting both @xmath697 and @xmath693 .
4 . @xmath698 .
@xmath699 consists of @xmath696 annuli , each meeting both boundary components of @xmath694 .
* proof : * let @xmath700 be the unit sphere @xmath701 .
let @xmath702 , @xmath703 , @xmath704 and @xmath705 be the reflections of @xmath700 in the planes @xmath706 , @xmath707 , @xmath708 and @xmath709 respectively , generating a group @xmath243 of order 8 .
the fixed point sets of @xmath702 and @xmath703 are the circles @xmath710 and @xmath711 where @xmath700 intersects the @xmath712 and @xmath713 planes respectively .
let @xmath714 .
let @xmath715 be disjoint closed disks in the interior of @xmath716 such that @xmath717 , for example , small round disks in @xmath700 with centers evenly spaced on the semicircle where @xmath716 meets the plane @xmath708 .
let @xmath718 and @xmath719 ; then @xmath149 is a @xmath243-invariant disjoint union of @xmath720 disks in @xmath700 .
( see figure 10 . )
set @xmath721 and @xmath722 , so @xmath80 is a @xmath720-punctured sphere on which @xmath243 acts .
let @xmath723 , where @xmath724 is a connected surface with @xmath725 such that the action of @xmath243 extends over @xmath724 .
for example , one could choose @xmath726 ; then @xmath41 is the double of @xmath80 , a surface of genus @xmath727 .
let @xmath729 , and let @xmath730 as a subset of of @xmath731 .
there is a homeomorphism @xmath732 such that @xmath733 and @xmath734 for all @xmath735 outside a small neighborhood @xmath29 of @xmath736 , and @xmath737 for every point @xmath738 .
let @xmath690 be the 3-manifold obtained from the disjoint union of @xmath739 and @xmath740 by identifying each point @xmath741 with @xmath742 . by using a collar neighborhood of @xmath232 in @xmath740 , construct an embedding @xmath743 such that @xmath744 for all @xmath738 , and @xmath745 is disjoint from the subset @xmath746 of @xmath690 .
let @xmath747 be disjoint arcs properly embedded in @xmath748 , with end - points on @xmath736 and such that @xmath749 separates @xmath750 from each @xmath751 with @xmath752 .
then @xmath753 is an embedded disk in @xmath690 with boundary in @xmath754 .
this exhibits @xmath690 as a compression body with @xmath755 and with @xmath756 as the cores of the 2-handles .
( see figure 11 . ) since @xmath724 was chosen to be connected , @xmath693 is connected . for @xmath757 let @xmath758 be a simple loop in @xmath748 that is parallel to the boundary component @xmath759 and invariant under @xmath704 .
enlarge @xmath760 to a maximal collection @xmath761 of disjoint essential non - parallel simple loops in @xmath748 in such a way that @xmath762 for all @xmath259 .
( see figure 12 . )
set @xmath763 ; then @xmath764 are disjoint properly embedded annuli in @xmath690 . if @xmath758 is chosen disjoint from the neighborhood @xmath29 of @xmath736 , then @xmath765 .
@xmath690 also contains disjoint properly embedded annuli @xmath766 and @xmath767 , where @xmath768 and @xmath769 .
let @xmath692 be the union of the annuli @xmath770 , @xmath771 and @xmath772 for @xmath773 .
let @xmath774 , where @xmath775 and @xmath776 are dehn twists in @xmath710 and @xmath711 respectively .
by corollary [ strong ] below , @xmath694 is strongly irreducible .
the intersection of @xmath692 with @xmath697 is the union of the curves @xmath758 , @xmath777 , @xmath778 and @xmath779 for @xmath773 , which is invariant under @xmath222 . in @xmath694 , each of @xmath780 and @xmath781 is a single properly embedded annulus .
[ wave ] let @xmath29 be a regular neighborhood of @xmath782 in @xmath41 .
suppose @xmath18 is an essential simple loop in @xmath41 which bounds a disk in @xmath690 and intersects @xmath783 minimally .
then some component of @xmath784 intersects @xmath710 and is disjoint from @xmath711 . *
proof : * a similar argument is given in ( * ? ? ? * appendix ) .
it follows from the construction of @xmath690 as @xmath785 that @xmath786 is an hnn extension of @xmath787 , so @xmath787 injects into @xmath786 . moreover , every loop in @xmath41 which bounds a disk in @xmath690 has zero homological intersection number with @xmath710 .
therefore every essential loop in @xmath41 which bounds a disk in @xmath690 intersects @xmath710 at least twice .
observe that @xmath711 bounds a disk in @xmath690 and intersects @xmath710 in exactly two points ; the regular neighborhood @xmath29 of @xmath782 is a 4-punctured sphere .
suppose that @xmath18 is an essential simple loop in @xmath41 which bounds a disk in @xmath690 and intersects @xmath788 minimally , but every component of @xmath784 either intersects @xmath711 or is disjoint from @xmath710 .
since some component of @xmath784 intersects @xmath710 , @xmath18 must intersect @xmath711 also .
therefore @xmath18 contains an arc @xmath210 ( a wave ) with @xmath789 and such that if @xmath434 is either arc of @xmath790 then @xmath791 bounds a disk in @xmath690 .
each component of @xmath792 not containing an end - point of @xmath210 is a component of @xmath784 which is disjoint from @xmath711 , and is therefore disjoint from @xmath710 also .
therefore every component of @xmath792 contains an end - point of @xmath210 .
it follows that @xmath793 is isotopic in @xmath794 to an arc @xmath795 not meeting @xmath710 ; let @xmath796 , @xmath797 be the arcs of @xmath798 . then @xmath799 , @xmath800 are essential simple closed curves in @xmath41 bounding disks in @xmath690 , and at least one intersects @xmath710 in less than two points , a contradiction .
[ end ] let @xmath41 be a closed orientable surface , possibly disconnected , and let @xmath801 be disjoint essential simple closed curves on @xmath41 . then there is a compact 3-manifold @xmath29 with incompressible boundary @xmath802 , containing disjoint annuli @xmath803 such that @xmath804 . * proof : * let @xmath803 be annuli , and construct @xmath805 where @xmath806 is a regular neighborhood of @xmath807 in @xmath808 .
the boundary of @xmath809 consists of @xmath41 together with a closed surface @xmath107 ; since @xmath758 and @xmath810 are essential , @xmath41 and @xmath107 are incompressible in @xmath809 .
choose a a compact 3-manifold @xmath811 with incompressible boundary @xmath812 , and set @xmath813 .
* proof of theorem [ example ] * choose @xmath814 and let @xmath690 , @xmath222 and @xmath692 be as in lemma [ unit ] .
let @xmath569 , @xmath815 be parallel essential simple closed curves on a torus @xmath286 .
let @xmath41 be the disjoint union of @xmath286 and @xmath693 , and let @xmath816 be the components of @xmath817 , in any order . by lemma [ end ]
, there is a 3-manifold @xmath29 with incompressible boundary @xmath818 , containing disjoint annuli @xmath819 such that @xmath820 if instead @xmath821 and @xmath822 are the components of @xmath823 in any order then lemma [ end ] gives a 3-manifold @xmath809 with incompressible boundary @xmath693 and containing disjoint annuli @xmath766 such that @xmath824 by ( * ? ? ?
* lemma 3 ) , @xmath828 has a strongly irreducible untelescoped heegaard splitting @xmath829 such that the components of @xmath830 and @xmath831 are essential sub - annuli of @xmath770 .
similarly , @xmath832 has an untelescoped heegaard splitting @xmath833 with similar properties .
let @xmath647 be the total number of 1-handles in these generalized heegaard splittings for @xmath29 and @xmath809 .
choose @xmath834 and let @xmath835 denote the union @xmath836 of @xmath537 copies of @xmath694 .
set @xmath837 with the generalized heegaard splitting @xmath838 and set @xmath839 , a single annulus properly embedded in @xmath20 . | we analyze how a family of essential annuli in a compact @xmath0-manifold will induce , from a strongly irreducible generalized heegaard splitting of the ambient manifold , generalized heegaard splittings of the complementary components .
there are specific applications to the subadditivity of tunnel number of knots , improving somewhat bounds of kowng @xcite .
for example , in the absence of @xmath1-bridge summands , the tunnel number of the sum of @xmath2 knots is no less than @xmath3 the sum of the tunnel numbers .
[ section ] [ thm]corollary [ thm]lemma [ thm]proposition [ thm]remark [ thm]definition [ thm]definitions |
the divide and color ( dac ) model is a natural dependent site percolation model introduced by hggstrm in @xcite .
it has been studied directly in @xcite , and as a member of a more general family of models in @xcite .
this model is defined on a multigraph @xmath10 , where @xmath11 is a multiset ( _ i.e. _ , it may contain an element more than once ) , thus allowing parallel edges between pairs of vertices . for simplicity , we will imprecisely call @xmath0 a _ graph _ and @xmath11 the _ edge set _ , even if @xmath0 contains self - loops or multiple edges . the dac model with parameters
@xmath12 $ ] , on a general ( finite or infinite ) graph @xmath0 with vertex set @xmath13 and edge set @xmath11 , is defined by the following two - step procedure : * first step : bernoulli bond percolation .
we independently declare each edge in @xmath11 to be open with probability @xmath9 , and closed with probability @xmath1 .
we can identify a bond percolation configuration with an element @xmath14 : for each @xmath15 , we define @xmath16 if @xmath17 is open , and @xmath18 if @xmath17 is closed . *
second step : bernoulli site percolation on the resulting cluster set . given @xmath14 ,
we call _ @xmath9-clusters _ or _ bond clusters _ the connected components in the graph with vertex set @xmath13 and edge set @xmath19 .
the set of @xmath9-clusters of @xmath20 gives a partition of @xmath13 . for each @xmath9-cluster @xmath21
, we assign the same color to all the vertices in @xmath21 .
the chosen color is black with probability @xmath2 and white with probability @xmath3 , and this choice is independent for different @xmath9-clusters .
these two steps yield a site percolation configuration @xmath22 by defining , for each @xmath23 , @xmath24 if @xmath25 is black , and @xmath26 if @xmath25 is white .
the connected components ( via the edge set @xmath11 ) in @xmath27 of the same color are called ( black or white ) _ @xmath2-clusters_. the resulting measure on @xmath28 is denoted by @xmath29 .
let @xmath30 denote the event that there exists an infinite black @xmath2-cluster . by standard arguments ( see proposition 2.5 in @xcite ) , for each @xmath31 $ ]
, there exists a _
critical coloring value _ @xmath32 $ ] such that @xmath33 the _ critical edge parameter _ @xmath34 $ ] is defined as follows : the probability that there exists an infinite bond cluster is @xmath35 for all @xmath36 , and positive for all @xmath37 .
the latter probability is in fact @xmath38 for all @xmath37 , whence @xmath39 for all such @xmath9 .
@xmath40 law shows that in the case when all the bond clusters are finite , @xmath41 ; nevertheless it is possible that @xmath42 for some @xmath43 ( _ e.g. _ on the square lattice , as soon as @xmath44 , one has @xmath45 ) .
our main goal in this paper is to understand how the critical coloring parameter @xmath5 depends on the edge parameter @xmath9 . since the addition or removal of self - loops obviously does not affect the value of @xmath4
, we will assume that all the graphs @xmath0 that we consider are without self - loops .
on the other hand , @xmath0 is allowed to contain multiple edges .
our first result , based on a stochastic domination argument , gives bounds on @xmath4 in terms of @xmath46 , which is simply the critical value for bernoulli site percolation on @xmath0 . by the _ degree _ of a vertex @xmath25 , we mean the number of edges incident on @xmath25 ( counted with multiplicity ) .
[ rcbounds ] for any graph @xmath0 with maximal degree @xmath47 , for all @xmath48 , @xmath49 as a direct consequence , we get continuity at @xmath8 of the critical value function : [ contin0 ] for any graph @xmath0 with bounded degree , @xmath4 is continuous in @xmath9 at @xmath35 .
one could think of an alternative approach to the question , as follows : the dac model can be seen as bernoulli site percolation of the random graph @xmath50 where @xmath51 is the set of bond clusters and two bond clusters are connected by a bond of @xmath52 if and only if they are adjacent in the original graph .
the study of how @xmath4 depends on @xmath9 is then a particular case of a more general question known as the _ locality problem _ : is it true in general that the critical points of site percolation on a graph and a small perturbation of it are always close ? here , for small @xmath9 , the graphs @xmath0 and @xmath53 are somehow very similar , and their critical points are indeed close .
dropping the bounded - degree assumption allows for the easy construction of graphs for which continuity does not hold at @xmath8 : [ nonbounded ] there exists a graph @xmath0 with @xmath54 such that @xmath5 is discontinuous at @xmath35 . in general , when @xmath55 , the graph @xmath53 does not have bounded degree , even if @xmath0 does ; this simple remark can be exploited to construct bounded degree graphs for which @xmath5 has discontinuities below the critical point of bond percolation ( though of course not at @xmath35 ) : [ boundeddegreeconstruction ] there exists a graph @xmath0 of bounded degree satisfying @xmath56 and such that @xmath4 is discontinuous at @xmath57 .
the value @xmath57 in the statement above is not special : in fact , for every @xmath58 , it is possible to generalize our argument to construct a graph with a critical bond parameter above @xmath59 and for which the discontinuity of @xmath60 occurs at @xmath59 .
our main results concerns the case @xmath6 , for which the above does not occur : [ continp ] the critical coloring value @xmath61 is a continuous function of @xmath9 on the whole interval @xmath7 .
the other , perhaps more anecdotal question we investigate here is whether @xmath5 is monotonic below @xmath62 .
this is the case on the triangular lattice ( because it is constant equal to @xmath57 ) , and appears to hold on @xmath63 in simulations ( see the companion paper @xcite ) . in the general case
, the question seems to be rather delicate .
intuitively the presence of open edges would seem to make percolation easier , leading to the intuition that the function @xmath64 should be nonincreasing .
theorem 2.9 in @xcite gives a counterexample to this intuition .
it is even possible to construct quasi - transitive graphs on which any monotonicity fails : [ propnonmononicity ] there exists a quasi - transitive graph @xmath0 such that @xmath5 is not monotone on the interval @xmath65 .
a brief outline of the paper is as follows .
we set the notation and collect a few results from the literature in section [ definitions - section ] . in section [ section - stochastic - domination ] ,
we stochastically compare @xmath66 with bernoulli site percolation ( theorem [ stdom ] ) , and show how this result implies proposition [ rcbounds ] .
we then turn to the proof of theorem [ continp ] in section [ contz2-section ] , based on a finite - size argument and the continuity of the probability of cylindrical events . in section [ tree - like - section ] , we determine the critical value function for a class of tree - like graphs , and in the following section we apply this to construct most of the examples of graphs we mentioned above .
we start by explicitly constructing the model , in a way which will be more technically convenient than the intuitive one given in the introduction .
let @xmath0 be a connected graph @xmath67 where the set of vertices @xmath68 is countable .
we define a total order `` @xmath69 '' on @xmath13 by saying that @xmath70 if and only if @xmath71 . in this way , for any subset @xmath72
, we can uniquely define @xmath73 as the minimal vertex in @xmath74 with respect to the relation `` @xmath69 '' . for a set @xmath75
, we denote @xmath76 by @xmath77 .
we call the elements of @xmath78 _ bond configurations _ , and the elements of @xmath79 _ site configurations_. as defined in the introduction , in a bond configuration @xmath80 , an edge @xmath81 is called _ open _ if @xmath16 , and _ closed _ otherwise ; in a site configuration @xmath82 , a vertex @xmath23 is called _ black _ if @xmath83 , and _ white _ otherwise . finally ,
for @xmath84 and @xmath23 , we define the _ bond cluster _ @xmath85 of @xmath25 as the maximal connected induced subgraph containing @xmath25 of the graph with vertex set @xmath13 and edge set @xmath86 , and denote the vertex set of @xmath85 by @xmath87 . for @xmath88 $ ] and a set @xmath75 , we define @xmath89 as the probability measure on @xmath77 that assigns to each @xmath90 value @xmath38 with probability @xmath91 and @xmath35 with probability @xmath92 , independently for different elements of @xmath75 .
we define a function @xmath93 where @xmath94 . for @xmath95 $ ]
, we define @xmath96 to be the image measure of @xmath97 by the function @xmath98 , and denote by @xmath99 the marginal of @xmath96 on @xmath100 .
note that this definition of @xmath101 is consistent with the one in the introduction .
finally , we give a few definitions and results that are necessary for the analysis of the dac model on the square lattice , that is the graph with vertex set @xmath102 and edge set @xmath103 of the square lattice is the graph with vertex set @xmath102 and edge set @xmath104 . in the same manner as in the introduction ,
we define , for a color configuration @xmath105 , ( black or white ) _ @xmath106-clusters _ as connected components ( via the edge set @xmath107 ) in @xmath82 of the same color .
we denote by @xmath108 the @xmath109-probability that the origin is contained in an infinite black @xmath106-cluster , and define @xmath110 for all @xmath111 $ ] note that this value may differ from @xmath112 .
the main result in @xcite is that for all @xmath113 , the critical values @xmath61 and @xmath114 satisfy the duality relation @xmath115 we will also use exponential decay result for subcritical bernoulli bond percolation on @xmath102 . let @xmath116 denote the origin in @xmath102 , and for each @xmath117 , let us define @xmath118 ( where @xmath119 denotes graph distance ) , and the event @xmath120 there is a path of open edges in @xmath80 from @xmath121 to @xmath122 .
then we have the following result : [ lexpdecay ] for @xmath123 , there exists @xmath124 such that for all @xmath125 , we have that @xmath126
in this section , we prove proposition [ rcbounds ] via a stochastic comparison between the dac measure and bernoulli site percolation . before stating the corresponding result , however , let us recall the concept of stochastic domination .
we define a natural partial order on @xmath127 by saying that @xmath128 for @xmath129 if , for all @xmath23 , @xmath130 .
a random variable @xmath131 is called _ increasing _
if @xmath128 implies that @xmath132 , and an event @xmath133 is increasing if its indicator random variable is increasing . for probability measures
@xmath134 on @xmath100 , we say that @xmath135 is _ stochastically larger _
than @xmath136 ( or , equivalently , that @xmath136 is _ stochastically smaller _ than @xmath135 , denoted by @xmath137 ) if , for all bounded increasing random variables @xmath131 , we have that @xmath138 by strassen s theorem @xcite , this is equivalent to the existence of an appropriate coupling of the measures @xmath135 and @xmath136 ; that is , the existence of a probability measure @xmath139 on @xmath140 such that the marginals of @xmath139 on the first and second coordinates are @xmath141 and @xmath136 respectively , and @xmath142 .
[ stdom ] for any graph @xmath143 whose maximal degree is @xmath47 , at arbitrary values of the parameters @xmath95 $ ] , @xmath144 before turning to the proof , we show how theorem [ stdom ] implies proposition [ rcbounds ] .
[ [ proof - of - propositionrcbounds . ] ] proof of proposition [ rcbounds ] .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + it follows from theorem [ stdom ] and the definition of stochastic domination that for the increasing event @xmath145 ( which was defined in the introduction ) , we have @xmath146 whenever @xmath147 , which implies that @xmath148 .
the derivation of the lower bound for @xmath149 is analogous .
now we give the proof of theorem [ stdom ] , which bears some resemblance with the proof of theorem 2.3 in @xcite .
[ [ proof - of - theoremstdom . ] ] proof of theorem [ stdom ] .
+ + + + + + + + + + + + + + + + + + + + + + + + + fix @xmath10 with maximal degree @xmath47 , and parameter values @xmath95 $ ]
. we will use the relation `` @xmath69 '' and the minimum of a vertex set with respect to this relation as defined in section [ definitions - section ] . in
what follows , we will define several random variables ; we will denote the joint distribution of all these variables by @xmath150 .
first , we define a collection @xmath151 of i.i.d .
bernoulli(@xmath9 ) random variables ( _ i.e. _ , they take value @xmath38 with probability @xmath9 , and @xmath35 otherwise ) ; one may imagine having each edge @xmath15 replaced by two directed edges , and the random variables represent which of these edges are open .
we define also a set @xmath152 of bernoulli(@xmath2 ) random variables . given a realization of @xmath153 and @xmath154 , we will define an @xmath155-valued random configuration @xmath156 with distribution @xmath157 , by the following algorithm . 1 .
let @xmath158 no @xmath82-value has been assigned yet to @xmath159 by this algorithm@xmath160 .
( note that @xmath25 and @xmath161 @xmath162 , defined below , are running variables , _
i.e. _ , their values will be redefined in the course of the algorithm . ) 2 .
we explore the `` directed open cluster '' @xmath74 of @xmath25 iteratively , as follows .
define @xmath163 . given @xmath164 for some integer @xmath165 , set @xmath166 for every edge @xmath167 incident to @xmath168 such that no @xmath80-value has been assigned yet to @xmath17 by the algorithm , and write @xmath169
can be reached from any of @xmath170 by using only those edges @xmath15 such that @xmath16 has been assigned to @xmath17 by this algorithm@xmath160 .
if @xmath171 , then we define @xmath172 , and continue exploring the directed open cluster of @xmath25 ; otherwise , we define @xmath173 , and move to step 3 . 3 .
define @xmath174 for all @xmath175 , and return to step 1 .
it is immediately clear that the above algorithm eventually assigns a @xmath27-value to each vertex .
note also that a vertex @xmath25 can receive a @xmath27-value only after all edges incident to @xmath25 have already been assigned an @xmath80-value , which shows that the algorithm eventually determines the full edge configuration as well .
it is easy to convince oneself that @xmath156 obtained this way indeed has the desired distribution .
now , for each @xmath23 , we define @xmath176 if @xmath177 and @xmath178 for all edges @xmath179 incident on @xmath25 ( _
i.e. _ , all directed edges towards @xmath25 are closed ) , and @xmath180 otherwise .
note that every vertex with @xmath176 has @xmath181 as well , whence the distribution of @xmath82 ( _ i.e. _ , @xmath66 ) stochastically dominates the distribution of @xmath182 ( as witnessed by the coupling @xmath150 ) .
notice that @xmath183 depends only on the states of the edges pointing to @xmath25 and on the value of @xmath184 ; in particular the distribution of @xmath182 is a product measure on @xmath127 with parameter @xmath185 at @xmath25 , where @xmath186 is the degree of @xmath25 , whence @xmath66 stochastically dominates the product measure on @xmath127 with parameter @xmath187 , which gives the desired stochastic lower bound .
the upper bound can be proved analogously ; alternatively , it follows from the lower bound by exchanging the roles of black and white .
in this section , we will prove theorem [ continp ] .
our first task is to prove a technical result valid on more general graphs stating that the probability of any event @xmath188 whose occurrence depends on a finite set of @xmath27-variables is a continuous function of @xmath9 for @xmath36 .
the proof relies on the fact that although the color of a vertex @xmath25 may be influenced by edges arbitrarily far away , if @xmath36 , the corresponding influence decreases to @xmath35 in the limit as we move away from @xmath25 .
therefore , the occurrence of the event @xmath188 depends essentially on a finite number of @xmath80- and @xmath189-variables , whence its probability can be approximated up to an arbitrarily small error by a polynomial in @xmath9 and @xmath2 .
once we have proved proposition [ cylcont ] below , which is valid on general graphs , we will apply it on @xmath102 to certain `` box - crossing events , '' and appeal to results in @xcite to deduce the continuity of @xmath61 .
[ cylcont ] for every site percolation event @xmath190 depending on the color of finitely many vertices , @xmath191 is a continuous function of @xmath192 on the set @xmath193 $ ] . [ [ proof . ] ] proof .
+ + + + + + in this proof , when @xmath194 is a measure on a set @xmath75 , @xmath195 is a random variable with law @xmath194 and @xmath196 is a bounded measurable function , we write abusively @xmath197 $ ] for the expectation of @xmath198 .
we show a slightly more general result : for any @xmath199 , @xmath200 and @xmath201 bounded and measurable , @xmath202 $ ] is continuous in @xmath192 on the product @xmath203 $ ] .
proposition [ cylcont ] will follow by choosing an appropriate family @xmath204 such that the states of the @xmath205 suffices to determine whether @xmath188 occurs , and take @xmath206 to be the indicator function of @xmath188 .
to show the previous affirmation , we condition on the vector @xmath207 which takes values in the finite set @xmath208 , and we use the definition of @xmath209 as an image measure . by definition ,
@xmath210 \\ & = \sum_{\boldsymbol{v } \in \boldsymbol{v } } \mathbb{p}_{p , r } ^g\left [ f(\xi(x_1),\ldots,\xi(x_k ) ) | \ { \boldsymbol{m}_{\boldsymbol{x}}=\boldsymbol{v } \ } \right ] \mathbb{p}_{p , r } ^g \left [ \{\boldsymbol{m}_{\boldsymbol{x}}=\boldsymbol{v } \ } \right ] \\ & = \sum_{\boldsymbol{v } \in \boldsymbol{v } } \nu_p^{\mathcal{e } } \otimes \nu_r^{\mathcal{v } } \left [ f(\kappa(v_1),\ldots,\kappa(v_k ) ) | \ { \boldsymbol{m}_{\boldsymbol{x}}=\boldsymbol{v } \ }
\right ] \nu_p^{\mathcal{e } } \left [ \{\boldsymbol{m}_{\boldsymbol{x}}=\boldsymbol{v } \ } \right ] \\ & = \sum_{\boldsymbol{v } \in \boldsymbol{v } } \nu_r^{\mathcal{v } } \left [ f(\kappa(v_1),\ldots,\kappa(v_k))\right ] \nu_p^{\mathcal{e } } \left [ \{\boldsymbol{m}_{\boldsymbol{x}}=\boldsymbol{v } \ } \right].\end{aligned}\ ] ] note that @xmath211 $ ] is a polynomial in @xmath2 , so to conclude the proof we only need to prove that for any fixed @xmath212 and @xmath213 , @xmath214 depends continuously on @xmath9 on the interval @xmath215 . for @xmath216 ,
write @xmath217 .
it is easy to verify that the event @xmath218 depends on the state of finitely many edges .
hence , @xmath219 $ ] is a polynomial function of @xmath9 .
fix @xmath220 . for all @xmath221 , @xmath222 - \nu_p ^{\mathcal{e}}\left [ \left \ { \boldsymbol{m}_{\boldsymbol{x}}=\boldsymbol{v } \right \ } \cap f_n \right ] & \leq & \nu_p^{\mathcal{e } } \left [ f_n^c \right ] \\ & \leq & \nu_{p_0}^{\mathcal{e } } \left [ f_n^c \right ] \end{aligned}\ ] ] where @xmath223= 0
$ ] , since @xmath220 .
so , @xmath224 $ ] is a uniform limit of polynomials on any interval @xmath225 , \ : p_0<p_c^g$ ] , which implies the desired continuity . in the proof
we can see that , for fixed @xmath36 , @xmath191 is a polynomial in @xmath2 .
if @xmath0 is a graph with uniqueness of the infinite bond cluster in the supercritical regime , then it is possible to verify that @xmath226 $ ] is continuous in @xmath9 on the whole interval @xmath227 $ ] . in this case
, the continuity given by the proposition [ cylcont ] can be extended to the whole square @xmath227 ^ 2 $ ] .
[ [ proof - of - theoremcontinp . ] ] proof of theorem [ continp ] .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + in order to simplify our notations , we write @xmath228 , @xmath229 , for @xmath230 and respectively .
fix @xmath231 and @xmath232 arbitrarily .
we will show that there exists @xmath233 such that for all @xmath234 , @xmath235 and @xmath236 note that by equation ( [ rcplusrcstaris1 ] ) , for all small enough choices of @xmath237 ( such that @xmath238 ) , ( [ minus ] ) is equivalent to @xmath239 below we will show how to find @xmath240 such that we have ( [ plus ] ) for all @xmath241 .
one may then completely analogously find @xmath242 such that ( [ plusstar ] ) holds for all @xmath243 , and take @xmath244 .
fix @xmath245 , and define the event @xmath246 there exists a vertical crossing of @xmath247\times [ 0,3n]$ ] that is black in @xmath248 . by `` vertical crossing
, '' we mean a self - avoiding path of vertices in @xmath247\times [ 0,3n]$ ] with one endpoint in @xmath247\times \{0\}$ ] , and one in @xmath247\times \{3n\}$ ] .
recall also the definition of @xmath249 in theorem [ lexpdecay ] . by lemma 2.10 in @xcite
, there exists a constant @xmath250 such that the following implication holds for any @xmath251 $ ] and @xmath252 : @xmath253{rcl } ( 3l+1)(l+1)\nu _ { a}(m_{\lfloor l/3\rfloor } ) & \leq & \gamma , \\
\textrm{and } \mathbb{p}_{p , a}(v_l ) & \geq & 1-\gamma \end{array } \right \ } \rightarrow a\geq r_c(p).\ ] ] as usual , @xmath254 for @xmath255 denotes the largest integer @xmath256 such that @xmath257 .
fix such a @xmath258 . by theorem [ lexpdecay ]
, there exists @xmath259 such that @xmath260 for all @xmath261 . on the other hand , since @xmath262 , it follows from lemma 2.11 in @xcite that there exists @xmath263 such that @xmath264 note that both @xmath265 and @xmath266 are continuous in @xmath9 at @xmath59 .
indeed , the former is simply a polynomial in @xmath9 , while the continuity of the latter follows from proposition [ cylcont ] .
therefore , there exists @xmath267 such that for all @xmath241 , @xmath268 by the choice of @xmath258 , this implies that @xmath269 for all such @xmath9 , which is precisely what we wanted to prove .
finding @xmath242 such that ( [ plusstar ] ) holds for all @xmath270 is analogous : one only needs to substitute @xmath271 by @xmath272 and `` crossing '' by `` @xmath106-crossing , '' and the exact same argument as above works .
it follows that @xmath273 is a constant such that both ( [ plus ] ) and ( [ plusstar ] ) hold for all @xmath274 , completing the proof of continuity on @xmath275 .
right - continuity at @xmath35 may be proved analogously ; alternatively , it follows from proposition [ contin0 ] .
it follows from theorem [ continp ] and equation ( [ rcplusrcstaris1 ] ) that @xmath114 is also continuous in @xmath9 on @xmath7 .
in this section , we will study the critical value functions of graphs that are constructed by replacing edges of an infinite tree by a sequence of finite graphs .
we will then use several such constructions in the proofs of our main results in section [ proofs - section ] .
let us fix an arbitrary sequence @xmath276 of finite connected graphs and , for every @xmath277 , two distinct vertices @xmath278 .
let @xmath279 denote the ( infinite ) regular tree of degree @xmath280 , and fix an arbitrary vertex @xmath281 .
then , for each edge @xmath282 , we denote the end - vertex of @xmath17 which is closer to @xmath283 by @xmath284 , and the other end - vertex by @xmath285 .
let @xmath286 be the graph obtained by replacing every edge @xmath17 of @xmath287 between levels @xmath288 and @xmath289 ( _ i.e. _ , such that @xmath290 ) by a copy @xmath291 of @xmath292 , with @xmath293 and @xmath294 replacing respectively @xmath284 and @xmath285 .
each vertex @xmath295 is replaced by a new vertex in @xmath296 , which we denote by @xmath297 .
it is well known that @xmath298 .
using this fact and the tree - like structure of @xmath299 , we will be able to determine bounds for @xmath300 and @xmath301 .
first , we define @xmath302 , and prove the following , intuitively clear , lemma .
[ lemmabond ] for any @xmath31 $ ] , the following implications hold : * if @xmath303 , then @xmath304 ; * if @xmath305 , then @xmath306 . [
[ proof.-1 ] ] proof .
+ + + + + + we couple bernoulli bond percolation with parameter @xmath9 on @xmath307 with inhomogeneous bernoulli bond percolation with parameters @xmath308 on @xmath309 , as follows .
let @xmath20 be a random variable with law @xmath310 , and define , for each edge @xmath311 ,
@xmath312 if @xmath313 and @xmath314 are connected by a path consisting of edges that are open in @xmath80 , and @xmath315 otherwise .
the tree - like structure of @xmath299 implies that @xmath316 depends only on the state of the edges in @xmath291 , and it is clear that if @xmath317 , then @xmath312 with probability @xmath308 .
it is easy to verify that there exists an infinite open self - avoiding path on @xmath307 from @xmath318 in the configuration @xmath20 if and only if there exists an infinite open self - avoiding path on @xmath309 from @xmath283 in the configuration @xmath319 .
now , if we assume @xmath303 , then there exists @xmath320 and @xmath259 such that for all @xmath261 , @xmath321 .
therefore , the distribution of the restriction of @xmath319 on @xmath322 is stochastically dominated by the projection of @xmath323 on @xmath324 .
this implies that , a.s .
, there exists no infinite self - avoiding path in @xmath319 , whence @xmath304 by the observation at the beginning of this paragraph .
the proof of b ) is analogous .
we now turn to the dac model on @xmath307 .
recall that for a vertex @xmath25 , @xmath325 denotes the vertex set of the bond cluster of @xmath25 .
let @xmath326 denote the event that @xmath293 and @xmath294 are in the same bond cluster , or @xmath293 and @xmath294 lie in two different bond clusters , but there exists a vertex @xmath25 at distance @xmath38 from @xmath327 which is connected to @xmath294 by a black path ( which also includes that @xmath328 ) .
this is the same as saying that @xmath327 is _ pivotal _ for the event that there is a black path between @xmath293 and @xmath294 , _
i.e. _ , that such a path exists if and only if @xmath327 is black .
it is important to note that @xmath329 is independent of the color of @xmath293 .
define @xmath330 , and note also that , for @xmath331 , @xmath332there is a black path from @xmath293 to @xmath333 .
[ lemmadac ] for any @xmath95 $ ] , we have the following : * if @xmath334 , then @xmath335 ; * if @xmath336 , then @xmath337 . [
[ proof.-2 ] ] proof .
+ + + + + + we couple here the dac model on @xmath307 with inhomogeneous bernoulli _
percolation on @xmath309 . for each @xmath338
, there is a unique edge @xmath339 such that @xmath340 .
here we denote @xmath291 ( _ i.e. _ , the subgraph of @xmath307 replacing the edge @xmath17 ) by @xmath341 , and the analogous event of @xmath329 for the graph @xmath341 by @xmath342 .
let @xmath343 with values in @xmath344 be a random variable with law @xmath345 . we define a random variable @xmath195 with values in @xmath346 , as follows : @xmath347 as noted after the proof of lemma [ lemmabond ] , if @xmath348 , the event @xmath342 is independent of the color of @xmath349 , whence @xmath350 are independent .
therefore , as @xmath351 with probability @xmath2 , and @xmath352 is realized with probability @xmath353 for @xmath354 with @xmath355 for some @xmath277 , @xmath195 is inhomogeneous bernoulli site percolation on @xmath309 .
our reason for defining @xmath195 is the following property : it holds for all @xmath338 that @xmath356 where @xmath357 denotes that @xmath159 and @xmath358 are in the same _ black _ cluster in the configuration @xmath182 . indeed , assuming @xmath359
, there exists a path @xmath360 in @xmath287 such that , for all @xmath361 , @xmath362 holds .
this implies that @xmath363 and that all the events @xmath364 occur , whence @xmath365 for @xmath366 , so @xmath367 is realized .
the proof of the other implication is similar .
it follows in particular from ( [ eq : connection0 ] ) that @xmath318 lies in an infinite black cluster in the configuration @xmath27 if and only if @xmath283 lies in an infinite black cluster in the configuration @xmath195 .
lemma [ lemmadac ] presents two scenarios when it is easy to determine ( via a stochastic comparison ) whether the latter event has positive probability .
for example , if we assume that , then there exists @xmath368 and @xmath259 such that for all @xmath261 , @xmath369 . in this case
, the distribution of the restriction of @xmath195 on @xmath370 is stochastically larger than the projection of @xmath323 on @xmath371 .
let us further assume that @xmath331 .
in that case , @xmath351 with positive probability , and @xmath372 for every @xmath373 .
therefore , under the assumptions @xmath374 and @xmath331 , @xmath283 is in an infinite black cluster in @xmath195 ( and , hence , @xmath318 is in an infinite black cluster in @xmath27 ) with positive probability , which can only happen if @xmath375 . on the other hand ,
if @xmath376 , then it is clear that @xmath374 ( whence @xmath337 ) for all @xmath331 , which implies that @xmath377 .
the proof of part a ) is similar .
in this section , we study two particular graph families and obtain examples of non - monotonicity and non - continuity of the critical value function .
the results in section [ tree - like - section ] enable us to prove that ( a small modification of ) the construction considered by hggstrm in the proof of theorem 2.9 in @xcite is a graph whose critical coloring value is non - monotone in the subcritical phase .
define for @xmath378 , @xmath379 to be the complete bipartite graph with the vertex set partitioned into @xmath380 and @xmath381 ( see figure [ graphdk ] ) .
we call @xmath382 and @xmath383 the edges incident to @xmath91 and @xmath384 respectively , and for @xmath385 , @xmath386 the edges incident to @xmath168
. consider @xmath387 the quasi - transitive graph obtained by replacing each edge of the tree @xmath309 by a copy of @xmath388 .
@xmath387 can be seen as the tree - like graph resulting from the construction described at beginning of the section , when we start with the constant sequence @xmath389 .
we will show below that it holds for all @xmath378 that @xmath390 furthermore , there exists @xmath378 and @xmath391 such that @xmath392 proving ( [ fact1])([fact4 ] ) will finish the proof of proposition [ propnonmononicity ] since these inequalities imply that the quasi - transitive graph @xmath387 has a non - monotone critical value function in the subcritical regime . throughout this proof
, we will omit superscripts in the notation when no confusion is possible .
for the proof of ( [ fact1 ] ) , recall that @xmath393 is strictly increasing in @xmath9 , and @xmath394 .
since @xmath395 is the @xmath396-probability of @xmath91 and @xmath384 being in two different bond clusters , we have that @xmath397 from this , we get that @xmath398 , which proves ( [ fact1 ] ) . to get ( [ fact2 ] ) , we need to remember that for fixed @xmath399 , @xmath400 is strictly increasing in @xmath2 , and @xmath401 .
one then easily computes that @xmath402 , whence ( [ fact2 ] ) follows from lemma [ lemmadac ] .
now , define @xmath188 to be the event that at least one edge out of @xmath403 , @xmath404 , @xmath405 and @xmath406 is open .
then @xmath407 which gives that @xmath408 , and implies ( [ fact3 ] ) by [ lemmadac ] . to prove ( [ fact4 ] ) , we consider @xmath409 to be the event that @xmath403 , @xmath404 , @xmath405 and @xmath406 are all closed and that there exists @xmath410 such that @xmath411 and @xmath412 are both open
. one can easily compute that @xmath413 which implies that we can choose @xmath391 ( small ) and @xmath414 ( large ) such that @xmath415 . then , @xmath416 whence inequality ( [ fact4 ] ) follows with these choices from lemma [ lemmadac ] , completing the proof .
it is elementary that @xmath417 , whence @xmath418 follows from lemma [ lemmabond ] , thus @xmath8 is subcritical . since @xmath419 , lemma [ lemmadac ] gives that @xmath420 . on the other hand , @xmath421 for all @xmath55 , which implies by lemma [ lemmadac ] that for @xmath422 , @xmath423 as @xmath424 , so @xmath5 is indeed discontinuous at @xmath425 . in the rest of this section , for vertices @xmath25 and @xmath426 ,
we will write @xmath427 to denote that there exists a path of open edges between @xmath25 and @xmath426 . our proof of theorem [ boundeddegreeconstruction ] will be based on the lemma 2.1 in @xcite , that we rewrite here : lemma [ sharpthlemma ] provides a sequence of bounded degree graphs that exhibit sharp threshold - type behavior at @xmath57 .
we will use such a sequence as a building block to obtain discontinuity at @xmath57 in the critical value function in the dac model .
we first prove the theorem in the case @xmath434 .
consider the graph @xmath435 @xmath436 as in lemma [ sharpthlemma ] .
we construct @xmath292 from @xmath433 by adding to it one extra vertex @xmath293 and one edge @xmath437 .
more precisely @xmath292 has vertex set @xmath438 and edge set @xmath439 .
set @xmath440 and let @xmath0 be the graph @xmath441 defined with the sequence @xmath442 as in section [ tree - like - section ] .
one can easily compute @xmath447 .
since the graph @xmath433 has degree at most @xmath448 and the two vertices @xmath449 are disjoint , the probability @xmath450 can not exceed @xmath451 .
this bound guarantees the existence of @xmath452 independent of @xmath289 such that @xmath453 for all @xmath289 , whence lemma [ lemmabond ] implies that @xmath454 .
for all @xmath31 $ ] , we have @xmath455 which gives that @xmath456 .
writing @xmath457 the positive solution of @xmath458 , we get that @xmath459 for all @xmath123 , which implies by lemma [ lemmadac ] that @xmath445 . on the other hand , @xmath460 , which gives by lemma [ sharpthlemma ] that @xmath461 . writing @xmath462 such that
@xmath463 , it is elementary to check that @xmath464 and that @xmath465 . then , using lemma [ lemmadac ] , we conclude that @xmath466 . | the divide and color model on a graph @xmath0 arises by first deleting each edge of @xmath0 with probability @xmath1 independently of each other , then coloring the resulting connected components ( _ i.e. _ , every vertex in the component ) black or white with respective probabilities @xmath2 and @xmath3 , independently for different components . viewing it as a ( dependent ) site percolation model , one can define the critical point @xmath4 . in this paper
, we mainly study the continuity properties of the function @xmath5 , which is an instance of the question of locality for percolation .
our main result is the fact that in the case @xmath6 , @xmath5 is continuous on the interval @xmath7 ; we also prove continuity at @xmath8 for the more general class of graphs with bounded degree .
we then investigate the sharpness of the bounded degree condition and the monotonicity of @xmath4 as a function of @xmath9 .
= 1 [ [ keywords ] ] keywords : + + + + + + + + + percolation , divide and color model , critical value , locality , stochastic domination .
[ [ ams-2000-subject - classification ] ] ams 2000 subject classification : + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + 60k35 , 82b43 , 82b20 |
a crucial part of the violation program in -factories is the measurement of the angle @xmath11 of the unitary triangle related to the ckm matrix @xcite .
decays of @xmath12 mesons that allows one to constraint the ckm angle @xmath0 , have either small asymmetry ( @xmath13 ) or small branching fractions ( @xmath14 ) .
this makes the violation effect hard to measure . furthermore , due to presence of two hadronic parameters in the observables ( @xmath15 and @xmath16 , the amplitude ratio and the strong phase difference between two amplitudes ) it is difficult to cleanly extract the weak phase information , although approaches based on su(3 ) symmetry exists .
the data used in the presented analyzes were recorded with the detector at the 2 asymmetric - energy storage rings , and consist of 211 fb@xmath17 collected on the @xmath18 resonance ( on - resonance sample ) , and 21 fb@xmath17 collected at an @xmath19 center - of - mass ( cm ) energy approximately 40 @xmath20below the resonance peak ( off - resonance sample ) .
this corresponds to approximately @xmath1 million @xmath2 recorded events .
the detector is described in detail in ref .
the decay modes @xmath22 have been proposed to measure @xmath8 @xcite . in the standard model the decays @xmath23 and @xmath24 proceed through the @xmath25 and @xmath26 amplitudes @xmath27 and @xmath28 , respectively .
the relative weak phase between these two amplitudes is @xmath29 .
when combined with @xmath30 mixing , this yields a weak phase difference of @xmath0 between the interfering amplitudes .
the decay rate distribution for @xmath31 is @xmath32 , \label{eq : pure - dt - pdf - b}\end{aligned}\ ] ] where @xmath33 is the @xmath34 lifetime averaged over the two mass eigenstates , @xmath35 is the @xmath36 mixing frequency , and @xmath37 is the difference between the time of the @xmath38 ( @xmath39 ) decay and the decay of the other @xmath3 ( @xmath40 ) in the event . the upper ( lower ) sign in eq .
[ eq : pure - dt - pdf - b ] indicates the flavor of the @xmath40 as a @xmath34 ( @xmath41 ) , while @xmath42 ( @xmath43 ) and @xmath44 ( @xmath45 ) for the @xmath39 final state @xmath46 ( @xmath47 ) .
the parameters @xmath48 and @xmath49 are given by @xmath50 here @xmath16 is the strong phase difference between @xmath27 and @xmath28 and @xmath51 .
since @xmath27 is doubly ckm - suppressed with respect to @xmath28 , one expects @xmath15 to be small of order 2% . due to the small value of @xmath15 ,
large data samples are required for a statistically significant measurement of @xmath52 . since the expected asymmetry in the selected @xmath3 decays is small , this measurement is sensitive to the interference between the @xmath53 and @xmath54 amplitudes in the decay of @xmath40 . to account for this `` tagside interference '' ,
we use a parametrization which is described in ref .
the @xmath49 coefficient are replaced with three others @xmath55 for each tagging category , independent of the decay mode @xmath56 , the tagside interference is parametrized in terms of the effective parameters @xmath57 and @xmath58 .
one notes , @xmath59 for the lepton tagging category .
two different analysis techniques , full reconstruction @xcite and partial reconstruction @xcite were used for the @xmath8 measurement with @xmath22 .
the full reconstruction technique is used to measure the asymmetry in @xmath22 and @xmath60 decays @xcite . from a time - dependent maximum likelihood
fit the following parameters related to the violation angle @xmath61 are obtained : @xmath62 where the first error is statistical and the second is systematic .
the systematic error for @xmath60 includes the maximum bias of asymmetry parameters due to possible dependence of @xmath15 on the @xmath63 invariant mass . for the measurement of @xmath64 parameter
only the lepton - tagged events are used due to a presence of tag - side violation effect @xcite . raw asymmetry for ( a ) lepton - tagged and ( b ) kaon - tagged events of @xmath65 decay mode using the method of the partial reconstruction .
the curves represent the projections of the pdf for the raw asymmetry.,title="fig : " ] raw asymmetry for ( a ) lepton - tagged and ( b ) kaon - tagged events of @xmath65 decay mode using the method of the partial reconstruction .
the curves represent the projections of the pdf for the raw asymmetry.,title="fig : " ] in the partial reconstruction of a @xmath65 candidate , only the hard ( high - momentum ) pion track @xmath66 from the @xmath3 decay and the soft ( low - momentum ) pion track @xmath67 from the decay @xmath68 are used . applying kinematic constraints consistent with the signal decay mode ,
the four - momentum of the non - reconstructed , `` missing '' @xmath69 is calculated .
signal events are peaked in the @xmath70 distribution at the nominal @xmath71 mass .
this method eliminates the efficiency loss associated with the neutral @xmath69 meson reconstruction .
the asymmetry independent on the assumption on @xmath72 measured with this technique is @xcite @xmath73 where the first error is statistical and the second is systematic .
this measurement deviates from zero by 2.0 standard deviations .
figure [ fig : sin2bg_asym ] shows the raw , time - dependent asymmetry @xmath74 in the absence of background and with high statistics , perfect tagging , and perfect @xmath37 measurement , @xmath75 would be a sinusoidal oscillation with amplitude @xmath76 .
two methods for interpreting these results in terms of constraints on @xmath77 are used .
both methods involve minimizing a @xmath78 function that is symmetric under the exchange @xmath79 , and applying the method of ref .
@xcite . in the first interpretation method ,
no assumption regarding the value of @xmath80 is made .
the resulting 95% lower limit for the mode @xmath81 is shown as a function of @xmath80 in figure [ fig : limit - vs - r ] .
the second interpretation assumes that @xmath82 can be estimated from the cabibbo angle , the ratio of branching fractions @xmath83 , and the ratio of decay constants @xmath84 .
the confidence level as a function of @xmath77 is shown in figure [ fig : s2bgcl ] .
this method yields the lower limits @xmath6 at @xmath85 @xmath86 c.l . lower limit on @xmath77 a t 90% cl as a function of @xmath87 , for @xmath88 .
, scaledwidth=48.0% ] the shaded region denotes the allowed range of @xmath77 for each confidence level .
the horizontal lines show , from top to bottom , the 68% and 90% cl .
, scaledwidth=48.0% ]
the decay modes @xmath89 have been proposed for determination of @xmath8 from measurement of time - dependent asymmetries @xcite . in the standard model
the decays of and mesons into final state @xmath90 proceed through the @xmath91 and @xmath92 amplitudes , respectively . due to relatively large asymmetry ( @xmath93 )
these decay channels look very attractive for such a measurement .
since the parameter @xmath94 can be measured with sufficient data sample by fitting the @xmath48 coefficient in time distributions , the measured asymmetry can be interpreted in terms of @xmath8 without additional assumptions
. however , the branching fractions of such decays are relatively small ( @xmath95 ) . that is way the large data sample is still required . from the measured signal yields @xcite
, we find @xmath96 where the uncertainties are statistical and systematic , respectively .
figure [ fig : dk ] shows the distributions of candidates with @xmath97 @xmath98for the sums of the reconstructed decay modes .
distribution of for a ) @xmath99 , b ) @xmath100 , c ) @xmath101 , and d ) @xmath102 candidates with @xmath103 .
the points are the data , the solid curve is the projection of the likelihood fit , and the dashed curve represents the background component .
, title="fig : " ] ( -102,74)a ) distribution of for a ) @xmath99 , b ) @xmath100 , c ) @xmath101 , and d ) @xmath102 candidates with @xmath103 .
the points are the data , the solid curve is the projection of the likelihood fit , and the dashed curve represents the background component .
, title="fig : " ] ( -102,74)b ) + distribution of for a ) @xmath99 , b ) @xmath100 , c ) @xmath101 , and d ) @xmath102 candidates with @xmath103 .
the points are the data , the solid curve is the projection of the likelihood fit , and the dashed curve represents the background component .
, title="fig : " ] ( -102,74)c ) distribution of for a ) @xmath99 , b ) @xmath100 , c ) @xmath101 , and d ) @xmath102 candidates with @xmath103 .
the points are the data , the solid curve is the projection of the likelihood fit , and the dashed curve represents the background component .
, title="fig : " ] ( -102,74)d ) the decay dynamics can modify the expectation for the ratio @xmath94 .
the magnitude of this ratio can be probed by measuring the rate for the decays @xmath104 and @xmath105 using the self - tagging decay @xmath106 .
the @xmath107 and @xmath108 decays are distinguished by the correlation between the charges of the kaons produced in the decays of the neutral @xmath69 and the .
this charge correlation in the final state is diluted by the presence of the doubly - cabibbo - suppressed decays @xmath109 , and @xmath110 .
the ratio @xmath94 is related to the experimental observable @xmath111 defined for the @xmath112 decay as @xmath113 where @xmath114 and @xmath115 and @xmath116 are strong phase differences between the two @xmath3 and @xmath69 decay amplitudes , respectively . from the measured branching fractions ( eq . [ eq : brdk ] ) , values of @xmath117 @xcite and eq .
[ eq : r_ads ] , one obtains @xmath118 at the 90% c.l . to conclude
, the present signal yields combined with this limit on @xmath15 suggest that a substantially larger data sample is needed for a competitive time - dependent measurement of @xmath8 in @xmath119 decays .
recently it was proposed to consider the @xmath10 decays for measurement of @xmath8 @xcite .
the decay amplitudes of mesons to light scalar or tensor mesons such as @xmath120 or @xmath121 , emitted from a weak current , are significantly suppressed due to the small decay constants @xmath122 .
thus , the absolute value of the ckm - suppressed and favored amplitudes become comparable . as a result
, the asymmetry in such decays is expected to be large . however , the theoretical predictions of the branching fractions for @xmath10 is expected of the order of @xmath123 @xcite .
the main uncertainty in the branching fractions of these decay modes is due to unknown @xmath124 transition form factors .
one way to verify the expectations and test a validity of the factorization approach is to measure the branching fractions for the more abundant decay modes @xmath125 . using a sample of about 230 million @xmath126 no evidence for these decays
were observed @xcite .
this allowed one to set upper limits at 90% c.l . on the branching fractions to be @xmath127 figure [ fig : dsa0 ]
shows the distributions for the reconstructed candidates @xmath128 , @xmath129 , @xmath130 and @xmath131 . for each decay mode
, an unbinned maximum - likelihood fit is performed using the candidates from the three decay modes .
distributions of for a ) @xmath132 , b ) @xmath133 , c ) @xmath134 , d ) @xmath135 candidates overlaid with the projection of the maximum likelihood fit .
contributions from the three @xmath136 decay modes are shown with different hatching styles : @xmath137 is cross hatched , @xmath138 is hatched , and @xmath139 is white.,title="fig:",scaledwidth=49.0% ] ( -230,150)a ) ( -230,75)c ) ( -115,150)b ) ( -115,75)d ) the upper limit value for @xmath140 is lower than the theoretical expectation , which might indicate the need to revisit the @xmath141 transition form factor estimate .
it might also imply the limited applicability of the factorization approach for this decay mode .
the measured upper limits suggest that the branching ratios of @xmath142 are too small for -asymmetry measurements given the present statistics of the @xmath3-factories .
the measurement of @xmath8 in @xmath142 decays is an interesting program for the future experiments such as super - factories .
the substantial constraint on the ckm angles @xmath0 comes from the measurements of time - dependent asymmetry in the @xmath4 and @xmath5 decays .
the experiment has used two techniques such as full and partial reconstruction to increase the signal yields in the @xmath143 channel .
the combined and _ belle _ results @xcite for violation in the most precisely measured decay channel @xmath143 is @xmath144 this measurement performed at the level of one per cent deviates from zero by 3.4 standard deviations .
future updates are therefore of a great interest .
we interpret the result in terms of @xmath8 and find @xmath145 @xmath146 at @xmath85 @xmath86 c.l . using a frequentist method .
the experiment has measured the branching fractions of @xmath9 and set up the limit on @xmath10 decays .
the present signal yields and established limits suggest that a substantially larger data sample is needed for a competitive time - dependent measurement of @xmath8 with these decay channels .
sachs , enrico fermi institute report , efi-85 - 22 ( 1985 ) ( unpublished ) ; i. dunietz and r.g .
sachs , phys .
* d37 * , 3186 ( 1988 ) [ e : phys .
* d39 * , 3515 ( 1989 ) ] ; i. dunietz , phys . lett . * b427 * , 179 ( 1998 ) ; p.f .
harrison and h.r .
quinn , ed .
, `` the physics book '' , slac - r-504 ( 1998 ) , chap . 7.6 . m. gronau and d. london , phys . lett . b * 253 * , 483 ( 1991 ) ; d. atwood , i. dunietz , and a. soni , phys . rev .
lett . * 78 * , 3257 ( 1997 ) ; b. kayser and d. london , phys .
d * 61 * , 116013 ( 2000 ) ; a.i.sandra , hep - ph/0108031 . | the recent experimental results on violation related to the angles of the cabibbo - kobayashi - maskawa ( ckm ) unitarity triangle @xmath0 are summarized in these proceedings .
these results are obtained with approximately @xmath1 million @xmath2 events collected with the detector at the pep - ii asymmetric - energy @xmath3-factory at slac . using the measurements on time - dependent asymmetries in @xmath4 and @xmath5 decays and theoretical assumptions
, one finds @xmath6 at @xmath7 confidence level .
the perspectives of @xmath8 measurement with @xmath9 and @xmath10 decay channels are also discussed . |
our transmission calculations for the satellite links take into account diffraction , pointing error and atmospheric transmittance @xcite .
diffraction is modeled using the rayleigh - sommerfeld formula @xcite .
pointing error , which arises due to jitter in the telescope and imprecision in the tracking system , will cause the beam to wander with a gaussian distribution .
we model the average combined effect of diffraction and pointing error by taking a two - dimensional convolution @xcite of the diffraction profile with the gaussian distribution of the pointing error .
finally the atmospheric transmittance , characterizing the probability of light to be scattered or absorbed by molecules in the atmosphere , is modeled using the modtran 5 software package .
the transmission calculation is performed for every point ( at 10 s intervals ) where the satellite is in view of the ground station . in fig .
si 1 we show results for the double - link loss for the leo and higher altitude satellites used in fig . 2 in the main text .
this allows us to find the average transmission and the flyby time for each configuration .
the period of the satellite is given by @xmath27 , where @xmath28 and @xmath29 is the rotation period of the earth .
@xmath30 and @xmath31 are the earth s radius and mass , @xmath1 is the satellite s height and @xmath32 is the gravitational constant .
here we assumed a simple circular orbit on the equatorial plane with no inclination . for all orbits except geo ( which by definition must move in the same direction as the earth s rotation ) ,
the overall performance is independent of the direction of the satellite .
if the direction is the same as the earth s rotation , the orbit will have longer but less frequent passes compared to an orbit with the opposite direction . in both cases
the average loss and average contact time per day remains the same . for our analysis
we chose the direction to be opposite of the rotation of the earth , allowing more frequent key exchange at the cost of key length .
we studied the dependence of the entanglement distribution rate over 20,000 km with respect to the efficiency of the different elements of our proposed architecture . as it can be seen in fig . 3 in the main text and in fig .
si 2 , the entanglement distribution rate has a similar dependence on source , qnd detection and memory write efficiency .
this is because these elements contribute similarly to the entanglement generation rate .
detector and memory read efficiency have the same impact on the rate as they have the same contribution to the entanglement swapping success probability .
a detailed nighttime background calculation @xcite showed a background on the order of 10@xmath33/s with a 1 m receiver diameter , 1 nm filter and a 50 @xmath4rad field of view . for our quantum repeater scenario
we have assumed a bandwidth of 10 mhz , allowing us to use a filter of that width ( approximately 10@xmath34 nm ) .
in addition the field of view of 50 @xmath4rad is a conservative estimate that can be reduced , reducing the background counts further by approximately 2 orders of magnitude .
this leads to a nighttime background count on the order of 10@xmath35/s , which has negligible impact .
the daylight background contribution has been studied in ref .
@xcite . they found the expected background counts to be on the order of 10@xmath36/s / m@xmath33/@xmath4m / str corresponding to around 100/s for a 1 m receiver diameter , 10 mhz filter and 10 @xmath4rad field of view .
the primary errors due to background light are caused by a noise photon arriving at one side in coincidence with a single photon from the source at the other side of an elementary link , which is wrongly interpreted as an entangled photon pair reaching the ground stations .
this coincidence probability is given by @xmath37 , where @xmath38 and @xmath39 . here , @xmath40(@xmath19 ) is the noise ( source ) rate , @xmath41 is the coincidence time window ( which can be set to the inverse of the source rate ) and @xmath42 is the maximum single - link transmission from the satellite to the ground . comparing this coincidence probability ( for daytime noise ) with the probability of receiving entangled photon pairs for different leo satellites shows that only about 1% error occurs for our quantum repeater scheme .
however , for the direct transmission scenario based on using a geostationary satellite the noise will become dominant for distances beyond 12,000 km .
using adaptive optics one can in principle achieve up - link losses approaching those of down - links . therefore , a more conventional quantum repeater scheme with up - links ( ground to satellite links ) will become feasible @xcite .
in such a scheme , pair sources and quantum memories remain on the ground and beam splitters and detectors for bell state measurement are placed on the satellite . in this case the scheme does not require qnd detectors . as the pair source is adjacent to the quantum memory at each node , photons from the pair source can be stored into quantum memories on the ground with ideally no loss .
however , this means that in order to operate the source at its maximum possible rate this scheme would require approximately two orders of magnitude larger memory mode capacity .
bourgoin , e. meyer - scott , b.l .
higgins , b. helou , c. erven , h. hbel1 , b. kumar , d. hudson , i. dsouza , r. girard , r. laflamme , and t. jennewein , new j. phys . *
15 * , 023006 ( 2013 ) .
goodman , introduction to fourier optics 2nd edn , greenwood village , co : roberts & company ( 1996 ) . n. baddour , j. opt . soc . am .
a * 26 * , 1767 ( 2009 ) .
www.ontar.com w.t .
buttler , r.j .
hughes , p.g .
kwiat , s.k .
lamoreaux , g.g .
luther , g.l .
morgan , j.e .
nordholt , c.g .
peterson , and c.m .
simmons , phys .
lett . * 81 * , 3283 ( 1998 ) .
d. rideout _
_ , class .
quantum grav .
* 29 * , 224011 ( 2012 ) . | we study entanglement creation over global distances based on a quantum repeater architecture that uses low - earth orbit satellites equipped with entangled photon sources , as well as ground stations equipped with quantum non - demolition detectors and quantum memories .
we show that this approach allows entanglement creation at viable rates over distances that are inaccessible via direct transmission through optical fibers or even from very distant satellites . over the last few decades the distribution of quantum entanglement has progressed from tabletop experiments to distances of over one hundred kilometers @xcite
. will it be possible to create entanglement over global distances ?
this is interesting from a fundamental point of view , but also from the perspective of trying to create a global `` quantum internet '' @xcite . in the context of quantum cryptography
, it would enable secure global communication without having to rely on any trusted nodes @xcite , as entanglement is the foundation for device - independent quantum key distribution @xcite .
it would also be useful for global clock networks @xcite and for very long baseline telescopes @xcite .
modern classical telecommunication relies on optical fibers .
unfortunately the direct transmission of photons through fibers is not practical for quantum communication over global distances because losses are too high .
the best available fibers have a loss of 0.15 db / km at the optimal wavelength .
this means , for example , that the time to distribute one entangled photon pair over 2000 km with a 1 ghz source exceeds the age of the universe .
two alternative approaches to try to overcome this problem are currently being pursued in parallel , namely fiber - based quantum repeaters and direct satellite links .
conventional quantum repeaters rely on first creating and storing entanglement for elementary links , then extending the distance of entanglement by entanglement swapping @xcite . based on the experimental and theoretical progress in this area over the last few years ,
it is plausible that this approach will make it possible to extend the distance of entanglement distribution significantly beyond what is possible with direct transmission through optical fibers @xcite .
however , truly global distances are still very difficult to envision for repeaters based on fiber links .
this is true also for related approaches based on quantum error correction @xcite , which tend to require repeater stations that are only a few kilometers apart .
the use of satellite links for quantum communication is also being pursued very actively .
there has been a lot of progress in terms of feasibility studies @xcite .
the launch of the first satellite carrying an entangled pair source has been announced for 2016 @xcite .
the advantage of quantum communication via satellites is that transmission loss is dominated by diffraction rather than absorption and thus scales much more favorably with distance .
for example , consider a pair source on a satellite at a height of 1000 km .
for realistic assumptions
( such as telescope size , see below ) , the combined transmission loss for the photon pair for a 2,000 km ground station distance is only of order 40 db .
this should be contrasted with 300 db for a fiber link of the same length .
however , global distances are still challenging even for satellite links .
direct transmission from low - earth orbit ( leo ) satellites , i.e. those below the van allen radiation belt , or up to about 2000 km in height , no longer works . even before the earth gets in the way
, the loss becomes forbidding for very grazing incidence due to long propagation distance in air .
one possible solution is to use satellites that are much further away , but this comes at significant cost , as satellites have to be much more robust to shield them from radiation .
moreover the greatest ground distances , approaching 20,000 km ( i.e. half the earth s circumference ) , are out of range even for very distant satellites . here
we propose to combine the two approaches discussed above .
we study quantum repeaters based on leo satellite links , as illustrated in figure 1 .
the satellites just need to be equipped with entangled pair sources , while the more complex components , such as quantum memories and quantum non - demolition ( qnd ) detectors , are on the ground and can be further developed even after the satellites are launched .
an important difference between satellite and fiber - based links is that the satellite - based links are active only during each time period when the satellite is visible from both ground stations ( the `` flyby time '' @xmath0 ) . for currently realistic quantum memory lifetimes all satellite links in figure 1 have to be active simultaneously , which implies that our architecture requires a number of satellites equal to the number of links .
however , our results show that four to eight links are sufficient to span global distances .
figure 2 compares the expected entanglement distribution rates per day for repeater architectures with leo satellites to those achievable by direct transmission from more distant satellites .
it is important to make the comparison on a per day basis since the flyby times and periods are different for satellites at different heights .
our results suggest that approach based on repeaters with leo satellite links is attractive for all but the shortest distances and is the only way to create entanglement for the longest distances .
we now describe the assumptions and requirements underlying these results in some detail .
one key ingredient for our analysis is the calculation of the probability for a pair of photons that are emitted from a satellite at height @xmath1 to be successfully transmitted to the ground .
our approach , which is based on ref .
@xcite , takes into account diffraction , pointing error and atmospheric transmittance . in figure 2
we assume a satellite transmitter size of 50 cm and ground telescope size of 1 m. for the quantum repeater scenarios we assumed a pair source that emits photons at 580 nm , which is motivated by our choice of quantum memory material ( eu - doped yttrium orthosilicate , see below ) . for direct transmission
we assumed a wavelength of 670 nm ( 470 nm ) for @xmath2 km ( @xmath3 km and geostationary ) , which results in optimal transmittance @xcite .
we include a satellite pointing error of 0.5 @xmath4rad and assume ground stations at rural atmosphere at sea level ; see the supplementary information @xcite for more details .
we assume that frequency shifts due to relativistic and gravitational effects @xcite are compensated ( e.g. by acousto - optic modulators on the ground ) .
timing jitter due to turbulence in the atmosphere is negligible for the relatively long pulses that we are considering @xcite . for the repeater scenarios ,
we have assumed a pair source with a repetition rate of 10 mhz .
this value is motivated primarily by the expected memory bandwidth for our choice of material , see below .
in contrast , we assume a 1 ghz repetition rate for direct transmission . in each case
the source could e.g. be a deterministic pair source based on a quantum dot in microcavity @xcite .
however , simpler implementations are possible based on parametric down - conversion sources with a small pair creation probability per pulse ( below 0.01 ) @xcite , in order to avoid errors due to multi - pair emissions . if one aims to achieve the same effective rate in this way , the underlying repetition rate ( and hence memory bandwidth , in the repeater scenario ) has to be increased correspondingly .
memory bandwidths up to 1 ghz have already been achieved in rare - earth doped materials ( e.g. in tm : linbo @xcite ) , but not yet in combination with long storage times .
we have not assumed any frequency multiplexing , neither for repeaters nor for direct transmission .
this could be used to boost rates in both scenarios , at the expense of more complex sources on the satellites .
the rates for the repeaters are calculated as in ref .
@xcite , assuming a `` nested '' approach .
that is , entanglement is first created and stored at the level of the elementary links .
then links are connected in a hierarchical fashion , forming links of length two , four etc .
for convenience let us define the average probability of a pair reaching the ground stations during one flyby of the satellite as @xmath5 , where @xmath6 is the time - dependent combined two - photon transmission , and @xmath7 is the flyby time of the satellite @xcite . the probability of successfully creating , transmitting and storing an entangled pair over one elementary link is @xmath8 , where @xmath9 , @xmath10 and @xmath11 are source , qnd detector and memory write efficiencies .
entanglement swapping relies on bell - state measurements ( bsm ) . in our scheme ,
a successful bsm requires successful readout of two photons from neighboring quantum memories with the efficiency of @xmath12 and two single - photon detections with @xmath13 efficiency . here
@xmath14 and @xmath15 are memory readout and detector efficiencies .
this gives the entanglement swapping efficiency of @xmath16 , where the factor of 1/2 is due to limited success probability of the bsm using linear optics with ancillary vacuum modes @xcite .
higher success probabilities are possible in principle using ancillary photons @xcite . for a repeater composed of @xmath17 links ,
the number of entangled pairs created during one flyby is given by @xmath18 , where @xmath19 is the source rate .
the factors of 2/3 take into account the fact that entanglement has to be created in two neighboring links before entanglement swapping can proceed @xcite . in fig .
[ rates ] , we assumed that the different elements including source , qnd detector , quantum memory ( write and read ) and detectors have efficiencies of 0.9 .
the effect of inefficiencies in these elements on the total entanglement distribution rate is shown in figure 3 and in the supplementary information @xcite .
significantly higher efficiencies have already been achieved for single - photon detectors @xcite .
we now discuss the other elements in more detail .
quantum memories for photons have been implemented in a range of physical systems @xcite .
memories based on rare - earth ion doped crystals @xcite are particularly attractive for our purpose because of their potential for highly multi - mode storage , e.g. using the atomic frequency comb ( afc ) protocol @xcite .
this is important because the quantum memories in each ground station will be exposed to a large number of photons , @xmath20 , before receiving the classical signals from the other end of each link that make it possible to decide which photons are part of an entangled pair and should thus be kept for entanglement swapping . here
@xmath21 denotes the maximum value of the single - photon transmission during one flyby . for the quantum repeater scenarios in fig .
[ rates ] , multimode storage of @xmath22 photons is required according to the above formula ( depending on satellite height and number of links ) .
a single afc type quantum memory based on eu - doped yttrium ortho silicate ( yso ) should be able to store @xmath23 photons in distinct temporal modes @xcite .
having several waveguides or using multiple locations on the same crystal makes the storage of a total of @xmath24 photons in distinct modes in a single crystal plausible .
our protocol also requires storage times of the order of the total communication time @xmath25 , where @xmath26 is the total distance , which corresponds to 67 ms for 20,000 km .
such long storage times should be achievable by transferring the optical memory excitations to ground spin states @xcite .
note that storage times up to one minute have already been achieved in rare - earth doped crystals in the classical regime @xcite .
the requirement of transferring the excitation to the ground state limits the repetition rate of the photon source as the bandwidth of the photons must be smaller than energy spacing between the ground spin states .
the 10 mhz bandwidth assumed in figure 2 is compatible with the ground level separations of eu : yso , which are of order 100 mhz @xcite .
high memory efficiencies , which are also important for our purpose ( see figure 3 ) , can be achieved in rare - earth doped crystals with the help of optical cavities @xcite .
our scheme also requires qnd detection of the photonic qubits .
qnd measurement of photons has recently been demonstrated using a single atom in a cavity @xcite . the cross - kerr effect induced by the ac - stark shift in atomic ensembles
also provides the possibility to realize qnd measurement of photons . in ref .
@xcite , 0.5 mrad cross - phase shift per photon has been shown using a hot atomic vapor inside a hollow - core photonic crystal fiber , which should already allow a qnd measurement of the photon number @xcite . here
we also require the qnd measurement to be insensitive to the photonic qubit state .
for example , if photon pairs with polarization entanglement are to be detected , the probe field must interact with both polarization modes .
a simpler implementation of the qnd detection of photonic qubits is possible for time - bin qubits based on the ac - stark shift in combination with quantum storage because the phase shift imparted to the stored probe beam is not sensitive to the precise timing of the signal photon propagating through the ensemble @xcite .
this approach should also make it possible to integrate the qnd detector with the quantum memory , e.g. a rare - earth doped waveguide @xcite .
another possibility is to use a heralded qubit amplifier based on linear optics and a deterministic pair source @xcite .
this achieves a qnd detection efficiency of up to 0.5 .
compared to the assumed qnd detection efficiency of 0.9 in fig .
[ rates ] , this would reduce the entanglement distribution rate by approximately a factor of 5 ( see fig . [ inefficiency ] ) .
we discuss the effect of background photons in the supplementary information @xcite . due to the narrow bandwidth of the photons used , which allows the use of correspondingly narrow filters ,
the background noise is completely negligible at nighttime . for the bright daytime sky ,
the related errors are below the 1% level for the repeater scenarios in figure 2 .
in contrast , for direct transmission from a geo satellite the background noise becomes dominant for distances beyond 12,000 km .
due to turbulence , which disturbs the wavefront of the photons , the use of adaptive optics is likely to be required in order to couple the photons into single - mode waveguides for qnd detection and quantum storage .
this also makes it possible to consider alternative repeater scenarios based on uplinks @xcite , which are significantly more lossy than downlinks without adaptive optics .
see the supplementary information @xcite for more discussion of this possibility .
we only performed a simple rate calculation for the proposed repeater architecture .
a more sophisticated analysis would characterize the fidelity of the distributed quantum state and extract a key rate for quantum key distribution applications @xcite .
however , assuming low noise levels in all components , and given the fact that we only consider small numbers of repeater links , the present estimates should give a reasonably accurate picture of achievable key rates .
we have argued that quantum repeaters based on leo satellite links are a viable approach to global quantum communication .
our proposed scheme relies on realistic advances in quantum memories and quantum non - demolition measurements and only requires a moderate number of satellites equipped with entangled photon pair sources .
ultimately global quantum repeater networks will likely combine satellite links for very long distances with fiber links for short and intermediate distances . _
acknowledgments . _
the authors would like to thank w. tittel for valuable discussions .
this work was supported by nserc , aitf , darpa , ontario ministry of research and innovation , era , industry canada , cifar , cfi , and feddev ontario .
11 s.j . freedman and j.f .
clauser , phys .
rev .
lett .
* 28 * , 938 ( 1972 ) ; a. aspect , j. dalibard , and g. roger , phys .
rev .
lett . * 49 * , 1804 ( 1982 ) ; g. weihs , t. jennewein , c. simon , h. weinfurter , and a. zeilinger , phys .
rev .
lett . * 81 * , 5039 ( 1998 ) ; w. tittel , j. brendel , h. zbinden , and n. gisin , phys .
rev .
lett . * 81 * , 3563 ( 1998 ) ; r. ursin _ et al . _ , nat . phys . * 3 * , 481 ( 2007 ) ; x .- s .
ma _
et al . _ ,
nature * 489 * , 269 ( 2012 ) .
h.j .
kimble , nature * 453 * , 1023 ( 2008 ) .
t. jennewein and b. higgins , physics world * 26*(3 ) , 5256 ( march 2013 ) .
a.k .
ekert , phys .
rev .
lett . * 67 * , 661 ( 1991 ) ; a. acin , n. brunner , n. gisin , s. massar , s. pironio , and v. scarani , phys .
rev .
lett .
* 98 * , 230501 ( 2007 ) .
p. kmr , e. m. kessler , m. bishof , l. jiang , a.s .
srensen , j. ye and m.d . lukin , nat . phys . * 10 * , 582 ( 2014 ) .
d. gottesman , t. jennewein , and s. croke , phys .
rev .
lett . * 109 * , 070503 ( 2012 ) .
h .- j .
briegel , w. dr , j.i .
cirac , and p. zoller , phys .
rev .
lett . * 81 * , 5932 ( 1998 ) ; l .- m .
duan , m.d .
lukin , j.i .
cirac , and p. zoller , nature * 414 * , 413 ( 2001 ) .
n. sangouard , c. simon , h. de riedmatten , and n. gisin , rev .
mod .
phys . * 83 * , 33 ( 2011 ) f. bussieres , n. sangouard , m. afzelius , h. de riedmatten , c. simon , and w. tittel , j. mod . opt . *
60 * , 1519 ( 2013 ) ; n. sinclair , e. saglamyurek , h. mallahzadeh , j.a .
slater , m. george , r. ricken , m.p .
hedges , d. oblak , c. simon , w. sohler , and w. tittel , phys .
rev .
lett . * 113 * , 053603 ( 2014 ) .
s. guha , h. krovi , c.a .
fuchs , z. dutton , j.a .
slater , c. simon , and w. tittel , in preparation .
w.j .
munro , a.m. stephens , s.j .
devitt , k.a .
harrison , and k. nemoto , nat .
photon .
* 6 * , 777 ( 2012 ) ; k. azuma , k. tamaki , h .- k .
lo , arxiv:1309.7207 ( 2013 ) ; s. muralidharan , j. kim , n. ltkenhaus , m.d .
lukin , and l. jiang , phys .
rev .
lett . * 112 * , 250501 ( 2014 ) .
w.t .
buttler , r.j .
hughes , p.g .
kwiat , s.k .
lamoreaux , g.g .
luther , g.l .
morgan , j.e .
nordholt , c.g .
peterson , and c.m .
simmons , phys .
rev .
lett .
* 81 * , 3283 ( 1998 ) .
c. kurtsiefer , p. zarda , m. halder , h. weinfurter , p.m. gorman , p.r .
tapster , and j.g .
rarity , nature ( london ) * 419 * , 450 ( 2002 ) .
j.g . rarity , p.r .
tapster , p.m. gorman , and p. knight , new j. phys . * 4 * , 82 ( 2002 ) . c. bonato , a. tomaello , v. da deppo , g. naletto , and p. villoresi , new j. phys . *
11 * , 045017 ( 2009 ) .
j .- y .
wang _
et al . _
nat .
photonics * 7*,387 ( 2013 ) . j .-
p .
bourgoin , e. meyer - scott , b.l .
higgins , b. helou , c. erven , h. hbel , b. kumar , d. hudson , i. dsouza , r. girard , r. laflamme , and t. jennewein , new j. phys .
* 15 * , 023006 ( 2013 ) .
t. zhongkan , r. chandrasekara , y.y .
sean , c. cheng , c. wildfeuer , and a. ling , arxiv:1404.3971 ( 2014 ) .
g. vallone , d. bacco , d. dequal , s. gaiarin , v. luceri , g. bianco , p. villoresi , arxiv:1406.4051 ( 2014 ) .
h. xin , science * 332 * , 904 ( 2011 ) .
see the supplementary information for more details on the transmission calculation , the satellite orbits , the effects of inefficiencies on entanglement distribution rates , the errors due to background light , and a potential repeater architecture based on uplinks rather than downlinks .
d.e .
bruschi , t. ralph , i. fuentes , t. jennewein , and m. razavi , arxiv:1309.3088 ( 2013 ) .
l. kral , i. prochazka , and k. hamal , opt .
lett . *
30 * , 1767 ( 2005 ) .
a. dousse , j. suffczynski , a. beveratos , o. krebs , a. lemaitre , i. sagnes , j. bloch , p. voisin , and p. senellart , nature * 466 * , 217 ( 2010 ) .
j.a .
slater , j .- s .
corbeil , s. virally , f. bussieres , a. kudlinski , g. bouwmans , s. lacroix , n. godbout , and w. tittel , opt .
lett . * 35 * , 499 ( 2010 ) .
e. saglamyurek , n. sinclair , j. jin , j.a .
slater , d. oblak , flix bussires , m. george , r. ricken , w. sohler and w. tittel , nature * 469 * , 512 ( 2011 ) .
j calsamiglia and n. ltkenhaus , appl .
phys .
b * 72 * , 67 ( 2001 ) .
w.p .
grice , phys .
rev .
a. * 84 * , 042331 ( 2011 ) . f. ewert and p. van loock , arxiv:1403.4841 ( 2014 ) .
a.e .
lita , a.j .
miller , and s.w .
nam , opt .
express * 16 * , 3032 ( 2008 ) .
k. hammerer , a.s .
sorensen , and e.s .
polzik , rev .
mod .
phys . * 82 * , 1041 ( 2010 ) ; a.i .
lvovsky , b.c .
sanders , and w. tittel , nat .
phot . * 3 * , 706 ( 2009 ) ; c. simon _ et al .
_ , eur .
phys .
j. d * 58 * , 1 ( 2010 ) .
h. de riedmatten , m. afzelius , m.u .
staudt , c. simon and n. gisin , nature * 456 * , 773 ( 2008 ) ; m.p .
hedges , j.j .
longdell , y. li and m.j .
sellars , nature * 465 * , 1052 ( 2010 ) ; w. tittel , m. afzelius , t. chaneliere , r.l . cone , s. kroll , s.a .
moiseev , and m. sellars , las .
phot . rev . * 4 * , 244 ( 2010 ) .
m. afzelius , c. simon , h. de riedmatten , and n. gisin , phys .
rev .
a * 79 * , 052329 ( 2009 ) .
m. afzelius , i. usmani , a. amari , b. lauritzen , a. walther , c. simon , n. sangouard , j. minar , h. de riedmatten , n. gisin , and s. krll , phys .
rev .
lett . * 104 * , 040503 ( 2010 ) .
p. jobez , i. usmani , n. timoney , c. laplane , n. gisin , and m. afzelius , arxiv:1404.3489 ( 2014 ) . g. heinze , c. hubrich , and t. halfmann , phys . rev
. lett . * 111 * , 033601 ( 2013 ) .
m. afzelius and c. simon , phys .
rev .
a * 82 * , 022310 ( 2010 ) ; m. sabooni , q. li , s. krll , and l. rippe , phys .
rev .
lett . * 110 * , 133604 ( 2013 ) .
a. reiserer , n. kalb , g. rempe and s. ritter , nature * 508 * , 237 ( 2014 ) .
v. venkataraman , k. saha , a.l .
gaeta , nat . phot .
* 7 * , 138 ( 2013 ) .
n. imoto , h.a .
haus , and y. yamamoto , phys .
rev .
a * 32 * , 2287 ( 1985 ) .
k. heshami _
et al . _ , in preparation .
m. curty and t. moroder , phys .
rev .
a * 84 * , 010304 ( 2011 ) ; e. meyer - scott , m. bula , k. bartkiewicz , a. ernoch , j. soubusta , t. jennewein , and k. lemr , phys .
rev .
a * 88 * , 012327 ( 2013 ) .
d. rideout _
et al . _ ,
class .
quantum grav .
* 29 * , 224011 ( 2012 ) . * supplemental material for global quantum communication with satellites and quantum repeaters " |
exclusive @xmath3 meson decays , especially @xmath4 modes , have aroused more and more interest both theoretically and experimentally . since the first observed charmless @xmath4 mode , the @xmath5 decay@xcite , many @xmath4decay channels have been studied in pqcd approach , such as @xmath6@xcite , @xmath7@xcite , @xmath8@xcite , @xmath9@xcite , and @xmath10@xcite .
it offers an excellent place to study the cp violation and search for new physics hints@xcite . because the hadronization process is non - perturbative in nature
, the essential problem in handling the decay processes is the separation of different energy scales , i.e. , the factorization assumption .
many approaches based on the factorization assumption have been developed , such as the naive factorization@xcite , the generalized factorization@xcite , the qcd factorization@xcite , and the perturbative qcd approach which is based on @xmath1 factorization@xcite .
recently , @xmath11 data reveal a large transverse polarization fraction , which has been considered as a puzzle , many theoretical efforts have been put to clarify it@xcite .
this suggests that @xmath4 modes must be more complicated than all the other modes and need to be studied deeply .
motivated by this , we study another @xmath4 mode in the perturbative approach ( pqcd ) within the standard model . for @xmath12 decay ,
only penguin operators contribute and we find that the branching ratio is at the order of @xmath13 . for @xmath14 decay ,
current - current operators and penguin operators can contribute and we find that the branching ratio is at the order of @xmath15 .
the longitudinal polarization predominates over transverse polarization and its fraction is found to go beyond @xmath16 .
our predictions are consistent with the current experimental values .
we hope that our study will help to resolve the above - mentioned puzzle a bit .
the remaining part of this paper is organized as follows . in sec.ii , we calculate analytically the related feynman diagrams and present the various decay amplitudes for the decay modes studied . in sec.iii , we give the numerical analysis for the branching ratios and polarization fraction of the related decay modes and compare them with the measured values . the summary and some discussion are included in the final section .
in pqcd approach , the decay amplitude is expressed as the convolution of the mesons light - cone wave functions , the hard scattering kernels and the wilson coefficients , which stand for the soft , hard and harder dynamics , respectively .
the formalism can be written as : @xmath17\end{aligned}\ ] ] where @xmath18 denotes the trace over dirac and color indices .
@xmath19 is wilson coefficient of the four - quark operator which results from the radiative corrections at short distance .
the wave function @xmath20 absorbs non - perturbative dynamics of the process , which is process independent .
the hard part @xmath21 is rather process - independent and can be calculated in perturbative approach .
the @xmath22 is the conjugate space coordinate of the transverse momentum , which represents the transverse interval of the meson .
@xmath23 is the largest energy scale in hard function @xmath21 , while the jet function @xmath24 comes from the resummation of the double logarithms @xmath25 , called threshold resummation@xcite , which becomes larger near the endpoint .
the sudakov form factor @xmath26 is from the resummation of double logarithms @xmath27 @xcite . in this paper
, we use the light - cone coordinates to describe the four - dimensional momentum as @xmath28 .
we work in the frame with the @xmath3 meson at rest , so the meson momentum can be written as @xmath29 in which @xmath30 is defined by @xmath31 and @xmath32 .
@xmath33 refer to @xmath34 respectively . to extract the helicity amplitudes , we parameterize the following polarization vectors .
the longitudinal polarization must satisfy the orthogonality and normalization : @xmath35 , and @xmath36
. then we can give the manifest form as follows : @xmath37 as to the transverse polarization vectors , we can choose the simple form : @xmath38 the decay width for these channel is : @xmath39 where @xmath40 is the three - dimensional momentum of the final state meson , and @xmath41 denotes the helicity states of the two vector mesons with @xmath42 standing for the longitudinal(transverse ) component . as discussed in ref.@xcite , the amplitude @xmath43 is decomposed into @xmath44 we can define the longitudinal @xmath45 , transverse @xmath46 helicity amplitudes @xmath47 where @xmath48 .
and we can deduce that they satisfy the relation @xmath49 there is another set of definition of helicity amplitudes @xmath50 with the normalization factor @xmath51 .
these helicity amplitudes satisfy the relation , @xmath52 where the notation @xmath53 denote the longitudinal , parallel , and perpendicular polarization amplitudes .
our next task is to calculate the matrix elements @xmath54 and @xmath55 of the operators in the weak hamiltonian with pqcd approach .
we have to use the mesons light - cone wave functions , they are universal for all decay channels . we employ the following wave functions as in other pqcd calculations @xcite .
@xmath56\ ; , \nonumber\\ & & \frac{1}{\sqrt{2n_c } } [ m_\phi\not \epsilon_2(t)\phi_\phi^v(x)+ \not\epsilon_2(t)\not p_2\phi_\phi^t(x ) + \frac{m_\phi}{p_2\cdot n_- } i\epsilon_{\mu\nu\rho\sigma}\gamma_5\gamma^\mu\epsilon_2^\nu(t ) p_2^\rho n_-^\sigma \phi_\phi^a(x)]\ ; , \nonumber\\ & & \frac{1}{\sqrt{2n_c}}[m_{\rho}\not \epsilon_3(l)\phi_{\rho}(x ) + \not\epsilon_3(l)\not p_3 \phi_{\rho}^{t}(x)+m_{\rho } i\phi_{\rho}^s(x)]\ ; , \nonumber\\ & & \frac{1}{\sqrt{2n_c } } [ m_{\rho}\not \epsilon_3(t)\phi_{\rho}^v(x)+ \not\epsilon_3(t)\not p_3\phi_{\rho}^t(x ) + \frac{m_{k^*}}{p_3\cdot n_+ } i\epsilon_{\mu\nu\rho\sigma}\gamma_5\gamma^\mu\epsilon_3^\nu(t ) p_3^\rho n_+^\sigma \phi_{\rho}^a(x)]\;. \nonumber\end{aligned}\ ] ] where @xmath57 and @xmath58 are dimensionless vectors on the light cone . @xmath59 is the momentum fraction .
the effective hamiltonian for the process@xmath60 is given as @xcite @xmath61 \nonumber\\-v_t\sum_{i=3}^{10}c_i(\mu)o_i^{(q)}(\mu)\},\end{aligned}\ ] ] where @xmath62 , @xmath63 , @xmath64 are the wilson coefficients , and the operators are @xmath65 here i and j stand for @xmath66 color indices .
the sum over @xmath67 runs over the quark fields that are active at the scale @xmath68 , i.e. , @xmath69 . from the effective hamiltonian
, we can see that the current - current operators have no contribution .
for factorizable diagrams , all the penguin operators contribute , but for the non - factorizable diagrams only the operators @xmath70 can contribute because of the color structure .
the leading order diagrams for these decays are shown in fig.1 .
we first calculate the usual factorization diagrams ( a)and ( b ) .
@xmath71\nonumber\\ & \times e_e(t_e^{(1)})h_e(x_1,x_3,b_1,b_3)\nonumber\\ & + 2r_{\rho}\phi_{\rho}^s(x_3)e_e(t_e^{(2)})h_e(x_3,x_1,b_3,b_1)\}\end{aligned}\ ] ] @xmath72\nonumber\\ & \times e_e(t_e^{(1)})h_e(x_1,x_3,b_1,b_3)\nonumber\\ & + r_{\rho}[\phi_{\rho}^v(x_3)+\phi_{\rho}^a(x_3)]e_e(t_e^{(2)})h_e(x_3,x_1,b_3,b_1)\}\end{aligned}\ ] ] @xmath73\nonumber\\ & \times e_e(t_e^{(1)})h_e(x_1,x_3,b_1,b_3)\nonumber\\ & + r_{\rho}[\phi_{\rho}^v(x_3)+\phi_{\rho}^a(x_3)]e_e(t_e^{(2)})h_e(x_3,x_1,b_3,b_1)\}\end{aligned}\ ] ] where @xmath74 is a color factor .
the function @xmath75 , including the jet function @xmath76(threshold resummation for non - factorizable diagrams is weaker and negligible ) , is the same as the @xmath75 in @xcite .
the factors @xmath77 contain the evolution from the @xmath78 boson mass to the hard scales @xmath23 in the wilson coefficients @xmath79 , and from t to the factorization scale @xmath80 in the sudakov factors @xmath26 : @xmath81 the wilson coefficients @xmath82 in eq.[aet ] are given by @xmath83 for the non - factorizable diagrams ( c ) and ( d ) , all the three meson wave functions are involved and the amplitudes @xmath84 are written as @xmath85\nonumber\\ & \times e_{e4}(t_d^{(1)})h_d^{(1)}(x_1,x_2,x_3,b_1,b_2)\nonumber\\ & + \phi_{\phi}(x_2)\left[(1-x_2)\phi_{\rho}(x_3)+r_{\rho}x_3(\phi_{\rho}^t(x_3)-\phi_{\rho}^s(x_3))\right]\nonumber\\ & \times e_{e4}(t_d^{(2)})h_d^{(2)}(x_1,x_2,x_3,b_1,b_2)\end{aligned}\ ] ] @xmath86\nonumber\\ & \times e_{e4}(t_d^{(1)})h_d^{(1)}(x_1,x_2,x_3,b_1,b_2)\nonumber\\ & + ( 1-x_2)(\phi_{\phi}^v(x_2)+\phi_{\phi}^a(x_2))\phi_{\rho}^t(x_3)\nonumber\\ & \times e_{e4}(t_d^{(2)})h_d^{(2)}(x_1,x_2,x_3,b_1,b_2)\}\end{aligned}\ ] ] @xmath87\nonumber\\ & \times e_{e4}(t_d^{(1)})h_d^{(1)}(x_1,x_2,x_3,b_1,b_2)\nonumber\\ & + ( 1-x_2)(\phi_{\phi}^v(x_2)+\phi_{\phi}^a(x_2))\phi_{\rho}^t(x_3)\nonumber\\ & \times e_{e4}(t_d^{(2)})h_d^{(2)}(x_1,x_2,x_3,b_1,b_2)\}\end{aligned}\ ] ] @xmath88\nonumber\\ & \times e_{e6}(t_d^{(1)})h_d^{(1)}(x_1,x_2,x_3,b_1,b_2)\nonumber\\ & + \left[-(1-x_2+x_3)\phi_{\rho}(x_3)+r_{\rho}x_3(\phi_{\rho}^t+\phi_{\rho}^s(x_3))\right]\nonumber\\ & \times e_{e6}(t_d^{(2)})h_d^{(2)}(x_1,x_2,x_3,b_1,b_2)\end{aligned}\ ] ] @xmath89\nonumber\\ & \times e_{e6}(t_d^{(2)})h_d^{(2)}(x_1,x_2,x_3,b_1,b_2)\}\end{aligned}\ ] ] @xmath90\nonumber\\ & \times e_{e6}(t_d^{(2)})h_d^{(2)}(x_1,x_2,x_3,b_1,b_2)\}\end{aligned}\ ] ] the evolution factors are given by @xmath91 with the sudakov factor @xmath92 .
the wilson coefficients @xmath93 appearing in the above formulas are @xmath94 the amplitudes for@xmath95 are written as @xmath96 where @xmath97 denotes the factorizable contributions and @xmath98 the non - factorizable contributions . for the other two decay channels of @xmath60 , the amplitudes are the following : for @xmath99 , @xmath100 and for @xmath101 , @xmath102 for the process
@xmath14 , it is a @xmath104 transition and we use the effective hamiltonian @xcite @xmath105 \quad .\end{aligned}\ ] ] we specify below the operators in @xmath106 for @xmath107 : @xmath108 @xmath109 from the effective hamiltonian we can see that the current - current operators and penguin operators can contribute .
the leading order diagrams for the decay are shown in fig.2 , the amplitude for @xmath110 mode is @xmath111 where @xmath112 . here
@xmath113 , @xmath114 and amplitude for the corresponding cp conjugate model is written as @xmath115 next we calculate the hard part in pqcd approach .
for the factorizable diagram ( e ) and ( f ) , we have @xmath116\nonumber\\ & \times e_e^{\prime}(t_e^{(1)})h_e(x_1,x_2,b_1,b_2)\nonumber\\ & + 2r_{\phi}^{\prime}\phi_{\phi}^s(x_2)e_e^{\prime}(t_e^{(2)})h_e(x_2,x_1,b_2,b_1)\}\end{aligned}\ ] ] @xmath117\nonumber\\ & \times e_e^{\prime}(t_e^{(1)})h_e(x_1,x_2,b_1,b_2)\nonumber\\ & + r_{\phi}^{\prime}[\phi_{\phi}^v(x_2)+\phi_{\phi}^a(x_2)]e_e^{\prime}(t_e^{(2)})h_e(x_2,x_1,b_2,b_1)\}\end{aligned}\ ] ] @xmath118\nonumber\\ & \times e_e^{\prime}(t_e^{(1)})h_e(x_1,x_2,b_1,b_2)\nonumber\\ & + r_{\phi}^{\prime}[\phi_{\rho}^v(x_2)+\phi_{\phi}^a(x_2)]e_e(t_e^{(2)})h_e(x_2,x_1,b_2,b_1)\}\end{aligned}\ ] ] with @xmath119 equal to @xmath120 except for @xmath121 replaced by @xmath122 . the expression for @xmath123
are the same as @xmath124 but with wilson coefficients @xmath93 replaced by @xmath125 .
as before , the factor @xmath126 are given by @xmath127 where the wilson coefficients are the following @xmath128 for non - factorizable diagram ( g ) and ( h ) , we find that @xmath129\times e_{e4}^{(q)\prime}(t_d^{(1)})h_d^{(1)}(x_1,x_2,x_3,b_1,b_3)\nonumber\\ & + \phi_{\rho}(x_3)\left[(1-x_3)\phi_{\phi}(x_2)+r_{\phi}^{\prime}x_2(\phi_{\phi}^t(x_2)-\phi_{\phi}^s(x_2))\right]\nonumber\\ & \times e_{e4}^{(q)\prime}(t_d^{(2)})h_d^{(2)}(x_1,x_2,x_3,b_1,b_3)\}\end{aligned}\ ] ] @xmath130\times e_{e6}^{(q)\prime}(t_d^{(1)})h_d^{(1)}(x_1,x_2,x_3,b_1,b_3)\nonumber\\ & + \left[-(1-x_3+x_2)\phi_{\phi}(x_2)+r_{\phi}^{\prime}x_2(\phi_{\phi}^t(x_2)+\phi_{\phi}^s(x_2))\right]\nonumber\\ & \times e_{e6}^{(q)\prime}(t_d^{(2)})h_d^{(2)}(x_1,x_2,x_3,b_1,b_3)\\end{aligned}\ ] ] @xmath131\nonumber\\ & \times e_{e4}^{(q)\prime}(t_d^{(1)})h_d^{(1)}(x_1,x_2,x_3,b_1,b_3)\nonumber\\ & + ( 1-x_3)(\phi_{\rho}^v(x_3)+\phi_{\rho}^a(x_3))\phi_{\phi}^t(x_2)\nonumber\\ & \times e_{e4}^{(q)\prime}(t_d^{(2)})h_d^{(2)}(x_1,x_2,x_3,b_1,b_3)\}\end{aligned}\ ] ] @xmath132\times e_{e6}^{(q)\prime}(t_d^{(1)})h_d^{(1)}(x_1,x_2,x_3,b_1,b_3)\nonumber\\ & + [ ( 1-x_3)\left(\phi_{\rho}^v(x_3)-\phi_{\rho}^a(x_3)\right)\phi_{\phi}^t(x_2)-2r_{\phi}^{\prime}\nonumber\\ & ( 1-x_3+x_2 ) \left(\phi_{\rho}^v(x_3)\phi_{\phi}^v(x_2)-\phi_{\rho}^a(x_3)\phi_{\phi}^a(x_2)\right)]\nonumber\\ & \times e_{e6}^{(q)\prime}(t_d^{(2)})h_d^{(2)}(x_1,x_2,x_3,b_1,b_3)\}\end{aligned}\ ] ] @xmath133\nonumber\\ & \times e_{e}^{(q)\prime}(t_d^{(1)})h_d^{(1)}(x_1,x_2,x_3,b_1,b_3)\nonumber\\ & + ( 1-x_3)(\phi_{\rho}^v(x_3)+\phi_{\rho}^a(x_3))\phi_{\phi}^t(x_2)\nonumber\\ & \times e_{e}^{(q)\prime}(t_d^{(2)})h_d^{(2)}(x_1,x_2,x_3,b_1,b_3)\}\end{aligned}\ ] ] @xmath134\phi_{\phi}^t(x_2)\nonumber\\ & \times e_{e}^{(q)\prime}(t_d^{(1)})h_d^{(1)}(x_1,x_2,x_3,b_1,b_3)\nonumber\\ & + [ ( 1-x_3)\left(\phi_{\rho}^v(x_3)-\phi_{\rho}^a(x_3)\right)\phi_{\phi}^t(x_2)-2r_{\phi}^{\prime}\nonumber\\ & ( 1-x_3+x_2 ) \left(\phi_{\rho}^v(x_3)\phi_{\phi}^a(x_2)-\phi_{\rho}^a(x_3)\phi_{\phi}^v(x_2)\right)]\nonumber\\ & \times e_{e}^{(q)\prime}(t_d^{(2)})h_d^{(2)}(x_1,x_2,x_3,b_1,b_3)\}\end{aligned}\ ] ] the evolution factors are given by @xmath135 with the sudakov factor @xmath136 and the wilson coefficients are given by @xmath137 the parameters used in our calculations are : the fermi coupling constant @xmath138 , the meson masses @xmath139 ,
@xmath140 , @xmath141 , @xmath142 , the decay constant @xmath143 , @xmath144 , @xmath145 , @xmath146 , the central value of the ckm matrix elements @xmath147 , @xmath148 , @xmath149 and the meson lifetime@xmath150@xcite .
using the above parameters , we get the branching ratios and helicity amplitudes of @xmath0 decays ( the helicity amplitudes are in table 1 ) @xmath151 compared with the averaged results of qcdf@xcite @xmath152 or those from naive factorization @xcite @xmath153 our predictions for @xmath12 are consistent with those of qcdf , the predictions for @xmath154 are consistent with the result of naive factorization , because the nonfactorization effects in @xmath154 are little . presently , only the experimental upper limits are available at the 90% confidence level @xcite , @xmath155 obviously , our results are consistent with the data .
our predictions will be tested by the oncoming measurements . for @xmath60 decays ,
only penguin operators contribute , so there is no direct cp violation in the @xmath60 decays . for @xmath14 decays , the cp asymmetry is time dependent @xmath156 where @xmath157 is the mass difference of the two mass eigenstates of neutral @xmath158 mesons .
the direct cp violation parameter is defined @xmath159 is @xmath160 from table 1 , we can find the longitudinal fraction in @xmath161 decays go beyond @xmath16 , but the longitudinal fractions in @xmath12 decays are very small , which is similar to @xmath162 decay mode@xcite
. in @xmath12 decays , @xmath163 in @xmath106 do nt contribute via factorizable diagrams , the penguin operators contributing via factorizable diagrams are color suppressed , so that the nonfactorizable effects are the same order as the factorizable ones , which cause the @xmath12 decays not to be factorization dominated , besides the above reason , the nonfactorizable longitudinal amplitude is opposite in sign to that of the factorizable part , therefore the longitudinal amplitude gets a large cancellation between the factorizable effects and the non factorizable parts such that @xmath164 is reduced much . according to ref.@xcite
, we can get the @xmath165 vector meson transition form factors , @xmath166 which are consistent with the results with light cone sum rules @xcite .
.branching ratios and helicity amplitudes [ cols="^,^,^,^",options="header " , ]
in this paper , we calculate the branching ratios and polarization fractions of @xmath12 and @xmath14 decays in perturbative qcd approach , the predicted branching ratios are compared with the experimental data and results obtained with other approaches .
cp parameters in @xmath14 are given in our paper .
we compared with the experimental values , our results are consistent with the current experimental data .
0.3 cm we thank dong - sheng du and mao - zhi yang for helpful discussions and communications .
the @xmath167 and @xmath168 distribution amplitudes up to twist 3 are given by @xmath169\},\nonumber\\ \nonumber & & \phi^s_{\phi}(x)=\frac{f_{\phi}^t}{4\sqrt{2n_c}}[3(1 - 2x ) ( 4.5 - 11.2x+11.2x^2)+1.38\ln{\frac{x}{1-x}}],\nonumber\\ & & \phi^t_{\phi}(x)=\frac{3f_{\phi}^t}{2\sqrt{2n_c}}x(1-x ) \left[1 + 0.2c_4^{\frac{3}{2}}(1 - 2x ) \right],\nonumber\\ & & \phi^v_{\phi}(x)=\frac{f_{\phi}^t}{2\sqrt{2n_c}}\{\frac{3}{4 } [ 1+(1 - 2x)^2]+0.24[3(1 - 2x)^2 - 1 ] + 0.96c_4^{\frac{1}{2}}(1 - 2x)\},\nonumber\\ & & \phi^a_{\phi}(x)=\frac{3f_{\phi}^t}{4\sqrt{2n_c}}(1 - 2x)[1 + 0.93(10x^2 - 10x+1)].\nonumber\\\end{aligned}\ ] ] with the gegenbauer polynomials , @xmath170 for @xmath168 meson , its lorentz structures are similar to @xmath167 meson and the distribution amplitudes are given by @xmath171\ ; , \label{pwr}\nonumber\\ & & \phi_{\rho}^t(x)=\frac{f^t_{\rho}}{2\sqrt{2n_c } } \{3(1 - 2x)^2 + 0.3(1 - 2x)^2[5(1 - 2x)^2 - 3 ] + 0.21[3 - 30(1 - 2x)^2 + 35(1 - 2x)^4]\}\ ; , \label{pwt}\nonumber\\ & & \phi_{\rho}^s(x ) = \frac{3f_\rho^t}{2\sqrt{2n_{c } } } ( 1 - 2x)[1 + 0.76(10x^2 - 10x+1)]\ ; , \label{pws}\nonumber\\ & & \phi_\rho^t(x)=\frac{3f_\rho^t}{\sqrt{2n_c } } x(1-x)\left[1 + 0.2c_2^{3/2}(1 - 2x)\right]\ ; , \label{pwft}\nonumber\\ & & \phi_{\rho}^v(x)=\frac{f_{\rho}}{2\sqrt{2n_c } } \bigg\{\frac{3}{4}[1+(1 - 2x)^2]+0.24[3(1 - 2x)^2 - 1 ] + 0.12[3 - 30(1 - 2x)^2 + 35(1 - 2x)^4]\bigg\}\ ; , \label{pwv}\nonumber\\ & & \phi_{\rho}^a(x ) = \frac{3f_\rho}{4\sqrt{2n_{c } } } ( 1 - 2x)\left[1 + 0.93(10x^2 - 10x+1)\right]\;. \label{pwa}\nonumber\end{aligned}\ ] ] for the amplitudes of @xmath3 and@xmath172 , meson , we employ the following distribution amplitudes : @xmath173\ ] ] which satisfies the normalization @xmath174 we choose @xmath175 , @xmath176
. things for @xmath172 are similar if we ignore @xmath66 symmetry breaking effect . as discussed in ref .
@xcite , we choose @xmath177gev .
+ 99 chuan - hung chen , yong - yeon keum , hsiang - nan li , phys.rev .
d66 ( 2002 ) 054013 han - wen huang , cai - dian l , c.d .
l , toshiyuki morii , yue - long shen , ge - liang song , and jin zhu , phys .
d 73 , 014011 ( 2006 ) jin - zhu , yue - long shen , cai - dian l , phys.rev . d72 ( 2005 ) 054015 jin - zhu , yue - long shen , cai - dian l , j.phys .
g32 ( 2006 ) 101 - 110 ying - li , cai - dian l , phys.rev .
d73 ( 2006 ) 014024 cai - dian l , yue - long shen , jin - zhu , eur.phys.j . c41 ( 2005 )
311 - 317 see for example : i.i.bigi , a.i.sanda , cp violation , cambridge . m. wirbel , b. stech , m. bauer , z. phys .
c29 , 637 ( 1985 ) ; m. bauer , b. stech , m. wirbel , z. phys .
c34 , 103 ( 1987 ) ; l .- l .
chau , h .- y .
cheng , w.k .
sze , h. yao , b. tseng , phys .
d43 , 2176 ( 1991 ) , erratum : d58 , 019902(1998 ) a. ali , g. kramer and c.d .
l , phys . rev .
d 58 , 094009 ; c.d .
l , nucl .
proc . suppl .
74,227(1999 ) .
y.h.cheng , et al , phys . rev .
d 60 , 094014 ( 1999 ) ; h.y.cheng , k.c.yang , phys.rev.d 62,054029(2000 ) . m. beneke , g. buchalla , m. neubert and c.t .
sachrajda , phys .
83 , 1914 ( 1999 ) ; nucl . phys .
b 591 , 313 ( 2000 ) ; nucl .
b 606 , 245 ( 2001 ) .
keum , h - n .
li , and a.i .
sanda , phys .
b 504 , 6 ( 2001 ) ; phys . rev .
d 63 , 054008 ( 2001 ) ; y.y . keum and h - n li , phys . rev .
d 63 , 074006 ( 2001 ) . c. d. l , k. ukai , and m. z. yang , phys
d 63 , 074009 ( 2001 ) ; c. d. l and m.z .
yang , eur . phys .
j. c 23 , 275 ( 2002 ) .
y.grossman , int.j.mod.phys . , 2004 , a 19,907 .
y.d.yang , g.r.lu , and r.m.wang , phys.rev . , 2005 ,
d 72,015009 .
, 2004 , b 601 , 151 .
c.w.bauer , d.pirjol , i.z.rothstein and i.w.stewart , phys.rev . , 2004 ,
d 70 , 054015 .
p.colangelo , f.de.fazio , and t.n.pham , phys.lett . , 2004 , b 597 , 291 ;
m.ladisa , v.laporta , g.nardulli , p.santorelli , phys.rev . , 2004 ,
d 70 , 114025 ; h.y.cheng , c.k.chua and a.d.soni , phys.rev.,2005 , d 71 , 014030 .
w.s.hou , and m.nagashima , hep - ph/0408007 . h.n.li and s.mishima ,
phys.rev . , 2005 ,
d 71 , 054025 .
p.k.das , k.c.yang , phys.rev .
, 2005 , d 71 , 094002 ; k.c.ynag , phys.rev.d 72 , 034009 ( 2005 ) , erratum - ibid.d 72 , 059901 ( 2005 ) .
s. catani , m. ciafaloni and f. hautmann , phys .
, 1990 , b 242 , 97 ; nucl .
phys . , 1991 ,
b 366 , 135 .
li , phys .
, 2002 , d66 , 094010 ; h - n . li , k. ukai , phys .
lett . , 2003 ,
b 555 , 197 .
y.h.chen , h.n.li , phys.rev . , 1997 , d55 , 5577 ; t.w.yeh and h.n.li , phys.rev . , 1997 , d56 , 1615 . g. buchalla , a. j. buras and m. e. lautenbacher , rev .
phys . , 1996 , 68 , 1125 .
particle data group , s. eidelman et al .
, 2004 , b 592 , 1 .
hai - yang chen and kwei - chou yang , phys.lett .
2001,b511 40 - 48 .
dong - sheng du and li - bo guo , j.phys.g.nucl.part.phys.,1997 , 23,525 - 535 .
chuan - hung chen , hep - ph/ 0601019 .
cai - dian l , mao - zhi yang , eur .
j.,2003 , c28,515 . | we study @xmath0 decays in a perturbative qcd approach based on @xmath1 factorization . in this approach , we calculate factorizable and non - factorizable contributions , there are no annihilation contributions due to quark content .
we get the branching ratios and polarization fractions for @xmath2 decays .
our predictions are consistent with the current experimental data .
23.2 cm 17.0 cm -1.0 in -35pt |
galaxy morphology in the local universe provides significant information about the formation and evolution of galaxies .
massive galaxies in the nearby universe are well described by the hubble sequence , which correlates with the dominance of galaxy s central bulge , surface brightness and colors .
hubble types are also broadly correlated with physical parameters , such as the star formation rate and dynamical properties @xcite . in the classical picture ,
late - type spiral galaxies are active star forming structures with flattened , gas rich , rotationally supported exponential disks , while early - type galaxies are more luminous , massive and quiescent systems , supported by stellar velocity dispersion and have smooth elliptical isophotes with a so called de vaucouleurs " ( or similar ) light profile . in a color - magnitude ( or mass ) diagram of the local universe , the early- and late - type galaxies occupy two distinct regions , known as the red sequence and blue cloud , respectively ( blanton et al . 2003 , baldry et al . 2004 ) .
the red sequence consists of mostly non - star - forming galaxies with a bulge dominated structure and colors indicative of an old stellar population .
in contrast , the galaxies lying in the blue cloud have different star formation properties , including blue star - forming stellar populations and mostly disk - like structures .
the key question of how and over what time scale the hubble sequence has formed remains unanswered . at a basic level , exploring the origin of the hubble sequence can simply be done by investigating if high - redshift galaxies have distributions of morphological types ( early- and late - type ) and star - forming properties that resemble those in the nearby universe .
several surveys that have used the hubble space telescope ( hst ) have observed that the properties of galaxies at @xmath10 are broadly consistent with those in the local universe ( bell et al .
2004 : gems , papovich et al . 2005 : hdfn , cassata et al .
2007 : cosmos ) .
however , the morphological analysis is still controversial at the peak epoch of star - formation activity ( @xmath11 ) . until relatively recently
, most studies of galaxy morphologies at @xmath12 have been performed at rest
frame ultraviolet ( uv ) wavelengths using optical imagers ( such as hst / wfpc2 and hst / acs ) .
these works found that irregular or peculiar structures appear more common , and traditional hubble types do not appear to be present at these epochs @xcite .
this is generally explained as due to the fact that uv radiation predominantly traces emission from the star - forming regions @xcite , which tend to be more clumped and irregularly distributed than older stellar populations , and also by the fact that quenched galaxies were missing from the optical images .
the rest - frame optical regime is a better probe of the overall stellar distribution in galaxies , and early near infrared ( nir ) observations with _
hst _ and nicmos of star forming galaxies at @xmath12 from uv selected samples found that their morphology remains generally compact and disturbed also at rest - frame optical wavelengths and bear no obvious morphological similarities to lower redshift galaxies @xcite .
interestingly , however , @xcite showed that 19 spectroscopically confirmed massive galaxies ( @xmath13 ) at @xmath14 are clearly separated into two classes as a blue cloud with large star - forming galaxies , and a red sequence with compact quiescent galaxies . unlike late type galaxies , early - type galaxies ( etgs ) have been used to investigate the cosmic history of massive galaxies in many studies ( e.g. renzini ( 2006 ) , and references therein ) due to their simple elliptical morphologies and passively evolving stellar populations . at @xmath15 , there is general consensus that the majority of massive etgs ( @xmath16 ) were already in place at @xmath17 , with a number density comparable to that of local galaxies @xcite .
a number of studies have reported the emergence of massive and compact galaxies by @xmath18 , which are already quenched etgs @xcite . the number density of these galaxies rapidly increases , by a factor of five , from @xmath8 to @xmath10 , and they are up to 5 times more compact in size than local ones with similar mass @xcite .
recent works have suggested , however , that a large fraction of , and possibly even all , massive , quiscent galaxies at @xmath8 are disk dominated @xcite .
while the observation of such disks at @xmath12 is based on morphological analysis alone , typically distribution of apparent elongation , with no dynamical measures at present , the existence a sizeable fraction of compact disks at @xmath12 among massive , passive galaxies @xcite suggests that the physical mechanism responsible for quenching star - formation may be distinct from the process responsible for morphological transformation .
recently , studies of the morphologies of @xmath8 galaxies have advanced using the high resolution nir wide - field camera 3 ( wfc3 ) .
@xcite found a variety of galaxy morphologies , ranging from large , blue , disk - like galaxies to compact , red , early - type galaxies at @xmath8 with 16 massive galaxies in the hubble ultra deep field ( hudf ) .
@xcite also studied the rest - frame uv and optical morphologies with @xmath19 galaxies determined by the yhvz color - color selection in the hudf and the early release science ( ers ) field , and confirmed previous studies by showing in particular the presence of regular disk galaxies , which have been missing in previous studies , either because they are not detected at the available sensitivity or because their appearance is irregular at rest - frame uv wavelengths .
the results from these studies are generally interpreted as possible evidence at @xmath9 , at least among the brightest galaxies at that epoch , of the general correlations between spectral types and morphology that today define the hubble sequence
. the most important limitations of these works are that they are based on very small samples , which are not statistically significant ( less than 20 galaxies ) and not homogeneously selected , and their morphological analysis was restricted to only @xmath2 profile fitting or visual classifications .
significant improvements are now possible using larger samples of panchromatic images from the candels ( cosmic assembly near - infrared extragalactic legacy survey ) observations .
@xcite investigated how the structure of galaxies ( @xmath2 index and size ) depends on galaxy position in the sfr
stellar mass diagrams since @xmath20 , specifically showing strong trends of specific star formation rate ( ssfr ) with @xmath2 index using large data sets combined in 4 different fields , cosmos , uds , goods - south and north .
although the @xmath2 index , measured by fitting a single @xmath2 profile to a galaxy , is the most common approach to analyzing galaxy morphology , it is also useful to study morphologies with non - parametric measures such as gini ( @xmath3 ) @xcite , @xmath4 ( lotz et al .
2004 ) , multiplicity ( @xmath7 ) ( law et al . 2007 ) and cas ( abraham et al . 1996 , conselice et al . 2003 ) since not all galaxies are described by smooth and symmetric profiles . in a recent candels paper , @xcite used gini ( @xmath3 ) , @xmath4 and visual classifications to identify a correlation between morphologies and star - formation status at @xmath8 , showing two distinct populations , bulge - dominated quiescent galaxies , and disky or irregular star - forming galaxies , though they only use massive galaxies ( @xmath21 ) .
recent panchromatic surveys such as candels hold the promise of significant progress in investigating galaxy structures at high redshift because they combine sensitive _ hst _ morphology at rest - frame uv and optical wavelengths with the depth and accuracy of space borne photometry .
the candels project also adds coverage of a substantial amount of sky , which results in samples whose size and dynamic range in mass are about one order of magnitude , or more , larger than in previous works . in this study , we extend previous results using a statistically significant sample ( 1,671 galaxies ) down to a lower mass limit ( @xmath22 m@xmath23 at @xmath24 and @xmath25 m@xmath23 at @xmath26 , specifically for passive galaxies ) and using various morphological parameters ( non - parametric diagnostics such as @xmath3 , @xmath4 , @xmath7 , concentration ( c ) and asymmetry ( a ) , and the parametric @xmath2 light profiles , characterized by the @xmath2 index ( n ) and half - light radius @xmath27 ) , as well as visual inspection .
the combination of high angular resolution and sensitivity afforded by the _ hst_/acs and wfc3 images with the relatively large size of the sample allows us to probe in a statistical fashion the correlations between galaxy structures and star - formation activity at @xmath8 , i.e. the epoch when the cosmic star
formation activity reached its peak , to their counterparts in the local universe .
the structure of this paper is as follows .
the optical and infrared data and selection of our galaxy sample are introduced in section 2 .
the rest - frame color and mass distributions are described in section 3 .
we present the analysis of galaxy morphologies in the rest - frame optical using the distribution of non - parametric approaches , as well as @xmath2 index and half - light radius in section 4 .
comparison with galaxies from the local universe is shown in section 5 and the results of a comparison of rest - frame uv morphologies with rest - frame optical are presented in section 6 .
finally , we conclude with a discussion of our results and compare them to other studies in section 7 and summarize our results in section 8 .
all the data used in this work come from the observations acquired during the goods ( giavalisco et al .
2004 ) and candels ( grogin et al . 2012 ; koekemoer et al .
2012 ) projects in the goods south field , including both space born ( chandra , hubble and spitzer ) as well as ground based ( vlt ) data .
the candels _ hst _ observations , including the details of the data acquisition , reduction and calibration , source identification and photometry extraction , are thoroughly described elsewhere ( see grogin et al .
2012 ; koekemoer et al .
2012 ; guo et al .
2013 ) ; here we briefly review key features of the wfc3 images that are relevant to this work .
the _ hst _ component of candels consists of a multi - cycle treasury program to image five distinct fields ( goods - north and -south , egs , uds and cosmos ) using both wfc3 and acs .
the whole project is organized as a two
tier deep+wide survey .
the candels / deep survey covers about 125 square arc minutes to @xmath28-orbit depth within goods - north and -south @xcite at f105w(y ) , f125w(j ) and f160w(h ) , while the wide survey covers a total of @xmath29 square arc minutes to @xmath30-orbit depth within all five candels fields . in this study , we use the 4-epoch ( about 2 orbits ) candels f160w(h - band ) imaging that covers about 115 square arc minutes ( @xmath31 of the whole goods - s ) including the goods - s deep region , plus the ers @xcite .
this survey has a @xmath32 limiting depth of @xmath33 , and a drizzled pixel scale of @xmath34 .
a number of photometric catalogs exist based on candels data , and here we use one where sources have been detected using the package sextractor @xcite in the wfc3 h band images , and multi wavelength photometry has been obtained using a software package with an object template - fitting method ( tfit , laidler et al .
this catalog includes photometry from the _
hst_/ acs and wfc3 images in the bvizyjh bands ; from vlt / vimos u and vlt / isaac ks images ; and from the spitzer / irac images at 3.6 , 4.5 , 5.8 and 8.0 @xmath35 m ( guo et al . 2013 ) .
we identify galaxies at @xmath0 with a broad range of star
formation properties , from passive to star forming , and with different levels of dust obscuration using photometric redshifts and ssfr estimated by fitting the candels broad - band rest
frame uv / optical / nir spectral energy distribution ( sed ) to spectral population synthesis models ( hereafter , sed sample ) .
additionally , for comparison , we also select samples of galaxies using the bzk technique , a color selection based on the ( b - z ) vs. ( z - k ) color - color diagram , widely used to identify galaxies at @xmath8 relatively independently of their star formation activity and dust obscuration properties @xcite . while characterized by some contamination by agns and
redshift interlopers , as well as incompleteness to very young and dust free star forming galaxies ( see @xcite ) , the bzk selection is overall quite effective and particularly economic in that it only requires the acquisition of three photometric bands .
in contrast , the sed selection is observationally much more expensive because it requires a large number of photometric bands to yield robust photometric redshifts as well as robust measures of the stellar population properties , i.e. stellar mass , star
formation rate , dust obscuration and age of the dominant stellar population .
for the same reason , however , it is less prone to the effects of photometric scatter and characterized by a higher degree of completeness than the bzk criterion . in view of the fact that in candels the two goods fields have deeper and fully panchromatic images relative to the other fields of the survey , here we use the sed sample as our primary data set for our study , and compare it with the bzk sample to test if they yield similar conclusions about the general morphological properties of the galaxies mix at @xmath9 .
such a comparison , which at this level of sensitivity can only be made in the goods fields , is particulary useful for those other fields where data for selecting galaxies by means of sed fitting are not available or do not have sufficient wavelength coverage and/or sensitivity for accurate results . in our particular case ,
since the bzk sample is bsed on the ground based k - band images , which are significantly shallower ( @xmath32 limiting magnitudes of ks=24.4 ) than the wfc3 images , the depth , and hence the size , of the sample is smaller than the sed one . however , since the efficiency and simplicity of the bzk selection criteria offer a distinct advantage in other fields of the candels survey , where the rich complement of photometry of the goods south field is not available , the knowledge of the relative performance and possible limitations of both selection criteria will be very useful .
measures of the stellar mass , star formation rate , dust obscuration and age of the dominant stellar population have been obtained by guo et al .
( 2013 ) using the tfit panchromatic catalog of the goods south field ( see also guo et al . 2011 , 2012 , where key results and features of the sed fitting have been discussed ) .
prior to carrying out the sed fitting , photometric redshifts have been measured for all galaxies from the 13band ubvizyjhksi@xmath36i@xmath37i@xmath38i@xmath39 photometry using the pegase 2.0 spectral library templates @xcite , as well as the available sample of 152 spectroscopic redshifts ( about 4% of our final sample ) as a training set . in the redshift range considered here
the candels photometric redshifts are of good quality , as verified by comparing them against available spectroscopic ones .
this comparison yields a mean and scatter ( @xmath40 deviation after @xmath41 clipping ) in our photometric redshift measurement of 0.0005 and 0.03 , respectively .
the properties of the dominant stellar populations are subsequently derived by fitting the observed sed to the spectral population synthesis models by fixing the redshift to the photometrically derived value and using the redshift probability function , @xmath42 , to calculate the errors from a monte carlo bootstrap .
the multi band photometry is fit to the updated version of the @xcite spectral population synthesis library with a salpeter initial mass function .
we use either a constant star formation history or an exponentially declining one ( @xmath43 ) , depending on which functions result in a smaller @xmath44 with the data .
the calzetti law @xcite is used for the dust obscuration model together with the @xcite prescription for the opacity of inter galactic medium ( igm ) ( see guo et al .
2013 for a full description of the procedure ) . in the redshift range @xmath45 , arbitrarily ( but inconsequentially )
chosen to reproduce that of the bzk selection criteria ( see @xcite ) , the photo z plus sed fitting procedures yield 3,542 galaxies with signal to noise ratio in the h - band @xmath46 ( hereafter , sed sample ) .
star forming and passive galaxies are defined based on the measure of the ssfr , namely the ratio of the star formation rate to the stellar mass . specifically , we define passive galaxies as those with @xmath47 , and using this criterion we find 105 passive galaxies and 3,437 star forming ones out of the 3,539 comprising the sed sample .
thus , with our cut on the ssfr , 3% of all galaxies at @xmath9 are classified as passive .
we have constructed the bzk sample by adopting the bzk color
color selection by @xcite , where galaxies of various spectral types " are identified by their position in the @xmath48 versus @xmath49 color - color diagram .
the bzk selection is widely used to investigate the evolution of galaxies at @xmath8 @xcite .
to the extent that the average obscuration properties of the star forming galaxies are well described by the calzetti ( 2000 ) obscuration law , this rest uv / optical color selection is sensitive to galaxies at @xmath50 with a significantly broader range of dust obscuration than the uv selection alone ( e.g. reddy et al .
2009 , 2010 ; also guo et al .
2011 for a discussion ) .
it is also sensitive to passively evolving galaxies in a similar redshift range , which the uv selection misses altogether . as in any selection of distant galaxies that is based on colors , however , the details of the redshift distribution of the selected galaxies depend on the relative sensitivity of the images and the shape of the adopted bandpasses .
the original bzk criteria were implemented using a sample where source detections were carried out in the k - band images , since these had sufficient sensitivity and were such that every galaxy detected at least in the z - band was detected in the k one with higher s / n . since this is not the case with our data , where the _
hst_/acs z - band image is much deeper than the ground based vlt / isaac ks band image even for the reddest sed considered here , we contruct our bzk samples from the acs z - band selected source catalog @xcite , where we further require @xmath51 in the k - band and @xmath52 in the z - band to ensure robust color measurements .
we then use the selection criteria introduced by @xcite as shown in figure [ fig : bzk ] : @xmath53 out of a total of 1,043 bzk galaxies , we find 981 sbzks ( blue circles in figure [ fig : bzk ] ) and 62 pbzks ( red circles ) , namely 6% of the sample is made of passive galaxies .
this fraction is twice as large as the one of the sed sample , and the reason is that the bzk selection defines galaxies as passive solely based on their colors , while in the sed sample galaxies are defined as passive based on the ssfr . if the threshold were defined as ssr@xmath54 gyr@xmath55 , then the sed sample and bzk would both have the same 6% fraction of passive galaxies . finally , it is important to keep in mind that all bzk galaxies also have detection in the wfc3/f160w candels images , which we have used to analyze their rest frame optical morphology . compared to the sed selection , the bzk selection is relatively easy and economical to implement , requiring only three photometric bands , and it is largely independent of dust reddening .
the combination of photometric scatter ( which depends on the sensitivity of the data ) and the intrinsic variance of galaxies sed , however , result in some degree of contamination by interlopers ( i.e. galaxies that are not in the targeted redshift range ) as well as of incompleteness , namely loss of galaxies that are scattered away from the selection windows .
for the same reasons , the separation between passive galaxies and star forming ones suffers from scatter , in the sense that dust reddened star
forming galaxies might be observed in the selection window of passive ones and vice versa ( see , e.g. daddi et al . 2004 , 2005 , 2007 ) . to diminish the contamination by interlopers from outside the redshift range considered here and how it affects our morphological analysis
, we can use the photometric redshift to restrict our sample to galaxies with @xmath56 .
this further cut serves two purposes .
first , it filters away the high z tail of our bzk sample , which very likely results from the combination of the relatively shallow depth of the acs b
band images , and the fact that the sample is z band selected .
the cut also serves the purpose of creating our reference cosmic epoch to assess the state of galaxy morphology evolution .
this leaves a final bzk sample of 736 galaxies down to @xmath57 , of which 46 are classified as passively evolving ( pbzk ) , i.e. 6.3% of the sample , and 690 are star forming ( sbzk ) galaxies .
we explicitly note that using the photometric redshifts to eliminate likely interlopers has left the passive fraction essenctially unchanged . in the following
we will refer to this photo z filtered sample simply as the `` bzk sample '' , while the original sample will be called the `` pure bzk '' one .
the top panel of the figure [ fig : numcount ] compares the photometric redshift distribution of the bzk sample to that of the whole h
band selected tfit sample ( e.g. sed sample not restricted by redshift ) , while the bottom one shows the number counts of the sed sample , of the two bzk samples , and also of the whole h band selected tfit sample for comparison .
it can be seen from the photometric redshift histograms that the redshift distribution of bzks is tapered at both ends of the corresponding selection window as a result of the color cuts built in the selection criteria .
clearly , this is not presented in the sed sample .
it can also be seen that the number counts of the sed and bzk samples are very similar in shape , especially at @xmath58 , the former being slightly larger than the latter simply due to the larger completeness and the fact that the redshift distribution is not set by color cuts .
since the magnitude ( mass ) distribution is not similar especially at the faint end , we cut bzk and sed samples with @xmath59 ( @xmath60 : over the 90% completeness limit of the candels @xmath61 band in the goods - s deep field ) to study and compare the morphologies directly . as we shall see later , the morphological properties of the sed and bzk samples are essentially identical , suggesting they are both representative of the mix of bright galaxies at @xmath9 .
in figure [ fig : cm ] , we report the distribution of rest - frame u v ( 3730 6030 ) color versus stellar mass for the bzk ( left ) and sed ( right ) samples .
the blue crosses and red circles represent the star - forming ( sbzk ) and passive ( pbzk ) galaxies , respectively .
the rest - frame colors are measured using the best - fitting sed template , which is scaled to pass through the closest observed photometric point for each rest - frame wavelength we consider to derive the k - correction and subsequently the rest - frame magnitude .
figure [ fig : cm ] shows that there is a distinct difference in the color - mass diagram between star - forming and passive galaxies . in both our samples ,
the colors of passive galaxies ( pbzks ) span a much smaller range than those of star - forming ones ( sbzks ) .
an important question to answer , therefore , is whether or not the limited excursion of the intrinsic colors of the pbzks is simply the result of their selection and not due to the characteristics of their sed .
after all , these galaxies are selected specifically to be red , namely to have the observed colors expected from passively evolving galaxies ( or galaxies with a relatively small specific star formation rate ) observed at @xmath62 . to test this possibility ,
we have compared the scatter of the observed colors and of the intrinsic colors of our pbzk sample in bins of both apparent and absolute magnitude , which is shown in figure [ fig : p_sigma ] . as it can be seen , the scatter _
always _ increases when going from the intrinsic colors to the observed ones , as one would expect in a sample with a relatively large dispersion of redshift .
moreover , the pbzks occupy a significantly smaller range of stellar mass , and , at the same time , the two types occupy a disjoint range of ssfr ( color - coding of points at figure [ fig : cm ] ) . taken all together , this is evidence that while pbzks are selected to be red , thus covering a restricted range of both the _ observed _ b - z and z - k colors , their _ intrinsic _ colors are all very similar , since they span a range significantly smaller than the observed ones , denoting a similarity of physical properties .
this conclusion is further reinforced by their small range of mass , since the color selection does not in principle set any restrictions on the stellar mass .
we note that although our star - forming and passive samples seperate well in figure [ fig : cm ] .
a small fraction of massive star - forming galaxies overlap with the passive ones .
the majority of those massive star - forming galaxies are red due to uv colors which are reddened by dust , and similar trends are observed in the red sequence at @xmath63 @xcite . in the bzk sample , 11 red sbzks with @xmath64 are all massive ( @xmath65 ) and generally red ( rest - frame @xmath66 except two galaxies with @xmath67 ) .
they are visually characterized by compact structures , with the exception of one having large size ( @xmath68 kpc ) and a light profile characteristic of an exponential disk ( @xmath69 )
. these could be passive galaxies that are not classified as such by the bzk criterion either because of photometric scatter in the photometric measures or because of the galaxies intrinsic sed variations ( see section 2.2 ) .
we find 70 ( 50 for the bzk sample ) x - ray detected galaxies among our sample , marked as diamond symbols in figure [ fig : cm ] .
most of them ( @xmath70 : sed sample , @xmath71 : bzk sample ) are star - forming galaxies ( sbzks ) , and those x - ray detected galaxies are generally massive and compact .
we do not exclude them from further study since they also follow a similar trend in the color - mass diagram , and have similar morphologies as non x - ray detected galaxies .
in order to investigate further the morphologies of galaxies within @xmath72 , we turn next to several non - parametric morphological diagnostics such as the gini parameter ( @xmath3 ) , the second - order moment of the brightest 20% of the galaxy pixels ( @xmath4 ) and the multiplicity parameter ( @xmath7 ) .
many studies have used these parameters to quantify galaxy morphology @xcite locally and at high redshift , generally showing that they are an effective and automated way to measure galaxy morphologies for large samples .
these parameters quantify the spatial distribution of galaxy flux among the pixels , without assuming a particular analytic function for the galaxy s light distribution .
thus they may be a better characterization of the morhology of irregulars , as well as standard hubble - type galaxies ( lotz et al .
2004 , hereafter , lpm04 ) . before measuring these parameters
, we need to identify the pixels belonging to each galaxy . for each galaxy , we calculate the `` elliptical petrosian radius '' , @xmath73 , which is defined like the petrosian radius @xcite but uses ellipses instead of circles ( lpm04 ) .
we use the segmentation map generated by sextractor when making the h - band detections @xcite , and use the ellipticities and position of peak flux determined by sextractor for each galaxy .
we then set the semi - major axis @xmath73 to the value where the ratio of the surface brightness at @xmath73 to the mean surface brightness within @xmath73 is equal to 0.2 .
the surface brightness at each elliptical aperture , @xmath74 , is measured as the mean surface brightness within an elliptical annulus from @xmath75 to @xmath76 .
there are 10 galaxies in the sed sample and one sbzk whose images comprise less than 28 pixels ( corresponding to a circle with a radius of 3 pixels ) , which we have excluded from further analysis . note again that we use galaxies with @xmath77 from both samples for our morphology analysis , which leaves us with 46 pbzks and 669 sbzks , and 104 passive and 1567 star - forming galaxies of the sed sample . using the sed and bzk samples with stellar mass @xmath78 , we first compute the g parameter defined in lpm04 , which measures the cumulative distribution function of a galaxy s pixel values ( light ) . therefore ,
g of 1 would mean all light is in one pixel while g of 0 would mean every pixel has an equal share .
hence , g is used to distinguish between the galaxies for which fluxes are concentrated within a small region or uniformly diffuse .
we also compute the @xmath4 parameter , which traces the spatial distribution of any bright nuclei , bars , spiral arms , and off
center star clusters .
typically , galaxies with high values ( @xmath79 ) are extended objects with double or multiple nuclei , whereas low values ( @xmath80 ) are relatively smooth with a bright nucleus ( see lmp04 for a detailed explanation of @xmath4 ) .
the third non - parametric coefficient is the multiplicity @xmath7 @xcite , designed to discriminate between sources based on how " multiple the source appears .
galaxies with lower @xmath7 are compact galaxies with generally one nuclei while irregular galaxies with multiple clumps have higher @xmath7 @xcite ( the definitions of each diagnostic are presented in the above references ) .
figure [ fig : nonpar ] shows the relative distribution of the @xmath3 , @xmath4 and @xmath7 for star forming and passive galaxies of the bzk and sed samples ( blue and red histogram , respectively ) .
the @xmath3 values are mostly in the range @xmath81 with a mean of 0.43 for star - forming galaxies ( 0.48 for sbzks ) and 0.53 for passive ones ( 0.58 for pbzks ) .
passive galaxies are shifted to higher @xmath3 than star - forming ones .
the majority of pbzks ( 90% ) and about 70% of passive galaxies have @xmath82 .
the mean values of the @xmath4 for star - forming ( sbzk ) and passive ( pbzk ) galaxies are -1.47 ( -1.54 ) and -1.68 ( -1.73 ) , respectively .
the middle panel of figure [ fig : nonpar ] shows that the passive galaxies ( pbzks ) have lower values and show a peak at @xmath83 while the star - forming ones ( sbzks ) exist in a wide range of @xmath4 values that are slightly skewed to higher @xmath4 .
lastly , the @xmath7 values of star - forming galaxies ( sbzks ) have a range of values up to @xmath84 , but most of the passive galaxies ( sed : 90% , bzk : 94% ) have @xmath85 .
@xcite find that spectroscopically confirmed star - forming galaxies at z=1.53.6 have @xmath86 for isolated regular galaxies , @xmath87 for sources that show some morphological irregularities , and larger values for sources having multiple clumps that are separated .
therefore , all passive galaxies ( pbzks ) tend to be dominated by one main clump while star - forming ones ( sbzks ) can have two or more significant components in addition to a main nucleation .
there is some degree of correlation between the @xmath3 and @xmath4 , @xmath7 measurements ( see figure [ fig : dist ] ) .
the passive galaxies ( pbzks ) reside in a narrow region with higher @xmath3 and lower @xmath4 and @xmath7 indicating that they consist of one bright central source .
in contrast with passive galaxies ( pbzks ) , star - forming ones ( sbzks ) with lower @xmath3 have higher @xmath4 and @xmath7 because galaxies with diffuse morphology tend to have a spread out flux distribution .
figure [ fig : nonpar ] and [ fig : dist ] also show that there is an overlap in the distributions of morphological parameters of star - forming galaxies and passive ones .
for example , there are star - forming galaxies with high @xmath3 and low @xmath7 or @xmath4 that are located in same region where the bulk of passive galaxies are observed , and vice versa . to illustrate and further explore these galaxies in the overlapped region , we have chosen star - forming galaxies with @xmath88 and passive galaxies with @xmath89 for visual inspection and classification .
we indeed found that the star forming galaxies can generally be classified as blue spheroids and the passive ones as red disks .
in agreement with @xcite , the 35 star - forming galaxies with high @xmath3 visually appear as compact structures in figure [ fig : sfg ] .
note that all these images have s / n ratio per pixel ( @xmath90 ) greater than 2.5 , the threshold used in lpm04 for reliable measurements , and most of them ( 85% ) are relatively bluer than normal passive galaxies . about 40% of star - forming galaxies in the sample of @xcite
were visually classified as such compact structures as well . among passive galaxies in the sed sample , 10 of them
have @xmath91 , and all have @xmath89 .
this indicates that we can not measure reliable morphology due to the low @xmath90 ratio . in figure
[ fig : pg ] , we show the 25 passive galaxies with @xmath89 and @xmath92 .
most of them have smooth structures , and some are elongated or have secondary structures .
they are intrinsically red with rest - frame @xmath93 , and 16 galaxies are massive ( @xmath94 ) .
red ( passive ) disks at high redshift have also been recently studied by other groups .
for example , @xcite found that 30% in quiescent galaxies of their sample with @xmath95 at @xmath96 can be morphologically classified as disks .
this is generally consistent with the findings presented here , although we note that due to their low @xmath90 , some of our `` passive disks '' might actually be morphological mis - classifications or even be dust
obscured star forming galaxies .
finally , we observe that , overall , the bzk and the sed samples have essentially identical distributions of morphological parameters , although the sed sample includes galaxies with lower @xmath3 , namely those with @xmath97 , which is the result of their lower surface brightness .
in addition to @xmath3 , @xmath4 and @xmath7 , we measure the concentration ( @xmath5 ) and asymmetry ( @xmath6 ) of our samples .
the concentration index @xmath5 @xcite measures the concentration of flux .
typical values of @xmath5 range from @xmath98 for the least compact to @xmath99 for most compact galaxies .
note that asymmetry @xmath6 @xcite compute the @xmath100 degree rotational asymmetric light distribution of all galaxy and hence the most symmetric galaxies have @xmath101 .
we present the distribution of @xmath5 and @xmath6 for the bzk ( left ) and sed ( right ) samples in figure [ fig : cas ] .
as expected , the passive galaxies ( pbzks ) are more similar to ellipticals in their @xmath5 and @xmath6 values , while the star - forming ones ( sbzks ) are more spiral and merger - like . in the @xmath102 plane ,
passive galaxies ( pbzks ) are different from star - forming ones ( sbzks ) , but the difference is not as significant compared to the difference in @xmath103 and @xmath7 .
as expected , the distribution of @xmath5 and @xmath6 for both samples are similar as shown in the previous figures . in summary , the distributions of the non parametric morphological diagnostics that we have considered here for both the bzk and sed samples are essentially the same in each spectral type class .
star forming and
passive galaxies clearly show different distributions of non - parametric morphological measures , and they are separated well in @xmath3@xmath4 and @xmath7 spaces .
passive galaxies ( pbzks ) are mostly compact , spheroidal structures , and the majority of star - forming ones ( sbzks ) are somewhat extended or have multiple clumps , similar to disks or irregular galaxies in the local universe .
these results agree with those of @xcite who also studied the morphologies of massive galaxies ( @xmath16 ) at @xmath8 with @xmath3 , @xmath4 and visual classifications .
they found that the quiescent galaxies are bulge dominated and star - forming galaxies have disks or irregular morphologies visually as well as in the @xmath3 and @xmath4 analysis .
we extend their study with a larger sample down to a lower mass limit , and obtain almost the same conclusion about galaxy morphologies at @xmath9 . in appendix
a , we investigate the robustness of non - parametric measures ( @xmath3 , @xmath4 and @xmath7 ) , mainly used in this study for morphological analysis , using goods - s and the hubble ultra deep field ( udf ) images in the h - band .
the udf overlaps part of the goods - s imaging , but goes much deeper ( 5 @xmath104 depth of 28.8 ) , and thus offers an opportunity to test the dependence of parameters on the signal - to - noise per pixel ( @xmath90 ) .
we show that any difference between the two different fields , which have different exposure times , is relatively small for three parameters , with the scatter in measured properties increasing as s / n decrease .
we find that most ( @xmath105 ) of bzk galaxies have @xmath106 , and @xmath107 of the sed sample have @xmath106 .
we note that we do not exclude galaxies with @xmath108 since they rarely change our results in this study .
@xmath2 index and half - light radius ( @xmath27 ) have been successfully used to characterize galaxy morphology in many previous works , both at low and high redshift ( low - z : blanton et al .
2003 , driver et al .
2006 ; mid - z : cheung et al .
2012 ; high - z : ravindranath et al .
2006 , bell et al .
recently , @xcite and @xcite show that the @xmath2 index correlates well with quiescence in galaxies at @xmath109 .
therefore , we investigate how galaxy morphologies with @xmath110 and @xmath7 correlate with @xmath2 index ( n ) and @xmath27 .
we use the @xmath2 index and @xmath27 @xcite obtained by fitting a @xmath2 profile to the galaxy image using galfit @xcite .
passive galaxies in both samples have @xmath111 , and over 96% of them have @xmath112 , with 50% having @xmath113 .
in contrast , star
forming galaxies have @xmath114 , with 85% of them having @xmath115 .
this suggests that the majority of star - forming ( sbzk ) galaxies have disk - like ( exponential light profile ) or irregular structure with a light profile shallower than an exponential one .
in contrast , all passive galaxies ( pbzks ) have a dominant bulge including some bulge+disk structures .
a similar analysis of morphologies at @xmath116 using the @xmath2 index in the sfr - mass diagram was carried out by @xcite , who found that the main sequence ( ms ) consists of star - forming galaxies with near exponential profiles , and passive galaxies below the ms have higher @xmath2 indices close to a de vaucouleur profile ( @xmath117 ) .
@xcite also reported similar results with 16 massive galaxies at @xmath8 , and found that star - forming galaxies have diskier ( low n ) profiles than passive galaxies .
we present the distribution of @xmath110 and @xmath7 as a function of @xmath2 index in two different mass bins divided by a threshold mass , @xmath118 in figure [ fig : sersic ] ( sed sample ) and [ fig : sersicb ] ( bzk sample ) . in both samples
, we find that there are significant correlations between @xmath2 index and @xmath110 and @xmath7 , with galaxies with high @xmath119 having high @xmath3 and low @xmath120 , and vice versa .
as we have already noted in figure [ fig : cm ] , most of the passive galaxies ( pbzks ) have masses greater than @xmath121 , and the majority of star - forming galaxies ( 78% ) and @xmath122 for sbzks have @xmath123 . in massive systems ( @xmath124 ) , the two populations show a well - separated bimodal distribution in their morphologies and colors ( see figures 10 and 11 ) .
red passive galaxies ( pbzks ) show spheroidal - like structures with high @xmath119 , @xmath3 and low @xmath4 , @xmath7 , while blue star - forming ones ( sbzks ) exhibit a larger variety of morphologies , but mainly have low @xmath119 , @xmath3 and high @xmath4 , @xmath7 .
there are some star - forming galaxies ( sbzks ) with high @xmath2 index ( @xmath125 , vertical dotted line in figure [ fig : sersic ] , [ fig : sersicb ] ) .
they follow mostly the same trend in morphologies with higher @xmath3 and lower @xmath4 , @xmath7 , indicating the presence of a bright center ( see the blue spheroids in figure [ fig : sfg ] ) .
@xcite showed examples of such systems in their sample , and found that those appear to be spheroidal - like structures , but in many cases also have significant asymmetries , or faint secondary sources and tidal tails . a loose relation between non - parametric measures and @xmath2 index
is observed for massive galaxies in the right panel of figure [ fig : sersic ] , [ fig : sersicb ] , but not for low mass galaxies at all .
this means that the commonly used @xmath2 index is not enough to study morphology of those galaxies .
therefore , it is important to use the non - parametric diagnostics in addition to the @xmath2 profile to quantify the morphology of galaxies towards the low end of the mass distribution . in figure
[ fig : re ] ( sed sample ) and [ fig : reb ] ( bzk sample ) , we plot the non - parametric measures as a function of @xmath27 in both small and large mass systems to see how the size varies along with spectral type and stellar masses . we find that @xmath27 also correlates well with all the non - parametric measures in general , as galaxies with low @xmath3 and high @xmath4 , @xmath7 have smaller sizes and relatively low @xmath2 index ( @xmath115 : empty symbols ) in both samples .
overall , star - forming galaxies ( sbzks ) tend to have larger half - light radii than passive ones ( pbzks ) , even in the case of massive systems ( @xmath126 ) and about half of passive ( pbzk ) galaxies show very compact morphologies , with @xmath127 .
this is consistent with previous results , which find that passive galaxies are more compact than star forming galaxies at @xmath9 @xcite , and the same general trend is observed at @xmath63 among massive galaxies @xcite .
the strong correlation between galaxy color ( and sfr ) and morphology shown in the previous sections is reminiscent of the hubble sequence at @xmath128 . however , to understand if actually the hubble sequence is in place at @xmath8 , it is important to examine how galaxy morphologies at @xmath8 differ from the local galaxies . in general , comparing morphological parameters of local galaxies , which are observed at relatively high resolutions to their conterparts in high - z samples , whose resolution is lower , is not straightforward because because most morphological diagnostics do depend on the resolution . a fortunate case , however , is that of the comparison between local galaxies at redshift @xmath129 from the sdss survey to galaxies at @xmath130 observed with _ hst _ , since the difference of angular diameter distance at these two redshifts nearly perfectly compensates for the difference in the angular resolution of the two instrumental configuration @xcite . at this purpose , we should also keep in mind that in our adopted cosmology , the fractional variation of the angular diameter distance in the redshift range @xmath131 is only @xmath132% . furthermore , even though @xmath3 and @xmath4 measures are robust given the resolution of the observation , particularly if the data is deep enough to allow the petrosian radius to be used to measure the parameters @xcite , we should nonetheless be careful when directly comparing the @xmath3 and @xmath4 from observations with different resolutions ( lpm04 ; lisker et al . 2008 ) .
therefore , in this study , we compare the @xmath8 galaxy morphologies in the sed and bzk samples to those of the local galaxy sample of lpm04 after we simulate how they would appear in the candels images if they were observed at redshift @xmath9 .
for this reason , we have used the b band and g band images of the local galaxies @xcite , which at this redshift correspond to the h band ( for details about the local galaxy observations , see lpm04 ) .
we have restricted the comparative analysis with the high redshift galaxies to only those at @xmath133 to minimize the possible effects of the morphological k - correction .
furthermore , we have only considered galaxies within a magnitude range of @xmath134 ( lpm04 ) , where we take @xmath135 @xcite for local galaxies , and @xmath136 for galaxies at @xmath137 @xcite , assuming that the local galaxies were brighter in the past but did not evolve morphologically ( lpm04 ) . in this simulation
, we first modify the angular sizes and surface brightness of local galaxy images to account for distance and cosmological effects .
the images are rebinned to the pixel scale of the galaxies observed at @xmath138 ( wfc3 pixel size is 0.06 " ) and the flux in each pixel is rescaled so that the total magnitude of the galaxy corresponds to some preassigned value , for example to that of an m * galaxy at @xmath138 .
the modified images are then convolved with the wfc3 psf and , lastly , we add poisson noise appropriate to the wfc3/nir observations using the iraf task mknoise . in figure [
fig : local ] , we present the @xmath3 , @xmath4 and @xmath7 measured from the redshifted modified local galaxy images .
the measures of the redshifted galaxies ( from the left , 1st and 2nd panels ) are quite different from those of the original local galaxy images ( 3rd panel ) .
this is in agreement with lpm04 , who conclude that @xmath8 lyman break galaxies ( lbgs ) do not have morphologies identical to local galaxies .
overall , the distributions of galaxies at @xmath139 in both samples and that of the redshifted local galaxies are similar in @xmath3-@xmath4 space , as shown in the 1st and 2nd panel of figure [ fig : local ] , but the high redshift star forming galaxies have a broader distribution of @xmath4 for a given value of @xmath3 than the redshifted local late types .
this trend is reflected in the @xmath3-@xmath7 plane ( in the bottom panels ) , which shows lower @xmath7 values for the redshifted local galaxies .
large and luminous star
forming disks are mostly responsible for this excess of galaxies with higher @xmath4 and broader ( slightly higher ) distribution of @xmath7 , another manifestation of the fact that disks at @xmath9 are not simply scaled up versions of the local ones in terms of star formation rates , but are intrinsically different ( e.g. papovich et al .
2005 , law et al .
comparatively , the @xmath9 passive galaxies have @xmath3 and @xmath4 values that are much more similar to those observed for the present day e / s0 galaxies .
overall , the qualitative similarity of values , shapes and trends of the distributions of morphological parameters at low and high redshift suggests that the hubble sequence is essentially in place at @xmath9 .
our comparative study also shows that while the morphology of the oldest systems at any epoch , i.e. the passively evolving galaxies , in general changes relatively little from @xmath63 to the present , at least as traced by our diagnostics , disk galaxies underwent strong structural evolution over the same cosmic period .
a noticeable exception is the evolution of the size of massive ellipticals , which at @xmath9 were dominated by very compact galaxies , which had stellar density up to two order magnitudes higher than today s counterparts of similar mass , while at present such systems have essentially disappeared ( see @xcite ) .
also , it is interesting to note that @xcite find that at even higher redshifts , i.e. @xmath140 , the size distribution of star forming galaxies is significantly larger than that predicted from the spin parameter distribution observed in cosmological n body simulations , a marked difference from local disks which follow the simulation predictions very well .
between the rest - frame uv ( z - band ) and optical ( h - band ) [ bzk : left , sed : right ] .
the top , middle and bottom panels show @xmath4 , @xmath3 and @xmath7 , respectively .
blue crosses and red circles represent the sbzks ( star - forming ) and pbzks ( passive ) , respectively , and a dotted black line in each panel shows a linear correlation .
the inset in each panel show the distribution of the fractional differences ( @xmath141 ) of the parameters in the two rest - frame bands defined as @xmath142/m_{20}(z)$ ] .
negative @xmath143 and positive @xmath144 means that parameters are bigger in the z - band in each plot.,width=340,height=604 ] it is important to compare the rest
frame uv and optical morphologies , since the former traces the spatial distribution of star formation , and thus contains information on how , and where , galaxies grow in mass and evolve morphologically , while the latter traces the structure of their stellar components . with the candels images we can study the relation between rest - frame optical ( 5300 ) and uv ( 2800 ) morphology with a large sample . since non - parametric measures can vary systematically with the psf and pixel size of the images as the resolution decreases ( lotz et al .
2004 ; law et al . 2012 ) , we made a version of the acs z band image which is psf matched to the wfc3 h band one ( by convolution with an ad hoc kernel ) and which we have re - binned to the same pixel scale of 0.06" / pixel . to make a meaningful comparison
, we use the same segmentation map and the elliptical petrosian radius " estimated from the h - band image and apply it to the psf matched z - band image to measure the @xmath145 and @xmath7 in the rest - frame uv . in figure
[ fig : zh ] we compare the @xmath110 and @xmath7 of the bzk and sed samples in the uv and optical rest - frames ( passive ( pbzk ) and star - forming ( sbzk ) galaxies in red and blue , respectively ) .
what we find from our measurements is that all three parameters are different between the h- and z - bands in general .
first , the h - band derived @xmath3 is higher for both passive ( pbzks ) and star - forming ( sbzks ) galaxies at @xmath8 .
the values of @xmath4 at the two wavelengths are well correlated for both populations , but generally the z - band measurements have slightly higher values . in particular , we find that the scatter increases as @xmath4 increases .
this means that galaxies are clumpier in the z - band and the difference in @xmath4 between the two bands is bigger in the case of multi - clump structures ( e.g. higher @xmath4 ) .
the @xmath7 values from the h - band and z - band for star - forming ( sbzk ) galaxies are well correlated , while most of the passive ones ( pbzks ) in the z - band have higher @xmath7 value than in the h band .
we compute the fractional differences of the three parameters between optical and uv defined as @xmath142/m_{20}(z)$ ] , shown in the insets of figure [ fig : zh ] to check for offsets from the linear correlation .
negative @xmath4 , positive @xmath3 and @xmath7 imply that parameters in the z - band are higher than the ones in the h - band .
overall , in the rest - frame uv , galaxies appear to have higher @xmath7 and @xmath4 and lower @xmath3 values than in the rest - frame optical since observations in the rest - frame uv show more fragmented structures than the rest - frame optical , especially for star - forming galaxies .
we additionally find that the passive galaxies ( pbzks ) in the rest - frame optical tend to have a higher @xmath3 and lower @xmath7 and @xmath4 than in the rest - frame uv because the rest
frame optical light from old stellar populations is typically more concentrated than that from younger stellar populations , consistent with the result from @xcite that the inner region of passive galaxies at @xmath8 have a redder color gradient .
a similar trend was noted by @xcite , who found that the median galaxy size and @xmath4 are reduced ( less clumpy ) , while @xmath3 and @xmath5 increase from rest - frame 2800 and u band to the optical , using star - forming galaxies at @xmath146 . and
labels indicate the redshifts ( @xmath147 ) and morphological parameters .
star - forming galaxies show a variety of morphologies , while all passive galaxies show bulge like structures .
the morphology between z and h - band of passive galaxies are almost identical , but this is not the case for star - forming ones . , width=359,height=491 ] sample images of galaxies in both rest - frame optical and uv are shown in figure [ fig : mont ] to visually present the morphological differences between passive and star - forming galaxies and to see how nonparametric measures are correlated with the visual classifications .
we have selected 16 galaxies ( eight pbzks and eight sbzks ) included in both bzk and sed samples . these images have relatively smooth and regular morphologies in both bands .
the eight images for each spectral type are sorted by their h - band magnitude , with the brightest one located at the bottom and decreasing upward , and non - parametric statistics and redshifts of each galaxy labeled .
this figure illustrates the good correspondence between the measured parameters and the visual morphologies of the galaxies .
most sbzks are extended , and exhibit a broad range of morphologies , from isolated systems with a central bulge and galaxies with bulge and disk components , to irregular systems with multi - clumps .
the pbzks have morphologies similar to those of present day spheroids , bulges and ellipticals .
visually , the sbzks appear clumpier in the rest - frame uv compared to optical , showing a dependence on wavelength with the exception of a few isolated cases , while pbzks look very similar in both wavelengths .
overall , visual inspection shows that the morphological types in both bands are generally similar , in agreement with @xcite who found that passive galaxies ( @xmath148 ) have a weak morphological k - correction " , with size being smaller in the rest - frame optical than in the uv .
however , the comparison with non - parametric measures show that galaxies in the rest - frame uv are somewhat clumpier than rest - frame optical for both galaxy populations
. for star - forming galaxies at @xmath149 , @xcite and @xcite measured the internal color dispersion ( icd ) , and found that the morphological differences between the rest - frame uv and optical are typically small .
however , the argument that the majority of icds for star - forming galaxies are larger than those for passive galaxies @xcite is not consistent with our finding of relatively large offset for pbzks in figure [ fig : zh ] .
since most of our pbzks ( @xmath150 ) are massive ( @xmath151 ) , one possibility is that high mass galaxies tend to exhibit greater morphological differences with large icd @xcite .
furthermore , pbzks are typically brighter and rather compact at rest - frame optical wavelengths , which results in higher @xmath3 and lower @xmath4 and @xmath7 than in the rest - frame uv . on the other hand
, @xcite found a strong dependence of the morphology on wavelength in a visual study of 16 massive galaxies at @xmath8 .
in this section we briefly compare our measures of the morphologies of the mix of galaxy populations in the redshift range @xmath62 to the predictions of theoretical models .
in particular , we discuss the evidence that the bimodal distribution of galaxies in the color - mass ( or luminosity ) diagram , namely the `` red sequence '' and `` blue cloud '' , has already started to appear at @xmath8 , and compare it with existing measures at lower redshift .
the reliability of our non - parametric measures , @xmath3 , @xmath4 and @xmath7 and their performance in quantifying the morphology of galaxies at @xmath8 , especially for less massive ones , are also discussed .
lastly , we discuss the comparison between bzk and sed selected samples . our analysis of the various morphological indicators , both parametric and non - parametric ones , as well as a simple visual inspection ( e.g. figure [ fig : mont ] ) , have shown that star - forming galaxies ( sbzks ) exhibit a broad variety of morphological structures , ranging from galaxies with a predominant disk morphology and varying degree of bulge to disk ratio to irregular ( clumpy ) structures to very compact and relatively regular galaxies .
generally , the mix of star forming galaxies at @xmath9 looks rather different from its counterpart in the local universe , showing a much larger fraction of irregular and disturbed morphologies , especially among massive and luminous galaxies , although bright galaxies that closely resemble local spirals are also observed .
we do observe luminous , clumpy galaxies whose light profile is consistent with a disk ( of course , we do not have dynamical information on these galaxies ) and whose overall morphology is in broad qualitative agreement with the theoretical predictions of violent disk instability ( vdi , dekel et al .
2009a ) as seen in high resolution hydrodynamic cosmological simulations @xcite .
these simulations show that galaxy disks are built up by accretion of continuous , intense , cold streams of gas that dissipate angular momentum in a thick , toomry
unstable disk , where star forming clumps form .
subsequently , clumps migrate toward the center and edge , giving rise to bulges and pseudo bulges
. observations of the star formation rate , stellar mass and age of the clumps , as well as their average radial dependence relative to the center of the galaxies are also broadly consistent with this scenario ( e.g. guo et al . 2011 ) .
on the other hand , passive galaxies ( pbzks ) are mostly spheroidal - like , comparatively more regular and compact structures , a fact that has been consistently observed in previous works @xcite .
it is important to keep in mind , however , that there is scant spectroscopic information on the dynamical properties of these galaxies , namely whether they are primarily supported by velocity dispersion or by rotation .
while the modicus of spectroscopic observations currently available @xcite is certainly consistent with the high redshift passive galaxies being spheoroids , a significant or even dominant contribution by rotation can not be ruled out given the limited angular resolution of the existing spectra , and some have indeed proposed that a significant fraction or maybe even most @xcite of these galaxies are actually compact rotating disks . from the theory point of view , there are three distict scenarios for the formation of the compact spheroids , namely major mergers , multiple minor mergers or the migration of clumps driven by violent disk instabilities @xcite to the disks center , building a massive bulge .
cosmological hydrodynamical simulations indicate that these processes , operating alone or in combination , can form compact spheroids . in most cases , the inner parts of the compact spheroids formed by vdi are rotating , and the outer parts are non - rotating , formed mostly by minor mergers .
while the observed morphological properties of galaxies certainly include cases that are broadly consistent with these scenarios , it is clear that to make progress comparisons between the dynamical properties of the galaxies and the predictions of the simulations are necessary .
these , however , require spectroscopic observations with sensitivity and spatial resolution that are not currently available . in general very compact and massive galaxies
are thought to be the result of a highly dissipative process , either a major wet merger ( e.g. wuyts et al . 2010 ) or direct accretion of cold gas .
accretion of cold gas from the inter
galactic medium @xcite can lead to the formation of compact , massive galaxies , either via vdi in a compact disk @xcite or via direct accretion of the gas traveling directly to the galaxy center rapidly and forming stars in
situ @xcite .
quenching of the star formation subsequently takes place late when the supply of gas is halted .
the simulations suggest that cold accretion is naturally interrupted in dark matter halos more massive @xmath152 m@xmath23 once the shocked halo gas become too hot to allow the cold flows to penetrate the halo before they themselves get shock - heated @xcite , leading the formation of a massive compact passive galaxy .
additional feedback mechanisms from star formation itself @xcite and agn @xcite can also contribute to suppress the accretion of cold gas .
as mentioned earilier , @xcite have studied the morphologies of massive galaxies at @xmath153 in the candels - udf field using @xmath2 and bulge+disc models , finding that at @xmath12 massive galaxies are dominated by disk - like structures and 25 - 40% of quiescent galaxies have disk - dominated morphologies .
following their classification of disks , namely @xmath115 ( even though they also use bulge- to - total h - band flux ratio ) , we find that about 50% of our passive galaxies
( with @xmath154 ) have exponential light profiles with @xmath155 , i.e consistent with exponential disks .
these roughly classified passive disks have mostly @xmath156 , @xmath157 , suggesting that they generally are not clumpy structures , and their morphology is characterized by a central bright nucleus surrounded by low surface brightness features .
the presence of passive disks seems inconsistent with models where galaxy morphology transforms from a disk structure into a bulge followed by quenching of star formation as the galaxy evolves .
the existence of passive disks is , however , predicted by hydrodynamic simulations @xcite , which show that these structures form when cold gas inflows are halted , thus quenching star formation without the transformation of morphology .
for example , @xcite argue that the morphological properties and volume density of massive , compact passive galaxies at @xmath9 and those of compact star forming galaxies at @xmath158 are generally consistent with such a scenario . in this study , to the extent that passive and star forming galaxies can effectively be identified by means of broad band colors , e.g. either the bzk selection criteria or via sed fitting , we have shown that passive ( pbzk ) and star - forming ( sbzk ) galaxies occupy regions of the rest - frame u - v and stellar mass diagram ( e.g. figure [ fig : cm ] ) that are essentially the same as the `` red sequence '' and `` blue cloud '' observed in the local universe ( e.g. blanton et al . 2003 ,
bell et al . 2004 ) .
passive galaxies ( pbzks ) are intrinsically red and massive , whereas , star - forming galaxies ( sbzks ) are generally bluer and have lower mass than passive ( pbzk ) ones ( with the exception of about 7% red massive sbzks ) .
the majority of the exceptions are massive dusty star - forming galaxies , and are largely redder than low mass star - forming galaxies ( sbzks ) at @xmath8 .
thus , they can contaminate the red sequence sample by a significant fraction , if selected based only on a single rest - frame ( u - v ) color @xcite , since most of the uv emission from high - redshift star formation is at least somewhat obscured by dust .
an intriguing property of this color - mass diagram is the lack of passive galaxies with mass @xmath159 m@xmath23 . to a minor extent , this is the result of incompleteness , since passive galaxies with lower mass , and hence luminosity , become harder to detect .
from the simulation using model galaxies in the goods - s deep field mosaics , we confirm that the early - type galaxies with de vaucouleurs " light profile ( @xmath2 index = 4 ) are 90% complete with @xmath160 .
clearly , however , such a small incompleteness alone can not explain the lack of low
mass passive galaxies at @xmath9 , and in fact such low mass galaxies are actually detected by the sed selection , as shown in the right panel of figure [ fig : cm ] which illustrates how massive red galaxies are rarely star - forming , and more actively star - forming galaxies are bluer and have lower masses .
furthermore , obscured star forming galaxies in the same redshift range and with similar rest
frame ( u - v ) colors are detected in significantly larger number even at lower masses , as the right panel of the figure shows .
this fact strongly suggests that the quenching of star - formation at this epoch is tightly correlated with the mass of the galaxies , with the most massive ones being significantly more likely to cease their star formation activity . at mass
@xmath161 m@xmath23 , galaxies appear much less likely to stop forming stars , a fact that has been observed by other groups .
for example , kauffmann et al.(2003 ) and bell et al.(2007 ) observed that at @xmath15 the stellar mass value of about @xmath162 appears to be the transition mass point between galaxies that belong to the blue cloud ( younger stellar populations ) and those of the red sequence .
there is evidence that this transition mass between quenched and star - forming galaxies has evolved significantly over cosmic time @xcite further supporting the downsizing scenario whereby more massive galaxies appear to quench first , and subsequently lower mass galaxies quench later . at high redshift , quenching appears to depend on galaxy stellar mass , perhaps through some internal process that is tied to the total mass of the galaxy of which the stellar one is a good proxy in passive systems , whereas later , environmental processes can contribute to galaxy quenching and can affect lower - mass galaxies ( e.g. peng et al 2010 , peng et al 2012 ) .
this process effectively builds up the lower - mass end of the red sequence over time by quenching lower - mass star - forming galaxies later when environmental processes become more influential .
our observed deficiency of lower - mass passive galaxies at @xmath8 is consistent with this scenario and the implied mechanisms by which galaxies quench their star - formation .
results from other deep extragalactic surveys have provided constraints on the buildup of stellar mass locked up in the red sequence , by studying how the bimodality of galaxy properties has changed over cosmic time .
for example , using the combo-17 and deep2 surveys , @xcite have studied the evolution in the rest - frame color bimodality of galaxy samples out to @xmath10 , finding evidence that the buildup of the red sequence must be accounted for by a combination of merging of galaxies already on the red sequence , as well as migration of star - forming galaxies that have quenched . recently
, brammer et al ( 2011 ) extended the study of rest - frame color bimodality in galaxies to higher redshift , showing that star - forming and passive galaxies are still robustly separated in color over the redshift range @xmath163 , and coming to the similar conclusion that the growth of the red sequence must come from both merging and migration , particularly for galaxies above the apparent quenching threshold , @xmath164 . in this context , a possibility is that that the morphological bimodality we have observed in this study may imply that some degree of morphological transformation must accompany the quenching of star - forming galaxies at @xmath165 , if they are to match the properties of the red sequence after quenching . regardless of the dominant mechanisms building the red sequence , and weather or not the dominant mechanisms evolve with redshift , our result suggests that the process was already underway at redshift 2 .
in other words , at this epoch , the formation of the hubble sequence is already underway .
further detailed study of the evolution of morphological properties of galaxies as a function of mass and star - formation properties will be required to identify the specific mechanisms contributing to the growth of the red sequence .
we mainly use the @xmath3 , @xmath4 and @xmath7 to study the morphologies of galaxies at @xmath8 .
those non - parametric measures are widely used to study galaxy morphologies at high redshift , especially for large samples @xcite .
we investigate the robustness of the @xmath3 , @xmath4 and @xmath7 parameters in relation to the snr in appendix a and show that any differences in the estimated parameters for the same galaxies observed in the goods - s and udf images , whose only difference is the vastly different total exposure time , are relatively small .
most of the galaxies in our samples ( above @xmath71 of bzk and @xmath122 of sed selected galaxies ) have reliable morphological measurements with @xmath106 at @xmath8 .
moreover , the good correspondence between those parameters and visual inspection ( in figure [ fig : mont ] ) , as well as model - dependent parameters indicates that our @xmath3 , @xmath4 and @xmath7 measures are not biased by low signal - to - noise .
we note that @xmath2 index alone is generally not sufficient to quantify the morphology of low mass galaxies since we find no correlation between @xmath2 index and non - parametric measures in the lower mass systems ( @xmath166 ) .
also non - parametric measures more effectively characterize the morphology of irregular galaxies ( lpm04 ) .
therefore , it is crucial to use non- parametric diagnostics instead of , or at least in parallel with , the commonly used @xmath2 profile to study the morphology of lower mass galaxies and to explore the origin of the hubble sequence at @xmath8 , and epoch when many galaxies appear irregular . both the bzk and sed samples show very similar morphological distributions in all the analysis done here .
the comparison of the two samples confirms that they are similarly representative of the mix of bright galaxies at @xmath8 .
we additionally compare the distribution of morphological parameters for the 136 spectroscopically confirmed bzk galaxies ( 8 pbzks and 128 sbzks , specz sample ) to the parent distribution ( bzk sample ) .
as expected , the relative distributions of @xmath3 , @xmath4 and @xmath7 are similar to our parent sample , and the median values of each parameter are almost the same , with the exception of @xmath3 in the case of sbzks . the average @xmath3 value of bzks in the specz sample is slightly higher , since over @xmath122 of the specz sample are bright and massive ( @xmath167 ) galaxies , which tend to have higher @xmath3 .
thus , in conclusion , the very similar results derived from both samples proves the effectiveness of the bzk selection criteria in sampling the full diversity of the mix of massive galaxies at @xmath9 , at least as far as the morphological properties of relatively massive galaxies are considered .
the bzk selection will be particularly effective , for example , in the other three candels fields where the broad band photometry is neither as deep or as broad in wavlength as in the two goods fields .
in this paper we have explored general trends between galaxy morphology and broad band spectral types at @xmath8 using the _ hst_/wfc3 h band images taken in the goods south field as part of the candels survey , in combination with the existing goods acs z - band images , as well as sensitivity matched images at other wavelngths that are part of the goods data products @xcite . combining the deep and high - resolution nir data to optical data , we are able to study the dependence of morphologies on wavelengths and expand the scope of previous studies of galaxy morphologies at the same redshift @xcite with significantly larger sample size and lower mass limit ( @xmath78 ) .
the galaxies of our primary sample are selected in the redshift range @xmath131 to cover a broad range of spectral types ( star formation properties and dust obscuration ) using sed fitting to spectral population synthesis models ( sed sample ) . for comparative reasons we also selected galaxies using the bzk creterion which culled galaxies in the same redshift range and with essentailly identical spectral properties , modulo a large contamination from interlopers and agn .
analyses of the two samples show consistent results suggesting that the bzk and sed selection criteria are equivalent in sampling the mix of spectral types at @xmath9 .
we investigate the rest - frame optical morphologies using five non - parametric approaches , mainly @xmath3 , @xmath4 and @xmath7 in addition to @xmath5 and @xmath6 , and two model - dependent parameters obtained by fitting @xmath2 profiles , namely @xmath119 and @xmath27 .
the major findings of this study are presented below . 1 . in the rest - frame ( u - v )
color and mass diagram , our sample clearly separates into red massive passive galaxies with low ssfr and blue star - forming ones with less massive , high ssfr occupying the same regions in the color - mass diagram as the galaxies observed in the local universe .
we find that galaxies with different spectral types are distinctly classified morphologically as two populations , especially for massive systems ( @xmath168 ) : 1 ) star - forming galaxies are heterogeneous , with mixed features including bulges , disks , and irregular ( or clumpy ) structures , with relatively low @xmath3 , @xmath119 and high @xmath4 , @xmath7 ; 2 ) passive galaxies are spheroidal - like compact structures with higher @xmath3 , @xmath119 and lower @xmath4 and @xmath7 .
generally , the sizes of star - forming galaxies are larger than passive ones , even in massive systems , but some have very compact morphologies , with @xmath169 . we confirm using a variety of measures that star formation activity is correlated with morphology at @xmath8 , with the passive galaxies looking similar to local passive ones although smaller , while star - forming galaxies show considerably more mophological diversity than massive star - forming galaxies on the hubble sequence today .
we show that the morphological analysis only using the @xmath2 index is not sufficient to charaterize differences in morphologies especially for lower mass galaxies ( @xmath170 ) .
therefore we conclude that it is important to use non - parametric measures to investigate the morphologies of high redshift galaxies in a broad range of stellar masses . in this study ,
the combination of large samples with a suite of morphological diagnaostics , both parametric and non parametric ones , as well as visual inspections , gives us a significantly improved description of the state of galaxy morphologies at @xmath8 and its correlations with the spectral type , i.e. mostly the star
formation activity , expanding the significance and the scope of previous studies which were based on much smaller samples and only massive galaxies at the same epoch @xcite .
4 . generally , @xmath8 galaxies show a similar trend in morphologies with those measured from the redshifted images of local galaxies , even though many of the star - forming galaxies have @xmath4 values higher than seen for galaxies in the local sample .
the passive galaxies at @xmath8 have @xmath3 and @xmath4 values that are much more similar to those observed for the present day e / s0 galaxies .
comparison of visual images between the rest - frame optical and near uv show that the morphological k - correction is generally weak , however , the comparison with non - parametric measures indicates that galaxies observed in the rest - frame uv are slightly clumpier , with lower @xmath3 and higher @xmath4 and @xmath7 , than rest - frame optical .
taken all together , our results show that the correlations between morphology as traced by a suite of common diagnostics , and broad band uv / optical spectral types of the mix of relatively massive galaxies ( i.e. @xmath171 m@xmath23 ) at @xmath9 are quantitatively and qualitatively similar to those observed for their counterparts in the local universe .
we interpret these results as evidence that the backbone of the hubble sequence observed today was already in place at @xmath9 .
this work is based on observations taken by the candels multi - cycle treasury program with the nasa / esa hst , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas5 - 26555 .
the reliability of model - independent parameters has been tested and discussed by many previous studies @xcite .
these parameters are robust for large bright sources ( with high signal - to - noise ratio ) in general , but it can be unreliable for small faint sources . in particular , the @xmath3 value is very sensitive to the signal - to - noise ratio ( lpm04 and lisker et al .
our galaxy sample goes deeper ( @xmath172 ) , and to lower stellar mass limits ( @xmath173 ) than other works that study morphologies at @xmath8 ( e.g. kriek et al .
2009 , szomoru et al .
2011 , wang et al .
therefore , it is crucial to understand how noise and other limitations affect parameters we use here .
we test the dependence of the parameters , @xmath145 and @xmath7 , on the signal - to - noise ratio , i.e. , the depth of an image .
hst / udf wfc3 imaging data , which differ only in exposure time from goods - s data , is an ideal comparison dataset for investigating the effect of the different image depth on these parameters .
previously , lisker et al .
2008 tested the dependence of signal - to - noise on @xmath145 in the udf and goods - s using @xmath174-band images . to investigate how this affects our measurements with the nir wfc3 images
, we use the udf and shallower goods - s f160w ( h - band ) data ( both with 60mas pixel scales ) and select 959 sources in the udf image using goods - s source catalog .
then , we estimate @xmath73 from the goods - s h - band image , and use it to measure the parameters for both the udf and goods - s images .
the final sample is 520 objects which have stellarity @xmath175 , after excluding sources for which the number of pixels within @xmath73 is less than 28 pixels ( same exclusion with lisker 2008 ) .
signal - to - noise ratio per pixel ( @xmath90 ) is defined as @xmath176 where , @xmath177 indicates the total number of pixels within the @xmath73 , @xmath178 is the total image exposure time .
@xmath179 and @xmath180 are the pixel values in the drizzled image and rms image , respectively .
the strong dependence of @xmath181 on the @xmath90 are shown in figures [ fig : gini_sn ] , [ fig : m20_sn ] , and [ fig : psi_sn ] , respectively .
the two top panels in figure [ fig : gini_sn ] present how values of @xmath3 vary with @xmath90 in the goods - s ( red circles ) and udf ( blue circles ) .
obviously , the udf has a higher @xmath90 ( difference of depth is about 1.8 mag ) , but the overall trend of @xmath3 looks similar to that for goods - s .
both @xmath3 values overlap almost exactly when we move the gini of the udf to the relatively small median difference ( -0.03 ) shown in the bottom left panel .
however , the @xmath3 of galaxies with different exposure times is dependent on the @xmath90 as shown in the bottom right panel .
the difference of in @xmath3 between the udf and goods - s decrease as @xmath90 increases . @xmath4 and
@xmath7 also follows the same trend and distribution as @xmath3 as shown in figure [ fig : m20_sn ] and [ fig : psi_sn ] , but with small median differences of about 0.01 and 0.11 , respectively .
consequently , the model - independent parameters depend on signal - to - noise ratio , but the difference between two fields with different exposure times are less significant than @xcite .
lpm04 find that @xmath3 and @xmath4 are reliable for galaxy images with @xmath182 ( vertical dashed line on the bottom right panel in each figure ) .
given this threshold in reliability , we investigate how this may affect our results by measuring the @xmath90 of each galaxy .
we find over 90% of bzk sample and 70% of sed sample having @xmath183 , which excludes some faint star - forming galaxies and passive galaxies with low @xmath3 from our sample .
repeating our analysis with these faint objects removed does nt change the results .
therefore , we confirm that the results from our measurements did not suffer from this @xmath184 effect .
abazajian , k. , et al .
2003 , , 126 , 2081 abraham , r. g. , tanvir , n. r. , santiago , b. x. et al . 1996 , , 279 , l47 , r. g. and van den bergh , s. and nair , p. 2003
, , 588 , 218 abraham , r. g. , nair , p. , macarthy , p. j. et al .
2007 , , 669 , 184 andrae , r. , jahnke , k. , melchior , p. 2011
, , 411 , 385 baldry , i. k. , glazebrook , k. , brinkmann , j. et al .
2004 , , 600 , 681 baldry , i. k. , balogh , m. l. , bower , r. g. et al .
2006 , , 373 , 469 bell , e. f. , mcintosh , d. h. and barden , m. et al .
2004 , , 600l , 11 , e. f. and zheng , x. z. and papovich , c. et al .
2007 , , 633 , 834 bell , e. f. , van der wel , a. , papovich , c. et al .
2012 , , 753 , 167 , e. and arnouts , s. 1996 , , 117 , 393 , y. and dekel , a. 2003 , , 345 , 349 blanton , m. r. , hogg , d. w. , bahcall , n. a. et al .
2003 , , 594 , 186 , m. r. and hogg , d. w. and bahcall , n. a. et al .
2003 , , 592 , 819 bond , n , a. and gawiser , e. and koeoemoer , a. m. 2011 , , 729 , 48 brammer , g. b. , whitaker , k. e. , van dokkum , p. g. et al .
2009 , , 706 , l173 brammer , g. b. , whitaker , k. e. , van dokkum , p. g. et al .
2011 , , 739 , 24 , v. a. and dunlop , j. s. and cirasuolo , m. et al .
2012 , , 427 , 1666 bruzual , g. , charlot , s. 2003 , , 344 , 1000 , k. and treu , t. and ellis , r. s. 2007 , , 665l , 5 calzetti , d. , armus , l. , bohlin , r. et al . 2000 , , 533 , 682 , e. and carollo , c. m. and oesch , p. a. et al .
2011 , , 743 , 146 , p. and guzzo , l. and franceschini , a. et al .
2007 , , 172 , 270 , p. and cimatti , a. and kurk , j. et al .
2008 , , 483 , 39 cassata , p. , giavalisco , m. , guo , y. et al .
2010 , , 714 , l79 cassata , p. , giavalisco , m. and guo , y. et al .
2011 , , 743 , 96 cassata , p. , giavalisco , m. and guo , y. et al .
2013 , , in press ceverino , d. , dekel , a. , bournaud , f. 2010 , , 404 , 2151 , e. and faber , s. m. and koo , d. c. et al .
2012 , arxiv1210.4173 , a. and daddi , e. and mignoli , m. et al .
2002 , , 381 , l68 , a. and daddi , e. and renzini , a. et al .
2004 , , 430 , 184 conselice , c. j. and bershady , m. a. , jangren , a. 2000 , , 529 , 886 conselice , c. j. 2003 , , 147,1 conselice , c. j. , rajgor , s. , myers , r. 2008 , , 386 , 909 daddi , e. , cimatti , a. , renzini , a. et al .
2004 , , 617 , 746 daddi , e. , renzini , a. , pirzkal , n. et al .
2005 , , 626 , 680 daddi , e. , dickinson , m. , morrison , g. et al .
2007 , , 670 , 156 , a. and birnboim , y. 2006 , , 368 , 2 , a. and birnboim , y. 2008 , , 383 , 119 dekel , a. , sari , r. , ceverino , d. 2009a , , 703 , 785 dekel , a. , birnboim , g. , engel , g. , et al . 2009b , nature , 457 , 451 , a. m. and moustakas , j. and tremonti , c. a. et al .
2012 , , 755l , 26 dickinson , m. 2000 , philos .
london a , 358,2001 driver , s. p. , allen , p. d. , graham , a. w. et al .
2006 , , 368 , 414 , s. m. and willmer , c. n. a. and wolf , c. et al .
2007 , , 665 , 265 , g. and kong , x. and chen , y. and lin , x. 2012 , , 751 , 109 fioc , m. , rocca - volmerange , b. 1997 , , 326 , 950 , m. and van dokkum , p. g. and schreiber , n. m. f. et al .
2008 , , 688 , 770 frei , z. , guhathakurta , p. , gunn , j. , & tyson , j. a. 1996 , /aj , 111 , 174 genzel , r. , newman , s. , jones , t. et al .
2011 , , 733 , 101 , m. and steidel , c. c. and macchetto , f. d. 1996a , , 470 , 189 , m. and livio , m. and bohlin , r. c. et al . 1996b , , 112 , 369 giavalisco , m. , gerguson , h. c. , koeoemoer , a. m. et al .
2004 , , 600 , 93 grogin , n. a. , kocevski , d. d. , faber , s .
2011 , , 197 , 35 guo , y. , giavalisco , m. , cassata , p. et al .
2011 , , 735 , 18 , y. and giavalisco , m. and cassata , p. et al .
2012 , , 749 , 149 guo , y. et al .
2013 , , in press huang , k .- h . ,
ferguson , h. c. , ravindranath , s. et al .
2013 , arxiv1301.4443 , m. and simard , l. and faber , s. m. et al .
2002 , , 571 , 136 , p. h. and naab , t. and ostriker , j. p. 2012
, , 754 , 115 , g. and heckman , t. m. and white , s. d. m. et al .
2003 , , 341 , 54 , d. and katz , n. and weinberg , d. h. et al .
2005 , , 363 , 2 koekemoer , a. m. , faber , s. m. , ferguson , h. c. et al .
2011 , , 197 , 36 kriek , m. , van dokkum , p. g. , franx , m. et al .
2009 , , 705 , l71 laidler , v. g. , papovich , c. , grogin , n. a. et al .
2007 , pasp , 119 , 1325 law ,
d. r. , steidel , c. c. , erb , d. k. et al .
2007 , , 656 , 1 law , d. r. , steidel , c. c. , shapley , a. e. et al .
2012 , , 745 , 85 , l. and dickinson , m. and jian , h .- y .
2012 , , 756 , 71 lisker , t. 2008 , , 179 , 319 , j. d. and koo , d. c. and guzman , r. et al .
1997 , , 481 , 673 lotz , j. m. , primack , j. , madau , p. 2004
, , 128 , 163 lotz , j. m. , madau , p. , giavalisco , m. et al .
2006 , , 636 , 592 madau , p. 1995
, , 441 , 18 mobasher , b. , idzi , r. , bentez , n. et al .
2004 , , 600l , 167 nair , b. p. , abraham , r. g. , 2010 , , 186 , 427 , m. and renzini , a. and carollo , m. et al .
2012 , , 755 , 260 , r. a. and heckman , t. m. and schiminovich , d. et al .
2010 , , 710 , 979 , c. and giavalisco , m. and dickinson , m. et al .
2003 , , 598 , 827 papovich , c. , dickinson , m. , giavalisco , m. et al .
2005 , , 631 , 101 peng , c. y. , ho , l. c. , impey , c.d . , & rix , h .- w .
2002 , , 124 , 266 , y .- j . and
lilly , s. j. and kova , k. et al . 2010 , , 721 , 193 , y .- j . and
lilly , s. j. and renzini , a. et al .
2012 , , 757 , 4 petrosian , v. 1976 , , 209 , l1 , s. and giavalisco , m. and ferguson , h. c. et al .
2006 , , 652 , 963 reddy , n. a. , erb , d. k , steidel , c. c. et al .
2005 , , 633 , 748 , n. a. and steidel , c. c. 2009 , , 692 , 778 , n. a. and erb , d. k. and pettini , m. et al .
2010 , , 712 , 1070 , a. 2006 , , 44 , 141 roberts , m. s. , haynes , m. p. 1994
, ara&a , 32 , 115 , d. and lilly , s. j. and crampton , d. et al . 1995 , , 451l , 1 shapley , a. , steidel , c. c. , adelberger , k. l. et al .
2001 , , 562 , 95 , v. 2005 , , 364 , 1105 , c. c. and giavalisco , m. and dickinson , m. et al .
1996 , , 112 , 352 , d. and franx , m. and bouwens , r. j. et al .
2011 , , 735 , l22 toft , s. , franx , m. , van dokkum , p. et al .
2009 , , 705 , 255 , i. and feulner , g. and goranova , y. et al .
2006 , , 373 , 36 , i. and conselice , c. j. and bundy , k. et al .
2007 , mras , 382 , 109 , t. and fritz , j. and poggianti , b. m. et al .
2010 , , 712 , 226 , a. and rix , h .- w . and wuyts , s. et al .
2011 , , 730 , 38 , a. and bell , e. f. and hussler , b. et al .
2012 , , 230 , 24 van dokkum , p. g. , franx , m. , kriek , m. et al .
2008 , , 677 , l5 , p. g. and brammer , g. and fumagalli , m. et al .
2011 , , 743l , 15 wang , t. , huang , j .,- s . ,
faber , s. m. et al .
2012 , , 752 , 134 , r. j. and quadri , r. f. and franx , m. et al .
2009 , , 691 , 1879 williams , r. j. , quadri , r. f. , franx , m. et al .
2010 , , 713 , 738 williams , c. c .. giavalisco , m. , cassata p. , et al .
2013 in submitted , r. a. and cohen , s. h. and hathi , n. p. et al .
2011 , , 193 , 27 , s. and labb , i. and franx , m. et al .
2007 , , 655 , 51 , s. and cox , t. j. and hayward , c. c. et al .
2010 , , 722 , 166 , s. and frster schreiber , n. m. and van der wel , a. et al .
2011 , , 742 , 96 , s. and frster schreiber , n. m. and genzel , r. et al .
2012 , , 753 , 114 , s. and ohta , k. and yabe , k. 2012 , , 761 , 19 | we discuss the state of the assembly of the hubble sequence in the mix of bright galaxies at redshift @xmath0 with a large sample of 1,671 galaxies down to @xmath1 , selected from the hst / acs and wfc3 images of the goods south field obtained as part of the goods and candels observations .
we investigate the relationship between the star formation properties and morphology using various parametric diagnostics , such as the @xmath2 light profile , gini ( @xmath3 ) , @xmath4 , concentration ( @xmath5 ) , asymmetry ( @xmath6 ) and multiplicity ( @xmath7 ) parameters .
our sample clearly separates into massive , red and passive galaxies versus less massive , blue and star forming ones , and this dichotomy correlates very well with the galaxies morphological properties . star
forming galaxies show a broad variety of morphological features , including clumpy structures and bulges mixed with faint low surface brightness features , generally characterized by disky - type light profiles .
passively evolving galaxies , on the other hand , very often have compact light distribution and morphology typical of today s spheroidal systems .
we also find that artificially redshifted local galaxies have a similar distribution with @xmath8 galaxies in a @xmath3@xmath4 plane .
visual inspection between the rest - frame optical and uv images show that there is a generally weak morphological k - correction for galaxies at @xmath8 , but the comparison with non - parametric measures show that galaxies in the rest - frame uv are somewhat clumpier than rest - frame optical .
similar general trends are observed in the local universe among massive galaxies , suggesting that the backbone of the hubble sequence was already in place at @xmath9 . |
two main kinds of representation for belief change have been suggested in the literature . in the agm theory @xcite ,
belief change is considered as a change of _ belief sets _ that are taken to be deductively closed theories . according to the so - called base approach ,
however , a belief set should be seen as generated by some ( finite ) _ base _ ( see , e.g. , @xcite ) .
consequently , revisions of belief sets are determined on this approach by revisions of their underlying bases .
this drastically reduces the set of alternatives and hence makes this approach constructive and computationally feasible .
the agm theory and base approach constitute two seemingly incomparable representations for belief change , each with advantages of its own .
still , the framework of epistemic states suggested in @xcite allows to formulate both as different species of a single general representation .
the reformulation makes it possible , in particular , to reveal common constitutive principles behind them .
these common principles embody , however , also common shortcomings that lead to a loss of information in iterated changes .
this suggests epistemic states in their full generality as a justified representational alternative .
epistemic states could be primarily seen as a generalization of the agm models , namely , they form an abstract logical representation for belief change . though the base - oriented models can be translated into epistemic states , the translation loses some important advantages of such models , namely their inherent finiteness and constructivity .
hence it would be desirable to have a constructive representation for epistemic states that would preserve these features of base models while avoiding their problems .
fortunately , such an alternative representation has already been suggested in @xcite , and it amounts to using sets , or _ flocks _ , of bases . as we will show ,
a certain modification of the original flock models will provide us with a constructive representation for an important class of epistemic states and belief change processes in them .
the theory described below , is different from that suggested in @xcite , since the latter has turned out to be flawed in its treatment of expansions .
the shortcoming has created the need for shifting the level of representation from pure epistemic states used in the above papers to more general persistent epistemic states ( see below ) .
a common feature of the agm and base representations is a preference structure on certain subsets of the belief set . in the case of the agm paradigm
, it is a preference structure on the maximal subtheories of the belief set , while for the base representation it is a preference structure on subtheories that are generated by the subsets of the base .
the notion of an epistemic state , defined below , reflects and generalizes this common structure of the two representations . an _ epistemic state _
@xmath0 is a triple @xmath1 , where @xmath2 is a set of _ admissible belief states _ , @xmath3 is a preference relation on @xmath2 , while @xmath4 is a labelling function assigning a deductively closed theory ( called _ an admissible belief set _ ) to every admissible belief state from @xmath2 .
any epistemic state determines a unique set of propositions that are _ believed _ in it , namely , the set a propositions that hold in all maximally preferred admissible belief states .
thus , even if an epistemic state contains conflicting preferred belief states , we can still believe in propositions that hold in all of them .
still , the belief set associated with an epistemic state does not always constitute an admissible belief set by itself .
the latter will hold , however , in an ideal situation when an epistemic state contains a unique most preferred admissible belief state .
such epistemic states will be called _
both agm and base models correspond to determinate epistemic states .
nevertheless , non - determinate epistemic states have turned to be essential for an adequate representation of both nonmonotonic reasoning and belief change . in particular , the necessity of accepting non - determination arises most clearly in the analysis of contractions ( see below ) . in what follows
, we will concentrate on a special kind of persistent epistemic states that will provide a representation for a non - prioritized belief change .
an epistemic state will be called _ persistent _ if it satisfies persistence : : if @xmath5 , then @xmath6 . for persistent epistemic states , the informational content of admissible belief states
is always preserved ( persist ) in transitions to more preferred states .
this property reflects the idea that the informational content of an admissible state is an essential ( though presumably not the only ) factor in determining the place of this state in the preference structure .
persistent epistemic states are formally similar to a number of _ information models _ suggested in the logical literature , such as kripke s semantics for intuitionistic logic , veltman s data semantics , etc .
( see @xcite for an overview ) .
all such models consist of a partially ordered set of informational states .
the relevant partial order represents possible ways of information growth , so it is assumed to satisfy the persistence requirement by its very meaning . as will be shown
, persistent epistemic states constitute a smallest natural class of epistemic states that is closed under belief change operations that do not involve prioritization .
moreover , it is precisely this class of epistemic states that admits a constructive representation in terms of flocks of bases .
an epistemic state will be called _ pure _ if it satisfies pure monotonicity : : @xmath7 if and only if @xmath6 .
pure epistemic states is a special kind of persistent states for which the preference relation is determined solely by the informational content of admissible belief states .
accordingly , a pure epistemic state can be defined simply as a set of deductively closed theories , with the intended understanding that the relation of set inclusion among such theories plays the role of a preference relation . in other words , for pure epistemic states
, preference is given to maximal theories .
pure epistemic states have been used in @xcite as a basis for a foundational approach to belief change .
as we mentioned , however , the approach has turned out to be flawed , since it does not provide an adequate representation of belief expansions .
the present study stems from a more general approach to representing belief change suggested in @xcite .
assume that a belief set @xmath8 is generated by some base @xmath9 with respect to a certain consequence relation @xmath10 ( that is , @xmath11 ) .
this structure is representable as an epistemic state of the form @xmath12 , where admissible belief states are the subsets of @xmath9 , @xmath4 assigns each such subset its deductive closure , while @xmath3 is a preference relation on @xmath13 . in the simplest ( non - prioritized ) case ,
this preference relations is definable via set inclusion : @xmath14 iff @xmath15 . in the latter case
base - generated epistemic states are equivalent to pure epistemic states consisting of theories of the form @xmath16 , where @xmath17 ranges over the subsets of @xmath9 .
notice that any such epistemic state will be determinate , since it contains a most preferred theory , namely @xmath18 .
unfortunately , we will see later that base - generated epistemic states ( and bases themselves ) are arguably inadequate for representing belief contractions . a generalization of bases that overcomes this shortcoming has been suggested in @xcite and consists in using sets ( or ` flocks ' ) of bases . by a _ flock
_ we will mean an arbitrary set of sets of propositions @xmath19 , for @xmath20 .
such a flock can be considered as a collection of bases @xmath21 , and the following construction of the epistemic state generated by a flock can be seen as a natural generalization of base - generated epistemic states .
any flock generates an epistemic state @xmath22 defined as follows : * @xmath23 is a set of all nonempty sets @xmath17 such that @xmath24 , for some @xmath25 ; * @xmath26 , for each @xmath27 ; * @xmath14 holds iff @xmath15 .
as can be seen , flocks constitute a generalization of bases .
namely , any base @xmath9 can be identified with a singular flock @xmath28 .
as in our study , flocks were used in @xcite as a framework for belief change operations .
our subsequent results will be different , however .
the main difference can be described as follows .
let us say that two flocks are _ identical _ if they generate the same epistemic state .
now , let @xmath29 be a flock and @xmath30 a set of propositions such that @xmath31 , for some @xmath25 .
then it is easy to see that flocks @xmath29 and @xmath32 produce the same epistemic state , and consequently they are identical in the above sense .
this shows that a flock is determined , in effect , by its inclusion - maximal elements . according to @xcite , however , the above two flocks are distinct , and hence the validity of propositions with respect to a flock
is determined , in effect , by _
minimal _ sets belonging to the flock .
this makes the resulting theory less plausible and more complex than it could be .
the above feature , though plausible by itself , gives rise , however , to high sensitivity of flocks with respect to the syntactic form of the propositions occurring in them .
let us consider the flock @xmath33 , where @xmath34 is logically equivalent to @xmath35 . replacing @xmath34 with @xmath35 , we obtain a different flock @xmath36 , which is identical to @xmath37 .
note that @xmath38 is believed in the epistemic state generated by the latter flock , though only @xmath35 is believed in the epistemic state generated by the source flock @xmath29 .
the above example shows that flocks do not admit replacements of logically equivalent propositions , at least in cases when such a replacement leads to identification of propositions with other propositions appearing elsewhere in the flock .
it should be kept in mind , however , that the epistemic state generated by a flock is a syntax - independent object , though purely syntactic differences in flocks may lead to significant differences in epistemic states generated by them .
flock - generated epistemic states are always persistent ; this follows immediately from the fact that @xmath14 holds in such a state if and only if @xmath15 , and hence @xmath39 .
moreover , it has been shown in @xcite that any finitary persistent epistemic state is representable by some flock .
this means that flocks constitute an adequate syntactic formalism for representing persistent epistemic states .
unfortunately , this also shows that flocks are not representable by pure epistemic states , as has been suggested in @xcite .
since belief sets are uniquely determined by epistemic states , operations on epistemic states will also determine corresponding operations on belief sets .
two kinds of operations immediately suggest themselves as most fundamental , namely removal and addition of information to epistemic states .
admissible belief states of an epistemic state constitute all potential alternatives that are considered as ` serious possibilities ' by the agent . in accordance with this
, the contraction of a proposition @xmath35 from an epistemic state @xmath0 is defined as an operation that removes all admissible belief states from @xmath0 that support @xmath35 .
we will denote the resulting epistemic state by @xmath40 .
the contraction operation has quite regular properties , most important of which being _ commutativity _ : a sequence of contractions can be performed in any order yielding the same resulting epistemic state .
let us compare the above contraction with base contraction . according to @xcite , the first step in performing a base contraction of @xmath35 consists in finding preferred ( selected ) subsets of the base that do not imply @xmath35 .
so far , this fits well with our construction , since the latter subsets exactly correspond to preferred admissible theories of the contracted epistemic state .
then our definition says , in effect , that the contracted belief set should be equal to the intersection of these preferred theories .
unfortunately , such a solution is unacceptable for the base paradigm , since we need to obtain a unique contracted base ; only the latter will determine the resulting belief set .
accordingly , @xcite defines first the contracted base as the intersection of all preferred subsets of the base ( ` partial meet base contraction ' ) , and then the contracted belief set is defined as the set of all propositions that are implied by the new base .
the problems arising in this approach are best illustrated by the following example ( adapted from @xcite ) .
two equally good and reliable friends of a student say to her , respectively , that niamey is a nigerian town , and that niamey has a university .
our student should subsequently retract her acquired belief that niamey is a university town in nigeria .
let @xmath35 and @xmath41 denote , respectively , propositions
niamey is a town in nigeria " , and there is a university in niamey " .
then the above situation can be described as a contraction of @xmath38 from the ( belief set generated by the ) base @xmath42 .
as has been noted in @xcite , this small example constitutes a mayor stumbling block for the base approach to belief change .
actually , we will see that none of the current approaches handles satisfactorily this example . to begin with , it seems reasonable to expect that the contracted belief set in the above situation should contain @xmath43 , since each of the acceptable alternatives support this belief .
this result is also naturally sanctioned by the agm theory . using the base contraction ,
however , we should retreat first to the two sub - bases @xmath44 and @xmath45 that do not imply @xmath38 , and then form their intersection which happens to be empty ! in other words , we have lost all the information contained in the initial base , so all subsequent changes should start from a scratch . next , it seems also reasonable to require that if subsequent evidence rules out @xmath35 , for example , we should believe that @xmath41 . in other words ,
contracting first @xmath38 and then @xmath35 from the initial belief state should make us believe in @xmath41 .
this time the agm theory can not produce this result .
the reason is that the first contraction gives the logical closure of @xmath43 as the contracted belief set , and hence the subsequent contraction of @xmath35 will not have any effect on the corresponding belief state .
notice that this information loss is not ` seen ' in one - step changes ; it is revealed , however , in subsequent changes . as we see it ,
the source of the above problem is that traditional approaches to belief change force us to choose in situations we have no grounds for choice .
and our suggested solution here amounts to retaining all the preferred alternatives as parts of the new epistemic state , instead of transforming them into a single ` combined ' solution .
this means , in particular , that we should allow our epistemic states to be non - determinate .
this will not prevent us from determining each time a unique current set of beliefs ; but we should remember more than that . returning to the example , our contraction operation results in a new belief set @xmath46 , as well as a new epistemic state @xmath47 consisting of two theories .
this latter epistemic state , however , is not base - generated , though it will be generated by a flock @xmath48 .
flocks emerge as a nearest counterpart of bases that will already be closed with respect to our contraction operation .
actually , the latter will correspond to the operation of deletion on flocks suggested in @xcite . a _ contraction of a flock _
@xmath19 with respect to @xmath35 ( notation @xmath49 ) is a flock consisting of all maximal subsets of each @xmath21 that do not imply @xmath35 .
the following result confirms that the above operation on flocks corresponds to contraction of associated epistemic states . if @xmath50 is an epistemic state generated by a flock @xmath29 , then , for any @xmath35 , @xmath51 . despite the similarity of the above definition of contraction with that given in @xcite , the resulting contraction operation
will behave differently in our framework .
the following example illustrates this .
contraction of the flock @xmath52 with respect to @xmath35 is a flock @xmath53 , which is identical to @xmath54 according to our definition of identity .
consequently , @xmath41 will be believed in the resulting epistemic state .
this behavior seems also to agree with our intuitions , since eliminating @xmath35 as an option will leave us with @xmath41 as a single solution . in the representation of @xcite , however , @xmath53 is reducible to @xmath55 . consequently , nothing will be believed in the resulting epistemic state .
furthermore , this also makes the corresponding operation of deletion _ non - commutative _ : whereas deletion of @xmath38 and then @xmath35 from the base @xmath42 results in @xmath53 , deletion of @xmath35 first and then @xmath38 will give a different flock , namely @xmath54 .
the operation of expansion consists in adding information to epistemic states . in the agm theory , this is achieved through a straightforward addition of the new proposition to the belief set , while in the base approach the new proposition is added directly to the base .
the framework of epistemic states drastically changes , however , the form and content of expansion operations .
this stems already from the fact that adding a proposition to an epistemic state is no longer reducible to adding it to the belief set ; it should determine also the place of the newly added proposition in the structure of the expanded epistemic state .
this will establish the degree of firmness with which we should believe the new proposition , as well as its dependence and justification relations with other beliefs . as a way of modelling this additional information
, we suggest to treat the latter as a special case of merging the source epistemic state with another epistemic state that will represent the added information .
accordingly , we will describe first some merge operation on epistemic states .
then an expansion will be defined roughly as a merge of @xmath0 with a rudimentary epistemic state that is generated by the base @xmath44 .
merge is a procedure of combining a number of epistemic states into a single epistemic state , in which we seek to combine information that is supported by the source epistemic states .
it turns out that this notion of merging can be captured using a well - known algebraic construction of _ product_. roughly , a merge of two epistemic states @xmath56 and @xmath57 is an epistemic state in which the admissible states are all _ pairs _ of admissible states from @xmath56 and @xmath57 , the labelling function assigns each such pair the deductive closure of the union of their corresponding labels , while the resulting preference relation agrees with the ` component ' preferences ( see @xcite for a formal description ) .
since the primary subject of this study is non - prioritized change , we will consider below only one kind of merge operations , namely a pure merge that treats the source epistemic states as having an equal ` weight ' . a _ pure merge _ of epistemic states @xmath58 and @xmath59 is an epistemic state @xmath60 such that * @xmath61 , for any @xmath62 ; * @xmath63 iff @xmath64 and @xmath65 .
a pure merge is a merge operation that treats the two component epistemic states as two equally reliable sources .
it is easy to see , in particular , that it is a commutative operation .
being applied to base - generated epistemic states , pure merge corresponds to a straightforward union of two bases : if @xmath56 and @xmath57 are epistemic states generated , respectively , by bases @xmath66 and @xmath67 , then @xmath68 is equivalent to an epistemic state generated by @xmath69 . to begin with , the following result shows that pure merge preserves persistence of epistemic states .
a pure merge of two persistent epistemic states is also a persistent epistemic state .
since finitary persistent epistemic states are representable by flocks , a pure merge gives rise to a certain operation on flocks .
this operation can be described as follows : let us consider two flocks @xmath70 and @xmath71 that have no propositions in common .
then a _ merge _ of @xmath70 and @xmath71 will be a flock @xmath72 thus , the merge of two disjoint flocks is obtained by a pairwise combination of bases belonging to each flock .
note , however , that the assumption of disjointness turns out to be essential for establishing the correspondence between merge of flocks and pure merge of associated epistemic states .
still , this requirement is a purely syntactic constraint that can be easily met by replacing some of the propositions with logically equivalent , though syntactically different propositions .
a suitable example will be given later when we will consider expansions of flocks that are based on the above notion of merge .
the following result shows that merge of flocks corresponds exactly to a pure merge of associated epistemic states .
if @xmath50 and @xmath73 are epistemic states generated , respectively , by disjoint flocks @xmath29 and @xmath74 , then @xmath75 is isomorphic to @xmath76 . for any proposition @xmath35 , we will denote by @xmath77 the epistemic state generated by a singular base @xmath44 .
this pure epistemic state consists of just two theories , namely @xmath78 and @xmath79 .
accordingly , it gives a most ` pure ' expression of the belief in @xmath35 .
now the idea behind the definition below is that an expansion of an epistemic state with respect to @xmath35 amounts to merging it with @xmath77 .
a _ pure expansion _ of an epistemic state @xmath0 with respect to a proposition @xmath35 ( notation @xmath80 ) is a pure merge of @xmath0 and the epistemic state @xmath77 that is generated by a base @xmath44 . in a pure expansion
the new proposition is added as an independent piece of information , that is , as a proposition that is not related to others with respect to priority .
being applied to base - generated epistemic states , pure expansion corresponds to a straightforward addition of a new proposition to the base .
if @xmath0 is generated by a base @xmath9 , then @xmath80 is isomorphic to an epistemic state generated by @xmath81 .
since pure merge is a commutative operation , pure expansions will also be commutative . as any other kind of change in epistemic states , expansions generate corresponding changes in belief sets of epistemic states .
it turns out that expansions generate in this sense precisely agm belief expansion functions : if @xmath8 is a belief set of @xmath0 , then the belief set of @xmath80 coincides with @xmath82 .
thus , belief expansions generated by expansions of epistemic states behave just as agm expansions : the underlying epistemic state plays no role in determining the resulting expanded belief set , since the latter can be obtained by a direct addition of new propositions to the source belief set .
it should be kept in mind , however , that identical expansions of belief sets can be produced by expansions of different epistemic states , and even by different expansions of the same epistemic state .
these differences will be revealed in subsequent contractions and revisions of the expanded belief set . as we have seen earlier
, pure merge generates a corresponding merge operation on flocks .
consequently , pure expansion corresponds in this sense to a certain expansion operation on flocks .
an _ expansion of a flock _
@xmath83 with respect to a proposition @xmath35 that does not appear in @xmath29 is a flock @xmath84 .
thus , an expansion of a flock is obtained simply by adding the new proposition to each base from the flock .
our earlier results immediately imply that this operation exactly corresponds to a pure expansion of the associated epistemic state with @xmath35 : if @xmath50 is an epistemic state generated by the flock @xmath29 , and @xmath35 does not appear in @xmath29 , then @xmath85 is isomorphic to @xmath86 .
the above operation is quite similar in spirit to the operation of insertion into flocks used in @xcite , though the latter was intended to preserve consistency of the component bases , so they defined , in effect , the corresponding revision operation based on contraction and expansion in our sense .
note , however , that our flock expansion is defined only when the added proposition does not appear in the flock .
the need for the restriction is illustrated by the following example .
let us return to the flock @xmath52 , where @xmath35 and @xmath41 denote , respectively , propositions
niamey is a town in nigeria " , and there is a university in niamey " .
recall that this flock is obtainable by contracting @xmath38 from the base @xmath42 .
in other words , it reflects an informational situation in which we have reasons to believe in each of these propositions , but can not believe in both .
now let us expand the epistemic state @xmath50 with @xmath41 .
this expansion can be modeled by expanding @xmath29 with some proposition @xmath87 that is logically equivalent to @xmath41 .
in other words , the epistemic state generated by the flock @xmath88 will be equivalent to the expansion of @xmath50 with @xmath41 .
this flock sanctions belief in @xmath41 in full accordance with our intuitions .
actually , it can be shown that the latter flock is reducible to a flock @xmath89 in the sense that the latter flock will produce an equivalent epistemic state .
however , @xmath87 can not be replaced with @xmath41 in these flocks : the flock @xmath90 is already reducible to a single base @xmath37 in which both @xmath35 and @xmath41 are believed , contrary to our intuitions about the relevant situation : receiving a new support for believing that there is a university in niamey should not force us to believe also that niamey is a town in nigeria .
this also shows most vividly that a straightforward addition of @xmath41 to each base in the flock @xmath48 does not produce intuitively satisfactory results .
an additional interesting aspect of the above representation is that , though we fully believe in @xmath41 in the flock @xmath91 , the option @xmath35 has not been forgotten ; if we will contract now @xmath41 from the latter flock , we will obtain the flock @xmath92 which supports belief in @xmath35 .
a little reflection shows that this is exactly what would be reasonable to believe in this situation .
the purpose of this study was to give a formal representation for iterative non - prioritized change .
as has been shown , such a representation can be achieved in the framework of persistent epistemic states , with flocks providing the corresponding constructive representation . moreover , these representations overcome shortcomings of both the agm and base - oriented models that incur loss of information in iterative changes .
contraction and expansions are two basic operations on epistemic states that allow to define the majority of derived belief changes .
thus , a _ revision _ of an epistemic state is definable via levi identity ( on the level of epistemic states ) , namely as a combination of contraction and expansion .
as can be shown , the resulting operation will be sufficiently expressive to capture any relational belief revision function in the sense of agm . to conclude the paper ,
we want to mention an interesting problem concerning expressivity of our belief change operations vis - a - vis flocks .
namely , it has been shown in @xcite that there are flocks that are not constructible from simple primitive flocks using the contraction and expansion operations ( a simplest example being the flock @xmath93 ) .
this apparently suggests that our stock of belief change operations is not complete and need to be extended with other operations that would provide functional completeness with respect to constructibility of flocks
. the resulting theory would give then a truly complete constructive representation of non - prioritized belief change . | we address a general representation problem for belief change , and describe two interrelated representations for iterative non - prioritized change : a logical representation in terms of persistent epistemic states , and a constructive representation in terms of flocks of bases . |
voids form a prominent aspect of the distribution of galaxies and matter on megaparsec scales .
they are enormous regions with sizes in the range of @xmath0 mpc that are practically devoid of any galaxy and usually roundish in shape .
forming an essential ingredient of the _ cosmic web _
@xcite , they are surrounded by elongated filaments , sheetlike walls and dense compact clusters .
together they define the salient weblike pattern of galaxies and matter which pervades the observable universe .
voids have been known as a feature of galaxy surveys since the first surveys were compiled @xcite . following the discovery by @xcite of the most dramatic specimen , the botes void , a hint of their central position within a weblike arrangement came with the first cfa redshift slice @xcite .
this view has recently been expanded dramatically as maps of the spatial distribution of hundreds of thousands of galaxies in the 2dfgrs @xcite and sdss @xcite have become available
. voids are a manifestation of the cosmic structure formation process as it reaches a non - linear stage of evolution .
structure forms by gravitational instability from a primordial gaussian field of small amplitude density perturbations , where voids emerge out of the depressions ( e.g. * ? ? ?
* ; * ? ? ?
they mark the transition scale at which perturbations have decoupled from the hubble flow and organized themselves into recognizable structural features .
early theoretical models of void formation @xcite were followed and generalized by the first numerical simulations of void centered universes @xcite . in recent years
the huge increase in computational resources has enabled n - body simulations to resolve in detail the intricate substructure of voids within the context of hierarchical cosmological structure formation scenarios @xcite .
they confirm the theoretical expectation of voids having a rich substructure as a result of their hierarchical buildup .
theoretically this evolution has been succesfully embedded in the extended press - schechter description @xcite .
@xcite showed how this can be described by a two - barrier excursion set formalism ( also see * ? ? ?
the two barriers refer to the two processes dictating the evolution of voids : their merging into ever larger voids as well as the collapse and disappearance of small ones embedded in overdense regions ( see * ? ? ?
besides representing a key constituent of the cosmic matter distribution voids are interesting and important for a variety of reasons .
first , they are a prominent feature of the megaparsec universe .
a proper and full understanding of the formation and dynamics of the cosmic web is not possible without understanding the structure and evolution of voids @xcite .
secondly , they are a probe of cosmological parameters .
the outflow from the voids depends on the matter density parameter @xmath1 , the hubble parameter @xmath2 and possibly on the cosmological constant @xmath3 ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
. these parameters also dictate their redshift space distortions @xcite while their intrinsic structure and shape is sensitive to various aspects of the power spectrum of density fluctuations @xcite . a third point of interest concerns the galaxies in voids .
voids provide a unique and still largely pristine environment for studying the evolution of galaxies @xcite .
the recent interest in environmental influences on galaxy formation has stimulated substantial activity in this direction @xcite . despite the considerable interest in voids
a fairly basic yet highly significant issue remains : identifying voids and tracing their outline within the complex spatial geometry of the cosmic web .
there is not an unequivocal definition of what a void is and as a result there is considerable disagreement on the precise outline of such a region ( see e.g. * ? ? ?
because of the vague and diverse definitions , and the diverse interests in voids , there is a plethora of void identification procedures @xcite .
the `` sphere - based '' voidfinder algorithm of @xcite has been at the basis of most voidfinding methods
. however , this succesful approach will not be able to analyze complex spatial configurations in which voids may have arbitrary shapes and contain a range and variety of substructures . a somewhat related and tessellation based voidfinding technique that still is under development is zobov @xcite .
it is the voidfinder equivalent to the voboz halofinder method @xcite .
here we introduce and test a new and objective voidfinding formalism that has been specifically designed to dissect the multiscale character of the void network and the weblike features marking its boundaries .
our _ watershed void finder _ ( wvf ) is based on the watershed algorithm @xcite .
it stems from the field of mathematical morphology and image analysis .
the wvf is defined with respect to the dtfe density field of a discrete point distribution @xcite .
this assures an optimal sensitivity to the morphology of spatial structures and yields an unbiased probe of substructure in the mass distribution ( see e.g * ? ? ?
* ; * ? ? ?
because the wvf void finder does not impose a priori constraints on the size , morphology and shape of a voids it provides a basis for analyzing the intricacies of an evolving void hierarchy .
indeed , this has been a major incentive towards its development .
this study is the first in a series . here
we will define and describe the watershed void finder and investigate its performance with respect to a test model of spatial weblike distributions , voronoi kinematic models .
having assured the success of wvf to trace and measure the spatial characteristics of these models the follow - up study will address the application of wvf on a number of gif n - body simulations of structure formation @xcite . amongst others ,
wvf will be directed towards characterizing the hierarchical structure of the megaparsec void population @xcite . for a comparison of the wvf with other void finder methods we refer to the extensive study of @xcite . in the following sections we will first describe how
the fundamental concepts of mathematical morphology have been translated into a tool for the analysis of cosmological density fields inferred from a discrete n - body simulation or galaxy redshift survey point distribution ( sect . 2 & 3 ) . to test our method
we have applied it to a set of heuristic and flexible models of a cellular spatial distribution of points , voronoi clustering models .
these are described in section 4 . in section 5
we present the quantitative analysis of our test results and a comparison with the known intrinsic properties of the test models . in section 6
we evaluate our findings and discuss the prospects for the analysis of cosmological n - body simulations .
the new void finding algorithm which we introduce here is based on the _ watershed transform _ of @xcite and @xcite .
a more extensive and technical description of the basic concepts of mathematical morphology and the basic watershed algorithm in terms of homotopy transformations on lattices @xcite is provided in appendix [ app : mathmorph ] and [ app : wshedimpl ] .
the watershed transform is used for segmenting images into distinct regions and objects .
the watershed transform ( wst ) is a concept defined within the context of mathematical morphology , and was first introduced by @xcite . the basic idea behind the wst finds its origin in geophysics .
the wst delineates the boundaries of the separate domains , i.e. the _ basins _ , into which yields of , for example , rainfall will collect . the word _ watershed
_ refers to the analogy of a landscape being flooded by a rising level of water .
suppose we have a surface in the shape of a landscape ( first image of fig .
[ fig : proc ] ) .
the surface is pierced at the location of each of the minima . as the water - level rises a growing fraction of the landscape
will be flooded by the water in the expanding basins .
ultimately basins will meet at the ridges corresponding to saddle - points in the density field .
this intermediate step is plotted in the second image of fig .
[ fig : proc ] .
the ridges define the boundaries of the basins , enforced by means of a sufficiently high dam .
the final result ( see last in fig . [ fig : proc ] ) of the completely immersed landscape is a division of the landscape into individual cells , separated by the _ ridge dams_. in the remainder of this study we will use the word _
`` segment '' _ to describe the watershed s cells .
the watershed algorithm holds several advantages with respect to other voidfinders : * within an ideal smooth density field ( i.e. without noise ) it will identify voids in a parameter free way .
no predefined values have to be introduced . in less ideal , and realistic , circumstances a few parameters have to be set for filtering out discreteness noise .
their values are guided by the properties of the data .
* the watershed works directly on the topology of the field and does not reply on a predefined geometry / shape . by implication
the identified voids may have any shape . *
the watershed naturally places the _ divide lines _ on the crests of a field .
the void boundary will be detected even when its boundary is distorted . *
the transform naturally produces closed contours . as long as minima
are well chosen the watershed transform will not be sensitive to local protrusions between two adjacent voids .
obviously we can only extract structural information to the extent that the point distribution reflects the underlying structure .
undersampling and shotnoise always conspire to obfiscate the results , but we believe the present methodology provides an excellent way of handling this .
the watershed void finder ( wvf ) is an implementation of the watershed transform within a cosmological context .
the watershed method is perfectly suited to study the holes and boundaries in the distribution of galaxies , and holds the specific promise of being able to recognize the void hierarchy that has been the incentive for our study .
the analogy of the wst with the cosmological context is straightforward : _ voids _ are to be identified with the _ basins _ , while the _ filaments _ and _ walls _ of the cosmic web are the ridges separating the voids from each other .
an outline of the steps of the watershed procedure within its cosmological context is as follows : * * dtfe * : given a point distribution ( n - body , redshift survey ) , the delaunay tessellation field estimator ( dtfe , * ? ? ?
* ) is used to define a continuous density field throughout the sample volume .
this guarantees a density field which retains the morphological character of the underlying point distribution , i.e. the hierarchical nature , the web - like morphology dominated by filaments and walls , and the presence voids is warranted . + * * grid sampling * : for practical processing purposes the dtfe field is sampled on a grid .
the optimal grid size has to assure the resolution of all morphological structures while minimizing the number of needed gridcells .
this criterion suggests a grid with gridcells whose size is in the order of the interparticle separation .
+ * * rank - ordered filtering * : the dtfe density field is adaptively smoothed by means of _ natural neighbour maxmin and median _ filtering .
this involves the computation of the median , minimum or maximum of densities within the _ contiguous voronoi cell _ , the region defined by a point and its _ natural neighbours _ .
+ * * contour levels * : the image is transformed into a discrete set of density levels .
the levels are defined by a uniform partitioning of the cumulative density distribution . + * * pixel noise * : with an opening and closing ( operation to be defined in appendix .
[ app : mathmorph ] ) of 2 pixel radius we further reduce pixel by pixel fluctuations . + * * field minima * : the minima in the smoothed density field are identified as the pixels ( grid cells ) which are exclusively surrounded by neighbouring grid - cells with a higher density value . + * * flooding * : the _ flooding procedure _ starts at the location of the minima . at successively increasing flood levels the surrounding region with a density lower than the corresponding density threshold
is added to the _ basin _ of a particular minimum .
the flooding is illustrated in fig .
[ fig : proc ] . + * * segmentation * : once a pixel is reached by two distinct basins it is identified as belonging to their segmentation boundary . by continuing this procedure up to the maximum density level the whole region has been segmented into distinct _
void patches_. + * * hierarchy correction * : a correction is necessary to deal with effects related to the intrinsic hierarchical nature of the void distribution .
the correction involves the removal of segmentation boundaries whose density is lower than some density threshold
. the natural threshold value would be the typical void underdensity @xmath4 ( see sect .
[ sec : threshold ] ) . alternatively
, dependent on the application , one may chose to take a user - defined value .
+ 0.4truecm a direct impression of the watershed voidfinding method is most readily obtained via the illustration of a representative example . in fig .
[ fig : wvf ] the watershed procedure has been applied to the cosmological gif2 simulation @xcite . the n - body particle distribution ( lefthand fig .
[ fig : wvf ] ) is translated into a density field using the dtfe method .
the application of the dtfe method is described in section [ sec : dtfe ] , the details of the dtfe procedure are specified in appendix [ app : dtfe ] .
the dtfe density field is sampled and interpolated on a @xmath5 grid , the result of which is shown in the top righthand frame of fig .
[ fig : wvf ] .
the gray - scales are fixed by uniformly sampling the cumulative density distribution , ensuring that all grayscale values have the same amount of volume .
the dtfe density field is smoothed by means of the adaptive _ natural neighbour median _
filtering described in sect .
[ sec : natnghb ] .
this procedure determines the filtered density values at the location of the particles .
subsequently , these are interpolated onto a grid .
this field is translated into a grayscale image following the same procedure as that for the raw dtfe image ( bottom lefthand panel ) .
the minima in the smoothed density field are identified and marked as the flooding centres for the watershed transform .
the resulting wvf segmentation is shown in the bottom righthand frame of fig .
[ fig : wvf ] . the correspondence between the cosmic web , its voids and the watershed segmentation is striking .
there is an almost perfect one - to - one correspondence between the segmentation and the void regions in the underlying density field .
the wvf method does not depend on any predefined shape . as a result
, the recovered voids do follow their natural shape .
a qualitative assessment of the whole simulation cube reveals that voids are very elongated and have a preferential orientation within the cosmic web , perhaps dictated by the megaparsec tidal force field ( see e.g. * ? ? ?
clearly , the watershed void finder is able to extract substructure at any level present in the density distribution . while this is an advantage with respect to tracing the presence of substructure within voids it does turn into a disadvantage when seeking to trace the outline of large scale voids or when dealing with noise in the dataset . while the noise - induced artificial segments are suppresed by means of the full machinery of markers ( sect .
[ sec : marker ] ) , void patch merging ( sect .
[ sec : merging ] ) and natural neighbour rank filtering ( sect . [ sec : natnghb ] ) , it are the latter two which may deal with intrinsic void hierarchy .
the follow - up study @xcite will involve a detailed quantitative analyze of volume and shapes of the voids in the gif2 mass distribution for a sequence of timesteps .
in order to appreciate the various steps of the watershed void finder outlined in the previous section we need to describe a few of the essential steps in more detail . to process a point sample into a spatial density field we use dtfe . to detect voids of a particular scale
it is necessary to remove statistically insignificant voids generated by the shotnoise of the discrete point sample as well as physically significant subvoids . in order to retain only the statistically significicant voids we introduce and apply natural neighbour rank - order filtering .
hierarchy merging is used for the removal of subvoids which one would wish to exclude from a specific void study .
the input samples for our analysis are mostly samples of galaxy positions obtained by galaxy redshift surveys or the positions of a large number of particles produced by n - body simulations of cosmic structure formation . in order to define a proper continuous field from a discrete distribution of points computer particles or galaxies we translate the spatial point sample into a continuous density field by means of the delaunay tessellation field estimator ( dtfe , * ? ? ?
the dtfe technique @xcite recovers fully volume - covering and volume - weighted continuous fields from a discrete set of sampled field values .
the method has been developed by @xcite and forms an elaboration of the velocity interpolation scheme introduced by @xcite .
it is based upon the use of the voronoi and delaunay tessellations of a given spatial point distribution to form the basis of a natural , fully self - adaptive filter in which the delaunay tessellations are used as multidimensional interpolation intervals .
a typical example of a dtfe processed field is the one shown in the top row of fig .
[ fig : wvf ] : the particles of a gif n - body simulation @xcite are translated into the continuous density field in the righthand frame . the primary ingredient of the dtfe method is the delaunay tessellation of the particle distribution .
the delaunay tessellation of a point set is the uniquely defined and volume - covering tessellation of mutually disjunct delaunay tetrahedra ( triangles in 2d ) .
each is defined by the set of four points whose circumscribing sphere does not contain any of the other points in the generating set @xcite .
the delaunay tessellation and the voronoi tessellation of the point set are each others _ dual_. the voronoi tessellation is the division of space into mutually disjunct polyhedra , each polyhedron consisting of the part of space closer to the defining point than any of the other points @xcite dtfe exploits three properties of voronoi and delaunay tessellations @xcite .
the tessellations are very sensitive to the local point density .
dtfe uses this to define a local estimate of the density on the basis of the inverse of the volume of the tessellation cells .
equally important is their sensitivity to the local geometry of the point distribution .
this allows them to trace anisotropic features such as encountered in the cosmic web .
finally , dtfe exploits the adaptive and minimum triangulation properties of delaunay tessellations in using them as adaptive spatial interpolation intervals for irregular point distributions . in this way it is the first order version of the _ natural neighbour method _ @xcite . within the cosmological context a major and crucial
characteristic of a processed dtfe density field is that it is capable of delineating three fundamental characteristics of the spatial structure of the megaparsec cosmic matter distribution .
it outlines the full hierarchy of substructures present in the sampling point distribution , relating to the standard view of structure in the universe having arisen through the gradual hierarchical buildup of matter concentrations .
dtfe also reproduces any anisotropic patterns in the density distribution without diluting their intrinsic geometrical properties .
this is particularly important when analyzing the the prominent filamentary and planar features marking the cosmic web .
a third important aspect of dtfe is that it outlines the presence and shape of voidlike regions . because of the interpolation definition of the dtfe field reconstruction voids
are rendered as regions of slowly varying and moderately low density values .
a more detailed outline of the dtfe reconstruction procedure can be found in appendix [ app : dtfe ] .
dtfe involves the estimate of a continuous field throughout the complete sample volume . to process the dtfe field through the wvf machinery we sample the field on a grid
it is important to choose a grid which is optimally suited for the void finding purpose of the wvf method . on the one hand , the grid values should represent all physically significant structural features ( voids ) in the sample volume . on the other hand
, the grid needs to be as coarse as possible in order to suppress the detection of spurious and insignificant features .
the latter is also beneficial from a viewpoint of computational efficiency .
this is achieved by adopting a gridsize in the order of the mean interparticle distance .
( a ) ( b ) the dtfe grid sampling is accomplished through monte carlo sampling within each grid cell . within each gridcell
the dtfe density value is measured at 10 randomly distributed sample points .
the grid value is taken to be their average .
a major and novel ingredient of our wvf method intended to eliminate shot noise in the dtfe density field reconstructions is that of a natural non - linear filtering extension : the _ natural neighbour rank - ordered filtering _
we invoke two kinds of non - linear adaptive smoothing techniques , _ median filtering _ and _ max / min filtering _ , the latter originating in mathematical morphology ( mm ) .
both filters are rank order filters , and both have well known behaviour .
they have a few important properties relevant for our purposes .
median filtering is very effective in removing shot noise while preserving the locations of edges .
the max / min filters are designed to remove morphological features arising from shot noise ( see appendix [ app : mathmorph ] ) .
the filters are defined over neighbourhoods .
these are often named connectivity or , alternatively , structure elements .
image analysis usually deals with regular two - dimensional image grids .
the most common situation for such grids are straightforward 4-connectivities or 8-connectivities ( see fig .
[ fig : connect ] ) .
when a more arbitrary shape is used one usually refers to it as a structure element . in the situation of our interest we deal with
irregularly spaced data , rendering it impossible to use any of the above neighbourhoods .
it is the delaunay triangulation which defines a natural neighbourhood for these situations . for any point it consists of its
_ natural neighbours _ , i.e. all points to which it is connected via an edge of the delaunay triangulation ( see fig .
[ fig : natnbrs ] ) .
this may be extended to any higher order natural neighbourhood : e.g. a second order neighbourhood would include the natural neighbours of the ( first order ) natural neighbours .
the advantages of following this approach are the same as those for the dtfe procedure : the natural neighbour filtering
shortly named nn-_median filtering _ or nn-_min / max filtering _
forms a natural extension to our dtfe based formalism .
it shares in the major advantage of being an entirely natural and self - adaptive procedure .
the smoothing kernel is compact in regions of high point concentrations , while it is extended in regions of low density .
implementing the min / max and median natural neighbour filters within the dtfe method is straightforward .
the procedure starts with the dtfe density value at each of the ( original ) sample points
. these may be the particles in an n - body simulation or the galaxies in a redshift survey .
for each point in the sample the next step consists of the determination of the median , maximum or minimum value over the set of density values made up by that of the point itself and those of its natural neighbours .
the new `` filtered '' density values are assigned to the points as the first - order filter value .
this process is continued for a number of iterative steps , each step yielding a higher order filtering step .
the number of iterative steps of the natural neighbour smoothing is dependent on the size of the structure to be resolved and the sampling density within its realm .
testing has shown that a reasonable order of magnitude estimate is the mean number of sample points along the diameter of the structure . as an illustration of this criterion
one may want to consult the low noise and high noise voronoi models in fig .
[ fig : segorg ] .
while the void cells of the low noise models contain on average 6 points per cell diameter , the void cells of the high noise model contain around 16 .
fifth - order filtering sufficed for the low noise model , 20-th order for the high noise model ( fig .
[ fig : vor11 ] and fig .
[ fig : vor15 ] ) in the final step , following the specified order of the filtering process , the filtered density values determined at the particle positions are interpolated onto a regular grid for practical processing purposes ( see sec .
[ sec : dtfegrid ] ) . [ cols="^,^,^,^,^,^,^,^,^,^ " , ] [ tab : void ] a direct application of the watershed
transform results in a starkly oversegmented tessellation ( fig .
[ fig : vor11 ] and fig .
[ fig : vor15 ] ) . amongst the overabundance of mostly artificial , noise - related , segments we may also discern real significant watersheds .
their boundary ridges ( divide lines ) are defined by filaments , walls and clusters surrounding the voids .
many of these genuine voids are divided into small patches .
they are the result of oversegmentation induced by the noisy poisson point distribution within the cells .
the local minima within this background noise will act as individual watershed flood centres marking corresponding , superfluous , watershed segments .
while for a general cosmological distribution it may be challenging to separate genuine physical subvoids from artificial noise - generated ones , the voronoi kinematic models have the unique advantage of having no intrinsic substructure .
any detected substructure has to be artificial , rendering it straightforward to assess the action of the various steps intent on removing the noise contributions .
the first step in the removal of insignificant minima consists of the application of the iterative natural neighbour median filtering process .
this procedure , described in sect .
[ sec : natnghb ] , removes some of the shot noise in the low density regions . at the same time it is edge preserving .
the result of five nn - median filtering iterations on the high noise version of the voronoi kinematic clustering model is shown in fig .
[ fig : vor11 ] . with the exception of a few artificial edges the resulting watershed segmentation
almost perfectly matches the intrinsic voronoi tessellation .
figure [ fig : vor15 ] shows the result for the high noise version of the same voronoi kinematic clustering model . in this case pure nn - median filtering is not sufficient .
a much more acceptable result is achieved following the application of the watershed hierarchy segment merging operation and the removal of ridges with a density contrast lower than the 0.8 contrast threshold .
for both the low - noise and high - noise realizations we find that the intrinsic and prominent edges of the voronoi pattern remain in place .
nonetheless , a few shot noise induced artificial divisions survive the filtering and noise removal operations .
they mark prominent coherent but fully artificial features in the noise . given their rare occurrence we accept these oversegmentations as inescapable yet insignificant contaminations .
the watershed segmentation retrieved by the watershed voidfinder is compared with the intrinsic ( geometric ) voronoi tessellation .
the first test assesses the number of false and correct wvf detections . a second test concerns the volume distribution of the voronoi cells and the corresponding watershed void segments . for our performance study we have three basic models : the intrinsic ( geometric ) voronoi tessellation , and the low noise and high noise voronoi clustering models ( table [ tab : vorkinmpar ] ) .
the voronoi clustering models are processed by wvf . in order to assess the various steps in the wvf procedure
the models are subjected to different versions of the wvf .
the second column of table [ tab : void ] lists the differently wvf processed datasets .
these are : 1 . _
original _ : the pure dtfe density field , without any smoothing or boundary removal , subjected to the watershed transform .
minmax _ : only the nn-_min / max _ filtering is applied to the dtfe density field before watershed segmentation .
med@xmath6 _ : @xmath6 iteratations of median natural - neighbour filtering is applied to the dtfe density field . in all situations
this includes max / min filtering afterwards .
_ hierarch _ : following the watershed transform , on the pure non - filtered dtfe density , a density threshold is applied .
the applied hierarchy threshold level is @xmath7 : all segment boundaries with a density lower than @xmath8 are removed as physically insignificant .
med@xmath6hr _ : mixed process involving an @xmath6 times iterated median filtered dtfe density field , followed by the watershed transform , after which the segment boundaries below the hierarchy threshold @xmath8 are removed .
note that the physically natural threshold of @xmath4 is not really applicable to the heuristic voronoi models . on the basis of the model specifications
the threshold level has been set to @xmath9 .
each of the resulting segmentations is subjected to a range of detection assessments .
these are listed in the 3rd to 7th column of table [ tab : void ] .
the columns of the table contain respectively the number of wvf void detections , the amount of false splits , the amount of false mergers , the number of correctly identified voids , and the correctness measure .
while the top block contains information on the intrinsic ( geometric ) voronoi tessellation , the subsequent two blocks contain the detection evaluations for the _ low noise _ and _ high noise _ models .
the false detections are split into two cases .
the first case we name _ false splits _ : a break up of a genuine cell into two or more watershed voids .
the second class is that of the _ false mergers _ : the spurious merging of two voronoi cells into one watershed void .
the splits , mergers and correct voids are computed by comparing the overlap between the volume of the voronoi cell and that of the retrieved watershed void .
a split is identified if the overlap percentage w.r.t .
the voronoi volume is lower than a threshold of 85 percent of the overlapping volume . along the same line
, a merger concerns an overlap deficiency with respect to the watershed void volume .
when both measures agree for at least 85 percent a void is considered to be correct .
the correctness of a certain segmentation is the percentage of correctly identified voids with respect the 180 intrinsic voronoi cells
. judging by the number of voids in the low noise model , it is clear that smoothing or any other selection criterion remain necessary to reduce the number of minima from 850 to a number close to the intrinsic value 180 .
the second row shows the results for the case when just the maxmin filter is applied .
this step already reduces the number of insignificant minima by already 60 percent .
it is an indication for the local character of the shot noise component .
the next three rows list the results for various iterations of the median filtering . with just 2 iterations almost 90 percent of the voids
are retrieved .
most of the splits are removed at 2 iterations .
this result does not improve with more median filtering , even up to 20 iterations this just increases the number of mergers as more walls are smoothed away .
the number of splits also increases as minima begin to merge . in general the same conclusion
can be drawn for the high noise model .
rank - ordered nn - median and nn - min / max filters manage to reduce the number of insignificant minima by a factor of 80 percent ( cf . the number of voids in the second and third row ) .
these models attain a correctness of approximately fifty percent .
mere rank - ordered filtering is evidently insufficient .
we also ran a _
model which did not include median filtering .
instead only insignificant boundaries were removed .
it achieved a recovery of fifty percent .
combining both methods ( _ med5hr _ and _ med20hr _ ) recovers 80 till 90 percent of the voids .
the succes rate may be understood by the complementarity of both methods : while the median filtering recovers the coherent structures , the thresholding will remove those coherent walls that are far underdense .
the translation to a cosmological density field is straightforward .
the rank - ordered filtering ensures that insignificant minima are removed and that the watershed will pick up only coherent boundaries .
thresholding is able to order these walls by significance and to remove the very underdense and insignificant walls . in fig .
[ fig : voiddis ] we compare the distribution of the void volumes .
the histogram shows the distribution of equivalent radii for the segment cells , @xmath10 the solid line histogram shows the ( geometric ) volume distribution for the intrinsic voronoi tessellations . on top of this
we show the distributions for the various ( parameterized ) watershed segmentation models listed in table [ tab : void ] .
not surprisingly the best segmentations have nearly equivalent volume - distributions .
for the lownoise models this is _ med2 _
( lefthand ) , for the highnoise models _
med20hr _ ( righthand ) .
this conclusion is in line with the detection rates listed in table [ tab : void ] .
the visual comparison of the intrinsic geometric voronoi tessellations and the two best segmentations - _ med2 _ for the lownoise model and _ med20hr _ for the highnoise version confirms that also the visual impression between these watershed renderings and the original voronoi model is very much alike .
we have also assessed the cell - by - cell correspondence between the watershed segmentations and the voronoi model . identifying each watershed segment with its original voronoi cell
we have plotted the volume of all watershed cells against the corresponding voronoi cell volumes .
the scatter plots in fig .
[ fig : voidvol ] form a convincing confirmation of the almost perfect one - to - one relation between the volumes derived by the wvf procedure and the original volumes .
the only deviations concern a few outliers .
these are the _ hierarchy merger _ segments for which the watershed volumes are too large , resulting in a displacement to the right . while the volumes occupied by the watershed segments in fig .
[ fig : best ] do overlap almost perfectly with that of the original voronoi cells , their surfaces have a more noisy and erratic appearance .
this is mostly a consequence of the shot noise in the ( dtfe ) density field , induced by the noise in the underlying point process .
the crests in the density field are highly sensitive to any noise , in addition to assess the impact of the noise on the surfaces of the watershed segments we compared the watershed segement surface areas with the voronoi cell surface areas .
the results are shown in fig .
[ fig : voidsurf ] .
we tested the lownoise _
med2 _ and the highnoise _
med20hr_. in both cases we find a linear relationship between the watershed surface and the genuine voronoi surface area .
both cases involve considerably more scatter than that for the volumes of the cells .
in addition to an increased level of scatter we also find a small be it significant offset from the 1 - 1 relation .
the slope of the lownoise model is only slightly less than unity , the highnoise model slope deviates considerably more . these offsets do reflect the systematically larger surface areas of the watershed segments , a manifestation of their irregular surfaces .
it is evident that the level of irregularity is more substantial for the highnoise than for the lownoise reconstructions ( cf .
[ fig : best ] ) .
the scatter plots do also reveal several cells with huge deviations in surface area . unlike expected there is no systematic trend for smaller cells to show larger deviations .
some of the small deviating cells can be recognized in fig .
[ fig : best ] as highly irregular patches .
the large deviant cells correspond to watershed segments which as a result of noisy boundaries got wrongly merged . while the irregularity of the surface areas forms a good illustration of the noise characteristics of the watershed patches , for the purpose of void identification it does not pose a serious problem .
smoother contours may always be obtained by applying the flooding process on a properly smoothed field . some suggestions for this
may be achieved follows in the discussion .
the wvf void finder technique is based on the watershed transform known from the field of image processing .
voids are identified with the basins of the cosmological mass distribution , the filaments and walls of the cosmic web with the ridges separating the voids from each other . stemming from the field of mathematical morphology ,
watershed cells are considered as patches locally minimizing the `` topographic distance '' .
the wvf operates on a continuous density field .
because the cosmological matter distribution is nearly always sampled by a discrete distribution of galaxies or n - body particles , the first step of the wvf is to translate this into a density field by means of the delaunay tessellation field estimator ( dtfe ) . because the wvf involves an intrinsically morphological and topological operation , the capability of dtfe to retain the shape and morphology of the cosmic web is essential .
it guarantees that within this cosmological application the intrinsic property of the watershed transform to act independent of scale , shape , and structure of a segment is retained . as a result , voids of any scale , shape and structure may be detected by wvf .
in addition to the regular watershed transform the wvf needs to invoke various operations to suppress ( discreteness ) sampling noise .
in addition , we extend the watershed formalism such that the wvf will be capable of analyzing the hierarchy of voids in a matter distribution , i.e. identify how and which small scale voids are embedded within a void on larger scales @xcite . markers indicating significant void centers , false segment removal by hierarchy merging and natural neighbour filtering all affect an efficient noise removal . hierarchy merging manages to eliminate boundaries between subvoids .
natural neighbour median filtering , for various orders , is an essential new ingredient for highlighting the hierarchical embedding of the void population .
it allows a natural selection of voids , unbiased with respect to the scale and shape of these structures .
the voids that persist over a range of scales are expected to relate to the voids that presently dominate the cosmic matter distribution . in other words ,
wvf preserves the void hierarchy @xcite .
the present work includes a meticulous qualitative and quantitative assessment of the watershed transform on the basis of a set of voronoi kinematic models .
these heuristic models of spatial weblike or cellular galaxy or particle distributions facilitate the comparison between the void detection of the wvf and that of the characteristics of the cells in the original and intrinsically known voronoi tessellation .
it is found that wvf is not only succesfull in reproducing the qualititative cellular appearance of the voronoi models but also in reproducing quantitative aspects like an almost perfect 1 - 1 match of cell size with wvf segment volume and the corresponding void size distribution .
we foresee various possible improvements of the wvf .
these concern in particular the identification of significant edges .
one possibility is that extension proposed by @xcite , in which not only the `` topographic costs '' but also the lengths of the contours should be minimized .
the length minimization will result in smoother boundaries .
additional improvements may be found in better filtering procedures in order to facilitate studies of hierarchically structured patterns .
we expect considerable improvements by anisotropic diffusion techniques @xcite and are currently implementing these in the wvf code .
given the results of our study , we are confident for applying wvf to more elaborate and realistic simulations of cosmic structure formation and on large galaxy redshift surveys . the analysis of a set of gif cosmological simulations will be presented in an upcoming paper .
we wish to thank miguel aragn - calvo for permission to use fig . [
fig : vorkinmschm ] .
we are grateful to the referee , mark neyrinck , for the incisive , detailed and useful comments and recommendations for improvements .
we particularly thank the participants of the knaw colloquium `` cosmic voids '' in amsterdam , dec .
2006 , for the many useful discussions and encouraging remarks .
99 k. , et al .
( the sdss collaboration ) , 2003 , aj , 126 , 2081 j. , mhnen p.,1998 , apj , 497 , 534 s. , mller v.,2002 , mnras , 332 , 205 a.j . , hoyle f. , fernando t , vogeley m.s . , 2003 , mnras , 340 , 160 f. , van de weygaert r. , 1996 , mnras , 279 , 693 e. , 1985 , apjs , 58 , 1 s. , lantuejoul c. , 1979 , in proceedings international workshop on image processing , ccett / irisa , rennes , france s. , meyer f. , 1993 , mathematical morphology in image processing , ed .
m. dekker , new york , ch .
12 , 433 m.j .
, marimont d.h . , 1998 , ieee image processing , 7 , 421 j.r . , cole s. , efstathiou g. , kaiser n. , 1991 , apj , 379 , 440 j.r .
, kofman l. , pogosyan d. , 1996 , nature , 380 , 603 g.r.,da costa l.n . ,
goldwirth d.s .
, piran t. , 1992 , apj , 388 , 234 j. , sambridge m. , 1995 , nature , 376 , 655 l. , padilla n.d . , valotto c. , lambas d.g , 2006 , mnras , 373 , 1440 g. , rood h.j . , 1975 , nature , 257 , 294 j.m . , sheth r.k .
, diaferio a. , gao l. , yoshida n. , 2005 , mnras , 360 , 216 j.m . , pearce f. , brunino r. , foster c. , platen e. , basilakos s. , fairall a. , feldman h. , gottlber s. , hahn o. , hoyle f. , mller v. , nelson l. , neyrinck m. , plionis m. , porciani c. , shandarin s. , vogeley m. , van de weygaert r. , 1999 , mnras , subm . m. , et al . , ( the 2dfgrs team ) , 2003 , vizier online data catalog , 7226 a. , rees m.j . , 1994 , apj , 433 , l1 v. , geller m. , huchra j.p . , 1986 , apj , 302 , l1 b. n. , 1934 , bull . acad . sci .
clase sci .
, 7 , 793 j. , da costa l.n . , goldwirth d.s . , lecar m. , piran t. , 1993 , apj . , 410 , 458 j.,joeveer m. , saar e , 1980 , mnras , 193 , 353 h. , piran t. , 1997 , apj , 491 , 421 h.h . , triay r. , 2006 , gr - qc/0607090 s.r . , piran t. , 2006 , mnras , 366 , 467 y. , piran t. , 2001 , apj , 548 , 1 d.m .
, vogeley m.s . , 2004 , apj , 605 , 1 s. , okas e.l .
, klypin a. , hoffman y. , 2003 , mnras , 344 , 715 l.a . ,
thompson s.a . , 1978 ,
apj , 222 , 748 n.a .
, geller m.j . , 1999 , aas , 195 , 1384 o. , porciani c. , carollo m. , dekel a. , 2007 , mnras , 375 , 489 h.j.a.m .. , 1994 , morphological image operators , academic press m. , yepes g. , gottlber s. , springel w.,2006 , mnras , 371 , 401 y. , shaham j. , 1982 , apj , 262 , l23 y. , silk j. , wyse r.f.g . , 1992 , apj , 388 , l13 f. , vogeley m. , 2002 , apj , 566 , 641 , v. , 1984 , mnras , 206 , 1 g. , fairall a.p . , 1991 , mnras , 248 , 313 g. , colberg j.m .
, diaferio a. , white s.d.m . , 1999 ,
mnras , 303 , 188 r.p . ,
oemler a. , schechter p.l .
, shectman s.a . , 1981 ,
apj , 248 , 57 r.p . , oemler a. , schechter p.l . ,
shectman s.a . , 1987 , apj , 314 , 493 r. , 1998 , in mathematical morphology and its applications to image and signal processing , eds . h.j.a.m .
heijmans , j.b.t.m .
roerdink , p. 35 - 42
, kluwer j. , park d. , 2006 , apj , 652 , 1 b. , weinberg d.h . , 1994 ,
mnras , 267 , 605 h. , wasserman i. , 1990 , apj , 348,1 g. , 1975 , random sets and integral geometry , wiley h. , white s.d.m . , 2002 , mnras , 337 , 1193 f. , 1994 , signal processing , 38 , 113 f. , beucher s. , 1990 , j. visual comm .
image rep .
, 1 , 21 m. c. , gnedin n. y. , hamilton a. j. s. , 2005 , mnras , 356 , 1222 m. , 2007 , in cosmic voids , proc .
knaw colloquium , knaw d. , colombi s. , dore o. , 2006 , mnras , 366 , 1201 t.n . , worring m. , van den boomgaard r. , 2003 , ieee transactions on pattern analysis and machine intelligence , vol . 25 , no . 3 , 330 a. , boots b. , sugihara k. , chiu s.n . , 2000 , 2nd ed .
, wiley n.d . , ceccarelli l. , lambas d.g . ,
mnras , 363 , 977 s.g .
, betancort - rijo j.e .
, prada f. , klypin a. , gottlber s. , mnras , 369 , 335 s.g .
, prada f. , holtzman j. , klypin a. , betancort - rijo j.e . , mnras , 372 , 1710 p.j.e . , 2001 ,
apj , 557 , 495 e. , van de weygaert r. , jones b.j.t .
, 2007 , mnras , in prep .
m. , basilakos s. , 2002 , mnras , 330 , 399 w.h . , schechter p. , 1974 ,
apj , 187 , 425 e. , geller m.j .
, 1991,apj , 373 , 14 j. , meijster a. , 2000 , fundamenta informaticae , vol .
41 , 187 r.r .
, vogeley m.s . , hoyle f. , brinkmann j. , 2005 , apj , 624 , 571 b.s . , melott a.l . , 1996 ,
apj , 470 , 160 w. , 2007 , the delaunay tessellation field estimator , ph.d .
thesis , university of groningen w. , van de weygaert r. , 2000 , a&a , 363 , l29 w. , van de weygaert r. , 2007 , a&a , to be subm .
ryden b.s . ,
melott a.l . , 2001 , apj , 546 , 609 j. , 1983 , image analysis and mathematical morphology , academic press s. , feldman h.a . , heitmann k. , habib s. , 2006 , mnras , 376 , 1629 r.k . , 1998 , mnras , 300 , 1057 r.k . , van de weygaert r. , 2004 , mnras , 350 , 517 n. , 1998 , the natural element method in solid mechanics . ph.d .
thesis , theoretical and applied mechanics , northwestern university , evanston , il , usa a. , van gorkom j.h . , greg m.d .
, strauss m.a . , 1996 ,
aj , 111 , 2150 r. , 1991 , voids and the geometry of large scale structure , ph.d .
thesis , leiden university r. , 1994 , a&a , 283 , 361 r. , 2002 , in modern theoretical and observational cosmology , proc .
2nd hellenic cosmology meeting , ed .
m. plionis , s. cotsakis , assl 276 , p. 119 - 272 , kluwer r. , 2007 , a&a , to be subm .
r. , icke v. , 1989 , a&a , 213 , 1 r. , van kampen e. , 1993 , mnras , 263 , 481 r. , sheth r. , platen e. , 2004 , iau colloq .
195 : outskirts of galaxy clusters : intense life in the suburbs , 58 l. , soille p. , 1991
, ieee transactions on pattern analysis and machine intelligence , vol .
13 , no . 6 , pp .
583 g. f. , 1908 , j. reine angew .
134 , 198 d.f . , 1992 ,
contouring : a guide to the analysis and display of spatial data , pergamon press
0.3 cm appendix [ app : mathmorph ] and [ app : wshedimpl ] provide some formal concepts and notations necessary for appreciating the watershed transform .
the wvf formalism introduced in this study is largely based upon concepts and techniques stemming from the field of image analysis .
although they are used within the context of the morphological analysis of spatial patterns in cosmological density fields the presentation is in terms of the original image analysis nomenclature . in this
we remind the cosmology reader to translate `` image '' into `` density field '' , `` basin '' into `` void interior '' , etc .
mathematical morphology ( mm ) is the field of image analysis which aims at characterizing an image on the basis of its geometrical structure .
it provides techniques for extracting image components which are useful for representation and description and was originally developed by g. matheron and j. serra . for more details
we refer to @xcite , also see @xcite .
it involves a set - theoretic method of image analysis and incorporates concepts from algebra ( set theory , lattice theory ) as well as from geometry ( translation , distance , convexity ) .
applications of mathematical morphology may be found in a large variety of scientific disciplines , including material science , medical imaging and pattern recognition .
a cosmological density field @xmath11 may be mapped onto an image @xmath12 , @xmath13 the image @xmath12 is a function in @xmath6-dimensional lattice space ( usually @xmath14 or @xmath15 ) @xmath16 , @xmath17 although in principle images may be continuous , in practice they usually attain a finite number of discrete values .
two important classes are : * _ binary image _ : + image with only 2 intensity values ( fig .
[ fig : mm_oper ] , top lefthand ) , @xmath18 we follow the convention to identify the binary image by the set @xmath19 for which @xmath20 . *
_ grayscale image _
: + image with a discrete number @xmath21 of values ( fig .
[ fig : mm_oper ] , bottom lefthand ) , @xmath22 mathematical morphology was originally developed for binary images , and was later extended to grayscale images .
the two basic operators of mathematical morphology are _ erosion _ and _ dilation _ of a binary image @xmath23 . in order to define these operators we need to invoke the _ translation _ and _ reflection _ of a set @xmath24 , @xmath25 the dilation or erosion of a binary image @xmath23 by a structuring element @xmath24 identifies
whether the translated set @xmath26 has an overlap with or is contained in a certain part of @xmath23 , @xmath27 in other words , dilation consists of the minkowski addition ( @xmath28 ) of the binary image @xmath23 with a structuring element @xmath24 while erosion is the minkowski substraction ( @xmath29 ) with @xmath24 .
a structuring element may be any object in @xmath16 .
an example is the circle which functioned as a structuring element in fig .
[ fig : mm_oper ] . erosion and dilation
have a number of properties : * translation invariance * global scaling invariance * addition : @xmath30 * complementarity : * erosion of a set is dilation of complement * ( and vice versa ) * adjunction relationship : @xmath31 the complementarity and adjunction relationship are two aspects of the existing duality between erosion and dilation . in general erosion and dilation induce a loss of information .
erosion followed by dilation , or vice versa , will only result in a restoration of the original image if the sets @xmath23 , @xmath32 and @xmath33 are convex .
in fact , various combinations of the erosion and dilation operators do result in new operators .
the most straightforward combination of dilation and erosion is that of the consecutive application of an erosion and a dilation .
this introduces two new operators , the _ opening _ and _ closing _ operators . on a binary image they have the effects shown in fig .
[ fig : mm_oper ] ( top centre , top righthand ) .
opening amends caps , removes small islands and opens isthmuses .
closing , on the other hand , closes channels , fills small lakes and ( partly ) the gulfs .
an additional combination of erosion and dilation is subtraction of the first from the latter , called a _
morphological gradient_. formally , an opening is an erosion followed by a dilation , a closing a dilation followed by erosion . @xmath34
\\ \textrm{closing : } & & \phi_{b } & \,\equiv\,&[(x \oplus b)\ominus b ] \\ \ \\ \end{array}\ ] ] characteristics of the opening and closing operators are : * increasing * idempotent : * applying the operator twice yields the same output * opening is anti - extensive : @xmath35 * closing is extensive : @xmath36 note that the extensivity and/or anti - extensivity of operators define the prime conditions for a _
morphological operator_. the morphological operators which we discussed above can be generalized to grayscale images .
a grayscale image is composed of subsets @xmath37 , @xmath38 with @xmath39 and @xmath40 the support of the full image .
the erosion and dilation of a grayscale image involves their application to each individual subset @xmath37 .
extension of the binary image definitions ( eqn .
[ eqn : bintrref ] ) implies the following definition wrt . a grayscale image , @xmath41 the effect of erosion on a grayscale image is the shrinking of the bright regions .
bright spots smaller than the structuring element @xmath24 disappear completely while valleys ( dark ) expand .
dilation has the opposite effect : dark regions shrink while bright regions expand .
it illustrates the duality between erosion and dilation . for our purposes
, this formal definition translates into the following practical implementation for 2-d grayscale images . given a grayscale image @xmath42 with grid elements @xmath43 , @xmath44\,+\,{\hat b}[i , j]\}\\ & & & \forall [ r , s ] \in a\\ { \mathcal f}\ominus b & \,=\,&{\min_{(i , j ) \in b}}\ \{a[r+i , s+j]\,+\,b[i , j]\}\\ & & & \forall [ r , s ] \in a\\
\ \\ \end{array}\ ] ] as in the case of binary images new operators may be defined through combinations of erosions and dilations .
the closing and opening operators are defined in exactly the same way as that for the binary images .
their effect is shown in the lower column of fig .
[ fig : mm_oper ] .
the morphological gradient @xmath45 is a dilation minus erosion operation .
the gradient operator is often used in object detection because an object is usually associated with a change in grayscale with respect to the background .
a variety of additional operators involving openings and closings may be defined .
interesting ones are the _ granulometries _ , a sequence of erosions with increasing scale , and _ distance transforms_.
the segmentation of images is defined on the basis of a distance criterion , referring to the concept of _ distance _ between subsections of an image . for appreciating the concept of watershed segmentation we consider two distance concepts , the _ geodesic _ and _ topographic _ distance .
geodesic distances are used in the case of binary images while topographic distances form the basis for the segmentation of grayscale images .
let @xmath46 be a set and x and y two points in an @xmath6-dimensional lattice space @xmath16 .
we may define the geodesic distance @xmath47 is the length of the shortest ( geometric ) path in @xmath23 connecting @xmath48 and @xmath49 ( see lefthand frame , fig .
[ fig : skiz ] ) .
accordingly , the distance between two subsets @xmath33 and @xmath24 in @xmath23 may be defined as follows .
considering the set of all paths between any of the elements of @xmath33 and those of @xmath24 , the distance @xmath50 between @xmath33 and @xmath24 is defined to be the minimum length of any of these paths .
based upon the concept of @xmath50 one may formulate a distance function of a set @xmath51 . for each point @xmath52 ,
the distance @xmath53 to its complement @xmath54 is computed .
the distance function @xmath55 is the resulting map of distance values @xmath56 .
regions whose distance @xmath56 is at least @xmath57 can be identified and equated by erosion of @xmath32 with a disk of radius @xmath57 ( defining a set @xmath58 ) . each of these regions forms a section @xmath59 ( see sec . [
app : grayscale ] ) , @xmath60 in which @xmath61 is the disk of radius @xmath57 .
the map @xmath62 may be regarded as a stack of these sections . for illustration
the distance transform of the the binary image fig .
[ fig : mm_oper ] is depicted in the central frame of fig [ fig : skiz ] .
the topographic distance @xmath63 between two points in @xmath16 is defined with respect to the image map @xmath64 . taking the limit of a continuous map @xmath12 , the topographic distance from @xmath48 to @xmath49 is the path which attains the minimum length through the `` image landscape '' , @xmath65 in this definition , the integral denotes the _ image pathlength _
@xmath66 along all paths @xmath67 in the set of all possible paths , @xmath68 .
this concept of distance is related to the geodesics of the surface @xmath12 : the path of steepest descent , specifing the track a droplet of water would follow as it flows down a mountain surface . based on the specific definition of distance , we may segment a binary image @xmath12 through the identification of the zones of influence of well - defined subsets of @xmath12 .
in general the binary image @xmath12 contains @xmath6 connected subsets @xmath69 ( @xmath70 ) ( the black regions of fig .
[ fig : mm_oper ] ) and a set x ( white region ) which contains all points @xmath48 which do not belong to any of the subsets @xmath69 , i.e. @xmath71 the geodesic zone of influence @xmath72 of a subset @xmath73 is the set of all the points @xmath74 that are stricktly closer to @xmath69 than to any other subset @xmath75 , @xmath76 the _ zone of influence _
@xmath77 of @xmath78 is the union of all influence zones of @xmath79 , @xmath80 the boundary set in @xmath23 consists of those points which do belong to @xmath23 yet are not contained in any of the zones of influence @xmath72 .
these boundary points define the geodesic _ skeleton _ @xmath81 of @xmath78 , @xmath82 in fig .
[ fig : skiz ] ( righthand frame ) the skeleton in set @xmath23 is outlined by white lines .
the skeleton is superimposed on its defining distance function landscape , its values indicated by a red colour gradient scheme ( the corresponding landscape profile is depicted in the central frame ) .
we should point out that here we follow the definition of mathematical morphology although the name of _ skeleton _ of the large scale cosmic matter distribution has been used for different be it related concepts ( see e.g. * ? ? ?
* ; * ? ? ?
it is interesting to note that if we restrict the subsets @xmath69 to single points the skeleton naturally evolves into a _ ( first - order ) voronoi tessellation_. it is the definition of the concept of skeleton @xmath81 within the specific context of grayscale images which brings us to the definition of the _ watershed transform _
( see next section ) .
grayscale images consist of a finite number of discrete levels .
this results in a slightly more complicated situation for its segmentation . in the case of a binary image
an image is segmented on the basis of a _ geodesic distance_. for the segmentation of grayscale images the distance concept needs to be generalized to that of _ topographic distance_. * definition : * each _ watershed basin _ is the collection of points which are closer in topographic distance to the defining minimum then to any other minimum .
the literature is replete with algorithms for the construction of the _ watershed transform _ of an image into its constituting _ watershed segments_. they may be divided into two classes .
one class simulates the _ watershed basin _ immersion process .
the second aims at detecting the watershed skeleton on the basis of the distance .
_ watershed by immersion _ was introduced and defined by @xcite .
the first step of the procedure concerns the identification of the minima .
formally , a minimum is a plateau at altitude @xmath83 from which it is impossible to reach a point of lower height .
starting with the lowest grayscale level @xmath84 and recursively proceeding to the highest level @xmath85 the algorithm allocates the zone of influence of each minimum by gradually filling up the surrounding catchment basin . at a particular grayscale level @xmath86 , with altitude @xmath87 ,
the algorithm has hypothetically inundated a landscape region with an altitude @xmath88 .
the total of inundated area , @xmath89 is the complement of the section @xmath90 . having arrived at level @xmath86 and proceeding to level @xmath91 the algorithm has three possibilities , * encounter a new minimum ( at level @xmath91 ) *
add new points to existing catchment basins ( condition : points connected to only one existing basin ) .
* encounter new points that belong to several basins situation ( 1 ) signals the event that a new minimum becomes active in the image .
the second option concerns the extension of an existing basin by an additional collection of points .
these are points identified at level @xmath92 which find themselves within the realm of a single basin existent at level @xmath86 .
they belong to the zone of influence of @xmath93 and find themselves embedded in its extended counterpart @xmath94 at level @xmath92 . in situation
( 3 ) more than one basin may be connected to the stack @xmath94 .
its correct subdivision is determined by computing the influence zones of all connected basins . defining @xmath95 to be the union of all catchment basins at level @xmath86 the union of catchment basins at level @xmath92 becomes @xmath96 the watershed procedure
may be viewed as iteratively computing the zone of influence at each new grey scale level . following the rationale above the final `` immersion '' definition for the watershed @xmath97 of the image @xmath12 within a domain @xmath23 is @xmath98 on completion of the procedure the union of points attached to every minimum @xmath99 in @xmath23 is equal to the union of catchment basins , @xmath100 .
the skeleton remains as the watershed segmentation .
the alternative strategy for determining the watershed transform is that of following the strict definition of segmentation by minimum topographic distance . @xcite
give a summary of the most notable schemes .
these algorithms seek to find all points ( pixels ) whose topographic distance to a particular marker - ie .
a significant minimum in the density field is the shortest amongst that to all other markers in the image .
the formalism bears some resemblance to dijkstra s graph theoretical problem of tracing the shortest path forest in a point distribution .
based on this similarity an image is seen as a connected ( di)graph in which the pixels of the image function as the nodes of the graph . each point @xmath101 is reachable from each other point @xmath102 via the graph s edges .
the latter usually define a network on the basis of @xmath103connectivities or @xmath104-connectivities .
the shortest path between two points ( nodes ) @xmath101 and @xmath102 is found by traversing the graph and keeping track of the walking cost .
critical for the procedure is the assignment of a proper measure of cost to each path . by definition
it should be a non - negative increasing function and be related to the definition of topographic distance ( eq . [ eq : topdist ] ) .
this suggests the use of the quantity @xmath105 the maximum slope linking the two pixels @xmath101 and @xmath102 .
this leads to the following _ cost function _ for the link between two neighbouring pixel @xmath101 and pixel @xmath102 , @xmath106 @xmath107 the total cost @xmath108 for a path @xmath109 connecting any two points @xmath110 and @xmath111 via the points @xmath112 will then simply be the sum @xmath113 the topographic distance @xmath114 is the infimum of @xmath115 over all paths connecting @xmath110 and @xmath111 .
@xmath116 given this definition for the topographic distance within a grayscale image we can pursue the segmentation process described in section [ app : segment ] , ultimately yielding the watershed segmentation .
we follow the watershed transform algorithm by @xcite .
their method implicitly incorporates the concept of markers .
these markers are the minima used as sources of the watershed flooding procedure . as such they form a select subgroup amongst all minima of an image @xmath12 .
the code for the watershed proceudre involves the following steps : * * initialization * all pixels of the cube are initialized and tagged to indicate they have not yet been processed .
each grayscale level is allocated a queue and all pixels are attached to the queue corresponding to their level . * * minima * each minimum plateau is tagged by a unique `` minimum tag '' .
the pixels corresponding to a minimum are inserted into the corresponding queue .
* * flooding * all pixels in the grayscale level queues are processed , starting at the lowest grayscale level .
unless a pixel is surrounded by a complex of unprocessed neighbours it will be assigned to the queue of the corresponding minimum .
pixels which also border another minimum obtain a boundary tag .
* * final stage * for any grayscale level the flooding stops when the queue has emptied .
the procedure continues with processing the pixels in the queue for the next grayscale level .
the process is finished once all level queues have been emptied .
voronoi clustering models are a class of heuristic models for cellular distributions of matter @xcite .
they use the voronoi tessellation as the skeleton of the cosmic matter distribution , identifying the structural frame around which matter will gradually assemble during the emergence of cosmic structure .
the interior of voronoi cells correspond to voids and the voronoi planes with sheets of galaxies .
the edges delineating the rim of each wall are identified with the filaments in the galaxy distribution .
the most outstanding structural elements are the vertices , corresponding to the very dense compact nodes within the cosmic web , the rich clusters of galaxies .
we distinguish two different yet complementary approaches .
one is the fully heuristic approach of voronoi element models .
they are particularly apt for studying systematic properties of spatial galaxy distributions confined to one or more structural elements of nontrivial geometric spatial patterns .
the second , supplementary , approach is that of the voronoi kinematic models , which attempts to `` simulate '' foamlike galaxy distributions on the basis of simplified models of the evolution of the megaparsec scale distribution .
the voronoi kinematic model is based upon the notion that voids play a key organizational role in the development of structure and make the universe resemble a soapsud of expanding bubbles @xcite .
it forms an idealized and asymptotic description of the outcome of the cosmic structure formation process within gravitational instability scenarios with voids forming around a dip in the primordial density field .
for plausible structure formation scenarios , most notably the concordance @xmath3cdm cosmology , this evolution will proceed hierarchically . a detailed assessment of the resulting void hierarchy by @xcite demonstrated that this leads to a selfsimilarly evolving peaked void size distribution .
by implication , most voids have comparable sizes and excess expansion rates .
the geometrically interesting implication is that the asymptotic limit of the `` peaked '' void distribution degenerating into one of only one characteristic void size .
it yields a cosmic matter distribution consisting of _ equally sized _ and expanding spherical voids , a geometrical configuration which is precisely that of a voronoi tessellation .
this is translated into a scheme for the displacement of initially randomly distributed galaxies within the voronoi skeleton ( see sect .
[ app : vorclustform ] for a detailed specification ) . within a void ,
the mean distance between galaxies increases uniformly in the course of time .
when a galaxy tries to enter an adjacent cell , the velocity component perpendicular to the cell wall disappears .
thereafter , the galaxy continues to move within the wall , until it tries to enter the next cell ; it then loses its velocity component towards that cell , so that the galaxy continues along a filament .
finally , it comes to rest in a node , as soon as it tries to enter a fourth neighbouring void .
0.5truecm 0.2 cm the initial conditions for the voronoi galaxy distribution are : * distribution of @xmath117 nuclei , _ expansion centres _
, within the simulation volume @xmath118 .
the location of nucleus @xmath99 is @xmath119 . *
generate @xmath21 model galaxies whose initial locations , @xmath120 @xmath121 , are randomly distributed throughout the sample volume @xmath118 .
* of each model galaxy @xmath6 determine the voronoi cell @xmath122 in which it is located , ie .
determine the closest nucleus @xmath123 .
all different voronoi models are based upon the displacement of a sample of @xmath21 `` model galaxies '' .
the initial spatial distribution of these @xmath21 galaxies within the sample volume @xmath118 is purely random , their initial locations @xmath120 ( @xmath124 defined by a homogeneous poisson process .
a set of @xmath117 nuclei within the volume @xmath118 corresponds to the cell centres , or _
expansion centres _ driving the evolving matter distribution .
the nuclei have locations @xmath119 @xmath125 . following the specification of the initial positions of all galaxies , the second stage of the procedure consists of the calculation of the complete voronoi track for each galaxy @xmath126 ( sec .
[ sec : vortrack ] ) .
once the voronoi track has been determined , for any cosmic epoch @xmath127 one may determine the displacement @xmath128 that each galaxy has traversed along its path in the voronoi tessellation ( sec .
[ sec : vortrack ] ) .
the first step of the formalism is the determination for each galaxy @xmath6 the voronoi cell @xmath122 in which it is initially located , ie . finding the nucleus @xmath123 which is closest to the galaxies initial position @xmath120 . in the second step
the galaxy @xmath6 is moved from its initial position @xmath120 along the radial path emanating form its expansion centre @xmath129 , ie . along the direction defined by the unity vector @xmath130 .
dependent on how far the galaxy is moved away from its initial location @xmath120 - set by the _ radius of expansion _
@xmath131 to be specified later the galaxies path @xmath128 ( see fig . [
fig : vorkinmschm ] ) may be codified as @xmath132 in which the four different components are : 1 .
@xmath130 : unity vector path within voronoi cell @xmath122 2 .
@xmath133 : unity vector path within voronoi wall @xmath134 3 .
@xmath135 : unity vector path along voronoi edge @xmath136 4 .
vertex @xmath137 the identity of the neighbouring nuclei @xmath123 , @xmath138 , @xmath139 and @xmath140 , and therefore the identity of the cell @xmath122 , the wall @xmath134 , the edge @xmath136 and the vertex @xmath137 , depends on the initial location @xmath120 of the galaxy , the position @xmath141 of its closest nucleus and the definition of the galaxies path within the voronoi skeleton .
the cosmic matter distribution at a particular cosmic epoch is obtained by calculating the individual displacement factors @xmath142 for each model galaxy .
these are to be derived from the global `` void '' expansion factor @xmath143 .
this factor parameterizes the cosmic epoch and specifies the ( virtual ) radial path of the galaxy from its expansion centre @xmath123 . at first , while still within the cell s interior , the galaxy proceeds according to @xmath144 as a result within a void the mean distance between galaxies increases uniformly in the course of time .
once the galaxy tries to enter an adjacent cell @xmath138 and reaches a voronoi wall , i.e. when @xmath145 , the galaxy s motion will be constrained to the radial path s component within the wall @xmath134 .
the galaxy moves along the wall until the displacement supersedes the extent of the path within the wall and it tries to enter a third cell @xmath139 , i.e. when @xmath146 .
subsequently , it moves along @xmath136 until it comes to rest at the node @xmath137 as soon as it tries to enter a fourth neighbouring void @xmath140 when @xmath147 .
a finite thickness is assigned to all voronoi structural elements .
the walls , filaments and vertices are assumed to have a gaussian radial density distribution specified by the widths @xmath148 of the walls , @xmath149 of the filaments and @xmath150 of the vertices .
voronoi wall galaxies are displaced according to the specified gaussian density profile in the direction perpendicular to their wall .
a similar procedure is followed for the voronoi filament galaxies and the voronoi vertex galaxies . as a result
the vertices stand out as three - dimensional gaussian peaks .
for a detailed specification of the dtfe density field procedure we refer to @xcite . in summary ,
the dtfe procedure for density field reconstruction from a discrete set of points consists of the following steps : 1 .
* point sample * + given that the point sample is supposed to represent an unbiased reflection of the underlying density field , it needs to be a general poisson process of the ( supposed ) underlying density field .
* boundary conditions * + the boundary conditions will determine the delaunay and voronoi cells that overlap the boundary of the sample volume .
dependent on the sample at hand , a variety of options exists : 1 .
_ empty boundary conditions : _ + outside the sample volume there are no points
_ periodic boundary conditions : _ + the point sample is supposed to be repeated periodically in boundary boxes , defining a toroidal topology for the sample volume .
_ buffered boundary conditions : _ + the sample volume box is surrounded by a bufferzone filled with a synthetic point sample .
* delaunay tessellation * + construction of the delaunay tessellation from the point sample . while we also still use the voronoi - delaunay code of @xcite , at present
there is a number of efficient library routines available .
particularly noteworthy is the ` cgal ` initiative , a large library of computational geometry routines + 4 .
* field values point sample * + the estimate of the density at each sample point is the normalized inverse of the volume of its contiguous voronoi cell @xmath151 of each point @xmath86 .
the _ contiguous voronoi cell _ of a point @xmath86 is the union of all delaunay tetrahedra of which point @xmath86 forms one of the four vertices .
we recognize two applicable situations : + : + the point sample is an unbiased sample of the underlying density field .
typical example is that of @xmath21-body simulation particles . for @xmath152-dimensional space
the density estimate is , @xmath153 with @xmath154 the weight of sample point @xmath86 , usually we assume the same `` mass '' for each point . + : + sampling density according to specified selection process . the non - uniform sampling process is quantified by an a priori known selection function @xmath155
this situation is typical for galaxy surveys , @xmath155 may encapsulate differences in sampling density @xmath156 as function of sky position @xmath157 , as well as the radial redshift selection function @xmath158 for magnitude- or flux - limited surveys . for @xmath152-dimensional space
the density estimate is , @xmath159 5 .
* field gradient * + calculation of the field gradient estimate @xmath160 in each @xmath152-dimensional delaunay simplex @xmath99 ( @xmath161 : tetrahedron ; @xmath162 : triangle ) by solving the set of linear equations for the field values @xmath163 at the positions @xmath164 of the @xmath165 tetrahedron vertices , + @xmath166 evidently , linear interpolation for a field @xmath167 is only meaningful when the field does not fluctuate strongly .
6 . * interpolation*. + the final basic step of the dtfe procedure is the field interpolation .
the processing and postprocessing steps involve numerous interpolation calculations , for each of the involved locations @xmath168 . given a location @xmath168 , the delaunay tetrahedron @xmath99 in which it is embedded is determined . on the basis of the field gradient @xmath160 the field value is computed by ( linear ) interpolation , @xmath169 in principle , higher - order interpolation procedures are also possible .
two relevant procedures are : + for nn - interpolation see @xcite and @xcite .
implementation of natural neighbour interpolations is presently in progress .
* processing*. + though basically of the same character , for practical purposes we make a distinction between straightforward processing steps concerning the production of images and simple smoothing filtering operations and more complex postprocessing .
the latter are treated in the next item .
basic to the processing steps is the determination of field values following the interpolation procedure(s ) outlined above .
straightforward `` first line '' field operations are _ image reconstruction _ and _ smoothing / filtering_. 1 .
* image reconstruction*. + for a set of image points , usually grid points , determine the image value . formally the average field value within the corresponding gridcell . in practice
a few different strategies may be followed + the choice of strategy is mainly dictated by accuracy requirements .
for wvf we use the monte carlo approach in which the grid density value is the average of the dtfe field values at a number of randomly sampled points within the grid cell .
smoothing * and * filtering * : + a range of filtering operations is conceivable .
two of relevance to wvf are : of the field @xmath170 + convolution of the field @xmath170 with a filter function @xmath171 , usually user - specified , @xmath172 + ( see sec . [ sec : natnghb ] ) .
8 . * post - processing*. + the real potential of dtfe fields may be found in sophisticated applications , tuned towards uncovering characteristics of the reconstructed fields .
an important aspect of this involves the analysis of structures in the density field .
the wvf formalism developed in this study is an obvious example . | on megaparsec scales the universe is permeated by an intricate filigree of clusters , filaments , sheets and voids , the cosmic web . for the understanding of its dynamical and hierarchical history it is crucial to identify objectively its complex morphological components .
one of the most characteristic aspects is that of the dominant underdense voids , the product of a hierarchical process driven by the collapse of minor voids in addition to the merging of large ones .
in this study we present an objective void finder technique which involves a minimum of assumptions about the scale , structure and shape of voids .
our void finding method , the watershed void finder ( wvf ) , is based upon the watershed transform , a well - known technique for the segmentation of images .
importantly , the technique has the potential to trace the existing manifestations of a void hierarchy .
the basic watershed transform is augmented by a variety of correction procedures to remove spurious structure resulting from sampling noise .
this study contains a detailed description of the wvf .
we demonstrate how it is able to trace and identify , relatively parameter free , voids and their surrounding ( filamentary and planar ) boundaries .
we test the technique on a set of kinematic voronoi models , heuristic spatial models for a cellular distribution of matter .
comparison of the wvf segmentations of low noise and high noise voronoi models with the quantitatively known spatial characteristics of the intrinsic voronoi tessellation shows that the size and shape of the voids are succesfully retrieved .
wvf manages to even reproduce the full void size distribution function .
[ firstpage ] cosmology : theory large - scale structure of universe methods : data analysis |
in this paper we proceed with the study of the random interlacements introduced by a.s .
sznitman in @xcite .
this model is for instance related to the trace left by a random walk on the discrete torus @xmath2 ( @xmath3 ) and on the discrete cylinder @xmath4 ( @xmath5 ) when the walk runs for times of order @xmath6 and @xmath7 respectively , see @xcite and @xcite .
intuitively , random interlacements describe the microscopic ` texture in the bulk ' left by the random walk in these contexts , see @xcite and @xcite . in @xcite ,
this model is the main ingredient to improve the upper bound on the disconnection time of a large discrete cylinder , and in @xcite they are used to extend the the lower bound obtained in @xcite to the case @xmath5 . loosely speaking ,
the interlacement at level @xmath1 ( denoted by @xmath8 ) is given by the trace left by a poisson cloud of doubly infinite random walk trajectories in @xmath9 , where @xmath1 controls the density of the cloud .
the so - called vacant set at level @xmath1 ( denoted with @xmath10 ) is the complement of the interlacement , or in other words , the set of sites in @xmath9 which are not visited by any trajectory in this cloud .
the random sets @xmath8 are constructed simultaneously for all values of @xmath1 on the same probability space @xmath11 .
although we postpone the precise description of the process to section [ sec : review ] , we state here a characterization of the law @xmath12 of the indicator function of @xmath10 , regarded as a random element of @xmath13 .
namely , @xmath12 is the only probability measure on @xmath13 such that @xmath14 = \exp \{-u \ , \text{cap}(k)\ } , \text { for all finite sets } k \subset \mathbb{z}^d,\ ] ] where @xmath15 denotes the capacity of @xmath16 ( see ( [ eq : cap ] ) ) and @xmath17 stand for the canonical coordinates on @xmath13 , see remark 2.2 2 ) of @xcite .
percolation of the vacant set of random interlacements presents a phase transition in the parameter @xmath1 .
more precisely , it is known that there is a @xmath18 , such that when @xmath19 , @xmath10 contains @xmath20-a.s .
an infinite connected component , see @xcite theorem 3.4 , and when @xmath21 , @xmath10 almost surely consists of finite clusters , see @xcite theorem 3.5 .
moreover it is known that if an infinite connected component of the vacant set exists , it is almost surely unique , see @xcite theorem 1.1 . in this article
we further investigate this model in the regime of small intensity @xmath1 , and give a partial answer to the question posed in @xcite , remark 3.5 2 ) . denoting by @xmath22 the closed ball with respect to the @xmath23 norm on @xmath9 with radius @xmath24 and centered at the origin , our main theorem [ th : main ] states that @xmath25 \geqslant 1- c_1 e^{-c_2 n^\alpha } , \end{split}\ ] ] when @xmath26 and @xmath27 .
here , @xmath28 and @xmath29 are positive constants solely depending on @xmath30 and @xmath31 , see theorem [ th : main ] .
we will now underline the importance of this main result by stating some of its consequences , see also section [ sec : applications ] .
as a first application of ( [ eq : maintheorem ] ) , one can show the ubiquity of @xmath32 , the unique infinite component in @xmath10 .
more precisely , for @xmath33 , @xmath34 and @xmath27 , with the same notation as in ( [ eq : maintheorem ] ) , we show in theorem [ th : cinftydense ] that @xmath35 \geqslant 1 - c_3 \exp(-c_4 n^\alpha ) , \end{split}\ ] ] where @xmath36 and @xmath37 are positive constants solely depending on @xmath30 and @xmath31 .
if @xmath38 stands for the connected component of @xmath10 containing the origin , with ( [ eq : statecinftydense ] ) , we are able to control the diameter and the volume of @xmath38 , when this cluster is finite .
this answers in part a question of @xcite , remark 4.4 3 ) . indeed in theorems
[ th : taildiam ] and [ th : tailvol ] , we show that , for @xmath0 and @xmath27 , @xmath39 \leqslant
c_6 \exp(-c_7 n^\alpha ) \,\,\ , ( n \geqslant 1),\\ \label{eq : volestimate } c_8(u ) \exp \big(-c_9v^{\frac{d-2}{d } } \log v \big ) \leqslant \mathbb{p}\big [ v \leqslant |\mathcal{c}_0^u| < \infty \big ] \leqslant c_6 \exp(-c_7 v^{\alpha / d } ) \,\,\ , ( v \geqslant 1),\end{gathered}\ ] ] where all the constants appearing above are positive and , except for @xmath40 and @xmath41 , only depend on @xmath31 .
the estimate ( [ eq : volestimate ] ) has the flavor of the open problem posed in @xcite , remark 4.7 1 ) . as another application of ( [ eq : maintheorem ] )
we prove in theorem [ th : logcomp ] that for @xmath0 , @xmath27 and any @xmath42 and integer @xmath43 , @xmath44 = 0 . \end{split}\ ] ] let us mention that in the case of bernoulli independent site percolation , similar results are already known to hold under weaker hypotheses .
however , the techniques used in the bernoulli context are not directly applicable to random interlacements , see remark [ rem : bernoulli ] 1 ) and 2 ) .
some difficulties we mention here are the high dependence featured by the measure @xmath12 , see @xcite ( 1.68 ) , and the fact that for every @xmath45 the interlacement at level @xmath1 is almost surely an infinite connected subset of @xmath9 , see corollary 2.3 of @xcite .
this property motivates the precise formulation of ( [ eq : maintheorem ] ) .
as we further explain in remark [ rem : bernoulli ] 3 ) , one can hope to find a vacant path joining the boundary of two large connected sets in @xmath46 , but not necessarily a vacant path joining the sets themselves .
we now describe the strategy adopted to prove ( [ eq : maintheorem ] ) .
in essence , the proof is based on the following two basic ingredients : * if the interlacement separates two macroscopic components of a box , then in many sub - boxes it also separates some macroscopic components . in other words , the property of separating macroscopic components ` cascades to finer scales ' , * a fixed number of random walk paths can hardly separate macroscopic components in a large box .
these claims are made precise and proved in sections [ sec : coarse ] and [ sec : local ] , respectively . in section
[ sec : proof ] we prove ( [ eq : maintheorem ] ) using these results .
consider the following sequence of scales @xmath47 where @xmath48 are integers .
we now provide a short overview of the proof of and give an idea of the role plaid by the parameters @xmath49 , @xmath50 and @xmath1 .
our aim is to bound the probability of the so - called separation event ( in essence the complement of the event appearing in ) . in the above mentioned ingredient @xmath51 of the proof ,
we show that separation ` cascades down to finer scales ' .
this allows us to bound the probability of the separation event in a box at scale @xmath52 ( having diameter @xmath53 ) by the probability that such separation occurs simultaneously in @xmath54 well - spaced boxes at the bottom scale ( each with diameter @xmath50 ) .
the bound on the latter probability has to be good enough to offset the number of possible choices for the boxes at the bottom scale .
the combinatorial complexity of this choice roughly amounts to choosing a binary sub - tree of depth @xmath55 in a rooted tree having @xmath56 descendants at each generation .
we thus need to control the probability of the simultaneous occurrence of separation events in each of the @xmath54 boxes at the bottom scale @xmath50 . for any such
given collection of boxes , we first bound the mutual dependence of the separation events in each of them .
for this purpose , we keep track of the number of excursions that the random walk trajectories ( composing the random interlacements ) perform between these boxes .
we now choose a large enough @xmath49 , consequently increasing the mutual distance between the boxes in this collection .
in this fashion we are able to make the large deviation cost of observing too many excursions offset the combinatorial complexity of the choices of the @xmath54 boxes .
this step is delicate because increasing @xmath49 also increases this combinatorial complexity . for this competition to work in our favor , we need to impose the boxes at the bottom scale to receive an average number of excursions ( say @xmath57 ) such that @xmath58 .
we take @xmath59 since this will do the job , see also remark [ rem : g100 ] 1 ) .
the dependence control described in this paragraph works for every @xmath50 , once @xmath1 is chosen small enough depending on @xmath49 and @xmath50 .
the previous step enables us to treat the separation event in the @xmath54 boxes as roughly independent .
we thus choose @xmath50 large enough so that the probability of these essentially independent @xmath54 separation events in boxes at scale @xmath50 also offsets the combinatorial choice of the boxes . now the above ingredient @xmath60 of the proof comes into play .
when @xmath50 is chosen large enough ( depending on @xmath49 ) , a fixed number ( as we said , @xmath61 will do the job ) of independent random walk excursions can hardly separate components in a given box of size @xmath50 .
now that @xmath50 is fixed , we can choose @xmath62 small to make sure that observing an average of @xmath61 random walks per box in the collection above is indeed a large deviation as we described .
we now give a more precise description of the proof .
for each depth @xmath52 , we partition @xmath9 into boxes of diameter @xmath53 and label the boxes in this partition with a set of indices @xmath63 . for a given box ( say indexed by @xmath64 ) , we consider the random variable @xmath65 , which loosely speaking indicates the separation of two macroscopic connected sets of this box by @xmath8 .
we refer to ( [ eq : chi ] ) for the precise definition . as a reduction step
, we prove that it is enough to establish ( [ eq : maintheorem ] ) in the case where @xmath66 and @xmath67 is taken along the sequence @xmath53 , @xmath68 .
in other words , according to proposition [ prop : reduct ] , in order to prove ( [ eq : maintheorem ] ) we only need to obtain a bound ( decaying exponentially in @xmath54 ) on the probability that @xmath69 when @xmath64 .
we also rely on the concept of a skeleton , which captures the possible ways in which the separation event can propagate to finer scales . roughly speaking
a skeleton is a set @xmath70 of indices in the finest scale ( @xmath71 ) satisfying some conditions on the distance between the boxes indexed by @xmath70 , see definition [ def : skeleton ] .
the notion of skeleton resembles the wiener criterion , see for instance @xcite , theorem 2.2.5 p.55 .
the main purpose of this definition appears in ( [ eq : boundexcursion ] ) , where we derive a bound on the probability that a random walk , starting in one of the boxes of a skeleton , hits another box of the skeleton before escaping to infinity .
the ingredient @xmath51 of the proof , which we call ` coarse graining ' argument , is the content of theorem [ th : coarse2 ] , section [ sec : coarse ] . loosely speaking ,
this theorem states that @xmath72 hence , the problem is reduced to estimating the probability that the ` separation event ' ( @xmath73 ) occurs simultaneously for the @xmath54 indices @xmath74 in a given skeleton @xmath70 as above .
this bound has to be able to offset the combinatorial complexity factor @xmath75 .
let us now indicate how the above bound is related to the second ingredient , which we call ` local estimates ' .
for this , fix a skeleton @xmath71 and a collection of @xmath54 boxes associated to indices @xmath76 in @xmath70 .
loosely speaking , we use a large deviation estimate to bound the total number of excursions performed between different boxes of this collection by all the interlacement trajectories , see ( [ eq : boundfirst ] )
. then we condition each of these excursions on their return and departure points from different neighborhoods of each box .
this procedure will reduce our problem to the analysis of what happens in the surroundings of one fixed box of diameter @xmath50 .
the ingredient @xmath60 of the proof is obtained in section [ sec : local ] .
it can be summarized as follows : @xmath77 moreover , this estimate is uniform on the points in which we condition these random walks to enter and exit a large neighborhood of the box .
this is the content of theorem [ th : local ] , section [ sec : local ] and is the last piece to establish ( [ eq : maintheorem ] ) .
theorem [ th : local ] is the only part of the proof of our main result in which we need the hypothesis @xmath0 . finally , let us outline of the proof of ( [ eq : statelocal ] ) .
first we introduce the definition of a cut - point for a double infinite trajectory , see ( [ eq : cuttime ] ) . loosely speaking ,
we regard each random walk excursion as a finite set of ` sausages ' connected by cut - points . given two connected subsets @xmath78 and @xmath79 of a box , we show that : @xmath80 see corollary [ lem : chifillx ] . roughly speaking , we construct this path by `` traveling along the boundaries of the sausages '' .
finally we show that with high probability ( as @xmath50 grows ) the diameters of the ` sausages ' are small when compared with @xmath50 and the excursions performed in the box @xmath81 are mutually far apart , so that they can be treated separately , see lemma [ lem : farapart ] .
this article is organized as follows . in section [ sec : review ] we give a precise description of the random interlacements and state some results which are used throughout the article . in section [ sec : applications ] , our main theorem [ th : main ] is stated .
in addition , we derive several applications of theorem
[ th : main ] . in theorems
[ th : cinftydense ] , [ th : taildiam ] , [ th : tailvol ] and [ th : logcomp ] we prove ( [ eq : statecinftydense ] ) , ( [ eq : diamestimate ] ) , ( [ eq : volestimate ] ) and ( [ eq : statelogcomp ] ) respectively . in section [ sec : proof ]
we prove the main theorem [ th : main ] assuming theorems [ th : coarse2 ] and [ th : local ] and lemma [ lem : indicespath ] , which are proved in the subsequent sections .
the main result of section [ sec : coarse ] is theorem [ th : coarse2 ] , which implements the ` coarse graining ' argument ( [ eq : coarse ] ) used to reduce the problem to a microscopic scale . in section [ sec : local ] we describe the local picture of the process ( see ( [ eq : statelocal ] ) ) .
this is the content of theorem [ th : local ] .
finally we comment on our use of constants . throughout this article , @xmath82 and @xmath83
will be used to denote positive constants depending only on @xmath31 ( except when explicitly mentioned ) , which can change from place to place .
we write @xmath84 for fixed positive constants ( also depending only on @xmath31 ) , which refer to their first appearance in the text . *
acknowledgments * - we are grateful to alain - sol sznitman for important suggestions and encouragement .
in this section we introduce some notation and describe the model of random interlacements . in addition , we recall some useful facts concerning the model . for @xmath85 , we write @xmath86 for the largest integer smaller or equal to @xmath57 and recall that @xmath87 , \text { for all } a , b \in \mathbb{z } \text { and } t \in [ 0,1].\ ] ] we denote by @xmath88 the canonical basis of @xmath89 and write @xmath90 for the corresponding orthogonal projections . for @xmath91 we denote by @xmath92 the element @xmath93 of @xmath9 such that @xmath94 for @xmath95 .
given @xmath96 we write @xmath97 if they are orthogonal for the usual scalar product .
we let @xmath98 and @xmath99 respectively denote the @xmath23 and the @xmath100 norms on @xmath89 and @xmath101 stand for the closed @xmath23-ball in @xmath9 , i.e. @xmath102 .
we say that two points @xmath103 are neighbors if @xmath104 ( we also write @xmath105 ) and if @xmath106 we say that @xmath93 and @xmath107 are @xmath108-neighbors ( and write @xmath109 ) . these definitions respectively induce the notions of connectedness and @xmath108-connectedness in @xmath9 .
if @xmath110 , we denote by @xmath111 its complement , by @xmath112 its cardinality and by @xmath113 the @xmath114-neighborhood of @xmath16 for the @xmath23-distance , i.e. the union of the balls @xmath101 for @xmath115 .
the diameter of @xmath16 ( denoted by @xmath116 ) is the supremum of @xmath117 with @xmath118 .
we define the boundary @xmath119 ( respectively the @xmath108-boundary @xmath120 ) by @xmath121 ( respectively by @xmath122 ) .
analogously , we define the interior boundary @xmath123 .
we also write @xmath124 and if @xmath16 is finite , we denote by @xmath125 the smallest box containing @xmath16 ( by box we mean a set of type @xmath126 \times \dots \times [ a_d , b_d ] \cap \mathbb{z}^d$ ] ) . for @xmath127 , the distance @xmath128 is given by @xmath129 while @xmath130 .
we will need the following [ def : fill ] given a finite set @xmath131 , we define @xmath132 as the complement of the unique unbounded connected component of @xmath133 .
two important features of the set @xmath132 are stated in the following theorems .
@xmath134 for the proofs , see for instance , @xcite , lemma ( 2.23 ) p.139 and @xcite , lemma 2.1 . during this article the term _ path _ ( respectively @xmath108-_path _ ) always denote finite , nearest neighbor ( resp .
@xmath135-neighbor ) paths , i.e. some @xmath136 such that @xmath137 ( resp .
@xmath138 ) for @xmath139 . in this case
we say that the length of @xmath140 is @xmath141 and denote it by @xmath142 .
we denote with @xmath143 and @xmath144 the spaces of infinite , respectively doubly infinite , transient trajectories @xmath145 { } \infty \big\},\\ w = \big\{w:\mathbb{z } \rightarrow \mathbb{z}^d ; w(l ) \leftrightarrow w(l+1 ) , \text { for each } l \in \mathbb{z } \text { and } \lvert w(l ) \rvert \xrightarrow[|l| \rightarrow \infty ] { } \infty \big\}. \end{split}\ ] ] we endow them with the @xmath146-algebras @xmath147 and @xmath148 generated by the coordinate maps @xmath149 and @xmath150 . for @xmath151 ( or @xmath144 ) , we write @xmath152}$ ] for the set @xmath153\}$ ] and analogously for @xmath154}$ ] , @xmath155 and @xmath156 . we state a useful link between @xmath132 and paths in @xmath143 .
namely , [ eq : charfill ] @xmath157 _ proof .
_ we first show that the complement of the above set is included in @xmath158 .
indeed , the existence of @xmath159 with @xmath160 implies that @xmath161 belongs to the infinite connected component of @xmath133 and hence to @xmath162 .
conversely , since @xmath162 is infinite and connected , given @xmath163 one can find a trajectory @xmath159 starting in @xmath164 which is disjoint from @xmath132 .
the lemma [ eq : charfill ] follows .
@xmath165 we also introduce the entrance and exit times of a finite set @xmath110 @xmath166 and for @xmath151 , we define the hitting time of @xmath16 @xmath167 let @xmath168 stand for the time shift given by @xmath169 ( where @xmath170 could also be a random time ) . given finite sets @xmath171 , we consider on @xmath143 and @xmath144 the sequence or returns to @xmath172 and departures from @xmath173 @xmath174 notice that the stopping time @xmath175 is also defined in ( [ eq : entranceexit ] ) for trajectories in @xmath144 . for @xmath176 , ( recall that @xmath3 ) we can define the law @xmath177 of a simple random walk starting on @xmath93 on the space @xmath178 .
if @xmath179 is a measure on @xmath9 , we write @xmath180 .
their expectations are respectively denoted by @xmath181 and @xmath182 . in some calculations
, we may consider different dimensions , in this case we will explicitly write @xmath183 to avoid confusion .
we need some estimates in the hitting probability of a given set .
first , let us define the green function @xmath184 , \text { for } x , y \in \mathbb{z}^d.\ ] ] we refer to @xcite , theorem 1.5.4 p.31 for the following estimate @xmath185 we will use the following inequalities : @xmath186 \leqslant \sum_{y\in k } g(x , y ) \big/ \inf_{z\in k } \big(\sum_{y \in
k } g(z , y)\big),\ ] ] see @xcite ( 1.9 ) .
they follow by considering the bounded martingale @xmath187 and remarking that it converges in @xmath188 towards @xmath189 .
the equality between the starting value of this martingale and the expectation ( with respect to @xmath177 ) of its limit leads to the inequalities above . using and theorem 1.5.4 of @xcite , p.31 , we conclude that , for @xmath93 such that @xmath190 , @xmath191 \leqslant \frac{c r^d / ( |x|-r)^{d-2}}{\sum_{h=1}^{r/2 } c '' h^{d-1}/h^{d-2 } } \leqslant c \big(\frac{r}{|x| } \big)^{d-2}\ ] ] note that in the first inequality , we have split the last sum appearing in according to the @xmath23 distance from @xmath164 to @xmath107 .
see also ( 2.16 ) and proposition 2.2.2 in @xcite , p.53 .
moreover , using the invariance principle , we obtain that @xmath192 \leqslant c_{10 } < 1.\ ] ] we introduce , for a finite @xmath110 , the equilibrium measure @xmath193 , \text { for } x \in \mathbb{z}^d,\ ] ] the capacity of @xmath16 @xmath194 and the normalized equilibrium measure @xmath195 we mention the following bound on the capacity of a ball of radius @xmath114 @xmath196 let @xmath197 stand for the space of doubly infinite trajectories in @xmath144 modulo time shift , @xmath198 endowed with the @xmath146-algebra @xmath199 which is the largest @xmath146-algebra making the canonical projection @xmath200 measurable . for a finite set @xmath110 , we denote as @xmath201 the set of trajectories in @xmath144 which meet the set @xmath16 and define @xmath202 .
now we are able to describe the intensity measure of the poisson point process which governs the random interlacements . for a finite set @xmath110 , we consider the measure @xmath203 in @xmath204 supported in @xmath201 such that , given @xmath205 and @xmath115 , @xmath206 = p_x[a,\tilde h_k = \infty]p_x[b].\ ] ] theorem 1.1 of @xcite establishes the existence of a unique @xmath146-finite measure @xmath207 in @xmath197 such that , @xmath208 we then introduce the spaces of point measures on @xmath209 and @xmath210 @xmath211 ) < \infty\\ & \mbox { for every finite } k \subset \mathbb{z}^d \mbox { and } u\geqslant 0 \bigg\}. \end{split}\\ & \begin{split } m = \bigg\{\mu = \sum_{i \in i } \delta_{(w_i , u_i ) } ; & i \subset \mathbb{n } , w_i \in w_+ , u_i \in \mathbb{r}_+ \mbox { and } \\ & \omega(w_+ \times [ 0,u ] ) < \infty \mbox { for every } u\geqslant 0 \bigg\ } , \end{split } \end{split}\ ] ] endowed with the @xmath146-algebras @xmath212 and @xmath213 generated by the evaluation maps @xmath214 for @xmath215 and @xmath216 for @xmath217 . here @xmath218 denotes the borel @xmath146-algebra .
we let @xmath20 be the law of a poisson point process on @xmath219 with intensity measure @xmath220 , where @xmath221 denotes the lebesgue measure on @xmath222 .
given @xmath223 , we define _ the interlacement _ and the _ vacant set _ at level @xmath1 respectively as the random subsets of @xmath9 : @xmath224 we introduce the critical value @xmath225 = 0\},\ ] ] c.f .
@xcite , ( 0.13 ) .
it is known that for all @xmath3 , @xmath226 see @xcite , theorem 3.5 and @xcite , theorem 3.4 . moreover ,
its is also proved that if existent , the infinite connected component of the vacant set must be unique , see @xcite , theorem 1.1 . for a finite set @xmath110 , we define the law @xmath227 on @xmath228 of a poisson point process in @xmath210 with intensity measure @xmath229 .
the point processes defined above are related by the following .
consider , for a finite set @xmath110 , the map @xmath230 defined as @xmath231 as well as the map @xmath232 defined via @xmath233 for @xmath234 non - negative , measurable in @xmath210 . one can prove that @xmath235 see @xcite proposition 1.3 . and defining the point process @xmath236 on @xmath219 with values in the set of finite point measures on @xmath178 @xmath237 ) , \text { for } \omega \in \omega,\ ] ] we have @xmath238= \exp\{u e_{e_k}[e^{- g } -1]\ } , \\
\text { for every non - negative $ \mathcal{w}_+$-measurable function $ g$}. \end{array}\ ] ] to see why this holds , define @xmath239 by @xmath240 and use ( 1.20 ) and ( 1.43 ) of @xcite .
in this section we state our main result , theorem [ th : main ] which translates ( [ eq : maintheorem ] ) and , although we postpone the proof of this theorem to the next section , we now establish some of its consequences . in theorems
[ th : cinftydense ] we prove the existence of a unique infinite component in @xmath10 , for @xmath0 and @xmath1 small enough , and show that with high probability this component is ` dense ' in the sense of ( [ eq : statecinftydense ] ) .
theorems [ th : taildiam ] and [ th : tailvol ] respectively provide estimates on the tail distribution of the diameter and the volume of the vacant component containing the origin when it is finite , see ( [ eq : diamestimate ] ) and ( [ eq : volestimate ] ) .
finally , we bound the probability of finding a vacant component contained in @xmath46 with diameter at least @xmath241 , as in ( [ eq : statelogcomp ] ) .
this appears in theorem [ th : logcomp ] .
[ def : separate ] we say that two subsets @xmath78 and @xmath79 of @xmath9 are separated by @xmath242 in @xmath243 if 1 .
@xmath244 and 2 .
every path in @xmath243 joining @xmath245 to @xmath246 meets @xmath242 .
we remark that @xmath247 we now state the main result of the present article .
[ th : main ] ( @xmath0 ) there are @xmath248 and @xmath249 such that , for every @xmath250 , @xmath251 < c_1 \cdot \exp(-c_2 n^\alpha),\ ] ] for all @xmath34 and @xmath27 , where @xmath252 and @xmath253 . note that the event appearing above decreases with @xmath30 . throughout the rest of this section , @xmath254 and @xmath255 stand for the values appearing above .
before going into the proof of the main theorem , we illustrate some of its applications . as a first consequence of theorem [ th : main ] , we show that , for @xmath27 , there is almost surely a unique infinite connected component @xmath32 of @xmath10 and with overwhelming probability @xmath32 neighbors all the macroscopic connected subsets of @xmath46 .
more precisely , [ th : cinftydense ] ( @xmath0 ) with @xmath254 and @xmath255 as in theorem [ th : main ] , for every @xmath27 , there is @xmath20-a.s . , a unique infinite connected component @xmath32 of @xmath10 , and for every @xmath250 , @xmath256 \geqslant 1 - c_3 \cdot \exp(-c_4 n^\alpha ) , \end{split}\ ] ] for all @xmath34 .
again , @xmath257 and @xmath258 .
[ rem : u*pos ] theorem [ th : cinftydense ] yields an alternative proof that @xmath259 when @xmath0 ( in fact @xmath260 ) .
the positivity of @xmath261 appeared first in @xcite , theorem 4.3 for @xmath262 .
later the result was established for all @xmath3 , see theorem 3.4 of @xcite .
_ proof of theorem [ th : cinftydense ] .
_ we can suppose @xmath263 . applying theorem [ th : main ] with @xmath264 to a sequence of boxes centered at the origin with radius @xmath265 ( @xmath266 ) , we have @xmath267\\ & \leqslant \sum_{j=1}^\infty c_1 \exp(-c_2 ( 2^j n)^\alpha ) \leqslant c_3 \cdot \exp(-c_4 n^\alpha ) . \end{split}\ ] ] now we prove that for @xmath268 , @xmath269 once we establish the statement above , the @xmath20-a.s .
existence of the infinite connected component will follow from @xcite , ( 2.4 ) . and
its uniqueness will be a consequence of @xcite , theorem 3.1 .
fix a connected set @xmath270 with @xmath271 and suppose we are in the complement of the event in ( [ eq : boundj ] ) .
taking @xmath272 , we conclude that @xmath273 is not separated from @xmath274 by @xmath8 in @xmath275 .
this implies that we can find a path in @xmath276 starting at @xmath277 and ending in @xmath278 so that its diameter is at least @xmath279 .
let @xmath280 denote the range of this path .
now suppose we have constructed connected sets @xmath281 in @xmath282 with diameters at least @xmath283 respectively .
we take @xmath284 in the complement of the event in ( [ eq : boundj ] ) to conclude that there is a path in @xmath285 connecting @xmath286 to @xmath287 . hence , defining the connected set @xmath288 as the union of @xmath289 and the range of this path , we have @xmath290 and @xmath291 . letting @xmath292 be @xmath293 , we obtain an infinite connected subset of @xmath10 , which intersects @xmath277 ( since it contains @xmath280 ) .
this proves ( [ eq : complemevent ] ) implying theorem [ th : cinftydense ] .
@xmath294 as another application of theorem [ th : main ] , we consider @xmath295 , the connected component of @xmath10 containing the origin and bound the tail of @xmath296 , when @xmath295 is finite .
[ th : taildiam ] ( @xmath0 ) let @xmath254 and @xmath255 be as in theorem [ th : main ] . for every @xmath27 , @xmath297 \leqslant c_6 \cdot \exp(-c_7 n^\alpha),\ ] ] where all the constants but @xmath298 depend only on @xmath31 .
_ the upper bound follows from theorem [ th : cinftydense ] . to prove the lower bound
, we estimate the probability that @xmath295 is precisely the segment @xmath299 , where @xmath300 is the first vector in the canonical basis of @xmath89 , see below ( [ eq : floorconvex ] ) .
define @xmath301 . using the transience of the @xmath302-dimensional simple random walk , one concludes that for @xmath303 @xmath304 \geqslant p_x[\widetilde h_{\overline{i } } = \infty ] = c > 0.\ ] ] and straightforwardly that @xmath305 \geqslant c > 0 , \text { for all $ x \in \partial i_n$}.\ ] ] for @xmath306 , write @xmath307 . using ( [ eq : segmentscape ] ) ( [ eq : qk ] ) and ( [ eq : nuqk ] ) , we conclude that @xmath308 , for every @xmath306 . since @xmath309 and @xmath310 are pairwise disjoint , the poisson point processes obtained by restricting @xmath311 to these sets are independent , so that @xmath312 \geqslant \mathbb{p}\big[\mathcal{c}_0^u = i_n\big]\\ & \geqslant \mathbb{p}\big[\omega(w^{*,x } \times [ 0,u ] ) \geqslant 1 , \text { for all $ x \in \partial i_n$ and } \omega(w_{i_n}^ * \times [ 0,u ] ) = 0\big ] \\ & \geqslant \left(1-\exp\big(-u \cdot c^2 \big ) \right)^{2dn } \exp(-u \cdot \text{cap}(i_n))\\ & \overset{(\ref{eq : cap})}{\geqslant } \exp(-c(u))^{2dn } \exp(-u \cdot n ) \geqslant \exp(-c_5(u )
n ) . \end{split}\ ] ] this concludes the proof of theorem [ th : taildiam ] .
@xmath294 we now bound the tail of the distribution of @xmath313 ( the volume of the vacant cluster containing the origin ) when this cluster is finite .
[ th : tailvol ] for @xmath27 ( @xmath254 as in theorem [ th : main ] ) , @xmath314 \leqslant c_6 \cdot \exp(-c_7 v^{\alpha / d}),\ ] ] with all the constants but @xmath315 depending only on @xmath31 . _
the upper bound follows from theorem [ th : taildiam ] once we use the fact that for some @xmath316 , @xmath317 .
we prove the lower bound by estimating the probability that the box @xmath318 is contained in @xmath10 while the sphere @xmath319 is contained in @xmath8 .
this is a case where @xmath320 .
we quote @xcite , remark 2.5 2 ) for the following estimate @xmath321 } \big ] \geqslant c \cdot \exp(-c ' n^{d-2 } \log n).\ ] ] for @xmath322 ( we assume without loss of generality that @xmath323 ) we consider the projection of the random walk starting at @xmath93 in the first coordinate . using a gamblers ruin argument , one sees that with probability at least @xmath324 one reaches @xmath325 before @xmath326 . from ( [ eq : hitball ] )
we obtain @xmath327 \geqslant \frac{c}{n}. \end{split}\ ] ] we want the set @xmath319 to be contained in @xmath8 and this will be the case if @xmath328 @xmath329 @xmath8 for all @xmath93 with @xmath330 .
we define @xmath331 for @xmath93 such that @xmath330 . using ( [ eq : qk ] ) and ( [ eq : nuqk ] ) ,
we estimate @xmath332}\big]\\ \overset{(\ref{eq : hitball } ) , ( \ref{eq : escape5n } ) , ( \ref{eq : coverbox})}{\geqslant } & \frac{c}{n } \cdot \exp(-c'n^{d-2 } \log n ) \geqslant c \cdot \exp(-c'n^{d-2 } \log n ) . \end{split}\ ] ] finally , since the sets @xmath333 and @xmath334 are pairwise disjoint , @xmath335 \geqslant \mathbb{p}[b(0,n ) \subset \mathcal{v}^u , { \partial_{\textnormal{int}}}b(0,5n ) \subset \mathcal{i}^u]\\ & \geqslant \mathbb{p}\bigg[\bigg ( \bigcap_{x \in \mathbb{z}^d ; \lvert x \rvert_\infty = 5}\omega(w^{*,x } \times [ 0,u ] ) \geqslant 1 \bigg ) \cap \omega(w_{b(0,n)}^ * \times [ 0,u ] ) = 0 \bigg]\\ & \overset{(\ref{eq : boundnuwx})}{\geqslant } \big ( 1- \exp \big(-u \cdot c \cdot e^{-c'n^{d-2}\log n}\big ) \big)^{9^d \cdot 2d } \cdot c \exp ( -u \cdot \text{cap}(b(0,n)))\\ & \overset{(\ref{eq : estimatecap})}{\geqslant } \big ( c(u ) \cdot
e^{-c'n^{d-2}\log n } \big)^{9^d \cdot 2d } \cdot c \exp ( -u \cdot c ' \cdot n^{d-2 } ) \geqslant c(u ) \exp(c ' n^{d-2 } \log n ) .
\end{split}\ ] ] and the proof is finished if one takes @xmath336 to be @xmath337 .
@xmath294 [ rem : intorus ] @xmath338
\1 ) the upper bound in theorem [ th : tailvol ] has a flavor of the open problem posed in remark 4.7 1 ) of @xcite .
\2 ) although the lower bound on theorem [ th : tailvol ] is not expected to be sharp , it has a slower decay than the corresponding upper bound in the case of bernoulli independent percolation , see for instance @xcite , ( 8.66 ) p.216 , c.f .
remark 1.1 of @xcite .
@xmath339 as a last application of theorem [ th : cinftydense ] , we prove [ th : logcomp ] ( @xmath0 ) let @xmath254 and @xmath255 be as in theorem [ th : main ] .
for every @xmath27 , @xmath42 and @xmath43 , @xmath340 = 0.\ ] ] _ proof .
_ for a fixed @xmath341 , one has @xmath342\\ & \leqslant \mathbb{p } \bigg [ \begin{array}{c } \text{there is a connected set $ c \subset b(x,2(\log n)^{(1 + \epsilon)/\alpha})$ which is}\\ \text{not neighbor of $ \mathcal{c}^u_\infty$ and has diameter larger or equal to $ \lfloor ( \log n)^{(1 + \epsilon)/\alpha } \rfloor$ } \end{array } \bigg]\\ & \overset{(\ref{eq : cinftydense})}{\leqslant } c \exp(-c ' ( \log n)^{(1 + \epsilon ) } ) .
\end{split}\ ] ] summing over the points @xmath341 , we obtain @xmath343\\ & \leqslant c n^{-c \log n + d } , \text { and the claim ( \ref{eq : logbound } ) follows}. \quad \square \end{split}\ ] ] [ rem : bernoulli ] let us compare the results of this section with what is known to hold in the case of bernoulli independent site percolation ( where to every site one independently assigns the value @xmath344 with probability @xmath345 and @xmath346 with probability @xmath347 ) .
\1 ) in the context of bernoulli percolation , a result with a similar flavor to theorem [ th : main ] can be proved .
it is valid for any @xmath5 , any @xmath348 and providing an exponential bound instead of a stretched exponential , see for instance @xcite , lemma ( 7.89 ) p.186 .
however , its proof strongly rely on the independence of the state of distinct sites , in contrast with the high dependence featured by the interlacement model , c.f .
@xcite , ( 1.68 ) .
\2 ) in the bernoulli independent case , one can use a peierls - type argument to show that , for @xmath345 sufficiently close to one , @xmath349 together with ( [ eq : stfillconn ] ) , this provides a simple proof of the fact that : for this choice of @xmath345 , the probability that the diameter of the cluster of @xmath344 s containing the origin equals @xmath336 decays exponentially in @xmath336 . again
this argument fails in the case of random interlacements .
actually , it is proved in @xcite , corollary 2.3 , that @xmath350 and according to ( [ eq : qucharacterized ] ) , @xmath8 meets @xmath46 with overwhelming probability . hence ( [ eq : bernoullibound ] ) does not hold for any @xmath45 under the measure @xmath12 .
see also @xcite , remark 2.5 2 ) .
\3 ) as mentioned in the introduction , theorem [ th : main ] does not hold true if one replaces @xmath245 and @xmath246 by @xmath78 and @xmath79 in definition [ def : separate ] .
we now give a brief justification for this claim . from the remark above we conclude that , with overwhelming probability as @xmath336 goes to infinity , one can find a self avoiding path @xmath140 contained in @xmath8 connecting @xmath46 to @xmath274 . in this case , it is possible to extract two connected subsets @xmath78 and @xmath79 of @xmath351 which have diameter at least @xmath352 and are far from each other . since @xmath78 and @xmath79 are contained in @xmath8 , there is no path in @xmath10 which joins these two sets . in the best case
, we can hope to find a path in @xmath10 which connects @xmath245 to @xmath246 , as in definition [ def : separate ] , see also the proof of lemma [ lem : avoid ] . in the context of bernoulli site percolation
, we do not expect to need such restriction in the definition of separation , for a theorem analogous to theorem [ th : main ] to hold for @xmath345 close enough to one .
\4 ) as an alternative definition of separation , one could for instance require that the sets @xmath78 and @xmath79 are disjoint from @xmath242
. this definition would be closer in spirit to the one appearing in @xcite , lemma ( 7.89 ) p.186 .
however , in order to prove theorem [ th : main ] , we will need the separation event to have a monotonicity property , see .
finding a suitable definition of separation is an important issue if one wishes to prove a theorem analogous to theorem [ th : main ] that holds for any @xmath19 . @xmath165
in this section we establish theorem [ th : main ] using the lemma [ lem : indicespath ] and the theorems [ th : coarse2 ] and [ th : local ] , which are going to be proved in sections [ sec : coarse ] and [ sec : local ] .
since the proof will follow a renormalization scheme , lets introduce the basic notation for the scales . for integers @xmath48 ,
we write @xmath353 and define the set of indices in the scale @xmath52 @xmath354 given @xmath355 , for @xmath68 , we consider the box @xmath356 and its @xmath357-th neighborhood @xmath358 note that with this notation @xmath359 .
we introduce a definition which has a flavor of ( 1.6 ) in @xcite . for @xmath64 , and @xmath360 , write @xmath361 recall the definition [ def : separate ] .
consider the function @xmath362 from @xmath13 into the set of subsets of @xmath9 that , given @xmath363 , is defined through @xmath364 ( in other words , @xmath362 returns the subset of @xmath9 where @xmath365 is one ) . since @xmath366 depends only on @xmath242 intersected with @xmath367 , one sees that @xmath368 depends only on @xmath369 for @xmath370 , therefore it is a cylinder event in the product @xmath146-algebra of @xmath13 .
it follows from the definition [ def : separate ] that @xmath371 first we prove the following reduction step .
[ prop : reduct ] to prove theorem [ th : main ] , it is suffices to show that : @xmath372 \leqslant c \cdot 2^{-2^{\kappa } } , \text { for every $ u \leqslant
\bar u$ and $ \kappa \geqslant 0$}. \end{array}\ ] ] _ proof .
_ fix @xmath27 as above and recall the definition of @xmath373 above definition [ def : fill ] . as an intermediate step towards ( [ eq : mainbound ] )
, we first prove that ( [ eq : reduct ] ) implies that for any @xmath250 and @xmath374 , @xmath375\\ & \leqslant c(l , l_0,\gamma ) \ , 2^{-l_{\kappa}^\alpha } , \text{with $ \alpha = \alpha(l ) > 0$}. \end{split}\ ] ] note that the condition on the diameters of @xmath78 and @xmath79 above is less restrictive than the condition appearing in ( [ eq : chi ] ) .
hence , on the event appearing in ( [ eq : reductinterm ] ) there is no guarantee that @xmath376 for some @xmath64 so that we can not apply ( [ eq : reduct ] ) with this @xmath52 to prove ( [ eq : reductinterm ] ) .
instead , we choose an appropriate @xmath377 , find an @xmath378 such that @xmath379 and then use ( [ eq : reduct ] ) .
more precisely , let @xmath380 be such that @xmath381 and note that , if @xmath382 , @xmath383 on the event appearing in ( [ eq : reductinterm ] ) , one can find @xmath384 such that the boxes @xmath385 and @xmath386 are contained in @xmath387 and intersect @xmath78 and @xmath79 respectively . by ( [ eq : kkosmall ] ) , @xmath388 .
we now join @xmath389 and @xmath390 by a path @xmath140 in @xmath391 and note that all boxes @xmath392 ( @xmath393 ) are contained in @xmath394 . by lemma [ lem : indicespath ]
there is an index @xmath395 such that @xmath396 and @xmath397 . as a result
we have the bound @xmath398\\ & \qquad \leqslant \mathbb{p } \bigg [ \begin{split } \text{there } & \text{is some } \bar m = ( { \kappa}-{\kappa}_o , \bar i ) , \text { such that } \\ & c_{\bar m}^3 \subset c_{({\kappa},0)}^2 \text { and } \chi_{\bar m}(\mathcal{i}^u ) = 1 \end{split } \bigg ] \\ & \qquad \leqslant ( 3 \cdot ( 80l)^{{\kappa}_o})^d \cdot \mathbb{p}[\chi_{({\kappa}- { \kappa}_o,0)}(\mathcal{i}^u ) = 1 ] \overset{(\ref{eq : reduct})}{\leqslant } c(l,\gamma ) 2^{-2^{{\kappa}-{\kappa}_o}}. \end{split}\ ] ] to obtain ( [ eq : reductinterm ] ) we now set @xmath399 and note that @xmath400 { } \frac{\log 2}{\log ( 80l ) } > \alpha,\ ] ] so that @xmath401 , for @xmath52 larger or equal to some @xmath402 . possibly increasing @xmath403 to some @xmath404
we obtain ( [ eq : reductinterm ] ) from ( [ eq : reducetok ] ) . finally , for a given @xmath336 ( which we assume for the moment to be larger or equal to @xmath50 ) we choose a depth @xmath405 such that @xmath406 is comparable with @xmath336 , more precisely , @xmath407 the boxes which appear in ( [ eq : mainbound ] ) are centered at the origin ( unlike the box @xmath387 in ( [ eq : reductinterm ] ) ) .
however , note that @xmath46 and @xmath408 can be respectively mapped ( under the same translation ) to the boxes @xmath409^d \cap \mathbb{z}^d$ ] and @xmath410^d \cap \mathbb{z}^d$ ] . hence ,
using that the law @xmath20 is invariant under translation , see @xcite proposition 1.3 , we conclude that it suffices to prove ( [ eq : mainbound ] ) with @xmath46 and @xmath408 respectively replaced by @xmath411 and @xmath412 .
note , by ( [ eq : nn ] ) , that @xmath413 hence , @xmath414\\ & \overset{(\ref{eq : b1b2 in } ) , ( \ref{eq : monotoneinb})}{\leqslant } \mathbb{p } \bigg [ \begin{split } \text{there } & \text { exist $ a_1 , a_2 \subset c_{({\kappa}(n),0)}$ with $ \operatorname{\textnormal{diam}}(a_1 ) , \operatorname{\textnormal{diam}}(a_2 ) \geqslant \tfrac{\gamma}{4 } l_{{\kappa}(n ) - 1}$}\\ \text{which } & \text{are connected and separated by $ \mathcal{i}^u$ in $ b(\text{sbox}(a_1 \cup a_2 ) , \tfrac{\gamma}{4 } l_{{\kappa}(n)-1})$ } \end{split } \bigg ] \\ & \overset{(\ref{eq : reductinterm})}{\leqslant } c(l , l_0,\gamma ) \ , 2^{-l_{{\kappa}(n)}^\alpha } \overset{(\ref{eq : nn})}{\leqslant } c_1(l , l_0,\gamma ) \cdot \exp(-c_2 n^\alpha ) .
\end{split}\ ] ] by possibly increasing @xmath28 we obtain ( [ eq : mainbound ] ) for all @xmath34 . @xmath294 in view of the above proposition , it suffices to show ( [ eq : reduct ] ) in order to establish theorem [ th : main ] .
this will be done using the two main ingredients mentioned in the introduction , see ( [ eq : coarse ] ) and ( [ eq : statelocal ] ) . to make this rough description precise , we will need the following [ def : skeleton ] a set of indices of boxes in the finest scale @xmath71 is said to be a _ skeleton _ if , for any @xmath415 , we have @xmath416 this is a type of wiener criterion , see for instance @xcite , theorem 2.2.5 , p.55 .
_ proof of theorem [ th : main ] .
_ the first ingredient of the proof of ( [ eq : reduct ] ) is stated in the following theorem , which will be a direct consequence of theorem [ th : coarse2 ] proved in section [ sec : coarse ] .
[ th : coarsefirst ] for each @xmath417 ( @xmath68 ) , there is a family of skeletons @xmath418 , such that @xmath419 the above statement will reduce the proof of ( [ eq : reduct ] ) to the derivation of an appropriate bound on @xmath420 $ ] uniformly over the skeleton @xmath70 . roughly speaking , to obtain such bound we will analyze the excursions of the interlacement trajectories between neighborhoods of the boxes of the skeleton . given a skeleton @xmath70 , we consider ( see ( [ eq : boxcml ] ) for the notation ) , @xmath421 recalling that in ( [ eq : boxcml ] ) we did not require @xmath357 to be an integer . by ( [ eq : skeleton ] ) @xmath60
, we conclude that for two distinct @xmath422 , @xmath423 .
hence , for @xmath93 in @xmath424 , we can write @xmath425 for the unique index @xmath426 such that @xmath427 .
define the successive times of return to @xmath172 and departure from @xmath173 , @xmath428 and @xmath429 as in ( [ eq : departreturn ] ) and note that on @xmath430 , @xmath431 . for @xmath432 , we define the number of excursions performed by @xmath433 , as @xmath434 and write @xmath435 & \leqslant \mathbb{p}\big[\langle \mu_{\sigma , u } , g_m \rangle > 100 \cdot 2^{\kappa}\big]\\ & \quad + \mathbb{p}\bigg[\langle \mu_{\sigma , u } , g_m \rangle \leqslant 100 \cdot 2^{\kappa } , \bigcap_{m\in m } \ { \chi_m(\mathcal{i}^u ) = 1\}\bigg ] .
\end{split}\ ] ] to estimate the first term in the right - hand side of the above equation , we will use an exponential chebychev - type inequality . with ( [ eq : expexpectation ] )
, this will amount to bounding the exponential moments of @xmath436 under @xmath177 .
this will be performed by choosing @xmath49 sufficiently large , consequently reducing the probability that a random walk starting on @xmath437 hits @xmath172 before escaping to infinity .
more precisely , for @xmath438 , @xmath439 \overset{(\ref{eq : hitfarball } ) , \text{(\ref{eq : skeleton})}}{\leqslant } p_y[h_{c_{m(y)}^5 } < \infty ] + \sum_{h=0}^\infty 2^{h+1 } \cdot c \cdot \left ( \frac{l_0}{l \cdot l_h } \right)^{d-2 } \leqslant \frac{c_{11}}{l^{d-2}}.\ ] ] using the strong markov property at time @xmath440 , we obtain , for @xmath441 , @xmath442 \leqslant e^\lambda ( 1 + e_x[1\{r_2 < \infty\ } \phi(x_{r_2})])\\ & \overset{(\ref{eq : boundexcursion})}{\leqslant } e^\lambda \big(1 + \frac{c_{11}}{l^{d-2 } } \sup_{z \in \sigma } \phi(z ) \big ) .
\end{split}\ ] ] this inequality implies that @xmath443 now choose @xmath444 so that , c.f .
( [ eq : skeleton ] ) @xmath51 , @xmath445 as a result , when we take @xmath446 , @xmath447 then , using ( [ eq : expexpectation ] ) , an exponential chebychev - type inequality and @xmath448 , we get the desired bound on the first term of ( [ eq : twoterms ] ) , namely : @xmath449
\leqslant \exp(-100 \lambda \ , 2^{\kappa}+ u e_{e_\sigma } [ e^{\lambda g_m } -1 ] ) \overset{(\ref{eq : boundphi})}{\leqslant } \left [ e^{-100\lambda + c\cdot u \cdot l_0^{d-2 } e^\lambda } \right]^{2^{\kappa } } \negthickspace.\ ] ] the term in the right hand side of the equation above will be controlled at the very end of the proof of theorem [ th : main ] using ( [ eq : etolambda ] ) and choosing @xmath1 small enough .
we now bound the second term of ( [ eq : twoterms ] ) .
for this we need to invoke theorem [ th : local ] of section [ sec : local ] , that was referred as the second ingredient of the proof of theorem [ th : main ] , see ( [ eq : statelocal ] ) .
we need to introduce some notation .
for @xmath450 , @xmath451 , recall that @xmath177-a.s . @xmath431 and @xmath452 .
define @xmath453 = p_x[\,\cdot\ , |x_{d_1 } = y].\ ] ] given @xmath454 and @xmath455 , we define @xmath456 and denote with @xmath457 the canonical coordinates in @xmath458 .
we introduce the stopping time @xmath459 .
we now need the second ingredient of the proof which is theorem [ th : local ] . for the reader s convenience we state here
this theorem and reefer to section [ sec : local ] for its proof .
[ th : localfirst ] for @xmath0 , given @xmath42 , @xmath460 and @xmath461 , for large enough @xmath462 @xmath463 } \bigg ) = 1 \right ] < \epsilon . \end{array}\ ] ] the above result concerns only what happens in one fixed box in the finer scale . loosely speaking , to bound the second term in the right hand side of ( [ eq : twoterms ] )
, we are going to condition all the interlacement trajectories which intersect @xmath172 on their return and departure points from @xmath172 and @xmath173 ( i.e. , @xmath464 and @xmath465 ) .
we first need some further notation .
it is known that @xmath466 has the same distribution as @xmath467 , where @xmath468 is a poisson variable with parameter @xmath469 and @xmath470 ( @xmath471 ) are i.i.d . , @xmath472-distributed and independent of @xmath473 , recall the notation in ( [ eq : normalizedek ] ) and ( [ eq : muku ] ) . write @xmath474 ( @xmath475 ) for the canonical coordinates of the trajectories @xmath476 .
we consider the number of excursions performed by @xmath477 , @xmath478 ( @xmath471 ) and for @xmath479 , we denote the range of the @xmath480-th excursion by @xmath481}^i$ ] . given an index @xmath76 in the skeleton @xmath70 , we collect the indices of all excursions performed in @xmath482 : @xmath483 note that , by ( [ th : coarse ] ) _ i ) _ , when @xmath484 , the set of @xmath485 with @xmath486 has cardinality at must @xmath487 , and hence @xmath488 therefore , we can bound the last term of ( [ eq : twoterms ] ) as follows : @xmath489\\ & \overset{(\ref{eq : countm})}{\leqslant } \sum_{\substack{m'\subset m ; \\ |m'| = 2^{{\kappa}-1 } } } \mathbb{p } \bigg [ \bigcap_{m'\in m ' } \big(\ { \chi_{m'}(\mathcal{i}^u ) = 1\ } \cap \ { |\phi_{m'}| \leqslant 200 \ } \big ) \bigg ] \\ &
\overset{(\ref{eq : pkmup})}{\leqslant } \sum_{\substack{m'\subset m ; \\
|m'| = 2^{{\kappa}-1 } } } \mathbb{p}_k \bigg [ \bigcap_{m'\in m ' } \big ( \ { \chi_{m'}(\cup_{i \leqslant n } \operatorname{\textnormal{range}}(x^i ) ) = 1\ } \cap \ { |\phi_{m'}| \leqslant 200 \ } \big ) \bigg]\\ & \leqslant 2^{2^{\kappa } } \sup_{\substack{m'\subset m ; \\ |m'| = 2^{{\kappa}-1 } } } \sum_{n\geqslant 0 } \mathbb{p}[s = n ] p_{\bar e_\sigma}^{\otimes n } \bigg [ \bigcap_{m'\in m ' } \big ( \ { \chi_{m'}(\cup_{i \leqslant n } \operatorname{\textnormal{range}}(x^i ) ) = 1\ } \cap \ { |\phi_{m'}| \leqslant 200 \ } \big ) \bigg ] .
\end{split}\ ] ] we now decompose the event under the above probability over all possible values of the number of excursions @xmath490
( performed by each of these @xmath141 trajectories ) and on the departure points of these excursions ( @xmath491 and @xmath492 , for @xmath493 ) . in the sums below , we tacitly assume that @xmath494 and @xmath495 , for @xmath496 and @xmath497 .
the term above is thus smaller or equal to @xmath498 p_{\bar e_\sigma}^{\otimes n } \bigg [ g_i = g_i , x_{r_j}^i = x_j^i , x_{d_j}^i = y_j^i,&\\ \qquad \text { for $ i \leqslant n$ and $ j \leqslant g_i$ } , \bigcap_{m'\in m ' } \ { \chi_{m'}(\cup_{(i , j ) \in \phi_{m ' } } e_j^i ) = 1\}\bigg]\\ \leqslant 2^{2^{\kappa } } \sup_{\mbox{\fontsize{8}{8 } \selectfont $ \substack{m'\subset m ; \\ |m'| = 2^{{\kappa}-1}}$ } } \,\ , \sum_{\mbox{\fontsize{8}{8 } \selectfont $ \substack{n \geqslant 0 , g_1 , \dots , g_n \geqslant 1\\ x_j^i , y_j^i \text { such that } \forall m ' \in m ' \\ \#\ { ( i , j ) ; m(x_j^i ) = m ' \ } \leqslant 200 } $ } } \mathbb{p}[s = n ] e_{\bar e_\sigma}^{\otimes n } \bigg [ g_i = g_i , x_{r_j}^i = x_j^i , x_{d_j}^i = y_j^i,&\\ \qquad \text { for $ i
\leqslant n$ and $ j \leqslant g_i$ } , \prod_{m'\in m ' } p_{\mbox{\fontsize{8}{8 } \selectfont $ \substack{(x_j^i)_{(i , j ) \in \phi_{m'}},(y_j^i)_{(i , j ) \in \phi_{m'}}}$ } } \big [ \chi_{m'}(\cup_{(i , j ) \in \phi_{m ' } } e_j^i ) = 1\big]\bigg ] .
\end{split}\ ] ] choosing @xmath499 and @xmath500 in theorem [ th : local ] , we can bound the product above by @xmath501 and remove it from the sum ( which sum up to one ) , obtaining the desired bound @xmath502 \leqslant ( 2^{9}(5 \cdot 80l)^{4d})^{-2^{\kappa}},\ ] ] for an appropriate choice of @xmath503 which came from ( [ th : reftolocal ] ) . with ( [ eq : twoterms ] ) , ( [ eq : boundfirst ] ) , ( [ eq : etolambda ] ) and ( [ eq : boundsecond ] )
, we obtain @xmath504 \leqslant 2 \cdot ( 2^{9}(5 \cdot 80l)^{4d})^{-2^{\kappa}}\ ] ] for @xmath1 smaller or equal to some @xmath505 .
this , together with ( [ th : coarse ] ) leads to ( [ eq : reduct ] ) , proving theorem [ th : main ] .
@xmath294 note the order in which we chose the parameters for the proof : first @xmath506 in ( [ eq : lchoice ] ) , then @xmath507 in ( [ eq : boundsecond ] ) and finally @xmath508 in ( [ eq : baruchoice ] ) .
[ rem : g100 ] @xmath338 \1 ) let us comment on the constant @xmath61 appearing in the equation ( [ eq : boundfirst ] ) .
in order to bound the first term of we use an exponential chebychev inequality , which provides a decay of @xmath509 $ ] of order @xmath510 , see . notice that this decay should be fast enough in order to offset the growth of the combinatorial factor in , @xmath60 ( which is @xmath511 ) .
the way we have to tune these two competing terms is by choosing @xmath49 large , but both of them depend on this parameter . in order to make this competition to work to our advantage
, we need to choose a large number @xmath57 above , in such a way that @xmath58 . for our purposes @xmath59 will do the job .
this delicate balance is well illustrated in and , where we choose a large @xmath49 .
\2 ) note that the only part of the proof of theorem [ th : main ] in which we use the hypothesis @xmath0 is theorem [ th : local ] .
this will be further discussed in the section [ sec : local ] .
in this section we study the hierarchical property of the function @xmath512 , see theorem [ th : coarsefirst ] , which was used to prove theorem [ th : main ] .
the main step to establish ( [ th : coarse ] ) is the lemma below , which is interesting by itself , see the paragraph above ( [ eq : reducetok ] ) . loosely speaking it states that
, if a path of boxes connects two large connected sets which are separated by @xmath360 , then at least one of this boxes ( say indexed by @xmath74 ) satisfies @xmath513 , see figure [ fig : lbar ] .
more precisely , [ cl][cl][3][0]@xmath78 [ cl][cl][3][0]@xmath79 [ cl][cl][3][0]@xmath242 [ cl][cl][3][0]@xmath514 [ cl][cl][3][0]@xmath515 [ cl][cl][3][0]@xmath516 , see lemma [ lem : indicespath].,title="fig:",scaledwidth=35.0% ] + [ lem : indicespath ] let @xmath517 be connected sets and @xmath360 be such that @xmath78 is separated from @xmath79 by @xmath242 in @xmath243 .
consider some scale @xmath68 and a path @xmath140 in @xmath9 , as well as the path of indices @xmath518 , @xmath519 , so that the following holds @xmath520 then there is an @xmath521 such that @xmath522 .
see figure [ fig : lbar ] .
_ we consider the set @xmath523 we claim that @xmath524 indeed , given @xmath525 , we join them to @xmath245 by paths as in ( [ eq : seta ] ) ( and then to @xmath78 in case they were not already there ) and use the connectedness of @xmath78 .
moreover @xmath526 indeed , suppose that there is some @xmath527 .
since @xmath244 ( see definition [ def : separate ] ) we have that @xmath528 . by ( [ eq : seta ] )
, there is a path in @xmath243 from @xmath245 to @xmath93 which does not intersect @xmath242 . possibly stopping this path at the first time it meets @xmath246
, we obtain a contradiction with the fact that @xmath78 is separated from @xmath79 by @xmath242 in @xmath243 .
we then claim that @xmath529 suppose by contradiction the existence of some @xmath530 , say with @xmath93 neighbor of @xmath531 . if @xmath532 , we have @xmath533 which implies by ( [ eq : seta ] ) that @xmath534 , a contradiction .
the other possibility is that @xmath107 is connected to @xmath245 by some path in @xmath243 which is disjoint from @xmath242 .
adding one step to this path we can assume that it meets @xmath93 , obtaining again the contradiction @xmath535 , see ( [ eq : seta ] ) .
this proves ( [ eq : boundainu ] ) . in order to choose @xmath536 , we define @xmath537 and consider the two following possibilities _ case 1 : @xmath538 is empty .
_ in other words , the set @xmath539 meets all the boxes with indexes in the path @xmath140 . in this case
we take @xmath540 . from ( [ eq : seti ] )
, there is some point @xmath541 and by ( [ eq : indicespath ] ) @xmath60 , there is some @xmath542 . by ( [ eq : afarfroma2 ] ) and ( [ eq : boundainu ] ) , @xmath539 and @xmath79
are separated by @xmath242 in @xmath243 . since ( [ eq : indicespath ] ) @xmath51 implies that all paths in @xmath543 are in @xmath243 , all we need in order to show that @xmath544 is to extract connected components @xmath545 with @xmath546 .
we know that @xmath547 if @xmath548 , since @xmath539 is itself connected , we take @xmath549 . otherwise , again using the connectedness of @xmath539 , we take @xmath550 to be the range of a path from @xmath93 to @xmath551 . with a similar argument
we obtain @xmath552 , since it is connected , intersects @xmath516 and its diameter is at least @xmath553 .
_ case 2 : @xmath538 is non - empty .
_ in this case , we have by ( [ eq : indicespath ] ) @xmath60 , that @xmath554 , so that @xmath555 and we define @xmath536 to be @xmath556 . we know that @xmath557 and as in the previous case , we are able to find some connected set @xmath558 such that @xmath559 . define @xmath552 to be @xmath560 , which is connected and contained in @xmath561 . by the definition of @xmath536
, we have that @xmath562 , hence , @xmath563 .
from ( [ eq : boundainu ] ) we conclude that any path in @xmath243 from @xmath550 to @xmath552 meets @xmath564 and hence @xmath242 . and since @xmath565 , we find that @xmath544 .
this finishes the proof of lemma [ lem : indicespath ] .
@xmath294 now we prove the following [ th : coarseinduct ] let @xmath566 and @xmath360 .
if @xmath567 , we can find @xmath568 and @xmath569 such that @xmath570 _ proof .
_ since @xmath567 , we can find connected sets @xmath571 such that @xmath572 , @xmath573 @xmath574 and @xmath78 , @xmath79 are separated by @xmath242 in @xmath575 . with the lemma [ lem : indicespath ]
, we conclude that theorem [ th : coarseinduct ] follows once we show that @xmath576 since @xmath577 ( @xmath578 ) , we can find @xmath579 after relabeling , we can suppose that @xmath580 and we set @xmath581 . we claim that @xmath582 indeed , with ( [ eq : noprimeismin ] ) and the triangle inequality , @xmath583 as a result of ( [ eq : i1i2andprime ] ) and ( [ eq : distprimeandno ] ) , we find that @xmath584 to find the desired paths @xmath140 and @xmath585 , we consider two cases : and @xmath585 in _ case 1 . _ ( left ) and _ case 2 . _ ( right),title="fig:",scaledwidth=100.0% ] + _ case 1 : _ @xmath586 . in this case
we connect @xmath389 to @xmath390 by any path @xmath140 contained in @xmath587 , see figure [ fig : paths ] . in order to find @xmath588
, we first connect @xmath589 to @xmath590 by any @xmath591 in @xmath592 . since @xmath593 @xmath594 ,
we have @xmath595 by ( [ eq : distprimetoi1 ] ) and the connectedness of @xmath596 , see ( [ eq : stfillconn ] ) , we can find a modification @xmath585 of the path @xmath591 which joins the points @xmath589 and @xmath590 , avoiding @xmath597 and satisfying @xmath598 .
so that @xmath599 and consequently @xmath600 for @xmath601 .
finally , @xmath602 and @xmath603 imply that @xmath604 .
_ case 2 : _ @xmath605 .
first we claim that @xmath606 see figure [ fig : paths ] . indeed ,
if we take @xmath607 ( recall the definition below ( [ eq : floorconvex ] ) ) , we have 1 .
@xmath608 for @xmath609 , since @xmath605 , 2 .
@xmath610 , applying to each coordinate the equation ( [ eq : floorconvex ] ) . since , a non - empty intersection of boxes is a box , ( [ eq : setf ] ) follows .
we also claim that @xmath611 for this , take @xmath612 , using once more ( [ eq : floorconvex ] ) we conclude that @xmath610 .
we then bound @xmath613 as follows : @xmath614 we thus find that @xmath615 . in a similar manner
we see that @xmath616 , finishing the proof of the claim ( [ eq : fcapballs ] ) .
we now connect @xmath389 to @xmath390 by any path @xmath140 in @xmath617 .
since @xmath618 , we have @xmath619 for all @xmath620 .
finally , define @xmath621 .
equation ( [ eq : distprimetoi1 ] ) implies that @xmath622 . by ( [ eq : noprimeismin ] )
, we have @xmath623 .
hence , @xmath624 and neither @xmath589 nor @xmath590 belongs to @xmath625 .
take @xmath107 in @xmath626 .
since only three of the @xmath627 half - lines ( parallel to the canonical basis ) connecting @xmath107 to infinity can meet @xmath625 ( @xmath625 is the union of three boxes ) , we see with lemma [ eq : charfill ] that @xmath628 .
so that @xmath629 and ( [ eq : stfillconn ] ) implies that @xmath630 is connected . as in the _ case 1 _ , we choose some path @xmath591 in @xmath631 , connecting @xmath589 to @xmath590 and modify it to get some @xmath585 joining @xmath589 to @xmath590 which is disjoint from @xmath625 and contained in @xmath632 . together with ( [ eq : allweneed ] ) , this concludes the proof of theorem [ th : coarseinduct ] .
@xmath294 in section [ sec : proof ] , we stated theorem [ th : coarsefirst ] , which we now deduce from the theorem [ th : coarseinduct ] .
note that the statement of theorem [ th : coarse2 ] below is slightly stronger than theorem [ th : coarsefirst ] ( as needed in the induction procedure used in the proof ) and thus the following implies theorem [ th : coarsefirst ] .
[ th : coarse2 ] given @xmath417 , @xmath68 , there is a family of skeletons @xmath418 , see definition [ def : skeleton ] , such that @xmath633 _ proof .
_ we proceed by induction on @xmath52 .
if @xmath634 , ( [ eq : coarse2 ] ) holds for @xmath635 . assume now that ( [ eq : coarse2 ] ) holds for some @xmath52 and consider for @xmath636 , @xmath637 note that ( [ eq : coarse2 ] ) @xmath638 directly follows from theorem [ th : coarseinduct ] and the induction hypothesis .
next we will show that @xmath418 is in fact a family of skeletons , see definition [ def : skeleton ] . to this end , consider @xmath639 , for @xmath640 as in ( [ eq : mmprimeequals ] ) .
we suppose without loss of generality that @xmath641 . by the induction hypothesis
, @xmath642 is a skeleton , so that @xmath643 since all boxes @xmath644 for @xmath645 are contained in @xmath646 ( @xmath647 ) and @xmath648 , the sets in ( [ eq : m1skeleton1 ] ) are empty for @xmath649 . from the inequality @xmath650
, we deduce that @xmath651 hence , @xmath652 and @xmath653 this shows that @xmath654 is a skeleton . by ( [ eq : distm1m2 ] ) , @xmath642 and @xmath655
are disjoint , so that @xmath656 and ( [ eq : coarse2 ] ) @xmath51 holds .
moreover , @xmath657 and ( [ eq : coarse2 ] ) @xmath60 is verified .
this concludes the proof by induction of theorem [ th : coarse2 ] .
the aim of this section is to establish theorem [ th : local ] ( already stated in section [ sec : proof ] and used in the proof of our main result , theorem [ th : main ] , above ( [ eq : boundsecond ] ) . roughly speaking , theorem [ th : local ] states that with overwhelming probability
, a fixed number of random walk trajectories does not separate components of macroscopic diameter in a large enough box .
moreover , the statement of theorem [ th : local ] holds uniformly over the points at which we condition these random walk trajectories to enter and exit large neighborhoods of this box . actually , the current section is the only part of the proof of theorem [ th : main ] where the hypothesis ( @xmath0 ) is used , so that in order to extend theorem [ th : main ] to lower dimensions , it would be enough to prove a version of theorem [ th : local ] for this case , see also remark [ rem : d3 ] .
we now give a rough overview of the proof of theorem [ th : local ] which relies on lemmas [ lem : conditioned ] , [ lem : farapart ] and [ lem : gfarapart ] .
the first step of the proof consists in proving an analogue of theorem [ th : local ] for one single trajectory .
more precisely , in lemma [ lem : conditioned ] we prove that the probability that one random walk trajectory separates components of macroscopic diameter inside a box goes to zero as the size of this box increases .
moreover , this limit is uniform over the points @xmath93 and @xmath107 at which we condition this random walk to enter and exit large neighborhoods of the box . to prove this lemma , we regard a ` chunk ' @xmath242 of the random walk trajectory as a set of ` sausages ' connected by cut - points , see ( [ eq : cuttime ] ) .
an important concept here is the notion of @xmath658-avoidable sets , where @xmath659 is an integer , see definition [ def : avoidable ] . loosely speaking
, a set @xmath539 is said to be @xmath658-avoidable if any path traversing @xmath539 can be modified ( within a distance of at most @xmath658 ) in order to go around @xmath539 through its boundary . in remark
[ rem : b01 ] we exemplify this definition showing that @xmath660 is @xmath661-avoidable , while @xmath662 is not .
the heart of the proof of lemma [ lem : conditioned ] is lemma [ lem : avoid ] , which roughly states the following : for a piece of trajectory @xmath242 , if the diameters of its ` sausages ' are bounded by @xmath658 , then @xmath663 is @xmath664-avoidable .
the strategy to prove this lemma can be informally described as `` to travel through the skins of the sausages '' .
this proof clarifies and solves the geometric restrictions mentioned in remark [ rem : bernoulli ] 3 ) . as a direct consequence of lemma [ lem : avoid ] , we conclude that for any pair of sets @xmath78 and @xmath79 , which are large when compared to the diameter of the ` sausages ' in @xmath242 , we can connect @xmath245 to @xmath246 avoiding @xmath242 . in corollary
[ lem : chifillx ] , we conclude that if the diameters of these ` sausages ' are bounded by @xmath665 , then @xmath242 does not separate components in a box of diameter @xmath666 ( in the sense of ( [ eq : chi ] ) ) . for @xmath0
, we are able to bound the diameter of the ` sausages ' occurring on a typical random walk trajectory before it exits the neighborhood of a given box , see lemma [ lem : estimates ] .
the proof of lemma [ lem : estimates ] relies on known results on intersections of random walks , see for instance @xcite .
the above mentioned results conclude the proof of lemma [ lem : conditioned ] .
the uniformity of this lemma over the points @xmath93 and @xmath107 ( where we condition the random walk to enter and exit large neighborhoods of the given box ) follows from the harnack inequality , see ( [ eq : harnack ] ) .
the second step in the proof of theorem [ th : local ] is to extend lemma [ lem : conditioned ] from one trajectory to a fixed number , say @xmath667 , of independent random walk trajectories .
first , consider @xmath667 connected subsets of @xmath9 such that : none of them separates components in a certain box , they are mutually far apart and they are @xmath668-avoidable . in lemma
[ lem : farapart ] we show that the union of these @xmath667 sets also does not separate components in the box .
thus , all that remains to prove is that @xmath667 independent random walks are , with high probability , mutually far apart .
this is the content of lemma [ lem : gfarapart ] , which again uses arguments on intersections of random walks for @xmath0 .
finally , we bring together lemmas [ lem : conditioned ] , [ lem : farapart ] and [ lem : gfarapart ] to obtain theorem [ th : local ] .
let us introduce the concept of cut - times for a doubly infinite trajectory .
let @xmath432 and recall that @xmath669 denote its canonical coordinates .
@xmath670 in this case @xmath671 is called a _ cut - point_. note that our definition differs from the usual definition of a cut - time , which does not require a strict inequality as above , see for instance @xcite .
what we informally described as a sausage will be determined by the range of @xmath433 between two chosen cut - times .
note , however , that the definition in ( [ eq : cuttime ] ) does not exclude the possibility that two cut - times are adjacent ( e.g. for the trajectory @xmath672 , every integer is a cut - time ) .
so , given a finite sequence of cut - times @xmath673 which are not adjacent , i.e. @xmath674 we define the ` pieces of trajectory ' : @xmath675 we stress here that the definition of sausages depend on the choice of the ( non - adjacent ) cut - times , which in general will not be the whole set of cut - times in a given interval .
there is still another technical reason for the introduction of the condition ( [ eq : farcuts ] ) , see the proof of lemma [ lem : avoid ] , above ( [ eq : utilde ] ) .
the sets @xmath676 are what we informally referred to as sausages , recall definition [ def : fill ] .
the next lemma states some useful properties of @xmath132 which we exclusively need in this section .
its proof , although short , digresses from our main purpose here and can be found in the appendix .
[ prop : fill ] if @xmath131 is finite , then @xmath677 now let @xmath678 be connected and finite .
if @xmath679 ( respectively @xmath680 ) then exactly one of the following possibilities holds : @xmath681 as we described at the beginning of this section , our main argument to show that certain sets do not separate components relies on the definition of ` avoidable set ' .
this is made precise in the following [ def : avoidable ] for @xmath659 and @xmath682 with @xmath539 finite , we say that @xmath539 is @xmath658-avoidable in @xmath273 if for every path @xmath140 in @xmath273 with endpoints
@xmath683 not in @xmath539 we can find a modification @xmath585 of @xmath140 such that : @xmath684 when @xmath685 , we simply write that @xmath539 is @xmath658-avoidable . and if the value of @xmath658 is not relevant , we omit it in the notation . note that we do not require the path @xmath585 or set @xmath539 to be contained in @xmath273 .
however , @xmath686 is @xmath661-avoidable , however , @xmath687 is not.,title="fig:",scaledwidth=75.0% ] + [ rem : b01 ] the property of being @xmath658-avoidable is not monotonic . consider for instance the set @xmath688 and some path with endpoints in @xmath133 , see figure [ fig : b01 ] .
every excursion this path performs inside the set @xmath689 can be replaced by an excursion entirely contained in @xmath690 ( according to ( [ eq : stfillconn ] ) , @xmath691 is connected ) . since @xmath692 , we conclude that @xmath539 is @xmath661-avoidable . although @xmath539 is @xmath661-avoidable and the set @xmath693 is contained in @xmath539 , we check that @xmath693 fails to be avoidable .
indeed , no path @xmath140 connecting the origin to some point in @xmath133 can be modified to another path which is disjoint from @xmath693 but have the same endpoints as @xmath140 , see figure [ fig : b01 ] .
note also that , @xmath694 in the lemma below , we show that @xmath695 is @xmath664-avoidable , where @xmath658 is a bound on the diameter of the sets @xmath696 .
loosely speaking , we first show that the union of the ` sausages ' ( @xmath676 ) and the cut - points ( @xmath697 ) is @xmath664-avoidable , by `` traveling through the skins of the sausages '' . then we show that @xmath695 is in fact precisely this union .
[ lem : avoid ] ( @xmath3 ) if @xmath698 are cut - times chosen as in ( [ eq : farcuts ] ) and for some @xmath699 , @xmath700 then @xmath695 is @xmath664-avoidable .
_ first we describe the neighborhood of the cut - points .
note that @xmath701 indeed , @xmath702 , and since @xmath703 , @xmath704 and @xmath234 must be equal .
we now define @xmath705 , as the set of @xmath135-neighbors of @xmath706 which lay in the @xmath302-dimensional plane by @xmath706 perpendicular to @xmath704 , i.e. @xmath707 see figure [ fig : ring ] . and the perpendicular vector @xmath704.,title="fig:",scaledwidth=20.0% ]
+ we claim that for any @xmath708 , @xmath709 in fact , the definition of @xmath705 in ( [ eq : ringj ] ) implies that @xmath710 so that by ( [ eq : cuttime ] ) , @xmath107 is disjoint from both @xmath711 and @xmath712 . since @xmath713 , this proves ( [ eq : ringdisj ] )
we also claim that the ` sausages ' are exterior to each other ,
i.e. for @xmath714 , @xmath715 indeed , by the definition of cut - times , the trajectory @xmath716 does nt meet the set @xmath696 ( analogously , @xmath717 does nt meet @xmath718 ) .
so , using the characterization of @xmath719 given in lemma [ eq : charfill ] , we conclude that @xmath720 and @xmath721 . by the the definition of the cut - time @xmath722
, we have @xmath723 , and using ( [ eq : threecases ] ) of lemma [ prop : fill ] , we obtain ( [ eq : distsauss ] ) . as a consequence of ( [ eq : ringdist ] ) and ( [ eq : distsauss ] ) , we have that @xmath724 and from ( [ eq : stfillconn ] ) we conclude that @xmath725 we now show that the union of the ` sausages ' with the cut - points @xmath726 \cup \{x_{n_j}\}_{j=1,\dots , j-1}\ ] ] is @xmath664-avoidable and it will only remain to prove that @xmath727 . as in the definition [ def : avoidable ] of a @xmath664-avoidable set , take a path @xmath140 such that @xmath728 loosely speaking , we first modify @xmath140 into a path @xmath591 which `` surrounds each sausage through its skin '' and finally we will modify @xmath591 to a path @xmath585 surrounding the cut - points using the rings . using ( [ eq : distsauss ] ) , we conclude that the visits performed by @xmath140 to the sets @xmath729 occur in time intervals which do not neighbor each other , i.e. there is a sequence of times @xmath730 , and a sequence of indices @xmath731 in @xmath732 such that @xmath733 if @xmath734 ( @xmath735 ) and @xmath736 otherwise . using ( [ eq : skinconn ] ) and ( [ eq : tauextr ] ) , we define a first modification @xmath591 of @xmath140 having the same end - points as @xmath140 and which is disjoint from all the ` sausages ' .
we do this by replacing all the pieces @xmath737 by some path in @xmath738 connecting @xmath739 to @xmath740 . by ( [ eq : distsauss ] ) we conclude that @xmath591 is disjoint from @xmath741 and using ( [ eq : diamcondit ] ) , we obtain that @xmath742 .
moreover , @xmath743 , which by ( [ eq : distsauss ] ) , is contained in @xmath744 . in order to find a path @xmath585 which avoids @xmath745 , we still need to modify @xmath591 in a way that it does not intersect the cut - points @xmath746 .
recall that the path @xmath591 is disjoint from @xmath676 , @xmath747 and by ( [ eq : farcuts ] ) , all the neighbors of @xmath706 which are not in @xmath748 are in @xmath705 .
one can define @xmath585 by the following : whenever @xmath749 , the piece @xmath750 is replaced by some path in @xmath705 connecting @xmath751 to @xmath752 . since @xmath753 and @xmath754 , see ( [ eq : ringskin ] ) and ( [ eq : distsauss ] ) , we conclude that @xmath755 is @xmath664-avoidable .
to finish the proof of the lemma , we show that @xmath756 by the lemma [ eq : charfill ] , we conclude that @xmath757 for any finite sets @xmath539 and @xmath243 , so that @xmath758 .
since @xmath759 , we only need to prove that @xmath760 . to this end
we will show that @xmath761 is connected . given @xmath762 ,
we connect them by some path @xmath146 .
using the fact that @xmath755 is avoidable , replace this path by some @xmath585 disjoint from @xmath755 , which also joins @xmath93 to @xmath107 .
thus @xmath761 is connected and with lemma [ eq : charfill ] we find that @xmath763 .
this shows ( [ eq : fillx ] ) , finishing the proof of lemma [ lem : avoid ] .
@xmath294 as a consequence of the result above , we prove in the next corollary that , if the diameter of each ` sausage ' is smaller than @xmath764 , @xmath695 does not separate components .
we write @xmath512 and @xmath765 instead of @xmath766 and @xmath767 for simplicity .
[ lem : chifillx ] ( @xmath3 ) if @xmath673 are cut - points of @xmath768 satisfying ( [ eq : farcuts ] ) and @xmath769 then @xmath770 ( recall the definition of @xmath512 in ( [ eq : chi ] ) ) .
note that the cut - times @xmath771 are not necessarily all the cut - times in a given interval , see comment under ( [ eq : uj ] ) .
_ take @xmath772 such that @xmath773 , @xmath774 .
we first show that the conditions on the diameters of the @xmath775 s and @xmath696 s ensures the existence of a point @xmath776 in @xmath777 , ( @xmath778 ) , c.f .
remark [ rem : bernoulli ] 3 ) .
_ case 1 : there is some @xmath779 such that @xmath780 .
_ in this case , take @xmath107 in this intersection . if @xmath781 ( respectively @xmath782 ) re - choose @xmath107 as one of its neighbors in @xmath705 ( resp .
@xmath783 ) . in any of these situations , although @xmath107 can still be an interior point of @xmath775 , we know that @xmath784 , see ( [ eq : fillx ] ) , ( [ eq : distsauss ] ) and ( [ eq : ringskin ] ) . now ,
take a path @xmath140 from @xmath107 to some @xmath785 and use the fact that @xmath695 is avoidable ( see lemma [ lem : avoid ] ) to find a @xmath585 from @xmath107 to @xmath786 which is disjoint from @xmath695 . finally , we take @xmath776 to be the first point of this path in @xmath787 .
_ case 2 : _
@xmath788 by claim [ eq : diamfill ] , @xmath789 implying that @xmath790 for any @xmath791 .
this , together with the connectedness of the @xmath775 s and ( [ eq : aiufill ] ) imply that @xmath792 in this case , we can choose @xmath776 to be any point in @xmath787 and it will automatically be out of @xmath695 . otherwise , by ( [ eq : fillx ] ) , its neighbor in @xmath775 would contradict ( [ eq : aidisj ] ) . now that we have ( for @xmath774 ) a point @xmath776 in @xmath793
, we take any path @xmath140 in @xmath794 connecting @xmath795 to @xmath796 . using ( [ eq : maxdiam ] ) and the lemma [ lem :
avoid ] we obtain a modified path @xmath585 connecting @xmath787 to @xmath246 which is disjoint from @xmath695 and contained in @xmath797 @xmath798 .
we proved that @xmath78 and @xmath79 are not separated by @xmath695 in @xmath799 and since the choice of @xmath78 and @xmath79 was arbitrary , @xmath800 , concluding the proof of corollary [ lem : chifillx ] .
@xmath294 we now obtain estimates on the the diameter of the ` sausages ' and on the stopping time @xmath801 for a typical random walk trajectory , when @xmath0 .
recall that our cut - times are defined for doubly infinite trajectories ( in fact this has simplified the exposition of lemma [ lem : avoid ] ) .
so , we now artificially introduce a negative time for our random walk trajectory by considering an independent copy of @xmath177 .
more precisely , let @xmath802 denote the canonical coordinates of the second process on @xmath803 .
[ lem : estimates ] ( @xmath0 ) given @xmath42 , and integers @xmath460 and @xmath461 , for large enough @xmath462 and every @xmath804 , with @xmath803-probability at least @xmath805 we can find cut times @xmath806 such that : @xmath807 the number @xmath808 is deterministic and only depends on @xmath809 , @xmath667 and @xmath49 .
the strategy to prove this lemma can be roughly described as follows .
given integers @xmath810 , @xmath70 and @xmath43 , consider the time interval @xmath811 .
we split this interval into @xmath812 subintervals of length @xmath813 , where we expect to find the cut - times @xmath814 .
we now make a brief comment on how we are going to pick the constants @xmath70 , @xmath810 and @xmath16 .
heuristically , we choose : @xmath815 finally , given an @xmath816 , we find a @xmath16 such that @xmath817 and consider the partition of @xmath818 as above . according to ( [ eq : descrestimates ] ) @xmath638 , with high probability we can find @xmath812 cut - times , one in each subinterval of our partition .
we retain only @xmath819 of these cut - times ( one in every other subinterval ) to ensure that they are not adjacent .
moreover , the choice of constants in ( [ eq : descrestimates ] ) @xmath51 and @xmath60 will ensure that ( [ eq : estimates ] ) @xmath60 and @xmath638 hold with high probability .
_ proof of the lemma [ lem : estimates ] .
_ as we now explain , there is a cut - time in the interval @xmath820 with high @xmath803-probability as @xmath76 grows , i.e. @xmath821 \xrightarrow[m \rightarrow \infty ] { } 0.\ ] ] recall that our definition of cut - times is slightly different from the definition that for instance appears in @xcite or @xcite , p.88 .
nevertheless , a slight modification of the argument in @xcite , p.89 proves that @xmath822 > 0 $ ] ( in ( [ eq : bounddist ] ) we perform a similar calculation ) .
the statement ( [ eq : qmtozero ] ) now follows from the ergodicity of the increments of @xmath823 under @xmath824 .
let @xmath825 ( for @xmath24 ) denote the exit time of the ball @xmath22 , recall the definition at the beginning of section [ sec : review ] .
we now choose the integers @xmath826 and @xmath827 , see ( [ eq : descrestimates ] ) @xmath51 . to this end , using the invariance principle , we note that , for large enough @xmath828 , @xmath829 = b < 1.\ ] ] applying the markov property at the times @xmath830 , we have , for large enough @xmath831 , @xmath832 \leqslant p_0[t_{2lf(k+1 ) } > ( fk)^2]^{m_0 } \leqslant b^{m_0 } < \epsilon/3,\ ] ] for every @xmath833 and @xmath834 .
this completes our choice of @xmath835 and @xmath827 in ( [ eq : descrestimates ] ) @xmath51 .
we now establish estimates on the diameter of the paths performed in each subinterval of length @xmath813 , see ( [ eq : descrestimates ] ) @xmath60 .
let @xmath836 stand for the one - dimensional simple random walk , see below ( [ eq : departreturn ] ) .
it follows from a variation of azuma s inequality , see for instance @xcite , ( 41 ) , p.28 , that @xmath837 & \leqslant 2d\ , p^1_0\big[\max_{l \leqslant k^2 } s_l > \tfrac{fk}{90g}\big ] \leqslant 4d \ , e^{- \tfrac{f^2}{(90g)^2 } } , \text { for every $ f$ , $ k$ and $ g \geqslant 1$}. \end{split}\ ] ] thus we can choose a large enough @xmath838 such that @xmath839 < \epsilon/3 , \text { for every $ k \geqslant 1$}.\ ] ] as described in ( [ eq : descrestimates ] ) @xmath638 , we want to find cut - times in all the @xmath840 intervals ( of length @xmath813 ) of our partition . for this , using ( [ eq : qmtozero ] ) , we pick @xmath841 so that for every @xmath842 , @xmath843 finally , given an @xmath816 , we choose @xmath842 such that @xmath844 .
the bound below is the precise implementation of ( [ eq : descrestimates ] ) .
@xmath845\\ & \quad \cup \big\ { t_{lf_0(k + 1 ) } > m_0(f_0 k)^2 \big\ } \bigg ] \overset{(\ref{eq : manycuts})(\ref{eq : manyexits})(\ref{eq : exitslow } ) } { < } \epsilon .
\end{split}\ ] ] on the complement of the event appearing above , we choose @xmath819 cut - times in every other time interval ( consequently they are not adjacent ) . for instance , we can choose the first cut - time of the intervals below @xmath846 this ensures ( [ eq : estimates ] ) @xmath51 . recall the definition of the sets @xmath847 and the comment below ( [ eq : uj ] ) .
since @xmath844 and @xmath848 on the event in ( [ eq : goodcuttimes ] ) , we have that ( [ eq : estimates ] ) @xmath60 and @xmath638 hold . @xmath294 [ rem : d3 ] there are results concerning `` monolateral '' cut - times of random walks when @xmath849 .
in particular , the number of cut - times between zero and @xmath141 ( with a different definition ) grows sub - linearly .
see for instance , @xcite .
the next result concludes what we called the first step of the proof of theorem [ th : local ] .
it establishes that the probability that one random walk trajectory separates macroscopic components inside a box goes to zero as the diameter of the box grows .
moreover , this limit is uniform in the points where we condition the random walk to enter and exit large neighborhoods of the box .
recall that a similar uniformity was important to obtain ( [ eq : boundsecond ] ) .
the proof of the following lemma combines lemmas [ lem : avoid ] and [ lem : estimates ] , corollary [ lem : chifillx ] and the harnack inequality ( [ eq : harnack ] ) .
consider the random time @xmath850 note that @xmath473 is not a stopping time .
[ lem : conditioned ] ( @xmath0 ) given @xmath851 , @xmath460 and @xmath461 , for large enough @xmath462 , @xmath852 } ) ) = 0 , \operatorname{\textnormal{fill}}(x_{[0,s ] } ) \text { is $ ( \tfrac{l_0}{2g})$-avoidable in $ c^3 $ } \big ] > 1-\epsilon.\ ] ] see ( [ eq : chi]),([eq : pxy ] ) and definition [ def : avoidable ] for the notation . _
proof_. as an intermediate step , we show that with @xmath667 and @xmath49 as above , for every @xmath853 , there is a large enough @xmath50 such that @xmath854 } ) ) = 0 \textnormal { and } \operatorname{\textnormal{fill}}(x_{[0,t ] } ) \textnormal { is } \big ( \frac{l_0}{2 g } \big)\textnormal{-avoidable in } c^3 \end{array } \right ] > 1-\epsilon'.\ ] ] to this end , we first take @xmath50 as in lemma [ lem : estimates ] . we know that on an event with probability at least @xmath855 we can choose the cut - times @xmath856 satisfying ( [ eq : estimates ] ) , recall the observation below ( [ eq : uj ] ) . on this event , for any @xmath857 such that @xmath858 , we take @xmath859 such that @xmath860 . using ( [ eq : chimonotone ] ) and corollary [ lem : chifillx ] , we conclude that @xmath861 } ) \big ) \leqslant \chi \big(\operatorname{\textnormal{fill}}(x_{(n_0,n_{\bar j + 1 } ) } ) \big ) = 0.\ ] ] since ( [ eq : estimates ] ) @xmath60 holds , we can take @xmath862 in lemma [ lem : avoid ] to obtain that @xmath863 is @xmath864-avoidable in @xmath9 and by ( [ eq : avoidmonotone ] ) , @xmath865 since @xmath866 } ) \subset \operatorname{\textnormal{fill}}(x_{(n_0,n_{\bar j + 1})}),\ ] ] according to ( [ eq : avoidremark ] ) and ( [ eq : x0j1avoidable ] ) , all we need in order to show that @xmath867})$ ] is also @xmath864-avoidable in @xmath799 is that @xmath868 using ( [ eq : fillx ] ) , we obtain that @xmath869 is contained in @xmath870 .
but since @xmath871 and @xmath872 , we know by ( [ eq : estimates ] ) @xmath60 that @xmath870 is disjoint from @xmath873 .
this establishes ( [ eq : samefill ] ) and consequently ( [ prop : freewalk ] ) .
we now introduce the stopping time @xmath874 } ) ) = 1 \text { or}\\ & \operatorname{\textnormal{fill}}(x_{[0,s ] } ) \text { is not $ ( \tfrac{l_0}{2g})$-avoidable in $ c^3 $ } \big\ } , \end{split}\ ] ] and note that the event appearing in ( [ eq : conditioned ] ) contains @xmath875 .
define , for @xmath876 and @xmath877 the function @xmath878 $ ] which is harmonic in @xmath879 . given @xmath880 and @xmath876
, we use the strong markov property at time @xmath881 to obtain @xmath882 = \frac{p_x[s'<d , x_d = y]}{h^y(x ) } \leqslant \sup_{x , z \in c^4 } \frac{h^y(z)}{h^y(x ) } p_x[s'<d].\ ] ] by the harnack inequality , see @xcite theorem 1.7.6 p.46 , we have @xmath883 and the lemma [ lem : conditioned ] follows from ( [ prop : freewalk ] ) by choosing @xmath884 .
@xmath294 in theorem [ th : local ] , one considers @xmath667 independent paths instead of just one as in the above lemma , see also remark [ rem : g100 ] .
the following lemma is the key step to obtain this extension .
as explained at the beginning of this section , lemma [ lem : farapart ] shows that for any family of @xmath667 connected sets , which do not separate components , are avoidable and mutually far apart , the union of these sets also does not separate components .
more precisely , [ lem : farapart ] let @xmath885 be connected sets .
setting @xmath886 , if the following holds : @xmath887 then , @xmath888 _ proof .
_ as we now show , we can assume without loss of generality that @xmath889 indeed , if @xmath890 for some distinct pair @xmath891 , we eliminate this @xmath892 .
so we can assume ( [ eq : redfarapart ] ) .
hence , with lemma [ prop : fill ] and ( [ eq : chioi ] ) @xmath60 we conclude that @xmath893 to obtain @xmath894 ( see ( [ eq : chi ] ) ) , we will prove that @xmath895 as a further reduction , we are going to prove that ( [ eq : da1a2anddiam ] ) follows if one shows ( [ eq : da1a2anddiam ] ) when @xmath896 to prove the above reduction , it suffices to show the following fact .
given any pair @xmath897 satisfying ( [ eq : da1a2anddiam ] ) , but such that @xmath898 , @xmath899 in other words , if @xmath900 separates @xmath78 from @xmath79 in @xmath799 , it also separates @xmath550 from @xmath552 in @xmath799 .
let us now explain how this fact is proved . using ( [ eq : da1a2anddiam ] ) and
lemma [ prop : fill ] , since @xmath898 , we know that either @xmath901 or @xmath902 . we suppose without loss of generality that we are in the former case .
we choose the sets @xmath903 and @xmath550 to be some face of @xmath904 , see figure [ fig : a1a2f ] .
since @xmath905 , to establish ( [ eq : pathtouches ] ) , is enough to show that any path connecting @xmath906 to @xmath907 must also intersect @xmath245 .
this is done in the next paragraph . to @xmath907
meets @xmath245.,title="fig:",scaledwidth=25.0% ] + suppose by contradiction that there is a path from @xmath906 to some @xmath908 , which does not meet @xmath245 ( and which by ( [ eq : da1a2anddiam ] ) does not meet @xmath78 ) , we could continue this path to the neighbor of @xmath107 in @xmath550 , then to @xmath909 and finally to infinity , without touching @xmath78 .
this would contradict , using lemma [ eq : charfill ] , the fact that @xmath901 .
using a similar , although simpler reasoning , we also obtain that @xmath910 , implying that @xmath911 . since @xmath912 , we proved ( [ eq : pathtouches ] ) and consequently ( [ eq : redfarapart2 ] ) . given a pair @xmath897 , satisfying the conditions in ( [ eq : da1a2anddiam ] ) and ( [ eq : redfarapart2 ] ) , we will exhibit a path in @xmath913 connecting @xmath245 to @xmath246 .
first we find a point @xmath795 in @xmath914 .
for this , note that by ( [ eq : fillstconn ] ) , @xmath915 and by ( [ eq : redfarapart2 ] ) , we can assume @xmath916 , so that @xmath917 in order to find a point @xmath795 in @xmath914 , we first take any @xmath918 , which by ( [ eq : boundfill ] ) is also in @xmath245 . if @xmath919 , we are done , otherwise , let @xmath920 be such that @xmath921 . by ( [ eq : chioi ] ) @xmath51 and ( [ eq : pfilla1separates ] ) , we are able to find some @xmath922 .
using the @xmath108-connectedness of @xmath923 ( see ( [ eq : pfilla1conn ] ) ) we join @xmath924 to @xmath925 by a @xmath108-path @xmath146 in @xmath923 , and take @xmath795 to be the first point of @xmath146 out of @xmath926 .
we conclude from ( [ eq : doioj ] ) that @xmath927 . in the same way
, we find some @xmath928 and join @xmath795 to @xmath796 by any path @xmath929 . roughly speaking , to conclude the proof we will modify @xmath667 times the path @xmath140 ( using ( [ eq : chioi ] ) @xmath638 ) in order to avoid each set @xmath930 . since @xmath931 is @xmath864-avoidable in @xmath799 ( see ( [ eq : chioi ] ) @xmath638 )
, we can find a modification @xmath932 of @xmath140 , joining @xmath795 to @xmath796 , which is disjoint from @xmath931 and such that @xmath933 .
we proceed by induction .
suppose that for some @xmath934 we found some @xmath935 joining @xmath795 to @xmath796 such that @xmath936 we use ( [ eq : chioi ] ) @xmath638 , see also definition [ def : avoidable ] , to find a path @xmath937 joining @xmath795 to @xmath796 such that @xmath938 the existence of @xmath939 as above implies @xmath894 and consequently , lemma [ lem : farapart ] . @xmath294 the next lemma is the final ingredient to prove the main result of this section .
it will ensure that with high probability a set of @xmath667 independent random walks satisfy the hypothesis ( [ eq : chioi ] ) @xmath60 of the lemma [ lem : farapart ] , or in other words : they are mutually far apart .
the proof of this lemma is an adaptation of known arguments concerning intersection of random walk trajectories for @xmath0 , see for instance , @xcite p.89 .
we denote by @xmath940 , @xmath941 and @xmath942 the times @xmath625 , @xmath943 and @xmath473 pertaining to the walks @xmath944 , recall ( [ eq : times ] ) . [ lem : gfarapart ] ( @xmath0 ) given @xmath42 and @xmath461 , for large enough @xmath462 , @xmath945}^1 , x_{[h_{c^4}^2 , s^2]}^2 ) \leqslant 1 \big ] < \epsilon \end{split}\ ] ] _ proof . _ as in the proof of lemma [ lem : conditioned ] , we define @xmath878 $ ] , for @xmath946 , @xmath947 , recall that for @xmath948 we write @xmath109 if they are @xmath108-neighbors . for @xmath949 , @xmath950 and @xmath951 , we have @xmath952}^1 , x_{[h_{c^4}^2 , s^2]}^2 ) \leqslant 1 \big]\\ \vspace{2 mm } \displaystyle \leqslant p_{x_1,y_1}\otimes p_{x_2,y_2 } \bigg [ \begin{array}{c } \text{there are } s_1 \leqslant t_1 < d^1 , s_2 \leqslant t_2 < d^2;\\ x_{t_1}^1 , x_{t_2}^2 \in c^4 \text { and } x_{s_1}^1 \overset{*}{\leftrightarrow } x_{s_2}^2 \end{array } \bigg]\\ \displaystyle \leqslant \sum_{s_1 \geqslant 0 } \sum_{s_2 \geqslant 0 } \,\,\ , \sum_{\substack{z_1,z_2 \in c^{l/4 } ; z_1 \overset{*}{\leftrightarrow } z_2 } } p_{x_1,y_1}[s_1 < d , x_{s_1 } = z_1 ] p_{z_1,y_1}[h_{c^4 } < d]\\ \displaystyle \phantom{= \sum_{s_1 \geqslant 0 } \sum_{s_2 \geqslant 0 } \sum_{z_1 \overset{*}{\leftrightarrow } z_2 } } \quad p_{x_2,y_2}[s_2 < d , x_{s_2 } = z_2 ] p_{z_2,y_2}[h_{c^4 } < d]\\ \displaystyle \leqslant \sum_{s_1 \geqslant 0 } \sum_{s_2 \geqslant 0 } \,\,\ , \sum_{z_1,z_2 \in c^{l/4 } ; z_1 \overset{*}{\leftrightarrow } z_2 } p_{x_1}[s_1 < d , x_{s_1 } = z_1 ] \frac{h^y(z_1)}{h^y(x_1 ) } \sup_{w_1 \in c^4 } \frac{h^y(w_1)}{h^y(z_1)}\\ \displaystyle \phantom{= \sum_{s_1 \geqslant 0 } \sum_{s_2 \geqslant 0 } \sum_{z_1 \overset{*}{\leftrightarrow } z_2 } } \quad p_{x_2}[s_2
< d , x_{s_2 } = z_2 ] \frac{h^y(z_2)}{h^y(x_2 ) } \sup_{w_2 \in c^4 } \frac{h^y(w_2)}{h^y(z_2 ) } , \end{array}\ ] ] using reversibility and harnack s inequality ( [ eq : harnack ] ) for @xmath953 , @xmath578 , ( note that we can not use it with the @xmath954 s since they could be out of @xmath955 ) , @xmath956 p_{z_2}[s_2 < d , x_{s_2 } = x_2 ] , \end{array}\ ] ] linking @xmath957 to @xmath958 in at most @xmath31 steps , we bound the expression above by @xmath959 .
\end{array}\ ] ] finally we use the heat kernel estimate @xmath960 \leqslant cn^{d/2}\exp(\tfrac{|x - y|^2}{cn})$ ] , see @xcite ( 2.4 ) , to obtain @xmath961}^1 , x_{[h_{c^4}^2 , s^2]}^2 ) \leqslant 1 \big ] \leqslant c \sum_{r \geqslant 0 } \frac{r+1}{r^{d/2 } } e^{-c'\tfrac{l_0 ^ 2}{r}}.\ ] ] this last quantity , independently on the choice of @xmath795 , @xmath796 , @xmath962 , @xmath963 , goes to zero as @xmath50 goes to infinity ( recall that @xmath0 ) , and lemma [ lem : gfarapart ] follows .
@xmath294 the next theorem is the main result of this section , recall that it was already stated and used in section [ sec : proof ] .
its proof combines lemmas [ lem : conditioned ] , [ lem : farapart ] and [ lem : gfarapart ] .
recall the definition of the time @xmath473 in ( [ eq : times ] ) and the sets @xmath964 , @xmath965 below ( [ eq : pxy ] ) .
[ th : local ] ( @xmath0 ) given @xmath42 , @xmath460 and @xmath461 , for large enough @xmath462 , @xmath966 } \bigg ) = 1 \right ] < \epsilon.\ ] ] _ proof . _
fix @xmath967 and @xmath968 and recall that for any @xmath360 , @xmath969 depends only on @xmath970 . roughly speaking , in lemma [ lem : farapart ] we have seen that if a union of @xmath667 sets separates components in @xmath799 then : these sets are not mutually far apart , or one of them either separates components in @xmath799 or is not @xmath864-avoidable in @xmath799 . more precisely , lemma [ lem : gfarapart ] yields the following : @xmath971}^i \big ) = 1 \bigg ] } \leqslant p_{\vec{x},\vec{y } } \left [ \begin{array}{c } \text{there are } 1 \leqslant i < j \leqslant g \text { such that $ h_{c^4}^i < d^i$,}\\ h_{c^4}^j < d^j \text { and } d_\infty(x_{[h_{c^4}^i , s^i]}^i , x_{[h_{c^4}^j , s^j]}^j ) \leqslant 1 \end{array } \right ] \end{array}\ ] ] @xmath972}^i ) ) = 1 \text { or } \operatorname{\textnormal{fill}}(x_{[h_{c^4}^i , s^i]}^i ) \text { is not $ ( \tfrac{l_0}{2g})$-avoidable in $ c^3 $ } \end{array } \right ] , \end{array}\ ] ] and using in both terms the strong markov property for @xmath973 at time @xmath974 , @xmath975}^1 , x_{[h_{c^4}^2 , s^2]}^2 ) \leqslant 1 \big]\\ \end{array}\ ] ] @xmath976 } ) ) = 1 \text { or } \operatorname{\textnormal{fill}}(x_{[0,s ] } ) \text { is not $ ( \tfrac{l_0}{2g})$-avoidable in $ c^3 $ } \big ] , \end{array}\ ] ] which , by the lemmas [ lem : gfarapart ] and [ lem : conditioned ] , can be made arbitrarily small once we choose @xmath50 large enough . finishing the proof of the theorem [
th : local ] .
@xmath294 this concludes the proof of ( [ th : reftolocal ] ) and provides the last missing piece of the proof of our main theorem [ th : main ] .
the present work leaves several questions untouched .
for instance : - can one improve theorem [ th : taildiam ] in such a way that the exponents of @xmath336 in the lower and upper bound match ?
if the answer is affirmative , what is this exponent ? the same questions can be asked for @xmath977 in theorem [ th : tailvol ] . -
which results of this article can be extended to any @xmath19 or for @xmath849 ? see the remark [ rem : bernoulli ] 4 ) , the beginning of section [ sec : local ] and remark [ rem : d3 ] .
- how does the size of @xmath38 behave in the sub - critical phase @xmath21 ?
in this appendix we prove lemma [ prop : fill ] which we used often in section [ sec : local ] .
_ proof of lemma [ prop : fill ] .
_ first we prove that @xmath978 , see ( [ eq : boundfill ] ) . if @xmath979 , one can join @xmath164 to @xmath162 and then to infinity without meeting @xmath539 , a contradiction to the assumption that @xmath980 , in view of lemma [ eq : charfill ] .
this proves ( [ eq : boundfill ] ) .
finally we prove ( [ eq : threecases ] ) , the last part of lemma [ prop : fill ] .
consider finite and connected sets @xmath983 .
we claim that , if @xmath679 , then @xmath984 .
indeed , by claim [ eq : boundfill ] , every point of @xmath985 is a neighbor of @xmath539 , so that they can not belong to @xmath243 .
the same argument applies replacing @xmath539 with @xmath243 .
it is not possible that @xmath986 and @xmath988 .
this would imply @xmath990 and by claim [ eq : boundfill ] that @xmath991 .
so we have three remaining possibilities which correspond to the cases enumerated in lemma [ prop : fill ] .
first we show that @xmath994 .
since @xmath995 is finite , it is enough to show that the boundary of this set is empty .
suppose @xmath996 and let @xmath107 be a neighbor of @xmath93 . if @xmath997 , then by claim [ eq : boundfill ] @xmath535 in contradiction with the fact that @xmath989 .
using a symmetric argument for @xmath998 , we conclude that @xmath999 , showing that the boundary of this set is empty .
this implies that @xmath994 .
we then prove that @xmath992 by excluding the possibility that there are neighbors @xmath1000 and @xmath1001 .
suppose by contradiction the existence of such @xmath93 and @xmath107 .
since @xmath994 , @xmath997 , implying that @xmath1002 , by claim [ eq : boundfill ] , a contradiction with @xmath1003 .
if in addition @xmath680 , we now show that @xmath993 . indeed ,
suppose by contradiction that @xmath1000 and @xmath1001 are @xmath108-neighbors .
since @xmath994 , we conclude that @xmath1004 and @xmath1005 .
take a path @xmath140 joining @xmath93 to @xmath107 such that @xmath1006 .
we denote by @xmath924 the last visit of @xmath140 to @xmath132 and by @xmath786 the first visit of @xmath140 to @xmath998 .
since @xmath1007 and @xmath1008 , we obtain a contradiction with the hypothesis @xmath680 .
@xmath294 lllll i. benjamini , a.s .
sznitman - giant component and vacant set for random walk on a discrete torus , j. eur .
, 10(1 ) , 1 - 40 , ( 2008 ) .
a. dembo , a.s .
sznitman - on the disconnection of a discrete cylinder by a random walk , probab .
theory relat .
fields , 136(2 ) , 321 - 340 , ( 2006 ) .
a. dembo , a.s .
sznitman , - a lower bound on the disconnection time of a discrete cylinder , progress in probability , vol . 60 , in and out of equilibrium 2 , birkhuser , basel , 211 - 227 , ( 2008 ) .
deuschel , a. pisztora - surface order large deviations for high - density , probab .
theory relat .
fields 104 , 467 - 482 , ( 1996 ) .
p. erds , s.j .
taylor - some intersection properties of random walk paths , acta math .
hungar . 11 , 231 - 248 , ( 1960 ) .
a. grigoryan , a. telcs - sub - gaussian estimates of heat kernels on infinite graphs , duke math .
j. , 109(3 ) , 451 - 510 , ( 2001 ) .
g. grimmett - percolation , springer verlag , second edition , new york ( 1999 ) .
h. kesten .
- aspects of first passage percolation , ecole dt de probabilit de saint - flour xiv-1984 125 - 264 .
lecture notes in math .
1180 , springer verlag , new york ( 1986 ) .
lawler - intersections of random walks , birkhuser , basel ( 1991 ) .
lawler - cut times for simple random walk , elect .
journal of probab .
, vol 1 , 1 - 24 , ( 1996 ) . c. mcdiarmid , concentration .
probabilistic methods for algorithmic discrete mathematics , 195 - 248 , algorithms combin . , 16 , springer , ( 1998 ) .
resnick - extreme values , regular variation and point processes , springer verlag , ( 1987 ) .
v. sidoravicius , a.s .
sznitman - percolation for the vacant set of random interlacements , comm .
pure appl .
math . , 62(6 ) , 831 - 858 , ( 2009 ) .
a.s . sznitman - vacant set of random interlacements and percolation , to appear in the annals of mathematics , also available at arxiv:0704.2560 , ( 2010 ) .
sznitman - random walks on discrete cylinders and random interlacements , probab .
theory relat .
fields , 145 , 143 - 174 , ( 2009 ) .
a.s . sznitman - upper bound on the disconnection time of discrete cylinders and random interlacements , annals of probability , 37(5 ) , 1715 - 1746 , ( 2009 ) .
sznitman - on the domination of random walk on a discrete cylinder by random interlacements , elect .
journal of probab .
, vol 14 , 1670 - 1704 , ( 2009 ) .
a.q . teixeira - on the uniqueness of the infinite cluster of the vacant set of random interlacements , annals of applied probability .
19(1 ) , 454 - 466 , ( 2009 ) .
d. windisch - on the disconnection of a discrete cylinder by a biased random walk , annals of applied probability .
18(4 ) , 1441 - 1490 , ( 2008 ) .
d. windisch - random walk on a discrete torus and random interlacements , elect .
communic . in probab .
, vol 13 , 140 - 150 , ( 2008 ) . | in this paper we establish some properties of percolation for the vacant set of random interlacements , for @xmath0 and small intensity @xmath1 .
the model of random interlacements was first introduced by a.s .
sznitman in @xcite .
it is known that , for small @xmath1 , almost surely there is a unique infinite connected component in the vacant set left by the random interlacements at level @xmath1 , see @xcite and @xcite .
we estimate here the distribution of the diameter and the volume of the vacant component at level @xmath1 containing the origin , given that it is finite .
this comes as a by - product of our main theorem , which proves a stretched exponential bound on the probability that the interlacement set separates two macroscopic connected sets in a large cube . as another application
, we show that with high probability , the unique infinite connected component of the vacant set is `` ubiquitous '' in large neighborhoods of the origin . |
quantum magnets exhibit a wealth of interesting phenomena , in particular on low - dimensional frustrated lattices , where both enhanced quantum fluctuations and geometric frustration can destroy semi - classical magnetic order .
indeed , various compounds have been characterized to provide realizations of the above paradigm .
recent examples include the valence bond solids found in ( c@xmath2h@xmath3)(ch@xmath4)@xmath4p[pd(dmit)@xmath2]@xmath2 @xcite and zn@xmath6cu@xmath7(od)@xmath8cl@xmath2 @xcite .
another material , that has been intensively studied is srcub@xmath2(bo@xmath4)@xmath9 ( we refer to ref . for a detailed review of the various experimental and theoretical explorations on this system ) .
this compound is described well by the dimer singlet ground state proven exactly previously by shastry and sutherland to exist @xcite in the spin-1/2 heisenberg model on the orthogonal dimer lattice , shown in fig .
recently , new results on the magnetization process of srcub@xmath2(bo@xmath4)@xmath9 have been presented . in particular , magnetization plateaus at @xmath10 , @xmath11 , @xmath12 , @xmath13 , and @xmath14 of the saturated magnetization
have been reported @xcite , in addition to the previously established plateaus at @xmath15 , @xmath16 and @xmath0 .
while the existence of some of the reported plateaus is at the moment controversial @xcite , new b@xmath17 nmr data @xcite and magnetic torque measurements @xcite provide evidence in favor of a persistent crystalline structure of magnetic excitations also above the @xmath15 plateau .
the presence of intra - dimer dzyaloshinskii - moriya interactions in srcub@xmath2(bo@xmath4)@xmath9 however calls for a more complex scenario than a direct interpretation in terms of supersolidity of triplet excitations in this regime @xcite . as another realization of the shastry - sutherland geometry the rare earth tetraborid tmb@xmath18 @xcite
has recently been studied in finite magnetic fields .
in contrast to srcub@xmath2(bo@xmath4)@xmath9 , this metallic compound exhibits stable long - range antiferromagnetic order in zero field below about 9.8 k. since full saturation can be obtained for magnetic fields parallel to the c - axis above 5 t , tmb@xmath18 allows for a complete scan of its magnetization process . for this compound magnetization plateaus
have been observed e.g. at fractions 1/2 , 1/7 , 1/8 and 1/9 of magnetic saturation . despite the metallic nature of tmb@xmath18 , its magnetism has been suggested to realize an easy - axis anisotropic version of the shastry - sutherland model close to the ising limit with similar intra- and inter - dimer coupling strengths @xcite .
( solid lines ) and nearest neighbor ( inter - dimer ) coupling @xmath19 ( dashed lines).,scaledwidth=40.0% ] in light of the progress in realizing novel quantum phases in frustrated quantum magnets , it is important to explore in detail the interplay between geometric frustration and quantum fluctuations in such systems based on effective spin models . in many aspects ,
numerical studies have become especially important as an unbiased approach to quantum magnetism .
however , numerical studies of even simple models of frustrated spin systems suffer from severe restrictions on the finite sizes accessible to current simulation techniques .
in particular , quantum monte carlo ( qmc ) simulations are tampered by a notorious sign - problem @xcite due to odd - length spin - exchange paths appearing on non - bipartite lattices .
this usually restricts unbiased numerical studies to the small lattices accessible to exact numerical diagonalization .
noteworthy in this respect are however recent studies employing the density matrix renormalization group algorithm on the triangular and kagome lattice heisenberg model @xcite . here , we employ a different approach in order to explore the interplay between quantum fluctuations and frustration , by studying a model of quantum magnetism in a parameter regime , where geometric frustration is restricted to the classical sector , and does not lead to qmc sign problems .
this allows us to employ large - scale qmc simulations to study quantum effects on a frustrated spin system .
in particular , we study the ground state properties of the spin-1/2 easy - axis heisenberg model on the orthogonal dimer lattice considered by shastry and sutherland @xcite .
namely , we consider the xxz hamiltonian @xmath20\nonumber\\ + j'&\sum_{\langle\langle i , j \rangle\rangle } & \left[-\delta(s^x_i s^x_j+s^y_i s^y_j)+ s^z_i s^z_j\right],\end{aligned}\ ] ] a variant of the model considered in ref . with ferromagnetic transverse exchange ( @xmath21 ) and antiferromagnetic ising exchange interactions @xmath22 . here
, @xmath23 denotes a spin-1/2 degree of freedom on site @xmath24 of the square lattice , and the first sum extends over all nearest neighbor bonds .
the second sum runs over a staggered subset of the next - nearest neighbor bonds , as indicated in fig . 1 . the model in eq .
( 1 ) maintains the frustrated nature of the antiferromagnetic ising interactions , and introduces ferromagnetic spin exchange terms . employing the well known mapping between spin-1/2 degrees of freedom and hard - core bosons ,
the model can be mapped onto an extended bosonic hubbard model of hard - core bosons hopping along the bonds of the shastry - sutherland lattice and experiencing repulsive interactions proportional to the strength of the ising exchange .
recently , similar hard - core boson models have been studied on different lattice geometries , and were found to exhibit interesting order - by - disorder phenomena when quantum fluctuations lift an extensive ground state degeneracy from the ising limit @xmath25 , with new quantum phases emerging .
examples include a supersolid phase on the triangular lattice @xcite , valence - bond - solids @xcite and a @xmath26 spin liquid @xcite on the kagome lattice , and a @xmath27 liquid on the pyrochlore lattice @xcite . in the limit of dominating kinetic terms ,
such models stabilize a superfluid phase on both bipartite and non - bipartite lattices . in magnetic language
, the superfluid corresponds to a transverse ferromagnetic spin alignment , driven by the ferromagnetic nature of the transverse spin exchange .
for the remainder of the paper , we prefer using the spin language , but occasionally find it convenient to also employ the bosonic notation . as reviewed in the following section , the antiferromagnetic ising model on the shastry - sutherland lattice exhibits regions of extensive ground state degeneracy , similar to the ising model on the triangular and kagome lattices .
motivated by the above mentioned studies on these frustrated geometries , we here assess the effects of quantum fluctuations on the classical degenerate ground states on the shastry - sutherland lattice , and explore the phase diagram of the full quantum model .
we find in this system a dimer triplet state , discussed in detail below , to emerge out of the classical degenerate region .
in addition , the system shows a nel ordered phase and a superfluid regime .
we study the quantum phase transitions between these different phases , and consider the effects of a magnetic field .
we do not obtain indications for supersolidity in this model , but find that quantum effects lead to the stabilization of a magnetization plateau at 1/2 of the full saturation , that does not persist down to the ising limit .
this is in contrast to the case of e.g. the triangular and kagome lattice , where all plateaus found in the quantum regime persist down to the ising limit , where they have a largest extension .
the remainder of the paper is organized as follows : in the next section , we review the properties of the antiferromagnetic ising model on the shastry - sutherland lattice .
then , we present in sec .
iii our numerical results on the phase diagram of the model introduced above .
in order to explain in a simple picture the emergence of the dimer triplet phase , we employ degenerate perturbation theory around the ising limit , which will be discussed in sec .
iv . in sec .
v , we analyze the properties of the system in finite magnetic fields , discuss the appearing magnetization plateaus , and scan for supersolid phases .
finally , we conclude in sec .
vi by relating our numerical finding to the properties of the isotropic heisenberg antiferromagnet ( @xmath28 ) on the shastry - sutherland lattice , and discuss connections to recent studies on its magnetization process .
we also comment on the recent work on the compound tmb@xmath18 , suggested to realize the easy - axis antiferromagnetic shastry - sutherland model close to the ising limit ( @xmath29 ) @xcite .
before exploring in detail the phase diagram of the quantum spin model introduced above , it is convenient to review the properties of the ising limit , @xmath25 , discussed in ref . .
( b ) additional configurations allowed on a non - void plaquette at @xmath30 . full ( open ) circles denote spin up ( down ) states.,scaledwidth=45.0% ] in the ising limit , the model in eq .
( 1 ) stabilizes an antiferromagnetic nel phase for sufficiently weak @xmath31 , up to @xmath32 .
for @xmath33 the classical system has a macroscopically degenerate ground state manifold with an extensive ground state entropy @xmath34 k_b n=0.347 k_b n$ ] , from all configurations that cover each of the @xmath31 dimer bonds with a pair of opposite spins . here
, @xmath35 denotes the number of spins .
exactly for @xmath30 , the degeneracy of the ground state manifold is further enlarged , as additional low - energy configurations proliferate .
shastry and sutherland proved a lower bound @xmath36 on the ground state entropy at @xmath30 , via mapping the model to a 10-vertex model and using brading techniques @xcite .
a simple estimate of the ground state degeneracy can be obtained by employing the argument from pauling s estimate of the residual entropy of ice @xcite . for this purpose , consider one of the filled plaquettes on the shastry - sutherland lattice . while for @xmath33 the eight configurations shown in fig .
2 ( a ) provide minimal contributions of this plaquette to the total energy , for @xmath30 the two configurations shown in fig .
2 ( b ) also lead to a minimal energy contribution . given that out of the 16 possible configurations of the four spins forming the plaquette these 10 configurations are thus feasible
, we obtain an estimate for the ground state entropy @xmath37 = 0.458 \ k_{b } n,\ ] ] to be compared to the above bound by shastry and sutherland .
we note , that for @xmath33 the pauling estimate @xmath38 recovers the exact result . besides the ising limit , shastry and sutherland considered the effects of antiferromagnetic transverse spin exchange terms ( @xmath39 in our notation ) , and proved that the system possesses an exact dimer - singlet product eigenstate , that at least for @xmath33 becomes the system s ground state @xcite .
later studies by various groups considered the full quantum phase diagram of this model , which up to date is not conclusively established ( c.f .
ref . for a review of the various theoretical and numerical proposals ) , even though numerical evidence has been put forward , that the @xmath40 symmetric model ( @xmath28 ) features indeed three phases : ( i ) a low-@xmath31 antiferromagnetically ordered nel phase , ( ii ) the large-@xmath31 dimer singlet phase , and ( iii ) an intermediate valence bond crystal ( vbc ) phase , which breaks the lattice symmetry by forming resonating plaquette singlet states on one of the subsets of the void plaquettes of the shastry - sutherland lattice @xcite .
this vbc phase has a two - fold degenerate ground state and a finite spin excitation gap . for the remainder of this work , we study the quantum model in the region @xmath21 , where large - scale qmc simulations are possible , in contrast to the previously studied case of @xmath39 .
furthermore , this model relates directly to a model of hard - core bosons , as mentioned in sec .
in this section , we present our numerical results on the phase diagram of the model in eq . ( 1 ) . these results are based on qmc simulations of finite systems with up to @xmath41 lattice sites , using period boundary conditions . in the simulations , we scaled the inverse temperature as @xmath42 in order to access ground state properties . here , @xmath43 denotes the linear system size .
the qmc simulations were performed employing a generalized directed - loop update @xcite in the stochastic series expansion ( sse ) algorithm @xcite . for the results obtained on the larger lattices ,
in particular in finite magnetic fields , we employed a decoupling of the hamiltonian in plaquette terms , instead of the more conventional bond decoupling for the sse formulation . in fig .
3 , we present the ground state phase diagram resulting from our calculations . .
the dashed - dotted line gives the estimated phase boundary of the dimer triplet phase within perturbation theory around the point @xmath44 , discussed in sec .
, scaledwidth=45.0% ] we find that the extend of the antiferromagnetically ordered nel phase shrinks essentially linearly upon increasing @xmath45 from @xmath46 up to the heisenberg point at @xmath47 ( for @xmath48 , the model reduces to a spin model on the bipartite square lattice , and the sign of @xmath45 can be inverted by an unitary transformation , thus relating the point @xmath47 to the isotropic heisenberg model at @xmath49 ) . in hard - core bosonic language
, the nel phase corresponds to a checkerboard solid with alternating occupation of the lattice sites . in our qmc simulations , we determine the corresponding structure factor @xmath50 for antiferromagnetic order , @xmath51 where @xmath52 , depending on the sublattice , to which lattice site @xmath24 belongs .
nel order is present , if in the thermodynamic limit @xmath53 scales to a finite value . for dominant transverse exchange , @xmath54
, the model reduces to an ferromagnetic xy model on the shastry - sutherland lattice , which in bosonic language relates to an non - frustrated tight - binding hopping model .
hence , the system is expected to exhibit a bosonic superfluid phase for large values of @xmath21 , which in spin language relates to a ferromagnetic ordering within the xy plane .
such a phase is characterized by a finite value of the superfluid density , or spin stiffness , which in the qmc simulations can be obtained from measuring the spin winding number fluctuation @xcite @xmath55 as @xmath56 at fixed @xmath57 as a function of @xmath58 for different system sizes .
, scaledwidth=45.0% ] as an example , we show in fig .
[ spinstiffness_delta0.5_fullscale.fig ] the behavior of @xmath59 for different system sizes as a function of @xmath58 for fixed @xmath57 . a region with finite spin stiffness
is found for @xmath60 .
the strong discontinuity of @xmath59 at @xmath61 indicates that the quantum phase transition between the nel ordered phase and the superfluid is strongly first order .
such behavior is also seen by monitoring the antiferromagnetic structure factor @xmath50 upon crossing the phase boundary , as shown in fig .
[ saf_delta0.5.fig ] , again for @xmath57 . combining the results for @xmath50 and @xmath59
, we obtain no indication for an intermediate region exhibiting both finite superfluidity and diagonal long - range order as inside a supersolid phase , as expected from the commensurate half - filling of the lattice . as a function of @xmath58 for different system sizes .
, scaledwidth=45.0% ] for dominant @xmath31 ( e.g. , @xmath62 at @xmath57 ) , both @xmath50 and @xmath59 eventually vanish .
we explicitly verified that inside this regime the model does not exhibit long - ranged correlations in the longitudinal nor the transverse spin - spin correlation function .
in addition , also the bond - order - wave structure factors do not exhibit long - ranged order in the spin exchange correlation function ( corresponding to kinetic energy correlations in the bosonic model ) .
since the parameter region with @xmath33 approaches the degenerate region of the ising limit for @xmath63 , quantum effects indeed select a unique phase from this degenerate ground state manifold .
in particular , for small values of @xmath64 , the ground state in this large-@xmath31 regime can be obtained using degenerate perturbation theory around the ising limit , discussed in detail in the next section . within first - order perturbation theory in @xmath45 one then finds that for @xmath33 quantum fluctuations select the following dimerized state of localized @xmath65 triplet states on each dimer : latexmath:[\[\label{idealdimernematicstate } here , the direct product extends over all @xmath31 dimer bonds on the lattice .
obviously , this symmetric linear local combination results from the ferromagnetic nature of the transverse spin exchange @xmath67 considered here . for @xmath39 , one
instead recovered the exact dimer singlet state found by shastry and sutherland @xcite .
in contrast to the dimer singlet state however , the above state @xmath68 is not an eigenstate of the hamiltonian for finite values of @xmath19 . as discussed in the following sections , processes in higher order perturbation theory lead to local correlations between the dominant resonances on the dimers .
hence , the ground state in the large-@xmath31 region of the quantum phase diagram does not take the above direct product form , but approaches it for @xmath69 . from the ground state energy
, we still find that the state @xmath68 provides a appropriate variational state for the true ground state in the large-@xmath31 regime .
this can be seen even at @xmath57 , i.e. significantly away from the ising limit , from a comparison between the system s ground state energy @xmath70 and the variational energy of @xmath68 , @xmath71 shown in fig .
[ fig : energycomparison ] . as a function of @xmath58 for @xmath57 and the variational energy of the dimerized state of localized @xmath65 triplet states , @xmath72 .
, scaledwidth=45.0% ] due to the dominant formation of @xmath65 triplet states on the dimer bonds , we denote this magnetically disordered phase as a dimer triplet phase .
since no spatial symmetry is broken in the dimer triplet phase , we expect the quantum phase transition from the superfluid with broken @xmath27 symmetry to the dimer triplet phase to be continuous , and to belong in the universality class of the three - dimensional ( 3d ) @xmath73 model , with a dynamical critical exponent @xmath74 . in order to study the nature of this quantum phase transition in the qmc simulations
, we scanned the transition region at fixed values of either @xmath45 or @xmath58 , varying the other parameter through the phase boundary . denoting the varied parameter by @xmath75 , at a continuous quantum phase transition the spin stiffness scales as @xmath76 with a scaling function @xmath77 , and the correlations length exponent @xmath78 .
here , @xmath79 denotes the relative distance away from the critical point at @xmath80 . from the above scaling relation ,
it follows that @xmath81 can be determined as the crossing point of finite size data for the rescaled spin stiffness @xmath82 .
furthermore , with appropriate values of the critical exponents @xmath83 and @xmath78 , the scaling function @xmath84 is then obtained by plotting @xmath85 vs. @xmath86 for a fixed value of @xmath87 . as an example
, we consider a scan in @xmath88 at a fixed value of @xmath57 , for which the finite size data of @xmath59 is shown in fig [ spinstiffness_delta0.5.fig ] ( a ) . for @xmath74 , we obtain @xmath89 from a clear crossing point in fig [ spinstiffness_delta0.5.fig ] ( b ) , and a clear data collapse within a finite critical region , taking @xmath90 for the 3d @xmath73 model @xcite , as shown in fig .
[ spinstiffness_delta0.5.fig ] ( c ) . at fixed @xmath57 as a function of @xmath58 for different system sizes ( a ) , and rescaled with linear system size @xmath43 ( b ) , with @xmath91 marked by the dashed line .
part ( c ) shows the data collapse expected from a finite size scaling analysis for the 3d @xmath73 universality class .
, scaledwidth=47.5% ] proceeding this way for other values of @xmath45 , we eventually obtained the phase boundary shown in fig .
3 . from this analysis
, we find that at low values of @xmath92 , the dimer triplet phase extends below the line @xmath93 , indicated by the dashed line in fig .
3 . in order to illustrate this explicitly
, we show in the left panels of fig .
[ spinstiffness_delta0.167_f1.928.fig ] the results of the finite size scaling analysis at @xmath94 , where the transition point is located at @xmath95 . at
fixed @xmath94 as a function of @xmath58 for different system sizes ( a ) , and rescaled with linear system size @xmath43 ( b ) , with @xmath96 marked by the dashed line .
part ( c ) shows the data collapse for the 3d @xmath97(2 ) universality class .
right panel : spin stiffness @xmath59 at fixed @xmath96 as a function of @xmath45 for different system sizes ( d ) , and rescaled with linear system size @xmath43 ( e ) , with @xmath94 marked by the dashed line .
part ( f ) shows the data collapse for the 3d @xmath97(2 ) universality class .
, scaledwidth=49.0% ] as a crosscheck , the right panel of fig .
[ spinstiffness_delta0.167_f1.928.fig ] shows the data of the finite size scaling analysis at fixed @xmath96 , where the transition is indeed observed at @xmath94 .
similarly , when varying @xmath45 at fixed @xmath93 a transition point between the dimer triplet phase and the superfluid regime is found at @xmath98 , as extracted from fig .
[ spinstiffness_f2.0.fig ] .
these results suggest that either ( i ) the superfluid region separating the nel ordered phase and the dimer triplet phase persists down to the ising limit , or ( ii ) the first order transition line and the second order transition line meet at a finite value of @xmath45 , or ( iii ) an additional phase appears near @xmath30 for even smaller values of @xmath99 .
such an additional phase might be expected to be selected by quantum effects from the state of enlarged degeneracy of the ising model at @xmath30 for finite @xmath45 .
the qmc simulations could not be extended to significantly smaller values of @xmath45 , due to an reduced efficiency in parameter regions dominated by the frustrated diagonal part of the hamiltonian . however , as discussed in the following section , degenerate perturbation theory in @xmath45 indicates that at @xmath30 the dominant effect of a finite @xmath45 is to effectively drive the system away from @xmath30 towards the region @xmath33 .
this leads us to exclude option ( iii ) from the above list .
scenario ( ii ) would imply a direct first order transition between the nel ordered phase and the dimer triplet state for sufficiently low values of @xmath21 , whereas within scenario ( i ) the superfluid phase would always separate the two phases . at fixed @xmath93 as a function of @xmath45 for different system sizes ( a ) , and rescaled with linear system size @xmath43 ( b ) , with @xmath100 marked by the dashed line .
part ( c ) shows the data collapse expected from a finite size scaling analysis for the 3d @xmath97(2 ) universality class .
, scaledwidth=47.5% ]
in order to study more closely the emergence of the dimer triplet phase from the ising limit upon introducing transverse exchange interactions , we employed degenerate perturbation theory , starting from the degenerate ground state manifold in the ising limit @xmath25 .
first , we consider the region @xmath33 , where the degenerate ground state manifold is spanned by independently placing two opposite spins on each @xmath31-dimer , as discussed in sec .
ii . for each such @xmath31-dimer @xmath101
, we denote these two lowest energy states as latexmath:[\ ] ] where @xmath111 is diagonal in the basis of @xmath103 and @xmath97 introduced above .
the effective hamiltonian @xmath112 , that describes the effective dynamics induced by @xmath113 within the model space @xmath103 is given by degenerate perturbation theory up to third order in @xmath45 as @xcite @xmath114 where @xmath115 is the projection operator onto the model space in terms of the above constructed basis states @xmath116 , and @xmath117 is the resolvent operator with @xmath118 the degenerate ground state energy of @xmath111 , and @xmath119 the energy of the basis state @xmath105 of the orthogonal space @xmath97 .
we start by considering the first order contribution to @xmath112 .
the only possible process in this case is a single spin - flip along a diagonal bond , graphically represented in fig .
[ fig : firstorderperturbation ] . in first order perturbation theory .
full ( open ) circles denote spin up ( down ) states.,scaledwidth=45.0% ] in terms of effective spin operators @xmath120 and @xmath121 , that act on the effective dimer spin states @xmath122 , the effective hamiltonian in first order perturbation thus reads , @xmath123 corresponding to a uniform transverse magnetic field acting on the effective dimer spins .
the lowest - energy eigenstate of @xmath112 for @xmath21 is the direct product state @xmath124 referred to already in eq .
( [ idealdimernematicstate ] ) of sec .
iii , corresponding to the decoupled dimer state with each dimer forming a @xmath65 triplet state . in case of an antiferromagnetic transverse exchange , @xmath39 ,
the lowest energy state of @xmath112 is the dimer singlet state @xmath125 proven to be an exact eigenstate of the full hamiltonian @xmath126 for @xmath39 by shastry and sutherland @xcite .
however , @xmath68 is not an eigenstate of the full hamiltonian , and correlations between the effective dimer spins are introduced in higher order perturbation theory . in order to assess the nature of these correlations , we consider processes occurring in second order perturbation theory . these involve two spin exchanges along axial bonds .
apart from diagonal terms , the result of such processes is again to flip the effective spin on one of the dimers , as shown in fig .
[ fig : secondorderperturbation ] . the matrix element of each such process depends in detail on the specific local spin configuration on the dimers neighboring the dimer that undergoes the spin flip . among the various possibilities , one with the largest amplitude
is shown in fig .
[ fig : secondorderperturbation ] ( a ) . in the second order
effective hamiltonian @xmath127 , this process contributes a term @xmath128 which provides a further contribution to the transverse field operator at site @xmath101 , dressed by diagonal operators from the neighboring dimers , that project out the specific configuration of dimer spin states according to the configuration shown in fig .
[ fig : secondorderperturbation ] . while the transverse field in @xmath129 acts locally on each dimer , the dressed transverse field operators deriving from @xmath127 lead to correlations among nearest neighbor dimer spins .
for example , the spin exchange process on dimer @xmath101 shown in fig . [ fig : secondorderperturbation ] ( a ) could not take place as indicated by the arrows , if dimer @xmath130 was in the opposite spin state . instead
, a different process could take place , as shown in fig .
[ fig : secondorderperturbation ] ( b ) , that leads to a similar term in @xmath127 , but with a different energy denominator than in eq .
( [ eq : secondorderperturbation ] ) . in this way , details of the local dimer configuration enter the effective hamiltonian in a rather complex manner .
the explicit form of the total effective hamiltonian up to second order in @xmath45 involves several terms , containing products of up to five effective spin operators , such as the term given explicitly in eq .
( [ eq : secondorderperturbation ] ) .
while the ground state of this effective hamiltonian is not directly accessible , the general structure of these terms indicate that it will be a dressed version of @xmath68 , with local inter - dimer correlations induced by the above virtual spin exchange processes .
this reflects the qmc result , that the true ground state is close in energy to @xmath68 , and does not exhibit long ranged correlations .
the effective hamiltonian up to second order in perturbation theory does not include transverse spin exchange terms , that would contain products of off - diagonal effective spin operators such as @xmath131 on two neighboring dimers @xmath101 and @xmath130 . however , such terms appear in third order of the degenerate perturbation theory , e.g. from the process shown in fig . [
fig : thirdorderperturbation ] .
similar to the transverse magnetic field operator in second order , the matrix elements depend on the details of the spin configuration on the neighbors of the two dimers that undergo an effective spin exchange .
the transverse effective spin exchange operators are thus similarly dressed by additional projection operators . becoming more relevant at large values of @xmath45 , we expect that such exchange terms eventually drive the transition from the dimer triplet phase to the superfluid region , that was observed in the qmc simulations ( c.f .
[ fig : groundstatephasediagram ] ) .
we now turn to the special point @xmath30 , where the ground state in the ising limit has an enhanced degeneracy , as discussed in sec .
ii . consider one of the @xmath31-dimer of the shastry - sutherland lattice in the ising limit @xmath25 .
for @xmath30 , this dimer can be either in one of the local configurations of fig .
2 ( a ) , which are also allowed configurations for @xmath33 , or it is part of one of the local configurations shown in fig .
2 ( b ) . introducing a finite @xmath21 in first order perturbation theory , a splitting in the energy between the configurations of fig .
2 ( a ) and those of fig .
2 ( b ) results , since the configurations of fig .
2 ( b ) can not gain exchange energy by a transverse spin exchange along the dominant @xmath31 bonds , in contrast to the configurations of fig .
2 ( a ) . to first order in @xmath45 ,
this energy difference equals @xmath132 near the point @xmath133 .
a finite @xmath134 thus leads to a partial lifting of the ground state degeneracy , and only the local configurations of fig .
2 ( a ) remain to span the low - energy sector . the configurations of fig .
2 ( b ) are split - off by a finite energy difference @xmath134 from the ground state manifold .
since this energy difference remains positive for @xmath135 we expect that as long as @xmath136 close to @xmath133 , the system is driven by quantum effects towards the same phase as for @xmath33 .
in fact , this expectation is in agreement with the qmc phase diagram of fig .
3 , where we found that ( i ) for small @xmath21 at @xmath30 , the system enters the dimer triplet state as it does for @xmath33 , and ( ii ) the phase boundary of the dimer triplet phase closely follows the limiting line @xmath137 according to @xmath138 for small @xmath45 ( compare to the dashed - dotted line in fig .
the above argument was based on energy considerations on an isolated dimer . due to this restriction
, we are not able here to discern , if the superfluid phase indeed terminates at a small , but finite values of @xmath45 , or if it persists down to any finite value of @xmath21 .
this would require the intra - dimer exchanges to be taken into account within higher orders of perturbation theory .
however , from our analysis of the case @xmath33 discussed above , we expect that also in this case , the effective model will not allow for an explicit solution , thus leaving this question unanswered .
hence , here we do not attempt to extend on this issue , but instead move on to study the magnetization process in the model under consideration .
after exploring the ground state phase diagram , we now consider the model of eq .
( 1 ) in the presence of a finite magnetic field @xmath139 , which couples to the spins by the standard zeeman term , @xmath140 from considering a singly flipped spin with respect to the fully polarized state , one finds that the system is fully polarized for magnetic fields @xmath139 beyond @xmath141 before discussing the magnetization process of the full quantum model , it is again useful to consider first the ising limit .
the authors of ref .
state that the ising model on the shastry - sutherland lattice exhibits a magnetization process with a single plateau at @xmath142 , where @xmath143 denotes the magnetization , and @xmath144 its saturation value , at least for @xmath145 . in our notation
, this @xmath1 plateau should extend between @xmath146 .
in particular , in case @xmath147 , well within the nel ordered zero - field regime , this range becomes @xmath148 .
this conclusion in ref . was based on analyzing a finite system with 16 spins only . in order to check this conclusion on larger system sizes
, we performed a systematic finite size analysis , using classical monte carlo ( mc ) simulations on lattices with up to @xmath149 spins in the canonical ensemble . for these simulations
, we employed a single spin - flip metropolis algorithm , and allowed for a simulated annealing of the system from a large initial temperature @xmath150 down to the final temperature during an initial state of equilibration .
this way we were able to obtain mc results down to @xmath151 for @xmath152 , and @xmath153 for @xmath154 . while for the purpose of the current study , one can draw relevant conclusions on the magnetization process from these data , it will be interesting to obtain more refined numerical data on the magnetization process in the ising limit , using extended ensemble sampling methods , similar to the approach taken in ref . for the square and triangular lattice
. however , this lies beyond the scope of the current study , which is directed towards the quantum regime . upon presenting the mc results
, we first discuss the case @xmath147 considered in ref . .
our mc data for the magnetization process on different lattices are collected in fig .
[ fig : magisinga ] . the inset of fig .
[ fig : magisinga ] shows our data for a @xmath155 system , which appear to confirm the conclusion of ref . .
however , upon increasing the system size , we find that different plateau structures appear .
in general , in order to allow a system to establish a certain magnetization plateau , appropriate lattice sizes and boundary conditions must to be chosen . otherwise , geometric constraints could frustrate certain magnetization patterns . in the present case
, it can be seen from the numerical data , that only for @xmath43 a multiple of 3 , the system establishes a wide @xmath156 plateau , which is consistently observed for @xmath157 and @xmath158 . instead
, the @xmath159 and @xmath160 systems can not establish the corresponding magnetic superstructure , and hence lead to rather different magnetization curves with strong finite size effects , that do not represent thermodynamic limit behavior .
the magnetization curve in fig .
[ fig : magisinga ] for @xmath161 , taken at @xmath151 , exhibits a magnetization plateau at @xmath156 , extending from @xmath162 up to the saturation field at @xmath163 .
the data shown for the two larger systems , for which higher temperatures had to be taken in the mc simulations , are consistent with a thermal smothering of the magnetization jumps out of the plateau towards @xmath164 and full saturation , respectively .
we conclude that for @xmath147 the ising model exhibits a single intermediate magnetization plateau at @xmath156 , extending from @xmath162 up to the saturation field at @xmath163 .
the different conclusion of ref . appears to be due to the usage of inappropriate finite lattice sizes .
note , that all values of @xmath43 considered here were even , i.e. a plateau at @xmath142 , if it would exist in this model , would not be frustrated by finite lattice effects . in fact , as discussed below , we find that such a @xmath1 plateau appears for finite values of @xmath45 due to quantum effects .
next , we consider the case @xmath165 , well inside the degenerate region of @xmath33 .
below , we will compare the ising model result to qmc data on the magnetization process for finite @xmath21 at the same value of @xmath58 .
the mc results for the magnetization process in the ising limit are shown in fig .
[ fig : magisingb ] , where we now consider linear system sizes @xmath43 that are a multiple of @xmath166 .
again , we observe a wide @xmath0 magnetization plateau , which extends from @xmath167 up to the saturation field at @xmath168 .
a precise estimate of the lower boundary of the plateau is not accessible from the current finite size data , as we could not collect data at sufficiently low temperatures on larger systems .
there appears however a finite magnetization with a smooth increase well before the plateau is entered .
it will be interesting to explore this low@xmath169 regime in more detail using extended ensemble methods .
the point that is important for the following discussion , and which follows also from the current mc data , is the absence again of a magnetization plateau at @xmath142 .
instead , the magnetization exhibits a jump from the @xmath0 plateau up to magnetic saturation at @xmath170 . as we show next , a plateau at @xmath142 , while not obtained in the ising limit , emerges in the quantum model at finite values of @xmath21 .
[ fig : magquantum ] shows the magnetization process for @xmath165 and @xmath171 from qmc simulations .
we again find a magnetization plateau at @xmath156 as well as a plateau at @xmath142 . adding a transverse exchange to the ising model thus leads to a softening of the large magnetization jump from @xmath156 to @xmath172 in the ising limit , and an additional plateau region appears with @xmath142 .
in hard - core bosonic language , the transverse exchange maps onto a non - frustrating hopping amplitude . in the current situation , this finite boson hopping does not lead always to superfluidity , but drives the system into an insulating phase at filling @xmath173 , corresponding to @xmath142 ( due to particle - hole symmetry , a similar insulating region emerges also for @xmath174 , corresponding to @xmath175 ) .
indeed , we find from fig .
[ fig : magquantum ] , that the spin stiffness @xmath176 vanishes inside both plateau regions .
it is then in order to study , if the magnetic excitations in both plateau phases form period crystals , and what the structures of such solid arrays would be .
long - range crystalline ordering of magnetic excitations inside magnetic plateau regions has previously been analyzed for the isotropic spin-1/2 heisenberg model on the shastry - sutherland model @xcite , based on different approximative schemes .
the relevant structures expected from these studies are shown in the inset of fig .
[ fig : magsuperstructures ] for the @xmath0 ( upper panel ) , and the @xmath1 ( lower panel ) plateau , respectively .
they are expressed in terms of periodic arrangements of @xmath177 dimer triplet states @xmath106 , denoted by dumbbells , in a background of less polarized , e.g. @xmath65 states on the remaining dimers .
we now assess , if these crystalline patterns appear also in the current quantum model , which represents an anisotropic version of the heisenberg model considered thus far . in a qmc simulation
, one can probe exactly for these magnetic superstructures by measuring an appropriate structure factor .
hence , we analyze the ordering pattern of the magnetic excitations by measuring the triplet excitation structure factor @xmath178 where @xmath179 is a projector on the @xmath177 dimer triplet state @xmath106 on dimer @xmath101 .
@xmath180 equals the number of dimers in the finite system of @xmath35 sites .
the positions @xmath181 of the dimers and the momentum space vector @xmath182 are defined with respect to the convenient coordinate system formed by the square lattice of dimers , with the distance between the centers of two neighboring dimers taken as unity . in this way ,
the solid order shown in the upper panel of fig .
[ fig : magsuperstructures ] for @xmath156 corresponds to a peak in @xmath183 at @xmath184 , and at @xmath185 for the order shown in the lower panel of fig .
[ fig : magsuperstructures ] for @xmath142 .
the main panels of fig . [ fig : magsuperstructures ] show these components of the structure factor for different system sizes , respectively .
we checked , that no other signal in @xmath186 appeared , apart from those explicitly shown . from these data
we conclude , that indeed the crystalline orders of the magnetic excitations shown in fig .
[ fig : magsuperstructures ] are stabilized within the two magnetization plateau regions , respectively . in order to study the real - space distribution of the magnetization among the lattice sites in more detail
, we also measured the mean local values of the magnetization .
the obtained distributions of the local magnetization are shown for the 1/3 and 1/2 plateau in fig .
[ fig : realspace ] ( a ) and ( b ) , respectively .
they agree well with the distributions reported for the isotropic heisenberg model in the studies mentioned above . upon varying @xmath139 beyond these wide magnetization plateaus ,
the magnetization undergoes discontinuous jumps , as seen in fig .
[ fig : magquantum ] , alert when entering the @xmath1 plateau from below , where such a jump appears not that well resolved .
the spin stiffness @xmath176 exhibits a similar behavior , with clear jumps upon entering the plateaus , expect when entering the @xmath1 plateau from below .
apart from this case , the superfluid - insulator transitions are thus clearly first order . concerning the region below the @xmath1 plateau
, one might instead consider the presence of a supersolid state of magnetic excitations , with both a finite superfluid density and a crystalline superstructure . in order to assess ,
if supersolid behavior is indeed present in this regime , we show in fig . [
fig : supersolid ] the finite size scaling of both the superfluid density @xmath176 and the relevant structure factors from the neighboring magnetization plateaus at a magnetic field of @xmath187 , inside the possibly supersolid region . from the finite size
scaling we can exclude the presence of a supersolid phase in this parameter regime . between the two magnetization plateaus , the system is thus in a uniform superfluid state .
our analysis of the magnetization process at @xmath188 and @xmath171 did not exhibit the presence of any additional plateaus at lower values of @xmath143 .
indeed , at low-@xmath143 the magnetization curve as well as the superfluid density appear smooth in fig .
[ fig : magquantum ] . in particular , we did not find any of the fractional magnetization plateaus mentioned in sec .
i for the compounds srcub@xmath2(bo@xmath4)@xmath9 and tmb@xmath18 , nor those found for the isotropic spin-1/2 heisenberg model on the shastry - sutherland model in recent studies @xcite .
we postpone a discuss on this issue to the concluding section . finally , we present our results for the magnetization process upon varying over a wider range of parameters space
. the results of qmc simulations similar to those discussed in detail above are summarized in fig .
[ fig : emergence ] , which shows the magnetic phase diagram for a generic fixed ratio of @xmath188 .
the figure shows both the @xmath156 magnetization lobe that extends from the ising limit up to @xmath189 , as well as the emergent @xmath142 lobe , that extends up to a similar value of @xmath45 , but shrinks upon approaching the ising limit , where no @xmath1 plateau persists .
we studied the ground state phase diagram and the magnetization process of the spin-1/2 easy - axis xxz model of eq .
( 1 ) with ferromagnetic transverse spin exchange on the shastry - sutherland lattice .
the model exhibits a nel ordered phase for small values of @xmath58 and @xmath45 , and a superfluid phase for dominant @xmath45 . for sufficiently large values of @xmath58 , a dimerized phase with the dominant formation of @xmath190 triplet states on the @xmath31-dimer bonds is stabilized , that connects to the degenerate phase of the model in the ising limit .
we like to compare this findings to the case of the isotropic heisenberg model on the same lattice .
also in this model there is strong evidence for an intermediate phase @xcite , separating the nel ordered phase from the dimer singlet phase .
in contrast , this intermediate phase is however not a superfluid , but appears to be realize a valence bond crystal driven by the frustrated nature of the transverse exchange interactions in this region @xcite .
using classical monte carlo simulations to assess the magnetization process in the ising limit , we found in contrast to previous claims @xcite no magnetization plateau at @xmath1 of full saturation , but instead a @xmath0 plateau , with a jump towards full saturation .
the quantum model shows the presence of a @xmath1 plateau in addition to the @xmath0 plateau .
this emergence of a magnetization plateau upon adding quantum fluctuations ( via a finite @xmath45 ) to the ising model remains of the previously observed emergence of a magnetization plateau via thermal fluctuations in the classical heisenberg model on the kagome lattice @xcite .
while we examined the case of a ferromagnetic transverse spin exchange , the experimental observation of a @xmath1 plateau in the compounds tmb@xmath18 and srcub@xmath2(bo@xmath4)@xmath9 can be taken as indication , that such a plateau also emerges for antiferromagnetic transverse exchange .
indeed , a @xmath1 plateau is consistently reported also in the studies on the magnetization process of the isotropic heisenberg model on the shastry - sutherland lattice @xcite .
with respect to other reported magnetization plateaus both from experiments and in recent theoretical work , we did not obtain evidence from our quantum monte carlo simulations , that they are stabilized in the current model .
in particular , the pronounced @xmath11 plateau , consistently reported in recent theoretical work @xcite , has not been found for the parameter region we examined in our simulations ( we performed various scans inside the range @xmath191 , down to @xmath192 , with results similar to those presented in detail above ) , even though the proposed magnetic unit cell @xcite is commensurate with the finite lattices that we employed in our simulations .
we explicitly checked that no signal in the relevant component of @xmath186 at @xmath193 appears near @xmath194 in this model .
hence , we conclude that the model considered here exhibits magnetization plateaus at @xmath1 and @xmath0 of the full saturation , only .
future work could explore in more detail the magnetization process in the ising model inside the low - field region using extended ensemble methods , similar to the approach taken in ref .
. it would be interesting to apply the theoretical approaches employed in refs . and to the current model , in order to to check their predictions against the unbiased large - scale quantum monte carlo results present here .
we thank a. luchli , f. mila , r. mssner , a. muramatsu and d. poilblanc for useful discussions . financial support by the deutsche forschungsgemeinschaft under grant no .
we 3649/2 - 1 and through sfb / tr 21 is gratefully acknowledged , as well as the allocation of cpu time by hlrs stuttgart and nic jlich .
the employed numerical simulation code was partially based on the alps libraries @xcite .
l. c. pauling , _ the nature of the chemical bond and the structure of molecules and crystals : an introduction to modern structural chemistry _ ( cornell university press , 20th edition , ithaca , new york , usa , 1986 ) . | we examine ground state properties of the spin-1/2 easy - axis heisenberg model on the shastry - sutherland lattice with ferromagnetic transverse spin exchange using quantum monte carlo and degenerate perturbation theory . for vanishing transverse exchange
, the model reduces to an antiferromagnetic ising model that besides nel order harbors regions of extensive ground state degeneracy . in the quantum regime
, we find a dimerized phase of triplet states , separated from the nel ordered phase by a superfluid .
the quantum phase transitions between these phases are characterized .
the magnetization process shows a magnetization plateau at @xmath0 of the saturation value , that persists down to the ising limit , and a further plateau at @xmath1 only in the quantum regime . for both plateaus
, we determine the crystalline patterns of the localized triplet excitations .
no further plateaus nor supersolid phases are found in this model . |
geometrical scaling is a phenomenon predicted by the gluon saturation in hadronic interactions @xcite .
it was first observed when the total @xmath10 cross sections in deep inelastic scattering ( dis ) at small @xmath11 were expressed in terms of the variable @xmath12 , where @xmath13 is the photon virtuality , @xmath11 is the bjorken variable and @xmath14 is the saturation radius @xcite .
similar geometrical scaling was found in dis on nucleus @xcite and diffractive processes @xcite . in recent years
, geometrical scaling was presented in the @xmath1 spectra of inclusive and identified charged hadrons in proton - proton ( @xmath15 ) collisions at @xmath16 = 0.9 , 2.76 and 7 tev @xcite .
as nuclei are made of nucleons , it is nature to extend geometrical scaling to the inclusive charged hadron production in nucleus - nucleus ( @xmath17 ) collisions at rhic and lhc energies @xcite . in @xcite , the authors showed when the multiplicity distributions in the central au - au and cu - cu collisions at @xmath0 = 62.4 and 200 gev were plotted as a function of @xmath18 , where @xmath19 is number of nucleons in the nuclei , @xmath20 gev , @xmath21 , @xmath5 = 0.30 , a tendency towards geometrical scaling was clearly seen , although the alignment of au and cu spectra was not perfect for small and medium values of @xmath22 .
reference @xcite illustrated that when the @xmath1 spectra in pb - pb 0 - 5@xmath2 , 5 - 10@xmath2 and 40 - 50@xmath2 collisions at @xmath0 = 2.76 tev were expressed in the variable @xmath23 , where @xmath24 is the saturation scale and @xmath25 , the data points in the 0 - 5@xmath2 and 5 - 10@xmath2 centralities fell into the same line for @xmath26 , while the data points in the 40 - 50@xmath2 centrality presented some departure . in this paper
, we will analyze the data points at 0 - 5@xmath2 , 5 - 10@xmath2 , 10 - 20@xmath2 , 20 - 30@xmath2 , 30 - 40@xmath2 , 40 - 50@xmath2 , 50 - 60@xmath2 , 60 - 70@xmath2 and 70 - 80@xmath2 centralities in pb - pb collisions at 2.76 tev .
we find that the inclusive charged hadron @xmath1 spectra at these centralities exhibit geometrical scaling when they are presented in terms of a new scaling variable @xmath27 . here
@xmath27 is defined as @xmath3 , @xmath5 is the scaling parameter which is determined by the method of ratios @xcite .
we then use a single tsallis distribution to parameterize the @xmath1 spectra at different centralities , and discuss the consequences of the geometrical scaling .
this paper is organized as follows . in section [ sec : scaling_variable ] , we will illustrate how the new scaling variable @xmath27 is proposed in terms of the gluon saturation . in section [ sec :
gs_inclusive ] , we will show how the scaling parameter @xmath5 in @xmath27 is determined and then present the geometrical scaling in the inclusive charged hadron @xmath1 spectra . in section [ sec : gs_consequence ]
, we will use the single tsallis distribution to parameterize the spectra at different centralities and discuss the consequences of the geometrical scaling .
finally , the conclusion is made in section [ sec : conclusion ] .
geometrical scaling is a property of particle densities at high energies .
it is based on the underlying mechanism which is called the gluon saturation @xcite .
the geometrical scaling means that the particle spectra are in fact a dimensionless function of the scaling variable @xmath28 , rather than independently of @xmath11 and @xmath1 . here
@xmath1 is the transverse momentum of an observed particle in high energy collisions , @xmath29 is the saturation momentum , @xmath30 , @xmath31 .
for example , in @xmath15 collisions at @xmath16 = 0.9 , 2.76 and 7 tev @xcite , the inclusive charged hadron @xmath1 spectra measured at different energies in the central region depend on a single variable , @xmath32 . here
@xmath33 is the proton saturation momentum scale and is given by @xmath34 where @xmath35 , @xmath36 = 1 gev and @xmath37 . in @xmath17 collisions , in order to consider the geometrical scaling in the @xmath1 spectra with different centralities , the saturation momentum is modified as @xcite @xmath38 as a result , the corresponding scaling variable for pb - pb collisions is @xmath39 when the @xmath1 spectra of charged hadrons at different centralities in pb - pb collisions at 2.76 tev are plotted in terms of @xmath40 , they will migrate into a universal curve @xmath41 , @xmath42 where @xmath43 . here ,
the transverse areas of the two colliding nuclei for different centralities are deemed to be proportional to @xmath4 , rather than to @xmath44 @xcite .
this is due to the fact that when considering the effect of running coupling @xmath45 the transverse areas prefer a linear dependence on @xmath4 @xcite . for simplicity ,
@xmath46 will be referred as the @xmath1 spectrum at @xmath4 in the following .
if the @xmath1 spectra at @xmath4 and @xmath47 are equal , @xmath48 , then they are the distributions of the same @xmath40 .
thus , @xmath49 which implies @xmath50 this formula allows us to rescale the @xmath1 spectrum at @xmath4 to @xmath47 . as a convention
, we choose the @xmath1 spectrum at the first centrality class ( 0 - 5@xmath2 , @xmath51 ) as a reference and try to rescale the @xmath1 spectra at the other 8 centrality classes ( 5 - 10@xmath2 , 10 - 20@xmath2 , 20 - 30@xmath2 , 30 - 40@xmath2 , 40 - 50@xmath2 , 50 - 60@xmath2 , 60 - 70@xmath2 and 70 - 80@xmath2 ) to this reference .
therefore , @xmath3 .
this @xmath27 is exactly the scaling variable we propose in this paper .
apparently , for the spectrum at @xmath4 , the corresponding @xmath27 only depends on the scaling parameter @xmath5 , while @xmath40 in eq .
( [ eq : nuclei_tau ] ) relies on @xmath36 , @xmath52 and @xmath5 .
thus @xmath27 is a more appropriate variable in searching for the geometrical scaling of the @xmath1 spectra at different centralities in pb - pb collisions .
the @xmath1 spectra of inclusive charged hadron at different centralities in pb - pb collisions at 2.76 tev were published by the alice collaboration @xcite .
these spectra will exhibit geometrical scaling when they are presented in terms of @xmath27 with a suitable scaling parameter @xmath5 . in order to estimate this parameter
, we adopt the method of ratios @xcite . in this method ,
a ration is defined in terms of @xmath27 , @xmath53 where @xmath54 is the @xmath1 spectrum rescaled from @xmath4 to @xmath47 , @xmath55 is the @xmath1 spectrum at @xmath47 .
since there are no data points at @xmath3 on the @xmath1 spectrum of @xmath47 , in order to calculate the ratio @xmath56 , we need to interpolate the values at these points . as the spectrum at @xmath47 lies on a curve as a function of @xmath1 ,
we do a shape - preserving piecewise cubic interpolation . in section [ sec : scaling_variable ]
, we have chosen @xmath47 = 191.5 for the 0 - 5@xmath2 centrality and @xmath4 = 165 , 130.5 , 93 , 64.5 , 42.5 , 26.5 , 15 and 7.9 for the 5 - 10@xmath2 , 10 - 20@xmath2 , 20 - 30@xmath2 , 30 - 40@xmath2 , 40 - 50@xmath2 , 50 - 60@xmath2 , 60 - 70@xmath2 and 70 - 80@xmath2 centralities respectively .
thus there are 8 such ratios .
if the geometrical scaling is true , then @xmath57 should be equal to 1 .
however , in practice , @xmath57 is around 1 in a certain @xmath27 range . in order to determine the best value of @xmath5
, we define the mean square deviation between @xmath57 and 1 as follows @xcite , @xmath58 where @xmath59 is the number of data points in the region @xmath60 , @xmath61 is the experimental uncertainty of the ratio @xmath57 . here
we have ignored the interpolation error and the theoretical uncertainty of geometrical scaling hypothesis @xcite , as they are small ( less than 2@xmath2 and around 3@xmath2 respectively ) when compared to the experimental errors ( 8 - 20@xmath2 ) in the region where the geometrical scaling holds .
@xmath62 ( @xmath63 ) is the @xmath27 ( @xmath1 ) of the first data point where the geometrical scaling starts to appear .
it is determined by @xmath64 .
when @xmath65 , the absolute difference between @xmath57 and 1 is either within @xmath61 or above @xmath61 .
@xmath66 ( @xmath67 ) is the @xmath27 ( @xmath1 ) of the last data point where the geometrical scaling holds up .
it is determined by @xmath68 . for data points with @xmath69 ,
the geometrical scaling is violated .
thus the number of data points which exhibit geometrical scaling is exactly @xmath59 introduced in eq .
( [ eq : chi_square ] ) .
obviously , @xmath59 depends on @xmath4 .
[ fig : n_lambda_vs_lambda ] shows @xmath59 as a function of @xmath5 for different centralities .
the best value of @xmath5 is determined to be the one which maximizes the sum of @xmath59 at 5 - 10@xmath2 , 10 - 20@xmath2 , 20 - 30@xmath2 , 30 - 40@xmath2 , 40 - 50@xmath2 , 50 - 60@xmath2 , 60 - 70@xmath2 and 70 - 80@xmath2 centralities .
there are five such values , @xmath5 = 1.58 , 1.59 , 1.60 , 1.61 and 1.68 .
however , only @xmath70 is chosen to be the best scaling parameter , as it gives the smallest @xmath71 , which means the deviation of the ratios in eq .
( [ eq : ratio ] ) from unity is the minimum .
this @xmath5 value is larger than the scaling parameter ( @xmath37 ) utilized in the geometrical scaling of the inclusive charged hadron @xmath1 spectra in @xmath15 collisions at 0.9 , 2.76 and 7 tev @xcite .
however , it is interesting to see that the effective scaling parameter @xmath72 ( 0.091 ) in pb - pb collisions is very close to @xmath73 ( 0.108 ) in @xmath15 collisions . the number of data points which exhibit geometrical scaling as a function of @xmath5 in the inclusive charged hadron @xmath1 spectra at pb - pb 5 - 10@xmath2 , 10 - 20@xmath2 , 20 - 30@xmath2 , 30 - 40@xmath2 , 40 - 50@xmath2 , 50 - 60@xmath2 , 60 - 70@xmath2 and 70 - 80@xmath2 collisions . ]
now since @xmath70 , we can figure out @xmath74 , @xmath59 , @xmath75 , @xmath63 , @xmath67 , @xmath62 and @xmath66 for the 5 - 10@xmath2 , 10 - 20@xmath2 , 20 - 30@xmath2 , 30 - 40@xmath2 , 40 - 50@xmath2 , 50 - 60@xmath2 , 60 - 70@xmath2 and 70 - 80@xmath2 centralities .
they are tabulated in table [ tab : n_lambda ] . here
, for each centrality , @xmath74 is the total number of data points , @xmath75 is the position number of the first data point which exhibits geometrical scaling . with @xmath75 and @xmath59 , we can determine @xmath63 and @xmath67 , and the corresponding @xmath62 and @xmath66 . from this table ,
we observe that the number of data points which present geometrical scaling increase with centrality . for the 5 - 10@xmath2 centrality class ,
all the data points show geometrical scaling , while for the 70 - 80@xmath2 centrality class , only 18 out of 61 data points exhibit geometrical scaling . as a consequence ,
the corresponding interval of @xmath27 for geometrical scaling also increase with centrality .
.[tab : n_lambda]the quantities @xmath74 , @xmath59 , @xmath75 , @xmath63 , @xmath67 , @xmath62 and @xmath66 for the 5 - 10@xmath2 , 10 - 20@xmath2 , 20 - 30@xmath2 , 30 - 40@xmath2 , 40 - 50@xmath2 , 50 - 60@xmath2 , 60 - 70@xmath2 and 70 - 80@xmath2 centralities . the units for @xmath63 , @xmath67 , @xmath62 and @xmath66 are gev / c .
[ cols="^,^,^,^,^,^,^,^ " , ]
in this paper , we have shown that the @xmath1 spectra of inclusive charged hadrons at 0 - 5@xmath2 , 5 - 10@xmath2 , 10 - 20@xmath2 , 20 - 30@xmath2 , 30 - 40@xmath2 , 40 - 50@xmath2 , 50 - 60@xmath2 , 60 - 70@xmath2 and 70 - 80@xmath2 centralities exhibit geometrical scaling when they are presented in terms of the scaling variable @xmath3 .
this variable is addressed according the gluon saturation mechanism .
the scaling parameter @xmath5 is determined to be 1.68 by the method of ratios .
the spectra at different centralities with @xmath76 gev / c can be parameterized by the single tsallis distribution in eq .
( [ eq : tsallis_distribution ] ) , and the temperature @xmath6 is proportional to @xmath7 , which is predicted by the geometrical scaling .
the geometrical scaling also predicts that @xmath8 grows as a power of @xmath4 , @xmath9 , which is confirmed by the data collected by the alice collaboration .
the author would like to thank prof . c. b. yang at central china normal university for valuable discussions .
this work was supported by the fundamental research funds for the central universities of china under grant no .
gk201502006 , by the scientific research foundation for the returned overseas chinese scholars , state education ministry , and by the national natural science foundation of china under grant nos .
11447024 and 11505108 .
00 l. d. mclerran and r. venugopalan , phys .
d * 49 * , 2233 ( 1994 ) .
l. d. mclerran and r. venugopalan , phys .
d * 49 * , 3352 ( 1994 ) .
l. d. mclerran and r. venugopalan , phys .
d * 50 * , 2225 ( 1994 ) .
a. m. stasto , k. golec - biernat , and j. kwiecinski , phys .
* 86 * , 596 ( 2001 ) .
a. freund , k. rummukainen , h. weigert and a. schafer , phys .
lett . , * 90 * , 222002 ( 2003 ) .
n. armesto , c. a. salgado and u. a. wiedemann , phys .
lett . , * 94 * , 022002 ( 2005 ) . c. marquet and l. schoeffel , phys .
b , * 639 * , 471 ( 2006 ) .
m. praszalowicz , phys .
* 106 * , 142002 ( 2011 ) .
m. praszalowicz and anna francuz , phys . rev .
d , * 92 * , 074036 ( 2015 ) .
m. praszalowicz , phys .
b , * 727 * , 461 ( 2013 ) .
m. praszalowicz , acta phys . pol .
b , * 42 * , 1557 ( 2011 ) . c. andres , a. moscoso and c. pajares , nucl .
a , * 901 * , 14 ( 2013 ) . c. klein - boesing and l. mclerran , phys .
b , * 734 * , 285 ( 2014 ) .
j. schaffner - bielich , d. kharzeev , l. mclerran and r. venugopalan , nucl .
phys . a , * 705 * , 494 - 507 ( 2002 ) .
b. abelev , _ et al . _
( alice collaboration ) , phys .
b , * 720 * , 52 ( 2013 ) .
m. praszalowicz and t. stebel , jhep , * 03 * , 090 ( 2013 ) .
w. c. zhang and c. b. yang , j. phys .
g : nucl . part .
* 41 * , 105006 ( 2014 ) .
w. c. zhang , j. phys .
g : nucl . part .
phys . , * 43 * , 015003 ( 2016 ) .
m. rybczynski , g. wilk and z. wlodarczyk , epj web of conference , * 90 * , 01002 ( 2015 ) . k. aamodt _
( alice collaboration ) , phys .
lett . , * 106 * , 032301 ( 2011 ) . | in pb - pb collisions at @xmath0 = 2.76 tev , we report on a geometrical scaling in the transverse momentum ( @xmath1 ) spectra of inclusive charged hadrons at 0 - 5@xmath2 , 5 - 10@xmath2 , 10 - 20@xmath2 , 20 - 30@xmath2 , 30 - 40@xmath2 , 40 - 50@xmath2 , 50 - 60@xmath2 , 60 - 70@xmath2 and 70 - 80@xmath2 centralities .
this geometrical scaling is exhibited when these spectra are expressed in terms of a new scaling variable , @xmath3 , where @xmath4 is the number of participating nucleon pairs , @xmath5 is the scaling parameter . with
the method of ratios , this scaling parameter is determined to be 1.68 .
we then use a single tsallis distribution to parameterize the spectra at different centralities , and find that the tsallis temperature @xmath6 is proportional to @xmath7 , which is exactly the prediction of the geometrical scaling .
the geometrical scaling also predicts that the charged - particle density per participating nucleon pair ( @xmath8 ) grows as a power of @xmath4 , @xmath9 , which is confirmed by the data collected by the alice collaboration . |
the rhic beam energy scan ( bes ) is mainly focused on a search for the qcd critical point and the 1st - order phase boundary in the qcd phase diagram . in the year 2010
, the star experiment at rhic took data for the beam energies @xmath37.7 , 11.5 and 39 gev as a first phase of the bes program .
analysis of the collected data for various observables sensitive to the phase transition and critical point continues for the successful accomplishment of the program .
directed flow , @xmath0 , is the first harmonic coefficient in the azimuthal distribution of the particles with respect to the reaction plane , @xmath4 where @xmath5 denotes the azimuthal angle of an outgoing particle and @xmath6 is the orientation of the reaction plane defined by the beam axis and the impact parameter vector@xcite . both hydrodynamic and nuclear transport models @xcite indicate that directed flow is a sensitive signature for the phenomena related to the possible phase transition specially in the bes region @xcite . in particular , the shape of @xmath7 may exhibit flatness at midrapidity due to a strong , tilted expansion of the source giving rise to anti - flow or a 3rd flow @xcite component .
if the tilted expansion is strong enough , it can cancel and reverse the motion in the bounce - off direction and results in a negative @xmath8 slope at midrapidity , potentially producing a wiggle like structure in @xmath7 .
a wiggle for baryons is a possible signature of a phase transition between hadronic matter and quark gluon plasma ( qgp ) , although qgp is not the only possible explanation @xcite .
if strong but incomplete baryon stopping is assumed together with positive space - momentum correlations caused by transverse radial expansion , then a wiggle structure might be explained even in a hadronic system @xcite .
as a function of @xmath9 scaled by @xmath10 for au + au collisions at @xmath11 = 7.7 , 11.5 , 39 , 62.4 and 200 gev for 3060% central collisions .
the right panel shows charged hadron @xmath12 as a function of @xmath9 - @xmath10 .
the results for 62.4 and 200 gev are for 3060% centrality , previously reported by star @xcite . ] as a function of rapidity for positive and negative hadrons from au+au collisions at @xmath13 7.7 , 11.5 and 39 gev respectively with proton , antiproton , @xmath14 and @xmath15 shown in the first , second and last column of panels for mid - central ( 10 - 40% ) collisions . ] in these proceedings , we report @xmath12 measurements by the star experiment from @xmath11 = 7.7 , 11.5 and 39 gev au + au collisions .
data were taken from run 10 ( 2010 ) .
the star time projection chamber ( tpc ) @xcite was used as the main detector for charged particle tracking at midrapidity and forward time projection chambers ( ftpcs ) were used for charged particle tracking at forward rapidities .
the centrality was determined by the number of tracks from the pseudorapidity region @xmath16 .
two beam beam counters covering @xmath17 were used to reconstruct the first order event plane for this analysis .
the pseudorapidity gap between bbc and tpc suppresses the the non - flow effects .
we analyzed minimum bias events with the following conditions : event vertex radius ( @xmath18 cm , where @xmath19 and @xmath20 are the vertex positions along the @xmath21 and @xmath22 directions , respectively , and @xmath23 within 70 , 50 and 30 cm of the center of the detector at 7.7 , 11.5 and 39 gev respectively .
tracks are also required to have transverse momenta @xmath24 gev@xmath25 , pass within 3 cm of the primary vertex , have at least 15 space points in the main tpc acceptance @xmath26 or 5 space points in the case of tracks in the ftpc acceptance ( @xmath27 ) , and we require the ratio of the number of actual space points to the maximum possible number of space points to be greater than 0.52 . protons and antiprotons up to 2.8 gev/@xmath28 and @xmath1 and @xmath2 up to 1.6 gev/@xmath28 in transverse momentum were identified based on specific energy loss in the tpc and the time - of - flight information from multi - gap resistive plate chamber tof detectors in combination with the momentum .
beam energy and centrality dependence of directed flow of charged particles as a function of pseudorapidity and transverse momentum has been previously reported @xcite .
fig [ fig1 ] ( left panel ) shows charged hadron @xmath0 as a function of @xmath9 scaled by the corresponding beam rapidity(@xmath10 ) and right panel as a function of @xmath9 - @xmath10 for au+au collisions at @xmath11=7.7 , 11.5 , 39 , 62.4 and 200 gev for 3060% central collisions .
the new results reported here are the charged hadron @xmath29 in au+au collisions for 3060% centrality at @xmath3 7.7 , 11.5 and 39 gev .
we observe an approximate beam energy scaling behavior of directed flow @xcite , also observed at sps energies @xcite .
[ fig2 ] shows directed flow @xmath30 of identified particles proton , antiproton , @xmath14 and @xmath15 as a function of rapidity for from au+au collisions at @xmath31 = 7.7 , 11.5 and 39 gev .
we observe the differences in @xmath30 for protons and antiprotons . in fig .
[ fig3 ] , proton @xmath0 as a function of rapidity ( @xmath22 ) for central ( 0 - 10% ) , midcentral ( 10 - 40% ) and peripheral ( 40 - 80% ) collisions at @xmath11 = 7.7 , 11.5 and 39 gev are compared to the urqmd model prediction . a sign change in proton slope going from central to peripheral collisions
is observed .
urqmd is qualitatively consistent with the data but does not predict the right magnitude at these energies .
ampt(default ) model calculation is similar to urqmd results . at mid - rapidity as a function of beam energy .
antiproton data are not available from na49 and e895 . ] in fig .
[ fig4 ] , proton and antiproton slope @xmath32 around midrapidity is plotted as a function of collision energy , where @xmath33 is the scaled rapidity , defined as @xmath34 .
values for the slope of @xmath35 are extracted via a polynomial fit of the form @xmath36 .
the proton slope decreases rapidly with increasing energy , changes its sign to negative between 7.7 and 11.5 gev , and remains small and negative up to 200 gev .
in contrast , the antiproton slope , which was not reported by the na49 or e895 collaborations , remains always negative in the measured @xmath11 range from 7.7 to 200 gev .
in this proceedings , we present star results for directed flow as a function of transverse momentum , pseudorapidity and centrality for au+au collisions at @xmath11 = 7.7 , 11.5 and 39 gev .
our findings demonstrate that @xmath37 shows a beam energy scaling behavior , though not perfect , that has already been established at higher rhic energies .
difference in directed flow of protons and antiprotons is observed . for midcentral collisions ( 10 - 40% ) ,
the @xmath1 , @xmath2 , and antiprotons have a negative @xmath38 at mid rapidity , but proton @xmath39 at 7.7 gev becomes positive .
the proton @xmath40 slope decreases rapidly with increasing energy up to 7.7 gev .
its sign changes to negative between 7.7 and 11.5 gev , and remains small and negative at 11.5 , 17.3 , 39 and 200 gev .
however antiproton slope is always negative in the measured range from 7.7 gev to 200 gev .
tested models do not predict the right magnitude of directed flow but urqmd and ampt ( default ) qualitatively follow some of the features shown by the data .
the beam energy scan at rhic will continue in the future to map in more detail the interesting energy range between 11.5 and 39 gev .
hence results from 19.6 gev and 27 gev , which were collected in 2011 , will provide further important information to the observed beam energy dependence of the directed flow . | we present the star measurements of directed flow , @xmath0 , for @xmath1 , @xmath2 , protons and antiprotons , as well as for all detected charged particles in au + au collisions at @xmath3 7.7 , 11.5 and 39 gev as a function of transverse momentum , rapidity and centrality .
results are compared to the predictions from transport models . |
whether diagonal long range orders [ such as charge - density - wave ( cdw ) and spin - density - wave ( sdw ) ] and off - diagonal long range orders [ such as superconducting and superfluid ( sf ) states ] can coexist homogeneously in correlated electronic systems is a central issue in condensed matter physics .
coexistence of superconductivity and cdw has been studied in three - dimensional systems @xcite ( such as @xmath1 doped with @xmath2 or @xmath3 ) , quasi - two - dimensional systems @xcite ( such as the dichalcogenide @xmath4 and @xmath5 ) and quasi - one - dimensional systems @xcite ( such as the trichalcogenide @xmath6 and doped spin ladder @xmath7 ) .
supersolidity , defined as homogeneous coexistence of superfluidity and crystalline order , was theoretically proposed more than 4 decades ago @xcite and has been a subject of debate since then .
the first experimental claim @xcite of observing supersolidity in helium-4 further intensified the debate and enhanced the interest in understanding the phenomena .
the general consensus @xcite is that a supersolid ( ss ) is not realizable in a perfect hcp crystal of @xmath8 ; nevertheless , superflow can occur along vacancies that can collect near extended structural defects @xcite .
thus , the occurrence of supersolidity in bulk solid helium-4 is being seriously questioned .
cold - atom systems offer another opportunity for realization of supersolidity .
theoretically @xcite , lattice bosons with various types of interactions in diverse geometries have yielded supersolidity .
however , there has been no experimental creation of optical lattices with effective long - range interactions that produce supersolidity .
furthermore , experimental techniques to detect signatures of supersolidity also need to be developed for optical lattices @xcite .
there have been numerous studies of supersolidity involving hard - core - bosons ( hcbs ) @xcite . a lattice model for quantum liquids , such as
the interacting bosonic helium-4 at low temperatures , needs a hard - core constraint to account for the exclusion of occupation of more than one atom at each lattice point @xcite .
the behavior of the ground state and low temperature excitations of such systems are largely controlled by couplings with phonons .
furthermore , local cooper pairs [ comprising of two electrons ( of opposite spin ) at a site ] can also be regarded as hcbs @xcite . in bismuthates , such hcbs couple to the breathing mode of the oxygen cage surrounding the bismuth ions @xcite . here , in this article , we study a two - dimensional ( 2d ) bose - holstein ( bh ) model for hcbs on a square lattice where they can hop to nearest - neighbor ( nn ) sites and experience the hcb - phonon interactions via a holstein - type term .
previously , exact diagonalization calculations were done on this model @xcite for a small system ( i.e. , @xmath9 lattice ) to study the resulting phase diagram .
here we use stochastic - series - expansion ( sse ) based quantum monte carlo ( qmc ) technique to simulate large - size lattices so that various phases in the thermodynamic limit can be identified more clearly . unlike the @xmath10 model , a ss is realized in our bh system due to non - negligible transport within the same sublattice . at densities
not far from half - filling and at sufficiently large hcb - phonon couplings , phase coexistence occurs ; furthermore , in the phase - coexistence region , the system tends to phase separate at stronger couplings .
our paper is organized as follows : section ii deals with a discussion of the bh hamiltonian , its transformations and its mapping to the equivalent spin model .
section iii covers a description of the numerical method and the observables employed to characterize the orderings . in section iv , we detail our results and the corresponding analysis , both with and without the presence of same - sublattice hopping terms .
lastly , in section v , we summarize our results and draw conclusions .
the bh hamiltonian is given by @xmath11 where @xmath12 and @xmath13 denote the annihilation operators for phonons and hcb particles , respectively , and @xmath14 @xmath15 is the number operator for hcbs at site @xmath16 .
furthermore , @xmath17 is the amplitude for hopping at nn sites denoted by @xmath18 and @xmath19 is the frequency of optical phonons @xcite .
an effective hamiltonian for the hcb particles is obtained by first transforming this bh hamiltonian to the polaronic frame of reference ( using the lang - firsov transformation ) and then performing perturbation theory as detailed in refs . .
interestingly , second - order perturbation theory yields a two - step hopping which produces , in a 2d square lattice , both next - nearest - neighbor ( nnn ) and next - to - next - nearest neighbor ( nnnn ) hopping terms besides the usual nn hopping term in the hamiltonian .
moreover , a nn repulsion also results from two - step virtual hopping back and forth .
so we get an effective @xmath20 hamiltonian for hcb particles on a 2d square lattice @xcite : @xmath21 where @xmath22 and @xmath23 denote nnn and nnnn sites respectively ; @xmath24exp@xmath25 , @xmath26 , @xmath27 and @xmath28 $ ] ; @xmath29 and @xmath30 $ ] . in the regime @xmath31 , we can make the approximations @xmath32 and @xmath33 \sim \frac{e^{2g^2}}{2g^2}$ ] with the approximations becoming exact for @xmath34 .
the small parameter is @xmath35 and is obtained from @xmath36^{1/2}$ ] ( see ref . for details ) ; our perturbation analysis is done in the nonadiabatic regime ( @xmath37 ) and at strong coupling ( @xmath31 ) .
since all the hoppings are non - frustrated , a qmc simulation of the system does not suffer from the negative sign problem .
we employ sse technique in our qmc simulation and investigate the co - existence or competition of cdw and superfluidity in various regimes of the parameter space . here
we should mention that our @xmath20 model for hcbs is equivalent to an extended xxz spin-1/2 hamiltonian as shown below : @xmath38 + \nonumber\\&\frac{j_{2xy}}{2}\sum_{<<i , j>>}(s_i^+s_j^-+{\rm h.c.})+\nonumber\\ & \frac{j_{3xy}}{2}\sum_{<<<i , j>>>}(s_i^+s_j^-+{\rm h.c.})-h_0\sum_is_i^z , \label{xxz_ext}\end{aligned}\ ] ] where @xmath39 , @xmath40 , and @xmath41 stand for nn , nnn , and nnnn pairs respectively .
furthermore , the operators for the hcbs are related to those of spin-1/2 particles as : @xmath42 and @xmath43 . a comparison of the parameters in eqs .
( [ he ] ) and ( [ xxz_ext ] ) yields : @xmath44 , @xmath45 , @xmath46 , @xmath47 and @xmath48 .
now , the magnetization of the system can be tuned by using an external magnetic field ; then , a term @xmath49 should be added to the hamiltonian in eq .
( [ xxz_ext ] ) where the magnetic field @xmath50 is given in units of @xmath51 .
p30.00pt|p30.00pt|p30.00pt|p30.00pt|p30.00pt|p30.00pt|p30.00pt|p30.00pt| @xmath52 & 0.5 & 1.0 & 1.5 & 2.0 & 2.5 & 3.0 + @xmath53 & 0.444 & 1.355 & 2.725 & 8.017 & 45.485 & 478.571 + @xmath54 & 0.415 & 0.970 & 0.960 & 0.647 & 0.395 & 0.255 + [ tab1 ] presence of hopping terms for hcbs indicates that superfluidity ( i.e. , spontaneous breaking of the global u(1 ) gauge symmetry ) can exist in the system . on the other hand ,
a large interaction strength suggests the possibility of a cdw .
thus , our objective is to study the compatibility of these two long range orders .
now , these two orders can coexist either in a phase separated form or homogeneously as a ss .
it should be pointed out that , a @xmath10 model on a square lattice does not show a thermodynamically stable ss phase for hcbs @xcite . on the other hand , striped ss behavior is found away from half filling when nnn repulsion ( @xmath55 ) is considered in the @xmath56 model@xcite .
we now give details of the sse - qmc simulation of our @xmath20 model , or equivalently , our extended xxz spin-1/2 model . finding the phase diagram in the present problem requires exploring various limits of the parameters ( including high anisotropy in our spin model ) . in our numerical computations ,
we used directed loop update for efficient sampling of the configurations @xcite .
the ground state properties are captured by simulating at low enough temperatures , i.e. , @xmath57 with @xmath58 being the linear dimension of the square lattice@xcite .
we employ @xmath59 since our calculations with @xmath60 yield the same values for the observables ( within the error bars of the calculations ) . from calculations involving various large system sizes
, we infer the results in the thermodynamic limit .
as can be seen from the expressions of the two - spin matrix elements for heisenberg spin models used in various sse - qmc studies @xcite , a positive parameter @xmath61 is introduced to ensure the positivity of all the matrix elements ( see appendix a for details ) .
this is necessary so that they can be treated as probabilities .
the value of @xmath61 can also affect the autocorrelation time of the desired variable .
we found that keeping the numerical value of @xmath61 equal to at least a quarter of the anisotropy parameter @xmath62 , particularly near the transition region , helps keep the data in various bins uncorrelated when each bin contains data from large number of monte carlo sweeps ( i.e. , at least 15,00,000 ) ; for the value of @xmath61 used in the xxz model , see ref .
. in this work we are concerned with the diagonal order parameter @xmath63 [ i.e , the structure factor at the neel ordering vector @xmath0 ] and the off - diagonal order parameter of the sf density @xmath65 . a general expression for @xmath66 ( for our hcb system ) is given as @xmath67 where @xmath68 denotes the ensemble average . since @xmath14 ( or @xmath69 in our spin model ) are diagonal in the basis ,
a qmc average can be computed easily . the sf density @xcite is given by @xmath70 , where @xmath71 is the free energy in the presence of twisted boundary conditions with angle of twist @xmath72 .
this is an off - diagonal order parameter . in a qmc calculation
, the sf density along @xmath73-direction is calculated using @xmath74 where @xmath75 and @xmath76 represent the total no . of hamiltonian operators transporting spin in the positive and negative @xmath73-directions , respectively @xcite .
a few benchmarking comparisons of our sse results with exact results for a @xmath77 model are shown in the supplementary material .
.1 in , @xmath63 and @xmath78 vs magnetic field h in @xmath79 square lattices with l = 10 ( line ) , 14 ( circle ) and 16 ( triangle up ) for @xmath80 and @xmath81 when ( a ) @xmath82 and ( b ) @xmath83.,width=288,height=192 ] our calculations for the 2d @xmath20 model aim at obtaining the ground state phase diagram of the system and also at extending ( to the thermodynamic limit ) the findings presented by a lanczos study on a small cluster in ref . .
we consider @xmath84 in the present study . to obtain the phase diagram for our system , the interplay between the diagonal and off - diagonal orders and the transition between them needs thorough investigation .
as antiferromagnetic order breaks su(2 ) symmetry whereas a sf phase breaks u(1 ) symmetry , a phase transition between them ( according to landau theory ) can not be of second - order type .
furthermore , a transition to a phase separated state occurs when the system undergoes a first - order transition . * .
the calculations are for a @xmath85 lattice and for our bh system [ using eq .
( [ he ] ) ] by ( a ) considering @xmath86 and ( b ) including all the interactions and hoppings.,width=288,height=384 ] .1 in we work in the grand canonical ensemble where there is no constraint on the hcb particle number ( or the magnetization in the equivalent extended xxz model ) . in fig .
[ figure 1 ] , we show the variation of @xmath78 , @xmath63 , and @xmath87 with magnetic field @xmath50 ( expressed in units of @xmath88 ) for g=1.75 for both our extended model and its xxz version ( i.e. , @xmath83 ) in @xmath79 square lattices with @xmath89 and @xmath90 .
the comparison shows that the results for different size lattices almost coincide . at small values of the magnetic field @xmath50 ,
the system manifests half - filling in the case of hcbs ( or zero magnetization in the case of the equivalent spin model ) ; the cdw phase is formed with maximum values of @xmath63 with sf density simultaneously assuming zero value . at large values of @xmath50 , before the system is at complete filling , @xmath63 decreases to zero while @xmath65 becomes finite manifesting the sf phase . in the intermediate magnetic field region ,
phase transition occurs with both the orders coexisting in our extended xxz model .
the plot of magnetization @xmath78 [ or ( @xmath91 ) where @xmath92 is the density ] in fig .
[ figure 1](b ) , indicates a discrete jump during the transition in the case of the xxz model . this is a typical signature for a first - order transition .
contrastingly , a continuous smooth increase in @xmath78 is observed for the extended xxz model clearly ruling out the possibility of a phase separation .
calculations using a canonical ensemble shows an inhomogeneous phase coexistence due to phase separation ( ps ) @xcite . in the region adjacent to @xmath93 in fig .
[ figure 1](a ) , where overlap of @xmath63 and @xmath65 are observed , the system displays a ss phase . analysis of various large system sizes confirms the picture in fig . [ figure 1](a ) that there is indeed a homogeneous coexistence of cdw and sf long - range orders in the system .
next , we will do a further detailed case - by - case study for the system with and without the effects of same - sublattice ( i.e. , nnn and nnnn ) hoppings . here
we study the case of @xmath86 in eq .
( [ he ] ) , @xmath94 , the bare xxz model .
calculations are done on a @xmath95 lattice .
we find that the system loses its cdw order at half - filling for values of the coupling @xmath96 below the critical value of @xmath97 corresponding to the heisenberg point of the xxz model . for smaller values of @xmath96
, superfluidity develops for all values of filling between 0 and 1 . the phase diagram for the xxz model
is depicted in fig .
[ figure 2](a ) .
our calculated phase diagram of the xxz model is compatible with fig .
1 , fig . 2a and
2b of ref . . at half filling ( i.e. , @xmath98 or @xmath93 ) , the heisenberg point denotes the boundary between cdw and sf phases .
[ figure 3](a ) shows the jumps in magnetization @xmath78 as soon as we increase @xmath96 beyond @xmath99 .
the magnitude of the jump increases as the coupling @xmath96 increases .
this jump implies a first - order transition and indicates a phase separated coexistence of the cdw and sf phases .
a histogram analysis can also capture the jump in @xmath78 values . in fig .
[ figure 3](b ) , plotted for @xmath100 , the @xmath78 histograms change when magnetic field values are varied .
for instance , when the magnetic field is set at @xmath101 , @xmath93 even when large number of monte carlo sweeps were used in our qmc computation . on increasing the magnetic field to @xmath102 , the @xmath78 values start showing a double - peaked structure ( with one peak at @xmath93 and another peak at @xmath103 ) which is indicative of phase separation [ see also inset in fig .
[ figure 3](b ) ] .
then , at a slightly higher magnetic field of @xmath104 , all the @xmath78 values seem to be centered around a mean value close to 0.18 .
this manifests the first - order transition .
all the discontinuous transitions are due to inhomogeneous coexistence of the cdw and sf phases .
the phase diagram for our bh model [ obtained from eq .
( [ he ] ) ] is depicted in fig .
[ figure 2](b ) . on including the effects of nnn and nnnn hoppings ( i.e. , @xmath105 ) in the @xmath20 model of eq .
( [ he ] ) , there can be a difference in the densities in the two sublattices owing to nn repulsion and same - sublattice hopping . as shown in table .
[ tab1 ] , at intermediate values of @xmath96 ( i.e. , @xmath106 ) , nnn hopping is comparable to nn hopping ; consequently , a ss state can occur .
whereas at larger values of @xmath96 ( i.e. , for @xmath107 ) , nnn hopping is fairly smaller than nn hopping and we can expect the same behavior as in the @xmath10 ( or the xxz ) model . on account of particle - hole symmetry in our model ,
the phase diagram is symmetric about half - filling .
we will now examine various features of the phase diagram of fig .
[ figure 2](b ) . the phase diagram for our model
was obtained by identifying the transition regions and the nature of the various phases .
the variation of magnetization with magnetic field in a @xmath95 lattice is shown in fig .
[ figure 6 ] in the transition region .
the results for @xmath108 show that the magnetization increases gradually without any jump as the magnetic field is increased .
hence , in fig . [ figure 6](a ) , for @xmath109 cdw and sf phases coexist homogeneously resulting in a ss state .
here we must mention that fig .
[ figure 1](a ) ( depicting simultaneous coexistence of cdw and sf phases through non - zero values of @xmath63 and @xmath65 ) corroborates this conclusion . as a matter of fact , large anisotropy [ i.e. , large values of @xmath110 in eq .
( [ he ] ) ] in the model requires large simulation time .
thus , as we increase the value of @xmath96 and thereby increase the value of @xmath111 or @xmath112 ( see table [ tab1 ] ) , the numerical calculation suffers from appreciable slowing down and with our computational constraints we can not study for large values of @xmath96 ( i.e. , @xmath113 ) .
we can set a cut - off for the anisotropy parameter @xmath114 above which the essential physics for the system does not change much ; for values of @xmath115 above the cut - off , we expect that nn occupancy is in effect projected out .
more precisely , we find that we can deal with large values of @xmath96 ( i.e. , @xmath116 ) by setting @xmath117 and still get the correct behavior of the observables thereby saving computational time . fig .
[ figure 4 ] shows @xmath78 versus @xmath50 and @xmath63 against @xmath78 for @xmath118 ; results for different values of @xmath115 are compared in order to decide on a cut - off value @xmath119 . in the large anisotropic limit
, a change in the anisotropy parameter @xmath120 can be shown to produce an additional effective field of @xmath121 ( at the mean - field level ) .
so in fig .
[ figure 4](a ) , we shifted the @xmath50 scale accordingly in an attempt to make the @xmath78 plots coincide .
the good agreement between the two cases for @xmath122 and @xmath123 gives us the freedom to use a cut - off of @xmath124 at large values of @xmath96 ( i.e. , @xmath116 ) .
when @xmath125 .
* results of ( a ) @xmath78 vs h and ( b ) @xmath63 vs @xmath78 for a @xmath95 lattice , for @xmath80 , and for @xmath118 but with @xmath126 and 20.,width=288,height=192 ] with the above simplification , we do the sse - qmc simulation for larger values of @xmath96 ( such as @xmath127 ) and observe phase - separated phases ; we explain this based on fig .
[ figure 5 ] plotted at @xmath118 . at values of magnetic field @xmath128 , a single peaked structure occurs . on increasing h , at @xmath129 , a double - peaked structure results showing simultaneous existence of two phases ( with magnetizations centered at @xmath130 and @xmath131 ) .
a further small increase to @xmath132 , leads to again a single peak ( centered around @xmath133 ) signalling that a discontinuous phase transition has occurred .
thus a phase separated state at @xmath118 is clearly captured in fig .
[ figure 5](a ) . in fig .
[ figure 5](b ) , we show the evolution of the sf density @xmath65 and @xmath63 as @xmath50 is varied and capture the first - order transition at @xmath134 . lattice for @xmath84 and for different values of @xmath96.,width=288,height=345 ] [ autocor ] the results for the magnetization versus magnetic field in the transition region for values of @xmath135 and @xmath136 are shown in figs .
[ figure 6](b ) , ( c ) , and ( d ) .
there is a gradual increase in the quantum of the magnetization jump as the coupling @xmath96 is increased .
we should also mention here that , particularly in the calculations involving large @xmath115 , we wanted to ensure that the numerical data in adjacent bins are not correlated . to this end , we compute the autocorrelation time @xmath137 defined as @xcite @xmath138=\frac{1}{2}+\sum_{t=1}^\infty a_{m_z}(t ) , \end{aligned}\ ] ] where @xmath139 with @xmath140 and @xmath17 being monte carlo times defined in units of monte carlo sweeps ( mcs ) . typically , we see that for moderately high values of @xmath115 , @xmath141 can restrict @xmath137 from attaining very large values .
thus , for @xmath142 , we use a large bin size ( i.e. , 15,00,000 mcs ) in our simulations and keep @xmath137 well within the bin size in order to produce meaningful results from our simulation .
however , for larger @xmath115 , even choosing @xmath141 can not keep autocorrelation times sufficiently smaller than such large bin sizes .
so , in those cases , we use larger @xmath61 values ( i.e. , @xmath61=6 and 8 for @xmath115=15 and 20 , respectively ) and even larger bin sizes ( i.e. , 22,00,000 mcs ) . the values of the autocorrelation times , for @xmath118 and at various fields @xmath50 close to the transition , are shown in table [ autocor ] . it should also be noted that we can not take @xmath61 too large , as a calculation with @xmath61 larger than @xmath115 does not produce meaningful results .
lastly , we mention that our phase diagrams , both for the xxz model and our extension of it , are similar ( though not identical ) to what were obtained using lanczos method in ref . .
in this work , we studied the effective hamiltonian of a bose - holstein model using the sse - qmc technique and obtained the ground state phase diagram .
we found that supersolidity is realized at intermediate couplings ; whereas , at large couplings the system phase separates because the double - hopping terms ( that produce transport in the same sublattice ) are not dominant .
our results on a large @xmath95 lattice represent well the system behavior in the thermodynamic limit , as demonstrated in fig .
[ figure 1 ] using different finite - size calculations .
our results are only qualitatively similar to those obtained earlier using modified lanczos technique on a much smaller @xmath9 cluster @xcite .
we overcame computational difficulties for large repulsive interactions by devising a cutoff repulsive strength ; above the cutoff , the system properties ( as shown in fig .
[ figure 4 ] ) become essentially independent of the strength of repulsion ( because repulsive interactions project out nearest - neighbor occupation ) . to mimic the results of statistically independent configurations , we considered sufficiently large @xmath61 values and kept the auto - correlations within acceptable limits .
our work is an exercise in sse - qmc study of a simple but important model which has significance in various fields .
we hope that the present results will stimulate further investigations in allied areas such as frustrated quantum magnets in various geometries and at various magnetizations ; coherence dynamics of excitons / spins in the presence of phonon environments ( pertinent to quantum computation and artificial light harvesting ) ; dimer formation and dimer correlations in hubbard - holstein model @xcite , hcbs coupled to multimode phonons @xcite , etc .
in sse - qmc study of heisenberg spin systems , the hamiltonian is written as a bond hamiltonian .
particularly , in our case we write @xmath143 where @xmath144 denote the nn , nnn , and nnnn bonds in our spin model , respectively .
each of such @xmath145 consists of the diagonal ( @xmath146 ) and the off - diagonal @xmath147 ) parts and is given as @xmath148 with expressions @xmath149\nonumber\\
h_{2,b_i}&=-\frac{j_{ixy}}{2}[s^+_{i(b_i)}s^-_{j(b_i)}+{\rm h.c .
} ] , \end{aligned}\ ] ] where @xmath150 , @xmath151 , @xmath152 , and @xmath153 with the coordination number @xmath154 . in our model ,
a two - spin matrix element of any of these operators can never become negative .
99 s. h. blanton , r. t. collins , k. h. kelleher , l. d. rotter , z. schlesinger , d. g. hinks , and y. zheng , infrared study of @xmath155 from charge - density - wave insulator to superconductor , phys .
b * 47 * , 996 ( 1993 ) ; a. m. gabovich , a. i. voitenko , and m. ausloos , charge- and spin - density waves in existing superconductors : competition between cooper pairing and peierls or excitonic instabilities , phys . rep . *
367 * , 583 ( 2002 ) . for a review ,
see r. l. withers and j. a. wilson , an examination of the formation and characteristics of charge - density waves in inorganic materials with special reference to the two- and one - dimensional transition - metal chalcogenides , j. phys .
c * 19 * , 4809 ( 1986 ) . w. w. fuller , p. m. chaikin , and n. p. ong , superconductivity and charge - density waves in ta- and ti - doped @xmath156 , phys .
b * 24 * , 1333 ( 1981 ) .
a. rusydi , w. ku , b. schulz , r. rauer , i. mahns , d. qi , x. gao , a. t. s. wee , p. abbamonte , h. eisaki , y. fujimaki , s. uchida , and m. rbhausen , experimental observation of the crystallization of a paired holon state , phys .
lett . * 105 * , 026402 ( 2010 ) ; p. abbamonte , g. blumberg , a. rusydi , a. gozar , p. g. evans , t. siegrist , l. venema , h. eisaki , e. d. isaacs , and g. a. sawatzky , crystallization of charge holes in the spin ladder of @xmath157 , nature ( london ) * 431 * , 1078 ( 2004 ) .
a. j. leggett , can a solid be `` superfluid '' ?
lett . * 25 * , 1543 ( 1970 ) ; a. andreev and i. lifshits , quantum theory of defects in crystals , zh . eksp .
. fiz . * 56 * , 2057 ( 1969 ) ; g. chester , speculations on bose - einstein condensation and quantum crystals , phys .
rev . a * 2 * , 256 ( 1970 ) .
e. kim and m. h. w. chan , probable observation of a supersolid helium phase , nature ( london ) * 427 * , 225 ( 2004 ) ; observation of superflow in solid helium , science * 305 * , 1941 ( 2004 ) .
d. y. kim and m. h. w. chan , absence of supersolidity in solid helium in porous vycor glass , phys .
* 109 * , 155301 ( 2012 ) .
m. boninsegni and n. v. prokofev , colloquium : supersolids : what and where are they ?
mod . phys . * 84 * , 759 ( 2012 ) .
m. boninsegni , a. b. kuklov , l. pollet , n. v. prokofev , b. v. svistunov , and m. troyer , luttinger liquid in the core of a screw dislocation in helium-4 , phys .
lett . * 99 * , 035301 ( 2007 ) .
g. g. batrouni and r. t. scalettar , phase separation in supersolids , phys .
lett . * 84 * , 1599 ( 2000 ) .
f. hebert , g. g. batrouni , r. t. scalettar , g. schmid , m. troyer , and a. dorneich , quantum phase transitions in the two - dimensional hardcore boson model , phys .
b * 65 * , 014513 ( 2001 ) .
d. heidarian and k. damle , persistent supersolid phase of hard - core bosons on the triangular lattice , phys .
lett . * 95 * , 127206 ( 2005 ) .
a. sen , p. dutt , k. damle , and r. moessner , variational wave - function study of the triangular lattice supersolid , phys .
100 * , 147204 ( 2008 ) .
r. g. melko , a. paramekanti , a. a. burkov , a. vishwanath , d. n. sheng , and l. balents , supersolid order from disorder : hard - core bosons on the triangular lattice , phys .
rev . lett . * 95 * , 127207 ( 2005 ) .
p. sengupta , l. p. pryadko , f. alet , m. troyer , and g. schmid , supersolids versus phase separation in two - dimensional lattice bosons , phys .
lett . * 94 * , 207202 ( 2005 ) .
v. w. scarola and s. das sarma , quantum phases of the extended bose - hubbard hamiltonian : possibility of a supersolid state of cold atoms in optical lattices , phys .
lett . * 95 * , 033003 ( 2005 ) .
v. w. scarola , e. demler , and s. das sarma , searching for a supersolid in cold - atom optical lattices , phys .
a * 73 * , 051601(r ) ( 2006 ) .
t. matsubara , h. matsuda , a lattice model of liquid helium , i , prog .
phys . * 16*,569 ( 1956 ) .
h. matsuda , t. tsuneto , off - diagonal long - range order in solids , prog .
theo . phys . * 46*,411 ( 1970 ) . for a treatment of polarized triplet on a dimer as a hcb ,
see k. k. ng and t. k. lee , numerical study of magnetic field induced ordering in @xmath158 and related systems , phys .
b * 73 * , 014433 ( 2006 ) . c. m. varma , missing valence states , diamagnetic insulators , and superconductors , phys . rev .
* 61 * , 2713 ( 1988 ) .
a. taraphder , h. r. krishnamurthy , rahul pandit , and t. v. ramakrishnan , negative - u extended hubbard model for doped barium bismuthates , phys .
b * 52 * , 1368 ( 1995 ) .
s. datta , s. yarlagadda , supersolidity for hard - core - bosons coupled to optical phonons , sol . state comm . * 150*,2040 ( 2010 ) .
g. d. mahan , `` many particle physics '' ( plenum press , ny , 1990 ) .
s. reja , s. yarlagadda , p. b. littlewood , phase diagram of the one - dimensional hubbard - holstein model at quarter filling , phys .
b * 84*,085127 ( 2011 ) .
r. pankaj and s. yarlagadda , study of cooperative breathing - mode in molecular chains , phys .
b * 86 * , 035453 ( 2012 ) .
r. t. scalettar , g. g. batrouni , a. p. kampf , g. t. zimanyi , simultaneous diagonal and off - diagonal order in the bose - hubbard hamiltonian , phys . rev .
b * 51*,8467 ( 1995 ) .
o. f. syljusen , a. w. sandvik , quantum monte carlo with directed loops , phys .
e * 66 * , 046701 ( 2002 ) .
o. f. syljusen , directed loop updates for quantum lattice models , phys . rev .
e * 67 * , 046701 ( 2003 ) .
g. g. bartrouni , r. t. scalettar , g. t. zimanyi , a. p. kampf , supersolids in the bose - hubbard hamiltonian , phys .
lett . * 74*,2527 ( 1995 ) .
a. w. sandvik , computational studies of quantum spin systems , aip conf .
1297 , 135 ( 2010 ) . g. schmid , s. todo , m. troyer , a. dorneich , finite - temperature phase diagram of hard - core bosons in two dimensions , phys .
* 88*,167208 ( 2002 )
. y. murakami , p. werner , n. tsuji , and h. aoki , supersolid phase accompanied by a quantum critical point in the intermediate coupling regime of the holstein model , phys .
lett . * 113 * , 266404 ( 2014 ) ; s. reja , s. yarlagadda , and p. b. littlewood , correlated singlet phase in the one - dimensional hubbard - holstein model , phys . rev .
b * 86 * , 045110 ( 2012 ) ; s. kumar and j. van den brink , charge ordering and magnetism in quarter - filled hubbard - holstein model , phys . rev .
b * 78 * , 155123 ( 2008 ) .
zi cai , u. schollwck , and l. pollet , identifying a bath - induced bose liquid in interacting spin - boson models , phys .
lett . * 113 * , 260403 ( 2014 ) .
the authors thank pinaki sengupta and keola wierschem for useful discussions on the sse technique and its implementation for our hcb system .
we also thank amrita ghosh for helping with the calculations and for cross - checking .
we have benchmarked our qmc calculations by comparing our calculated values with those obtained by exact - diagonalization ( ed ) methods .
we consider both the xxz model and its simple extension , namely , the @xmath77 model .
we find that the energy , magnetization , structure factor @xmath63 and sf density @xmath65 of our sse calculations match quite well with those from the ed results .
the comparisons of the calculated @xmath63 and @xmath65 for the @xmath77 model are shown in fig .
[ figure 7](a)-(b ) .
99 e. einarsson , h. j. schulz , direct calculation of the spin stiffness in the @xmath77 heisenberg antiferromagnet , phys .
b * 51*,6151(1995 ) .
k. runge , numerical study of the onset of superfluidity in the two - dimensional , disordered , hard - core bosons , phys .
b * 45*,13136(1992 ) . g. g. batrouni and r. t. scalettar , phase separation in supersolids , phys .
* 84 * , 1599 ( 2000 ) .
f. hebert , g. g. batrouni , r. t. scalettar , g. schmid , m. troyer , and a. dorneich , quantum phase transitions in the two - dimensional hardcore boson model , phys .
b * 65 * , 014513 ( 2001 ) .
r. t. scalettar , g. g. batrouni , a. p. kampf , g. t. zimanyi , simultaneous diagonal and off - diagonal order in the bose - hubbard hamiltonian , phys .
b * 51*,8467 ( 1995 ) . y .-
c . chen , r. g. melko , s. wessel , y .- j .
kao , supersolidity from defect condensation in the extended boson hubbard model , phys .
b * 77*,014524 ( 2008 ) . k .-
ng , y .- c .
chen , supersolid phases in the bosonic extended hubbard model , phys .
b * 77*,052506 ( 2008 ) . | exploring supersolidity in naturally occurring and artificially designed systems has been and will continue to be an area of immense interest . here , we study how superfluid and charge - density - wave ( cdw ) states cooperate or compete in a minimal model for hard - core - bosons ( hcbs ) coupled locally to optical phonons : a two - dimensional bose - holstein model .
our study is restricted to the parameter regimes of strong hcb - phonon coupling and non - adiabaticity .
we use quantum monte carlo simulation ( involving stochastic - series - expansion technique ) to study phase transitions and to investigate whether we have homogeneous or phase - separated coexistence .
the effective hamiltonian involves , besides a nearest - neighbor hopping and a nearest - neighbor repulsion , sizeable double - hopping terms ( obtained from second - order perturbation ) . at densities
not far from half - filling , in the parameter regime where the double - hopping terms are non - negligible ( negligible ) compared to the nearest - neighbor hopping , we get checkerboard - supersolidity ( phase separation ) with cdw being characterized by ordering wavevector @xmath0 . |
patrolling is continuously traveling through an environment in order to supervise or guard it .
although mostly used to refer to humans guarding an area , the term patrolling is also used to describe surveying through a digital , virtual environment .
consider , for example , the task of repeatedly reading web pages from the world - wide - web in order to keep an updated database representing the links between pages , possibly for the purpose of later retrieval of pages in an a accurate and prompt manner .
these problems exhibit similarities , in the sense that they can be represented as traveling through the vertices of a graph .
but there are also differences : a physical area is usually fixed in size , whereas the virtual area is , in general , prone to constant change .
the number of human guards is , generally , fixed for the particular area being patrolled , while the number of software agents or `` bots '' performing a large scale patrolling task may be subject to change as well . partitioning a graph into similar sized components is an important and difficult task in many areas of science and engineering . to name few examples , we can mention the partitioning of a netlist of an electronic vlsi design @xcite , the need for clustering in data mining @xcite , and the design of systems that balance the load on computer resources in a networked environment @xcite .
the general graph partition problem is loosely defined as dividing a graph into disjoint , connected components , such that the components are _ similar _ to each other in some sense .
practical considerations impose additional constrains .
for example , an important problem , known as the _ graph k - cut _ , requires a partition where the sum of the _ weights _ of vertices belonging to each component _ _ is more or less equal , and additionally , the number and/or the sum of weights of edges that _ connect disjoint components _ is minimized @xcite .
the _ k - cut _ problem can model the distribution of tasks between computers on a network , while minimizing communication requirements between them . in this work
we define a patrolling strategy that fairly divides the work of patrolling the environment among several a(ge)nts by partitioning it into regions of more or less the same size .
we have no constraints on edges connecting different components , but we impose strict restrictions on our patrolling agents in search for a heuristic multi - agent graph partitioning algorithm that may continuously run in the background of a host application .
we are interested in programming the same behavior for each individual ant - like agent , which should be very simple in terms of resources , hardware or communications .
furthermore the agents have no i d s , hence are part of a team of units that are anonymous and indistinguishable to each other .
our a(ge)nts should have very little knowledge about the system or environment they operate in , have no awareness on the size or shape of the graph , no internal memory to accumulate information , nor a sense of the number and locations of other agents active in the system .
these limitations mean that such a multi agent process has inherent scalability ; the environment might be large , complex , and subject to changes , in terms of vertices , edges and even the number of agents , and our simple agents should still be able to patrol it , while also evolving towards , and ultimately finding balanced partitions , if such partitions exist . to simplify the discussion , we will think of a graph to be partitioned as a planar area , and the task at hand will be to partition the area into regions of more or less the same size . the area is modeled as a grid , where each vertex is a unit area , thus a balanced partition should have components of roughly the same number of vertices . in our scheme , agents are each given the task to _ patrol _ and define a region of their own , and have the ability to expand their region via conquests . like ants ,
our agents leave pheromone marks on their paths . the marks decay with time and are subsequently used as cues by all the agents to make decisions about their patrolling route and about the possibility to expand their region . by assumption , each agent operates _ locally _ , thus it can sense levels of pheromones or leave pheromone marks on the vertex it is located , on its edges and on adjacent vertices . while patrolling its region , an agent visits a vertex and reads the intensity of pheromone marks that remain from previous visits
it then uses the reading , and the known rate of pheromone decay , to calculate the vertex s idle - time the time that passed since the previous visit . using the decaying pheromone mark
we can chose a patrolling rule according to which the agents visit the vertices of their region in repetitive cycles , each vertex being marked with a pheromone once on each cycle .
the patrolling process hence ensures that the idle - time measured by agents on visits to their region s vertices is the same , effectively encoding their region s _ cover time _
the time that takes for an a(ge)nt to complete a full patrolling cycle and therefore it can also be used to estimate the region s size : the shorter the cover time , the smaller the region .
we assume that each agent detects pheromones without being able to distinguish between them , except for recognizing its own pheromone .
when an a(ge)nt hits a border edge an edge that connects its region with one that is patrolled by another agent - it can use the neighbor s idle - time ( encoded in its pheromone marks ) to calculate the size of the neighboring region , and thereby decide whether to try to conquer the vertex `` on the other side of the border '' .
this causes an effect that mimics pressure equalization between gas - filled balloons : at two vertices on opposing ends of a border edge , the agent that hits the border more frequently is the agent with a shorter cover time ( patrolling the smaller region ) hence it may attempt a conquest .
we define that in a balanced partition , any pair of neighboring regions have a size difference of at most one vertex .
this means that for a graph @xmath0 and @xmath1 agents , our partitioning heuristics ensures a worst case difference of @xmath2 vertices between the largest and smallest of the regions , once a balanced partition is reached .
for example in a graph of 1 million vertices ( e.g. database entries , each representing a web page ) and 10 agents ( network bots patrolling the pages ) , this difference is truly negligible
. additionally , the length of the patrolling path is predetermined , and is proportional to the size of the region being patrolled , therefore when a balanced partition is reached , the algorithm guarantees that the idleness of any of the vertices of the graph is bounded by a number of steps equal to @xmath3 , about twice the size of the largest possible region ( note that @xmath4denotes the number of vertices in the graph @xmath0 ) . in figure
[ fig : evolution intro ] one can see a series of snapshots depicting 8 patrolling agents working to partition a 50x50 grid .
the first snapshot shows an early phase of the joint patrolling algorithm , where agents already captured some of the vertices around their initial random locations , in the second , the area is almost covered and most of the vertices of the graph are being patrolled , the third exhibits a phase when all the area is covered but the regions are not balanced , and finally , the last snapshot shows a balanced partition that the system evolved into .
evolution of a 50x50 grid graph partitioning by 8 a(ge)nts , title="fig : " ] evolution of a 50x50 grid graph partitioning by 8 a(ge)nts , title="fig : " ] evolution of a 50x50 grid graph partitioning by 8 a(ge)nts , title="fig : " ] evolution of a 50x50 grid graph partitioning by 8 a(ge)nts , title="fig : " ] this figure exhibits typical stages in the evolution of such a system , for which balanced partitions exist , and the environment graph remains stationary for a time long enough for agents to find one of them .
often , the agents will relatively quickly find a partition that covers the graph , and is _ close _ to being balanced . then
, on stationary graphs , they may spend a rather long time to reach a perfectly balanced partition . in a time
varying environment the system will continuously adapt to the changing conditions .
the concept of partitioning a graph with a(ge)nts patrolling a region and exerting pressure on neighboring regions was first presented by elor and bruckstein in @xcite .
they proposed a patrol algorithm named bdfs balloon dfs and this work is a follow up research on this problem . according to bdfs , an agent patroling a smaller region conquers vertices from a neighboring larger region . to achieve the goal of patrolling an area
, bdfs uses a variation of multi - level depth - first - search ( mldfs ) , an algorithm presented by wagner , lindenbaum and bruckstein in @xcite .
the task of the mldfs too , was to distribute the work of covering `` tiles on a floor '' among several identical agents .
the floor - plan mapping of the tiles is unknown and may even be changing , an allegory for moving furniture around while agents are busy cleaning the floor .
mldfs implements a generalization on dfs : agents leave decaying pheromone marks on their paths as they advance in the graph , and then use them either to move to the vertex least recently visited or to backtrack . when none of the choices are possible , either when the graph covering ends , or following to changes in the graph or loss of tracks due to noise in the pheromone marks , agents _ reset , _ thus starting a new search .
the time of reset , named `` the time where new history begins '' , is stored in the agents memory , as a * search - level * variable . after a _ reset , _
the cycle repeats , hence an agent traces pheromone marks left in an earlier cycle .
the mere existence of a pheromone mark is , however , not sufficient for agents to choose a path not yet taken during the current search cycle . to select the next step , agents use the value stored in the * search - level * variable as a threshold :
any pheromone that was marked on a vertex or edge prior to this time must have been the result of marking in an earlier cycle . in mldfs , pheromones of all agents are the same , and agents are allowed to step on each other s paths
. for the task of partition a graph , in bdfs each agent has its own pheromone and it performs mldfs cycles on its `` own '' region of the graph , leaving its particular pheromone marks .
as long as the region is stationary , bdfs agents exactly repeat their previous route .
if the region changes , either expands or shrinks , it will cause bdfs to look for a new and possibly substantially different route before settling into the next search cycle .
this occurs due to a subtlety in the way that depth - first - search defines a spanning tree , a special type of tree called a _ palm tree _
, were each edge @xmath5 in the spanning tree connects a vertex @xmath6 with one of its ancestors , see e.g. tarjan @xcite .
the spanning tree defined during a bdfs search cycle does not consider all edges emanating from _ all _ of the region s vertices , simply because some of the edges connect to vertices _ on neighboring _
_ when bdfs conquers a vertex , it is possible that this vertex has more than one edge connecting to the region .
all these edges will now be considered during the next search cycle , a process that may dictate a different palm tree .
we call this event - _ respanning_. in the algorithm we define here , named _ ant patrolling and partitioning , or antpap _ , we use a different generalization of dfs that avoids respanning .
furthermore we reduce the requirements on the agent s capabilities .
for example , our agents have no memory , and also can not control the levels of pheromones they leave , the pheromone level at the time of a marking is always the same .
we further add the possibility for agents to _ lose _ a vertex if a conquest fails , and we provide a proof of convergence to balanced partitions , while experimentally observing much faster evolution towards such partitions . the subject of multi - agent patrolling has been extensively studied .
lauri and charpillet @xcite also use an `` ants paradigm '' , where a method based on ant colony optimization ( aco ) , introduced by dorigo , maniezzo , and colorni in @xcite .
aco provides multi - agent solutions for various problems , for example the traveling salesman problem ( tsp ) in complete and weighted graphs by a so - called _ ant - cycle _ algorithm .
ants move to the next vertex according to a probability that is biased by two parameters : the closest neighbor vertex ( corresponding to the lowest edge weight ) and the level of pheromone on the edge . during their search
, ants record their path to avoid visiting the same vertex twice .
since at each step all ants traverse one edge to a neighboring vertex , all ants complete their travel at the same time .
thereafter each ant leaves pheromone marks on the entire path it took .
due to the probability bias , shorter edges have a higher probability to be traversed , thus it is probable that multiple ants traversed them , hence they tend to accumulate stronger pheromone levels .
the cycles repeat , and with each cycle the biasing gets stronger towards the shortest path .
the process ends after a a - priori given number of cycles complete or when the ants all agree on the shortest path . for the patrolling problem , lauri et .
@xcite used this method to find multiple paths , one for each agent , _ before _ the agents begin their joint work .
their algorithm employs multiple colonies of ants where each agent is assigned one ant on each colony .
ants in a colony cooperate ( exchange information regarding their choices ) to divide the exploration into disjoint paths , leading the agents to eventually cooperate in the patrolling task .
unlike for the tsp , the environment graph is not required to be complete , and ants are allowed to visit a vertex more than once when searching for a patrolling route .
chevaleyre , yann , sempe , and ramalho @xcite compared cyclic patrolling strategies , in which agents tend to follow each other , to partitioning strategies , in which agents patrol each its own region . by applying several algorithms on several graphs examples
, they found that the choice of strategy should be based on the shape of the graph .
the partitioning based strategy gets better results on `` graphs having long corridors connecting sub - graphs '' , i.e. if there are high weight edges that are slow to traverse , it is better not traverse them at all by allocating them to connect disjoint partitions .
there is substantial research on heuristics for partitioning of a graph , and some of it even related to multi - agent scenarios .
inspired by ants , comellas and sapena @xcite presented yet another _ ants _ algorithm to find a _
k - cut _ solution to a graph .
the system is initiated by randomly coloring all the graph vertices in a more or less even number of colors and positioning the agents randomly on the graph .
then a _ local cost _ value is calculated for each of the vertices , storing the percentage of neighbors that have the same color as its own .
agents will then iteratively move to a neighboring vertex @xmath6 that has the lowest cost ( i.e. with the most neighbors of a _ different _ color than its own ) , and then switch colors with a random vertex @xmath7 on the graph , where the color of @xmath7 is the one most suitable for @xmath6 , i.e. similar to most of @xmath6 s neighbors .
@xmath7 is selected from from a random list of vertices colored with the same color as @xmath7 , by choosing from this list the one with lowest cost ( most neighbors colored _ differently _ than @xmath7 ) .
then the cost value is refreshed for both @xmath6 and @xmath7 . on each iteration
the number of cuts , defined as the number of edges connecting vertices of different colors , is calculated over all the edges of the graph and the lowest value is stored .
the choices of agent moves are stochastic , i.e. agents have a probability @xmath8 to select the next vertex to move to by using the cost value , otherwise it selects another neighbor vertex at random .
this allows the system to escape from local minima . unlike our algorithm , agents of comellas and sapena _ _ s ants _ _ aim to find a _ k - cut _ , and while doing so do not leave pheromones to be used as cues on vertices and edges they visit as our agents do . also , their _ _
agents are assumed to have the ability to look at vertices that are anywhere in the graph and change their values , thus their sensing is not local as in our algorithm . in _
ants _ , each iteration relies on a global calculation that involves access to values on _ all _ edges of the graph , in order to measure the quality of the partition so far determined , as well as storing the result .
inspired by bee foraging , mccaffrey @xcite simulates a bee colony in order to find a _
k - cut _ graph partition .
each of the agents , in this case called _ bees _ , is assumed to know in advance the size and shape of the graph , as well as the number of components desired .
the agent must have an internal memory to store an array of vectors listing the vertices of all sub - graphs of a proposed solution , as well as the number of cuts this partition has , as a measure of its quality . in a hive ,
some 10% of the bees are considered _ scouts , _ all other agents being in one of two states , _ active _ or _ inactive .
_ emulated scouts select a random partition of the graph . if the selection is better than what the scout previously found , it stores it in its memory and communicates it to other bees in the hive that are in an _ inactive _ state .
some of those store the scout s solution in their own memory , change their state to _ active _ and begin to search for a better partition around this solution .
if an active bee finds an improved solution it communicates it to the bees that are left in the hive . after looking at neighboring solutions for a long enough time , the _ active _ bee returns to the hive and becomes _ inactive _ again .
the algorithm , therefore , is constantly searching for improvements in the quality of the partition that the bees collectively determine . the partitioning and patrolling multi - agent algorithms that we have surveyed above ,
all assume that agents posses substantial internal memory .
some algorithms assume that the agents are able to sense and even change values of vertices and/or edges in graph locations that are distant from their position in the graph , and sometimes they can even sense and/or store a representation of the whole graph in their memory .
patrolling algorithms may be partition based , and then the task is divided into two stages . in the first stage
the graph is partitioned into disjoint components , and at the second stage each of the agents patrols one of those components . in our case , partitioning the graph , and thereby balancing the workload among our agents , is a requirement .
our algorithm does not have stages , the agents simply perform pheromone directed local steps thereby carrying out a _ patrolling _ algorithm , and while doing so also implicitly cooperate in partitioning the graph .
our agents have no internal memory at all .
their decisions are based on pheromone readings from vertices and edges alone , and they can only sense or leave pheromone marks around their graph location .
one may view our solution for patrolling and partitioning the graph environment as using the graph as a shared distributed memory for our oblivious agents .
the task analyzed here is the partition of an area or environment into regions of similar size by a set of agents with severe restrictions on their capabilities .
the inspiration for the algorithm are gas filled balloons ; consider a set of elastic balloons located inside a box , and being inflated at a constant and equal rate , until the balloons occupy the entire volume of the box . while inflating , it may be that one balloon disturbs the expansion of another balloon .
this may cause a momentarily difference of the pressure in the balloons , until the pressure difference is large enough to displace the disturbing balloon and provide space for the expansion of the other . since the amount of gas is equal for all balloons
, they will each occupy the same part of the volume , effectively partitioning it into equal parts .
our agents mimic this behavior by patrolling a region of the area `` of their own '' , while continuously aiming to expand it .
the area is modeled by a graph and the region is a connected component of the graph .
when expanding regions touch , the agent on the smaller region may conquer vertices of the larger region .
we assume that initially a given number of agents are randomly placed in the environment , they start the process of expanding and this process goes on forever .
eventually the expansion is `` contained '' due to the interaction between the regions of the agent , hence the process will lead to an equalization of the sizes of the regions patrolled by the agents . in the discrete world of our agents a partition to regions of exactly the same size may not exist , therefore we define a balanced partition as such that any two neighboring regions may have a size difference of at most one vertex . * * for simplicity , a(ge)nts operate in time slots , in a strongly asynchronous mode , i.e. within a time interval every agent operates at some random time , so that they do not interfere with each other . during a given time slot
, each agent may move over an edge to another vertex , and may leave pheromone marks on a vertex and/or edges .
the marks , if made , are assumed to erase or coexist with the pheromone that remained there from the previous visit .
agents have no control over the amount of pheromone they leave , its initialization level being always the same .
thereafter , the pheromone level decays in time .
each agent has its own pheromone , thus pheromones are like colors identifying the disjoint components and hence the partitioning of the graph .
the agents themselves can only tell if a pheromone is their own or not .
agents are oblivious , i.e. have no internal memory . on each time slot ,
an agent reads remaining pheromone levels previously marked on the vertex it is located and its surroundings , and bases its decisions upon these readings .
the readings and decisions are transient , in the sense that they are forgotten when the time slot advances .
decaying pheromone marks on vertices and edges linger , serving both as distributed memory as well as means of communication . in our model
, agents leave pheromones in two patterns : one pheromone pattern is marked when agents advance in their patrolling route , and the second pattern is used when agents decide to remain on the same vertex .
pheromones are decaying in time , thus once marked on a vertex or edge , their level on the vertex or edge decreases with each time step .
a straightforward way for implementing such behavior in a computer program , is to use the equivalent `` time markings '' , i.e. stamping the _ time _ at which a pheromone is marked on the vertex or edge .
we therefore denote by @xmath9 the time of pheromone marking on vertex @xmath6 , hence @xmath10 means that an agent left a pheromone on vertex @xmath6 at time @xmath11 . as time advances , the age of the pheromone on vertex @xmath6 ,
i.e. the time interval since it was marked , which can be calculated as @xmath12 where @xmath13 is the current time , advances as well .
this is equivalent to measuring the level of the temporally decaying pheromone on vertex @xmath6 , and using its value along with the known rate of decay to calculate its `` age '' .
similarly , @xmath14 is a time marking , equivalent to the decaying pheromone level on the edge , where @xmath15 and @xmath16 are not necessarily the same .
the use of time markings require the computer program implementation to know the current time @xmath13 in order to be able to calculate the age of pheromones .
however , the knowledge of current time is strictly limited to its use in the emulation of temporally decaying pheromones by equivalent time markings , thus it does not depart from our paradigm of obliviousness and local decisions based on decaying pheromone markings only .
when an agent decides to leave a pheromone mark on a vertex , it may avoid erasing the pheromone that remains from the previous ( most recent ) visit .
we denote the previous time marking as @xmath17 , thus when an agent marks a pheromone on vertex @xmath6 , the computer program implementation moves the value stored in @xmath9 to @xmath17 and afterwards sets the new time mark to @xmath9 .
hence , the value , @xmath18 encodes the _ idle time _ of the vertex @xmath6 . * * agents patrol their region in a dfs - like route , in the sense that they advance into each vertex once and backtrack through the same edge once during a complete traversal of their region . when an agent completes traversing the region it resets ( i.e. it stays at the same location for one time step and refreshes its pheromone mark ) , and
subsequently starts the search again .
the cycles repeat the same route as long as the region is unchanged . when the region does change either expanding or shrinking out agents persist on keeping advancing and backtracking into a vertex through the same edge that was used to conquer the vertex .
this is implemented by marking _
`` pair trails '' , _ i.e. leaving pheromones over edges as well as vertices , when conquering and ( subsequently ) advancing into a vertex .
a pair trail is a directed mark from a vertex @xmath7 to an adjacent vertex @xmath6 , of the form @xmath19 , and is one of the two pheromone patterns that agents leave .
this behavior results in a patrolling process that _ follows the pair trails _ , were agents advance through the earliest marked pair trail , refreshing the marks while doing so . when all pair trails to advance through are exhausted , it backtracks through the same pair trail it entered .
an example of a route and the spanning tree it defines are depicted in figure [ fig : patrolling route and tree ] .
the departure from the classic dfs is that edges that are not marked as pair trails are ignored .
the pair trails mark a _ spanning tree _ ( which is not necessarily a palm tree ) of the region , where its root is the vertex where the search cycle begins , and each pair trail marks the path advancing upwards the tree .
when an agent backtracks to the root , it has no untraveled pair trail to advance through , and it restarts the search cycle remaining one time slot in the root .
it then uses the second marking pattern which is simply leaving a pheromone on the vertex , denoted as @xmath20 , where @xmath7 is the root . since agents advance and
backtrack once from each vertex in their region ( except the root ) and then restart a patrolling cycle in the root , the number of steps in one patrolling cycle , called the _ cover time _
, is @xmath21 .
@xmath22 denotes the region of agent @xmath23 , the set of vertices that are part of @xmath24 patrolling cycle . patrolling cycles
repeat the exact same route as long as the region remains unchanged , hence the _ idle time _ of any vertex @xmath6 of the region is also the region s cover time @xmath25 .
thus the pheromone markings on the vertex can be used to calculate the size of the region @xmath26 . * * to expand their region , agents may conquer vertices adjacent to ( vertices of ) their region . for agent @xmath23 to attempt to launch a conquest from a vertex @xmath27 to a target vertex @xmath6
, the following conditions must apply : 1 .
@xmath6 is not part of @xmath23 s region , @xmath28 , let us then assume that @xmath29 of another agent @xmath30 2 .
@xmath7 is subject to a _
double visit _ by @xmath23 , i.e. @xmath23 visits @xmath7 leaving pheromone marks _ twice _ , while @xmath6 was not visited even once by @xmath30 during the same period of time .
since the time difference between the two visits by @xmath23 is the _ cover time _ and the cover time is proportional to the size of the region , it means that @xmath23 s region is smaller than @xmath30 s , @xmath31 . an agent may check for this condition by evaluating if @xmath32 .
if the double visit condition is met , thus @xmath30 s region is larger , allow a conquest attempt if it is not larger by _ exactly one _ vertex since a difference of one vertex is considered balanced .
4 . if the double visit condition is met and @xmath30 s region is larger by _ exactly one _ vertex , allow a conquest attempt if vertex @xmath6 is stagnated
it s pheromones are older than their purported cover time .
an agent checks this by comparing the idle time @xmath12 to the cover time @xmath18 .
depending on the above conditions , an agent may stochastically _ attempt _ the conquest of vertex @xmath6 , with a predefined probability @xmath33 .
this mechanism works even if @xmath6 is not part of any of the other agent s regions , @xmath34 . in such case the pheromone marks on @xmath6
will never be refreshed and the conquest conditions hold . * * when a region expands or shrinks as result of conquests , its becomes _ inconsistent _ in the sense that the size of the region changed , but at least some of the pheromone marks on its vertices encoding the _ cover time _
@xmath35 do not reflect that immediately . to regain consistency on a vertex @xmath7 , the pheromone marks on @xmath7
must be _ refreshed _ , hence an agent must leave there a fresh pheromone , and that may occur only when the agent advances into vertex @xmath7 through a pair trail . therefore , there is a delay in the propagation of the change , thus there will be a temporary inconsistency between the actual size of the region and the cover time encoded on region s vertices .
that inconsistency is certainly not desirable since it might result in a miscalculation of conquest conditions .
consider an agent @xmath23 with a larger region than two of its neighbors @xmath30 and @xmath36 .
both neighbors will be attempting to conquer vertices from @xmath23 .
since all agents awareness is local , @xmath30 and @xmath36 have no means to know that @xmath23 is shrinking due to the work of the other as well , and as @xmath30 and @xmath36 repeatedly conquer vertices from @xmath23 , the combined conquests may accumulate to `` eat up '' too much out of @xmath23 s region up to a point where the imbalance is reversed , and the areas of both @xmath30 and @xmath36 are now larger than @xmath23 s . nonetheless , the inconsistency is temporary .
it is convenient to analyze this issue by considering the spanning tree of pair trails .
when an agent conquers a vertex and expands its region , it results in adding a leaf to the spanned tree , and losing a vertex to another agent results in the _ pruning _ of the tree , the splitting of the tree into two or more branches , while the losing agent remains on one of them . in either case ,
the _ follow the pair trails _ strategy ensures that the new route remains well defined .
it is therefore sufficient for an agent to patrol its region twice , to ensure that the region is consistent , as described in the following lemma : [ lem : a - region - is - consistent]a region is consistent if it has remained unchanged for a period of time which is twice its cover time . on the first cycle
, the agent leaves a fresh pheromone , @xmath37 , on each vertex , while the previous most recently visit , @xmath38 , may reflect an inconsistent state .
the second cycle repeats the exact same route as the first , since the region remains unchanged , and now both the most recent visit as well as the one preceding it , indicated by pheromone levels @xmath37 and @xmath38 , are updated , thus @xmath39 reflects the cover time on all vertices of the route and the region becomes consistent . * * when balloons are inflated in a box , to the observer it looks as a smooth evolution where the balloons steadily grow and occupy more of the volume until the box is filled .
but unlike gas inside a balloon that exerts pressure in all directions concurrently , our discrete agents work in steps , where at each step they attend one vertex of their region , while the other vertices may be subject to conquests by other agents .
since regions are defined by patrolling routes , an agent @xmath23 , by conquest of a single vertex from @xmath30 , may prune @xmath30 s region in a way that leaves @xmath30 to patrol a much smaller region effectively rendering it smaller than @xmath23 s .
now the `` balance tilts '' , as the region that was larger prior to the conquest becomes the smaller .
pruning may cut a spanning tree into two or more sections , but in many cases the sections of the tree may still be _ connected _ by edges that are _ not _ marked by _ _ a pair trail .
in such case , @xmath30 has an opportunity to mark a pair trail over such an edge and regain access to a branch still marked with its own pheromones . yet ,
sometimes the pruning divides the region into two unconnected components .
we call these _ balloon explosions , _ and when these occur it is more difficult for the agent that lost part of its region to regain its loss . therefore , when an agent launches a conquest attempt it is not always clear if its success will advance or set - back the evolution towards convergence .
it is then natural to add the following _ vertex loss _ rule : should an agent fail the conquest attempt , there is a predefined probability @xmath40 for _ losing _ the vertex from which the attempt was launched .
_ losing _ the vertex may indeed be a better evolution step than succeeding in that conquest , resembling actions of withdrawal from local minima used in _
simulated annealing_. in fact , this property becomes instrumental in our convergence proof for the antpap algorithm . in order to prevent
an agent from `` cutting the branch it is sitting on '' , we limit vertex loss events to steps of the patrolling process in which agents backtrack , and , symmetrically restrict conquests to steps in which the agents advance . we next list the algorithm describing the work of each agent on the graph environment . -2cm-2 cm > p9cm >
p7 cm * * & + _ _ entry point of an agent @xmath23 at time step @xmath13 . _ _ _ upon entry , the agent is located on vertex @xmath7 . _ & + & + * * for each * * @xmath41 * * if * * @xmath42 * * @xmath43 or * * @xmath44 * then * * * if * * @xmath45 * * then * * * * if @xmath46 then * * @xmath14 ; @xmath47 ; @xmath48 @xmath49 * * goto * * @xmath6 ; * * return * * @xmath50 * else * @xmath50 * * if * * @xmath51 * then * @xmath14 ; @xmath47 ; @xmath48 ; @xmath49 * * goto * * @xmath6 ; * * return * * * * if * * @xmath52 * then * * * if * * @xmath53 * * then * * @xmath54 @xmath14 ; @xmath47 ; @xmath48 * * goto * * @xmath6 ; * * return * * * * if * * @xmath55 * * then * * * if * @xmath56 * * then * * * * @xmath57 * * goto * * @xmath6 ; * * return * * @xmath58 + * return * & * * * * for each neighbor @xmath6 of @xmath7 , marked by @xmath23 s pheromone , and meeting the _ double hit . _
if size difference@xmath59 , + or @xmath6 is stagnated ( pruned or empty ) then , if not backtracked to @xmath7 if chance allows conquest ( x random @xmath60 $ ] ) + * * * * mark pair - trail into @xmath6 move to @xmath6 and exit this step if no conquest , set _ _ lose _ _ flag ( y random @xmath60 $ ] ) * * if @xmath6 has @xmath24 ( self ) mark but _ _ double visit _ _ is met , rejoin @xmath6 .
move to @xmath6 and exit this step select @xmath6 of the oldest pair - trail + * * if pair - trail points into @xmath6 , and mark on @xmath6 is older keep previous time - mark refresh the pair - trail pointing to @xmath6 move to @xmath6 and exit this step * * if pair - trail points into @xmath7 * * if _ _ lose _ _ flag is set remove time marks of @xmath7 and its pair - trails move to @xmath6 and exit this step * * refresh vertex @xmath7 + exit this step +
during a patrolling cycle , an agent attempts conquests over all the border edges of all vertices of its region .
hence its region may expand with additional vertices bordering its route .
this causes the spanning tree defined by its dfs - route to have an ever growing number of branches as the patrolling cycles continue , resulting in a tree shape resembling a `` snow flake '' .
additionally , agents have a strong tendency to form `` rounded '' regions , if the environment and other agents regions allow it .
this happens because vertices that are candidates for conquest and are adjacent to more than one of the region s vertices , have a higher probability to be conquered and incorporated into the patrolled region .
these two properties , the snow - flake like spanning tree and rounded build - ups , often assist in achieving a smooth evolution towards convergence .
thick regions make the possibility of major `` balloon explosions '' unlikely .
the thiner branches at the ends of the snow - flake - like region , cause the pruning of the region by another agent to merely `` shave '' off small fragments from the region , and are less likely to cut out a large portion of its vertices . moreover , if a greater portion was cut out by pruning , it is highly likely that the lost portion is connected by edges that were not pair trails , making it is easier for the losing agent to regain its vertices .
figure [ fig : chart region size vs time ] depicts a chart showing a typically observed evolution of 12 regions in a 50x50 square grid .
the chart describes how the sizes of the regions change with time .
the images of figure [ fig : plausible rounded and thick ] are two snapshots taken during the same evolution , the second snapshot being of the balanced partition that the agents reached .
the chart shows that some regions grew faster than others , then at some point , the regions grew enough so that the graph was covered ( or close to being covered due to continuous pruning ) , and the smaller regions began to grow at the expense of the larger regions until they all reach a very similar size , albeit not yet balanced .
for practical purposes arriving to this state in the graph might suffice , especially if the graph is constantly changing and a partition that is balanced is not well defined .
this process occurs quickly , then there is a much longer phase towards convergence .
the `` turbulence '' seen at about @xmath61 are due to some mild `` balloon explosions '' followed by recovery and convergence .
the last _ plateau _ towards convergence is short in this particular example , but in some other simulated examples it appeared much longer .
this is especially noticeable in cases which the evolution results in the system having two large regions that are adjacent and have very similar sizes but are not yet balanced , e.g. a size difference of 2 vertices .
such scenarios may increase the time required for a double visit , which is a condition for conquest .
figure [ fig : convergence time as func of graph size ] is a chart depicting time to convergence for a system with 5 agents as a function of the graph size , overlaying results of multiple simulation runs .
it shows that the spread in the time to convergence grows with size , but it also clear that the majority , depicted as dense occurrences , are not highly spread in value , indicative of runs of `` typical evolution scenarios '' on square grid , as was the environment tested in these simulations . the dotted line is an interpolation of average convergence time among the results achieved for each graph size .
the above discussed experimental evidence showcases an evolution of the system towards a balanced partition ( when the topology of the environment / graph remains stationary ) , an evolution which is smooth without `` dramatic '' incidents , driven by the antpap algorithm mimicking pressure equalization . however , antpap is a heuristic process , and the experimentally observed smooth convergence is by no means guaranteed . in the evolution towards balanced partitions
there are various events that may substantially alter the size difference between regions , and lead the system to longer and chaotic excursions .
chance dictates the way regions expand and , for example , a region may build up with less thickness in some areas allowing other agents to cut across it causing `` major balloon explosions '' .
furthermore , even a quite well - rounded region may be subject to an `` unfair '' probabilistic attack , driven to cut through its width and eventually succeeding to remove a large portion of its area . to make things even worse
, the portion of the split region that ceased to be patrolled , becomes a readily available prey to neighboring agents .
therefore , although the system is relentlessly progressing towards a balanced partition due to the rules of `` pressure equalization '' , such `` balloon explosions '' are singular events that may significantly derail the smooth evolution towards convergence , slowing the process considerably .
one may wonder if there are conditions were these events occur repeatedly , making the convergence into a balanced partition an elusive target , that may even never be reached .
clearly , there are systems where a balanced partition can not be reached , simply because one does not exist .
an evident example is a graph having the shape of a star .
consider a graph of 7 vertices , one at the center and three branches of two vertices each .
a system with 7 vertices and 2 agents should be partitioned into two connected components , one of 3 vertices and the other of 4 .
but such partition does not exist , thus repeated balloon explosions will forever occur .
interestingly , a balanced partition does exist for the same graph with 3 agents .
therefore , our first step towards proof is to precisely define the systems of interest , which are based on environment - graphs for which a balanced partitions exist for _ any _ number of agents ( up to a bound ) .
the most general set of such graphs is an interesting question in itself . for our purposes
, we shall limit our analysis to systems of the following type : 1 .
the environment is a graph @xmath0 that has a hamiltonian path , a path that passes through every vertex in a graph exactly once , with any number of agents @xmath62 patrolling it . indeed , for any graph @xmath0 that has a hamiltonian path , we can find multiple possibilities for a balanced partition for any number of agents @xmath62 . to name one , the partition where each region includes vertices that are all adjacent one to another along the hamiltonian path , and the regions are chained one after another along the path .
some of the regions can be of size @xmath63 and the others of size @xmath64 .
note that for our purposes , the path does not need to be closed , so the existence of a _
hamiltonian cycle _ is not required .
2 . the environment is a _ k - connected _ graph , a graph that stays connected when any fewer than @xmath1 of its vertices are removed , with @xmath65 agents patrolling it . in @xcite
, gyori shows that a _ k - connected _ graph can always be partitioned into @xmath1 components , including _
k _ different and arbitrarily selected vertices
. we shall analyze the evolution of the system as a stochastic process , and base our proof of convergence on the theory of markov chains .
the remainder of this section is organized as follows : we define a `` system configuration '' by considering a simple evolution example and show that there always exists a mapping from configurations to well defined `` states '' .
we then look at more complex configurations and realize that although the set of configurations is not bounded , it can be divided into a finite set of equivalence classes , each class representing a state .
hence , we conclude that the number of states in the markov chain is finite , and the evolution of the configurations maps into corresponding transitions between the states of the chain .
next we use the concept of consistency of a region , as presented in lemma [ lem : a - region - is - consistent ] , to conclude that if a balanced partition is attained , it may persist indefinitely .
this means that balanced partitions map to recurrent states in the markov chain .
we use this result to analyze the structure of the stochastic matrix that describes the chain .
then we turn to prove that it is only the balanced partitions that are mapped to recurrent states .
we first abstract the complexity of the problem by classifying all possible graph partitions into mutually exclusive classes : uncovered , covered but unbalanced , balanced but unstable , balanced and stable
. then we proceed to analyze the changes that may cause the system to shift from a configuration in one class to a configuration in another .
finally we show that when the graph has a hamiltonian path or is _ k - connected _ , despite the possibility that the system may repeatedly transition between these classes , it can not do so indefinitely and will inevitably have to sink into a recurrent state that belongs to a set of states which are all assigned to the same balanced partition , forming a so - called `` recurrent class '' .
the vertices and edges of the graph with their respective pheromone markings , and the agent locations , will be called a configuration of the system , and denoted by @xmath66 .
recall that , as discussed in section [ agents model ] ( sub - section `` agents '' ) , more straightforward time markings , ( in which the current time , @xmath67 , is marked ) , can be used to emulate their equivalent pheromone markings with temporal decay .
the diagram of figure [ fig : 4 first steps ] depicts an example of transitions between configurations of a system of a 4-vertex graph on which two agents , the _ green _ and the _ cyan _ , are active__. _ _ @xmath68 is the initial configuration at _
@xmath69 _ .
agents , shown as dots , are placed at some random initial vertices that are colored according to the agent patrolling them , _ green _ at the top - left vertex and the _ cyan _ at the bottom - right vertex .
the pheromone markings on vertices are shown as ordered pairs of time markings @xmath70 , where @xmath37 is the most recent time that a vertex was marked with pheromone , and @xmath38 is the previous time that the vertex was marked ( so , generally , @xmath71 ) .
when the two agents _ wake up _ at time slot @xmath72 the readings of pheromone marks around are all zero , therefore the _
double visit _
condition @xmath32 , is not met , and conquests are prohibited .
hence the only possible action for the agents is _ reset _ ,
i.e. leaving a fresh pheromone mark @xmath73 ( the current time ) , and thus transitioning to the new configuration @xmath74 . at @xmath75 ,
the double visit condition is again not met , and the system transitions to @xmath76 ( recall that according to the antpap algorithm , at the time that a pheromone is marked on a vertex , the previous mark is moved from @xmath37 to @xmath38 ) . at @xmath77 , conquest conditions are met for both agents , and the system may now transition to any one of the configurations @xmath78 , according to whether one or more conquests succeed , or to @xmath79 , with a probability of @xmath80 , if all 3 conquest attempts fail .
it is important to notice that @xmath79 is equivalent to @xmath76 ( and in fact the configurations are identical in terms of temporally decaying pheromone markings since @xmath81 ) . in
both we have a @xmath37 pheromone that has been freshly marked , so @xmath82 , and a @xmath38 which has been marked on the immediately preceding time slot , hence @xmath83 .
we can , therefore , map each configuration of figure [ fig : 4 first steps ] to distinct `` states '' , states @xmath84 , as depicted in figure [ fig:4 first states ] , and group @xmath79 and @xmath76 into an _ equivalence class _ of system configurations , were configurations in such equivalence class map to the same `` state '' ( in this case the equivalence class that includes @xmath79 and @xmath76 maps to state 3 ) .
calculating transition probabilities between `` states '' is straightforward , but sometimes subtleties arise , for example : @xmath85@xmath86 , @xmath87 ( both agents attempt conquests , but fail ) , and @xmath88 ( due to the strong asynchronous assumption , the _ cyan _ agent has a probability of @xmath89 to move first within the time slot , and if so , its attempted conquest has a probability @xmath90 to succeed .
the result is multiplied by a probability @xmath91 that the _ green _ agent fails in its conquest .
if the _ green _ moves first , there is a probability @xmath92 that it would fail both attempts , and then a probability @xmath90 that the _ cyan _ succeeds ) .
an edge @xmath93 which is part of a pair - trail pattern marking , so that @xmath94 , will be called a pair - trail edge . for a time interval in which no conquests or losses of a vertex in a region occur ,
the region is considered `` stable '' .
[ lem : consistent and zero = 00003d m states]a system comprising a graph @xmath0 with time - invariant topology , and @xmath1 agents , in a configuration @xmath95 , satisfying 1 .
the @xmath1 regions marked by the agents are consistent 2 .
pheromone marks exist only on vertices and pair - trail edges inside the k - regions , and no pheromone markings exist elsewhere on the graph , will transition through a finite sequence of m of states , where m is the least common multiple of the cover times of the k - regions ( i.e. @xmath96 , prior arriving to a configuration @xmath97 equivalent to @xmath95 ( i.e. @xmath98 , as long as the k regions are stable ( no conquests or losses occur ) . a consistent region @xmath99 is a region for which @xmath100 , where @xmath101 is the cover time of region @xmath99 ( see lemma [ lem : a - region - is - consistent ] ) . therefore , as long as the region is stable
, all vertices and all pair - trail edges of @xmath99 , cyclically return to the exact same decaying pheromone levels , i.e. exactly the same temporal differences @xmath12 ( where @xmath13 is the current time ) every @xmath101 steps .
we can thereby consider the @xmath1 regions in the partition , each repeating its pheromone level markings independently , as @xmath1 cyclic processes each with its own cycle time .
hence , all the processes complete an integer number of cycles every @xmath102 steps , where @xmath102 is the least common multiple of the cycle times , @xmath103 , which ensures that all vertices of all the @xmath1 regions in the configuration _
@xmath97 _ that was reached have exactly the same temporal differences as in _ @xmath95 , _ and therefore _
_ in the above lemma , we required to have no pheromones at all on edges that are not pair - trail edges .
but , if there were markings on such edges , the patrolling agents would simply ignore them , according to the antpap algorithm .
therefore such markings have no influence on the possible future evolutions of the system .
we shall formally define states of the system by grouping together configurations that have `` the same future evolutions '' , i.e. same possible future configuration transition sequences with the same probabilities .
for example , as seen above , configurations that differ only by levels of pheromones on non pair - trail edges form such equivalence classes , hence each class defines a distinct state . in systems
theory , this is the classical _ nerode _ equivalence way of defining states .
accordingly , two configurations that do not have the exact same patrolling routes ( either not having the same @xmath1 regions , or the agents have developed different patrolling paths within the regions ) can not have the same future evolutions , since , even without any conquests or loses , the future sequences of configurations that the systems go through are different due to the different patrolling steps .
therefore these two configurations can not belong to the same equivalence class thus represent distinct states .
next we turn to discuss pheromone markings that may exist on vertices and edges that are part of any current patrolling route , hence outside of all the regions .
such scenarios may occur as result of a successful conquest by an agent that disconnects the region of a neighbor and hence prunes the spanning tree of that agent , splitting it into two or more disjoint branches .
clearly , the latter agent remains on one of these branches , while the others cease to be part of its patrolling route and remain `` isolated '' .
a segment of a spanning tree ( i.e. a set of vertices marked with pheromones and connected by pair trails ) that is part of a patrolling route , thus not included in any of the regions , forms what we shall call an isolated branch . for completeness , a single such vertex that is not connected by a pair - trail is also considered an isolated branch . in our pheromone marking model we have not limited the pheromone decay , thus , on an isolated branch , pheromones may decay indefinitely .
this means that there is no bound to the set of configurations , and raises the question of whether there exists a bound to the set of equivalence classes to which they can belong , hence a bound on the number of states of the system .
we shall , therefore , consider configurations that include isolated branches , and analyze the effect of pheromone decay in these branches on the evolution of the system , or more precisely , how such decay influences `` future '' system states .
we have already seen that two configurations that do not have the exact same patrolling routes must represent different states , thus we shall verify that this distinction , by itself , does not produce an unbounded number of states . the number of permutations of possible @xmath1 stable regions is finite ( in a finite graph ) , and for each such permutation , the number of permutations of possible routes for the @xmath1 agents must be finite too ( since each of the @xmath1 regions have a finite number of edges ) . we are therefore left to show that starting at any arbitrary configuration with @xmath1 agents patrolling @xmath1 regions that also include isolated branches , all future evolutions in an arbitrarily large interval in which all @xmath1 regions remain stable , can be grouped into a finite number of equivalence classes .
let us consider a setup of @xmath1 regions and an isolated branch , where the regions are stable in an arbitrarily large interval , and further assume that a vertex @xmath6 of the isolated branch is adjacent to a vertex @xmath7 in one of the regions ( see figure [ fig : isolated branch prunning ] ) . since the regions are stable , every patrolling cycle the value @xmath105 is refreshed , thus its time - marking increases with each cycle . on the other hand ,
the time marking @xmath9 of the vertex in the branch remains unchanged .
when the agent is on vertex @xmath7 the following scenarios may arise : 1 .
the vertex @xmath6 on the branch might be marked with the same agent s pheromone , and hence moving into the adjacent vertex consists of the action of _ rejoining _ a vertex previously lost . according to antpap
, agents check for a double visit condition , i.e. @xmath32 , prior to this action .
when traversing onto the vertex , e.g. at time @xmath13 , the agent marks there a fresh pheromone @xmath48 .
this may result in splitting the isolated branch into two or more disjoint branches .
the agent will then follow the pair - trails emanating from vertex @xmath6 at which it is presently located , oblivious to the fact that pheromone marks on pair - trails and vertices of the branch are old .
thereafter the agent traverses the section of the previously - isolated branch it is located on , thus refreshing its marks , until all the section is visited ( e.g. , in the example of figure [ fig : isolated branch prunning ] , the agent will visit all vertices on the lower section ) , then it returns to the vertex @xmath7 from which the conquest was launched .
the other disjoint branches ( in the example of figure [ fig : isolated branch prunning ] , the upper section ) remain isolated .
if the branch is marked with another agent s pheromone , and conquest conditions are met ( e.g. double visit ( @xmath32 ) and the regions size difference is not exactly one vertex ( or equivalently the difference in cover time is not 2 , i.e. , @xmath42 ) , the agent may attempt a conquest on the vertex @xmath6 and thereafter on all the vertices of the branch , one by one .
3 . if the branch is marked with another agent s pheromone but the double visit condition is met ( i.e. @xmath106 ) , it may remain so only for at most two cycles of patrolling .
note that @xmath107 is growing with each agent s visit , while @xmath9 remains unchanged and as a result a double visit condition will necessarily arise .
furthermore , meeting the double visit condition also ensures that all the additional conquest conditions are met at the same time , since either the cover time encoded in the vertex of the isolated branch indicates a region size conducive to conquests , or the agent recognizes that the neighboring region is stagnated ( not being patrolled for too long , i.e. , @xmath44 ) .
therefore any further decay of the pheromone mark on the isolated branch will not influence the future behavior of the system .
a double visit is , therefore , a sufficient condition for an agent to conquer or rejoin a vertex on an adjacent isolated branch , hence we conclude the following : [ lem : n+m states]a system comprising a graph @xmath0 with time - invariant topology , and @xmath1 agents , in a configuration @xmath66 that includes exactly one isolated branch will transition through a finite sequence of at most @xmath108 states , where @xmath109 and @xmath103 , where @xmath101 is the cover time of region @xmath99 , as long as all the regions are stable ( no conquests or losses occur ) . the completion of two patrolling cycles of a region by its patrolling agent ensures that the double visit condition is met at any vertex @xmath7 of the region adjacent to a vertex @xmath6 of the isolated branch ( see discussion above ) .
therefore , after an interval of _ @xmath109 _ ( i.e. when the agent on the largest region completed two patrolling cycles ) it is certain that the double visit condition is globally met ( i.e. for any vertex @xmath7 on any of the regions adjacent to any vertex @xmath6 in the isolated branch ) .
moreover , it will be met on any time step that follows ( as long as the regions are stable ) . hence any two configurations on which the double visit condition is globally met , and have the same levels of pheromones on vertices and pair - trail edges that are in the @xmath1 regions ( but may differ in levels of pheromones on the isolated branch ) are equivalent .
since we also know , based on lemma [ lem : consistent and zero = 00003d m states ] , that every + @xmath103 time steps , all pheromones in vertices and pair - trails included in the @xmath1 regions cyclically return to the exact same decaying pheromone levels ( i.e. exactly the same temporal differences ) , we conclude that a system in a configuratio__n _ _ _ @xmath66 , _ will transition at most @xmath110 distinct states to a configuration _ @xmath95 _ ( on which the double visit condition is globally met ) and then will cyclically transition through @xmath102 states reaching , at each cycle , a configuration _ @xmath104 . _
our next analysis is of the effect of multiple isolated branches on future evolutions of a system with stable regions .
consider two scenarios , both starting with the same configuration that has one isolated branch .
an arbitrary time later , a conquest creates another isolated branch , the second branch being the same in both , only the _ time _ of its creation is different .
hence , there can be an arbitrarily large time difference between the creation of the second isolated branch in the two scenarios .
contemplating the case of an arbitrary number of isolated branches created each at an arbitrary time , the complexity of such presented scenarios may substantially grow .
nevertheless , in term of system states the above complexity does not matter .
once the decay of pheromones on an isolated branch is such that the double visit condition is globally met , the conquest or rejoin threshold is triggered , and afterwards no amount of further decay affects the future evolutions of the system .
this insensitivity holds regardless of the presence of other isolated branches , simply because the double visit is a condition , limited to the time difference encoded in pheromones a on a vertex in a region and an adjacent vertex on the branch .
thus , any two configurations that differ only by level of pheromones on isolated branches for which the double visit condition is globally met , are equivalent .
particularly , there must exist a configuration such that the level of pheromones on isolated branches is at its `` highest level '' , i.e. the time marking on each vertex of each isolated branch is the highest that allows the double visit condition to be globally met .
a branch with such `` highest level '' will have one vertex @xmath6 , where @xmath111 with a time - mark value @xmath112 where @xmath13 is the current time ( i.e. a pheromone was left there @xmath113 steps before the current time ) , and all other vertices and edges with ( lower ) values that agree with the ordered directions of pair - trails .
hence we conclude , again , that any configuration @xmath66 that includes multiple isolated branches will transition at most @xmath113 distinct states as long as all the regions remain stable . for a system with a graph @xmath0 of stationary topology , and
@xmath1 agents , the set @xmath114 of states is finite .
based on the above analysis , we conclude : 1 .
any configuration @xmath66 is equivalent to a configuration @xmath115 identical to @xmath66 except for having no pheromone marking on edges that are not pair trails .
any configuration @xmath115 that includes isolated branches is equivalent to a configuration @xmath68 identical to @xmath115 except by the levels of pheromones on vertices and pair - trail edges in those isolated branches that globally meet the double visit condition on both configurations .
specifically , in @xmath68 , vertices and pair - trail edges on each such isolated branch , will be of a `` highest level '' , i.e. will have one vertex @xmath6 , where @xmath111 with a time - mark value @xmath112 where @xmath13 is the current time ( i.e. a pheromone was left there @xmath113 steps before the current time ) , @xmath109 and @xmath103 , and all other vertices and edges with values that agree to the directions of pair - trails
. therefore any configuration @xmath66 is grouped in an equivalence class with a correspondingly `` representative '' configuration @xmath68 . to find out how many such classes exist , we observe that a @xmath68 includes the following elements : @xmath1 regions with @xmath1 patrolling routes , isolated branches that do not globally meet the double visit condition and isolated branches of a highest level of time - markings that globally meet the double visit condition . however , we have that : 1 .
the number of possible choices of @xmath1 regions is finite ( in a finite graph ) .
2 . for any arbitrary set of @xmath1 regions ,
the number of possible routes in the @xmath1 regions is finite .
3 . for any arbitrary set of @xmath1 regions ( with a particular selection of @xmath1 routes )
the number of vertices included in these is finite , thus the number of possible isolated branches is finite ( and their possible assignments to whether they meet the double visit condition or not is also finite ) .
therefore the set of possible representative configurations @xmath68 is finite , each defines an equivalence class corresponding to a distinct state of the markov chain , hence the set @xmath114 of system states is finite . concluding the above analysis we see that in spite of the infinite number of configurations possible for a system , the number of system states , though quite large , is finite .
let us denote the finite set of states of a system by _ s. _
( gallager @xcite ) , a markov chain is an integer - time process , @xmath116 for which the sample values for each random variable @xmath117 lie in a countable set s and depend on the past only through the most recent random variable @xmath118 .
clearly , any state of the system at time @xmath119 , formally represented by @xmath120 , is dependent only on the previous state @xmath121 , since our agents have no memory , and their decisions are based solely on readings from vertices and edges of the configuration , which are completely described by @xmath118 .
we can , therefore , analyze the evolution of the system based on the theory of markov chains .
our aim is to prove that the markov chain is not irreducible ( i.e. given enough time , the probability to reach some of its states tends to zero ) , and that all its recurrent states represent balanced partitions . to proceed with our analysis
, we notice that the size of set @xmath122 grows very fast with the size of the graph .
calculations show that even the simple example of figure [ fig:4 first states ] develops to a surprisingly large chain . in order to be able to describe the evolution of the system in a simple manner
, we also define a _ partition _ of the environment .
the coloring of each vertex of a configuration @xmath66 by its patrolling agent along with the set of unvisited vertices form a partition @xmath123 of the graph .
partitions are unconcerned about the levels of pheromones on the vertices and indifferent to agent locations , thus only exhibit the regions of @xmath66 .
many different configurations ( and hence states too ) correspond to the same partition , therefore we can use the concept of a _ partition _ as an abstraction referring to all those configurations .
figure [ fig : float initial - partition ] is an example of a partition of the environment graph that the system we discussed above arrived to . from our previous discussion
we know that it represents a set of states of the underlying markov chain .
one characteristic of that set of states is that it contains a cyclic path .
this reflects the fact that agents may cyclically repeat their patrolling route for some period of time during which conquests or losses do not occur , and the partition remains stationary .
in fact , having a cyclic path in the underlying markov chain is characteristic of any reachable partition .
eventually conquests or losses are stochastically enabled leading to a different partition , and , as a result , to a different set of underlying states . in figure
[ fig : transition of partitions ] , the system may remain in partition a for a while , as the underlying chain cycles through the relevant states , but eventually it will probabilistically transition to one of the partitions b , c , d , e . note
that the transition from a to d means that both agents conquered one vertex each during the same time - slot .
a _ recurrent class _ in a markov chain is a set of states which are all accessible from each other ( possibly passing through other states ) , and no state outside the set is accessible ( gallager @xcite ) .
the following lemma shows that the set of states corresponding to any balanced partition includes recurrent classes : [ lem : c4]if a system remains in a balanced partition for a period of time equal to twice the cover time of its largest region , it will remain so indefinitely . we know
from lemma [ lem : a - region - is - consistent ] , that if a region remains unchanged for a period of time which is twice its cover time , it becomes consistent , so the pheromone levels in all of its vertices correctly indicate its cover time , @xmath124 .
therefore if the system remains in a balanced partition for a period twice the largest cover time ( the cover time of its largest region ) , it is guaranteed that all the regions are consistent .
hence , we conclude that no conquest attempts are subsequently possible , since the system is balanced and conquest conditions can not be satisfied across any border edge .
the conclusion of lemma [ lem : c4 ] is that a balanced partition with all its regions consistent , must correspond to a _ recurrent _ ( and _ periodic _ ) _ class _ in the markov chain__. _ _ the random process continuously repeats a series of states based on the individual agents patrolling cycles . since agents may reach different patrolling routes for the same region , a balanced partition may correspond to multiple recurring classes .
additionally we can conclude the following : [ cor : c3 - 1]a system may enter a state in which the partition is balanced , and then move into a state in which the partition is not balanced . clearly , we see while the conditions for consistency are not satisfied for one or more regions in the partition , a conquest or loss may possibly happen , hence the partition may become unbalanced . since recurrent classes exist , the stochastic transition matrix of the markov chain of the patrolling system must have the form : @xmath125 $ ] + where @xmath126 is a matrix of transitions between transient states ( for example those corresponding to non - balanced partitions ) , @xmath127 is a matrix of transitions between states in recurrent classes ( for example those corresponding to balanced partitions with consistent regions ) , and * tr * describes the transitions from transient to recurrent states ( for example those corresponding to balanced partitions with inconsistent regions as described in corollary [ cor : c3 - 1 ] ) . note that we assume an initial distribution given by a row vector @xmath128 and hence the future distributions evolve according to @xmath129 .
each recurrent class representing one particular occurrence of @xmath1 routes in the @xmath1 regions contributes a section * * @xmath130 to * r * , of the form of a shifted identity matrix , + @xmath131 $ ] + the rank of @xmath130 is @xmath103 ( i.e. the least common multiple of the cover times of the _ k _ regions ) , and is a function of the sizes of the regions in the partition .
the contribution of a particular balanced partition would be a matrix @xmath132 that comprises of a set of @xmath130 * * matrices on its diagonal , + * @xmath133\end{aligned}\ ] ] * + where @xmath134 here is the finite number of possible route combinations in the regions that form the partition . therefore our goal is to show that the structure of * r * is + @xmath135 $ ] + + listing contributions from a finite number @xmath119 of possible balanced partitions . to achieve this
we must prove that recurring classes that are representing balanced partitions do not exist .
this is done next . to visualize the problem
, we will classify all possible partitions as shown in table 1 , recalling that a partition @xmath123 represents a set of states in the underlying markov chain : cc > p4cm
> p6 cm * partition class * & & * name * & * * description * * + @xmath136 & & not covered & any partition @xmath123 that includes at least one vertex that is part of a patrolling cycle .
+ @xmath137 & & covered , not balanced & any partition @xmath138 that includes two adjacent regions with size difference greater than one .
+ @xmath139 & & balanced , unstable & any partition @xmath140 where one or more inconsistent regions .
+ @xmath141 & & balanced , stable ( convergence ) & any partition @xmath140 where all regions are consistent .
+ & & & + by construction of the classifications @xmath142 it is clear that these are mutually exclusive as well as complete , hence they divide the set of all possible partitions , so that any partition @xmath123 can belongs to one and only one of the four defined classes .
it also means that any state of the system @xmath143 can be classified to one and only one of the above classes .
@xmath141 includes the set of partitions that comply with lemma [ lem : c4 ] , and @xmath139 includes the partitions of lemma [ cor : c3 - 1 ] .
our goal is to show that from any initial state @xmath143 the system will reach some state in @xmath141 .
figure [ fig : system modes ] , which we will soon justify , depicts the possible _ transitions _ between the classes , i.e. from a state in one class , exists a transition sequence in the markov chain to a state in another class as depicted in the diagram .
we first observe that clearly , any state in @xmath136 or @xmath137 must be _ transient _ , the system can not remain in any one of them indefinitely . in @xmath136
the partition includes free vertices ( that are not part of any region ) . for an agent that persist in visiting a neighboring vertex ,
conquest conditions will be eventually met , and the agent will make repeated attempts to conquer it . eventually all free vertices will be conquered and the graph becomes covered .
similarly , on @xmath137 , when an agent visits a vertex bordering a larger region there is a chance it will conquer the neighboring vertex .
the conquest could cause the pruning of the spanning tree and even a balloon explosion of the larger region , and the system shifts to @xmath136 , or maybe , by chance , it will move to a balanced partition and end up in @xmath139 . in @xmath139 the system
is balanced but inconsistent .
we know from lemma [ lem : c4 ] that it can become consistent , thus shift to @xmath141 , since there is always a chance that no conquests or losses will occur in any finite period of time .
if a conquest does occur , it may become unbalanced , and then we are back to @xmath136 or @xmath137 .
so any state in @xmath139 is transient and may sink to a state in @xmath141 .
we are left to show that from any state in @xmath136 and @xmath137 there exists a path to @xmath139 .
a sequence of states that switches the system back and forth between @xmath136 and @xmath137 is the scenario were the system repeatedly evolves to a covered graph only to retract by events such as a balloon explosion .
we will show now that from any configuration there is a strictly positive probability to find a balanced partition , hence move to @xmath139 , and as result there is a strictly positive probability to get into @xmath141 .
this means that @xmath141 , the set of balanced partitions , are the only partitions that map to recurrent classes . to do that we recall our restriction to hamiltonian graphs ( or _
_ k-__connected graphs ) and invoke the property of vertex loss .
[ lem : c1-c2 are transient]given a system with an environment graph @xmath0 having a hamiltonian path ( or a k - connected graph ) , and @xmath119 agents , in an arbitrary configuration @xmath144 , there is a strictly positive probability for the system to evolve to a balanced partition .
let us assume the contrary , that there exists a configuration @xmath144 representing a state @xmath145 such that the probability of any arbitrarily long step sequence starting at @xmath144 to arrive into a state in @xmath139 is strictly zero .
we first note that there is a strictly positive probability that the next change of a region is the loss of a vertex .
namely , assume that for a sequence of configurations @xmath146 , where @xmath110 is finite , at configurations @xmath147 there were no changes to any of the regions in the partition of the environment , and at @xmath95 one or more of the regions lost one vertex each , following failed conquest attempts .
this could repeat and consequently , there is a strictly positive ( though very small ) probability for a sequence of configurations @xmath148 , where @xmath102 is finite , to arrive to a configuration @xmath97 where each of the regions is of size 1 .
now , we note that there is a strictly positive probability that the next change of a region is a conquest launched from one vertex to a neighbor along a given hamiltonian path .
this could repeat until the regions form one of the possible balanced partitions along the hamiltonian path and the system is now in @xmath139 .
this evolution contradicts our assumption regarding @xmath144 , and we therefore conclude that @xmath149 is a transient set , thus from any arbitrary configuration there is a strictly positive probability to arrive to a balanced partition . note that a similar argument can be made for a _
k_-connected environment graph .
we therefore conclude : a system with a hamiltonian or k - connected graph @xmath0 with stationary topology , and n agents implementing antpap converges in finite expected time to a balanced and stable partition with probability 1 . the conclusion from lemma [ lem : c1-c2 are transient ] , is that any state @xmath150 is transient .
we also know that any state @xmath151 is transient by lemma [ lem : c4 ] .
this means that all recurrent classes included in the matrix @xmath127 of the stochastic matrix @xmath125 $ ] represent balanced and stable partitions , classified as @xmath141 . a markov chain described by a stochastic matrix of this form ,
will eventually enter a recurrent state , regardless of the initial state , and the probability that this takes more than @xmath13 steps approaches zero geometrically with @xmath13 ( see , for example , gallager @xcite ) .
we conclude that a system with a graph _ @xmath0 _ ( of _ _ stationary _ _ topology ) and @xmath119 agents implementing antpap converges with probability 1 and a finite expected time to a balanced and stable partition .
we presented and thoroughly analyzed the antpap algorithm for continuously patrolling a graph - environment with simple finite state automaton agents ( or bots ) using `` pheromone traces '' .
the simulations presented so far were on an environment in the shape of a square .
on such an environment , we know that many balanced partitions do indeed exist .
practical scenarios are seldom so simplistic . in many important cases ,
the environment graph is , in fact , uncharted and much more complex in its structure .
still , agents implementing antpap will find a balanced partition with probability one ( almost surely ) , if such a partition exists , and will certainly divide their work fairly even when such partitions do not exist .
the shape of the environment considerably affects the time to convergence .
the number of balanced partitions that the environment graph has is , naturally , one of the major factors .
so is their diversity , i.e. how different the balanced partitions are from each other .
if the balanced partitions are similar to one another , the dependency of the time to convergence on the initial locations of the agents tends to be higher than if the solutions are further apart .
consider the system of figure [ fig : t with odd ] .
the initial positions of 7 agents are shown in the first snapshot at @xmath152 .
next , at @xmath153 , the lower section becomes almost covered . at @xmath154
the upper section is almost covered , and the two agents there clearly have larger regions than the agents in the lower section . at @xmath155 , the cyan agent is trapped in the upper section , and we witness a competition between the agents from the lower section to grow their regions into the prolonged section , that the cyan agent abandoned . at @xmath156 ,
the competition ends after the yellow agent traversed into the upper section .
now we have 3 agents on each of the upper and lower sections , and one on the prolonged section .
soon after , the system reaches a balanced partition . clearly , there are many balanced partitions for this system , but all of them have one agent on the prolonged section , and 3 agents on each of the upper and lower sections
. initial conditions with 3 agents on the upper and lower sections each will ensure faster convergence to a balanced partition . following this experiment and discussion , it is interesting to consider a system with the same environment graph and an number of agents .
since only one balanced partition exists , it is reasonable to predict that the required time for convergence might be substantial .
figure [ fig : t with 2 ] shows snapshots of an evolution of this system .
both the violet and yellow agents are initially located in the lower section .
after a while , the violet agent expands its region so that part of it extends into the upper section .
a while later , the violet region covers almost all of the upper section as well as the prolonged section .
then , the yellow agent begins to expand into the prolonged section , eventually causing a `` balloon explosion '' of violet s region .
soon enough , the violet agent responds , and causes a balloon explosion of the yellow s region . due to the shape of the graph
, this cycle may repeat over and over again .
it will stop only when the single possible balanced partition is reached , and subsequently the regions `` lock - in '' , and the system remains stable . for that to happen
, an agent must conquer the appropriate half of the vertices of the prolonged section . we know
that this will eventually happen with probability 1 , however the time it will take can be very very long .
the number of agents is also an important factor of convergence time .
generally , more agents hasten the convergence .
figure [ fig : plus w 5 agents ] shows an evolution of a system with 5 agents on a different environment .
we shall call this environment graph the `` cross '' .
figure [ fig : plus w 100 agents ] shows a system with the same `` cross '' graph and 100 agents .
here , the `` pressure '' that an agent `` feels '' from other `` balloons '' quickly accumulates around its region , and the convergence is swift .
figure [ fig : plus agents vs. convergence ] depicts results of multiple simulation runs , of systems with the `` cross '' graph of figure [ fig : plus w 5 agents ] , exhibiting convergence time as a function of the number of agents . in some systems
, particular numbers of agents may cause a substantially larger time to convergence .
in figure [ fig : rooms balanced ] we present a balanced partition in a graph environment that we call `` 6 rooms '' . systems with a `` 6 rooms '' graph and 6 agents sometimes require a substantially longer convergence time , as shown in figure [ fig : rooms agents vs. convergence ] . ignoring the outliers at 6 agents , figure [ fig : rooms agents vs. convergence no outliers ] shows that the chart exhibiting convergence time as a function of the number of agents is similar in shape to the one we have seen for the `` cross '' graph , in figure [ fig : plus agents vs. convergence ] . in the simulations described above , we tested the evolution of the multi - agent patrolling process until convergence to a stable and balanced partition .
however remarkably , the system evolves rather quickly to close - to - balanced partitions due to the `` balloon '' forces implicitly driving the agents behavior . therefore , for practical purposes we see that the antpap algorithm balances the work of the agents much earlier than its convergence time , and the partitioning becomes reasonably good rather quickly .
this property is crucial in case of time varying topologies .
hence , antpap is a versatile and adaptive process .
considering again the `` 6 rooms '' example with 10 agents , we see in figure [ fig : rooms region size evoultion ] a temporal evolution of antpap until a stable and balanced partition is achieved . as is clear on the chart displaying the time evolution of the sizes of the 10 regions , the system reached convergence at approximately @xmath157 steps .
however it is also clear that after approximately @xmath158 steps , the difference between the largest and smallest regions in the partition of the environment graph is already insignificant . in the chart ,
a system is defined as `` close to balanced '' when more than 99% of the graph is covered , and the difference between the largest and smallest regions is less than 5% the ideally balanced size ( i.e. the graph size divided by the number of agents ) .
figure [ fig : rooms balanced with contours ] exhibits a partition reached when the system was `` close to balanced '' .
both snapshots show the same partition ( at two different times ) .
the snapshot at the right also shows the borders between regions that `` reached balance '' ( i.e. their size difference is at most 1 vertex ) depicted in purple .
there is only one border which is not yet balanced , between the magenta region and the dark yellow region , located in the right `` corridor '' .
these regions are close in their sizes , and as a result the double visit condition does not occur very often . despite the partition not being balanced yet ,
the division of work between agents is already fair , hence for practical purposes , a `` close to balanced '' condition is good enough .
we note in summary that antpap does not produce _ k - cut _
partitions @xcite , and generally assumes that there are no constraints on the grouping of vertices .
some important real - world problems impose such constraints , for example , the allocation of users in a social network to hosting servers , according to their interconnections .
other real - world problems , however , do not impose such constrains , for example , the division of work patrolling the world - wide - web for content analysis and classification . in view of the good properties discussed above
, we envision that antpap could become a building block for distributed algorithms aiming to fairly divide between agents the labor of patrolling an environment , using very simple agents constrained to local interactions based on tiny `` pheromone '' marks left in the environment .
10 y. elor , a. m. bruckstein , multi - a(ge)nt graph patrolling and partitioning .
proceedings of the 2009 ieee / wic / acm international joint conference on web intelligence and intelligent agent technology - volume 02 .
ieee computer society , 2009 .
r. g. downey , v. estivill - castro , m. r. fellows , e. prieto , f. a. rosamond , `` cutting up is hard to do : the parameterized complexity of k -cut and related problems '' , electronic notes in theoretical computer science , 78():209222,2003 y. chevaleyre , f. sempe , g. ramalho , a theoretical analysis of multi - agent patrolling strategies .
proceedings of the third international joint conference on autonomous agents and multiagent systems - volume 3 .
ieee computer society , pp .
1524 - 1525 , 2004 . | a team of identical and oblivious ant - like agents a(ge)nts leaving pheromone traces , are programmed to jointly patrol an area modeled as a graph .
they perform this task using simple local interactions , while also achieving the important byproduct of partitioning the graph into roughly equal - sized disjoint sub - graphs .
each a(ge)nt begins to operate at an arbitrary initial location , and throughout its work does not acquire any information on either the shape or size of the graph , or the number or whereabouts of other a(ge)nts .
graph partitioning occurs spontaneously , as each of the a(ge)nts patrols and expands its own pheromone - marked sub - graph , or region .
this graph partitioning algorithm is inspired by molecules hitting the borders of air filled elastic balloons : an a(ge)nt that hits a border edge from the interior of its region more frequently than an external a(ge)nt hits the same edge from an adjacent vertex in the neighboring region , may conquer that adjacent vertex , expanding its region at the expense of the neighbor . since the rule of patrolling a region ensures that each vertex is visited with a frequency inversely proportional to the size of the region , in terms of vertex count , a smaller region will effectively exert higher `` pressure '' at its borders , and conquer adjacent vertices from a larger region , thereby increasing the smaller region and shrinking the larger .
the algorithm , therefore , tends to equalize the sizes of the regions patrolled , resembling a set of perfectly elastic physical balloons , confined to a closed volume and filled with an equal amount of air .
the pheromone based local interactions of agents eventually cause the system to evolve into a partition that is close to balanced rather quickly , and if the graph and the number of a(ge)nts remain unchanged , it is guaranteed that the system settles into a stable and balanced partition . |
optical excitation of graphene can result in the interband population inversion @xcite . at sufficiently strong excitation
, the interband emission of photons can prevail over the intraband ( drude ) absorption . in this case
, the real part of the dynamic conductivity re @xmath0 becomes negative .
due to the gapless energy spectrum of graphene @xcite , re @xmath0 can be negative at relatively low frequencies @xmath1 , for instance , those in the terahertz ( thz ) range .
this effect can be used in graphene - based thz optically pumped lasers with the fabri - perot resonators and the resonators based on dielectric or slot - line waveguides @xcite . as was previously pointed out @xcite and analyzed @xcite , apart from the lasing associated with the stimulated generation of electromagnetic modes , the stimulated generation of different plasmons ( with their conversion into electromagnetic radiation ) can also be of practical interest .
the plasma excitations in graphene were considered , in particular , in refs .
@xcite . in this paper
, we study the propagation and amplification of surface plasmons ( sps ) in graphene structures under nonequilibrium conditions when re @xmath2 due to optical pumping .
we consider a single - graphene layer ( sgl ) and a multiple - graphene layer ( mgl ) on a substrate .
as demonstrated recently ( see the review paper @xcite and the references therein ) , the mgls , which constitute stacks of disoriented ( non - bernal stacked ) gls , exhibit similar electron and optical properties as individual gls , while the electron momentum relaxation time in such mgls can be extremely long .
the latter circumstance implies that the intraband absorption of photons and plasmons can be effectively diminished .
this makes the mgls very prospective for different optoelectronic devices considering that the net dynamic conductivity along gls increases with increasing their number in the mgl structure .
the mgl structures can be particularly effective in lasers in which the thz modes are supported by the external resonator or waveguide @xcite , thz photodetectors @xcite , and transit - time thz oscillators @xcite .
however , sgl and mgl structures can play the dual role : the imaginary parts of their dynamic conductivity provides the mode localization near the sgl or mgl ( i.e. , the formation of sps ) , while the real part provide absorption or amplification of plasmons . due to this ,
the mgl structures as sp waveguides are not always superior over the sgl structures .
+ since the thickness , @xmath3 , of the mgl structure with even relatively large number , @xmath4 , of gls is rather small in comparison with the wavelength of the plasmons under consideration , we shall consider the sgl and mgl structure alike .
the real mgl usually include a highly conducting bottom gl ( at the interface between the substrate and other gls ) with rather high electron density @xcite .
although this bottom gl can be in some way removed , we will take it into account .
it is assumed that the gl or mgl structure is illuminated from the top by light with the energy of photons @xmath5 . under optical excitation ,
the electron and hole densities exceed substantially their equilibrium values . due to this
, one can consider the electron and hole systems under consideration as characterized by the quasi - fermi energies @xmath6 , respectively , and the effective temperature @xmath7 .
a distinction in the fermi energies in gls with different indices is due to the attenuation of incident optical radiation associated with its absorption in the gls closer to the structure top : @xmath8 .
the electron - hole system in the bottom gl in mgls is characterized by the unified fermi energy @xmath9 determined by the doping of this gl .
if the characteristic time , @xmath10 , of the emission of the optical phonon by an electron or a hole is much shorter than the time of pair collisions , the photoexcited electrons and holes emit a cascade of optical phonons and occupy low energy states in the conduction and valence bands , respectively . in this case , the contribution of optical excitation to the heating of the electron - hole system is small , so that the effective temperature @xmath7 is close to the lattice temperature @xmath11 @xcite . in the opposite case ,
the photoexcited electrons and holes are thermalized , and their effective temperature is determined by the pumping photon energy @xmath5 and the rate of electron and hole energy relaxation .
in such a situation , the effective temperature can be elevated , so that somewhat stronger optical pumping might be needed to fullfil the condition re @xmath2 @xcite .
the net dynamic conductivity in the lateral direction of a structure with @xmath4 gls at the signal frequency @xmath1 can be presented as the sum of the contributions of the individual gls @xmath12 ( @xmath13 ) and the bottom gl @xmath14 ; @xmath15 considering the expressions for @xmath12 and @xmath14 obtained previously ( see , for instance , refs , @xcite ) , one can arrive at the following : @xmath16 + \displaystyle \tanh\biggl(\frac{\hbar\omega - 2\varepsilon_f^{(k)}}{4k_bt}\biggr ) - \frac{4\hbar\omega}{i\pi}\int_0^{\infty}\frac{g(\varepsilon , \varepsilon_f^{k})- g(\hbar\omega/2 , \varepsilon_f^{k } ) } { ( \hbar\omega)^2 - 4\varepsilon^2}d\varepsilon\biggr\ } , % \ ] ] @xmath17\ ] ] @xmath18^{-1 } - \biggl[1 + \exp\biggl(\frac{\hbar\omega/2 + \varepsilon_f^b}{k_bt}\biggr)\biggr]^{-1 } - \frac{4\hbar\omega}{i\pi}\int_0^{\infty}\frac{g(\varepsilon , \varepsilon_f^b)- g(\hbar\omega/2 , \varepsilon_f^b ) } { ( \hbar\omega)^2 - 4\varepsilon^2}d\varepsilon\biggr\}.\ ] ] here , @xmath19 is the electron charge , @xmath20 and @xmath21 are the electron and hole momentum relaxation times in the bottom and other gls , respectively , @xmath22 is the reduced planck constant , @xmath23 is the boltzmann constant , and @xmath24 in the case of sgl structure , one should put @xmath25 and @xmath26 . for the mgl structures without the bottom gl , one can use eq .
( 3 ) formally setting @xmath26 .
the quasi - fermi energies in the gls with @xmath27 , @xmath28 , are mainly determined by the electron ( hole ) density in this layer @xmath29 , with @xmath30 ( at sufficiently strong degeneracy of the electron and hole systems ) , i.e. , by the rate of photogeneration @xmath31 by the optical radiation at the @xmath32th gl plane .
considering the attenuation of the optical pumping radiation due to its absorption in each gl , one can obtain that @xmath28 can be expressed via the quasi - fermi energy in the topmost gl @xmath33 , which , in turn , is a function of the the intensity of incident pumping radiation @xmath34 .
considering the mgl structures , the @xmath28 versus @xmath35 dependence was found as previously @xcite .
k ( solid lines ) and @xmath36 k ( dashed lines ) with the same as in fig . 2 values of the quasi - fermi energy . ,
width=283 ] in a sgl structure at @xmath37 k ( upper panel ) and @xmath36 k ( lowe panel ) and different quasi - fermi energies ( 1 -@xmath38 mev , 2 -@xmath39 mev , and 3 -@xmath40 mev ) .
markers show the dependences for equilibrium electron - hole systems calculated using the derived formulas as well as using @xcite for intrinsic and doped sgl structures .
, title="fig:",width=283 ] + in a sgl structure at @xmath37 k ( upper panel ) and @xmath36 k ( lowe panel ) and different quasi - fermi energies ( 1 -@xmath38 mev , 2 -@xmath39 mev , and 3 -@xmath40 mev ) .
markers show the dependences for equilibrium electron - hole systems calculated using the derived formulas as well as using @xcite for intrinsic and doped sgl structures .
, title="fig:",width=283 ] k. @xmath40 mev , and @xmath41 ps . , title="fig:",width=283 ] + k. @xmath40 mev , and @xmath41 ps . , title="fig:",width=283 ] using the the maxwell equations , in which the current associated with electron - hole system is given by of @xmath42 the components of the electric and magnetic fields , @xmath43 and @xmath44 can be presented in the following form : @xmath45 at @xmath46 , and @xmath47 at @xmath48 . here
, @xmath1 is the signal ( plasmon ) frequency , @xmath49 is the speed of light , @xmath50 is the refraction index ( generally complex ) of the substrate , @xmath51 is the complex propagation index ( which relates to the wavenumber @xmath52 ) , the axis @xmath53 is directed perpendicular to the structure plane , the axis @xmath54 corresponds to the in - plane direction along the sp propagation , and @xmath55 is the dirac delta function .
the latter describes the localization of the plasmon near the sgl or mgl structure plane .
considering eqs .
( 6 ) and ( 7 ) , from the maxwell equations we obtain the following dispersion equation for sps : @xmath56 equation ( 8) can be derived from that obtained previously using a different approach @xcite .
what is important for the present consideration is that @xmath0 is determined by both nonequilibrium electron and hole components with , in particular , re @xmath57 . solving eq .
( 8) with respect to @xmath51 , one can find the propagation index re @xmath58 and the sp absorption ( amplification ) coefficient 2im @xmath59 or the sp gain -2im @xmath60 . in the cases
when @xmath61 ( sgl or mgl structure levitating in the free space ) , eq .
( 8) obviously yields @xmath62
figure 2 presents the frequency dependences of the real part of the propagation index re @xmath63 shown by solid lines calculated for sgl structures with different substrate refraction indices @xmath50 at different temperatures @xmath7 and different values of the quasi - fermi energies @xmath64 ( i.e. , for different pumping intensities ) .
dotted lines in fig .
2 correspond to the frequency dependences of the group velocity of sps normalized by the speed of light in vacuum @xmath49 . in our model
we disregarded the effect of spatial dispersion on the dynamic conductivity of sgl and mgl structures , i.e. , the dependence of @xmath0 on the sp wavenumber @xmath65 in eqs .
( 1 ) - ( 4 ) .
this is valid if @xmath66 or @xmath67 @xcite , where @xmath68 cm / s is the characteristic velocity of electrons in holes in graphene . in the case of sgl structures with @xmath61 ,
these inequalities are satisfied in the frequency range under consideration ( see figs . 2 and 3 )
. however , as follows from fig . 2 at larger @xmath50 , the validity of the dependences obtained is limited by relatively low frequencies ( @xmath69 thz for @xmath70 at @xmath37 k ) .
one can see from fig . 2 that the sp group velocity can be negative ( at @xmath36 k ) at elevated frequencies .
this corresponds to backward waves @xcite and occurs at the frequencies at which the imaginary part of the dynamic conductivity is determened primarily by the interband transitions .
figure 3 shows the spectra of sps , i.e. , @xmath1 versus re@xmath71 dependences in the sgl structures at different temperatures at the same pumping conditions as in fig . 2 .
figure 4 shows the sp absorption coefficient 2im @xmath71 as a function of frequency calculated for different temperatures @xmath7 and different values of the quasi - fermi energies @xmath64 .
one can see that an increase in @xmath64 leads to widening of the frequency range where i m @xmath72 and a marked increase in the absolute value of the absorption coefficient in this range .
the dependences indicated by markers were calculated for equilibrium sgls using the above formulas which in the absence of optical pumping at @xmath61 coincide with those obtained previously @xcite .
the markers correspond to intrinsic ( with the fermi energy @xmath73 ) and doped sgl structures .
the dependences for doped gl structures at the equilibrium correspond to the fermi energies @xmath74 in doped sgls at the equilibrium equal to the quasi - fermi energies @xmath64 in the optically pumped sgls .
clear distinctions in these dependences in the equilibrium and pumped gls are associated with the contributions of both electrons and holes to the negative dynamic conductivity due to the interband transitions ( in the latter ) .
figure 5 shows the spatial distributions ( in the direction perpendicular to the sgl plane ) of the sp electric and magnetic fields in the sgl structures with different substrate refraction indices . :
1 - @xmath75 , @xmath76 , @xmath77 , and @xmath78 , and @xmath79 ( @xmath37 k , @xmath80 mev ) .
, width=279 ] mev , @xmath41 ps , @xmath81 mev , @xmath82 ps , @xmath83 mev , @xmath41 ps , and @xmath84 mev , @xmath82 ps . , title="fig:",width=279 ] + mev , @xmath41 ps , @xmath81 mev , @xmath82 ps , @xmath83 mev , @xmath41 ps , and @xmath84 mev , @xmath82 ps . , title="fig:",width=291 ] ps ) at @xmath85 thz , @xmath86 thz , @xmath87 thz . , width=279 ]
figure 6 shows the frequency dependences of the sp absorption coefficient calculated for sgls with the substrates with different real and imaginary parts of the refraction index ( different absorption in the substrate ) . as follows from fig . 6
, the contribution of the substrate to the sp absorption can be insignificant at the realistic values of the imaginary part of the substrate refraction index .
in particular , in the case of the substrate made of undoped si ( i m @xmath88 @xcite ) , the imaginary part of the refraction index can be smaller than those used in the calculations of curves 3- 5 in fig .
figure 7 shows the frequency dependences of the real part of the propagation index and the sp absorption coefficient .
these dependences were calculated for sgl structures with different electron and hole momentum relaxation times ( @xmath41 ps and @xmath82 ps ) for different @xmath64 , i.e. , for different pumping intensities assuming that @xmath37 k. as seen from fig . 7 , in the sgl structure with a relatively long momentum relaxation time ( @xmath41 ps ) , the sp absorption coefficient changes its sign at moderate values of @xmath64 ( @xmath89 mev ) .
however when @xmath82 ps , the sp absorption coefficient does not change the sign at least at @xmath90 mev ( see below ) , although it is markedly smaller than in the equilibrium conditions ( without pumping ) .
however , at elevated pumping intensities ( elevated values of the quasi - fermi energy ) the absorption coefficient in the slg structures even with short momentum relaxation times can become negative although at higher frequencies ( see fig .
8) . ) and mgl structure ( @xmath91 ) at @xmath40 mev .
, width=298 ] mev for sgl structure with 1- @xmath41 ps , 2- @xmath92 ps , and 3- @xmath93 ps and for mgl structure with @xmath91 ( dashed line ) and @xmath41 ps . , width=264 ] at @xmath37 k , @xmath80 mev , @xmath41 ps , and @xmath94 thz . , width=283 ] .
, width=279 ] in figs . 9 and 10
, we compare the real parts of the sp propagation index , absorption coefficients , the sp group velocities in the optically pumped sgls and mgls ( @xmath40 mev ) . as seen , the sp characteristics in sgls and mgls are markedly different .
in particular , sgls exhibit substantially stronger ( about 50 times in the peaks ) amplification ( higher plasmon gain ) in a wide frequencies range ( from 4 to 8 thz ) at @xmath41 ps .
this is attributed to markedly different values of the sp group velocities in sgls ( relatively small plasmon group velocity ) and in mgls ( in which this velocity is relatively high ) , as shown in fig .
9 , which , in turn , is due to different net electron and hole densities and different widths of the sp electric field localization that is clearly seen in fig . 11 .
an example of the dependence of the sp absorption coefficient in the structures with optically pumped gls on the number of the latter is demonstrated in fig .
thus , at the same values of the optical pumping intensity and the electron and hole momentum relaxation time , sps in sgl structures can exhibit stronger amplification than mgl structures .
however , as demonstrated @xcite , the momentum relaxation time in epitaxially grown mgls can be rather long , so mgl structures might be preferable in applications . taking into account the real possiblity of a large difference in the electron and hole relaxation times in sgl and mgl structures , in fig .
10 , we demonstrate that the sp amplification in the mgl structure with @xmath91 and @xmath41 ps can exceed that in the sgl structure with smaller @xmath21 .
hence , to achieve the sp maximum gain , the number of gl layers should be optimal .
a decrease in the sp gain with increasing @xmath4 is in contrast to the behavior the electromagnetic modes in optically pumped mgl structures with the dielectric waveguide considered by us previously @xcite . for comparison , the pertinent dependence taken from ref .
@xcite for the waveguide thickness @xmath95 m ( at the same quality of gls and pumping conditions ) is shown in fig .
12 as well .
we studied the sps in optically pumped sgl and mgl structures . using the developed model , we calculated the sp dispersion relations , spatial distributions of their electric and magnetic fields , and frequency dependences of the absorption coefficient as functions of the optical pumping and factors determining the intraband absorption in gls and the substrate .
it was demonstrated that at sufficiently strong but realistic optical pumping , the sp absorption coefficient can be negative ( so that the gain is positive ) in the certain range of frequencies .
the absolute value of the sp absorption coefficient ( plasmon gain ) in sgl structure can be fairly large and markedly exceed the gain of electromagnetic modes in the dielectric waveguides with optically pumped sgl or mgl structures .
in contrast to the cases of amplification of electromagnetic modes in the structures with optically pumped sgl and mgl and dielectric waveguide , the sp amplification weakens with increasing number of gls @xmath4 .
however , due to possibility of rather long momentum relaxation times of electrons and holes in mgl structures , the sp maximum gain can be achieved at optimal values of the nimber of gls .
the effect of sp amplification in optically pumped sgl and mgl structures can be used in thz lasers ( with the conversion of sps to the output electromagnetic modes ) .
the variation of the sp characteristics by optical pumping might be useful in different gl - based plasma - wave devices .
the authors are grateful to m. ryzhii for comments .
this work was supported by the japan science and technology agency , crest , japan and partially by the federal russian program `` scientific and educational staff '' , russia .
l. a. falkovsky and a. a. varlamov , eur .
j. b * 56 * , 281 ( 2007 ) .
v. ryzhii , a. satou , and t. otsuji , j. appl . phys .
* 101 * , 024509 ( 2007 ) .
g. w. hanson , j. appl .
phys . * 103 * , 064302 ( 2008 ) . m. jablan , h. buljan , and m. solijacic , phys .
b * 80 * , 245435 ( 2009 ) .
m. orlita and m. potemski , semicond .
* 25 * , 063001 ( 2010 ) .
v. ryzhii , m. ryzhii , v. mitin , and t. otsuji , j. appl . phys . * 107 * , 054512 ( 2010 ) .
v. ryzhii , m. ryzhii , v. mitin , and m. s. shur , appl .
express * 2 * , 034503 ( 2009 ) .
m. ryzhii , v. ryzhii , t. otsuji , v. mitin , and m. s. shur , phys .
b * 82 * , 075419 ( 2010 ) . f. varchon , r. feng , j. hass , x. li , b. ngoc nguyen , c. naud , p. mallet , j .- y .
veuillen , c. berger , e. h. conrad , and l. magaud , phys .
lett . * 99 * , 126805 ( 2007 ) .
v.ryzhii , m.ryzhii , and t.otsuji .
( c ) * 5 * , 261 ( 2008 ) . | we analyze the surface plasmons ( sps ) propagating along the optically pumped single - graphene layer ( sgl ) and multiple - graphene layer ( mgl ) structures .
it is shown that at sufficiently strong optical pumping when the real part of dynamic conductivity of sgl and mgl structures becomes negative in the terahertz ( thz ) range of frequencies due to the interband population inversion , the damping of the thz sps can give way to their amplification .
this effect can be used in graphene - based thz lasers and other devices . due to relatively small sp group velocity
, the absolute value of their absorption coefficient ( sp gain ) can be large , substantially exceeding that of the optically pumped structures with the dielectric waveguide .
the comparison of the sgl and mgl structures shows that to maximize the sp gain the number of gl layers should be properly choosen . |
the ability to confine ultracold bose and fermi gases inside artificial crystals generated by standing - wave laser light fields , i.e. , optical lattices , offers the possibility to create ideally clean and highly tunable strongly interacting quantum many - body systems @xcite .
the low - energy properties of these systems can be described using models borrowed from condensed matter systems @xcite whose parameters and dimensionality can be controlled with high precision . this fact opens new routes for understanding the physics of long - standing problems of strongly correlated systems , like the phase separation , metal - insulator transitions , superconductivity , etc . on the other hand ,
the quadratic confining potential , present in addition to the regular `` lattice '' potential leads to a number of fundamentally new phenomena .
for example , it was shown @xcite that in the presence of a confining potential the mott - insulating phase is restricted to the domain that coexists with a metallic phase , in contrast to the global character typical of solid - state systems .
moreover , mixtures of two - component atoms with different masses ( e.g. , @xmath0 and @xmath1 ) introduce an additional parameter , namely , the difference between the hopping amplitudes associated with each species of atoms in the optical lattice .
this may affect the stability of the possible quantum phases or even induce new ones @xcite .
for these reasons the mixtures of ultracold atoms in optical lattices belong to the most intensively studied subjects of contemporary experimental and theoretical physics . in this paper
we investigate the ground - state properties of mixtures of two species of fermionic atoms with strongly different masses in a harmonic potential .
since the tunneling rate decreases exponentially with the square root of mass of the atom , and particles of different species on the optical lattice interact only through the on - site interaction , this leads naturally to the falicov - kimball hamiltonian @xcite with a confining potential @xcite @xmath2 where @xmath3(@xmath4 ) and @xmath5(@xmath6 ) are the creation ( annihilation ) operators of heavy ( @xmath7 ) and light ( @xmath8 ) particles at lattice site @xmath9 .
the number of lattice sites is @xmath10 and is selected so that all the fermions are confined in the trap .
we denote the total number of fermions in the trap as @xmath11 and consider equal number of heavy ( @xmath12 ) and light ( @xmath13 ) atoms . the first term of eq
. 1 is the kinetic energy corresponding to quantum - mechanical hopping of the light @xmath8 atoms between the nearest - neighbor sites @xmath9 and @xmath14 .
these intersite hopping transitions are described by the matrix elements @xmath15 , which are @xmath16 if @xmath9 and @xmath14 are the nearest neighbors and zero otherwise ( in the following all parameters are measured in units of @xmath17 ) .
the second term represents the on - site coulomb interaction ( @xmath18 ) between the light and heavy atoms .
the last term is the energy of light and heavy atoms in the harmonic trapping potential . in accordance with similar studies for the asymmetric hubbard model @xcite
, we consider here the same trapping potential for both species of atoms .
since in this spinless version of the falicov - kimball model with a confining potential the @xmath7-heavy atom occupation number @xmath19 of each site @xmath9 commutes with the hamiltonian ( 1 ) , the @xmath7-heavy atom occupation number is a good quantum number , taking only two values : @xmath20 or 0 , according to whether or not the site @xmath9 is occupied by the heavy atom .
therefore , the hamiltonian ( 1 ) can be written as @xmath21 where @xmath22 .
thus for a given configuration of heavy atoms @xmath23 defined on the one - dimensional lattice , the hamiltonian ( 2 ) is the second - quantized version of the single - particle hamiltonian @xmath24 , so the investigation of the model ( 2 ) is reduced to the investigation of the spectrum of @xmath25 for different configurations of heavy atoms .
it is well known that in the absence of harmonic confinement ( @xmath26 ) the one - dimensional falicov - kimball model exhibits a rich spectrum of solutions including phase separated ( @xmath27 small ) as well as most homogeneous distributions of heavy atoms @xcite . however , due to the confining potential the lattice sites become , inequivalent , and thus , it is of fundamental importance to analyse the interplay between the on - site coulomb interaction and the confining potential . to describe the system at nonzero @xmath28 we have calculated various local quantities , like the local density of heavy atoms ( @xmath29 ) , the local density of light atoms ( @xmath30 ) , the total site occupation ( @xmath31 ) , the variance of the local density ( @xmath32 ) and the local double occupation ( @xmath33 ) , as functions of total number of the confined atoms , the coulomb interaction @xmath27 and the confining potential @xmath28 . the ground states are calculated by a well - controlled numerical method that we have elaborated @xcite for a description of the conventional falicov - kimall model ( @xmath26 ) .
later , the method was successfully used for various generalizations of the falicov - kimball model and different physical problems @xcite .
its generalization on systems with a harmonic potential is straightforward too .
in fig . 1 we present results of our numerical calculations for @xmath34 and @xmath35 obtained on the one - dimensional cluster of @xmath36 sites at @xmath37 , @xmath38 and different fillings .
we have added also the profiles of the local compresibility that has been proposed by rigol et al .
@xcite as a local order parameter to characterize the mott - insulator regions .
this quantity is defined as @xcite @xmath39 where @xmath40 is the density - density correlation function and @xmath41 , with @xmath42 the correlation length of @xmath43 in the unconfined system at half - filling for a given @xmath27 .
the factor @xmath44 is chosen within a range where @xmath45 becomes qualitatively insensitive to its precise value @xcite .
the insulating regions are then characterized by @xmath46 .
for the values of @xmath27 used here we usually have @xmath47-@xmath48 with @xmath49 .
the most interesting result obtained at low particle fillings ( we note that @xmath50 ) is the observation of the complete phase separation with heavy atoms in the center of the trap ( @xmath51 ) and light atoms in the surrounding regions ( @xmath52 ) .
such a behaviour is found for all particle fillings from @xmath53 to some critical value @xmath54 that rapidly decreases with increasing @xmath28 and is almost independent of the local coulomb interaction @xmath27 . in the regions where @xmath55 ( @xmath56 ) the variance of the local density @xmath35 and the local compresibility @xmath57 are finite ( the metallic phase ) , while in the middle of the trap where @xmath58 ( @xmath59 exponentially decreases in this region ) both @xmath35 and @xmath57 are equal to zero ( the insulating phase ) .
thus in accordance with results obtained for the hubbard model @xcite ( the hopping probabilities are same for both types of atoms ) we have found that also in the falicov - kimball model insulating domains coexist with metallic regions , such that global quantities are not appropriate to describe the system . at higher particle fillings the situation is more complex . above the critical filling @xmath54 ( @xmath60 , for @xmath61 ) the connected cluster by heavy atoms occupied sites ( in the center of the trap ) splits on smaller clusters , usually of the same size , separated by the empty site . as @xmath62 increases the size of clusters decreases from @xmath62 to 1 .
of course , the redistribution of heavy atoms has dramatic consequences on the distribution of light atoms .
now , the light atoms occupy preferably the empty sites in the middle of the trap ( @xmath63 ) what leads to the atomic - density waves in @xmath59 and @xmath64 profiles . in the region where @xmath65 the variance of the local density is finite but smaller than in surrounding metallic regions indicating @xcite the presence of mott - insulating phase in the center of the trap .
this conjecture supports the behaviour of the local compresibility @xmath66 that is equal to zero in the corresponding region . increasing the number of particles up to @xmath67 , the connected cluster of heavy atoms ( @xmath58 ) starts to form in the middle of the trap . in this region both
the variance of the local density and the local compresibility are finite what indicates the presence of metallic phase in the center of the trap . upon adding more fermions
, this new metallic phase widens spatially , while the mott - insulating regions of the atomic density waves are pushed to the borders and completely disappear at @xmath68 . at this filling practically the whole region of the trap , except the edges ( where @xmath69 ) , is metallic . with still a higher filling this metallic phase
is further stabilized , but at some critical filling ( @xmath70 ) a new insulating phase ( `` a band insulator '' ) starts to develop in the center of the trap ( @xmath71 ) .
this trend holds also for the highest particle fillings , the width of the band - insulating phase increases and the surrounding metallic regions are gradually suppressed . to reveal the role of the coulomb interaction @xmath27 and the confining potential @xmath28 on a formation of metallic and insulating domains ,
similar calculations have been also performed for various combinations of @xmath27 and @xmath28 .
the results of numerical calculations for the site occupation @xmath64 as a function of @xmath9 are displayed in fig .
one can see that the coulomb interaction @xmath27 and the confining potential @xmath28 exhibit precisely opposite effects on the stability of metallic and insulating domains .
indeed , with increasing @xmath28 ( at fixed @xmath27 ) the insulating domain @xmath72 is suppressed and the metallic domains are stabilized , while with increasing @xmath27 ( at fixed @xmath28 ) the metallic domains are suppressed and the insulating domain is stabilized . since the falicov - kimball model can be considered as a simplification of the hubbard model ( only one kind of particles , say with spin up can hop ) it is interesting to compare results obtained in these two different limits . such a comparison ( see ref . 3 and ref .
4 ) reveals obvious differences in behavior of these models in the confining potential .
for example , the ground - state of the hubbard model ( for @xmath73 ) is always metallic at low particle fillings , while the metallic regions coexist with the insulating region in the ground state of the falicov - kimball model .
moreover , the local density profiles exhibit obvious oscillations for the falicov - kimball model , while no sign of such oscillations has been observed for the repulsive hubbard model ( with the exception of hartree - fock @xcite and variational studies @xcite , that were not confirmed by a projector monte carlo simulations , however @xcite ) . to exclude the possibility that oscillations are a consequence of a finite size of clusters used in our numerical calculations , we have performed an exhaustive finite - size scaling analysis on finite clusters up to @xmath74 sites for all particle fillings from fig . 1 .
this analysis showed that the ground states found for @xmath36 hold also on clusters of @xmath75 and @xmath74 sites and thus they can be satisfactorily extrapolated on much larger clusters . in fig .
3 we present numerical results for extrapolated ground states obtained for a cluster of @xmath76 sites and the same values of @xmath77 and @xmath27 as used in fig . 1 .
these results clearly demonstrate that increasing @xmath10 suppresses the atomic - density oscillations in the metallic phase , but stabilizes the atomic - density waves in the insulating phase . since in the real experiments with ultracold atoms the hopping matrix elements ( of heavy atoms ) between the nearest - neighbor sites ( @xmath78 ) are not strictly equal to zero
, it is necessary to examine the stability of our solutions obtained for @xmath79 against the finite values of @xmath78 .
for this reason we have performed exhaustive exact - diagonalization studies of the asymmetric hubbard model @xcite ( @xmath80 ) in a confining potential for a wide range of model parameters ( @xmath81 and @xmath28 ) on finite clusters up to @xmath82 .
the representative examples of local density profiles are displayed in fig . 4 for several different values of @xmath78 .
they clearly show that results obtained for @xmath79 remain stable also at finite ( small ) @xmath78 . thus the falicov - kimball model with a confining potential
can be used satisfactorily to model the ground - state behavior of mixtures of two - component fermionic atoms ( with strongly different masses ) in a harmonic potential .
finally , we have also calculated the variance of the local density as a function of the local density @xmath64 .
it is known from the study of the one - dimensional hubbard model in a harmonic potential @xcite that this quantity shows universality with respect to the confining potential for @xmath83 .
it should be noted that the universal behavior was observed only for strong coulomb interactions @xmath27 , where systems have a mott - insulating phase at @xmath84 . in fig .
5 we present numerical results for the variance of the local density obtained within the falicov - kimball model with a harmonic potential as a function of the local density calculated for various particle fillings and various values of @xmath27 and @xmath28 .
one can see that all numerical data for the variance of the local density collapse on the same curve ( given by @xmath85 for @xmath86 and @xmath87 for @xmath88 ) .
thus in contrast to the similar studies on the hubbard model @xcite we have found that the variance of the local density of systems described by falicov - kimball model with a harmonic potential exhibits universality not only with respect to the confining potential but also with respect to the coulomb interaction @xmath27 . in addition , we have found that the universality hold not only for @xmath83 , but for all values of @xmath64 .
a similar universal behaviour has been observed also for other local quantities .
for example , the local double occupation @xmath89 for @xmath90 and @xmath91 for @xmath92 , independent of values of @xmath93 and @xmath10 . also the universal behavior is found for the local compresibility @xmath45 when @xmath83 .
however , the value of the critical exponent for the falicov - kimball model is equal to 1 , unlike the nontrivial value 0.68 - 0.78 found for the hubbard model @xcite . in summary
, we have studied the ground - state properties of fermionic mixtures with mass imbalance in a one - dimensional optical lattice within the spinless falicov - kimball model with a harmonic potential .
we have found that the system exhibits the phase separation at low particle fillings . in this case
the heavy atoms occupy the center of the trap while the light atoms are localized in the surrounding ( metallic ) regions . at higher fillings
we have observed a formation of mott - insulating domains with atomic - density waves . in all cases
mott - insulating phases coexist with metallic phases .
one of the most interesting results is , however , the observation of the universal behavior of the variance of the local density ( independent of the particle filling , the coulomb interaction and the strength of the confining potential ) over the whole region of the local density values .
m. greiner et al . , nature ( london ) * 415 * , 39 ( 2002 ) ; i. bloch , phys .
world * 17 * , 25 ( 2004 ) ; d. jacksch and p. zoller , ann.phys .
( n.y . ) * 315 * , 52 ( 2005 ) ; w. hofstetter , philos . mag . * 86 * , 1891 ( 2006 ) ; t. stoferle et al .
. lett . * 96 * , 030401 ( 2006 ) .
p. lemberger , j. phys .
a * 25 * , 715 ( 1992 ) ; c. gruber and d. ueltschi and j. jedrzejewski , j. stat .
phys . * 76 * , 125 ( 1994 ) ; j.k .
freericks , ch . gruber and n. macris , phys .
b * 53 * , 16189 ( 1996 ) ; j.k .
freericks , e.h .
lieb and d. ueltschi , phys .
lett . * 88 * , 106401 ( 2002 ) .
1 . profiles of various local quantities along the trap calculated for @xmath94 and four different values of particle fillings .
different panels ( from top to bottom ) correspond to : the local density of heavy atoms @xmath95 , the local density of light atoms @xmath96 , the total site occupation @xmath97 , the local double occupation @xmath98 , the variance of the local density @xmath35 and the local compresibility @xmath66 . fig . 3 .
the site occupation @xmath64 as a function of site position @xmath9 calculated for extrapolated ground states for different particle fillings
. the parameters involved are @xmath99 and @xmath76 .
the insets show details of @xmath64 profiles for the region of atomic - density waves .
the exact - diagonalization results ( the asymmetric hubbard model with a harmonic trap ) for the site occupation @xmath64 as a function of site position @xmath9 calculated for different values of @xmath100 and @xmath28 on a finite cluster of @xmath82 sites .
the variance of the local density @xmath101 as a function of the local density @xmath102 calculated for different values of @xmath103 and @xmath62 .
the numerical data are fitted by @xmath85 for @xmath86 and @xmath87 for @xmath88 . | ground - state properties of fermionic mixtures confined in a one - dimensional optical lattice are studied numerically within the spinless falicov - kimball model with a harmonic trap .
a number of remarkable results are found .
( i ) at low particle filling the system exhibits the phase separation with heavy atoms in the center of the trap and light atoms in the surrounding regions .
( ii ) mott - insulating phases always coexist with metallic phases .
( iii ) atomic - density waves are observed in the insulating regions for all particle fillings near half - filled lattice case .
( iv ) the variance of the local density exhibits the universal behavior ( independent of the particle filling , the coulomb interaction and the strength of a confining potential ) over the whole region of the local density values . |
measurements on @xmath0 colliding systems at rhic provide a crucial baseline measurement with which to compare measurements from au+au collisions .
these comparisons offer a means to distinguish nuclear medium effects from what might be expected in a superposition of @xmath4 nucleon+nucleon collisions .
however , @xmath0 systems are interesting in their own right .
previous measurements have shown a systematic dependence of the @xmath1 on particle mass as well as event multiplicity .
measurements from ua1@xcite of the @xmath1 of @xmath5 , @xmath6 , and charged hadrons as a function of the @xmath7 from @xmath8 collisions at @xmath2=630 gev show a clear increasing trend with @xmath7 and this trend is stronger for the @xmath5 than for the @xmath6 or the pion - dominated charged inclusives . lower energy isr data ( @xmath923 gev 63 gev)@xcite
seem to also show a mass dependence for @xmath1 which has been parameterized@xcite with the form @xmath10 .
an increase in the @xmath1 with particle mass should also be reflected in the inverse - slope parameter .
model calculations that include the effects of mini - jets " ( multi - parton interactions ) show such a dependence of the inverse - slope parameter on particle mass@xcite while hydro calculations show only a weak dependence on mass at best
. identified particle spectra ( @xmath11 , k@xmath12 , p & @xmath13 ) at isr energies have been successfully described@xcite by a universal " parameterization of the form shown in equation [ equ : univ ] @xmath14 where @xmath3 is the transverse mass ( @xmath15 ) .
this phenomenon is known as @xmath3scaling and would seem to suggest that over a given range of @xmath3 , spectra of particles with different masses would have similar slopes .
also , even if the spectra have no relative scaling factors applied , the yields will be similar at a given @xmath3 .
it has been suggested that in au+au collisions at sufficiently high energy the parton density increases to the point of saturation , forming a state referred to as the color glass condensate " ( cgc)@xcite . the intrinsic momentum scale of the cgc allows a single function to describe produced particle spectra over a range of energy , centrality , and system size .
this idea has been tested on data from @xmath0 at the cern sps and , more recently , on minimum bias @xmath16 gev au+au data from the phenix collaboration at rhic@xcite .
star is a multi - system experiment the heart of which is the large volume time projection chamber ( tpc ) .
the tpc provides excellent tracking of charged particles with good momentum resolution ( 2 - 7% for @xmath17 s out to a @xmath18 of 4 gev/@xmath19 ) over a broad rapidity window ( 1.8 units in @xmath20 ) .
particle identification is done via @xmath21 and , more recently , with a time of flight ( tof ) system@xcite .
the full details of the star experiment subsystems are available elsewhere@xcite .
identification of strange particle decays in star is done via a topological algorithm in which several geometrical cuts are applied to the charged particle tracks . once identification is done and the parent momenta determined , the resulting spectra must be corrected for finite detector acceptance and efficiency . for this
we use a technique known as embedding " . in embedding , monte - carlo generated tracks are propagated through a full geant@xcite simulation of the star detector response and then added at the pixel level to signals from real data events . the embedded event is then processed with our standard event reconstruction software and a search is made to see if the embedded particle s track is reconstructed . in this way we obtain correction factors which account for both efficiency and acceptance losses , as a function of @xmath18 , in a single procedure .
star has measured the transverse momentum spectra for particles with masses up to and including the @xmath22 baryon in @xmath0 systems with @xmath23 gev .
present statistics are insufficient for an @xmath24 spectrum in @xmath0 .
the @xmath18 ranges of the measured spectra depend on the particle species due to the differing means of particle identification .
pion , kaon , and proton spectra from the tpc have the greatest precision at low @xmath18 but are the most limited in extent ( out to @xmath251 gev/@xmath19 in total momentum ) due to the overlapping of the bethe - bloch bands .
the tof increases the upper limit to about 3 gev/@xmath19 .
strange particles decays can be reconstructed over a wide range of transverse momentum @xmath26 but are limited by the available statistics .
figure [ fig : ppxi ] shows the mid - rapidity @xmath18 spectra for @xmath27 and anti-@xmath27 baryons . and
@xmath28 from @xmath0 at @xmath23 gev , @xmath29 .
error bars are statistical only , width=245 ] the fits are exponentials in @xmath3 of the form : @xmath30 which describes the data reasonably well at low @xmath18 .
but we already begin to see the deviation of the power - law - like tail above @xmath254 gev/@xmath19 .
while the statistics do not allow binning in event multiplicity for the @xmath27 s , the situation is quite different for the @xmath6 and @xmath5@xcite . from the fits to the @xmath18 spectra
we may determine the @xmath1 in each multiplicity bin .
the results of this determination can be seen in the figure [ fig : lamk0meanpt ] below .
note the suppressed zero on the vertical scale . and
@xmath31 on uncorrected @xmath32 , @xmath33 .
error bars are statistical only.,width=245 ] both species show a clear rising trend with uncorrected @xmath32 , though the trend seems to be slightly stronger for the @xmath5 s .
both @xmath6 and @xmath5 production seems to be biased more toward the harder collisions that are indicated by the higher uncorrected @xmath32 .
we present in the figure [ fig : meanptvsmass ] the trend of the @xmath1 with particle mass for the 200 gev @xmath0 and central au+au systems .
vs. particle mass for @xmath0 at @xmath23 gev and au+au at @xmath34 gev.,width=245 ] the black line is from the @xmath10 parameterization of the isr data which included masses only up to the @xmath5 .
the difference between the au+au and @xmath0 @xmath1 for each particle is believed to be due mostly to the presence of radial flow@xcite in the au+au system . this effect is weaker for the more massive particles .
although the isr parameterization works well for the lower mass particles despite the large difference in beam energy , the deviations become significant for the higher masses where the increase in @xmath1 is nearly linear with mass .
figure [ fig : ppnoscale ] shows the spectra with no scaling factors applied .
the difference in yields at any given @xmath3 shows that @xmath3scaling does not work in the absolute sense and can only be approximate at best for our @xmath0 data .
taking the @xmath3 spectra for the various species from the @xmath0 and scaling them by arbitrary factors it seems likely that they would be describable by a single function .
gev @xmath0 collisions with no relative scaling factors applied .
error bars are statistical only.,width=245 ] the spectra in figure [ fig : pplowptscaled ] have been scaled by arbitrary factors to achieve the best overlap and were fit with a power law of the form : @xmath35^n } \label{equ : pwrlaw}\ ] ] figure [ fig : ppmtratio ] is the point - by - point ratio of the data to the fit .
the fit is good to within 20% out to almost 3 gev/@xmath19 .
adding in the data from the tof as well as the @xmath6 data , but retaining the fit to the spectra from figure [ fig : pplowptscaled ] we see in figure [ fig : ratiowithtof ] that the meson spectra trend away from the apparent curve established by the previous fit .
the trend becomes apparent due to the extended @xmath18 range availible in the tof and @xmath6 spectra .
gev @xmath0 collisions including time - of - flight and @xmath6 .
the tof spectra provide a greater range in @xmath18 than those from @xmath21 alone .
error bars are statistical only.,width=245 ] the onset of deviation from the original fit at @xmath361 gev/@xmath19 is interesting in that the hard scatterings are beginning to become an important contribution to the spectra in this region . the baryons seem to maintain agreement further out in @xmath3 and the delayed departure of the pions could be due to contributions feeding - down from higher mass decays .
when scaling factors are applied to the au+au @xmath3 spectra no absolute scaling is evident and only very little overlap can be achieved .
figure [ fig : scaledauau ] shows the scaled au+au spectra .
gev au+au collisions .
error bars are statistical only.,width=245 ] deviations from the universal fit in @xmath0 and au+au are very different in nature . in au+au the likely presence of radial flow changes the spectral shapes at low-@xmath3 eliminating the universal curve that was present in the @xmath0 spectra .
the au+au spectra are also different in shape , being flatter than those from p+p .
we have presented preliminary results from analyses of the @xmath18 spectra of @xmath11 , @xmath37 , p , @xmath13 , @xmath6 , @xmath5 , and @xmath27 from @xmath23 gev @xmath0 collisions and @xmath11 , @xmath37 , p , @xmath13 , @xmath6 , @xmath5 , @xmath27 , and @xmath38 from @xmath34 gev au+au collisions measured by the star collaboration at rhic .
the measured @xmath1 for @xmath6 and @xmath5 show a clear dependence on the uncorrected average charged particle multiplicity per unit @xmath20 in the @xmath0 collisions .
the dependence is slightly stronger for the @xmath39 than for the @xmath31 , while production of both species seems to be more probable in the harder , higher multiplicity collisions .
the mass dependence of the @xmath1 agrees well with the @xmath10 parameterization of the lower energy isr results for masses up to about the @xmath40 mass .
again , it should be noted that the isr measurements only included masses below the @xmath5 .
the strong trend of the @xmath1 with particle mass could be an indication of the increasing importance of the mini - jet contribution to the @xmath18 spectra at rhic energies as mentioned in @xcite .
the transverse mass spectra from the @xmath0 collisions are seen to obey an approximate @xmath3scaling in the low @xmath18 region but the greater range of the tof spectra show the scaling breaks down above @xmath251 gev in @xmath3 for mesons but seems to hold slightly further out in @xmath18 for baryons .
the au+au data show a different behavior from the @xmath0 in that the radial flow , which is believed to be present in the au+au system , depletes the low-@xmath18 region producing a flatter , hotter spectrum .
the low-@xmath18 region then can not be used for the scaling normalization , in effect breaking the approximate scaling observed in @xmath0 .
we wish to thank the rhic operations group and the rhic computing facility at brookhaven national laboratory , and the national energy research scientific computing center at lawrence berkeley national laboratory for their support .
this work was supported by the division of nuclear physics and the division of high energy physics of the office of science of the u.s .
department of energy , the united states national science foundation , the bundesministerium fuer bildung und forschung of germany , the institut national de la physique nucleaire et de la physique des particules of france , the united kingdom engineering and physical sciences research council , fundacao de amparo a pesquisa do estado de sao paulo , brazil , the russian ministry of science and technology , the ministry of education of china , the national natural science foundation of china , and the swiss national science foundation .
this work has been partially supported by nsf grant phy-03 - 11859 .
99 j. rafelski and b. mueller , phys .
* 48 * ( 1982 ) 1066 .
g. bocquet et al . , phys . lett .
* b366 * ( 1996 ) 441 .
b. alper et al .
b100 * ( 1975 ) 237 . m. bourquin and j. m. gaillard , nucl . phys .
* b114 * ( 1976 ) 334 .
a. dumitru et al . , phys .
b446 * ( 1999 ) 326 .
g. gatoff and c. y. wong , phys . rev .
* d46 * ( 1992 ) 997 .
l. mclerran and r. venugopalan , phys . rev .
* d49 * ( 1994 ) 2233 l. mclerran and j. schaffner - bielich , phys . lett .
* b514 * ( 2001 ) 29 j. schaffner - bielich et al .
, nucl . phys . *
a705 * ( 2002 ) 494 .
j. schaffner - bielich et al .
, arxiv nucl - th/0202054 w. j. llope et al . , arxiv nucl - ex/0308022 nucl
a 499 ( 2003 ) 624 http://wwwasd.web.cern.ch/wwwasd/geant/ j. adams and m. heinz , arxiv nucl - ex/0403020 j. adams et al .
, arxiv nucl - ex/0310004 | an enhancement in the number of strange particles produced in relativistic heavy ion collisions is expected to coincide with the formation of a deconfined state of partonic matter@xcite .
measurements of transverse momentum spectra for strange particles emerging from @xmath0 collisions are used as a baseline to which similar measurements from heavy ion collisions are compared .
in addition , several observations from @xmath0 collisions , such as the variation of @xmath1 with particle mass and with event multiplicity , are interesting in their own right .
we present measurements of the transverse momentum spectra and @xmath1 systematics for strange and non - strange particles from @xmath0 collisions at @xmath2=200 gev .
we show the dependence of the @xmath1 on measured charged multiplicity and on particle mass .
we will also demonstrate the ability to scale the transverse mass spectra of various species onto a single universal curve for our @xmath0 data ( an effect known as @xmath3scaling ) and the failure of this scaling when applied to our au+au data .
the work presented here was presented as a poster at quark matter 2004 . |
the use of magnetic materials in plasmonic structures allows the coupling between plasmonic and magnetic states .
for example , recent results in au / ferromagnetic layer / au heterostructures revealed the ability to control the propagation of surface plasmon polaritons ( spps ) at the metal - ferromagnetic interface , using an applied external magnetic field @xcite .
although the exploration of the effect on the magneto - optic ( mo ) activity in magneto - plasmonic structures has just started , the potential for utilization in optoelectronics @xcite , bio - sensors @xcite , and magnetic applications @xcite are already apparent .
the discovery of the extraordinary optical transmission through sub - wavelength hole arrays @xcite started an intensive research effort on the optical response of plasmonic structures .
the investigations of spps have almost exclusively been limited to metal - dielectric systems , where the metal is non - magnetic and highly conductive ( ag , au , al ) @xcite . in the case of highly absorbing magnetic metals , such as fe , co , and ni , spps undergo a strong internal damping , weakening the optical activity .
recently , however , plasmonic states in pure ferromagnetic patterned ni dots were revealed @xcite , while the generation of spps in perforated ferromagnetic metal ( fe , co ) nano - structures were demonstrated and have been used to enhance mo surface effects @xcite .
an enhancement of mo kerr rotation was found in hexagonal arrays of ni nanowires embedded in an alumina matrix @xcite due to a surface plasmon resonance of the ni nanowires .
ni gratings coated with ag @xcite also showed enhanced mo activity . however ,
there are still many unresolved questions concerning the basic principles of the underlying effects and the root of the mo enhancement arising from the generation of surface plasmons at a magnetic metal / air interface must be considered as an open scientific question .
here we focus on the effect of plasmonic states in an anti - dot structure of ferromagnetic material ( ni ) .
we examine the possibility to excite spps on the patterned surfaces and we explore the angular dependence effects , i.e. the orientation of the pattern with respect to the scattering plane .
the angular and energy dependent evolution of spps shows how the interactions of plasmonic excitation couple with the different length scales of the hexagonal pattern .
finally , we show to what extend these spps do influence the magneto - optic response .
the ni antidot arrays were prepared on si(111 ) substrates , using self - assembly nano - sphere lithography with polystyrene ( ps ) spheres of @xmath0 nm in diameter @xcite .
after the self - assembly , the diameter of the spheres was reduced to @xmath1 nm , using reactive ion etching .
these spheres were used as shadow masks , while growing a patterned ni , directly on the substrate .
the deposition started with a 2 nm thick ti layer , followed by a 100 nm thick ni film , and a 2 nm thick au layer on top .
the ti layer was used to enhance the adhesion to the substrate while the au is used to protect the surface of the film .
the self - passivation of ni oxide prevents the degradation of the structure , which is of importance for the bare ni surface at the rim of the holes .
ni can be regarded as a material of choice when making patterned magnetic structures , since the self - passivating oxide is very thin .
the nio layer reaches a maximum thickness of 1 nm at room temperature @xcite .
after the deposition , the ps mask was dissolved , resulting in a hexagonal anti - dot structure as illustrated in fig .
the ratio of the radius ( @xmath2)/pitch size ( @xmath3 ) was determined to be @xmath4 , while the fraction of the holes represent 31 % of the total surface .
a continuous film of the same structure was also fabricated as a reference sample .
ultraviolet ( uv)-visible reflectivity spectra were recorded at room temperature using a perkin elmer l-35 uv - visible spectrometer , in the energy ( wavelength ) range @xmath5 ev ( @xmath6 nm ) .
a 50 mm integrating sphere was used for the acquisition of the reflectivity curve .
the incident light had thereby an angle of 8with respect to the normal of the sample surface .
the magneto - optic spectra were recorded using a magneto - optic kerr spectrometer @xcite operating in a polar configuration .
the range in recorded energies ( wavelengths ) was @xmath7 ev ( @xmath8 nm ) .
the kerr spectra were obtained in the saturation state of the samples at a magnetic field of 1.64 t. angular dependent @xmath9-order reflectivity curves were obtained using a dedicated optical diffractometer based on a huber goniometer with a mc 9300 controller .
this optical diffractometer consists of six motorized stages for the angular ( step resolution of 1 millidegree ) and positional control , monochromatic laser light sources of different wavelengths ( 408 , 535 , 660 and 787 nm ) , and glan - taylor polarizers .
the scattering geometry is defined in fig .
[ scattering ] : @xmath10 is the angle of incident light while @xmath11 is the angular position of the detector . @xmath12
is defined as the azimuthal angle between the y - axis ( defined as one of the high symmetry directions of the hexagonal lattice as shown in fig .
[ sem ] ) and the scattering plane .
all the results presented here are with @xmath10=@xmath11=@xmath13 , often referred to as @xmath14 scans .
we used a scattering matrix method ( smm ) , specially developed to include mo effects @xcite in periodic media , when calculating the optical and mo response of the structures .
the optical and mo constants of ni were taken from ref . @xcite and those for si from ref @xcite .
the values for au and ti were obtained from ellipsometry measurements of 20 nm thick films .
[ fig : fig3 ] ( a ) shows the reflectivity spectra as a function of photon energy for the hexagonal anti - dot array sample together with the continuous ni film . as shown in the figure , the overall reflectivity of the patterned sample is lower than that of the continuous film .
pronounced minima are observed in the reflectivity from the patterned sample , the most distinctive one at a photon energy ( wavelength ) @xmath15 ev ( @xmath16 nm ) and a broader one from around @xmath17 ev ( @xmath18 nm ) to @xmath19 ev ( @xmath20 nm ) .
these are signatures of spps excitations , which are directly connected to the symmetry of the lattice and the optical constants of ni , as discussed below .
+ spps can be resonantly excited by the coupling between free surface charges ( electrons ) of the metal and the incident electromagnetic field .
the real part of the dispersion relation of spps propagating at the interface with a frequency - dependent spp wave - vector , @xmath21 can be obtained from maxwell s equations , using the appropriate boundary conditions @xcite , @xmath22 where @xmath23 is the frequency - dependent permittivity of the dielectric material and @xmath24 the real part of the frequency - dependent permittivity of the metal .
these must have opposite signs , if spps are to be supported at the interface , condition that is satisfied for metals .
for ni at @xmath25 nm ( wavelength that is most used in this work ) , the spp wavevector for the ni - air interface assuming optical constants of ni @xcite ( @xmath26 ) , is found to be @xmath27 using eq .
[ eq : dispersion ] .
coupling photons of a given energy with a grating structure ( here the @xmath28dimensional ( 2d ) hexagonal periodic array ) that provides some additional momentum , it is possible to excite surface plasmons according to the dispersion relation @xcite : @xmath29 where @xmath30 is the surface plasmon polariton wave vector , @xmath31 = @xmath32 is the component of the incident wave vector that lies in the plane of the sample , @xmath33 , and @xmath34 are the basis vectors of the reciprocal hexagonal lattice and i , j are integers .
the in - plane wave vector @xmath31 of the incident light is a function of the incident angle @xmath35 and the azimuthal angle @xmath36 between the plane of incidence and a specific lattice orientation ( see fig .
[ fig : fig1 ] , [ fig : fig2 ] ) and is given by : @xmath37 the reciprocal vectors @xmath33 , and @xmath34 ( fig .
[ fig : fig1 ] inset ) are inversely proportional to the pitch size @xmath38 , with a magnitude of @xmath39 = @xmath40 = @xmath41 .
the reciprocal lattice is rotated @xmath42with respect to the real lattice as it is shown in fig .
[ fig : fig1 ] where also the corresponding angles used in the experiment are presented .
the integers i , j specify the orders of the spps resonances . for a hexagonal lattice , when the component of the incident wave vector @xmath31 is along @xmath33 , the conservation of energy in eq .
[ eq : matching ] can be written as : @xmath43^{1/2}\end{aligned}\ ] ] combining eq.([eq : dispersion ] ) and ( [ eq : matchinghexagonal ] ) we find that the angular dependent dispersion relation for an hexagonal lattice along the @xmath33 direction is given by : @xmath44 similar calculations can be performed when the component of the incident wave vector , @xmath45 is along the @xmath33 + @xmath46 direction .
it is worth to notice that eq .
[ eq : angulardependent ] is derived based only on geometrical considerations and does not take into account the presence and the size of the holes and the subsequent losses and interference effects
. however , it serves as a guideline for the expected energy and angular dependence of spps . for the case of normal incidence and from eq .
[ eq : angulardependent ] the spps resonances are expected at wavelengths @xcite : @xmath47 we can now calculate the expected positions of the reflectivity minima for an angle of incidence of 0 , using eq .
[ eq : normalincidence ] .
the expected positions of the minima from a ni hexagonal pattern are indicated as vertical lines in fig .
[ fig : fig3 ] . as seen in the figure , the positions of the two first minima are captured , while the expected higher order minima are not clearly captured in the experimental data .
furthermore , these calculations do not provide any quantitative information on the shape of the reflectivity .
therefore , we performed quantitative calculation of the reflectivity , using the scattering matrix method ( smm)@xcite .
the results are presented as a continuous line in fig .
[ fig : fig3 ] ( a ) .
the simulated curve contains the main experimental features of the reflectivity : two characteristic troughs in the energy range of @xmath48 ev ( @xmath49 nm ) .
the second , broader , dip consists of three unresolved plasmonic resonances ( see eq .
[ eq : normalincidence ] ) which are not resolved due to the width of the resonances , giving rise to an overlap .
the energy dependence of the polar kerr rotation of the patterned sample is illustrated in fig .
[ fig : fig3](b ) , together with the results from the 100 nm thick reference layer ( continuous ni film ) .
as expected , the reference spectrum corresponds to a typical mo spectrum of a ni film @xcite with respect to the shape and the value of the kerr rotation .
two distinct minima at 1.6 ev and at 3.0 ev can be identified .
it also shows a change of sign , near the photon energies of 1 and 4 ev .
ni has the lowest mo activity compared to co and fe , corresponding roughly to the relative magnetization of these materials .
ab - initio and numerical calculations @xcite revealed a dependence of the polar kerr rotation on the inter - band transitions ( important for energies smaller than @xmath50 ev ) in combination with the spin - orbit interaction and spin - orbit coupling strength @xcite . however , the differences in the mo activity of ni , as compared to co and fe , are not fully understood microscopically @xcite .
the spectrum for the patterned sample shows a completely different behavior , while certain aspect of the reflectivity curve can be identified ( see fig .
[ fig : fig3](a ) ) .
a sharp trough occurs around @xmath51 ( @xmath52 ) ev ( nm ) , which coincides with a minimum in the reflectivity curve at @xmath53 ev ( nm ) . at this energy ,
the kerr rotation is four times larger than obtained from a continuous film .
this enhancement takes place at an energy defined by the periodicity of the structure ( here @xmath54 nm ) @xcite .
a second minimum is obtained around @xmath55ev(nm ) which corresponds to the second trough in reflectivity at the calculated resonance of @xmath56 ev(nm ) .
the kerr rotation at this energy is 2.5 times bigger than obtained from the continuous film .
a change in sign is observed around 3.8 ev , which resembles the changes obtained from the continuous layer , while the size of the kerr rotation is almost doubled at the photon energy of @xmath57 ev ( nm ) .
this enhancement is also correlated to the broad minimum in the reflectivity and the two calculated plasmonic resonances after @xmath58 ev(nm ) . at higher energies
the mo activity decreases , approaching the one for the continuous film .
the strong enhancement for the mo signal is a result of the creation of spps resonances along different crystallographic orientations of the hexagonal pattern .
it is known that at frequencies near the bulk plasma resonance the reflectivity shows a minimum and the mo - kerr signal is enhanced @xcite .
however in this case the spps represent another type of electronic resonances that propagates along the surface and influence the mo response .
in particular , we attribute the magneto - optic enhancement to the interplay between optical and magneto - optical effects .
the magneto - optical kerr effect has its origin in intrinsic magneto - optically active electronic transitions and is influenced by optical enhancement effects .
the reduction of the reflectivity through the excitation of surface plasmons gives rise to such an optical enhancement .
simultaneously the surface plasmons intensify the electromagnetic field sensed by the ni film .
this results in an increase of the polarization conversion ( or p - s light conversion ) due to the magnetization of the material and to the presence of the magnetic field .
the appearance of these two factors simultaneously ( reduced reflectivity and enhanced polarization conversion ) leads to the enhanced kerr rotation .
the aforementioned mechanism applies in several types of metal - ferromagnet - metal structures ( look for example ref . @xcite and references therein ) .
the calculated mo spectrum of the patterned sample is shown as a continuous line in fig .
[ fig : fig3 ] ( b ) .
the calculations are based on the smm code , specially developed to be able to include mo effects . these calculation confirms and strengthen the obtained correspondence between the changes in reflectivity and mo activity .
the qualitative and quantitive agreement between theory and experiment clearly demonstrate the possibility to control and tailor the plasmonic and magneto - plasmonic resonances .
to demonstrate the possibility to manipulate the mo activity of these samples through the spp , we now discuss the angular dependence of the reflectivity for @xmath59 and @xmath60 incident polarizations , in the two main crystallographic directions of the 2d anti - dot pattern ( @xmath61 , @xmath62 ) .
the specular reflectivity curves of the continuous and patterned ni structures are presented in fig .
[ fig : fig4 ] .
the incident light is either @xmath59 and @xmath60 polarized with a wavelength of @xmath63 nm . at a single metal - dielectric interface , no spps can be created for the case of @xmath59 incident radiation ( e vector has only one component , e@xmath64 , tangential to the interface ) @xcite .
no significant change is therefore observed when changing the angle of incidence @xmath13 .
however , a small feature appears in the direction @xmath65 . a @xmath60 component from the incident @xmath59polarization , reflection coefficient @xmath66 ,
can be induced magneto - optically since ni is ferromagnetic and can give rise to the small resonant feature which is seen in fig .
[ fig : fig4 ] ( a ) .
the results for @xmath60 polarized light are shown in fig .
[ fig : fig4 ] ( b ) .
a remarkable drop in reflectivity around @xmath67is generated on the ferromagnetic anti - dot array for @xmath65 .
the trough demonstrates the excitation of spps . by varying the crystallographic direction @xmath36 of the 2d antidot pattern we can tune the spps resonances by changing the incident angle @xmath13 .
the angle @xmath61corresponds to the high symmetry direction @xmath68 in the reciprocal space ( see fig . [
fig : fig1],[fig : fig2 ] ) .
the angle of 34.9agrees well with the calculated value from eq .
[ eq : angulardependent ] , for the case of the ( i , j)=(1,0 ) plasmonic mode at a ni / air interface .
ni optical constants were taken for the corresponding wavelength @xmath63 nm @xcite . for @xmath69(@xmath70 direction ) the resonance moves to @xmath71with significantly larger decrease in the reflectivity .
this behavior can be understood , when considering the length - scales probed when using these scattering geometries . at @xmath61
, the real space spacing is smaller than the one corresponding to @xmath72(see fig .
[ sem ] ) .
the resonances will therefore appear at smaller @xmath13 when @xmath61 , as compared to @xmath72(notice the difference in the optical and x - ray reflectivity conventions ) .
consequently , the interplay between symmetry of the hexagonal lattice and the plasmon excitation is demonstrated in fig .
[ fig : fig4 ] .
the coupling strength of the incoming light to the hexagonal lattice depends on the incident angle , influencing the generation of plasmon resonances .
the inset in fig .
[ fig : fig4 ] ( a ) and ( b ) contain the calculated reflectivity curves for the range 0 to 90 in @xmath13 .
the experimental and theoretical results are qualitative and partially quantitatively in agreement .
the largest discrepancy is in the relative change of the reflectivity , which can be attributed to both missing footprint corrections in the experiments and the exclusion of edge effects in the calculations .
when approaching the glancing incidence ( @xmath73 ) the role of the hole side - walls becomes significant , which will reduce the measured reflectivity .
full calculation of the angular and energy dependent reflectivity ( @xmath68 direction of the reciprocal lattice ) corresponding to @xmath65 , is presented in fig .
[ fig : fig5 ] ( a ) .
the angular dependence of the minima in reflectivity ( when varying the incident wave vector @xmath74 ) , yields a shift of the minimum , to higher @xmath13 angles as the energy is decreased .
the experimental results are displayed in fig .
[ fig : fig5 ] ( b ) .
these results were obtained at a constant energy , marked with dotted lines in fig .
[ fig : fig5 ] ( a ) .
the agreement between the experimental and calculated results is good , both with respect to the expected position of the minimum as well as the overall changes in the reflectivity . at 1.88 and 2.32 ev ,
we notice a local maxima in the intensity on the high angle side after the minima of reflectivity , which are identified as wood s anomalies @xcite .
the magnetic field is expected to modify the surface plasmon state in similar structures composed of metallic films @xcite where the applied magnetic field is a second order term in the off diagonal elements of the electric permitivity tensor .
whether similar behaviour is present in our case , where we have a magnetic metal with a smaller corresponding permitivity element ( for the case of ni : @xmath75 ) , needs further future investigations .
the optical and mo response of a hexagonal anti - dot pattern , with a pitch distance of 470 nm and hole diameter @xmath1 , are revealed to exhibit strong coupling between the spps and the mo activity .
the intensity minima in the reflectivity coincides with the excitation of spps , arising from the coupling between the incident light and the 2d hexagonal lattice .
the spectral position of the minima depends on the periodicity of the structures and subsequently on the different diffraction orders of the 2d grating .
large changes in the mo signal are observed at the corresponding minima in the reflectivity , confirming the importance of spps on the magnetic response .
angular and energy dependent measurements of the reflectivity reveal the importance of surface plasmons and the possibility to control the coupling of spps in the magnetic metal by changing the scattering geometry .
the use of theoretical simulations for optimizing the design and resulting properties is shown to be conceivable , which can be of importance for developing new ways to use combined magneto - optic and surface plasmon effects for manipulation of light .
financial support from swedish foundation for international cooperation in research and higher education ( stint ) is gratefully acknowledged .
m . acknowledges funding from eu ( nmp3-sl-2008 - 214107-nanomagma ) , the spanish micinn ( ` magplas ' mat2008 - 06765-c02 - 01/nan and ` funcoat ' consolider ingenio 2010 csd2008 - 00023 ) , and the comunidad de madrid ( microseres - cm s2009/tic-1476 ) .
gc acknowledges financial support by nwo - nano . | the influence of surface plasmons on the magneto - optic activity in a two - dimensional hexagonal array is addressed .
the experiments were performed using hexagonal array of circular holes in a ferromagnetic ni film .
well pronounced troughs are observed in the optical reflectivity , resulting from the presence of surface plasmons .
the surface plasmons are found to strongly enhance the magneto - optic response ( kerr rotation ) , as compared to a continuous film of the same composition .
the influence of the hexagonal symmetry of the pattern on the coupling between the plasmonic excitations is demonstrated , using optical diffraction measurements and theoretical calculations of the magneto - optic and of the angular dependence of the optical activity . 10 v. v. temnov , g. armelles , u. woggon , d. guzatov , a. cebollada , a. garcia - martin , j .-
m .
garcia - martin , t. thomay , a. leitenstorfer , and r. bratschitsch , `` active magneto - plasmonics in hybrid metal - ferromagnet structures , '' nature photon . * 4 * , 107111 ( 2010 ) .
v. i. belotelov , i. a. akimov , m. pohl , v. a. kotov , s. kasture , a. s. vengurlekar , g. a. venu , d. r. yakovlev , a. k. zvezdin , and m. bayer , `` enhanced magneto - optical effects in magnetoplasmonic crystals , '' nature nanotech . * 4 * , 17 ( 2011 ) .
j. f. torrado , j. b. gonzlez - daz , m. u. gonzlez , a. garca - martn , and g. armelles , `` magneto - optical effects in interacting localized and propagating surface plasmon modes , '' opt .
express * 18 * , 1563515642 ( 2010 ) .
b. seplveda , a. calle , l. m. lechuga , and g. armelles , `` highly sensitive detection of biomolecules with the magneto - optic surface - plasmon - resonance sensor , '' opt
. lett . * 31 * , 10851087 ( 2006 ) .
b. c. stipe , t. c. strand , c. c. poon , h. balamane , t. d. boone , j. a. katine , j .- l .
li , v. rawat , h. nemoto , a. hirotsune , o. hellwig , r. ruiz , e. dobisz , d. s. kercher , n. robertson , t. r. albrecht , and b. d. terris , `` magnetic recording at 1.5 pb m-2 using an integrated plasmonic antenna , '' nat photon * 4 * , 484488 ( 2010 ) .
t. w. ebbesen , h. j. lezec , h. f. ghaemi , t. thio , and p. a. wolff , `` extraordinary optical transmission through sub - wavelength hole arrays , '' nature * 391 * , 667669 ( 1998 ) .
h. raether , _ surface plasmons _
( springer - verlag berlin heidelberg , 1988 ) .
j. chen , p. albella , z. pirzadeh , p. alonso - gonzlez , f. huth , s. bonetti , v. bonanni , j. kerman , j. nogus , p. vavassori , a. dmitriev , j. aizpurua , and r. hillenbrand , `` plasmonic nickel nanoantennas , '' small * 7 * , 23412347 ( 2011 ) . e. t. papaioannou , v. kapaklis , p. patoka , m. giersig , p. fumagalli , a. garcia - martin , e. ferreiro - vila , and g. ctistis , `` magneto - optic enhancement and magnetic properties in fe antidot films with hexagonal symmetry , '' phys . rev .
b * 81 * , 054424 ( 2010 ) .
g. ctistis , e. papaioannou , p. patoka , j. gutek , p. fumagalli , and m. giersig , `` optical and magnetic properties of hexagonal arrays of subwavelength holes in optically thin cobalt films , '' nano lett .
* 9 * , 16 ( 2009 ) .
j. f. torrado , e. t. papaioannou , g. ctistis , p. patoka , m. giersig , g. armelles , and a. garcia - martin , `` plasmon induced modification of the transverse magneto - optical response in fe antidot arrays , '' phys .
status solidi ( rrl ) * 4 * , 271273 ( 2010 ) .
j. b. gonzalez - diaz , a. garcia - martin , g. armelles , d. navas , m. vazquez , k. nielsch , r. b. wehrspohn , and u. gsele , `` enhanced magneto - optics and size effects in ferromagnetic nanowire arrays , '' adv . mater . *
19 * , 26432647 ( 2007 ) .
d. m. newman , m. wears , r. j. matelon , and i. r. hooper , `` magneto - optic behaviour in the presence of surface plasmons , '' j. phys . :
condens .
matter * 20 * , 345230 ( 2008 ) .
p. poulopoulos , v. kapaklis , p. e. jonsson , e. t. papaioannou , a. delimitis , s. d. pappas , d. trachylis , and c. politis , `` positive surface and perpendicular magnetic anisotropy in natural nanomorphous ni / nio multilayers , '' appl .
phys .
lett . * 96 * , 202503 ( 2010 ) .
r. saiki , a. kaduwela , m. sagurton , j. osterwalder , d. friedman , c. fadley , and c. brundle , `` x - ray photoelectron diffraction and low - energy electron diffraction study of the interaction of oxygen with the ni(001 ) surface : c(2 2 ) to saturated oxide , '' surf .
sci . *
282 * , 3361 ( 1993 ) .
e. t. papaioannou , m. angelakeris , n. k. flevaris , p. fumagalli , c. mueller , a. troupis , a. spanou , v. karoutsos , p. poulopoulos , v. kapaklis , and c. politis , `` magnetism and magneto - optics of nanocrystalline ni / pt multilayers grown by e - beam evaporation at room temperature , '' j. appl . phys .
* 101 * , 023913 ( 2007 ) .
a. garca - martn , g. armelles , and s. pereira , `` light transport in photonic crystals composed of magneto - optically active materials , '' phys . rev .
b * 71 * , 205116 ( 2005 ) .
s. visnovsk , v. parzek , m. nvlt , p. kielar , v. prosser , and r. krishnan , `` magneto - optical kerr spectra of nickel , '' journal of magnetism and magnetic materials * 127 * , 135 139 ( 1993 ) .
palik , _ handbook of optical constants of solids part ii _ ( academic press , new york , 1985 ) .
w. l. barnes , a. dereux , and t. w. ebbesen , `` surface plasmon subwavelength optics , '' nature * 424 * , 824830 ( 2003 ) .
c. genet and t. w. ebbesen , `` light in tiny holes , '' nature * 445 * , 3946 ( 2007 ) .
a. k. zvezdin and v. a. kotov , _ modern magnetooptics and magnetooptical materials _
( iop publishing , 1997 ) .
p.
m. oppeneer , t. maurer , j. sticht , and j. kbler , `` ab initio calculated magneto - optical kerr effect of ferromagnetic metals : fe and ni , '' phys .
rev .
b * 45 * , 1092410933 ( 1992 ) .
p. fumagalli , c. spaeth , u. rudiger , and r. j. gambino , `` a new magneto - optic enhancement effect in macroscopic ferrimagnets , '' ieee trans . magn . * 31 * , 33193324 ( 1995 ) .
v. i. belotelov , l. l. doskolovich , and a. k. zvezdin , `` extraordinary magneto - optical effects and transmission through metal - dielectric plasmonic systems , '' phys .
rev .
lett . *
98 * , 077401 ( 2007 ) .
s. a. maier , _ plasmonics : fundamentals and applications _
( springer science + business media llc , 2007 ) .
r. w. wood , `` anomalous diffraction gratings , '' phys . rev .
* 48 * , 928936 ( 1935 ) .
y. m. strelniker and d. j. bergman , `` transmittance and transparency of subwavelength - perforated conducting films in the presence of a magnetic field , '' phys .
rev .
b * 77 * , 205113 ( 2008 ) . |
with the release of wmap-3yr temperature @xcite and polarization data @xcite , the standard @xmath4cdm cosmological model has passed another rigorous set of tests , being validated once again @xcite . in this era of precision cosmology
, the early universe is playing an increasingly more important role as a laboratory where physics not accessible on earth can be tested and verified .
thus , cosmology became an excellent tool to answer to many questions of fundamental physics , among which the time variation of fundamental constants of nature , predicted as observational consequences of some currently preferred unification theories @xcite .
one of the most exciting case is the fine structure constant @xmath5 , possible time variations being already reported from geophysical and astronomical observations . from the oklo natural nuclear reactor in gabon ,
the relative time variation of @xmath5 has been obtained as @xmath6 at @xmath7 @xcite .
terrestrial laboratory measurements give the constraint @xmath8 based on comparisons of rates between clocks with different atomic numbers during 140 days of observations @xcite . while the results presented above are only upper limits of @xmath9 , the estimations of its actual value have been firstly computed from spectral analysis of high - redshift quasar absorption systems .
some of the reported values are @xmath10 for a redshift range @xmath11 @xcite , @xmath12 @xcite , these determinations being a confirmation of the expectation that the fine structure constant is a nondecreasing function of time . from this point of view , the result @xmath13 ppm for @xmath14 and @xmath15 @xcite , is controversial .
in fact , there are also many other contradictory results which , perhaps , may be validated if the fine structure constant would not only be time dependent but also had a spatial variation .
the cmb presents the advantage of probing directly the decoupling epoch ( @xmath16 ) when the @xmath5 variations from its current value are expected to be more important and therefore , more easy to test .
a value of @xmath5 different from @xmath17 at the epoch of recombination changes the ionization evolution at that epoch , inducing effects on the cmb anisotropies observed today . in the last years
several groups have analyzed the variation of fine structure constant at recombination using cmb measurements .
a lower value of @xmath5 at recombination epoch compared with its present value , was taken into account as a possibility to solve the disagreement between the flat cosmological model and the bbn predictions on the one side and the first doppler peak position and the secondary peaks amplitudes in cmb power spectra obtained by boomerang and maxima on the other side .
negative values of few percent for @xmath9 have been reported from boomerang and maxima data analysis @xcite and from boomerang , dasi and cobe data analysis @xcite . from the analysis of the first year wmap data ,
a @xmath18 interval @xcite was reported . using the same data togheter with hst hubble key project , but considering simultaneous time variations of both @xmath5 and the electron mass @xmath19 , and using a different analysis method ,
has been obtained the interval @xmath20 when only @xmath5 varies @xcite . in the present work ,
the wmap-3yr cmb data are being analyzed in order to find new limits on the @xmath5 value at recombination , and the corresponding limits on the recombination redshift . in section 2
it is reviewed the standard recombination process and are described the changes involved by a different value of the fine structure constant at recombination .
section 3 presents the results of the analysis of wmap-3yr data in the context of the recombination process affected by time varying fine structure constant . in section 4
are summarized the conclusions of the work .
it is well known the effect of changing the value of the fine structure constant on the energy levels of hydrogen atom and , therefore , on the photon frequencies corresponding to the transitions between two such levels . as a consequence , a change in @xmath5 value will change the dynamics of the hydrogen recombination process in the universe , which affects the cmb fluctuations observed today . as the cmb primary anisotropies probe the recombination epoch , it is natural to consider imposing limits on possible variation in @xmath5 value at the recombination epoch relative to present value using cmb anisotropy data . in the following i
briefly review the hydrogen and helium recombination process and the implications of non - standard value of @xmath5 , neglecting the impact on helium . before recombination
, the photon gas was coupled to electron - baryon fluid through thomson scattering on free electrons , the cross section for this process being given by @xcite @xmath21 the cmb formation followed the recombination process , when the photons became free particles after electrons have been captured by ions . for modeling the recombination process i used the treatment implemented in the recfast code @xcite .
the equations describing the evolution of proton fraction @xmath22 , the singly ionized helium fraction @xmath23 and the matter temperature @xmath24 are : @xmath25,\\ \frac{dx_{heii}}{dz}&=&\frac{c_{he}}{h(z)(1+z)}\nonumber\\ & & \times[x_{heii}x_en_hr_{hei}-\beta_{hei}(f_{he}-x_{heii})e^{-h\nu_{hei}/kt_m}],\\ \frac{dt_m}{dz}&=&\nonumber\\ & & \frac{8\sigma_ta_rt_r^4}{3h(z)(1+z)m_e}\frac{x_e}{1+f_{he}+x_e}(t_m - t_r)+\frac{2t_m}{(1+z)}.\end{aligned}\ ] ] in the above equations , @xmath26 is the hubble expansion rate at redshift @xmath27 , @xmath28 is the planck constant , @xmath29 is the boltzmann constant , @xmath30 is the speed of light , @xmath31 is the blackbody constant , @xmath32 is the hydrogen number density , @xmath33 is the proton fraction , @xmath34 is the singly ionized helium fraction and @xmath35 is the electron fraction .
the number fraction of helium to hydrogen is @xmath36 , where @xmath37 is the primordial helium mass fraction .
the radiation temperature @xmath38 is identical to @xmath24 at high redshift because of the coupling of photons and baryons through thomson scattering .
@xmath39 is the case b recombination coefficient for h , and is given by the fit formula @xmath40 where @xmath41 , @xmath42 , @xmath43 , @xmath44 , @xmath45 @xcite and @xmath46 is the fudge factor @xcite introduced in order to reproduce the results of the multilevel calculation by speeding up recombination in the standard scenario .
@xmath47 is the photoionization coefficient @xmath48 and @xmath49 is the peebles reduction factor @xmath50}{[1+k_h(\lambda_h+\beta_h)n_h(1-x_p)]},\ ] ] which accounts for the presence of non - thermal ly-@xmath5 resonance photons . in the above
, @xmath51ev is the binding energy in the @xmath52 energy level , @xmath53 is the ly-@xmath5 frequency , @xmath54 is the rate of decay of the @xmath52 excited state to the ground state via @xmath55 photons , and @xmath56 . the quantities @xmath57 , @xmath58 , @xmath59 and @xmath60 from eq .
( 3 ) are the analogous for helium of the quantities from eq .
( 2 ) and their expressions may be found , for example , in ichikawa et al .
( 2006 ) . in this work ,
the small effect of changing @xmath5 on helium recombination process has been neglected . in order to take into account the changing value of @xmath5 in the recombination process ,
i have modified the evolution equations for proton fraction and for matter temperature , considering their @xmath5 dependence .
the quantities which depends on @xmath5 in these equations have been taylor expanded up to first order in @xmath9 according to their scaling relations @xcite @xmath61 where @xmath62 was adopted .
in this work , in order to search for new cmb limits on @xmath5 value at recombination , i have analyzed the wmap-3yr cmb anisotropy data @xcite in the framework of the extended cosmological model which includes the variation of the fine structure constant at recombination with respect to its present value .
the data analysis has been done using markov chain monte carlo ( mcmc ) techniques @xcite implemented in the cosmomc code @xcite . for this purpose
i modified the recfast code to compute the hydrogen recombination in the hypothesis of different value of @xmath5 at last scattering .
the relative variation @xmath63 was added as an additional parameter ; the modified equations for the evolution of hydrogen ionization fraction and matter temperature have been integrated with camb code @xcite , used by cosmomc to compute the theoretical cmb power spectra . together with @xmath9 the following cosmological parameters
have been varied : physical density in baryons @xmath64 , physical density in dm @xmath65 , hubble constant @xmath66 , reionization redshift @xmath67 , spectral index @xmath68 and amplitude @xmath69 of primordial fluctuations ; the cosmological constant @xmath70 and the optical depth to reionization @xmath71 have been derived . to realistically constrain the hubble expansion rate taking into account its degeneracy with @xmath9 @xcite , the hst key project prior
@xcite has been used in addition to wmap-3yr data .
the modified version of cosmomc software package has been run on 8 markov chains , using the `` variance of chain means''/``mean of chain variances '' r statistic @xcite as convergence criterion with the choise @xmath72 .
the most likely values of cosmological parameters obtained from mcmc simulations are given in table 1 .
the mean values of the standard cosmological parameters are in the limits reported by wmap team @xcite .
table 1 : @xmath73 with varying @xmath5 model parameters and their 68% confidence intervals obtained from wmap-3yr data . [ cols="^,^,^",options="header " , ] the marginalized distributions of the parameters obtained from simulations are presented in figure 1 .
the observed differences between the curves corresponding to the parameters @xmath66 , @xmath64 , @xmath74 , @xmath70 and the age of the universe cdm model with varying @xmath5 obtained from the analysis of wmap-3yr data and hst key project ( continuous lines ) compared with the distributions of the parameters in the standard @xmath4cdm model ( dashed lines ) from the same data . ] in the @xmath4cdm with varying @xmath5 model compared to those of the standard @xmath4cdm model are due to the degeneracy of these parameters with @xmath5 .
figure 2 presents the 2d marginalized distributions for pairs of parameters at 68% and 95% cl for @xmath4cdm with varying @xmath5 and the @xmath4cdm models .
the negative values for @xmath9 in the @xmath4cdm with varying @xmath5 model are correlated with smaller values of @xmath66 and @xmath64 then in the @xmath4cdm model . on the other side ,
the distributions for @xmath65 are almost the same in both cases , therefore , in the case with varying @xmath5 the parameter @xmath75 tends to go to higher values and @xmath70 to lower values then in the case with no @xmath5 variations .
consequently , the distribution for the age of the universe is widened , with mean value larger then in the case with constant @xmath5 .
figure 3 shows the @xmath76 , @xmath77 and @xmath78 cmb power spectra for the best fit parameters of the model with varying @xmath5 compared with the best fit cmb power spectra for the @xmath4cdm model and the wmap-3yr experimental data .
it is remarkable that the two best fit models are almost identical from the cmb power spectra point of view .
the 95 % confidence interval for the additional parameter obtained from wmap-3yr data is cdm model with varying @xmath5 obtained from the analysis of wmap-3yr data and hst key project ( continuous lines ) compared with the 2d - distributions of the parameters in the standard @xmath4cdm model from the same data ( dashed lines ) .
the contours are at 68% and 95% cl . ]
@xmath79 which is a tighter cmb constraint on the variation of fine structure constant at recombination than previously reported from the analysis of wmap-1yr data @xcite .
the interval for the recombination redshift corresponding to the limits on @xmath9 , @xmath80 ( a ) , @xmath77 ( b ) and @xmath78 ( c ) for the best fit @xmath4cmd model with varying @xmath5 ( continous lines ) and for the the best fit @xmath4cdm model with constant @xmath5 ( computed by the wmap team @xcite ) ( dash - dotted lines ) and the wmap-3yr binned data and their errors.,height=396 ] is tighter compared to those corresponding to limits on @xmath9 from wmap-1yr data analysis @xcite .
the recombination redshift corresponding to the best fit parameters of the model with varying @xmath5 , @xmath81 , indicate a delayed recombination compared with the results from wmap-1yr data .
the relative variation of @xmath5 in redshift unit corresponding to the best fit is @xmath82 or , equivalent , in time unit , @xmath83 both in agreement with the predicted constraints on @xmath5 from cmb @xcite .
based on the analysis of wmap-3yr data with mcmc techniques , i obtained a tighter cmb constraint on the value of fine structure constant at recombination epoch then was previously obtained from wmap-1yr data .
the results confirm the limits predicted to be achieved from cmb . the wmap-3yr constraints on other cosmological parameters like @xmath66 , @xmath64 , @xmath74 and @xmath70 are relaxed because of their degeneracy with the @xmath5 variation , confirming that a non - standard recombination weakens the constraints on other cosmological parameters @xcite . using the constraint on @xmath5 variation ,
i have also obtained the recombination redshift , finding a delay of the recombination epoch , similar to that of the reionization epoch and of the structure formation computed from wmap-3yr versus wmap-1yr data @xcite .
the author thanks l.a .
popa , a. vasile and o.m .
tantareanu for useful discussions and suggestions .
she also acknowledges the cosmic rays and nuclear astrophysics group of iss for providing the computing facilities .
10 avelino , p.p . ,
martins , c.j.a.p . , rocha , g. , viana , p. , 2000 , phys.rev .
d62 , 123508 avelino , p.p . , esposito , s. , mangano , g. , martins , c.j.a.p .
, melchiorri , a. , miele , g. , pisanti , o. , rocha , g. , viana , p. , 2001
, phys.rev .
d64 , 103505 battye , r.a . ,
crittenden , r. & weller , j. , 2001 , phys.rev .
d63 , 043505 bean , r. , melchiorri , a. & silk , j. , 2007 , astro - ph/0701224 brooks , s. & gelman , a. , 1998 , journal of computational and graphical statistics , 7 , 434 damour , t. & dyson , e. , 1996 , nucl . phys .
b 480 , 37 freedman , w.l .
et al . , 2001 , ap.j . , 553,47 hannestad , s. , 1999 , phys .
d60 , 023515 hinshaw , g. , nolta , m.r . ,
bennett , c.l . , bean , r. , dore , o. , greason , m.r . ,
halpern , m. , hill , r.s . ,
jarosik , n. , kogut , a. , komatsu , e. , limon , m. , odegard , n. , meyer , s.s . , page , l. , peiris , h.v .
, spergel , d.n . ,
tucker , g.s . ,
verde , l. , weiland , j.l .
wollack , e. , wright , e.l . , 2006 , ap .
j. submitted , astro - ph/0603451 ichikawa , k. , kanzaki , t. , kawasaki , m. , 2006 , phys.rev .
d74 , 023515 kaplinghat , m. , scherrer , r.j . , & turner , m.s . , 1999 ,
phys . rev .
d 60 , 023515 levshakov , s.a . ,
molaro , p. , lopez , s. , dodorico , s. , centurion , m. , bonifacio , p. , reimers , d. , 2006 , astro - ph/0611803 lewis , a. , challinor , a. & lasenby , a. , 2000 , ap.j .
538 , 473 , http://camb.info lewis , a. & bridle , s. , 2002 , phys .
d 66 , 103511 mackay , d.j.k .
, 2003 , _ information theory , inference , and learning algorithms _ , cambridge university press , http://www.inference.phy.cam.ac.uk/mackay/itila martins , c.j.a.p . , 2002 , phil . trans .
lond .. , a360 , 2681 ,
astro - ph/0205504 page , l. et al .
2006 , ap .
j. submitted , astro - ph/0603450 pquignot , d. , petitjean , p & boisson , c , 1991 , a&a , 251 , 680 popa , l.a . , 2006 , astro - ph/0605358 prestage , d. , toelker , r.l .
& maleki , l. , 1995 , phys .
74 , 3511 rocha , g. , trotta , r. , martins , c.j.a.p . , melchiorri , a. , avelino , p.p . ,
viana , p.t.p . , 2003 ,
47 , 863 rocha , g. , trotta , r. , martins , c.j.a.p . , melchiorri , a. , avelino , p.p . ,
bean , r. , viana , p.t.p .
, 2004 , m.n.r.a.s .
352 , 20 rubino - martin , j.a . ,
rebolo , r. , carreira , p. , cleary , k. , davies , r.d . ,
davis , r.j . ,
dickinson , c. , grainge , k. , gutierrez , c.m . , hobson , m.p . , jones , m.e . , kneissl , r. , lasenby , a. , maisinger , k. , oedman , c. , pooley , g.g . ,
sosa molina , p.j .
, rusholme , b. , saunders , r.d.e . ,
savage , r. , scott , p.f .
, slosar , a. , taylor , a.c . , titterington , d. , waldram , e. , watson , r.a . ,
wilkinson , a. , 2003 , m.n.r.a.s .
, 341 , 1084 seager , s. , sasselov , d.d . & scott , d. , 1999 , ap.j .
523 , l1 , astro - ph/9909275 spergel , d.n . ,
bean , r. , dore , o. , nolta , m.r . ,
bennett , c.l . , hinshaw , g. , jarosik , n. , komatsu , e. , page , l. , peiris , h.v . ,
verde , l. , barnes , c. , halpern , m. , hill , r.s . ,
kogut , a. , limon , m. , meyer , s.s . , odegard , n. , tucker , g.s . ,
weiland , j.l .
, wollack , e. , wright , e.l . , 2006 , ap .
j. submitted , astro - ph/0603449 uzan , j.p .
, 2003 , rev .
, 75 , 403 , hep - ph/0205340 varshalovich , d.a . , potekhin , a.y . & ivanchik , 2001 , a.v.comments on modern physics , 2 , d223-d232 , physics/0004068 webb , j.k . , murphy , m.t . , flambaum , v.v . ,
dzuba , v.a . , barrow , j.d . ,
churchill , c.w . ,
prochaska , j.x . ,
wolfe , a.m. , 2001 , phys .
lett . , 87 , 091301 , astro - ph/0012539 weinberg , s. , 1995 , _ the quantum theory of fields _ , vol . 1 , cambridge university press wmap website - http://lambda.gsfc.nasa.gov | this paper presents the constraints on the time variation of the fine structure constant at recombination relative to its present value , @xmath0 , obtained from the analysis of the wmap-3yr cosmic microwave background ( cmb ) data , with an additional prior on the hubble expansion rate from hst hubble key project .
i found out that @xmath1 at 95% c.l .
, which brings a @xmath2 improvement to the previous limits from wmap-1yr data .
the corresponding recombination redshift , @xmath3 , shows a delayed recombination epoch compared with the results from wmap-1yr data .
+ cosmology : cosmic microwave background , methods : data analysis , atomic processes + |
we examine the riemann - roch theorem on a finite graph @xmath0 , due to baker and norine @xcite , through the lens of combinatorial commutative algebra . throughout this paper , @xmath0 is undirected and connected , has @xmath1 nodes , and multiple edges are allowed , but we do not allow loops .
its laplacian is a symmetric @xmath2-matrix @xmath3 with non - positive integer entries off the diagonal and kernel spanned by @xmath4 .
divisors on @xmath0 are identified with laurent monomials @xmath5 .
the _ chip firing _
moves are binomials @xmath6 where @xmath7 and @xmath8 is in the lattice spanned by the columns of @xmath3 .
the lattice ideal @xmath9 spanned by such binomials is here called the _ toppling ideal _ of the graph @xmath0 .
it was introduced by perkinson , perlman and wilmes @xcite , following an earlier study of the inhomogeneous version of @xmath9 by cori , rossin and salvy @xcite . for any fixed node
, the toppling ideal @xmath9 has a distinguished initial monomial ideal @xmath10 .
this monomial ideal was studied by postnikov and shapiro @xcite , and it is characterized by the property that the standard monomials of @xmath11 are the _ @xmath0-parking functions_. we construct free resolutions for both @xmath9 and @xmath10 , and we study their role for riemann - roch theory on @xmath0 . for an illustration ,
consider the complete graph on four nodes , @xmath12 .
the chip firing moves on @xmath13 are the integer linear combinations of the columns of @xmath14 the toppling ideal is the lattice ideal in @xmath15 = \mathbb{k}[x_1,x_2,x_3,x_4]$ ] that represents @xmath16 : @xmath17 this ideal is _ generic _ in the sense of peeva and sturmfels @xcite , as each of the seven binomials contains all four variables .
the minimal free resolution is given by the _ scarf complex _
@xmath18 \,\longleftarrow \ , \mathbb{k}[{\bf x}]^7 \,\longleftarrow \ , \mathbb{k}[{\bf x}]^{12 } \,\longleftarrow \ , \mathbb{k}[{\bf x}]^6 \,\longleftarrow \ , 0.\ ] ] the seven binomials in ( [ eq : kayfour2 ] ) form a grbner basis of @xmath9 , with the underlined monomials generating the initial ideal @xmath11 . that monomial ideal has the irreducible decomposition @xmath19 the ideal @xmath11 is the _ tree ideal _ of @xcite .
its standard monomials are in bijection with the @xmath20 spanning trees .
its alexander dual is generated by the six socle elements @xmath21 these correspond to the _ maximal parking functions _ studied in combinatorics ; see @xcite .
we claim that the duality seen in figures 4.2 and 4.3 of @xcite is the same as that expressed in the riemann - roch theorem for @xmath0 .
this will be made precise in sections 3 and 4 .
the present article is organized as follows : section 2 is concerned with the case when @xmath0 is a _ saturated graph _ , meaning that any two nodes @xmath22 and @xmath23 are connected by at least one edge .
we show that here @xmath9 is a generic lattice ideal , and we determine its minimal free resolution and its hilbert series in the finest grading .
the scarf complex of the initial monomial ideal @xmath11 is supported on the barycentric subdivision of the @xmath24-simplex @xcite , and this lifts to the scarf complex of the lattice ideal @xmath9 by ( * ? ? ?
* corollary 5.5 ) . in section 3
we revisit the riemann - roch formula @xmath25 we prove this formula in an entirely new setting : the role of the curve is played by a monomial ideal , and that of the divisors @xmath26 and @xmath27 is played by monomials @xmath28 and @xmath29 . the identity ( [ eq : rr ] ) is shown for monomial ideals that are artinian , level , and reflection - invariant .
this includes the parking function ideals @xmath11 derived from saturated graphs @xmath0 .
in section 4 we extend our results to the case of graphs @xmath0 that are not saturated , and we rederive riemann - roch for graphs as a corollary . here
@xmath11 is still an initial ideal of @xmath9 , but the choice of term order is more delicate @xcite .
one choice is the cost function used by baker and shokrieh for the integer program in ( * ? ? ?
* theorem 4.1 ) .
the scarf complexes in section 2 support cellular free resolutions of @xmath9 and @xmath11 , but these resolutions are usually far from minimal .
we conclude with several open questions .
this paper demonstrates how riemann - roch theory embeds into combinatorial commutative algebra .
our main results are theorems [ satminres_theo ] , [ rr_theo ] and [ nonsatinit_theo ] .
these build on earlier works , notably @xcite and @xcite , but they go much further and are new in their current form .
when this collaboration started in the summer of 2011 , both authors were unaware of the articles @xcite written on similar topics by david perkinson and his students at reed college . as our point of departure , we chose to focus on chip firing in the most classical case of undirected graphs , but with the tacit understanding that our ideals and modules generalize to directed graphs , arithmetic graphs , simplicial complexes , matroids , abelian networks , or any of the other extensions seen in the recent chip firing literature ( cf . @xcite ) .
in this section , we assume that the graph @xmath0 has @xmath30 edges between node @xmath22 and node @xmath23 , where @xmath30 is a positive integer , for @xmath31 . however , we do not allow loops , so that @xmath32 .
thus , in the language of @xcite , @xmath0 is a _
saturated graph_. we shall see that , under this hypothesis , the lattice ideal @xmath9 is generic , and an explicit combinatorial description of its minimal free resolution can be given . throughout this paper
we work in the polynomial ring @xmath15 = \mathbb{k}[x_1,\ldots , x_n]$ ] over an arbitrary field @xmath33 .
we begin by explicitly showing the generators of the lattice ideal @xmath9 in the case @xmath34 .
[ ex : fourex ] if @xmath0 is a saturated graph on @xmath35 = \{1,2,3,4\}$ ] then @xmath9 is generated by @xmath36 @xmath37 @xmath38 here the @xmath30 are arbitrary positive integers .
these binomials form a grbner basis .
the initial ideal @xmath11 is generated by the underlined monomials .
the minimal free resolution of @xmath9 has the form ( [ eq : kayfour3 ] ) .
the same holds for @xmath11 , as was shown in ( * ? ? ?
* corollary 6.9 ) .
the minimal resolution of @xmath11 is given by the scarf complex , which is depicted in figure [ fig : eins ] .
we now state our main result in this section . for disjoint subsets @xmath39 and @xmath40 of @xmath41 $ ]
we set @xmath42 a _ split _ of the set @xmath41 = \{1,2,\ldots , n\}$ ] is an unordered pair @xmath43 of non - empty disjoint subsets @xmath39 and @xmath40 whose union equals @xmath41 $ ] .
the number of splits equals @xmath44 . with each split
@xmath43 we associate the following binomial which is well - defined up to sign : @xmath45 these are precisely the seven binomials in example [ ex : fourex ] , one for each split @xmath43 .
let @xmath46 denote the set of cyclically ordered partitions of the set @xmath41 $ ] into @xmath47 blocks .
each element of @xmath46 has the form @xmath48 , where @xmath49 $ ] is a partition .
we regard the @xmath50 as formal symbols , subject to the identifications @xmath51 we write @xmath15^{{\rm cyc}_{n , k}}$ ] for the free @xmath15$]-module generated by these symbols .
the rank of this free module equals the number of cyclically ordered partitions , namely @xmath52 where @xmath53 is the stirling number of the second kind , i.e. , the number of partitions of the set @xmath41 $ ] into @xmath47 blocks .
let @xmath54 denote the following complex of free @xmath15$]-modules : @xmath55^{{\rm cyc}_{n,1 } } \,\longleftarrow\,\mathbb{k}[{\bf x}]^{{\rm cyc}_{n,2 } } \,\longleftarrow\,\mathbb{k}[{\bf x}]^{{\rm cyc}_{n,3 } } \,\longleftarrow \,\,\cdots \
, \,\longleftarrow\,\mathbb{k}[{\bf x}]^{{\rm cyc}_{n , n } } \longleftarrow \ , 0,\ ] ] where the boundary map from @xmath56^{{\rm cyc}_{n , r}}\,$ ] to @xmath56^{{\rm cyc}_{n , r-1}}\,$ ] is given by the formula @xmath57 in this formula it is assumed that @xmath58 , so as to ensure that all signs are consistent .
-simplex supports minimal free resolutions for toppling ideals of saturated graphs with @xmath1 nodes and their initial ideals .
shown here is the scarf complex ( [ eq : kayfour3 ] ) for @xmath34 , with @xmath59 vertices , @xmath60 edges and @xmath61 triangles .
, width=597 ] [ satminres_theo ] let @xmath0 be a saturated graph .
the toppling ideal @xmath9 is a generic lattice ideal .
it is minimally generated by the @xmath62 binomials ( [ eq : ijbinomial ] ) , these form a reverse lexicographic grbner basis , the complex @xmath63 coincides with the scarf complex , and this complex minimally resolves @xmath64/i_g$ ] .
we begin by noting that @xmath65 actually lies in the ideal @xmath9 . to see this ,
let @xmath66 denote the incidence vector in @xmath67 that represents the subset @xmath39 of @xmath41 $ ] .
the @xmath22-th coordinate of the vector @xmath68 is equal to @xmath69 if @xmath70 , and it is @xmath71 if @xmath72 . hence @xmath68 is represented algebraically by @xmath73 , which is hence in @xmath9 .
fix any reverse lexicographic term order on @xmath15 $ ] that has @xmath74 as the smallest variable , and let @xmath75 denote the initial monomial ideal of @xmath9 . since @xmath9 is a lattice ideal , @xmath74 is a non - zerodivisor and it does not divide any of the generators of @xmath75 . we may thus regard @xmath75 as an artinian ideal in @xmath76 = \mathbb{k}[x_1,\ldots , x_{n-1}]$ ] .
the index of the laplacian lattice @xmath77 in its saturation @xmath78 equals the number @xmath79 of spanning trees of @xmath0 .
hence @xmath75 has @xmath79 standard monomials in @xmath76 $ ] .
let @xmath11 denote the ideal generated by the initial monomials of the binomials in ( [ eq : ijbinomial ] ) : @xmath80\backslash i } \,:\
, \hbox{$i$ non - empty subset of $ [ n-1]$ } \,\bigr\rangle.\ ] ] by construction , the inclusion @xmath81 holds .
the monomial ideal @xmath11 was studied in @xcite and shown to have precisely @xmath79 standard monomials . indeed , the standard monomials of @xmath11 are in bijection with the _ @xmath1-reduced divisors_. it is known in the chip firing literature ( cf .
@xcite ) that their number equals the number @xmath79 of spanning trees .
hence @xmath11 and @xmath75 are artinian of the same colength in @xmath76 $ ] , so they must be equal : @xmath82 therefore the binomials ( [ eq : ijbinomial ] ) form a grbner basis , and hence a generating set , of @xmath9 .
the ideal @xmath9 is a generic lattice ideal , in the sense of @xcite , because all @xmath1 variables @xmath83 occur in the binomial ( [ eq : ijbinomial ] ) .
here we are using that @xmath0 is saturated . by (
* theorem 4.2 ) , the scarf complex is the ( essentially unique ) minimal free resolution of @xmath9 .
it remains to be seen that the scarf complex is equal to @xmath84 .
postnikov and shapiro ( * ? ? ?
* corollary 6.9 ) showed that the scarf complex of the initial ideal @xmath11 is supported on the barycentric subdivision of the @xmath24-simplex , as shown in figure [ fig : eins ] .
the scarf resolution has the format ( [ eq : cycres1 ] ) , but with @xmath15 $ ] replaced by @xmath76 $ ] . here
, we label the cells in that barycentric subdivision with ordered partitions @xmath85 satisfying @xmath58 . the boundary maps in the scarf resolution are then given by ( [ eq : cycres2 ] ) , namely , by the sum ranging from @xmath86 to @xmath87 , but without the additional term @xmath88 .
we pass from the scarf resolution of @xmath11 to that of @xmath9 by the combinatorial rule in ( * ? ? ?
* theorem 5.4 ) .
this adds precisely one term to the boundary of each scarf simplex of @xmath11 . in our case ,
that additional term is precisely the one above , and we get ( [ eq : cycres2 ] ) .
[ ex : ressi ] returning to example [ ex : fourex ] , with the seven binomials in that order , here are the matrices over @xmath89 $ ] that represent the first and second syzygies in @xmath90 : @xmath91 @xmath92 note that the seven binomial generators of @xmath9 appear as the @xmath93-minors of the @xmath94-matrices seen in the six pairs of columns within the @xmath95-matrix of first syzygies .
the syzygies of the ideal @xmath11 generated by the underlined monomials in example [ ex : fourex ] are found by replacing with @xmath96 all monomials that have the symbol `` 4 '' to the left of the arrow .
one immediate application of our minimal free resolution is a formula for the hilbert series of the ring @xmath15/i_g$ ] in its natural grading by the group @xmath97 . as is customary in chip firing theory @xcite
, we consider the decomposition @xmath98 where @xmath99 records the degree of a divisor on @xmath0 , and @xmath100 is the finite subgroup of divisors of degree @xmath96 .
the order of @xmath100 is the number of spanning trees of @xmath0 .
let @xmath101 and @xmath102 denote the generators of the group algebra @xmath103 $ ] corresponding to this decomposition .
the hilbert series of @xmath15/i_g$ ] equals @xmath104 times the hilbert series of @xmath76/m_g$ ] , where @xmath105 is the initial ideal in ( [ eq : mg ] ) .
the latter series equals @xmath106 this finite sum is over all elements @xmath107 that represent parking functions on @xmath0 with respect to the last node @xmath1 , and @xmath108 denotes the class of the reduced divisor of degree @xmath96 given by the vector @xmath109
. see ( * ? ? ?
* theorem 6.14 ) for a nice formula , due to merino @xcite , which expresses this sum with @xmath110 in terms of the tutte polynomial of @xmath0 .
we fix the natural epimorphism from the semigroup algebra of @xmath111 to that of @xmath112 : @xmath113 \rightarrow \mathbb{z}[{\rm div}(g ) ] \,,\,\ , \
, { \bf x}^{\bf u } \mapsto t^{|{\bf u}| } q^{{\rm div}({\bf u } ) } .\ ] ] with this notation , our minimal free resolution in theorem [ satminres_theo ] implies the following result : [ cor : hs ] the hilbert series of @xmath114/i_g$ ] in the grading by the group @xmath112 equals @xmath115 it suffices to note that the @xmath116-degree of the basis element @xmath50 of the free @xmath76$]-module in the @xmath47-th step of the resolution of @xmath11 is the exponent vector of @xmath117 .
this monomial does not contain @xmath74 because @xmath118 .
we close this section with a combinatorial recipe for the socle monomials modulo @xmath11 .
these are the monomials @xmath119 that are not in @xmath11 but @xmath120 for @xmath121 .
each permutation of @xmath122 $ ] corresponds to a flag @xmath123 of subsets @xmath124 .
the flag is complete , meaning that each inclusion is strict and each @xmath125 is a singleton .
let @xmath126 denote the set complement of @xmath127 with respect to @xmath41 $ ] .
for instance , @xmath128 .
[ gensocle_theo ] the socle monomials of @xmath76/m_g$ ] are precisely the @xmath129 monomials @xmath130 where @xmath131 runs over all complete flags of subsets of @xmath122 $ ] . the scarf complex of @xmath11 is a minimal free resolution and it is supported on the barycentric subdivision of the @xmath24-simplex , by ( * ? ? ?
* corollary 6.9 ) and our discussion above .
each facet in that barycentric subdivision corresponds to a complete flag @xmath123 .
the vertices of that facet are labeled by @xmath132 in the scarf complex , and the monomial label of the facet is their least common multiple .
facets of the scarf complex are in bijection with the irreducible components of @xmath11 and also with the socle monomials modulo @xmath11 . by (
* corollary 6.20 ) , each socle monomial is multiplied by the product of all variables to give the monomial label of the corresponding facet .
our results hold verbatim for all generic sublattices of finite index in the root lattice @xmath133 , so we recover the voronoi theory of @xcite .
we posit that our commutative algebra derivation of their voronoi theory is a natural and useful one , and that it opens up new and unexpected connections .
for instance , grbner bases of lattice ideals are fundamental for integer programming @xcite .
one original source for that application is herbert scarf s seminal work on neighborhood systems in economics .
a key example that motivated scarf was the _
leontief system _ @xcite .
it turns out that the lattices representing leontief systems are precisely our generic lattices here . the grbner basis property stated in theorem [ satminres_theo ]
is in fact equivalent to ( * ? ? ?
* theorem 2.2 ) .
in this section we fix an arbitrary artinian monomial ideal @xmath134 in a polynomial ring @xmath135 = \mathbb{k}[x_1,\ldots , x_m]$ ] .
we focus on alexander duality @xcite , and we establish the riemann - roch formula ( [ eq : rr ] ) in this new context . towards the end of this section , and in the next section , we will recover the riemann - roch formula for graphs from the riemann - roch formula for monomial ideals . to begin with ,
we need to gather the ingredients , that is , we need to redefine the notions of divisor , genus , rank and degree .
the role of _ divisors _ on the monomial ideal @xmath134 is played by laurent monomials @xmath28 .
( * rank of a monomial * ) [ def : rank ] the _ rank _ of a monomial @xmath28 with respect to @xmath134 is one less than the minimum degree of any monomial @xmath136 that satisfies @xmath137 .
this definition is restricted to honest monomials @xmath28 , where @xmath138 . just before the statement of theorem [ rr_theo ]
, we shall extend the definition of rank to all laurent monomials .
the rank measures how deeply a monomial @xmath28 sits inside the ideal @xmath134 .
we have @xmath139 if @xmath140 and @xmath141 otherwise .
rank zero monomials form the _ border _ of @xmath134 .
let @xmath142 denote the set of _ socle monomials _ of @xmath15/m$ ] .
see figure [ fig : zwei ] for a picture of a monomial ideal .
the ideal generators are the large black circles , and monomials in @xmath134 are labeled by their rank .
the socle elements are the black squares , and other standard monomials are white squares .
( * reflection invariance * ) a monomial ideal @xmath134 is _ reflection - invariant _ if there exists a _
canonical monomial _
@xmath143 such that the map @xmath144 defines an involution of the set @xmath145 .
this requires that every socle monomial divides @xmath29 .
using notation as in @xcite , we note that our artinian monomial ideal @xmath134 is reflection - invariant with canonical monomial @xmath29 if and only if the following identity holds : @xmath146 } \,\,\ , = \,\,\ , \bigl\langle\ , { \rm monsoc}(m ) \,\bigr\rangle .\ ] ] here @xmath4 and @xmath147}$ ] is the _ alexander dual _ of @xmath134 with respect to @xmath148 .
( * genus * ) the ideal @xmath134 is _ level _ if all socle monomials have the same degree .
if this holds then one plus that degree is called the _ genus _ of @xmath134 , denoted @xmath149 .
let @xmath134 be the ideal generated by the seven underlined monomials in ( [ eq : kayfour2 ] ) .
then @xmath134 is level of genus @xmath150 , because all six socle monomials in ( [ eq : kayfour4 ] ) are cubics , and @xmath134 is reflection - invariant .
the canonical monomial @xmath151 has degree @xmath152 .
but , the rank of @xmath29 is equal to @xmath153 , as can be seen from the following lemma .
for @xmath154 we abbreviate @xmath155 .
[ rankformart_theo ] let @xmath134 be an artinian monomial ideal
. then every monomial @xmath28 satisfies @xmath156 the condition @xmath157 in definition [ def : rank ] is equivalent to @xmath158\backslash m$ ] .
maximizing the degree of @xmath159 subject to this condition is equivalent to minimizing the degree of @xmath136 .
but , since @xmath134 is artinian , the maximal degree among its finitely many standard monomials is attained by one of the socle monomials @xmath160 .
formula ( [ rank_form ] ) resembles the formula in ( * ? ? ?
* lemma 2.2 ) for the rank of a divisor on a finite graph .
we shall exploit this resemblance at the end of this section .
is riemann - roch of genus @xmath60 , with canonical monomial @xmath161 and @xmath162 .
generators and socle monomials highlighted in dark .
monomials in @xmath134 are labeled with their rank .
the square boxes correspond to the standard monomials of @xmath15/m$ ] .
the dotted lines mark the boundary of the staircase region of @xmath147}$ ] .
note that the identity ( [ eq : reflinv ] ) holds .
, width=415 ] the formula in ( [ rank_form ] ) can be rewritten to be reminiscent of s - pairs for grbner bases : @xmath163 we now define the _ rank _ of an arbitrary laurent monomial @xmath28 by the formula ( [ rank_form ] ) .
this is consistent with definition [ def : rank ] , and it is the natural extension to monomials some of whose exponents are negative .
the following main result gave this section its title : [ rr_theo ] let @xmath134 be a monomial ideal that is artinian , level , and reflection - invariant .
then @xmath134 satisfies the riemann - roch formula , i.e. , every laurent monomial @xmath164 satisfies @xmath165 we denote @xmath166 by @xmath167 . using the formula for rank shown in lemma [ rankformart_theo ] ,
the left hand side of ( [ eq : rrf ] ) equals @xmath168 for any socle monomial @xmath169 we have @xmath170 , and hence @xmath171 taking the minimum of @xmath172 over @xmath173 , equation ( [ dplus_eq ] ) implies @xmath174 since the map @xmath175 is an involution , we can replace @xmath176 by @xmath177 in the second row of ( [ eq : dreiii ] ) .
it then follows from ( [ eq : vierrr ] ) that ( [ eq : dreiii ] ) is equal to @xmath178 , as desired .
theorem [ rr_theo ] can be extended to monomial ideals @xmath134 that are artinian and reflection invariant but not necessarily level .
such @xmath134 arise as initial ideals from directed regular ( indegree = outdegree ) graphs .
following @xcite , we define @xmath179 as one minus the minimum degree of a socle monomial of @xmath134 , and @xmath180 as one minus the maximum degree of a socle monomial of @xmath134 . using a technique similar to that in the proof of theorem [ rr_theo ]
, we can derive the following riemann - roch inequalities : @xmath181 of course , the above inequality generalizes the riemann - roch formula ( [ eq : rrf ] ) : if the ideal @xmath134 is also level then @xmath182 , and the riemann - roch formula ( [ eq : rrf ] ) immediately follows from the inequalities ( [ rre_inq ] ) .
we say that a monomial ideal @xmath134 is _ riemann - roch _ if it is artinian , level , and reflection invariant .
see figure [ fig : zwei ] for an example in two variables . in what follows
we assume that @xmath134 is a riemann - roch monomial ideal .
the next corollaries are formal consequences of the riemann - roch formula , as is the case for algebraic curves and graphs .
[ cano_cor ] if @xmath28 is a multiple of @xmath29 then @xmath183 .
if @xmath29 divides @xmath164 , then @xmath184 . plugging this equation into the riemann - roch formula gives the assertion .
note that , by definition , the degree of the canonical monomial @xmath29 equals twice the socle degree .
we record the following general facts about the canonical monomial @xmath29 .
[ cor : xk ] the canonical monomial of a riemann - roch monomial ideal @xmath134 satisfies @xmath185 experts will note that the rank is off by one when compared to the canonical divisor of an algebraic curve or metric graph .
this discrepancy will be addressed in remark [ rmk : recon ] below .
we now prepare for an analogue of clifford s theorem on special divisors .
[ revtri_lem ] the rank is superadditive for monomials @xmath186 and @xmath164 of non - negative rank : @xmath187 consider an arbitrary monomial @xmath169 of degree at most @xmath188 such that @xmath169 divides @xmath189 .
the following formulas define monomials @xmath190 and @xmath191 such that @xmath190 divides @xmath136 and @xmath191 divides @xmath28 : @xmath192 by construction , @xmath193 .
this implies that either @xmath194 or @xmath195 . in other words
, either @xmath196 or @xmath197 is in @xmath134 , and hence , their product @xmath198 is also in @xmath134 . from this
we infer the inequality ( [ eq : superadd ] ) as follows : since @xmath169 is an arbitrary monomial of degree less than or equal to @xmath188 , and @xmath169 divides @xmath189 , we know that any monomial that `` defines '' the rank of @xmath189 ( i.e. , a monomial @xmath199 of minimum degree such that @xmath199 divides @xmath189 and @xmath200 ) has degree strictly greater than @xmath188 .
hence , @xmath201 .
( * clifford s theorem * ) let @xmath28 be a monomial dividing @xmath29 such that both @xmath202 and @xmath203 are non - negative . then @xmath204
. lemma [ revtri_lem ] and corollary [ cor : xk ] imply @xmath205 from the riemann - roch formula we have @xmath206 the desired conclusion follows by adding these two identities and dividing by @xmath207 .
the construction of all riemann - roch monomial ideals of genus @xmath208 works as follows .
we first fix a monomial @xmath209 with degree @xmath210 .
next we choose a set @xmath211 of monomials of degree @xmath212 that divide @xmath29 .
then there exists a unique artinian monomial ideal @xmath134 whose socle is spanned by the monomials in @xmath211 and their complements relative to @xmath29 : @xmath213 namely , the ideal @xmath134 is the intersection of the irreducible ideals @xmath214 where @xmath169 runs over the set @xmath215 .
then @xmath134 is artinian , level , and reflection - invariant .
we shall now make the connection to the riemann - roch theorem for graphs .
as in section 2 , we let @xmath0 denote a saturated graph on @xmath1 nodes , with @xmath216 edges between nodes @xmath22 and @xmath23 , and @xmath11 the initial monomial ideal in @xmath217 $ ] of the toppling ideal @xmath9 with respect to a reverse lexicographic term order having @xmath74 as smallest variable .
[ thm : isrr ] let @xmath0 be a saturated graph with @xmath1 vertices , @xmath218 edges , and node @xmath219 having degree @xmath220 . then the monomial ideal @xmath11 is riemann - roch with canonical monomial @xmath221 the monomial ideal @xmath11 is artinian , and it is level because all the socle monomials @xmath222 in corollary [ gensocle_theo ] have the same degree @xmath223 .
this quantity is the cyclotomic number ( or genus ) of the graph @xmath0 , which , by @xcite , coincides with the common degree of all maximal parking functions .
there is a natural involution @xmath175 on the set of @xmath129 maximal flags @xmath123 of subsets in @xmath122 $ ] .
it takes a flag @xmath224 to the reverse flag @xmath225 . using the identification between flags and socle monomials in corollary [ gensocle_theo ]
, we have @xmath226 where @xmath29 is the monomial defined in ( [ eq : m - n+2 ] ) . hence , @xmath11 is also reflection - invariant .
not every riemann - roch monomial ideal arises as an initial monomial ideal @xmath11 for a connected graph @xmath0 . to see this , note that the number of socle monomials of @xmath11 is at most @xmath227 where @xmath228 is the number of variables of @xmath11 . on the other hand , riemann - roch monomial ideals can in general have more
than @xmath227 socle monomials .
furthermore , the initial monomial ideal @xmath11 for a connected graph @xmath0 is not necessarily riemann - roch . to see this , consider the four - cycle @xmath229 .
its initial ideal is @xmath230 , with socle monomials @xmath231 , @xmath232 and @xmath233 , and @xmath234 is not reflection - invariant .
in the remainder of this section we show how the familiar riemann - roch theorem for graphs is derived from theorem [ thm : isrr ] .
while the proof still assumes that @xmath0 is saturated , that hypothesis will be removed in the next section .
the following algebraic definitions are valid for any undirected connected graph @xmath0 on @xmath41 $ ] . here
@xmath0 need not be saturated .
the laurent polynomial ring @xmath235 = \mathbb{k}[x_1^{\pm 1 } , \ldots , x_n^{\pm 1}]$ ] is a module over the polynomial ring @xmath15 = \mathbb{k}[x_1 , \ldots , x_n]$ ] .
the laplacian lattice @xmath236 is a sublattice of rank @xmath237 in @xmath238 , and we write @xmath239 for the corresponding _ lattice module _ , as in ( * ? ? ?
* definition 9.11 ) .
thus , @xmath239 is the @xmath15$]-submodule of @xmath235 $ ] generated by all laurent monomials @xmath240 that have degree zero in the grading by the group @xmath112 .
if @xmath0 is saturated then @xmath239 is generic and the scarf complex is a minimal free resolution by ( * ? ? ?
* theorem 9.24 ) .
that scarf complex is precisely the _ delaunay triangulation _ in @xcite , and our point here is to redevelop the amini - manjunath approach in the language of commutative algebra .
consider the set of all laurent monomials @xmath169 that are in the socle of the module @xmath239 : @xmath241 this socle is a set of laurent monomials on which the lattice @xmath239 acts with finitely many orbits , so the computation of @xmath242 is a finite algorithmic problem , as in @xcite . the problem s solution is given by the socle monomials ( [ eq : mgsoc ] ) of our monomial ideal @xmath11 .
[ laplatsoc_theo ] the socle monomials of the lattice module @xmath239 are precisely of the form @xmath243 , where @xmath244 and @xmath245 runs over the minimal generators of @xmath239 .
since @xmath0 is assumed to be saturated throughout this section , the lattice module @xmath239 is generic in the sense of ( * ? ? ?
* definition 9.23 ) , with @xmath11 being the reverse lexicographic initial ideal of the corresponding lattice ideal @xmath9 .
we claim that the stated characterization of the socle is valid for any generic lattice module that is artinian . indeed , by the proof of (
* theorem 5.2 ) , the @xmath238-degrees of the @xmath1-th syzygies of @xmath239 are the vectors @xmath246 , where @xmath247 runs over the @xmath116-degrees of the @xmath248-st syzygies of @xmath11 and @xmath249 is any vector in the lattice .
the socle degrees of @xmath11 are the vectors @xmath250 in @xmath116 , and the socle degrees of @xmath239 are the vectors @xmath251 in @xmath252 .
we now identify laurent monomials @xmath253 with divisors on the graph @xmath0 .
the @xmath22-th coordinate @xmath254 of the exponent vector @xmath255 is the multiplicity of node @xmath22 in the divisor @xmath253 .
the _ degree _ of the divisor @xmath253 is its total degree as a monomial , @xmath256 .
the _ rank _ of the divisor @xmath253 is defined by the same formula ( [ rank_form ] ) as in lemma [ rankformart_theo ] : @xmath257 thus , @xmath258 if and only if @xmath253 lies in @xmath239 .
our definition of rank of the divisor @xmath253 coincides with the rank of @xmath255 as in @xcite . to see this , use lemma 2.7 in @xcite and
note that the exponents of the socle laurent monomials of @xmath239 are the elements of the set @xmath259 in @xcite .
we finally define the _ canonical divisor _ of @xmath0 to be the monomial @xmath260 where @xmath261 is the degree of node @xmath22 . finally , we recall that the genus of @xmath0 is @xmath223 , where @xmath218 is the number of edges .
the following is precisely ( * ? ? ?
* theorem 1.12 ) : [ thm : bn ] riemann - roch holds for any divisor @xmath253 on the graph @xmath0 : @xmath262 it follows from lemma [ laplatsoc_theo ] that all socle monomials of @xmath239 have degree equal to the genus of @xmath0 minus one .
the lattice module @xmath239 is also reflection - invariant , in the sense that @xmath263 implies @xmath264 . using the representation in lemma [ laplatsoc_theo ] ,
the resulting involution @xmath175 on @xmath265 can be written as follows : @xmath266 where @xmath267 denotes the reverse flag as in ( [ eq : involution ] ) .
note that the image of @xmath175 is in @xmath265 , since @xmath268 is in @xmath269 .
the proof of theorem [ thm : bn ] is now entirely analogous to that of theorem [ rr_theo ] . in other words ,
our argument for the validity of the riemann - roch formula for reflection - invariant artinian level monomial ideals generalizes in a straightforward manner to reflection - invariant artinian level lattice modules .
[ rmk : recon ] the rank of the canonical monomial of @xmath11 equals the rank of the canonical divisor of the graph @xmath0 , but the degree of the former is two more than that of the latter .
we turn to graphs @xmath0 that are not necessarily saturated .
the binomials in ( [ eq : ijbinomial ] ) , their syzygies in ( [ eq : cycres2 ] ) , and the ideals @xmath9 and @xmath11 are still well - defined . however , the minimality in theorem [ satminres_theo ] is no longer true , and the choice of term order is more subtle , as the next example shows . let @xmath0 be the edge graph of a triangular prism , labeled so that @xmath9 equals @xmath270.\ ] ] this toppling ideal has @xmath271 minimal generators , and its free resolution has betti numbers @xmath272 .
the same holds for the ideal that represents parking functions : @xmath273 here the reverse lexicographic term order with @xmath274 was used .
however , if we take @xmath275 then the reverse lexicographic grbner basis requires two more binomials .
now , the initial monomial ideal has betti numbers @xmath276 . to explain the phenomenon in this example , we fix a spanning tree @xmath277 of the graph @xmath0 that is rooted at the node @xmath1 , and we order the unknowns according to a linear extension of @xmath277 .
thus , we fix an ordering of @xmath41 $ ] such that @xmath278 if the node @xmath22 is a descendant of the node @xmath23 in @xmath277 . a term order on @xmath15 $ ] is a _ spanning tree order _ if it is a reverse lexicographic term order whose variable ordering is compatible with some spanning tree rooted at @xmath1 .
one spanning tree order is the toppling order considered in ( * ? ? ?
* theorem 10 ) .
see also @xcite for a discussion of grbner bases of toppling ideals in the inhomogeneous case .
[ nonsatinit_theo ] the toppling ideal @xmath9 is generated by the binomials @xmath279 where @xmath43 runs over splits of @xmath41 $ ] such that the subgraphs of @xmath0 induced on @xmath39 and @xmath40 are connected . for any spanning tree order ,
these binomials form a grbner basis of @xmath9 with initial monomial ideal @xmath11 .
the complexes constructed in ( [ eq : cycres1 ] ) are cellular free resolutions .
the first paragraph in the proof of theorem [ satminres_theo ] is valid in the non - saturated case .
it shows that the binomials @xmath279 lie in @xmath9 . for the spanning tree term order ,
the leading monomials are @xmath280 , where @xmath281 , and hence the initial ideal @xmath75 contains the monomial ideal @xmath11 of ( [ eq : mg ] ) .
again , both ideals are artinian of the same colength in @xmath76 $ ] , and hence they are equal .
this establishes the grbner basis property .
the argument in the proof of ( * ? ? ?
* theorem 14 ) shows that the property that the subgraphs of @xmath0 induced on @xmath39 and @xmath40 are connected characterizes a minimal grbner basis of @xmath9 and the minimal generators of @xmath11 .
in particular , these binomials @xmath279 generate @xmath9 . our last assertion states that ( [ eq : cycres1 ] ) with differentials ( [ eq : cycres2 ] ) gives a free resolution of @xmath9 , and dropping the last term in ( [ eq : cycres2 ] ) gives a free resolution of @xmath11 .
this claim is proved by deformation to generic monomial modules , as explained in @xcite . to be precise , in our situation
we replace @xmath0 by a nearby saturated graph @xmath282 with fractional edge numbers @xmath283 between any pair of nodes .
the monomial ideal @xmath284 is generic and degenerates to @xmath11 .
the lattice ideals @xmath9 and @xmath285 are represented by the corresponding lattice modules @xmath239 and @xmath286 .
these are submodules of the laurent polynomial ring as in ( * ? ? ?
* definition 9.11 ) .
the lattice module @xmath286 is generic and degenerates to @xmath287 . according to ( * ? ? ?
* theorem 6.24 ) , the scarf complex of @xmath284 _ with labels from @xmath0 _ gives a free resolution of @xmath11 .
likewise , the scarf complex of the generic lattice module @xmath286 _ with labels from @xmath0 _ gives a free resolution of @xmath239 .
the resulting minimal free resolution of @xmath285 degenerates to a ( typically non - minimal ) resolution of @xmath9 , using ( * ? ? ? * corollary 9.18 ) .
these free resolutions of @xmath11 and @xmath9 are cellular because they are given by labeled simplicial complexes .
let @xmath34 and @xmath0 be the @xmath288-cycle @xmath289 . for @xmath290
consider the graph @xmath291 that has @xmath292 edges for every edge in @xmath0 and @xmath293 edges for every non - edge of @xmath0 .
then @xmath291 is saturated for @xmath294 .
the scarf complex in example [ ex : ressi ] gives the minimal free resolution of @xmath295 and this lifts to the minimal free resolution of @xmath296 . by theorem [ thm : isrr ]
, the monomial ideal @xmath297 is riemann - roch , its genus is @xmath298 , and its canonical monomial equals @xmath299 .
the involution @xmath300 on its six socle monomials is given by swapping the two rows below : @xmath301 setting @xmath302 and @xmath303 , we get the parking function monomial ideal of the @xmath288-cycle @xmath304 here , @xmath305 .
this ideal is not reflection - invariant and hence not riemann - roch .
the cellular resolution of @xmath11 induced from @xmath297 is not minimal .
we now take a closer look at the combinatorial structure of our resolutions .
let @xmath306 denote the first barycentric subdivision of the @xmath24-simplex , whose @xmath44 vertices , namely the non - empty subsets @xmath39 of @xmath122 $ ] , are labeled by the corresponding monomials @xmath307\backslash i}$ ] .
thus , @xmath306 is the cellular free resolution of @xmath105 referred to in theorem [ nonsatinit_theo ] .
each simplex in @xmath306 is labeled by the least common multiple of the monomials that label its vertices . for any @xmath308
we write @xmath309 for the subcomplex consisting of all simplices in @xmath306 whose labels properly divide @xmath169 .
we next state the analogous result for the lattice module @xmath239 , that is , the @xmath15$]-module generated by all laurent monomials whose exponent vector lies in the laplacian lattice @xmath311 .
we identify this lattice with @xmath312 by writing its elements as @xmath313 where each @xmath314 is unique modulo @xmath315 .
the _ tropical metric _ on @xmath312 is @xmath316 we write @xmath317 for the corresponding _ flag simplicial complex_. thus , @xmath317 is the simplicial complex whose simplices are subsets @xmath318 of @xmath312 such that @xmath319 for @xmath320 .
the notation `` apt '' refers to the fact that this infinite simplicial complex is the standard _
apartment _ in the affine building of lie type @xmath321 .
it is well - known that @xmath317 is pure of dimension @xmath237 and that it triangulates the @xmath248-dimensional affine space @xmath322 . for more on buildings and their connection to tropical geometry ,
see @xcite .
the apartment @xmath317 is precisely the same as the delaunay triangulation constructed in @xcite , and it also coincides with the scarf complex of @xmath282 that we used to give a cellular resolution of @xmath239 .
the number of @xmath22-faces of @xmath317 modulo the lattice action is given by ( [ eq : cycnumber ] ) .
each vertex @xmath323 of @xmath317 is labeled by the corresponding laurent monomial @xmath324 , and each face is labeled by the least common multiple of its vertex labels .
thus , each face of @xmath317 is labeled by a laurent monomial of degree @xmath325 . we write @xmath326 for the subcomplex of all simplices
whose label properly divides @xmath169 .
[ cor : minsyz2 ] the number of minimal @xmath327-st syzygies of the lattice module @xmath239 in degree @xmath177 is the rank of the reduced homology @xmath328 .
the sum of these ranks over all @xmath177 modulo @xmath329 counts the minimal @xmath22-th syzygies of the toppling ideal @xmath9 .
this conjecture has been verified for many graphs using the software macaulay2 .
we note that the two simplicial complexes appearing in ( [ eq : baryapt ] ) are different from the complex @xmath332 used in hochster s formula for the betti numbers of a lattice ideal ( * ? ? ?
* theorem 7.4 ) . the simplicial complexes @xmath326 can be large even for small graphs .
let @xmath0 be the graph on four nodes , labeled @xmath333 , with laplacian matrix @xmath334 both the toppling ideal and the ideal of parking functions are complete intersections : @xmath335 the monomial ideal @xmath11 has one minimal first syzygy in degree @xmath336 .
the simplicial complex @xmath337 consists of two isolated nodes @xmath338 and @xmath339 .
the simplicial complex @xmath340 is two - dimensional but it has the homology of a circle .
it consists of @xmath60 triangles , @xmath341 edges and @xmath342 vertices , labeled by the following generators of @xmath239 : @xmath343 the lattice module @xmath239 has one second syzygy in this degree , translating into a first syzygy of @xmath9 .
it is represented in @xmath340 by the @xmath288-cycle @xmath344 . at present , no explicit minimal free resolution of @xmath11 is known .
finding such a resolution was stated as an open problem by postnikov and shapiro in @xcite .
we do not even know whether the betti numbers depend on the characteristic of the field @xmath33 .
an explicit formula for the betti numbers of the toppling ideal @xmath9 was conjectured by wilmes in @xcite .
see also @xcite .
wilmes formula is combinatorial , and it has been verified for all graphs with @xmath345 nodes . at present
we do not know how to relate wilmes conjecture to the ranks of the homology groups in corollaries [ cor : minsyz1 ] and [ cor : minsyz2 ] .
the monomial ideal @xmath11 is artinian , so its maximal syzygies correspond to the socle elements .
these are the maximal parking functions , and , by ( * ? ? ?
* theorem 4.1 ) , they correspond to cyclic orientations of @xmath0 with node @xmath1 as unique sink . since all maximal syzygies of @xmath105 lift to maximal syzygies of @xmath9 , the same result holds for @xmath9 .
we now derive the riemann - roch theorem for non - saturated graphs @xmath0 .
let @xmath11 be the initial ideal with respect to a spanning tree order on the variables with @xmath74 as the least . by corollary [ maxsyz_cor ]
, we know that the socle monomials of the laurent monomial module @xmath239 are @xmath347 where @xmath348 runs over all socle monomials of @xmath11 and @xmath245 runs over minimal generators of @xmath239 .
unlike in the saturated case , the monomial ideal @xmath11 is generally not reflection - invariant .
but the laurent monomial module @xmath239 is always reflection - invariant . to see this
, we use lemma 3.2 of @xcite to deduce that @xmath349 , defined in ( [ eq : mgsoc ] ) , is not contained in @xmath239 for any complete flag @xmath123 of @xmath41 $ ] .
this implies that @xmath350 is a socle element of @xmath239 , since every laurent monomial of degree greater than @xmath351 is in @xmath239 .
we now immediately verify that @xmath239 is reflection - invariant with the involution on @xmath242 that takes @xmath352 to @xmath353 , where @xmath267 is the reverse flag of @xmath123 exactly as in the proof of theorem [ thm : bn ] .
the canonical monomial is @xmath260 where @xmath261 is the degree of node @xmath22 .
hence , @xmath239 satisfies the riemann - roch formula , in its monomial formulation ( [ eq : rrf ] ) , with @xmath354 and @xmath355 .
* acknowledgments * : we thank david perkinson , b.v .
raghavendra rao , frank - olaf schreyer , and john wilmes for helpful discussions .
bernd sturmfels was partially supported by the u.s . national science foundation ( dms-0757207 and dms-0968882 ) . | the riemann - roch theorem on a graph @xmath0 is related to alexander duality in combinatorial commutive algebra .
we study the lattice ideal given by chip firing on @xmath0 and the initial ideal whose standard monomials are the @xmath0-parking functions .
when @xmath0 is a saturated graph , these ideals are generic and the scarf complex is a minimal free resolution .
otherwise , syzygies are obtained by degeneration .
we also develop a self - contained riemann - roch theory for artinian monomial ideals . |
it is well known that the system comprising a squid ring ( i.e. a single josephson weak link enclosed by a thick superconducting ring ) , inductively coupled to a resonant circuit [ typically a parallel lc radio frequency ( rf ) tank circuit ] , can display very interesting dynamical behaviour @xcite .
this behaviour is due to the non - linear dependence of the ring screening supercurrent @xmath0 on the applied magnetic flux @xmath1 which originates ultimately from the cosine ( josephson ) term in the squid ring hamiltonian @xcite . in the past ,
limitations in the noise level , bandwidth and dynamic range of rf receivers used to investigate this coupled system have restricted the range of non - linear phenomena that could be observed , i.e. only relatively weak non - linearities in @xmath2 could be probed .
however , recent improvements in rf receiver design have changed this situation quite radically .
the regimes of behaviour of squid rings @xcite are usually characterized in terms of the parameter @xmath3 , where @xmath4 is the critical current of the weak link in the ring , @xmath5 is the ring inductance and @xmath6 is the superconducting flux quantum .
thus , when @xmath7 ( the inductive , dissipationless regime ) @xmath8 is always single valued in @xmath9 . by contrast , for @xmath10 ( the hysteretic , dissipative regime ) this current is multi - valued in @xmath9 .
both these regimes can lead to very strong non - linearities in @xmath11 @xcite . for the inductive regime
these will be strongest when @xmath12 from below , at which point @xmath13 becomes almost sawtooth in @xmath9 ( modulo @xmath14 ) .
for the hysteretic regime the existence of bifurcation points in @xmath15 , at particular values of @xmath9 , can lead to even stronger non - linearities . in the absence of external noise these bifurcations , linking adjacent branches in @xmath16 , occur when @xmath17 .
although the behaviour of single weak link squid rings ( as shorted turns ) can be monitored at any non - zero frequency , most studies have been carried out at radio frequency ( rf , typically @xmath18 ) for which a wide range of high performance electronics is available .
we have adopted this frequency range in the work described here . in practice ,
signal to noise is improved ( as in this work ) by driving the squid ring through the intermediary of a resonant circuit , almost invariably a parallel lc ( tank ) circuit with a high quality factor @xmath19 .
a schematic of the squid ring , inductively coupled to an rf tank circuit ( with parallel inductance and capacitance @xmath20 and @xmath21 , respectively ) , is shown in figure 1 . in the most
commonly used experimental scheme the dynamical behaviour of the ring is followed in the time domain through the so - termed squid magnetometer characteristics . in these ,
the rf voltage @xmath22 developed across the tank circuit is plotted as a function of the rf current @xmath23 used to excite it , where @xmath24 is linearly amplitude modulated in time ( figure 1 ) .
this current generates an rf flux ( peak amplitude @xmath25 ) in the tank circuit inductor ( coil ) , a fraction of which ( @xmath26 for a coil - ring mutual inductance of @xmath27 ) then couples to the squid ring . the supercurrent response in the ring
then back - reacts on the tank circuit , and so on .
any additional static or quasi - static magnetic flux @xmath28 that is required is usually supplied via a second coil coupled to the squid ring , again as depicted in figure 1 . for small @xmath29-values
( i.e. from @xmath30 to a few ) the @xmath31 versus @xmath32 dynamics contain as their main element a set of voltage features repeated periodically in @xmath24 at intervals @xmath33 .
the precise form of the squid magnetometer characteristics depends both on the type of rf detection employed ( e.g. phase sensitive or diode ) and the value of @xmath34 .
the non - linear screening current response can also be made manifest in the frequency domain .
this can be accomplished at a fixed level of ( peak amplitude ) rf flux @xmath35 at the squid ring , without the need for amplitude modulation .
for example , in the inductive regime , with @xmath36 very close to one from below , we have demonstrated that , at sufficiently large values of @xmath35 , fold bifurcations develop in the resonance lineshape of a squid ring - tank circuit system . indeed , using a bidirectional sweep ( up and down in frequency ) , and at certain values of @xmath35 and @xmath37 , the system can even display opposed ( hammerhead ) fold bifurcations in this lineshape , one on each side of the resonance @xcite .
opposed bifurcations are not only most unusual but , to our knowledge , the possibility of their existence is scarcely mentioned in the literature on non - linear dynamics @xcite . in the case of the inductive squid ring , with @xmath38 very close to one
, we have found that these bifurcations are always well separated in frequency .
however , in this work we will show that in the weakly hysteretic regime of squid behaviour @xmath39 this separation can become extremely , perhaps vanishingly , small . throughout this work we shall term this a pinch resonance .
the one sided ( above or below resonance ) weak fold bifurcation @xcite is taken as the text book example of the onset of non - linear behaviour in a strongly driven oscillator . the development of opposed fold bifurcations ( on both sides of the resonance ) in any non - linear system is another matter altogether .
as far as we are aware , opposed fold bifurcations had not be seen experimentally prior to our report of hammerhead resonances in the frequency domain of coupled inductive squid ring - tank circuit systems @xcite , and only once in the theoretical literature @xcite . in itself
, we consider that this constitutes a significant contribution to the body of knowledge concerning non - linear dynamical systems . that the cosine non - linearity of the squid ring can also generate , in part of its parameter space , something as exotic a a pinch resonance is simply remarkable , particularly since such resonances appear as solutions to the established equations of motion for the ring - tank system in the weakly hysteretic regime .
what this result , and others , point to is that the non - linear dynamics of squid rings coupled to other , linear , circuits is likely to prove very rich indeed @xcite , the more so since the parameter space that has so far been explored is still very limited . as such
, these coupled systems are likely to continue to be a focus of attention by the non - linear community , both in theory and experiment .
with the squid ring treated , as in this paper , quasi - classically , there would appear to be many potential applications of its intrinsic non - linear nature when coupled to external circuits , for example , in logic and memory ( single @xcite or multi - level @xcite ) , stochastic resonance @xcite and in safe communication systems based on chaotic transmission / reception @xcite .
however , in the last few years there has been a burgeoning interest in exploring the properties of squid rings in the quantum regime for use in quantum technologies , particularly for the purposes of quantum computing and quantum encryption @xcite .
this interest has been boosted by recent experimental work on quantum superposition of states in weak link circuits @xcite and very strongly in the last year by reports in the literature of experimental investigations of superposition states in squid rings @xcite .
quantum technologies will , of necessity , involve time dependent superpositions for their operation .
where quantum circuits such as squid rings are involved , this means the interaction with external electromagnetic ( em ) fields or oscillator circuit modes .
correspondingly , the effect of applied em fields ( classical or quantum ) of high enough frequency and amplitude is to generate quantum transition regions where energy is exchanged between the ring energy levels .
given the cosine term in the squid ring hamiltonian @xcite , these energy exchanges in general involve ( non - perurbatively ) coherent multiphoton absorption or emission processes @xcite .
these exchange regions extend over very small ranges in the static magnetic flux @xmath34 applied to the ring ( typically over @xmath40 to @xmath41 ) , comparable to the width of the anticrossing regions reported in the recent experiments on superposition of states in squid rings @xcite .
such narrow flux widths imply the concomitant existence of extremely strong non - linearities in the screening supercurrents flowing in the ring . in general
these current non - linearities generate non - linear dynamical behaviour in external ( classical ) circuits used to probe the superposition state of the ring . as we have emphasized in previous publications @xcite ,
following these non - linear dynamics is one way of extracting information concerning the quantum state of a squid ring , or any other quantum circuit based of weak link devices . from our perspective , therefore , it is of clear importance to the future development of quantum technologies based on squid rings and related devices that these non - linear dynamics be studied in great detail . given the results presented in this paper , the non - linear dynamics of coupled squid ring - probe circuit systems can display rich and yet unanticipated behaviour . with these rings operating in the quantum regime
this is even more likely to be the case .
although sometimes unappreciated , probing the non - linear dynamical behaviour of squid ring - tank circuit systems makes strong demands on the high frequency electronics used . in order to open up new non - linear phenomena in these systems we required rf electronics which combined ultra low noise performance , large dynamic range and high slew rate capability . in practice
we achieved this by using a liquid helium cooled ( 4.2 kelvin ) , gaasfet - based , rf amplifier ( noise temperature @xmath42 10 kelvin , gain close to 20db ) followed by a state of the art , low noise , room temperature rf receiver . with these electronics
we could choose diode or phase sensitive detection of @xmath31 . since phase sensitive detection generally provides better signal to noise , this was always our chosen mode of operation for acquiring squid magnetometer characteristics . in the work reported in this paper
the squid rings used were of the zimmerman two hole type @xcite .
these were fabricated in niobium with a niobium weak link of the point contact kind .
this weak link was formed by mechanical adjustment between a point and a post screw , the latter having a plane machined end face which had been preoxidized prior to its insertion in the squid block .
the weak link was made in situ at 4.2 kelvin via a room temperature adjustment mechanism .
as we have shown in previous publications @xcite , the control we can exert using this mechanism is very good , sufficient in fact to make reproducible point contact squid rings with any desired @xmath36 value from inductive @xcite to highly hysteretic @xcite .
in figure 2 we show what we term the equal amplitude magnetometer characteristics for a weakly hysteretic ( @xmath36 close to @xmath43 ) niobium point contact squid ring coupled to an rf tank circuit . in these characteristics , taken at 4.2 kelvin with a static bias flux of @xmath44 , the frequency of the drive current was set to the coupled squid block - tank circuit resonant frequency before a weak link contact was made @xmath45 and the voltage components ( @xmath46 and @xmath47 ) were maintained exactly @xmath48 out of phase .
however , their relative phases with respect to @xmath49 were rotated electronically until the average slopes of the two characteristics are essentially equal .
this approach was adopted for convenience - it presented the periodic features in @xmath24 of the orthogonal phase components of @xmath31 at essentially the same amplitude . here ,
the ( measured ) ring inductance @xmath50 , the measured ring - tank circuit coupling coefficient @xmath51 , @xmath52 , @xmath53 @xmath54 and the drive ( excitation ) frequency for @xmath24 , @xmath55 .
it is apparent from figure 2 that even in this weakly hysteretic regime the observed squid magnetometer characteristics are highly non - linear and contain a wealth of detail .
some of the time domain information of figure 2 can be recast in a quite dramatic form in the frequency domain . to extract this
experimentally requires the use of a high performance spectrum analyzer with a tracking generator output @xcite .
this output provides a voltage signal of constant peak to peak amplitude over the frequency range of interest . feeding this through a series high impedance generates an rf current in the tank circuit which
, in turn , creates an rf flux ( peak amplitude @xmath35 ) at the squid ring . as with our previous work on squid ring - tank circuit dynamics in the inductive regime @xcite ,
the details of this behaviour depend crucially on the value of the static bias flux @xmath34 . in figure 3
we show four frequency responses for unidirectional sweeps , from low to high frequency , for the squid ring - tank circuit system of figure 2 ( at t = 4.2 kelvin ) .
the data were taken at rf flux amplitudes @xmath35 ranging from @xmath56 to @xmath57 , with @xmath58 .
these frequency responses were recorded using a high performance , but conventional , commercial spectrum analyzer ( a rohde and schwarz fsas system ) which could only scan in this unidirectionally manner .
what is clear from figure 3 is that as @xmath59 becomes a significant fraction of a flux quantum the lineshape of the frequency response curves changes dramatically . in particular , for @xmath60 there is a punch through effect which leads to the creation of fold bifurcation cuts .
if we refer back to figure 2 , this value of peak rf flux @xmath35 equates to a drive current corresponding to the feature shown bracketed in figure 2 , but for the case of static bias flux @xmath61 .
the significance of this was revealed when bidirectional frequency sweeps ( from low to high and then high to low ) were taken using a spectrum analyzer of our own design .
this spectrum analyzer is shown in block form in figure 4 . as an illustration of its utility ,
we show in figure 5 the result of this bidirectional sweep for the curve ( d ) of figure 3 ( @xmath62 ) , again at @xmath63 .
as can be seen in figure 5 , there are four fold bifurcation cuts , one each at the lowest and highest frequency excursions in the response , and two inner cuts , one on the sweep up and the other on the sweep down . within the experimental resolution ,
as reflected in the data points , these inner cuts appear to overlay exactly . to our knowledge , this has not been observed before .
for this bidirectional response it seems reasonable to adopt the description pinch resonance .
the frequency response curve of figure 5 and those plotted in figure 3 as a function of coupled rf amplitude , are perfectly typical of the low @xmath36 value ( @xmath64 few ) hysteretic squid ring - tank circuit systems that we have studied . to aid in interpretation we have provided in figure 6 a schematic of three different bidirectional frequency response curves : ( a ) where the inner pair of opposed fold bifurcation cuts are well separated in frequency ( b ) where these cuts superimpose in frequency ( i.e. in our terminology , a pinch resonance ) and ( c ) where the fold bifurcations have moved beyond one another .
it seems clear that the data of figure 5 correspond to the bidirectional response of figure 6(b ) since this response can obviously be differentiated from those of figures 6(a ) and 6(c ) .
it is interesting to note that the existence of any one of the bidirectional response curves shown schematically in figure 6 could not be revealed using standard commercial spectrum analyzers since these always sweep the frequency unidirectionally , i.e. from low to high . for completeness
, we show in figure 7 the bidirectional frequency responses ( with both up and down sweeps ) for the ring - tank circuit system of figures 3(d ) and 5 @xmath65 at the static bias flux values @xmath66 , @xmath67 and @xmath68 .
as is apparent , the pinch resonance can only be seen in one of these responses , i.e. at @xmath69 .
in the standard dynamical description of a squid ring it is customary to invoke the well known ( quasi - classical ) resistively shunted junction plus capacitance ( rsj+c ) model @xcite of the weak link in the ring . here
, a pair ( josephson ) current channel is in parallel with an effective weak link capacitance @xmath70 and a normal current channel of resistance @xmath71 . for an applied current @xmath72 ,
the normal channel opens up with a value of @xmath71 which is characteristic of the particular weak link in the ring .
in this ( and other ) models the squid ring moves in the space of total included magnetic flux @xmath73 with an effective mass given by the weak link capacitance @xmath74 . within the rsj+c model
it is always assumed that the squid ring can be treated as a quasi - classical object , such that @xmath70 is relatively large , typically @xmath75 to @xmath76 . in the rsj+c description
the squid ring and tank circuit constitute a system of coupled oscillators with equations of motion which can be written in the form _ tank circuit _
@xmath77 where @xmath78 is the tank circuit resistance on parallel resonance of the coupled system and @xmath24 is the drive current ( figure 1 ) , which may contain both coherent and noise contributions .
_ squid ring _
@xmath79 where @xmath80 is the phase dependent josephson current for the weak link in the ring .
until recently little effort appears to have been directed to solving the coupled equations in their full non - linear form , understandably given the level of computational power required to deal with them .
instead it has often been assumed that it is sufficiently accurate to linearize the squid ring equation of motion .
this is equivalent to invoking the adiabatic condition that the ring stays at , or very close to , the minimum in its potential @xmath81 as this varies with time .
the squid equation ( [ squeqmot ] ) then reduces to @xmath82 however , as we have shown in recent work @xcite , this approximation should be treated with great care , particularly in the hysteretic regime of behaviour . in order to avoid this problem
we have chosen to solve the full coupled equations of motion [ ( [ tceqmot ] ) and ( [ squeqmot ] ) ] for the system using values of @xmath70 and @xmath71 considered typical of a weakly hysteretic squid ring .
we shall demonstrate that pinch resonances , such as we have presented in figure 5 , and which match with the experimentally determined circuit parameters , show up as solutions of these equations . with a knowledge of the system parameters , and an estimate of @xmath4
, we can solve the coupled equations of motion ( [ tceqmot ] ) and ( [ squeqmot ] ) numerically to find the value , at a given @xmath83 , of the rf tank circuit voltage @xmath84 as a function of the frequency of the fixed amplitude sinusoidal drive current @xmath49 .
this voltage response can then be used to compute the bidirectional frequency responses of the ring - tank circuit system . to do this
we considered an interval @xmath85 $ ] in the drive frequency which contained the ( peak amplitude ) resonant frequency of the system .
this interval was divided into @xmath86 bins which gave us a resolution of @xmath87 in the frequency domain .
we then drove the tank circuit at @xmath88 ( for @xmath24)@xmath89 for @xmath90 increments in frequency ( this being done without changing the initial conditions of the system ) , thus obtaining the power in the tank circuit at each of these frequencies via @xcite @xmath91 for a given integer @xmath92 .
we note that @xmath92 is taken so that the power is integrated over a few @xmath53 cycles of the tank circuit oscillation to reduce the effects of transients .
in the situation where the non - linearities in the coupled ring - tank circuit system are strong , and where , typically , the rf drive amplitudes are large , we have already shown @xcitethat fold bifurcation regions ( i.e. regions where more than one stable solution exists ) can develop in the resonance curves of the system . to make a comparison with the experimental data of figure 5 we estimated the value of the @xmath36-parameter from the @xmath93 squid magnetometer characteristic of figure 2 using the ratio ( on the current axis @xmath94 ) of the initial riser to the interval between the periodic features in this characteristic . this yielded a @xmath36 value of @xmath95 , which , for a ring inductance of @xmath96 , set @xmath97 .
in a recent publication we described a quantum electrodynamic model @xcite of a squid ring which has allowed us to derive an expression for the effective ring capacitance @xmath74 , knowing the superconducting material from which it is constructed and the value of the weak link critical current . with niobium as the material , and @xmath98 to a few @xmath99 ,
this yields @xmath100 .
an alternative model @xcite , using a different approach , gives rise to similar size capacitances . from the literature @xcite ,
a corresponding effective dissipative resistance in the ring @xmath101 appears perfectly reasonable , setting a ring @xmath102 time constant of @xmath103secs .
the theoretical bidirectional frequency response curves corresponding to the experimental data ( and circuit parameters ) of figure 5 are shown in figure 8 , with @xmath4 and @xmath102 taken to be @xmath104 and @xmath103 secs , respectively .
these were computed , sweeping forwards and backwards across the defined frequency window , using a 4@xmath105 order runge - kutta - merson routine with adaptive step size error control .
the squid ring - tank circuit coupling @xmath106 and @xmath53 parameters were set as for figures 3(d ) and 5 together with @xmath107 and @xmath108 . from numerous computations
we have found that the shape of the bidirectional response in this rsj+c approach is determined principally by the @xmath102 product and is only weakly dependent on the particular values of @xmath70 and @xmath71 chosen , at least in the ranges @xmath103 to @xmath76 and @xmath109 to @xmath110 . with the circuit parameters detailed above
, we found that the best fit to figure 5 ( the pinch off resonance ) is made with @xmath111 secs . as for the experimental data displayed in figure 5 , the theoretical inner fold bifurcation cuts superimpose very well , at least to the limits of our computational accuracy .
this is reflected in the computed data points in the region of these inner cuts . clearly , the correspondence between the theoretically calculated response curves of figure 8 and the experimental bidirectional responses of figure 5 is good , both in frequency range and lineshape .
for example , at @xmath63 the experimental and theoretical responses appear to meet exactly at one point in the frequency domain .
however , we must caution that due to the numerical nature of the calculation it is not possible theoretically to prove that this single frequency meeting point is exact .
nevertheless , by trying different amplitudes for @xmath24 we have found that we can narrow the separation between the solutions down to an arbitrarily small level .
this reflects the situation found experimentally .
we note that we have been unable to generate this theoretical fit to our experimental bidirectional reponse curves using the linearized squid equation ( [ lineqmot ] ) . with regard to the alternative frequency response curves of figure 6 , and the experimental data of figures 5 and 7
, we note that it may be appropriate to treat this weakly hysteretic squid ring - tank circuit system as an extremely sensitive finite state machine @xcite .
if so , we may be able to compare the transition state diagram for the different cycles shown in figure 6 [ ( a ) , ( b ) and ( c ) ] .
as an example , we may ask whether it is possible in the pinch resonance ( b ) to go between the lower left and lower right branches without jumping up to the loop .
this could happen occasionally but still be beyond the sensitivity limits of our present experimental apparatus .
if it does occur there will be a sharp transition between the full pinch loop resonance [ fine black line in figure 6(b ) ] and the resonance displaying sets of bifurcation cuts .
, i.e. in the one case the resonance loop is not accessed , in the other ( the latter ) it is . in principle , the two processes could be distinguished since the pathways around the loop involve very much larger voltages than the crossing between the lower branches of the resonance .
such contrasting processes would , of course , be of scientific interest but could also have potential in device applications .
in this paper we have shown that a single weak link squid ring , weakly hysteretic and coupled to a simple parallel resonance tank circuit , can display quite remarkable , and previously unexpected , non - linear behaviour in the frequency domain .
thus , we have demonstrated that this system exhibits opposed fold bifurcation cuts in its bidirectional frequency response curves which , at specific values of rf and static bias flux , and within our experimental resolution , superimpose exactly . since single fold bifurcations in the upper or lower branches of a strongly driven resonant system are usually taken as one of the standard examples of the onset of non - linear dynamics @xcite , the onset of opposed fold bifurcations is of special note . that this can result , as a limiting case , in the creation of a pinch resonance
would appear to be exceptional and , in our opininion , of very considerable interest to the non - linear community .
nevertheless , as we have also shown , such pinch off resonances can be modelled well within the full , non - linear rsj+c description of the ring - tank circuit system . together with other recent published work @xcite
, these results would appear to suggest that there is still a great deal to be explored in the non - linear dynamics of squid ring - resonator systems in the quasi - classical regime .
it is also apparent that the new non - linear dynamical results obtained in this regime point to future research needed to implement quantum technologies based on squid rings ( and related weak link devices)interacting with classical probe circuits @xcite . from both perspectives it is clear that the squid ring which , through the cosine term in its hamiltonian , is non - linear to all orders , is set to continue to reveal new and interesting non - linear phenomena of importance in the field of non - linear dynamics and in the general area of the quantum - classical interface as related to the measurement problem in quantum mechanics .
it is also apparent that in both the classical and quantum regimes the non - linear dynamics generated in coupled circuit systems involving squid rings could form the basis for new electronic technologies .
we would like to thank the engineering and physical sciences research council for its generous funding of this research .
see , for discussions of bifurcations , etc . in non - linear systems ,
d.zwillinger , `` handbook of differential equations '' ( academic press , boston , 1992 ) , p.16 and s.h.strogatz , `` non - linear dynamics and chaos '' ( perseus books , reading , massachusetts , 1994 ) , p.44 .
m.j.everitt , p.stiffell , t.d.clark , a.vourdas , j.f.ralph , h.prance and r.j.prance , `` a fully quantum mechanical model of a squid ring coupled to an electromagnetic field '' , to be published in phys .
b , april 1 , 2001 . | in this paper we present experimental data on the frequency domain response of a squid ring ( a josephson weak link enclosed by a thick superconducting ring ) coupled to a radio frequency ( rf ) tank circuit resonator .
we show that with the ring weakly hysteretic the resonance lineshape of this coupled system can display opposed fold bifurcations that appear to touch ( pinch off ) .
we demonstrate that for appropriate circuit parameters these pinch off lineshapes exist as solutions of the non - linear equations of motion for the system . |
organic solar cells ( osc ) manufactured from organic blends @xcite represent a promising low - cost , rapidly deployable strategy for harnessing solar energy @xcite . in contrast to traditional silicon based solar cells , organic ( or polymer ) solar cells are low weight , printable on flexible substrate , and most importantly , can be produced at room temperature at very low cost .
solvent - based thin - film deposition technologies @xcite ( e.g. , spin coating , drop casting , doctor blading , roll - to - roll manufacturing ) are the most common organic photovoltaic manufacturing techniques .
these techniques , especially doctor blading and roll manufacturing , are very attractive , due to their ease of scale - up for large commercial production .
all solution - processing techniques usually involve preparing dilute solutions of electron - donor and electron - accepting materials in a volatile solvent . after some form of coating onto a substrate , the solvent evaporates .
an initially homogeneous mixture separates into electron - accepting rich regions and electron - donor rich regions as the solvent evaporates ( see figure [ fig : evapse : scheme ] ) . depending on the specifics of the polymer blend and processing conditions ( spin coating time @xcite , solvent type @xcite , nature of substrates ) ,
different morphologies are typically formed in the active layer .
the active layer of osc is a blend of two types of materials : electron donating and electron accepting material .
the active layer is sandwiched between electrodes ( see figure [ fig : evapse : scheme ] right ) .
the morphological distribution of the electron - donor and electron - acceptor subregions in the device strongly determines the power conversion efficiency ( pce ) of oscs @xcite .
in fact , every stage of the photovoltaic process is affected by this morphology .
consequently , there is immense interest to understand morphology evolution during fabrication @xcite .
an important technological goal ( that understanding morphology evolution will help achieve ) is the ability to design fabrication processes to obtain tailored morphologies for high efficiency osc devices .
current state - of - the - art approaches towards tailoring the manufacturing process are limited to combinatorial trial - and - error - based experimental investigations @xcite .
furthermore , the inability to experimentally visualize morphology evolution hinders the ability to quantify the effect of various process and system variables ( such as evaporation rate , blend ratio , solvent type ) on morphology evolution .
experimental techniques provide only limited data for analysis : ( a ) mainly limited to observations of the lateral organization of the top layer , and ( b ) mainly limited to the final morphology .
in addition , visualizing 3d morphology remains challenging .
these challenges serve as a compelling rationale for developing a computational framework that can model morphology evolution , thus significantly augmenting experimental analysis . computational approaches to this problem
exist , but are mostly limited to one scale : atomistic scale or macro scale . from a macro scale perspective
the problem of thin film formation is well - studied .
the series of work summarized in @xcite link the macroscale film thickening during evaporation process with angular velocity of the coater , concentration , evaporation rate , and solution viscosity .
_ however , morphology evolution is not analyzed in these studies .
_ on the other end of the spectrum , morphology evolution in a typical osc system was recently studied using molecular dynamics simulations @xcite .
the authors were able to predict phase separation between two typical components in a cubical domain of size @xmath0 and only for @xmath1 , without incorporating the macroscale effects of evaporation or substrate .
to the authors best knowledge , there exists no meso - scale approach that links morphology evolution at the nano scale with macro - scale phenomena like evaporation , and substrate patterning .
the development of such a model will provide significant advantages , it will in particular : * serve as a powerful tool to analyze morphology evolution over time in three dimensions .
this can be used as a `` stereological microscope '' to visualize morphology evolution from early stages until the formation of the stable morphology .
it is worth mentioning that three dimensional experimental reconstruction of the final morphology is possible using electron tomography @xcite but this approach to polymeric systems is exceedingly rare because of the required proper contrast between components . *
allow for independent control over various process and system variables , thus making it easy to isolate factors affecting the process ( such as substrate patterning , solvent annealing , or blend ratio ) .
* ability to perform high throughput analysis .
such an ability allows automated exploration of the phase space of various manufacturing and system variables ( such as different blend ratios , evaporation rates , solvent choices , effect of substrates ) to understand their effect on morphology and performance .
this opens up the possibility of data - driven knowledge discovery to understand the effects of different competing phenomena and , subsequently , tailoring the manufacturing process . in this work
, we develop a computational framework to model morphology evolution during fabrication of organic solar cells .
we formulate a model that takes into account all the important processes occurring during solvent - based fabrication of oscs .
the model is based on a phase field approach to describe the behavior of multicomponent system with various driving forces .
we develop an efficient numerical framework that enables three dimensional , long time - scale analysis of the fabrication .
we illustrate the framework by investigating the effect of independent control over various external conditions : solvent evaporation , blend ratio , and interaction parameters .
we further quantify the interplay between the solvent evaporation and diffusion within the film .
to the best knowledge of the authors , this is the first comprehensive effort to construct a virtual framework to study 3d morphology evolution during solvent - based fabrication of organic solar cells .
all solution - processing techniques usually involve preparing dilute solutions of electron - donor and electron - accepting materials in a volatile solvent . after some form of coating onto a substrate , the solvent evaporates .
an initially homogeneous mixture separates into electron - accepting rich regions and electron - donor rich regions as the solvent evaporates . depending on the specifics of the blend and processing conditions ( spin coating time @xcite , annealing time @xcite , solvent type @xcite , nature of substrates ) ,
different morphologies are typically formed .
the two materials usually used in fabricating osc are a conjugated polymer and a fullerene derivative , the conjugated polymer is the electron donor material , while the fullerene derivative is the electron - acceptor .
we will use the terms electron - accepting material ( electron - donating material ) , acceptor ( donor ) and fullerene ( polymer ) interchangeably in this article .
there is a rich and complex collection of interacting phenomena that direct the morphology evolution during solvent - based fabrication .
_ phase - separation _
@xcite is a key phenomena triggered by the _ evaporation of the volatile solvent_. the atmosphere on the free surface determines the evaporation rate of the solvent .
the resulting morphology can form multiple phases , making the system a _ multi - component and multi - phase _ system .
in addition , the morphology on the substrate s surface may differ from that in the ` bulk ' state @xcite . both _
free surface and substrate _ influence the organization of the morphology and directly affect characteristics of the devices .
in particular , chemical and physical patterning of the substrate have been shown to affect morphology evolution . in order to enable predictive modeling ,
each of the following phenomena must be included in the computational model .
the rate at which solvent is removed from the top layer depends on various factors ( e.g. solvent volatility , spinning velocity during spin - coating ) . during evaporation ,
the initially dilute solution becomes enriched in the two solutes due to depletion of solvent .
this enrichment results in increased interaction between the solutes and triggers morphology evolution .
the evolution is critically determined by the evaporation profile and diffusion of solutes ( and solvent ) within the film . .
the binodal curve is related to the equilibrium phase boundary between the single phase and the phase separated region . during typical fabrication ,
initial dilute solution , @xmath2 , is pushed into two - phase region ( delimited by the spinodal line ) as solvent evaporates ( @xmath3 ) .
solution separated into polymer - rich ( @xmath4 ) and fullerene - rich phase ( @xmath5 ) .
the equilibrium compositions of two phases lie on the binodal curve .
line connecting corresponding equilibrium compositions are called tie - lines ( right ) . , title="fig:",scaledwidth=40.0% ] .
the binodal curve is related to the equilibrium phase boundary between the single phase and the phase separated region . during typical fabrication ,
initial dilute solution , @xmath2 , is pushed into two - phase region ( delimited by the spinodal line ) as solvent evaporates ( @xmath3 ) .
solution separated into polymer - rich ( @xmath4 ) and fullerene - rich phase ( @xmath5 ) .
the equilibrium compositions of two phases lie on the binodal curve .
line connecting corresponding equilibrium compositions are called tie - lines ( right ) . , title="fig:",scaledwidth=40.0% ]
as the solvent is removed , the blend is pushed into the spinodal range ( immiscible conditions ) and induces phase separation ( see figure [ fig : ps : scheme ] left ) .
phase separation ( or spinodal decomposition ) is a mechanism by which solution separates to create phases of different properties . under these conditions ,
the solution is unstable and even small fluctuations lead to fast phase separation .
the composition of phases changes and reaching equilibrium concentrations .
usually , the solution separates into two phases .
one phase is rich in donor , the other phase is rich in acceptor material .
the exact composition of phases is determined by thermodynamic conditions @xcite .
the creation of the two phases is followed by slow coarsening .
the kinetics of phase separation and coarsening is affected by the kinetics of solvent removal . in a confined thin film geometry , the evolution of the morphology close to the walls / interfaces can be significantly different from that of the bulk .
in particular , chemical interactions between the substrate and the solute as well as surface patterning can significantly affect morphology evolution .
recent experimental studies of osc explore this possibility @xcite . by changing substrate properties and by ( chemically or physically ) patterning the substrate , it was possible to direct vertical segregation , percolation and control phase separation . in certain systems ,
crystallization is an additional mechanism of morphology evolution . while incorporating crystallization is relatively straightforward in the current framework
, we primarily focus on modeling evaporation and substrate induced spinodal decomposition in this paper .
this is in line with experimental results @xcite which seem to suggest that phase separation is a key mechanism in morphology evolution for polymer - polymer blends .
phase field methods have been used to model morphology evolution in heterogeneous materials typically consisting of grains or domains characterized by different structure , orientation or chemical composition .
these methods are highly versatile ( due to a diffuse - interface formulation ) and easily represent the evolution of complex morphologies with no assumption made about shape or distribution of domains .
various thermodynamic driving forces for morphology evolution , e.g. bulk energy of the system , interfacial energy , substrate energy can be easily introduced .
additionally , the effect of different transport processes , such as mass diffusion , convection , or shear rates , can be directly introduced and exploited using this technique .
the advantages and flexibility offered by the phase field method provide an ideal framework to model morphology evolution during fabrication of osc . in particular , the mechanisms that direct morphology evolution ( spinodal decomposition and crystallization )
can be naturally modeled using this method .
they have been well studied for alloy system ( dendritic growth and spinodal decomposition in alloy system @xcite ) . additionally , various driving forces for phase transformation , such as effect of substrate and evaporation ,
can be introduced in a very straightforward way without substantial model reformulation .
finally , phase field methods can be scaled to predict morphology evolution for device - scale problems , which is currently impossible using any other framework , like molecular dynamics .
in this section , we formulate the phase field model to simulate morphology evolution in a ternary system with solvent evaporation and substrate interaction included into the model .
[ ch : mo ] formulating a phase field model usually consists of three stages : ( 1 ) identifying order parameters ; ( 2 ) postulating free energy functional that depends on the order parameters ; ( 3 ) constructing governing equations that describe the evolution of the system towards a minimum energy state .
we consider a ternary system consisting of polymer , fullerene , and solvent .
we denote the volume fractions of polymer , fullerene and solvent as @xmath6 , @xmath7 and @xmath8 , respectively .
we set the volume fractions @xmath9 as the conserved order variables ( since @xmath10 ) .
note that during evaporation , while volume is not conserved , the sum of the volume fractions by definition is conserved . in the second step ,
we construct the energy functional for this system .
energy , @xmath11 , consists of two bulk terms : homogeneous energy , and interfacial energy between phases .
the total energy is given as : @xmath12 dv+ f_s(\phi_p,\phi_f,\textbf{x } ) \label{eq : energy}\ ] ] the energy of the homogeneous system , @xmath13 , also called configurational energy or free energy of mixing , is the quantity which governs phase separation .
homogeneous energy depends only on local volume fractions and is at least double - welled .
wells correspond to the equilibrium concentrations of separated phases .
in contrast , the interfacial energy depends on the composition gradient and is scaled by an interfacial parameter @xmath14 ( see second term in eq . [ eq : energy ] ) . in the current work ,
we assume @xmath15 .
we construct homogeneous energy , @xmath13 , using the flory - huggins formulation @xcite , which is suitable for polymer solutions . , to the flory - huggins energy function .
this modification allows to improve the efficiency of computation , while not changing the rate of morphology evolution or the shape of forming structure .
the parameter @xmath16 is set to a small value @xmath17 . ]
according to this theory , the energy of the system is given by : @xmath18 \label{eqfloryhuggins}\ ] ] where : @xmath19 is the gas constant ; @xmath20 is the temperature ; @xmath21 is the volume fraction of component @xmath22 ; @xmath23 is the degree of polymerization of component @xmath22 . here
, @xmath24 is the molar volume ; @xmath25 is a reference molar volume ( e.g. solvent ) ; and @xmath26 is the flory - huggins binary interaction parameter between component @xmath22 and @xmath27 .
the interaction parameters and degree of polymerization define the shape of the binodal and spinodal lines ( figure [ fig : ps : scheme ] ) .
one of the advantages of the phase - field method is the ability to introduce additional driving forces to the model in a straightforward way .
for instance , the effect of the substrate is a surface energy term , @xmath28 that is simply added to the total energy of the system ( eq .
[ eq : energy ] ) .
a generalized form of the substrate energy is given in eq .
[ eq : fs ] .
patterning is introduced through the space dependent functions @xmath29 and @xmath30 in eq .
[ eq : fs ] .
these functions determine the geometry of patterning . at a point @xmath31 on the substrate that is chemically tuned to component @xmath22 , the value of @xmath32 , otherwise @xmath33 .
the chemical specificity of patterning @xmath22 is reflected in the parameters @xmath34 and @xmath35 .
parameter @xmath36 is a chemical potential favoring component @xmath22 at the substrate , and @xmath37 expresses the way interactions between the components near the substrate are modified by the presence of a pattern at the substrate @xcite . @xmath38 ds \label{eq : fs}\ ] ] once the energy of the system is specified , the governing equations of the evolution can be formulated .
this is usually done by defining the chemical potentials of the system .
the chemical potential quantifies how much the energy changes when the configuration changes .
the chemical potential for polymer and fullerene are @xmath39 and @xmath40 , respectively .
next , using fick s first law for the flux ( @xmath41 ) and the continuity equation ( @xmath42 ) we get the governing equation for each component .
we consider only two of the three variables as independent ( since @xmath43 ) .
the resulting cahn - hilliard - cook equations are given by : @xmath44 + \xi_p \label{eq : goveq1}\ ] ] @xmath45 + \xi_f \label{eq : goveq2}\ ] ] where @xmath46 is the mobility of component @xmath22 .
the diffusivity , @xmath47 , is a linear combination of self - diffusivities of all components and their volume fractions : @xmath48 .
the energy of ideal solution , @xmath49 , is used to link mobility with diffusivity and to comply with classic fick s law .
the ideal solution is one with zero interaction parameters .
the advection term ( second term in lhs of both equations ) accounts for the change in height of the film and is scaled according to the height . scaling term
@xmath50 is zero at the bottom surface ( @xmath51 ) and one at the top surface ( @xmath52 ) , where @xmath53 is the total current height of the film .
substrate effects are included in the third rhs term of both equations .
this term enters the equation only for the bottom surface ( zero height @xmath51 ) where the heaviside function , @xmath54 .
the last term of rhs ( in both equations ) is the langevin force term , @xmath55 and @xmath56 .
this term mimics the conserved noise due to fluctuations in composition .
we set the noise to be a gaussian space - time white noise with the following constrains : @xmath57 and @xmath58 , where @xmath59 is the kronecker delta .
variance is determined by the fluctuation - dissipation theorem ( fdt ) @xcite .
cahn - hilliard equation with noise considered is called the cahn - hilliard - cook equation @xcite .
this formulation allows for a natural extension to include crystallization . to do this ,
an additional phase variable must be considered , the energy functional expanded accordingly and the governing equation formulated . from a mathematical perspective , an evaporation process is classified as a moving boundary problem , since the height of the thin film changes over time ( @xmath60 ) .
following @xcite , we explicitly trace the surface evolution .
we assume that solvent is the only component that evaporates from the top surface .
solvent lost from the top layer results in height decrease
. the rate at which the height decreases , @xmath61 , is proportional to the flux of the solvent out of system , @xmath62 : @xmath63 where @xmath62 is the normal component of molar flux ( of solvent into the air ) .
this equation constitutes the mass balance for moving film - air interface : @xmath64 with two assumptions : only solvent evaporates ( @xmath65 ) and solvent content in the air is very low ( @xmath66 ) .
the superscript @xmath67 and @xmath68 denote air and film , respectively . the molar flux of solvent can be further linked with evaporation rate of the solvent , @xmath69 , and the average content of the solvent at the top layer , @xmath70 , using equation : @xmath71 in this work , we assume that the solvent evaporates uniformly from the top surface and film height decreases homogenously .
this assumption simplifies analysis and quantification of the competing effects of evaporation and substrate shown in the results section .
we investigate the effect of inhomogeneous evaporation on the film surface evolution in a forthcoming publication .
the evaporation rate , @xmath69 , depends on various parameters including the solvent partial pressure , solvent vapor pressure , temperature , air flow and is specific to the fabrication process . for spin - coating ,
the evaporation rate depends on the angular velocity @xmath72 @xcite , while for solvent annealing or drop - casting , the evaporation rate can be assumed constant ( @xmath73 ) .
we apply boundary conditions at the top surface to satisfy the balance of two other components within the film .
removal of the solvent from the top layer , results in enrichment of the volume fraction of polymer and fullerene .
therefore , to account for this enrichment we apply neumann boundary conditions at top surface for two other components : @xmath74 .
although the phase field method is a well used technique for simulating the morphological evolution of a wide variety of materials and processes @xcite , employing it to predict morphology evolution during fabrication of active layer for organic solar cells requires resolving several challenges .
we detail these challenges below .
morphology evolution during fabrication of organic solar cells is an _ intrinsically multiscale process both in time and space _ which makes this process difficult to solve accurately and efficiently using reasonable computational resources .
the governing equations ( eqns .
[ eq : goveq1 ] , [ eq : goveq2 ] ) are forth - order nonlinear partial differential equations , which are difficult to solve numerically .
the complexity of these equations is related to two competitive subprocesses : _ phase separation and coarsening_. these processes occur at two widely different temporal and spatial scales , all of which must be resolved properly .
initially , very fast phase separation is the dominant process and followed by slow coarsening .
phase separation is a fast process and results in thin layer creation .
in contrast , coarsening is slow process consisting of rare events and involving merging of bulky regions .
another numerical challenge is related to the _ evaporation process_. in case of the solvent - based thin - film deposition , the volume fraction of solvent changes from 99% to almost zero .
this poses a severe problem for numerical techniques that have to reliably model the huge change in domain size in 3d .
a key objective of the formulation is the ability to predict morphology evolution at _ the device scale_. the inherent complexity of the process makes this objective demanding . this is because we are interested in resolving nano scale morphological evolution while investigating device - scale domains . from a computational perspective
this involves solving differential equations with a very large number of degrees of freedom ( @xmath75 ) , which can not be solved using current serial processing machines .
this require developing modules heavily based on parallel processing , including applying domain decomposition strategies @xcite .
the phase field method is a generic technique , and can be applied to almost any type of system undergoing morphology evolution .
thus , in order to provide quantitative prediction , it is necessary to determine material - specific set of parameters , both thermodynamic and kinetic .
these parameters very often show compositional , directional or temperature dependence , which pose additional difficulties . in this
regard , significant work has been done to determine several parameters for materials and systems utilized for photovoltaic applications using both molecular dynamics @xcite and experimental techniques @xcite .
the strong form is formulated as follows : find @xmath76\to \mathbb{r}$ ] and @xmath77\to \mathbb{r}$ ] ( @xmath78 is an auxiliary variable ) such that : @xmath79 , \label{eq : sfchpa}\\ \frac{\partial \phi_f}{\partial t } + u\frac{h}{h^{curr}}\frac{\partial \phi_f}{\partial h } = \nabla \cdot \left(m_f\nabla \mu_f\right ) + \xi_f & in & \;\omega \times [ 0,t ] , \label{eq : sfchfa}\\ \mu_p = \displaystyle \frac{\partial f}{\partial \phi_p } - \epsilon^2 \nabla^2 \phi_p + \left(1-h(h)\right)\frac{\partial f_s}{\partial \phi_p } & in & \;\omega \times [ 0,t ] , \label{eq : sfchpb}\\ \mu_f= \displaystyle \frac{\partial f}{\partial \phi_f } - \epsilon^2 \nabla^2 \phi_f + \left(1-h(h)\right)\frac{\partial f_s}{\partial \phi_f } & in & \;\omega \times [ 0,t ] , \label{eq : sfchf}\\ \nabla ( m_p\mu_p ) \cdot \textbf{n } = h_{\mu_p } & on & \;\gamma_t \times [ 0,t ] , \label{eq : sfchpe}\\ \nabla ( m_f\mu_f ) \cdot \textbf{n } = h_{\mu_f } & on & \;\gamma_t \times [ 0,t ] , \label{eq : sfchfe}\\ \phi_p(\textbf{x},0)=\phi_p^0(\textbf{x } ) & in & \;\omega , \\ \phi_f(\textbf{x},0)=\phi_f^0(\textbf{x } ) & in & \;\omega .
\end{aligned}\ ] ] we apply neumann boundary conditions on the top surface , @xmath80 , to account for polymer and fullerene fraction increase at the top layer due to solvent evaporation .
flux of polymer @xmath81 equals to @xmath82 , while flux of the fullerene @xmath83 equals to @xmath84 . on other boundaries
, we apply zero flux conditions for both variables : volume fraction , @xmath9 , and chemical potential , @xmath85 . to account for domain change due to solvent removal from the top layer
, we introduce a coordinate transformation @xcite : @xmath86 where @xmath87 is the height coordinate scaled according to current total height of the film , @xmath53 .
this coordinate transformation permits recasting the model equations into a system of equations with fixed boundaries which is more convenient for numerical solution .
_ in this way , there is no need for remeshing . _
since we assume homogeneous evaporation , film has an uniform height @xmath88 , at any point in time , @xmath89 .
recasting the model equation using the coordinate transformation [ eq : eta ] involves defining the gradient operator in the new coordinate system : @xmath90 where the first component is the direction along the height of the film .
strong forms of the recast equations are given by : @xmath91 , \label{eq : t : sfchpa}\\ \frac{\partial \phi_f}{\partial t}+u\frac{\vartheta}{h^{curr}}\frac{\partial \phi_f}{\partial \vartheta}&=&\tilde{\nabla}\cdot\left(m_f\tilde{\nabla}\mu_f\right)+\xi_f\qquad in\;\omega \times [ 0,t ] , \label{eq : t : sfchfa}\\ \mu_p&=&\displaystyle \frac{\partial f}{\partial \phi_p } - \epsilon^2 \tilde{\nabla}^2 \phi_p + \left(1-h(h)\right)\frac{\partial f_s}{\partial \phi_p } \qquad in\;\omega \times [ 0,t ] , \label{eq : t : sfchpb}\\ \mu_f&=&\displaystyle \frac{\partial f}{\partial \phi_f } - \epsilon^2 \tilde{\nabla}^2 \phi_f + \left(1-h(h)\right)\frac{\partial f_s}{\partial \phi_f } \qquad in\;\omega \times [ 0,t ] , \label{eq : t : sfchf}\\ \tilde{\nabla } ( m_p\mu_p ) \cdot \textbf{n}&= & h_{\mu_p } \qquad on\;\gamma_t \times [ 0,t ] ,
\label{eq : t : sfchpe}\\ \tilde{\nabla } ( m_f\mu_f ) \cdot \textbf{n}&= & h_{\mu_f } \qquad on\;\gamma_t \times [ 0,t ] , \label{eq : t : sfchfe } \end{aligned}\ ] ] the advection term ( second lhs term of eqns [ eq : t : sfchpa ] and [ eq : t : sfchfa ] ) is non zero only along the height direction .
velocity corresponding to this term , @xmath92 , is proportional to the molar flux of the solvent out of system : @xmath93 ( eq . [ eq : u ] ) .
we use the finite element method to solve the governing equations ( eqns [ eq : t : sfchpa ] - [ eq : t : sfchfe ] ) .
we solve equations in the split form , to avoid constraints related to the continuity of the basis functions when using the finite element method with a primal variational formulation @xcite .
more precisely , the standard @xmath94-continuous finite element formulation is not suitable for forth - order operators , and consequently basis functions which are piecewise smooth and @xmath95-continuous should be utilized @xcite .
however , there are only a limited number of finite elements that fulfill the above continuity condition , especially in two and three dimensions . the weak form of the split cahn - hilliard equation is given by : @xmath96 @xmath97 @xmath98 @xmath99 where @xmath100 are weighting functions
, @xmath101 is the @xmath102 inner product on @xmath103 , @xmath104 is the energy inner product on @xmath103 , @xmath105 and @xmath106 define natural boundary conditions .
second term in lhs of eqns [ eq : wfchap ] and [ eq : wfchaf ] accounts for domain size change due to evaporation .
fourth term in lhs of eqns [ eq : wfchap ] and [ eq : wfchaf ] accounts for the conserved noise , where @xmath107 is the vector of the stochastic flux terms .
fourth term in lhs of eqns [ eq : wfchbp ] and [ eq : wfchbf ] accounts for substrate effect and are included only for surface elements belonging to bottom boundary @xmath108 .
we note that substrate term is not a typical boundary condition but an additional term resulting from the additional energy in the system .
we use the galerkin approximation to solve the two split cahn - hilliard equations .
we define @xmath109 and @xmath110 to be the finite dimensional approximation of polymer and fullerene volume fraction fields , @xmath111 @xmath112 to be the finite dimensional approximation of polymer and fullerene chemical potential fields and @xmath113 to be the finite dimensional weighting function .
the approximate solutions are computed on the following function spaces : @xmath114 @xmath115 @xmath116 @xmath117 @xmath118 with @xmath119 being the space of the standard polynomial finite element shape functions on element @xmath120 , where @xmath121 is the polynomial order .
we additionally introduce the supg stabilization term for the advection terms in eqs [ eq : wfchap ] and [ eq : wfchaf ] @xcite . we discretize the conserved noise by generating gaussian- distributed random numbers for each component of flux @xmath122 that satisfy @xmath123 and @xmath124 .
we generate @xmath125-dimensional vector ( @xmath126 depending on 2d or 3d problem ) for each component ( polymer and fullerene ) in each node at each time step , @xmath127 .
parameters @xmath128 and @xmath129 ( where @xmath130 is size of the time step ) account for fluctuation - dissipation theorem requirement . for more details
see @xcite .
we use linear basis functions for variables @xmath6 , @xmath7 , @xmath131 , @xmath132 , @xmath133 and @xmath134 . we also tested quadratic and cubic basis functions but have not noticed any significant improvements . this choice is based on an extensive analysis of the cahn - hilliard equation in @xcite . in our approach , we take advantage of implicit time schemes due to their unconditional stability .
explicit methods are often intractable due to severe restrictions on size of time step ( @xmath135 ) arising from the stiffness of the equation .
this makes them computationally prohibitive even for simple problems .
consequently , implicit methods arise as a natural alternative . since they are unconditionally stable they allow for much larger time step .
however , because of nonlinear nature of the cahn - hilliard equation , implicit schemes require nonlinear solvers .
the popular implicit time schemes in this context are euler backward and crank - nicholson methods @xcite .
because of the multiple temporal scales that occur during phase separation , we tested adaptive various time stepping strategies @xcite .
we noticed significant improvement in terms of number of times steps required to reach steady state as well as in total run times @xcite . in these time stepping strategies , step size is adjusted on the basis of the error between solutions of different order .
however , whenever noise is considered , the error computed using such strategies is highly affected by the noise and can not be used in standard time stepping strategies .
therefore , we use a euler backward scheme with a heuristic strategy to adjust the time step as used in @xcite ( see [ sec : app ] ) .
to solve the nonlinear system of two split cahn - hilliard equations , the formulation is linearized consistently and a newton raphson scheme is used . to solve large problems with several millions of degrees of freedom , we use a domain - decomposition based mesh - partitioner to divide the mesh and
distribute it across computational nodes . in our framework
we use the parmetis partitioner @xcite . additionally , to enable parallel solution of the algebraic systems
, we use the petsc solver library @xcite .
all results reported are obtained using the generalized minimal residual method .
in this section , we showcase the formulation by investigating the effect of evaporation rate , blend ratio , degree of polymerization as well as effect of choice of solvent on morphology evolution . finally , we investigate the possibility of additional control of morphology through substrate patterning .
we investigate a representative osc system .
such a system consists of polymer , fullerene and solvent . in the default configuration , we consider degree of polymerization of solvent @xmath136 , fullerene @xmath137 and solvent @xmath138 .
degree of polymerization is computed on the basis of molar volumes of the components .
we assume @xmath139 , @xmath140 , @xmath141 .
interaction parameter between polymer and fullerene reflects the low solubility of two components and is set as@xmath142 .
interaction parameters between solvent and fullerene or polymer are assumed to be much lower : @xmath143 . in subsection [ subsec : n ] , we investigate the effect of varying degree of polymerization to @xmath144 and @xmath145 . in subsection
[ subsec : solvent ] , we investigate the effect of solvent and various interaction parameters .
all these values are representative for osc and correspond well with experimentally determined values @xcite .
we take the solvent self - diffusion coefficient as @xmath146 , which are typical for solvents ( chlorobenzene , chloroform and xylene ) used in osc @xcite .
self - diffusion of polymer and fullerene is much lower .
we assume that @xmath147 .
we present one , two and three dimensional results . for each simulation
we generate one mesh in the reference coordinate system .
mesh density is adjusted to the estimated width of the interface between polymer and fullerene .
we use the interface width as a metric to determine mesh density to accurately capture the dynamics of the interface evolution .
our detailed analysis presented in @xcite showed that at least four elements per interface are required to capture the physics of phase separation and coarsening accurately . in @xcite , we showed that the analytical estimation of interface width is fairly accurate for two and three dimensional cases , even though it was derived for the one dimensional case in @xcite .
the interface width is defined as a distance required to span by the concentration profile across the accessible range of the composition : @xmath148 . for more details
see @xcite .
the profile of concentration , and subsequently width of the interface , depends on the interfacial parameter @xmath14 , interaction parameters , degree of polymerization . in ternary system range of the composition
changes with time at the intermediate stages .
consequently , the interface width also changes with evaporation . to estimate interface width
, we consider the case when the interface width is the smallest .
this corresponds to the fully evaporated , binary fullerene - polymer system . for the system modeled in this paper , the interface width between polymer and fullerene varies from @xmath149 ( for @xmath136 ) to @xmath150 ( for @xmath145 ) .
the interfacial parameter @xmath14 is also estimated for a binary system consisting of polymer and fullerene . following the analysis in @xcite
, one can link @xmath14 with the interfacial energy , @xmath151 , between polymer - rich and fullerene - rich phases .
we assume @xmath152 , which is comparable with interfacial energies for organic compounds . for more details regarding detailed analysis and computations ,
we refer the reader to @xcite . computed value of @xmath153 is @xmath154 ( for @xmath136 ) , @xmath155 ( @xmath144 ) and @xmath156 ( @xmath145 ) .
the initial solvent fraction is chosen to guarantee that the ternary solution is homogeneous and there is no phase separation at time @xmath157 . in 3d simulations , we start with a solvent fraction @xmath158 . in 2d simulations , we start with a solvent fraction @xmath159 . in 1d simulations , we start with a solvent fraction @xmath160 . the higher initial volume fraction of solvent in 2d simulations is required to cover the wide range of degree of polymerization and interaction parameters investigated .
simulation is stopped when the fraction of solvent within the film is @xmath161 . at this time , the diffusion coefficient reduces significantly , and the morphology is frozen . in two
dimensional simulation , we generate meshes that consist of @xmath162 linear elements .
the computational domain is rectangular with height @xmath163 , and width @xmath164 .
height changes from @xmath165 to @xmath166 , when all the solvent evaporates . in three dimensional simulations
, we generated meshes that consist of @xmath167 linear elements to discretize a computational domain that is a rectangular prism of length @xmath168 , breadth @xmath169 and height @xmath170 .
height changes from @xmath171 to @xmath172 , as the solvent evaporates .
in one dimensional simulations , we generate meshes that consist of @xmath173 linear elements to discretize a height @xmath174 .
height changes from @xmath175 to @xmath176 .
the large number of degree of freedom ( @xmath17715 million for 3d simulation ) require using parallel solvers and domain decomposition .
the average run time for a three dimensional case is around 50h solved using 256cpus .
the average run time for two dimensional case is around 1.5h using 8cpus .
detailed scalability analysis of the solver has been reported elsewhere @xcite .
evaporation of the solvent from the top surface is one of the key phenomena that induces phase separation .
the way morphology evolves is an interplay between two competing processes : ( i ) evaporation of the solvent from the top surface , and ( ii ) the diffusion of the solvent within the film .
this interplay can be expressed as a mass biot number .
biot number is defined in eq .
[ eq : bi ] and expresses the ratio between external mass transport by evaporation of the solvent from the top surface and internal mass transfer to the top surface by diffusion .
@xmath178 where @xmath179 is the characteristic length - height of the film , @xmath47 is the diffusion coefficient of the thin film and @xmath69 is the mass transfer coefficient - evaporation rate .
we compute the biot number for initial height and solvent diffusion coefficient . in such case ,
the solution consists mostly of solvent , and the diffusion is dominated by it . to better understand the interplay between evaporation and diffusion , we first perform experiments in 1d .
we consider the default case of system variables ( @xmath180 , @xmath138 , @xmath181 , @xmath182 ) . in figure
[ fig : res : bi ] , we plot change of height for three biot numbers as a function of time . in the same figure
we also show the 1d volume fraction profiles of one component , polymer , as function of height .
initially , polymer is distributed uniformly along the height and its volume fraction is 0.2 .
similarly , the initial volume fraction of fullerene is 0.2 , since the blend ratio between polymer and fullerene 1:1 . for the symmetric case , the fullerene profiles mirror images of polymer profiles ( and hence are not plotted ) .
we investigate the cases with three biot number : equal to one , much larger than one ( @xmath183 ) , and much lower than one ( @xmath184 ) . for this specific system ,
biot number @xmath185 correspond to initial height @xmath186 and evaporation rate @xmath187 .
when biot number is much larger than one , evaporation is the dominant process .
this is reflected in shorter total time of evaporation compared to other cases ( see figure [ fig : res : bi ] top ) .
consequently , the solvent removed from the top layer can not be balanced by mass diffusion within the film .
thus , a boundary layer lean in solvent and rich in other components is created close to the top ( figure [ fig : res : bi ] ( right ) ) .
enrichment of polymer and fullerene results in the initiation of phase separation . in this way ,
boundary layer becomes the region of the thin film when solution is unstable even to small fluctuations , which leads to phase separation .
phase separation subsequently propagates into the depth of the film .
this is clearly seen in figure [ fig : res : bi ] ( right ) .
close to the top layer , a blocking layer rich in polymer is created . because of low diffusion coefficient of polymer , the top layer blocks solvent evaporation from the top .
when biot number is much smaller than one , diffusion is the dominant process . correspondingly , the total time of solvent removal is much longer .
the system has more time to balance solvent lost from the top layer .
diffusion is not suppressed by fast solvent removal ( as for high biot numbers ) and consequently no top boundary layer rich in polymer is created .
there is no significant gradient of the solvent in the composition within the film .
consequently , phase separation in initiated homogeneously along the height .
biot number provides important insight into interplay between evaporation and diffusion .
it can be used to link two types of morphology evolution : homogeneous across the film or initiated close to the top surface . for symmetric systems ( @xmath188 ) and
low biot number , solvent content within the film is homogeneous .
for such a scenario , an assumption of homogeneous decrease of solvent volume fraction _ within the entire volume is valid_. such assumption was made in a recent work @xcite .
however , if one of the non - solvent components has larger molar volume , and consequently large @xmath189 which is the case in osc this assumption is invalid and evaporation must be included in the model explicitly .
+ + the evaporation rate affects not only the region where phase separation is initiated but also affects the average size of the separated phases . in figure
[ fig : res : bi:3d ] , we plot morphology evolution for three different evaporation rates characterized by biot number @xmath190 , @xmath191 and @xmath192 .
the magnitude of the evaporation rate affects the total time of the process ( as shown in figure [ fig : res : bi ] ) . for lower evaporation rate , and biot number ,
total process time is longer .
once phases separates , domains rich in each component have additional time to coarsen and create larger domain ( see figure [ fig : res : bi:3d ] left ) .
for higher evaporation rate , and biot number , total time is shorter .
once phases separates they have very short time to coarsen ( see figure [ fig : res : bi:3d ] right ) .
this is due to the fact that solvent is removed from the system rapidly , and consequently the diffusion coefficient of the film decreases significantly leaving the morphology frozen . ,
@xmath191 and @xmath192 .
consecutive rows correspond to the morphology at height : initial , @xmath193 , @xmath160 , @xmath194 and the final height.,title="fig:",scaledwidth=15.0% ] + the blend ratio between polymer and fullerene is a key system variable during fabrication of osc , that is additionally relatively easy to manipulate .
the blend ratio affects the type of morphology that develops during spinodal decomposition .
two basic classes of morphologies typically develop percolated morphology and morphology with islands .
intuitively , the former is more suitable for osc than the latter .
this is because the latter structure has more islands not connected to relevant electrodes and can not provide useful pathways for charges to reach the boundaries .
therefore , for osc application , it is necessary to clearly identify process and system variables that lead to the percolated type of structure .
we investigate two blend ratios 1:1 and 1:0.8 . in figure
[ fig : res:3d : br ] , we plot snapshots of the morphology evolution with time . percolated morphology develops for blend ratio 1:1 .
however , a small change in blend ratio of the analyzed system results in significantly different morphology type .
this is clearly seen for blend ratio 1:0.8 , where multiple stripes form spontaneously and break up as the solvent evaporates .
such multiple layer formation was also observed in experiments and is reported @xcite .
we note that although a layered morphology has no application in oscs , such morphology is desired in organic transistors @xcite . ,
@xmath195 , @xmath194 and the final height.,title="fig:",scaledwidth=15.0% ] + the degree of polymerization , @xmath196 , is another tunable variable that allows for controlling morphology evolution . in practice , for osc fabrication , only polymer degree of polymerization can be controlled and it has been shown experimentally that efficiency is affected by its choice @xcite . in figure
[ fig : res:2d : n ] , we compare the morphology evolution for three different degree of polymerization : @xmath136 , @xmath197 and @xmath173 .
all simulations have been performed for the same blend ratio 1:1 and biot number @xmath198 .
as we increase the degree of polymerization , the morphology changes significantly . for the symmetric case : @xmath180 , a percolated
morphology evolves ; while for asymmetric cases multiple layers are created .
notice also that with increasing degree of polymerization , phase separation is initiated earlier .
moreover , when polymer degree of polymerization is larger than fullerene degree of polymerization , morphology type changes from percolated into multiple layered morphology . , @xmath144 and @xmath145 .
consecutive rows correspond to the morphology at height : initial , @xmath193 , @xmath160 , @xmath195 , @xmath176 , @xmath194 , @xmath199 and the final height.,scaledwidth=18.0% ] in solvent - based fabrication , the solvent creates the environment for morphology evolution .
both components must be soluble in the common solvent to create an initially homogeneous solution .
choice of the solvent has significant effect on morphology evolution and provides additional system variable for morphology control .
this is manifested as differences in relative values of interaction parameters between all three components . in figure
[ fig : res:2d : solvent ] , we show morphology evolution for three combinations of the interaction parameters ( @xmath200 , @xmath201 , @xmath202 ) : ( @xmath203,@xmath191 , @xmath191 ) , ( @xmath203,@xmath191 , @xmath160 ) and ( @xmath203,@xmath160 , @xmath191 ) . in the first case ,
solvent is chosen such that interactions between solvent and two components are the same and much lower than interaction between polymer and fullerene .
this means that polymer and fullerene have similar solubilities in solvent . in the second case ,
fullerene is less soluble in the solvent compared to polymer .
interaction parameter between fullerene and solvent is two times higher than between polymer and solvent . in the third case ,
we assume that polymer is less soluble in the solvent . in all three case
, we assume the same evaporation rate ( @xmath204 ) , blend ratio ( 1:1 ) , and degree of polymerization ( @xmath145 , @xmath205 and @xmath138 ) . changing solvent results in dramatically different morphology evolution @xcite
. multiple layer formation that we observe for higher polymerization is broken in the third case , due to the different solvent used . , @xmath206 , @xmath207 , @xmath160 , @xmath195 , @xmath208 and the final height.,scaledwidth=18.0% ] the height of a typical osc active layer is around @xmath209 .
for such thin geometries , morphology close to the substrate can be significantly different than that of the bulk .
we consider the effect of chemical patterning of substrate to selectively affect each component .
we consider two cases . in the first case ,
the substrate is patterned to attract the polymer preferentially . in the second case ,
substrate is patterned with two chemistries : one preferentially attracting polymer and one preferentially attracting fullerene . in both cases , patterns of wavelength @xmath210
cover the substrate . in each case , we assume the chemical potential of patterned material @xmath211 , and @xmath212 . in most cases , @xmath87 is small compared to @xmath78 @xcite .
figure [ fig : res:2d : substrate ] shows morphology evolution on three different patterned substrates : neutral , preferentially attracting polymer , and preferentially attracting polymer and fullerene .
we run these tests for polymer of high degree of polymerization @xmath145 ( last column in figure [ fig : res:2d : n ] right ) , which becomes the reference simulation and is repeated in figure [ fig : res:2d : substrate ] ( left ) . _
these results clearly demonstrate that substrate patterning provides additional degree of control over morphology . _
multiple layers observed without patterning can be broken close to the substrate .
the breakage depends on the frequency of the patterning and combination of the material types ( figure [ fig : res:2d : substrate ] middle and right ) . patterning with alternating chemical preference allows for better control of domain size .
the size along the substrate of the polymer - rich induced by such patterning is maintained during evaporation .
this is not the case when patterning is purely of one preference .
the domains created close to the substrate grow in size along the substrate .
however , when substrate is patterned with one chemical preference , domains created close to the substrate penetrate deeper into film , compared to the other case .
, @xmath207 , @xmath160 , @xmath195 , @xmath208 and final.,scaledwidth=18.0% ] in previous subsections , we showed that by independently changing system and process variables we obtain various types of morphologies . in figure [ fig : res:2d : sum ] , we summarize these types of morphologies .
it is important to notice that significantly different morphologies can develop for various system and process variables . in general
, there are three main types of morphologies : percolated morphology ( b ) ; morphology with multiple layers ( c and d ) and morphology with islands ( e ) .
substrate patterning gives an additional means to initiate and direct morphology close to the substrate .
for example , adding substrate patterning leads to breaking the bottom layer , as shown in figure [ fig : res:2d : substrate ] .
_ it is interesting to note that controlling different variables may lead to the same type of morphology .
_ for example , increasing degree of polymerization from @xmath136 to @xmath144 leads to multiple layer creation ( figure [ fig : res:2d : n ] ) .
similar effect is observed by changed solvent ( figure [ fig : res:2d : solvent ] ) .
this emphasize the importance of further systematic studies of solvent - based fabrication .
morphology is a key element affecting the performance of organic solar cells .
the morphology evolution during solvent - based fabrication of organic solar cells is a complex , multi physics process that is affected by a variety of material and process parameters .
a virtual framework that can model 3-d morphology evolution during fabrication of osc can relate these fabrication conditions with morphology
. this will significantly augment organic photovoltaic research which has been predominantly based on experimental trial and error investigations .
such a framework will also allow high throughput analysis of the large phase space of processing parameters , thus yielding considerable insight into the process - structure - property relationships governing organic solar cell behavior that is currently in its infancy . in this work ,
we develop a phase field - based framework to study 3d nanomorphology evolution in the active layer of osc during solvent - based fabrication process .
in particular , we formulate physical and mathematical model that takes into account all important processes that occur during solvent - based fabrication of oscs .
we select phase field method to model the behavior multicomponent system with various driving forces for morphology evolution .
we outline an efficient numerical implementation of the framework , to enable three dimensional analysis of the process .
we showcase our framework by investigating the effect of various process and system variables , that lead to following observations : 1 .
mass biot number expresses the interplay between solvent evaporation from the top surface and diffusion within the thin film . for high biot number ,
evaporation is a dominant process which results in top boundary layer creation enriched in two main components .
consequently , phase separation initiates close to the top and propagates into the film . for low biot number , in turn
, diffusion is a dominant process , solvent removed from the top layer is diffused back from the resulting in homogeneous profile .
thanks to this uniform distribution phase separation initiates and evolve homogeneously within film .
the morphology evolution is affected not only by kinetics through evaporation and diffusion but also by thermodynamics .
in particular , interaction parameters between components and degree of polymerization have a large effect on morphology evolution .
the accessibility of possible configurations provided by the free energy landscape is controlled by system variable such as blend ratio .
small change of blend ratio lead to large variation in morphology evolution .
4 . finally , surface induced phase separation provides another opportunity to locally affect the morphology , by creating additional
sinks in the energy landscape .
we are currently investigating extensions of the framework along three directions : nonhomogeneous evaporation , fluid shear effects ( based on @xcite ) , and crystallization .
this research was supported in part by the national science foundation through teragrid resources provided by tacc under grant number tg - cts110007 and tg - cts100080 .
bg & ow were supported in part by nsf phy-0941576 and nsf ccf-0917202 .
we use euler backward scheme with the heuristic strategy to adjust size of time step @xcite .
this strategy is based on the number of newton s iterations required to solve a nonlinear problem for given time step .
if number of iteration is lower than 20 size of time step is increased by 25% in new time step , otherwise it is reduced to 25% of previous time step . when solution can not be found in 50 iterations ,
such step is rejected and time step decreased by half . in this way , time step is decreased when rare coarsening events occur and increased when morphology evolves slowly . in figure
[ fig : res : ts ] , we show example profile of time step size for various dimensionality considered .
size of time step is adjusted by few orders of magnitude .
efficient time stepping strategy allows to perform simulations that cover several time scales .
in the figure , we show the time scale spanning over two to three orders of magnitude when height decreases by up to 80% . 63 natexlab#1#1[2]#2 , , , , ( ) .
, , , ( ) .
, , , , , , , , , , , ( ) . , , ( ) . .
, , , , , , , , , , ( ) . , , , , , , , ( ) . , , , ( ) .
, , , , , ( ) . . , , , , , , , , ( ) . , , , , , ( ) . , , , , , , ( ) . , , , , , , , , ( ) . , , , , ( ) . , , , , ( ) . , , , , , ( ) . , , , , , , ( ) . , , , , , , ( ) . , , , , , ( ) . , , , , , , , ( ) . , , , , ( ) . , , , , , , , , , , ( ) . , , , , ( ) . , , , . , , , , , ( ) . , , , , , ( ) . , , , ( ) . , , , ( ) .
, , , ( ) .
, , , , , ( ) . , , , , . , , , .
, , , , , , ( ) . , , , ( ) .
, , , , , , ( ) . , ,
( ) . , , ph.d .
thesis , mit , .
, , , , in : , pp . . , , , , ( ) .
, , , , ( ) . , , , , , , , , , , ( ) .
, , , , , , , , , , .
, , , , ( ) . , , , , , ( ) . , , , , ( ) .
, , , , ( ) .
, , , , , ( ) . , , , ( ) .
, , , , ( ) .
, , , ( ) . , , , , , ( ) . , , , , , ( ) .
, , , , , , , , , . .
, , , , , , , , , , , argonne national laboratory , . , , , ( ) .
, , , , , , , ( ) . , , , , , , , , , , ( ) . , , , , , , , , , , ( ) .
, , , , ( ) . , , , , , , , ( ) . , , ( ) .
, , , , , ( ) . | solvent - based thin - film deposition constitutes a popular class of fabrication strategies for manufacturing organic electronic devices like organic solar cells .
all such solvent - based techniques usually involve preparing dilute blends of electron - donor and electron - acceptor materials dissolved in a volatile solvent . after some form of coating onto a substrate to form a thin film , the solvent evaporates .
an initially homogeneous mixture separates into electron - acceptor rich and electron - donor rich regions as the solvent evaporates . depending on the specifics of the blend , processing conditions , and substrate characteristics
different morphologies are typically formed .
experimental evidence consistently confirms that the resultant morphology critically affects device performance .
a computational framework that can predict morphology evolution can significantly augment experimental analysis .
such a framework will also allow high throughput analysis of the large phase space of processing parameters , thus yielding considerable insight into the process - structure - property relationships governing organic solar cell behavior . in this paper
, we formulate a computational framework to predict evolution of morphology during solvent - based fabrication of organic thin films .
this is accomplished by developing a phase field - based model of _ evaporation - induced and substrate - induced phase - separation in ternary systems_. this formulation allows most of the important physical phenomena affecting morphology evolution during fabrication to be naturally incorporated .
we discuss the various numerical and computational challenges associated with a three dimensional , finite - element based , massively parallel implementation of this framework .
this formulation allows , for the first time , to model three - dimensional nanomorphology evolution over large time spans on device scale domains .
we illustrate this framework by investigating and quantifying the effect of various process and system variables on morphology evolution .
we explore ways to control the morphology evolution by investigating different evaporation rates , blend ratios and interaction parameters between components . phase separation , evaporation , cahn - hilliard equation , substrate patterning , organic solar cells , morphology evolution . |
this paper is concerned with stability in the numerical solution of initial - boundary value problems for time - dependent multidimensional diffusion equations containing mixed spatial - derivative terms , @xmath1 here @xmath2 is any given integer and @xmath3 denotes any given real , symmetric , positive semidefinite @xmath4 matrix .
the spatial domain is taken as @xmath5 and @xmath6 .
multidimensional diffusion equations with mixed derivative terms play an important role , notably , in the field of financial option pricing theory . here
these terms represent correlations between underlying stochastic processes , such as for asset prices , volatilities or interest rates , see e.g. @xcite . in this paper
we consider for given @xmath7 $ ] the following useful condition on the relative size of the mixed derivative coefficients , @xmath8 clearly , @xmath9 yields the smallest possible mixed derivative coefficients ; they all vanish in this case .
subsequently , @xmath10 admits the largest possible mixed derivative coefficients , given that matrix @xmath11 is positive semidefinite . in most applications in financial mathematics
it turns out that ( [ gamma ] ) holds with certain known @xmath12 .
more precisely , such @xmath13 can be determined from the pertinent correlation coefficients .
the aim of our present paper is to effectively use this knowledge about @xmath13 in the stability analysis of numerical schemes , discussed below , so as to arrive at weaker sufficient stability conditions than those obtained when considering @xmath10 .
semidiscretization of initial - boundary value problems for multidimensional diffusion equations ( [ pde ] ) leads to initial value problems for very large systems of stiff ordinary differential equations , @xmath14 here @xmath15 is a given real @xmath16 matrix and @xmath17 ( for @xmath18 ) and @xmath19 are given real @xmath20 vectors . for the efficient numerical solution of ( [ ode ] ) , splitting schemes of the alternating direction implicit ( adi ) type form an attractive means .
these schemes employ a splitting of the matrix @xmath15 that corresponds to the different spatial directions , leading to a substantial reduction in the amount of computational work per time step compared to common implicit methods such as crank nicolson .
adi schemes constitute a popular class of numerical methods for multidimensional equations in financial mathematics , where mixed derivative terms are ubiquitous .
originally adi schemes were proposed and analyzed for equations without mixed derivatives , see e.g. @xcite . to formulate adi schemes in the presence of mixed derivative terms , consider a splitting of the matrix @xmath15 into @xmath21 where @xmath22 represents all mixed derivative terms in ( [ pde ] ) and @xmath23 represents the derivative term in the @xmath24-th spatial direction for @xmath25 . split the vector function @xmath26 in an analogous way , @xmath27 and define @xmath28 let parameter @xmath29 be given .
let time step @xmath30 and temporal grid points @xmath31 with integer @xmath32 . for the numerical solution of the semidiscrete system ( [ ode ] )
, we study in this paper four adi schemes , each successively generating for @xmath33 approximations @xmath34 to @xmath35 : 0.5 cm _ douglas ( do ) scheme _
@xmath36 + + _ craig sneyd ( cs ) scheme _ @xmath37 +
+ _ modified craig sneyd ( mcs ) scheme _ @xmath38 + + _ hundsdorfer verwer ( hv ) scheme _ @xmath39 0.5 cm a study of the above four adi schemes shows that the @xmath22 part , representing all mixed derivative terms , is always treated in an _ explicit _ fashion . in line with the original adi idea ,
the @xmath23 parts for @xmath25 are successively treated in an _ implicit _ fashion . in its general form ( [ do ] ) the douglas scheme has been considered e.g. in @xcite .
special instances include the well - known adi schemes of douglas & rachford @xcite , where @xmath40 and @xmath41 , and of brian @xcite and douglas @xcite , where @xmath40 and @xmath42 .
mckee & mitchell @xcite first proposed the do scheme for equations ( [ pde ] ) where a mixed derivative term is present , with @xmath43 .
it is readily shown that if @xmath22 is nonzero then the classical order of consistency of the do scheme is just one , for any given @xmath0 .
the cs , mcs and hv schemes can be viewed as different extensions to the do scheme .
they each perform a second explicit prediction stage followed by @xmath44 implicit correction stages .
the cs scheme was introduced in @xcite and attains a classical order equal to two ( independently of @xmath22 ) if @xmath42 .
the mcs scheme was defined in @xcite and possesses a classical order equal to two for any given @xmath0 .
note that the cs and mcs schemes are identical if @xmath42 . the hv scheme was constructed in @xcite and was proposed for equations with mixed derivative terms in @xcite .
like the mcs scheme , the hv scheme possesses the favorable property of having a classical order equal to two for any given @xmath0 . for the stability analysis in this paper the following linear scalar _ test equation
_ is relevant , @xmath45 with complex constants @xmath46 ( @xmath47 ) .
application of any given adi scheme in the case of test equation ( [ test ] ) gives rise to a linear iteration of the form @xmath48 here @xmath49 is a given , fixed , multivariate rational function and @xmath50 ( @xmath47 ) . the iteration is stable if @xmath51 for the schemes ( [ do ] ) , ( [ cs ] ) , ( [ mcs ] ) , ( [ hv ] ) it is readily verified that @xmath52 , respectively , where @xmath53 for the semidiscretization of ( [ pde ] ) we replace all spatial derivatives by second - order central finite differences on a rectangular grid with constant mesh width @xmath54 in the @xmath55direction ( @xmath56 ) : [ discretization ] ( u_x_ix_i ) _ & & , + ( u_x_ix_j ) _ & & + & & + , ij . here @xmath57 and the unit vectors @xmath58 denote multi - indices , @xmath59 stands for @xmath60 and @xmath61 ( @xmath62 ) are given real parameters with @xmath63 .
the righthand side of ( [ discretization]b ) is the most general second - order finite difference formula for the mixed derivative @xmath64 that is based on a centered 9-point stencil .
if @xmath65 , then ( [ discretization]b ) reduces to the standard 4-point formula @xmath66 in the literature also the choices @xmath67 and @xmath68 are frequently considered .
the aim of this paper is to investigate the stability of the four adi schemes ( [ do])([hv ] ) in the application to semidiscretized multidimensional diffusion equations ( [ pde ] ) where the condition ( [ gamma ] ) on the size of the mixed derivative coefficients is effectively taken into account .
our stability analysis is equivalent to the well - known von neumann ( fourier ) analysis .
accordingly , ( [ pde ] ) is considered with constant diffusion matrix @xmath11 and periodic boundary condition and stability pertains to the @xmath69-norm .
the obtained semidiscrete matrices @xmath23 ( @xmath47 ) are then all normal and commuting and can therefore be unitarily diagonalized by a single matrix . for any given adi scheme ,
one thus arrives at the stability requirement ( [ stab ] ) with @xmath46 eigenvalues of the matrices @xmath23 ( @xmath47 ) . upon substituting discrete fourier modes into ( [ discretization ] )
, it is readily shown that the scaled eigenvalues @xmath70 are given by [ eig ] z_0&=&_ij r_ijd_ij[-_i_j+_ij ( 1-_i)(1-_j ) ] , + z_j & = & -2r_jjd_jj(1-_j ) , j=1,2, ,k . the angles @xmath71 are integer multiples of @xmath72 with @xmath73 the dimension of the grid in the @xmath74direction ( @xmath75 ) .
further , @xmath76 stability results for adi schemes pertinent to ( [ gamma ] ) in the case of @xmath77 were derived by craig & sneyd @xcite for the do and cs schemes and by in t hout & welfert @xcite for the mcs and hv schemes . for
any given scheme , both necessary and sufficient conditions on its parameter @xmath0 were obtained such that the stability requirement ( [ stab ] ) with @xmath78 given by ( [ eig ] ) is unconditionally fulfilled , i.e. , for all @xmath30 and all @xmath79 ( @xmath56 ) . the stability analysis of adi schemes based on ( [ gamma ] ) with _ arbitrary _ given @xmath7 $ ] was recently started in in t hout & mishra @xcite . for the mcs scheme and @xmath43 ,
the useful result was proved here that the lower bound @xmath80 is sufficient for unconditional stability ( whenever @xmath81 ) .
the present paper substantially extends the work of @xcite reviewed above .
section [ main ] contains the two main results of this paper .
the first main result is theorem [ theorem1 ] , which provides for each of the do , cs , mcs , hv schemes in @xmath43 and @xmath82 spatial dimensions a _ sufficient _ condition on the parameter @xmath0 for unconditional stability under ( [ gamma ] ) for arbitrary given @xmath83 $ ] .
the second main result is theorem [ theorem2 ] .
this yields for any of the four adi schemes , any spatial dimension @xmath2 and any @xmath7 $ ] a _ necessary _ condition on @xmath0 for unconditional stability under ( [ gamma ] ) . for each scheme ,
the obtained necessary and sufficient conditions coincide if @xmath43 or @xmath82 .
section [ numexp ] presents numerical illustrations to the two main theorems .
the final section [ concl ] gives conclusions and issues for future research .
in the following we always make the minor assumption that the matrix @xmath84 with @xmath85 ( @xmath56 ) is positive semidefinite .
thus , in particular , @xmath86 for all @xmath87 .
we note that this assumption on @xmath88 is weaker than the corresponding assumption that was made in @xcite .
this section gives two lemmas that shall be used in the proofs of the main results below .
[ lemma2 ] let @xmath89 , @xmath90 be given real numbers with @xmath91 .
consider the polynomial @xmath92 then @xmath93 if and only if @xmath94 the critical points @xmath95 of @xmath96 are given by the equations @xmath97 a straightforward analysis , using @xmath91 , shows that there is precisely one critical point in the domain @xmath98 , and it is given by @xmath99 . inserting this into @xmath96 and rewriting yields @xmath100 hence @xmath101 if and only if the first inequality in ( [ pol ] ) holds .
consider next the polynomial @xmath96 on the boundary of the pertinent domain .
it is clear that @xmath102 if @xmath103 .
next , on the boundary part @xmath104 , @xmath105 , @xmath106 there holds @xmath107 thus @xmath108 ( whenever @xmath104 , @xmath105 ) if and only if the second inequality in ( [ pol ] ) holds . by symmetry ,
the result for the other two boundary parts is the same , which completes the proof .
the subsequent analysis relies upon four key properties of the scaled eigenvalues ( [ eig ] ) .
[ lemma1 ] let @xmath109 be given by ( [ eig ] ) . let @xmath83 $ ] and assume ( [ gamma ] ) holds . then : [ zproperties ] & & all z_j are real , + & & z_j0 _ for _ 1jk , + & & z+z_0 0 , + & & |z_0| _ ij . properties ( [ zproperties]a ) , ( [ zproperties]b ) are obvious .
we consider ( [ zproperties]c ) and ( [ zproperties]d ) .
using @xmath110 and the simple identity @xmath111 one readily verifies that @xmath112 = \mathbf{u}^{\rm t}d\mathbf{u}+\mathbf{v}^{\rm t}(d\circ b)\mathbf{v},\ ] ] where @xmath113 denotes the hadamard product of two matrices and @xmath114 it is well - known that the hadamard product of two positive semidefinite matrices is also positive semidefinite , see e.g. @xcite , and hence , @xmath115 is obtained .
next , using the cauchy
schwarz inequality and @xmath86 , it directly follows that @xmath116 together with ( [ gamma ] ) this yields @xmath117 in the following we shall use the notation @xmath118 ( for @xmath75 ) .
then @xmath119 and by ( [ zproperties]d ) there holds @xmath120 the first main result gives sufficient conditions on @xmath0 for unconditional stability of each of the do , cs , mcs , hv schemes in the application to ( [ pde ] ) for @xmath43 or @xmath82 under condition ( [ gamma ] ) with arbitrary given @xmath13 .
[ theorem1 ] consider equation ( [ pde ] ) for @xmath43 or @xmath82 with symmetric positive semidefinite matrix @xmath11 and periodic boundary condition .
let @xmath7 $ ] and assume ( [ gamma ] ) holds . let ( [ ode ] ) , ( [ splitting ] ) be obtained by central second - order fd discretization and splitting as described in section [ intro ] .
then for the following parameter values @xmath0 the do , cs , mcs , hv schemes are unconditionally stable when applied to ( [ ode ] ) , ( [ splitting ] ) : + + @xmath121 do scheme : @xmath122 @xmath121 cs scheme : @xmath123 @xmath124 mcs scheme : @xmath125 @xmath124 hv scheme : @xmath126 the proofs for the four schemes are similar . in view of this
, we shall consider here the hv scheme and leave the proofs for the do , cs and mcs schemes to the reader .. ] using the properties ( [ zproperties]a)([zproperties]c ) of the scaled eigenvalues , it is readily seen that for @xmath127 the stability requirement ( [ stab ] ) is equivalent to [ condhv ] & & 2p^2+(2p-1)(z_0+z)+(z_0+z)^2 0 , + & & 2p-1+(z_0+z ) 0 .
condition ( [ condhv]a ) is always fulfilled since the discriminant @xmath128 .
subsequently , using ( [ defyj ] ) and ( [ z0plusz ] ) it is easily seen that if there exists @xmath129 such that @xmath130 for all @xmath131 ( @xmath132 ) , then condition ( [ condhv]b ) is fulfilled whenever @xmath133 .
+ + the inequality ( [ kappa ] ) reads @xmath134 and this can be rewritten as @xmath135 hence , the inequality is satisfied if @xmath136 which is equivalent to @xmath137 selecting the rightmost value for @xmath138 , yields that ( [ condhv]b ) holds for @xmath43 whenever @xmath139 the inequality ( [ kappa ] ) reads @xmath140 which , by the identity @xmath141 is equivalent to @xmath142 let @xmath143 , @xmath144 , @xmath145
. then @xmath146 , @xmath147 , @xmath148 and the above inequality can be written as @xmath149 where @xmath150 with @xmath151 hence , if @xmath152 and @xmath153 , then ( [ kappa ] ) is satisfied for @xmath82 .
if @xmath154 , i.e. @xmath155 , then obviously @xmath156 .
next consider @xmath157 .
lemma [ lemma2 ] yields that @xmath153 whenever @xmath158 set @xmath159 . then @xmath160 and @xmath161 it follows that ( [ beta ] ) is equivalent to @xmath162 , which means @xmath163 .
hence , if @xmath164 then ( [ kappa ] ) holds for @xmath82 . selecting the upper bound for @xmath138 , yields that ( [ condhv]b ) is fulfilled for @xmath82 whenever @xmath165 upon setting @xmath10 in theorem [ theorem1 ] the resulting sufficient conditions on @xmath0 for the cs , mcs , hv schemes agree with those in ( * ? ?
2.2 , 2.5 , 2.8 ) .
the above theorem thus forms a proper generalization of results in @xcite .
for the do scheme , the obtained sufficient condition generalizes and improves the corresponding result for this scheme from @xcite . in particular ,
if @xmath10 and @xmath82 , then @xcite yields @xmath166 , whereas theorem [ theorem1 ] yields @xmath167 . in view of theorem [ theorem1 ] , a smaller parameter value @xmath0 can be chosen while retaining unconditional stability if it is known that ( [ gamma ] ) holds with certain given @xmath168 .
this is useful , in particular since a smaller value @xmath0 often yields a smaller , i.e. , more favorable , error constant .
the mcs scheme with the lower bound for @xmath0 given by theorem [ theorem1 ] has been successfully used recently in the actual application to the three - dimensional heston
white pde from financial mathematics , see haentjens & in t hout @xcite .
the following theorem provides , for each adi scheme , a necessary condition on @xmath0 for unconditional stability . as in the foregoing proof
, we consider here the hv scheme and leave the ( analogous ) proofs for the other three adi schemes to the reader .
consider the @xmath4 matrix @xmath174 with @xmath175 and @xmath176 whenever @xmath62 .
clearly , @xmath11 is symmetric positive semidefinite and ( [ gamma ] ) holds .
let @xmath177 , so that @xmath178 .
first , choose the angles @xmath71 in ( [ eig ] ) equal to zero for @xmath179 .
then the eigenvalues @xmath70 are given by @xmath180 , @xmath181 and @xmath182 .
in view of ( [ condhv]b ) , we have @xmath183 whenever @xmath184 , @xmath185 .
this immediately implies @xmath186 . next , choose all angles @xmath71 in ( [ eig ] ) the same , i.e. , @xmath187
. then the eigenvalues @xmath70 are given by z_0&=&-rk(k-1 ) , + z_j & = & -2r(1- ) , j=1,2,
,k , where @xmath188 note that @xmath189 . by ( [ condhv]b ) , it must hold that @xmath190 + 2\theta r ( 1-\cos\phi ) } { 2[\ , 1 + 2\theta r ( 1-\cos\phi)\,]^k-1}\ ] ] whenever @xmath184 , @xmath191 . setting @xmath192 , this yields @xmath193/2+\alpha}{2(1+\alpha)^k-1}\ ] ] for all @xmath194 , @xmath195 .
taking the supremum over @xmath196 , gives @xmath197 \quad { \rm for~all}~~\alpha > 0,\ ] ] where @xmath198 by elementary arguments ( cf .
@xcite ) it follows that the function @xmath199 has an absolute maximum which is equal to @xmath200 given in the theorem . upon setting @xmath10 in theorem [ theorem2 ] , the necessary conditions on @xmath0 for the cs , mcs
, hv schemes reduce to those given in ( * ? ? ?
2.3 , 2.6 , 2.9 ) .
further , for the do scheme and @xmath10 there is agreement with the necessary condition from @xcite .
it is readily verified that , for each adi scheme , the sufficient conditions of theorem [ theorem1 ] and the necessary conditions of theorem [ theorem2 ] are identical whenever @xmath43 or @xmath82 and @xmath201 .
hence , in two and three spatial dimensions , these conditions are sharp .
the interesting question arises whether the necessary conditions of theorem [ theorem2 ] are also sufficient in spatial dimensions @xmath202 . in @xcite it was proved that this is true for the hv scheme and @xmath203 .
it can be seen , however , that the proof from @xcite does not admit a straightforward extension to values @xmath204 .
further , in the case of the do , cs , mcs schemes a proof is not clear at present .
accordingly , we leave this question for future research .
in this section we illustrate the main results of this paper , theorems [ theorem1 ] and [ theorem2 ] .
we present experiments where all four adi schemes ( [ do])([hv ] ) are applied in the numerical solution of multidimensional diffusion equations ( [ pde ] ) possessing mixed derivative terms .
the pde is semidiscretized by central second - order finite differences as described in subsection [ fd ] , with @xmath205 , and the semidiscrete matrix @xmath15 is splitted as described in subsection [ adi ] .
the boundary condition is taken to be periodic ( in this case @xmath206 ) .
figure [ fig1 ] shows for @xmath211 the semidiscrete solution values @xmath212 and @xmath213 displayed on the grid in @xmath214 , so that they represent the exact solution @xmath146 at @xmath215 and @xmath216 .
0.3 cm [ cols="^,^ " , ] exactly the same observations can be made as in the experiment ( above ) for the two - dimensional case .
the large temporal errors for each adi scheme when applied with the smaller value @xmath0 , as seen in the right column of figure [ 3derrors ] , correspond to instability of the scheme .
when applied with their lower bound values @xmath0 , given by theorem [ theorem1 ] , the error behavior for all adi schemes is in agreement with unconditional stability of the schemes .
moreover , in this case a stiff order of convergence is observed that is equal to one for the do scheme and equal to two for the cs , mcs and hv schemes .
in this paper we analyzed stability in the von neumann sense of four well - known adi schemes - the do , cs , mcs and hv schemes - in the application to multidimensional diffusion equations with mixed derivative terms .
such equations are important , notably , to the field of financial mathematics .
necessary and sufficient conditions have been derived on the parameter @xmath0 for unconditional stability of each adi scheme by taking into account the actual size of the mixed derivative terms , measured by the quantity @xmath7 $ ] .
our two main theorems generalize stability results obtained by craig & sneyd @xcite and in t hout & welfert @xcite .
ample numerical experiments have been presented , illustrating the main theorems and also providing insight into the convergence behavior of all schemes . among issues for future research
, it is of much interest to derive sufficient conditions on @xmath0 for unconditional stability of each adi scheme in arbitrary spatial dimensions @xmath202 and arbitrary @xmath7 $ ] .
also , it is of much interest to extend the results obtained in this paper to equations with advection terms .
this leads to general complex , instead of real , eigenvalues , which forms a nontrivial feature for the analysis , cf .
e.g. @xcite .
k. j. in t hout & c. mishra , _ a stability result for the modified craig
sneyd scheme applied to 2d and 3d pure diffusion equations _
, in : numerical analysis and applied mathematics , eds
. t. e. simos et .
al . , aip conf . proc .
* 1281 * ( 2010 ) 20292032 .
k. j. in t hout & b. d. welfert , _ unconditional stability of second - order adi schemes applied to multi - dimensional diffusion equations with mixed derivative terms _ , appl .
numer . math .
* 59 * ( 2009 ) 677692 . | in this paper the unconditional stability of four well - known adi schemes is analyzed in the application to time - dependent multidimensional diffusion equations with mixed derivative terms .
necessary and sufficient conditions on the parameter @xmath0 of each scheme are obtained that take into account the actual size of the mixed derivative coefficients .
our results generalize results obtained previously by craig & sneyd ( 1988 ) and in t hout & welfert ( 2009 ) .
numerical experiments are presented illustrating our main theorems . |
consider a fluid system close to a critical point , for example a fluid near the liquid - gas critical point , a binary liquid near the consolute point , or liquid @xmath7 near the @xmath8point . due to the long correlation length a local perturbation , as provided by the wall of the container or by the surfaces of immersed colloid particles ,
is not screened within a few atomic layers but influences the system over appreciable distances .
this gives rise to a long - range interaction between immersed particles or a particle and a wall , as pointed out by fisher and de gennes @xcite .
the forces may be termed critical casimir forces @xcite on account of the close formal relationship with the usual casimir forces @xcite due to quantum fluctuations of the electromagnetic field .
an interesting feature of the critical casimir effect is that it can be turned on and off by changing the magnitude of the correlation length , e.g. by changing the temperature .
the simplest and most - considered geometry for the critical ( as well as the electromagnetic ) casimir effect is the parallel - plate geometry .
results have been obtained in spatial dimensions @xmath9 by using conformal invariance @xcite and in @xmath10 @xcite . in this paper
we consider _ spherical boundary surfaces _ ,
i.e. spheres immersed in a critical fluid .
many of the general model - independent properties for this geometry are also discussed in a recent letter by burkhardt and one of the present authors @xcite .
the spherical shape is of immediate relevance for colloidal particles @xcite , and besides the planar boundary it is the simplest boundary shape for theoretical analyses .
the reason is that systems right at the critical point display an invariance not only with respect to homogenous dilatations but also with respect to conformal or angle - preserving coordinate transformations @xcite .
the class of special conformal transformations @xcite which exists in arbitrary spatial dimension @xmath6 maps spheres onto spheres .
this has been used to study the critical properties of an infinite system with a single spherical hole ( particle ) and of a finite system with a spherical boundary by conformally mapping results for the half
space @xcite . an infinite system with _ two _ arbitrary non - overlapping spherical holes ( particles )
may be obtained by conformally mapping a highly symmetric finite system which is bounded by two concentric spheres @xcite .
this is shown in fig . 1 and follows from the special conformal transformation @xmath11 which maps the sphere of radius @xmath12 centered at the origin of @xmath13-space onto a plane in @xmath14-space that is perpendicular to @xmath15 and passes through the origin .
two spheres with fixed radii @xmath16 , @xmath1 centered about the origin of @xmath17-space ( and depicted in fig .
1a ) map onto two spherical particles in @xmath18-space , as shown in fig .
the points of intersection @xmath19 with the @xmath20-axis have the coordinates @xmath21 with @xmath22 .
two cases of particular interest will be studied in detail below : * for @xmath23 , the _ two spheres _ in @xmath24-space have _ equal size _ ( ses ) , since @xmath25 and @xmath26 . * for @xmath27 , the left sphere reduces to a plane through the origin , which is denoted by a broken line in fig .
this corresponds to _ one sphere _ with radius @xmath28 near a _ planar wall _ ( spw ) of the container .
one may also consider a single spherical particle immersed in a critical fluid inside a _ spherical container_. this geometry is obtained from the concentric geometry via ( [ cotr ] ) when @xmath12 is chosen _ outside _ the interval @xmath29 .
for @xmath30 the order of the intersection points is @xmath31 , as in the concentric geometry of fig . 1a , and sphere @xmath32 remains inside sphere @xmath33 .
for @xmath34 the order is @xmath35 , and sphere @xmath32 is located outside sphere @xmath33 .
an important parameter of the two - sphere geometry is the _ cross ratio _ of intersection points , which remains invariant under conformal transformations @xcite .
instead of working directly with the cross ratio , it is convenient to introduce the related invariant quantity @xmath36 which can be written as @xmath37 in either of the above geometries . here
@xmath28 and @xmath38 are the radii of spheres @xmath32 and @xmath33 , respectively , and @xmath39 is the distance between their centers . on comparing with the concentric geometry of fig . 1a , where @xmath40 and where @xmath41 and @xmath38 have been denoted by @xmath0 and @xmath1 , respectively
, one finds from ( [ kap2 ] ) @xmath42 with @xmath43 one can also obtain ( [ kap3 ] ) by inserting ( [ inte ] ) into ( [ kapp ] ) , which demonstrates the conformal invariance of @xmath44 since the parameter @xmath12 of ( [ cotr ] ) drops out .
it follows from ( [ kapp ] ) or ( [ kap2 ] ) that @xmath45 corresponds to two particles that touch , while @xmath46 describes the situation where one or both of the radii @xmath47 are much smaller than the remaining lengths in the two - sphere configuration .
now let us turn our attention to the free energy .
for two non - overlapping spherical particles in an infinite critical medium we consider @xmath48 here @xmath49 is the singular part @xcite of the free energy for immersing two particles into an infinitely extended fluid , while @xmath50 is that for immersing a single particle .
@xmath51 determines the casimir force , @xmath52 , and it vanishes as the separation @xmath39 tends to infinity with @xmath53 fixed . for the single particle ( @xmath54 ) in the fluid with spherical confinement ( @xmath55 ) one may consider a similar expression @xmath56 with @xmath57 , where @xmath58 and @xmath59 , the free energy of the spherical container without a particle , are subtracted from the total free energy .
this expression vanishes as @xmath38 tends to infinity with @xmath28 and @xmath39 fixed .
right at the bulk critical point the quantities @xmath51 are invariant under special conformal transformations @xcite and can only depend on the conformal invariant @xmath44 in ( [ kap2 ] ) : @xmath60 it follows from the above arguments that @xmath51 for two particles in the infinite critical liquid and @xmath51 for the single particle in the spherically confined liquid both obey ( [ delf ] ) with the same function @xmath61 .
we note two special cases .
first , for a single sphere with radius @xmath28 and with its closest point at a distance @xmath62 from a planar wall ( spw ) , the casimir interaction is @xmath63 this follows from ( [ delf ] ) by substituting @xmath64 and with @xmath65 .
second , for the concentric geometry ( con ) of fig .
1a , the result reads @xmath66 with @xmath67 given by ( [ kap3 ] ) . due to the high symmetry ,
the concentric geometry is especially convenient for the explicit evaluation of the universal scaling function @xmath61 .
the scaling function @xmath2 depends on both the bulk universality class @xcite and the surface universality classes @xmath68 @xcite of the critical system . for a liquid - gas critical point or a critical consolute point of a binary mixture , @xmath32 and @xmath33 correspond to the ising model with surface magnetic fields @xcite either parallel @xmath69 or antiparallel @xmath70 depending on whether the two surfaces favor the same or different signs of the order parameter .
liquid @xmath71he at the @xmath8point belongs to the bulk universality class of the @xmath72 model , and surfaces are expected to suppress the superfluid order @xcite , corresponding to the surface universality class @xmath73 of the ` ordinary transition ' @xcite .
one may also consider tricritical bulk universality classes both for mixtures of ordinary liquids and of @xmath74he@xmath71he .
it is also interesting to consider the multicritical surface universality class @xmath75 of the ` special transition ' @xcite .
exact results for @xmath2 have been obtained by burkhardt and eisenriegler @xcite for the critical ising model with pairs of surface universality classes @xmath76 , and @xmath70 in spatial dimension @xmath9 . in the present paper
we investigate @xmath2 for @xmath6 near the upper critical dimension @xmath77 . for the combinations @xmath78 , @xmath70 , @xmath79 , and @xmath80 , where the order parameter symmetry is broken
, the leading behavior in the @xmath81-expansion follows from mean - field theory .
this will be considered in sec .
[ 3 ] . here
we make close contact to the work of indekeu et al .
@xcite for the parallel - plate geometry and of gnutzmann and ritschel @xcite for the order - parameter profile in the concentric geometry for @xmath69 . for the symmetry - preserving combinations @xmath82 , @xmath83 , and @xmath84 ,
the leading behavior is that of the gaussian model . since the gaussian model is conformally invariant for all @xmath85 , it is instructive to consider the above pairs of boundaries in general @xmath6
this is carried out in sec .
the parallel - plate result ( krech and dietrich @xcite ) emerges as a special case . the explicit evaluation of the different universal scaling functions @xmath2 in secs .
[ 3 ] and [ 4 ] are preceded in sec . [ 2 ] by a discussion of general relationships involving @xmath2 which hold for all spatial dimensions and universality classes of interest .
we discuss the limits @xmath86 and @xmath46 , which have also been addressed in ref .
@xcite and correspond to @xmath87 and @xmath88 , respectively . for @xmath87 , the scaling functions @xmath2 are determined by the amplitudes of the casimir effect in _ parallel - plate _
geometry @xcite . for @xmath89
, we use the _ small - sphere _ expansion @xcite ( a kind of short - distance expansion ) to relate @xmath2 to the amplitudes of the bulk correlation function and of the profile in the half space of either the order parameter or the energy density , depending on the pair @xmath90 of universality classes .
we also discuss how @xmath4 can be conveniently calculated from the thermal average of the _ stress tensor _ @xcite .
furthermore , we discuss the profiles of thermodynamic densities in the presence of two spheres @xcite . for the particular case of a sphere and a planar wall ( spw )
, we use a short - distance expansion @xcite to relate the behavior of the profile near the wall to the average of the stress tensor at the wall , i.e. to @xmath2 .
all the general relationships are consistent with the explicit results in secs .
[ 3 ] and [ 4 ] .
[ 5 ] contains our conclusions .
here we discuss several important relationships which involve the scaling function @xmath2 of the free energy .
they are of a general character and apply to most of the bulk and surface universality classes and spatial dimensions @xmath6 which are of interest . while conformal coordinate transformations of the two - sphere geometry preserve @xmath44 ( and thus leave @xmath51 invariant ) ,
certain non - conformal transformations preserve the spherical shape of the boundaries but change @xmath44 . in equation ( [ aeq1 ] ) we consider a transformation which increases the distance @xmath39 of the two non - overlapping spheres of fig .
1b keeping @xmath28 and @xmath38 fixed .
this is directly related to the casimir force . in the concentric geometry of fig .
1a one may apply the infinitesimal transformation @xmath91 defined by @xmath92 which increases the radial component of @xmath93 by a fraction @xmath94 if @xmath95 .
choosing @xmath96 between @xmath0 and @xmath1 , one sees that @xmath1 increases while @xmath0 remains unchanged . consequently , @xmath97 and @xmath44 are changed . the effect of these transformations on @xmath51 can be written as an integral over the stress - energy tensor @xmath98 @xcite . in the concentric case
@xmath99 as we show in more detail in appendix a. here @xmath100 is the surface area of the @xmath6-dimensional unit sphere , and @xmath101 is the radial component ( normal to the spherical surfaces ) of the stress tensor at a distance @xmath96 from the center . the thermal average at the right hand side of eq .
( [ deef ] ) is taken at the critical fixed point , where the trace of the stress tensor vanishes @xcite . since @xmath96 is arbitrary in the interval @xmath102 , @xmath103 where we have included a factor @xmath104 for convenience below . in evaluating @xmath61 and the casimir forces in sec .
[ 3 ] and [ 4 ] , we have found it convenient to first calculate the scaling function @xmath105 defined in ( [ tenn ] ) and then integrate the result according to ( [ deef ] ) .
the limit @xmath87 corresponds to the parallel - plate geometry .
the concentric geometry becomes equivalent to the parallel - plate geometry if we let @xmath107 while keeping the width @xmath108 fixed . in this limit @xmath109
tends to zero . with the usual notation @xcite @xmath110 for the casimir free energy per @xmath111 and surface area of one of the two parallel plates @xmath90 , [ lim ] @xmath112 and @xmath113 as @xmath87 .
eq.([limf ] ) also determines the casimir interaction of non - overlapping spheres that nearly touch .
this follows from the form @xmath114^{-(d-1)/2}$ ] of @xmath2 for @xmath115 as implied by ( [ fcon ] ) and ( [ kap3 ] ) . in this geometry
the limiting form reads @xcite @xmath116^{-(d-1)/2}\>.\ ] ] here ( [ delf ] ) and ( [ kap2 ] ) have been used , @xmath117 is the distance between the closest points of the two spherical surfaces ( i.e. @xmath118 in fig .
1b ) , and we assume @xmath119 .
in the limit @xmath120 , the right hand side of ( [ plaw ] ) determines the casimir interaction @xmath121 of a single sphere close to a planar wall @xmath122 . in the other limit @xmath89 , where the radius of the inner sphere is much smaller than that of the outer boundary ,
one may use a short - distance expansion introduced in ref .
@xcite . in this expansion
the boltzmann weight @xmath124 which generates in the critical system a spherical hole @xmath125 with center at @xmath126 and radius @xmath127 is written in terms of local fluctuating quantities ( operators ) @xmath128 and non - fluctuating amplitudes @xmath129 in the form [ sms ] @xmath130 with @xmath131 the relation ( [ sms ] ) holds in thermal averages or correlation functions containing other fluctuating quantities ( operators ) or boundaries @xcite with distances from @xmath126 that are much larger than @xmath127 .
we mention a few examples where ( [ sms ] ) is useful : consider the thermal average @xmath132 of a _ primary _ operator @xmath133 @xcite such as the order parameter or the energy density in an infinite critical system ( bulk ) with a spherical hole @xmath125 . by means of ( [ sms1 ] )
one finds @xmath134 since the bulk average of @xmath135 vanishes . as another example consider the concentric geometry , which may be viewed as a critical system inside a full spherical container ( fc ) with hamiltonian @xmath136 , where a hole @xmath125 is generated by adding @xmath137 .
we are interested in the combination of free energies @xmath138 = -k_bt_c\>{\rm ln}\>\langle 1+\sigma\rangle_{\rm fc}\,,\end{aligned}\ ] ] the singular part of which is the function @xmath139 in ( [ fcon ] ) . as a third example one may use ( [ sms1 ] ) also to study the behavior of @xmath140 in the concentric geometry as @xmath141 .
one finds @xmath142 to first order in @xmath135 .
this relation also applies for @xmath143 , ( see ( [ tenn ] ) ) .
further , the expansion ( [ sms2 ] ) holds also if the distant boundary is not a sphere concentric to @xmath125 but , for example , a planar wall @xcite , and it should also apply slightly away from the critical point @xcite . which quantities @xmath144 appear in ( [ sms2 ] ) depends upon the universality class of the surface of @xmath125 .
in case of the surface universality classes @xmath73 and @xmath75 that do not break the @xmath145 symmetry of the @xmath146-component order parameter @xmath147 at the bulk critical point , all the @xmath144 s are also @xmath145 invariant , and the leading @xmath144 is the energy density @xmath148 @xcite . for symmetry - breaking surfaces
we only consider the case of a one - component order parameter @xmath149 and the class @xmath150 or @xmath151 . in this case @xmath144
s , which are odd and even under @xmath152 both appear in the expansion .
the leading @xmath144 is the order parameter @xmath149 and the leading even @xmath144 is the energy density . in ( [ sms ] )
@xmath127 is assumed to be large on a microscopic scale .
the amplitudes @xmath129 for @xmath153 or @xmath154 can be derived explicitly .
they follow from the known result @xcite for the profile @xmath155 of the primary operators @xmath156 or @xmath154 by comparing with ( [ hole ] ) . from the leading behavior for @xmath157 , @xmath158 for @xmath159 or @xmath160 ,
respectively @xcite . here
@xmath32 denotes the surface universality class of @xmath125 , @xmath161 is the amplitude in the critical profile @xmath162 of @xmath163 in a half space with surface universality class @xmath32 , and @xmath164 is the amplitude of the bulk correlation function @xmath165 which also identifies @xmath166 as the bulk exponent of @xmath163 .
now we may calculate @xmath167 for small @xmath97 as the singular part of ( [ spfc ] ) by using ( [ sms2 ] ) and ( [ iden ] ) , with the result [ frc ] @xmath168 where @xmath159 if both surfaces @xmath90 are symmetry - breaking and @xmath160 if one or both preserve the symmetry of the order parameter .
here we have identified @xmath127 with @xmath0 of fig .
1a and have used the fact that the profile @xmath169 at the center of a full spherical container with radius @xmath1 and surface universality class @xmath33 is given by @xmath170 ( see @xcite ) .
below we will also need the small-@xmath97 limit @xmath171 of the stress - tensor scaling function defined in ( [ tenn ] ) .
this follows either from ( [ frco ] ) by means of ( [ deef ] ) or by inserting ( [ sms2 ] ) and ( [ iden ] ) into ( [ tcon ] ) with @xmath172 and using @xmath173 and @xcite @xmath174 eq.([frco ] ) also implies , via ( [ fcon ] ) and ( [ kap3 ] ) , the form @xmath175 for the scaling function @xmath2 in ( [ delf ] ) if @xmath176 .
this leads to the casimir interaction @xcite [ fli ] @xmath177 between distant spheres ( @xmath178 ) and @xmath179 between a planar wall and a single distant sphere @xmath180 .
here eqs.([kap2 ] ) and ( [ fspw ] ) have been used . in this section
we study the profile @xmath181 as @xmath182 approaches one of the boundaries of the critical system .
the normal component @xcite @xmath183 of the stress - tensor average is _ regular _ near the boundaries @xcite . for the concentric geometry , this is apparent from ( [ tenn ] ) , and it holds quite generally @xcite .
the behavior of the order parameter and the energy density is quite different , however , and is described by power - law exponents . in general these are determined by the scaling dimensions of surface operators and are different from bulk exponents @xcite .
here we consider the particularly simple cases where @xmath163 approaches a _
boundary with universality class @xmath33 and @xmath184 is either @xmath185 , @xmath186 , or @xmath187 .
then the leading surface operator is @xmath188 @xcite , which leads to a surface exponent @xmath6 @xcite .
the corresponding short - distance expansion about the planar surface @xcite has the form @xcite @xmath189\>.\ ] ] the planar surface is located in @xmath24-space at @xmath190 .
the factor in front of the square bracket is the critical half
space profile as in ( [ ampl ] ) for surface universality class @xmath33 .
@xmath191 is a universal short - distance amplitude which generally depends on @xmath33 @xcite .
its behavior is known in the gaussian model with @xmath85 and in the @xmath192-theory near the upper critical dimension @xmath193@xcite .
the ellipses inside the square brackets in ( [ shor ] ) denote contributions of higher order in @xmath194 . the expansion ( [ shor ] ) holds in thermal averages or correlation functions where besides the boundary at @xmath190 and @xmath163 other boundaries and operators may be involved .
we are particularly interested in the small @xmath194 behavior of @xmath195 in the sphere - near - planar - wall ( spw ) geometry , which has been introduced as case ( ii ) just below eq.([inte ] ) and in eq.([fspw ] ) . for the stress tensor [ tsp ] @xmath196^{d/2}}{\left[{\rm
\bf r}_{\parallel}^{\,\prime\,2}+d\left(d+2r_1\right)\right]^d}\>y(\rho ) \>,\ ] ] with @xmath197/r_1\ ] ] and @xmath62 as defined in ( [ fspw ] ) . here
we have used the general transformation formula @xcite [ tra ] @xmath198^{-d}\>\langle t_{nn}(r)\rangle_{\rm \scriptscriptstyle con}\ ] ] with the local scale factor @xmath199 setting @xmath200 leads to the spw geometry on the left hand side of ( [ ttra ] ) with @xmath201 which follows from ( [ inte ] ) . using the form ( [ tenn ] ) for @xmath202 on the right hand side of ( [ ttra ] ) and letting @xmath203 so that @xmath204 leads to ( [ tspw ] ) .
eq.([rhod ] ) follows e.g. from ( [ kap3 ] ) , since the quantity @xmath44 from ( [ kap2 ] ) equals @xmath205 in the spw case .
we close this section by introducing some of the tools needed in secs . [ 3 ] and [ 4 ] .
critical systems with surfaces may be described @xcite by a configuration probability @xmath206 for the fluctuating @xmath146-component order parameter @xmath207 , where @xmath208 separates into a bulk part @xmath209 and surface parts @xmath210 .
while the integration in ( [ hbul ] ) is over the interior of the critical system , e.g. over @xmath211 in the concentric geometry , the surface contributions involve integrals ( of powers of @xmath149 and derivatives thereof ) over the surfaces of the system .
since we shall only consider systems right at the bulk critical point , we have not included a @xmath154 term in ( [ hbul ] ) @xcite .
we will also need the explicit form of the stress tensor at an arbitrary interior point of the system .
it is given by @xcite [ tkl ] @xmath212 with the canonical tensor supplemented by the improvement term @xmath213\,\roarrow\phi^{\ > 2}\>.\ ] ] the quantity denoted by @xmath214 is needed to renormalize the stress tensor @xcite . since it is of third or higher order in the renormalized coupling @xmath215 ( compare eq .
( [ relu ] ) below ) , it does not appear in our explicit low - order calculations .
the thermal average of the stress tensor in the concentric geometry takes the rotationally - invariant form @xmath216\>.\ ] ] here @xmath217 are dimensionless quantities which only depend on the position via @xmath218 .
a fundamental property of the stress tensor is the continuity equation @xcite . in the present case @xmath219 & = & r_k\,r^{-d-2}\left[-\left(\tau^{(1)}-\tau^{(2)}\right ) + r\,\frac{\mbox{d}}{\mbox{d}\ , r}\left(\tau^{(1)}- \frac{1}{d}\;\tau^{(2)}\right)\right]\>,\end{aligned}\ ] ] which implies the important relation @xmath220 between the @xmath221-dependence of the radial component and the trace of the stress tensor .
eqs.([cont ] ) and ( [ trel ] ) hold for arbitrary @xmath222 and for arbitrary ( rotationally - invariant ) conditions at the surfaces .
if the trace vanishes , as happens @xcite if @xmath223 is at the critical fixed point of the renormalization group ( rg ) , eq.([trel ] ) implies that @xmath224 is independent of @xmath221 .
this is consistent with eq .
( [ tenn ] ) . to set up the rg
, we use reparametrization by minimal subtraction of poles in @xmath81 @xcite .
we note in particular the relation @xmath225\ ] ] between the @xmath192 coupling constant @xmath226 and its renormalized counterpart @xmath215 . here
@xmath227 is the inverse length scale that determines the renormalization flow , and the @xmath228-contributions contain pole terms in @xmath81 .
the stress tensor ( [ tkl ] ) is already renormalized @xcite , and quantities such as @xmath229 in eq .
( [ aeq9 ] ) or the functions @xmath230 with @xmath231 from ( [ troi ] ) are finite for @xmath232 order by order in @xmath215 .
( [ tau1 ] ) follows from dimensional analysis , assuming that the boundary condition @xmath233 take their fixed - point form and do not introduce any new lengths .
if , in addition , the coupling @xmath215 equals its critical fixed - point value @xmath234 the renormalization - group equation @xcite implies that the @xmath235 are independent of @xmath236 . by eqs .
( [ troi])([trel ] ) and the vanishing trace , the @xmath235 are also independent of @xmath237 and @xmath238 .
thus , we find that @xmath239 only depends on @xmath97 , @xmath6 and the surface universality classes @xmath233 .
this provides another derivation of ( [ tenn ] ) .
in this section we consider the surface universality classes @xmath240 and @xmath241 , where the order parameter symmetry is _ broken _ at the bulk critical point . with applications to the liquid - gas or consolute critical points in mind , we only consider a one - component order parameter .
in the standard approach , the order parameter @xmath242 is decomposed into a non - vanishing average @xmath243 and the fluctuation @xmath244 around it
. then a fluctuation ( or loop ) expansion is carried out , which amounts to an expansion in powers of @xmath226 . to leading order
one may neglect fluctuations altogether and determine @xmath245 by minimizing @xmath246 and ignore @xmath244 . in the concentric geometry
@xmath247 only depends on the distance @xmath221 from the center and @xcite @xmath248 with @xmath249 which has the dimension of an inverse length . for the trace of the stress tensor one finds @xmath250 while the radial component is given by @xmath251 where @xmath252 here @xmath253 , and the superscript zero denotes the @xmath254 contribution to @xmath255 , which is obtained by neglecting fluctuations .
the first and second term on the right hand side of ( [ fiin ] ) come from the canonical tensor and the improvement term , respectively .
the quantities ( [ trst ] ) and ( [ trad ] ) determine the @xmath254 contributions to the right and left hand side of eq.([trel ] ) , respectively , and allow one to explicitly check ( [ trel ] ) to this order .
note that for @xmath232 with @xmath256 fixed the trace vanishes .
thus @xmath257 is independent of @xmath221 and represents a first integral of the differential equation ( [ mfeq ] ) if @xmath193 .
this integral was encountered in ref .
@xcite , where it is pointed out that eq.([mfeq ] ) is a special case of the generalized emden
fowler ( gef ) equation @xcite . for @xmath193 , ( [ mfeq ] ) belongs to a subclass of the gef equation , for which the first integrals are known explicitly @xcite . a first integral for the mean - field equation in the spherically symmetric geometry
may also be constructed by means of the stress tensor if @xmath258 in ( 2.20 ) is replaced by an arbitrary monomial interaction @xmath259 .
the leading contribution in @xmath226 of the stress tensor always has a vanishing trace at the upper critical dimension @xmath260 ( see ref .
all the corresponding mean - field equations belong to the above - mentioned subclass .
now we assume that the boundary conditions for @xmath261 at @xmath262 and @xmath263 take their fixed - point forms
. then the dimensionless quantity @xmath264 in ( [ fiin ] ) can only depend upon two ratios , say @xmath265 of the three lengths @xmath266 .
in particular , for @xmath193 @xmath267 i.e. , @xmath257 depends only on @xmath97 .
note that the term in @xmath268 of leading order in @xmath215 , which follows from substituting ( [ twor ] ) and ( [ relu ] ) into ( [ trad ] ) , does indeed have the form ( [ tau1 ] ) .
the scaling function @xmath269 in ( [ tenn ] ) follows from @xmath270 for @xmath271 . substituting ( [ relu ] ) , ( [ usta ] ) and ( [ orat ] ) into ( [ trad ] ) and keeping only the leading @xmath81-dependence
, one finds @xmath272 to obtain solutions @xmath261 in @xmath193 with the first integral @xmath273 , we follow the procedure in ref . @xcite . with the substitution @xmath274 eq.([fiin ] ) becomes [ dif ] @xmath275 with @xmath276 and solutions can be obtained in terms of elliptic integrals .
+ for @xmath277 the order parameter profile @xmath261 tends to @xmath278 near the two surfaces and has one minimum in between .
this implies @xmath279 and a solution with the parametric form @xcite [ sol1 ] @xmath280 @xmath281^{1/2}\right)\>,\ ] ] where @xmath49 is the elliptic integral @xmath282 and @xmath283 here @xmath284 is the radius with minimum @xmath285 , corresponding to @xmath286 . at @xmath287 where the profile diverges
, @xmath221 takes the value @xmath288 .
eq.([pr++ ] ) implies @xmath289 and [ f1 ] @xmath290 with @xmath291^{1/2}\right ) \>.\ ] ] here @xmath97 is from ( [ rhof ] ) , and @xmath292 is the complete elliptic integral @xmath293 eqs.([f1 ] ) , ( [ gamm ] ) , and ( [ yrho ] ) determine the scaling function @xmath269 for the stress tensor in ( [ tenn ] ) .
note that @xmath294 varies from @xmath295 to @xmath296 as @xmath97 varies from @xmath295 to @xmath297 .
+ the solution for @xmath298 or @xmath80 is readily obtained from the solution in ( i ) .
identifying @xmath299 in ( [ pr++ ] ) with the radius of the @xmath75 boundary , i.e. considering [ rre ] @xmath300 or @xmath301 leads to the relation @xmath302 which only differs from ( [ f1ln ] ) by a factor @xmath303 .
the reason for the identifications ( [ rre ] ) can be understood by conformally transforming the mean - field ( mf ) solution ( [ sol1 ] ) for @xmath69 in concentric geometry to spheres of equal size ( ses ) in @xmath24-space , with the mid - plane @xmath190 corresponding to the sphere with @xmath284 ( see the remark just below eq.([inte ] ) ) .
the transformed mf solution is invariant under @xmath304 , and the normal derivative vanishes at @xmath305 .
since the mf equation @xmath306 is conformally invariant in @xmath193 , this is the correct mf solution for a half
space @xmath307 in @xmath193 with a planar @xmath75 surface at @xmath305 and a spherical hole with @xmath150 boundary conditions . transforming this solution back to
the concentric geometry leads to ( [ rre ] ) and ( [ f1sb ] ) .
note that for the two geometries ( [ rre ] ) the radial derivative @xmath308 at the @xmath75 surfaces does not vanish but equals @xmath309 and @xmath310 for ( [ rre1 ] ) and ( [ rre2 ] ) , respectively @xcite .
thus the mf order - parameter profile increases ( decreases ) on approaching the convex ( concave ) @xmath75 surface from the interior of the critical system , which is reasonable .
the mf values @xmath311 for the ratio of the logarithmic normal derivative and the inverse radius of curvature in @xmath193 are consistent with results in ref .
@xcite . + for @xmath312
, the profile tends to @xmath313 for @xmath314 and has a zero in between .
this implies @xmath315 .
though the @xmath294-dependence of @xmath316 is perfectly regular at @xmath317 , it is advantageous to choose different parametric representations for the intervals @xmath318 and @xmath319 , respectively .
they are given by [ sol2 ] @xmath320 with @xmath321^{1/2}\right)\ ] ] for @xmath318 ( with @xmath322 from ( [ gamm ] ) ) and by [ sol3 ] @xmath323 with @xmath324^{1/2}\right)\ ] ] for @xmath319 .
the profile diverges at @xmath325 , corresponding to @xmath288 , and has a zero at @xmath286 , corresponding to @xmath326 .
eqs.([sol22 ] ) and ( [ sol32 ] ) determine the relation @xmath327 .
a form which holds for both @xmath318 and @xmath319 is [ f3 ] @xmath328 with @xmath329 here @xmath330^{1/2}\>.\ ] ] + the solution for @xmath331 follows from that in case ( iii ) by identifying @xmath332 with the radius @xmath0 of the @xmath73 boundary , where the order parameter vanishes @xcite .
this implies @xmath333 the four functions @xmath327 in eqs.([f1 ] ) , ( [ f1sb ] ) , ( [ f3 ] ) , and ( [ f3do ] ) determine via ( [ yrho ] ) the scaling function @xmath269 of the stress tensor and via ( [ delf ] ) , ( [ fcon ] ) , and ( [ deef ] ) the free energy and the casimir forces for the four pairs @xmath69 , @xmath334 , @xmath335 , @xmath336 of boundary conditions . integrating according to ( [ deef ] ) yields the parametric form @xmath337 for the scaling function , with @xmath338 from ( [ kap3 ] ) and @xmath327 defined as above .
on varying @xmath97 or @xmath44 , @xmath2 , @xmath339 , and @xmath294 remain negative in the cases ( i ) and ( ii ) and positive in the cases ( iii ) and ( iv ) , implying attractive and repulsive interactions , respectively .
the remaining integration in ( [ fkin ] ) has been carried out numerically .
results for the scaling function for the four combinations of boundary conditions are shown in fig . 2 .
since the prefactor in ( [ fkin ] ) behaves like @xmath340 as @xmath341 , we have plotted the finite quantity @xmath342 .
it is instructive to check that our solutions exhibit the limiting behavior discussed in subsecs .
[ 2b ] and [ 2c ] .
consider first the limit @xmath87 , where @xmath343 . in this case
@xmath344 substituting the corresponding @xmath345 into ( [ yrho ] ) , one finds that @xmath269 does indeed have the behavior ( [ limy ] ) with @xmath193 and that [ de ] @xmath346 ^ 4\ ] ] and @xmath347 the expressions in ( [ de ] ) reproduce the known results @xcite for the casimir amplitudes in the parallel - plate geometry for the symmetry - breaking cases and @xmath6 near @xmath348 .
next consider @xmath89 , where @xmath349 . in this case
[ li ] @xmath350 which implies @xmath351\ ] ] in cases ( i)-(iv ) . in order to compare with ( [ ylim ] )
we need the expressions @xmath352\>\ > , \quad d\rightarrow 4\ ] ] for @xmath353 in our four cases . here
$ ] as noted just below eq.([frco ] ) .
the results ( [ dra4 ] ) follow from the behavior @xmath355 to leading order in @xmath226 of @xmath356 , @xmath357 in the half
space with @xmath358 or @xmath151 ( see eq .
( [ ampl ] ) ) , from the corresponding behavior @xmath359 of @xmath357 in the half
space with an @xmath360 or @xmath75 surface , and from the leading - order contributions @xmath361 to the bulk correlation functions @xmath362 , @xmath363 ( cf .
( [ corr ] ) ) .
here @xmath364 and @xmath226 must be replaced by ( [ relu ] ) with the fixed - point value ( [ usta ] ) with @xmath365 for @xmath215 . using the results ( [ limi ] ) and ( [ dra4 ] ) , one finds by means of eq.([yrho ] ) and from @xmath366 , @xmath367 for @xmath341 that
our four solutions are indeed consistent with ( [ ylim ] ) .
now we turn to the density profiles and their behavior near a boundary .
consider first the @xmath203 behavior of @xmath261 in the concentric geometry for @xmath193 with universality class @xmath368 for the boundary at @xmath1 .
this follows directly from the differential equation ( [ dife ] ) , which implies @xmath369 here we have used @xmath370 and assumed that @xmath285 is larger than a possible minimum in @xmath371 . for @xmath221 close to @xmath1 , @xmath285 tends to infinity , and one may expand the integrand in ( [ lnrr ] ) in powers of @xmath372 .
the integration is elementary , and one finds @xmath373\>.\ ] ] the first factor on the right hand side is the profile inside a full spherical container with radius @xmath1 , which follows from ( [ lnrr ] ) with @xmath374 .
the second term in square brackets respresents the leading correction for @xmath203 due to the inner spherical boundary @xmath32 with radius @xmath0 .
the prefactor @xmath375 denotes @xmath294 in the above cases ( i ) , ( ii ) , ( iii ) , or ( iv ) , with @xmath358 , @xmath75 , @xmath151 , or @xmath73 ( compare eq .
( [ limi ] ) ) .
the result ( [ fidg ] ) also follows from the parametric representations ( [ pr++ ] ) , ( [ sol22 ] ) , ( [ sol32 ] ) for @xmath376 . the corresponding expansion near the planar wall with universality class @xmath368 in the spw geometry introduced below ( [ inte ] ) follows from substituting ( [ fidg ] ) into the transformation formula @xmath377^{-1}\>m_{a,\uparrow}(r)\ ] ] for mean - field solutions in @xmath193 . here @xmath378 is from ( [ losc ] ) with @xmath200 .
this leads to the result @xmath379 with @xmath380\>,\ ] ] for the leading behavior of @xmath381/r_+$ ] as @xmath382 .
here we use the relation between the profiles @xmath383 in half space and @xmath384 inside a full spherical container that follows from the conformal transformation ( [ trfo ] ) @xcite . using eqs.([tspw ] ) , ( [ rhod ] ) and ( [ yrho ] ) for @xmath385 and the result @xcite @xmath386 for the universal short - distance amplitude , we have checked that eqs .
( [ expw ] ) , ( [ calz ] ) are fully consistent with the short - distance expansion ( [ shor ] ) when applied to a @xmath387 average for @xmath341 . in the other cases such as @xmath388 with @xmath358 , the short distance expansion ( [ shor ] ) about the planar wall with universality class @xmath33 can be checked in a similar way @xcite .
the short - distance coefficient @xmath389 turns out @xcite to be of order @xmath390 ( see eq .
( [ cphi ] ) below ) .
as a consequence , the first term in square brackets in ( [ shor ] ) leads to a contribution which is smaller by a factor @xmath81 than the contribution from the second term and is not present in the mean - field solution . in appendix [ b1 ] we evaluate density profiles for @xmath141 for arbitrary fixed @xmath391 .
for @xmath87 , the density profile in the parallel - plate geometry @xcite is obtained .
this has been shown for @xmath392 in ref.@xcite , and similar arguments apply for other @xmath68 .
the boundary conditions @xmath393 do not destroy @xcite the bulk @xmath145 symmetry of the @xmath146-component order parameter at the bulk critical point . in these cases
the starting point of the @xmath10 expansion is @xcite the critical gaussian model .
this is defined by eq.([hbul ] ) with @xmath394 .
it is instructive @xcite to consider the critical gaussian model for arbitrary @xmath85 .
for all @xmath6 it is at its critical fixed point , i.e. dilatation and conformally invariant on all length scales .
this also holds in the presence of boundaries @xmath32 and @xmath33 if the boundary conditions take their fixed - point forms . in the concentric geometry the gaussian order - parameter correlation functions or ` propagators ' with @xmath395 , @xmath396 only depend on the positions @xmath397 through their distances @xmath398 from the center and the cosine of the enclosed angle [ pro ] @xmath399 here @xmath238 and @xmath400 denote components of the order parameter .
one finds @xmath401 with [ def ] @xmath402 and @xmath403 the @xmath404 are hyperspherical ( gegenbauer ) polynomials @xcite , and @xmath405 and @xmath406 the quantities @xmath407 depend on the ratio @xmath408 of the radii of concentric boundaries and are given by @xmath409^{-1}\nonumber\\*[2 mm ] & w_{o , sb}^{(l)}\:&=\:\left[p^{-1}+p\right]^{-1}\nonumber\\*[2 mm ] & w_{sb , o}^{(l)}\:&=\:-\left[p^{-1}+p\right]^{-1}\nonumber\\*[2 mm ] & w_{sb , sb}^{(l)}\:&=\:\left[p^{-1}-p\right]^{-1}\>,\end{aligned}\ ] ] with @xmath410 the @xmath411 remain unchanged for @xmath412 @xmath413 @xmath414 and satisfy @xmath415 with @xmath416 and @xmath417 the laplacian and the dirac @xmath417-function , respectively , in @xmath6 dimensions .
note that @xmath418 and @xmath419 vanish at the boundaries where @xmath420 .
this implies the fixed - point boundary conditions @xmath421 and @xcite @xmath422 here the first and second relations in ( [ fibc ] ) and ( [ bcsb ] ) hold for @xmath423 and @xmath424 , respectively .
a conformal transformation @xmath425^{-\vartheta } g_{ab}\left({\rm \bf r}_1,\,{\rm \bf r}_2\right)\ ] ] to the sphere near the planar wall ( spw ) geometry shows that ( [ bcsb ] ) is equivalent to the usual neumann boundary condition for a planar @xmath75 surface : consider e.g. ( [ cotr ] ) with @xmath200 , which transforms the outer concentric boundary @xmath426 to the plane @xmath190 ( see fig . 1 ) .
to prove that the derivative vanishes at @xmath427 for @xmath428 , we show that the continuation in @xmath429 of @xmath430 across the boundary is even in @xmath429 .
the points corresponding to @xmath431 ) and its mirror image @xmath432 are @xmath433 and @xmath434 respectively , as follows from inversion about the sphere @xmath263 . since @xmath435 and @xmath436 eq .
( [ coga ] ) shows that @xmath430 is indeed even in @xmath429 .
an inversion about the sphere @xmath437 exchanges the boundaries @xmath32 and @xmath33 and implies @xmath438 one easily verifies that @xmath439 , @xmath440 in ( [ prex ] ) are consistent with ( [ geba ] ) .
@xmath441 in ( [ prex ] ) should be compared with the propagator in the unbounded bulk @xmath442 this equation follows from @xmath441 in ( [ prex ] ) in the limits @xmath141 , @xmath107 with @xmath443 fixed .
below we will need the difference @xmath444^{-1}\cdot\nonumber\\ & & \left\{\frac{r_1^l}{r_2^{\vartheta+\lambda}}+ \frac{r_2^l}{r_1^{\vartheta+\lambda}}+ \sigma_{ab}^{(2)}\:\frac{(r_1r_2)^l}{r_-^{2\lambda } } + \sigma_{ab}^{(3)}\frac{r_+^{2\lambda}}{(r_1r_2)^{\vartheta+\lambda } } \right\}\>,\end{aligned}\ ] ] which is regular for @xmath445 at interior points . here
@xmath446 \sigma_{ab}^{(2)}&=&(-1,-1,1,1)\nonumber\\*[2 mm ] \sigma_{ab}^{(3)}&=&(-1,1,-1,1)\end{aligned}\ ] ] for @xmath447 in order to calculate @xmath448 in the gaussian model , we use ( [ tekl ] ) with @xmath394 in the hamiltonian ( [ hbul ] ) .
since the average @xmath449 in the unbounded bulk vanishes , we evaluate the averages @xmath450 and @xmath451 that arise from eq.([tekl ] ) by using @xmath452 from ( [ dege ] ) instead of @xmath441 .
this yields @xmath453^{-1 } { d-3+l \choose l } \left\{2l\,(d-2+l)+\frac12 ( d-2)^2\right\}\>.\end{aligned}\ ] ] using ( [ deef ] ) , one eventually obtains @xmath454 where the relation between @xmath44 and @xmath97 is given in ( [ kap3 ] ) .
the signs of both @xmath2 and @xmath455 are opposite to the sign of @xmath456 .
this holds for all values of @xmath44 or @xmath97 .
thus the interaction is always attractive if @xmath457 and repulsive if @xmath458 .
we have evaluated numerically the sum in ( [ nmof ] ) .
the function @xmath459 for @xmath193 is shown in fig . 2 .
the profile of the energy density @xcite is obtained from @xmath452 .
it takes the form @xmath460^{-1 } { d-3+l \choose l } \left\{2+\sigma_{ab}^{(2)}\;\left(\frac{r}{r_-}\right)^{2\lambda } + \sigma_{ab}^{(3)}\;\left(\frac{r_+}{r}\right)^{2\lambda}\right\}\>.\end{aligned}\ ] ] note the different sources of @xmath221-dependence in the expression ( [ endi ] ) for the energy density .
the @xmath221-dependence is very similar in each of the contributions to @xmath461 coming from eqs.([tkl ] ) .
@xmath462 and @xmath463 cancel in the sum , and @xmath461 in ( [ tgau ] ) turns out proportional to @xmath464 , in agreement with ( [ tenn ] ) .
the @xmath465 terms with @xmath466 parallel or perpendicular to @xmath17 that arise from the canonical tensor in ( [ tekl ] ) contribute half of the first term in the curly brackets on the right hand side of ( [ tgau ] ) , while the second term comes from the improvement @xmath467 in ( [ impr ] ) .
the sum of the two terms in curly brackets equals @xmath468 . from ( [ tgau ] ) and ( [ endi ] )
, one may again check that the two combinations @xmath469 and @xmath470 are related by an inversion about the sphere @xmath471 .
this operation preserves @xmath461 , while in @xmath472 the ratios @xmath473 and @xmath474 are exchanged .
note that @xmath461 is the same for @xmath475 and @xmath476 , too .
however , the latter identity is limited to the gaussian model , as we argue below .
we now consider the limiting behavior of the stress tensor in ( [ tgau ] ) . according to the discussion in subsec .
[ 2b ] , @xmath461 should approach the parallel - plate result when @xmath477 tends to @xmath297 . in this case
@xmath478 the exponential decay of the first factor in ( [ tgau ] ) is very slow , and the sum in ( [ tgau ] ) may be replaced by an integral @xmath479^{-1 } \;2\:l^{\,d-1}/\gamma(d-2)\>.\ ] ] here @xmath480 is from ( [ plim ] ) with @xmath8 replaced by @xmath481 , and the second and third factors in the sum of ( [ tgau ] ) have been replaced by their limiting forms for @xmath482 .
the integral in ( [ suli ] ) can be expressed in terms of the riemann zeta function @xmath483 @xcite .
comparison with eqs.([tenn ] ) and ( [ limy ] ) shows that @xmath461 does indeed have the expected limiting form , with the well - known casimir amplitudes @xcite [ caa ] @xmath484 and @xmath485 for the @xmath146-component gaussian model bounded by plates of type @xmath73 or @xmath75 . in the other limit @xmath89 the leading behavior of ( [ tgau ] ) comes from the @xmath486 term in the sum , which is given by @xmath487 this
should be compared with the amplitude ratios @xmath488 for the combinations ( [ abde ] ) of @xmath32 and @xmath33 .
these follow from eqs.([cobe ] ) and ( [ dokn ] ) . comparing with ( [ tenn ] ) and ( [ tisd ] ) , one obtains the expected limiting behavior ( [ ylim ] ) with @xmath489 .
we conclude the discussion of @xmath490 in the limits @xmath87 and @xmath89 with two remarks : ( i ) only the canonical part of @xmath98 , i.e. the first term in curly brackets in ( [ tgau ] ) , contributes in the limit @xmath87 .
this is to be expected , since the average of the improvement term , which contains a total derivative parallel to the plates , vanishes due to translational invariance in the parallel - plate geometry .
it is less obvious that the leading behavior for @xmath89 is only due to the improvement term ( second term in curly brackets in ( [ tgau ] ) ) , as we have noted in eq.([rho0 ] ) .
( ii ) for a normal critical system the @xmath192-interaction of ( [ hbul ] ) must be included
. then @xmath461 and thus @xmath105 and @xmath2 for @xmath475 and @xmath476 are no longer the same . from ( [ ylim ] )
there should already be a deviation in first order in @xmath5 for @xmath89 , since @xmath491 however , for @xmath87 , i.e. in the parallel - plate geometry there is no deviation in this order @xcite .
now we turn to the profile ( [ endi ] ) of the energy density .
this is the counterpart of the discussion at the end of sec . [
3 ] . for @xmath221
fixed and @xmath141 only the second term in curly brackets in eq.([endi ] ) survives , and the concentric profile tends to @xmath492 & & = \sigma_{ab}^{(3)}\;\tilde s_d\;\left\{r_+\,\left[1- \left(\frac{r}{r_+}\right)^2\right]\right\}^{-(d-2)}=n^{-1 } \,\langle \roarrow \phi^{\>2}({\rm \bf r})\rangle_{\rm fc}\>,\end{aligned}\ ] ] which is the profile inside a full spherical container ( fc ) with radius @xmath1 and universality class @xmath33 ( see the corresponding general expression in ref.@xcite ) . below
we will need the difference @xmath493= } \nonumber\\*[3 mm ] & & \hspace{5 mm } r^{2-d}\,\tilde s_d \sum_{l=0}^{\infty}\left[\sigma_{ab}^{(1)}\,p^{-2}-1\right]^{-1 } { d-3+l\choose l } \left\{2+\sigma_{ab}^{(3)}\;\left[\left(\frac{r}{r_+}\right)^{2\lambda } + \left(\frac{r_+}{r}\right)^{2\lambda}\right]\right\}\>.\end{aligned}\ ] ] note that the second term in curly brackets is different from that in ( [ endi ] ) . here
we study the behavior for @xmath203 with @xmath494 , @xmath495 . in this case
the expression in curly brackets in ( [ eddi ] ) tends to @xmath496 ^ 2 $ ] , and the sum over @xmath481 can be expressed in terms of @xmath497 in ( [ tgau ] ) . using ( [ tenn ] )
, this leads to @xmath498\ ] ] for @xmath203 and @xmath494 .
this is the counterpart of ( [ fidg ] ) . as in eqs.([trfo])@xmath499([sham ] ) , one may transform to an spw geometry and identify the leading coefficient @xmath389 in the short - distance expansion ( sde ) about the planar wall with @xmath494 .
now exponents @xmath500 and @xmath6 instead of @xmath501 and @xmath348 appear in eqs.([trfo ] ) and ( [ expw ] ) , respectively , and @xmath502 this result , obtained here from the @xmath503 geometry with @xmath360 or @xmath75 , is consistent with the result for @xmath504 geometry for @xmath341 , discussed in the paragraph after eq.([sham ] ) .
it is also in agreement with refs .
@xcite , where the sde of @xmath154 about an @xmath73 surface has been applied to the profile in the parallel - plate geometry and in the half space at @xmath505 and to correlation functions in the half space at @xmath506 . in ref.@xcite
@xmath389 for the gaussian model was called @xmath507 .
eq.([eddi ] ) can also be used to evaluate the density profiles for arbitrary fixed @xmath391 as @xmath141 , and one may check the predictions of the small - sphere expansion .
this is done in appendix [ b2 ] .
finally , we mention that for @xmath87 the profiles ( [ endi ] ) of the energy density reduce to the gaussian results in the parallel - plate geometry @xcite . this may be shown with arguments analogous to those near eqs .
( [ plim ] ) and ( [ suli ] ) .
we have analyzed the casimir interaction between two spherical particles in a fluid at the critical point with boundary conditions @xmath508 the casimir interaction is completely determined by a universal function @xmath509 of a single variable @xmath44 ( see eq .
( [ kap2 ] ) ) , which is a combination of the three lengths that characterize the configuration of two spheres .
we have discussed the limiting behavior for spheres that nearly touch @xmath510 and spheres that are far apart from each other @xmath511 and calculated the complete functions @xmath512 for the case where the spatial dimension @xmath6 is near 4 @xcite ( see fig . 2 ) . from these results and corresponding information for @xmath9 and @xmath513 @xcite , one can make fairly reliable estimates of the scaling functions @xmath514 in @xmath515 . since @xmath44 and @xmath2 are unchanged under conformal coordinate transformations , it is sufficient to consider a critical system bounded by two concentric spheres .
we have derived @xmath61 by calculating the average of the stress - energy tensor in this highly symmetric geometry . besides the global free energy local thermodynamic averages , e.g. profiles of the order parameter and the energy density in the presence of two spherical particles , are also of interest .
these can also be derived from the concentric geometry @xcite , and we have presented explicit results for @xmath6 near 4 and for various surface universality classes @xmath516 .
we have also demonstrated that the stress - tensor average determines the modification of the density profile near the surface of one sphere due to the other distant sphere @xcite . the casimir interaction and the density profiles for parallel plates
are obtained as a special case of the results for concentric spheres when the radii @xmath517 tend to infinity with fixed difference @xmath518 ( compare fig .
the other extreme @xmath141 with @xmath1 fixed is also of interest .
it determines the casimir effect and the density profiles when one sphere is much smaller than other lengths in the problem .
we have checked that the free energy contribution and the density profiles arising from the presence of the small sphere are consistent with a ` small - sphere operator expansion ' .
this method has been introduced in ref .
@xcite and is also explained in sec . [ 2c ] of this paper .
the density profiles show still non - trivial scaling forms in this limit ( cf .
appendix b ) .
we have also found simple expressions @xcite for the interaction between distant spheres in an infinite host fluid if the latter is not at the critical point . here
both the host correlation length @xmath519 and the distance @xmath520 between the spheres have to be much larger than the radii of the spheres , but the ratio @xmath521 is unrestricted .
similar expressions have been found for the interaction of a single sphere with a distant planar boundary .
we conclude by discussing possible experimental observations of casimir interactions between spheres at the critical point .
the case of fluid mixtures at their consolute point is of particular interest , since the attractive interaction between two immersed equal spheres decays extremely slowly , with a power law @xmath522 in @xmath515 as the separation @xmath39 becomes much larger than the radius @xmath28 of the spheres .
the exponent follows from eq.([fdis ] ) , with @xmath523 in @xmath515 @xcite . in the only previous theoretical work we know on this topic ,
de gennes @xcite obtained an extremely good estimate @xmath297 for this exponent by minimizing a phenomenological free energy functional .
the universal amplitude @xmath524 is estimated to be somewhat larger than @xmath525 @xcite .
the casimir interaction ( [ 5.1 ] ) should be compared with the van der waals interaction which decays much faster with an exponent @xmath526 ( non - retarded ) or @xmath527 ( retarded ) .
the prefactors are comparable .
for example , for fused silica spheres in water the van der waals prefactor is about @xmath528 kelvin @xcite .
we note that for spheres in @xmath71he at the @xmath8point ( @xmath72 model ) the casimir exponent is @xmath529 for @xmath515 .
thus the decay is faster than for fluid mixtures ( see ( [ 5.1 ] ) ) .
finally we mention two classes of experiments one could conceivably carry out to study the casimir interaction : + ( i ) interaction between only two objects , for example a single sphere near a planar wall : by measuring the force on the sphere for varying distance from the wall , one can in principle obtain the whole scaling function @xmath61 .
of course , the distance must not exceed the correlation length of the nearly critical host binary fluid .
correlation lengths of several @xmath530 are feasible , and one would like to have a sphere diameter of a few hundred .
measurements of the interaction potential of isolated pairs of spheres of somewhat larger size have been reported for a charge stabilized colloid @xcite , see also @xcite .
+ ( ii ) aggregation phenomena in colloids : the casimir interaction between more than two spherical particles at large @xcite separations is a sum of pair interactions , which have the gravitation - like form ( [ 5.1 ] ) for separations @xmath520 smaller than the correlation length @xmath519 of the pure host and vanish for @xmath531 . on approaching the critical point @xmath519
strongly exceeds the radii and mean separation of colloid particles .
this offers a possible mechanism for their aggregation @xcite .
* acknowledgements : * we thank w. fenzl , m. lssig , h. wagner , and especially t. w. burkhardt for helpful comments and discussions .
we also thank t. w. burkhardt for a critical reading of the manuscript .
u. r. would like to thank the deutsche forschungsgemeinschaft for partial support through sonderforschungsbereich 237 . + * note added in proof : * after this work was completed we realized that the widely used non - retarded hamaker approximation @xmath532 for the van der waals interaction of two spheres is also _ conformally invariant _ ( like the critical casimir energy @xmath51 studied in this paper ) .
this follows immediately from its representation @xmath533 with integrations over the two spheres and allows to express @xmath532 as a function of the invariant quantity @xmath44 defined in ( 1.4 ) .
the explicit result [ 57 ] @xmath534\right\}$ ] may be used for a first comparison of critical casimir and van der waals interaction . note that for _ close _ spheres ( @xmath44 near 1 ) @xmath532 has the _ same _ dependence on @xmath535 as @xmath51 and can be obtained from the right hand side of ( 2.5 ) by replacing @xmath536 by @xmath537 and setting @xmath515 .
an infinitesimal coordinate transformation @xmath538 changes a geometry @xmath539 to a geometry @xmath540 .
we consider two examples : + ( i ) for the geometry of non - overlapping spheres in fig .
1b where @xmath541 , @xmath542 with @xmath543 the unit step function , implies a rigid shift of the right sphere in the @xmath544 direction and increases the distance @xmath39 between the spheres by the amount @xmath94 .
+ ( ii ) for the concentric geometry in fig .
1a where @xmath545 , we consider the transformation ( [ thet ] ) .
the corresponding change in the singular part of the free energy is given by @xcite @xmath546 to first order in @xmath94 .
here @xmath98 is the stress tensor . in case ( i )
@xmath547 with the average taken in the geometry of fig .
in case ( ii ) @xmath548\>.\ ] ] the result simplifies at the critical fixed point since the trace of the stress tensor vanishes for @xmath549 @xcite ; a possible boundary contribution at @xmath263 drops out in the change of @xmath550 @xcite .
then only the second term on the right hand side of ( [ part ] ) contributes and picks out the radial component @xmath551 of the stress tensor .
this leads to @xmath552\;\right)\>\right|_{\,\alpha=0 } = - k_bt_c\ : s_d\,r_0^d\;\langle t_{nn}(r_0)\rangle_{\scriptscriptstyle\rm con}\ ] ] to first order in @xmath94 , which implies eq.([deef ] ) . since both @xmath553 and @xmath554 can be expressed in terms of the same function @xmath61 ( see sec .
[ 1 ] ) , the ratio between the left - hand sides of eqs .
( [ freg ] ) and ( [ deef ] ) is simply @xmath555 , with @xmath44 and @xmath67 from ( [ kap2 ] ) and ( [ kap3 ] ) , respectively .
it is instructive to verify from the transformation formula ( [ ttra ] ) for the stress - tensor average at the critical fixed point that the right - hand sides of ( [ freg ] ) and ( [ deef ] ) lead to the same simple ratio . with @xmath556 and @xmath557 in eq.([ttra ] ) , an easy @xmath558-integration leads to the ratio @xmath559 .
this equals the above - mentioned ratio , since @xmath560 as follows from writing @xmath561 in terms of the intersection points i , ii , iii , iv in ( [ inte ] ) .
( [ freg ] ) is not restricted to the critical fixed point but applies for arbitrary values @xmath215 of the renormalized @xmath192 coupling strength ( see eq .
( [ relu ] ) ) . this can be used to show the universality ( @xmath215-independence ) of the singular part of the free energy in the long - distance limit . for simplicity
consider a situation where the boundary conditions have their fixed - point form .
then by naive dimensional analysis the right hand side of ( [ freg ] ) can be written as @xmath562 in the ` shift operator ' @xmath563 the improvement term @xmath467 in ( [ impr ] ) does not contribute , and the average ( [ aeq9 ] ) is renormalized and obeys a simple _ homogeneous _ renormalization group equation @xcite @xmath564 { \cal k}\left(r_1/s_{12},\,r_2/s_{12}\,;\,\mu\,s_{12 } , \,u_r\right ) = 0\ ] ] with the usual @xmath565-function @xcite .
standard arguments then imply that in the long - distance ( infrared ) limit @xmath566 with @xmath567 and @xmath568 fixed @xmath569 with @xmath570 from ( [ usta ] ) .
this shows the @xmath215-independence of the asymptotic critical casimir force ( [ freg ] ) .
here we study the density profiles @xmath571 in the concentric geometry in the limit @xmath141 with @xmath572 fixed at an arbitrary positive value @xmath573 .
we show that this form is indeed in agreement with @xcite , i.e. @xmath574\>,\ ] ] which follows from the small - sphere expansion ( sse ) ( [ tcon ] ) , ( [ sms2 ] ) , and ( [ iden ] ) . as before the full container ( fc ) has a surface with universality class @xmath33 . in ( [ pro1 ] ) , @xmath159 if both @xmath575 and the correlation in square brackets are non - vanishing , and @xmath160 otherwise . here
the upper sign holds for @xmath494 and the lower sign for @xmath426 . to obtain ( [ beq3 ] )
we have used the vanishing of @xmath579 for @xmath494 and @xmath75 and the result for @xmath580 that follows via a simple conformal transformation @xcite from the half space correlation function , which for @xmath341 is that of the gaussian model with dirichlet or neumann boundary conditions @xcite .
we have also used ( [ leau ] ) and ( [ dokn ] ) to evaluate @xmath581 and @xmath582 , respectively . the predictions ( [ beq3 ] ) of the sse are consistent with the solutions of the differential equation ( [ dife ] ) . for @xmath494
this has the form @xmath583 note that @xmath294 is the same for @xmath584 and @xmath336 and equals @xmath585 by ( [ limi ] ) for our case @xmath89 .
it follows from ( [ beq3b ] ) that the @xmath586-term in the function @xmath407 in ( [ beq4 ] ) is of order @xmath587 and can be neglected , since @xmath294 and @xmath588 are both of order @xmath589 .
the integral in ( [ beq4 ] ) is then trivial and leads to the result ( [ beq3b ] ) with the upper sign . for @xmath426
, the solution follows from ( [ beq4 ] ) on replacing the lower limit of integration by the positive @xmath285-value @xmath590^{1/2}\ ] ] for which @xmath591 vanishes .
this guarantees that @xmath592 vanishes at @xmath263 @xcite . for @xmath89 , in which case @xmath593 by eq .
( [ limi ] ) , one may again neglect @xmath586 inside @xmath594 , and the lower boundary of integration becomes @xmath595 .
the integral is simple and leads to the result ( [ beq3b ] ) with the lower sign .
the leading contribution to the energy @xmath596 for @xmath341 is given by @xcite @xmath597 and follows immediately from ( [ beq3 ] ) .
since @xmath598 , the right hand side of ( [ pro1 ] ) is , in leading order , the square of its value in the previous cases , and eq .
( [ pro1 ] ) is again fulfilled .
with @xmath294 from ( [ limi ] ) for the four cases @xmath603 . here
( [ beq7b ] ) represents the non - vanishing profile @xmath604 in a full container ( fc ) .
it can be obtained by means of a conformal transformation from the half space profile ( [ leau ] ) as described e.g. in ref .
( [ beq7c ] ) respresents the right hand side of ( [ pro1 ] ) . here
@xmath605 is @xmath606 in the four cases ( see remark just below ( [ dra4 ] ) ) . for @xmath341 [ beq8 ] @xmath607
is determined by gaussian fluctuations @xmath244 about the mean - field profile @xmath608 .
the quantity @xmath609 can be obtained from its counterpart in the half space whose form is known ( see eq .
( 4.96 ) in ref .
the corresponding correlation for @xmath598 follows from @xmath610 again the prediction ( [ beq7 ] ) of the sse may be checked by comparing with the solution of the differential equation ( [ dife ] ) , which is now given by eq .
( [ lnrr ] ) .
since @xmath285 tends to the finite positive value @xmath613 in ( [ beq7b ] ) as @xmath614 , one may expand the integrand in ( [ lnrr ] ) with respect to @xmath294 and invert the function @xmath615 to obtain @xmath616 , order by order in @xmath294 . in first order
the result reads @xmath617^{-3}\ > , \right\}\ ] ] where @xmath618 is @xmath407 for @xmath374 .
the integral on the right hand side is straightforward , and one sees that eq .
( [ beq10 ] ) indeed implies the result ( [ beq7 ] ) . here
we consider the @xmath141 behavior of the left hand side of ( [ pro1 ] ) for @xmath619 and @xmath620 .
the result @xmath621/n \longrightarrow \tilde s_d\,\sigma_{ab}^{(2)}\;(r_-)^{d-2}\>r^{-2(d-2)}\left [ 1+\sigma_{ab}^{(3)}\tilde r^{d-2}\right]^2\ ] ] follows for the gaussian model from the @xmath486 term in ( [ p2li ] ) , using @xmath622 .
note that @xmath623 only depends on @xmath32 , while @xmath624 only depends on @xmath33 .
this is in agreement with the right hand side of ( [ pro1 ] ) .
it is easily verified that ( [ beq11 ] ) equals the right hand side of ( [ pro1 ] ) : the leading operator is @xmath160 , and the @xmath625 correlation function inside a full spherical container with an @xmath73 or @xmath75 surface , which appears on the right hand side of ( [ pro1 ] ) in square brackets , follows from its counterpart in the half space with dirichlet or neumann boundary conditions , respectively .
the prefactor @xmath626 in ( [ pro1 ] ) is obtained from ( [ cobe ] ) and ( [ dokn ] ) .
99 m. e. fisher and p .-
g . de gennes , c. r. acad .
b * 287 * , 207 ( 1978 ) .
g . de gennes , c. r. acad .
b * 292 * , 701 ( 1981 ) .
for a recent review , see m. krech , _ the casimir effect in critical systems _
( world scientific , singapore 1994 ) h. b. g. casimir , proc .
akad . wet .
* 51 * , 793 ( 1948 ) ; for a recent review on the casimir effect in qed , see g. plunien , b. mller , and w. greiner , phys .
* 134 * , 87 ( 1986 )
. h. w. j. blte , j. l. cardy , and m. p. nightingale , phys .
lett . * 56 * , 742 ( 1986 ) .
i. affleck , phys .
* 56 * , 746 ( 1986 ) .
j. l. cardy , nucl .
b * 275 * , 200 ( 1986 ) .
t. w. burkhardt and t. xue , phys .
* 66 * , 895 ( 1991 ) ; nucl .
b345 * , 653 ( 1991 ) .
t. w. burkhardt and e. eisenriegler , nucl .
b * 424 * [ fs ] , 487 ( 1994 ) . k. symanzik , nucl .
b190 * , [ fs3 ] , 1 ( 1981 ) .
m. p. nightingale and j. o. indekeu , phys .
lett . * 54 * , 1824 ( 1985 ) ; j. o. indekeu , m. p. nightingale , and w. v. wang , phys . rev .
b * 34 * , 330 ( 1986 ) . m. krech and s. dietrich , phys . rev . lett .
* 66 * , 345 ( 1991 ) ; * 67 * , 1055 ( 1991 ) ; phys .
a * 46 * , 1886 ( 1992 ) ; * 46 * , 1922 ( 1992 ) .
e. eisenriegler and m. stapper , phys .
b * 50 * , 10009 ( 1994 ) . t. w. burkhardt and e. eisenriegler , to be published .
d. beysens and s. leibler , j. physique .
* 43 * , l133 ( 1982 ) ; d. beysens and d. estve , phys .
lett . * 54 * , 2123 ( 1985 ) ; d. beysens , j .- m .
petit , p. narayanan , a. kumar , and m. l. broide , ber .
bunsen ges .
phys . chem . *
98 * , 382 ( 1994 ) .
w. a. ducker , t. j. senden , and r. m. pashley , nature * 353 * , 239 ( 1991 ) .
a. m. polyakov , zh .
* 12 * , 538 ( 1970 ) [ sov .
jetp lett .
* 12 * , 381 ( 1970 ) ] .
j. l. cardy , in _ phase transitions and critical phenomena _
vol 11 , eds . c. domb and j. l. lebowitz ( academic press , london , 1987 )
t. w. burkhardt and e. eisenriegler , j. phys .
a * 18 * , l83 ( 1985 ) .
note that the static sphere in electromagnetism or particle physics can not be discussed in this way .
the space - time continuum where there is conformal invariance corresponds to a cylinder and can not be mapped onto the half
space in @xmath85 .
s. gnutzmann and u. ritschel , z. phys .
b , to be published .
v. privman and m. e. fisher , phys .
b * 30 * , 322 ( 1984 ) .
it has been pointed out in j. l. cardy and i. peschel , nucl .
b * 300 * [ fs22 ] , 377 ( 1988 ) that @xmath627 or @xmath628 are , at least in @xmath9 , not scale invariant , i.e. not independent of @xmath12 , but show a @xmath629 behavior with a universal prefactor ; compare also the discussion in v. privman , phys .
b * 38 * , 9261 ( 1988 ) for systems in @xmath85 . according to cardy and peschel the reason is that the average of the trace of the stress tensor has at the critical fixed point a non - vanishing delta - function contribution on the curved boundary of the critical medium .
the prefactor of this contribution is given by the local curvature of the boundary ( @xmath630 for the spherical boundary ) times a number which depends only on the universality class of the bulk and , for @xmath85 , possibly of the surface .
thus the delta - function contribution from the surface of the @xmath32-particle , say , is independent of the presence of the @xmath33-particle and cancels for the difference of stress - tensor trace averages , which when multiplied with the local dilatation and integrated over space determines the change of @xmath51 in ( [ del1 ] ) under an infinitesimal conformal transformation ; compare ref .
@xcite and eq .
[ free ] below .
the conformal invariance of @xmath51 in ( [ del1 ] ) will be explicitly verified in appendix a ( see the paragraph containing eq .
( [ aeq8 ] ) ) by using the transformation formula ( [ ttra ] ) for the stress tensor at interior points and transforming to the concentric geometry .
k. g. wilson and j. kogut , phys .
* 12 * , 75 ( 1974 ) .
k. binder , in _ phase transitions and critical phenomena _
vol 8 , eds . c. domb and j. l. lebowitz ( academic press , london , 1983 ) .
h. w. diehl , in _ phase transitions and critical phenomena _
vol 10 , eds . c. domb and j. l. lebowitz ( academic press , london , 1986 ) t. w. burkhardt and j. l. cardy , j. phys .
a * 20 * , l233 ( 1987 ) .
t. w. burkhardt and h. w. diehl , phys .
b * 50 * , 3894 ( 1994 ) .
v. dohm , physica scripta t * 49 * , 46 ( 1993 ) . j. l. cardy , phys . rev
. lett . * 65 * , 1443 ( 1990 ) e. eisenriegler , m. krech , and s. dietrich , phys .
. lett . * 70 * , 619 ( 1993 ) and to be published ; m. krech , e. eisenriegler , and s. dietrich , to be published . in many respects a boundary acts qualitatively like an assembly of operators . here
the tilde serves to distinguish the primary operators @xmath631 from those which appear in the sum ( [ sms2 ] ) .
the latter ( which are special @xmath144 s ) are denoted by @xmath163 , compare eq .
( [ iden ] ) below . both @xmath631 and @xmath163
may be either the order parameter or the energy density .
an expansion of the form ( [ sms ] ) will also exist if the ` small object ' is not a sphere but , for example , a _ cube _ with edge length @xmath127 . while the leading operators @xmath144 in @xmath135 should be the same as for the sphere , the prefactors @xmath129 will not , in general , be expressible in terms of half space amplitudes .
we expect additional operators which reflect the anisotropy of the cube .
for an infinite _ cylinder _ or ` wire ' with diameter @xmath632 ( or a hyperprism with @xmath633 finite equilateral edges and one edge of infinite extent ) , we expect an expansion of the form ( [ sms ] ) , the leading @xmath144 being a line integral @xmath634 with the unit vector @xmath635 in the infinite direction and with prefactor @xmath636 . here we consider cases where @xmath637 is positive and the line integral is an ` irrelevant ' bulk perturbation ( compare a. bray and m. a. moore , j. phys .
10 , 1927 ( 1977 ) , and t. w. burkhardt and e. eisenriegler , phys .
b * 24 , 1236 ( 1981 ) for a corresponding discussion of defect planes ) .
the expansion has been checked for the simple case of the order parameter correlation in an infinite gaussian model with a cylindrical hole with dirichlet boundary conditions in @xmath638 .
( for @xmath639 the long - distance decay of the correlation along the cylinder is governed by an exponent larger than the bulk exponent @xmath640 , and the expansion does not apply . ) in this case @xmath160 , and the amplitude @xmath641 can be explicitely determined .
for the interaction free energy @xmath642 of _ two crossed wires _ with @xmath73 ( ordinary ) boundary conditions , diameters @xmath643 , @xmath644 , and a minimal separation @xmath645 , one finds @xmath646 from the arguments which led to eq .
( 15 ) in ref .
@xcite . here
@xmath647 for @xmath515 , @xmath648 , which should apply to @xmath71he .
the quantity @xmath649 is independent of @xmath39 but depends on the angle between the wires . for simplicity
consider @xmath650 .
both the finite bulk correlation length @xmath651 and the ` large ' distances mentioned right after eq .
( [ sms2 ] ) have to be much larger than @xmath127 in order that eqs .
( [ sms ] ) apply .
the amplitudes @xmath652 are independent of @xmath653 and are given by ( [ iden ] ) as before .
since the bulk average of @xmath135 is for @xmath505 in general non - vanishing , one obtains slightly different relations .
for example the right hand side of ( [ hole ] ) has to be supplemented by subtracting the product @xmath654 .
this may e.g. be verified for the simple case when @xmath655 is the energy density around a single spherical hole with an ordinary @xmath656 surface in an infinite gaussian model at @xmath505 .
then the left hand side of ( [ hole ] ) can for arbitrary ratios @xmath657 , @xmath658 be represented by a sum over angular momentum numbers ( similar to eqs.([prex ] ) or ( [ endi ] ) ) ; compare e. eisenriegler , z. phys .
b * 61 * , 299 ( 1985 ) . for two distant spheres at @xmath505 with @xmath659 , eq .
( [ sms1 ] ) implies that @xmath51 defined in ( 1.7 ) is @xmath660 $ ] in an obvious notation . using ( [ iden ] ) ,
the leading behavior of @xmath51 is given by the right hand side of ( [ fdis ] ) with @xmath661 replaced by the normalized bulk correlation function @xmath662/b_{\psi}$ ] at @xmath505 , where the two points 1,2 are separated by the distance @xmath39 .
for the sphere in a half space ( spw geometry ) at @xmath505 with @xmath663 , the interaction energy @xmath664 can be written like the left hand side of ( [ spfc ] ) ; only the full container fc in ( [ spfc ] ) has to be replaced by the half space .
( [ sms ] ) imply for @xmath664 a leading behavior given by the right hand side of ( [ fdsp ] ) with the half - space profile @xmath665 at @xmath666 replaced by the difference @xmath667 of the profile at @xmath505 and its bulk limit . in the landau
ginzburg model the energy density operator @xmath668 is proportional to @xmath669 on account of the thermodynamic bulk relation @xmath670 with @xmath671 the corresponding free energy density .
since we use dimensional regularization @xcite , the bulk average of @xmath672 vanishes at @xmath506 .
for the universal amplitude ratios considered below , the ( negative ) factor between @xmath673 and @xmath148 does not matter , and we refer to @xmath674 as the energy density .
we denote the normal normal component of the stress tensor by @xmath551 in the concentric geometry of fig . 1a , where the normal is in the radial direction and by @xmath188 in geometries of the type depicted in fig .
1b , where the normal is on curved lines which are perpendicular to all the spherical @xmath182-images of concentric @xmath17-spheres .
in particular the lines are perpendicular to the plane @xmath305 . here
we consider situations where both the bulk and the two surfaces are at critical fixed points of the renormalization group .
it is only in this case that dilatation and conformal invariance hold on all length scales .
d. m. mcavity and h. osborne , nucl .
b * 406 * , 655 ( 1993 ) . for our two - sphere geometries
, it follows from ( [ tenn ] ) by means of the conformal transformation ( [ cotr ] ) which itself is regular near the boundaries . in @xmath9
, @xmath675 is hyperuniversal , i.e. independent of @xmath33 @xcite . here
@xmath676 is the bulk conformal charge of the two - dimensional system , see a. a. belavin , a. m. polyakov , and a. b. zamolodchikov , j. stat .
phys . * 34 * , 763 ( 1984 ) or ref . @xcite .
below we consider the gaussian model ( @xmath394 ) for arbitrary spatial dimension @xmath85 and the @xmath192-model in dimensional regularization and within the @xmath81-expansion . in both cases
there is no @xmath154 term in @xmath677 at the bulk critical point . c. callan , s. coleman , and r. jackiw , ann .
( ny ) * 59 * , 42 ( 1970 ) .
l. s. brown , ann .
( ny ) * 126 * , 135 ( 1980 ) .
d. j. amit , _ field theory , the renormalization group and critical phenomena _ , ( mcgraw - hill , new york , 1978 ) .
j. zinn - justin , _ quantum field theory and critical phenomena _ , ( clarendon press , oxford , 1989 ) .
p. g. leach , j. math .
* 26 * , 2510 ( 1985 ) .
w. sarlet and l. j. bahar , int .
j. non - linear mech .
* 15 * , 133 ( 1980 ) .
@xmath678 vanishes , however , at the @xmath75 surface ( cf .
d. m. mcavity and h. osborne , nucl .
b * 394 * , 728 ( 1993 ) . here
the integrand on the right hand side of ( [ lnrr ] ) is integrated from zero , the value of @xmath285 at the @xmath73-surface , to @xmath285 .
close to the @xmath73-surface @xmath285 is small , and the integrand can be expanded in powers of @xmath679 .
apart from being the simplest example for a family of @xmath6-dependent conformal theories , the model describes tricritical behavior for @xmath680 , apart from possible logarithmic corrections in @xmath515 .
m. abramowitz and i.a .
stegun , _ handbook of mathematical functions _ , ( dover publications , new york , 1972 ) . i. s. gradshteyn and i. m. ryzhik , _ table of integrals , series , and products _ , ( academic press , new york , 1980 ) . while the complete functions @xmath2 have only been obtained to leading order in the @xmath81-expansion , most of the universal amplitudes that determine the limiting behaviors are available to leading and next - to - leading order .
corresponding results for the amplitude in ( [ fdis ] ) for @xmath681 can be inferred from @xcite and @xcite and are shown in eqs.(20 ) , ( 21 ) of @xcite .
the amplitudes in ( [ plaw ] ) for @xmath86 in the symmetry - preserving cases follow from ref .
j. c. le guillou and j. zinn - justin , j. physique lett .
( paris ) * 46 * , l137 ( 1985 ) .
r. j. hunter , _ foundations of colloid science _ , ( oxford , clarendon 1989 ) . j. c. crocker and d. g. grier , phys .
lett . * 73 * , 352 ( 1994 ) .
if the separation is not much larger than the radius , multiparticle interactions appear , which can be related to multispin correlation functions in the ising model @xcite . note that at @xmath666 ( @xmath682 ) the total potential energy is more than extensive : for a quasi - close - packed configuration of @xmath683 aggregated colloid particles , from ( [ 5.1 ] ) the potential energy per particle is @xmath684 . here
@xmath49 is the exponent 1.04 in the pair potential ( [ 5.1 ] ) and @xmath515 the space dimension .
contrary to the usual thermodynamic behavior ( which applies for @xmath685 ) , the energy per particle increases with particle number , and the prefactor depends on the shape of the aggregate . for long - distance pair attraction @xmath686 with @xmath687
the statistical mechanics appears to be quite interesting and unusual . for a discussion of gravitational interactions ( @xmath688 ) ,
see k. h. kiessling , j. stat .
phys . * 55 * , 203 ( 1989 ) and references therein .
see discussion of the solutions for @xmath689 below ( [ f1sb ] ) .
@xmath690 is non - vanishing for @xmath494 or @xmath75 , but does not contain a contribution of order @xmath254 .
d. jasnow , in _ phase transitions and critical phenomena _
vol 11 , eds . c. domb and j. l. lebowitz ( academic press , london , 1987 ) * figure captions * + * fig . 1 : * conformal mapping of concentric spheres with center at the origin of @xmath17-space ( fig
. 1a ) onto spheres in @xmath691-space with centers on the line @xmath692 ( fig .
1b ) as implied by eq .
( [ cotr ] ) . the sphere denoted by the broken line with radius @xmath12
is mapped onto the plane @xmath305 . the two points @xmath693 and @xmath694 of this sphere
are mapped onto @xmath695 and @xmath696 , respectively .
the two spheres with radii @xmath16 and @xmath697 are mapped on spheres with radii @xmath28 and @xmath38 , respectively .
the corresponding intersection points of the two spheres with the coordinate axes parallel @xmath698 are denoted by i , ii , iii , iv . for the case
@xmath22 shown in fig .
1a the images @xmath699 and @xmath700 of @xmath701 and @xmath702 are located inside the two spheres in fig .
1b and are closer to the origin than the centers of the spheres .
+ * fig . 2 : * scaling functions @xmath703 in a critical ising - type ( @xmath365 ) system for spatial dimension @xmath341 . the function @xmath704 , which is not shown , would nearly coincide with @xmath705 in this plot .
for the four symmetry - breaking combinations @xmath706 , @xmath707 , @xmath708 , @xmath709 , the functions @xmath2 behave as @xmath340 for @xmath341 , and we have plotted the finite limit @xmath710 . | mesoscopic particles immersed in a critical fluid experience long - range casimir forces due to critical fluctuations . using field theoretical methods , we investigate the casimir interaction between two spherical particles and between a single particle and a planar boundary of the fluid .
we exploit the conformal symmetry at the critical point to map both cases onto a highly symmetric geometry where the fluid is bounded by two _ concentric _ spheres with radii @xmath0 and @xmath1 . in this geometry
the singular part of the free energy @xmath2 only depends upon the ratio @xmath3 , and the stress tensor , which we use to calculate @xmath4 , has a particularly simple form .
different boundary conditions ( surface universality classes ) are considered , which either break or preserve the order - parameter symmetry .
we also consider profiles of thermodynamic densities in the presence of two spheres .
explicit results are presented for an ordinary critical point to leading order in @xmath5 and , in the case of preserved symmetry , for the gaussian model in arbitrary spatial dimension @xmath6 .
fundamental short - distance properties , such as profile behavior near a surface or the behavior if a sphere has a ` small ' radius , are discussed and verified . the relevance for colloidal solutions
is pointed out . |
let @xmath0 be a bounded conformal map defined on the unit disk @xmath1 .
the _ integral means spectrum _ of @xmath2 is the function @xmath3 the complex power @xmath4 is defined in terms of the complex logarithm . as
the derivative @xmath5 is never zero in @xmath1 , we can select a unique single - valued branch of @xmath6 by requiring that @xmath7 . for real @xmath8 ( positive or negative ) ,
integral means measure boundary expansion and compression associated with a given conformal map . allowing @xmath8 to be complex amounts to addressing rotational phenomena , as well .
the _ universal integral means spectrum _ is obtained by taking the supremum over all bounded conformal maps @xmath2 , @xmath9 we will actually work with conformal maps with quasiconformal extensions and correspondingly introduce for any @xmath10 , @xmath11 the quasiconformal extension means here that @xmath2 extends to a @xmath12-homeomorphism @xmath13 with the distortion inequality @xmath14 brennan s classical conjecture @xcite states that @xmath15 . note that the identity @xmath16 is rather easy ( follows readily from the finite area of @xmath17 ) . on the other hand , there is no real evidence that @xmath18 is even function .
even a stronger radial invariance was hinted by becker and pommerenke in @xcite who proposed the following .
[ conj : circular - brennan ] @xmath19 in this work we provide some evidence for this conjecture by establishing partial radial invariance for the function @xmath20 . for every @xmath21 we have [ thm : main ]
@xmath22 we believe that in fact the identity @xmath23 holds without restriction on the real part of @xmath8 .
this is basically equivalent to conjecture [ conj : circular - brennan ] . on one hand , in view of the fractal approximation principle @xcite , we have @xmath24 . in the other direction ,
we sketch an argument along the lines of @xcite .
we first embed our map @xmath2 into a standard holomorphic motion @xmath25 , with @xmath26 and @xmath27 .
we then consider the subharmonic function for a fixed radius @xmath28 .
@xmath29 conjecture [ conj : circular - brennan ] controls the growth rates in @xmath30 and the maximum principle leads to the upper bound in .
the lower bounds are easy , see remark [ rmk : trivialbounds ] .
the only issue with the above sketch is that we need uniform estimates ( independent of @xmath25 ) of the growth rates .
actually , becker and pommerenke postulated this stronger uniform version in @xcite .
the special case @xmath31 of theorem [ thm : main ] was proved in @xcite .
recently , hedenmalm obtained @xcite a general bound ( without restriction on @xmath8 ) of the form @xmath32 when specialised to the values of @xmath8 covered by theorem [ thm : main ] this is weaker by the factor @xmath33 . finally , let us note that the strongest conjecture in this area @xcite of the form @xmath34 has recently been disproved by ivrii @xcite .
namely , he shows that there are better bounds asymptotically as @xmath35 and @xmath36 .
ivrii s approach builds on the recent developments of @xcite and on @xcite . in this paper , we consider the opposite asymptotic regime , when @xmath37 and @xmath38 arbitrary .
this regime is related to the following phase transition phenomenon . using the convexity of @xmath39 and the easy bounds of remark [ rmk : trivialbounds ]
, it follows that theorem [ thm : main ] can be extended to @xmath40 consider an angle @xmath41 with @xmath42 .
it means that for large positive values of @xmath8 , namely when @xmath43 , the function @xmath44 is linear in @xmath8 . from considerations of small values of @xmath8 , in each of these directions
there is a phase transition point where the function becomes strictly convex .
the existence of such a phase transition phenomenon for @xmath18 in the negative direction was shown by carleson and makarov @xcite and later extended to arbitrary direction in @xcite . as explained in @xcite ,
this can be interpreted as the transition from isolated singularities to fractal distribution of singularities in the extremal contribution to integral means .
[ [ acknowledgements ] ] _ acknowledgements _ + + + + + + + + + + + + + + + + + + it is my pleasure to thank hkan hedenmalm , oleg ivrii and stas smirnov for several discussions on topics related to the present paper .
classical distortion estimates in conformal mappings describe the optimal rate of growth for the derivative . for instance
, we have @xcite [ prop : prop1 ] for @xmath45 , we have @xmath46 here @xmath47 denotes the class of univalent maps of the disk with normalizations @xmath48 , @xmath49 . similarly , we will use notation @xmath50 for maps in @xmath47 admitting @xmath38-quasiconformal extension to @xmath51 .
an immediate application of lehto s majorant principle @xcite shows [ prop : propk ] for @xmath52 , we have @xmath53 one can alternatively describe the pointwise bounds in terms of scaling and rotation exponents . for this ,
let us introduce a few definitions .
for a simply connected domain @xmath54 , consider the harmonic measure @xmath55 with a basepoint @xmath56 .
the choice of the basepoint will be irrelevant in what follows .
we say that the harmonic measure _
scales with exponent _ @xmath57 at @xmath58 if @xmath59 in a dual way , we will measure the rotation near @xmath60 .
let @xmath61 be the connected component of @xmath62 containing @xmath63 .
we say that @xmath17 _ rotates at rate _
@xmath64 if @xmath65 the branch of the argument is selected so that @xmath66 . in other words
, @xmath67 measures the rate of rotation ( in the sense of @xcite ) of how fast a curve should rotate around point @xmath60 in order to get to the @xmath30-neighbourhood of @xmath60 from @xmath63 within the domain @xmath17 .
of course , in general , the limit in these scaling and rotation exponents need not exist . to rectify this
, one could consider subsequential limits . in this case , it is important to require that we measure both scaling and rotation simultaneously along the same subsequence . a well - known estimate of beurling ( * ? ? ?
* corollary 9.3 ) says that @xmath68 . in terms of the scaling exponent , this means that @xmath69 ( if exists ) is at least @xmath70 .
a slightly more general version relates the scaling exponent to the twisting at a boundary point via the inequality , see ( * ? ? ?
* lemma 1 ) @xmath71 the basic examples for extremal pointwise behaviour are more conveniently described as conformal maps of the upper half plane @xmath72 .
conjugating with an appropriate mbius transformation provides examples defined on the unit disk .
[ ex : basicexamples ] the _ complex power map _
( with the principal branch ) @xmath73 , @xmath74 defines a conformal map as long as @xmath75 and @xmath76 . with the notation @xmath77 , @xmath78 , @xmath64 , the conformal map @xmath79 maps the positive and negative half - lines of @xmath80 to logarithmic spirals of rotation rate @xmath67 , while harmonic measure scales with exponent @xmath69 at the origin . the condition @xmath75
, @xmath76 is needed to ensure injectivity .
equivalently , this condition is exactly .
observe that @xmath81 has a @xmath82-quasiconformal extension to the lower half plane @xmath83 .
this follows readily from the @xmath84-lemma ( * ? ? ?
* section 12.3 ) .
indeed , @xmath81 embeds to the holomorphic family @xmath79 parametrised by @xmath85 . in section [ se :
example ] we describe the extensions and the corresponding holomorphic motion explicitly .
[ rmk : trivialbounds ] the pointwise estimates of proposition [ prop : prop1 ] and [ prop : propk ] give the following _ trivial upper bounds _ for integral means spectra @xmath86 the _ trivial lower bounds _ are obtained by considering example [ ex : basicexamples ] and read as @xmath87
we are going to make use of the following three - point nevanlinna - pick problem on the bidisk . in
what follows , @xmath88 will denote the right - half plane .
[ prob : threepoint ] given @xmath89 , define the set @xmath90 as @xmath91 where @xmath92 ranges over all holomorphic mappings of the bidisk @xmath93 .
describe @xmath90 in terms of the parameter @xmath94 .
first , let us observe that @xmath90 only depends on @xmath95 .
this can be seen by considering transformations of the form @xmath96 .
[ lemma : np ] let @xmath97 .
the solution of problem [ prob : threepoint ] is described by the convex hull of the union of two disks ( see figure [ fig:3point ] ) @xmath98 is the union of the red and blue disks together with their convex hull.,width=226 ] in connection with the nevanlinna - pick interpolation , it is more customary to consider holomorphic maps into @xmath99 .
we may achieve this by considering @xmath100 and the corresponding three - point problem .
we shall consider the two problems in parallel as we will need to move back - and - forth between the two settings . as is well - known , in the one variable case , the positive semi - definiteness of pick s matrix is equivalent to the existence of an interpolating map .
agler @xcite found an analogous necessary and sufficient condition for interpolation problems in the bidisk .
the three - point case has been analysed in detail in @xcite .
in fact , problem [ prob : threepoint ] is very much related to example 3.4 from @xcite .
the difference is that example 3.4 assumes that @xmath101 , while we need to consider complex values .
nevertheless , the inner functions found on @xcite will be relevant .
we go through the analysis based on @xcite .
first , let us consider only the two - point problem given by the nodes @xmath102 and @xmath103
. we must have @xmath104 in terms of the hyperbolic and kobayashi distances .
a short calculation reveals this is equivalent to the tangent cone displayed in figure [ fig:3point ] @xmath105 consider now the corresponding three - point problem in terms of @xmath106 , @xmath107 we may assume that @xmath108 and by multiplying with a positive number that the problem is _
extremal_. that is , @xmath109 and there are no solutions with smaller norm .
the problem is _ non - degenerate _ in the sense that there are no solutions that depend only on one of the coordinate functions ( because @xmath108 ) .
it may happen that a sub two - point problem is already extremal , but assume for now , that this is not the case and the problem is a _ genuine _ three - point problem . under these circumstances ( extremal , non - degenerate , genuine ) we have a unique solution which is a rational inner function of degree @xmath110 ( * ? ? ?
* theorem 0.1 ) .
moreover , the second order terms only involve the mixed product @xmath111 @xcite .
since @xmath112 , we have the following representation ( see @xcite ) @xmath113 where @xmath114 is a linear polynomial which does not vanish on @xmath115 , @xmath116 since the function @xmath117 solves the same three - point problem and the solution is unique we have the symmetry @xmath118 this leads to the following equations in terms of the coefficients @xmath119 there are two possibilities : either @xmath120 is real or purely imaginary .
observe , that we are free to multiply the coefficients by a real number without affecting the function @xmath121 .
thus the two cases can be normalized to @xmath122 or @xmath123 . in the first case ,
we are led to @xmath124 the condition @xmath125 is enforced because @xmath114 must not vanish on @xmath115 . solving for @xmath126 we arrive at @xmath127 in conclusion , @xmath128 and so @xmath129 .
this is the first disk in . in the second case , when @xmath123 , we are led to @xmath130 this time , we have @xmath131 in this case , @xmath132 and we now have @xmath133 .
this is the second disk in .
so far we covered the cases of genuine three - point ( extremal ) problems .
obviously , @xmath90 is a convex set , we will show that by taking the convex hull in we also cover cases when the three - point problem degenerates to a two - point problem .
it may happen that a sub two - point problem is already extremal but only for the pair @xmath102 and @xmath103 otherwise @xmath121 would depend only on one variable .
from now on we assume that two - point problem corresponding to these nodes is extremal , that is we have equality in .
in such a situation @xmath121 has to be a mbius transformation along an embedded analytic disk through the two nodes . by changing coordinates by an automorphism of @xmath115
we may arrange things so that the embedded disk is the diagonal @xmath134 and the function is identity on the diagonal .
this leads to the following situation in terms of the new holomorphic function @xmath135 @xmath136 we will show that necessarily @xmath137 . indeed , all such functions are of the following form ( see ( * ? ? ? * ( 11.81 ) ) ) @xmath138 where @xmath139 $ ] and @xmath140 . from @xmath141
we find that @xmath142 and conclude with @xmath143 in terms of @xmath144 , we arrive at @xmath145 where we also used the fact that @xmath146 .
condition corresponds exactly to the contribution of the convex hull in as claimed .
with all the cases analysed we concluded the proof of the lemma .
let @xmath147 be a measure space with its @xmath148 spaces of complex - valued measurable functions .
we will consider analytic families of measurable functions in @xmath17 parametrised by two variables .
that is , jointly measurable functions @xmath149 defined on @xmath150 such that except for a measure zero set @xmath151 , for each fixed @xmath152 the map @xmath153 is analytic in @xmath154 .
the family is said to be _ non - vanishing _
if @xmath151 can be chosen so that @xmath155 for all @xmath156 and for all @xmath157 .
we may use the holomorphic dependence to amplify a priori norm estimates in such families , see ( * ? ? ?
* interpolation lemma ) . with our application in mind
, we will carry this out in the following setting .
[ prop : interpolation ] suppose @xmath158 is a holomorphic family of measurable functions , such that there is an exceptional set @xmath151 with @xmath159 and for every @xmath160 , @xmath161 let @xmath162 and assume that @xmath163 then , for every @xmath164 and for every complex exponent @xmath165 satisfying @xmath166 the choice of the continuous branch used here is determined by @xmath167 .
the proof goes along the lines of ( * ? ? ?
* lemma 4.1 ) .
the key difference is that we will make use of the three - point nevanlinna - pick problem analysed in section [ se : np ] instead of the schwarz lemma used in @xcite . by considering the analytic family @xmath168
we may restrict our attention to the @xmath169 case .
observe that our assumption @xmath170 implies @xmath171 actually , by the maximum principle for analytic @xmath172-valued functions , we may further assume the strict inequality @xmath173 otherwise @xmath174 would be constant in the parameter @xmath175 , as we have @xmath176 for all @xmath160 .
we may also assume in the proof that @xmath177 uniformly for all @xmath178 , as the reduction of the general situation to this is done similarly as in ( * ? ? ?
* section 2 ) .
we choose an arbitrary positive probability density @xmath179 , uniformly bounded away from @xmath180 and @xmath181 , @xmath182 by jensen s inequality using the convexity of @xmath183 and ( [ eq : size ] ) we have @xmath184 for a fixed @xmath185 , we consider the holomorphic function @xmath92 in the bidisk @xmath154 @xmath186 again by jensen s inequality we have the bound @xmath187 thus @xmath92 maps the bidisk into a half - plane @xmath188 , while @xmath189 by our assumption @xmath190 .
furthermore , by our symmetry assumption in @xmath191 . at this stage
we appeal to the three - point nevanlinna - pick problem of problem [ prob : threepoint ] and deduce that for any @xmath192 , @xmath193 from now on we assume that the exponent @xmath165 satisfies ( with @xmath169 ) .
that is , @xmath194 for such @xmath8 , we have @xmath195 .
moreover , the tangent line to the the convex set @xmath196 at the point @xmath197 is vertical . in other words
, we have @xmath198 for any @xmath199 .
these statements follow from the description of @xmath200 in lemma [ lemma : np ] . in particular , @xmath201 equivalently , we have @xmath202 by specialising the choice of @xmath179 , that is , choosing @xmath203 , we obtain our assertion in the form @xmath204
towards the proof of theorem [ thm : main ] , we first prove a more technical version with various normalisations . in particular , we assume that the beltrami coefficient @xmath205 is supported in the set @xmath206 for some @xmath207 .
[ thm : integrability ] given @xmath208 and @xmath209 , let @xmath205 be measurable , @xmath210 , with some @xmath211 .
let @xmath212 be the homeomorphic solution to @xmath213 normalised by fixing @xmath214 .
we have @xmath215 for any @xmath216 with @xmath217 and @xmath218 . here
@xmath219 is a constant depending only on @xmath220 .
the complex powers in [ eq : intmeans11 ] are defined in terms of the holomorphic branch of @xmath221 which vanishes at @xmath222 .
we embed @xmath2 in a two - parameter holomorphic motion by setting for @xmath157 , @xmath223 let @xmath224 denote the unique solution to the beltrami equation @xmath225 normalised so that @xmath226 , @xmath227 and @xmath228 .
the uniqueness of the solution implies that @xmath229 .
we apply proposition [ prop : interpolation ] to the family @xmath230 since @xmath224 is conformal on @xmath231 , this is a non - vanishing holomorphic family .
we record the following symmetry of the construction , @xmath232 this , in turn implies @xmath233 standard quasiconformal estimates give the following global @xmath234-bounds , @xmath235 this is derived exactly as in ( * ? ? ?
* ( 4.18 ) ) .
the point is that @xmath224 are @xmath236-quasiconformal for all @xmath160 .
observe that @xmath190 since @xmath237 .
we now use proposition [ prop : interpolation ] with @xmath238 , @xmath239 and @xmath240 to deduce the conclusion of the theorem .
a final remark is in order on the choices of complex logarithms . in proposition
[ prop : interpolation ] we used the holomorphic flow to define complex logarithm .
this is consistent with the statement of theorem [ thm : integrability ] as can be seen by considering the continuous function @xmath241 on the simply connected region @xmath242 . in order to compare the growth of complex powers of @xmath5 with that of @xmath243 we need to control the growth of @xmath244 .
the growth is quite slow , for a bounded function @xmath245 , we have ( * ? ? ? * section 4 ) @xmath246 this would suffice for our purposes but it turns out that in our case , when @xmath2 has a quasiconformal extension , @xmath244 is even bounded .
[ lemma : foverz ] let @xmath0 be a conformal map in the class @xmath50
. then @xmath247 is a bounded function .
the bound only depends on the constant @xmath38 .
embed @xmath2 in the holomorphic motion by solving the beltrami equation for @xmath248 @xmath249 for a fixed @xmath250 , consider the holomorphic function @xmath251 . from distortion properties of quasiconformal maps the real part is bounded from above and below ( independently of @xmath252 ) , @xmath253 it follows automatically that the imaginary part also satisfies a similar bound .
indeed , let @xmath254 and @xmath255 . the function @xmath256 maps @xmath257 into a strip @xmath258 with @xmath259 .
in view of the schwarz - pick lemma , @xmath256 contracts in the hyperbolic metric @xmath260 . we deduce that @xmath261 with a different constant .
we are ready to deduce theorem [ thm : main ] .
consider a conformal map @xmath0 with a @xmath38-quasiconformal extension . by composing with a similarity transformation
, we may assume that @xmath52 .
let @xmath165 with @xmath217 and @xmath218 .
by lemma [ lemma : foverz ] @xmath262 we continue the inequality by applying theorem [ thm : integrability ] for the function @xmath263 . for a small @xmath208 we have
@xmath264 this shows that @xmath265 , and letting @xmath266 proves the theorem .
integration by polar coordinates gives the following corollary .
[ cor : integrability ] let @xmath267 be a conformal map with @xmath38-quasiconformal extension
. then we have @xmath268 for any @xmath165 with @xmath217 there exist a conformal map @xmath2 with @xmath38-quasiconformal extension such that @xmath269 namely , it is sufficient to consider example [ ex : basicexamples ] with @xmath270 .
it will be instructive to analyse examples [ ex : basicexamples ] from the point of view of holomorphic motions .
we will embed them in a motion parametrised by the bidisk as in the proof of theorem [ thm : integrability ] .
again , it will be more convenient to work with upper and lower half planes @xmath271 and @xmath83 instead of the unit disk and its exterior
. set the beltrami coefficient @xmath272 , @xmath273 .
we now describe the holomorphic motion @xmath224 solving the normalised beltrami equation with coefficient @xmath274 for @xmath160 . along the diagonal @xmath275
the solution is the radial stretching / twisting map @xmath276 in general , the solution @xmath224 will be given in @xmath271 by @xmath277 followed by a ( conformal ) complex power map and in @xmath83 by @xmath278 followed by another complex power map .
we have to choose the complex powers in such a way that the map in @xmath271 and in @xmath83 matches on the real line .
explicitly , with the notation @xmath279 , @xmath280 and @xmath281 , @xmath282 the stretching / twisting behaviour of the motion @xmath224 ( at the origin ) is described by the complex exponent @xmath283 where @xmath284 observe that @xmath285 satisfies the conditions in problem [ prob : threepoint ] and @xmath286 is equal to @xmath92 from with the choice of @xmath287 ( general values of @xmath288 may be covered by reparametrising the motion ) .
that is , we see that the first disk in lemma [ lemma : np ] is a ` physical example ' , in the sense that it appears from holomorphic functions built from holomorphic motion of conformal maps ( through the constructions in the proofs of theorem [ thm : integrability ] and proposition [ prop : interpolation ] ) .
conjecture [ conj : circular - brennan ] amounts to saying that this first disk is the only allowable region for such physical examples .
our method does not capture subtler properties of conformal maps and , in particular , is unable to rule out the ` non - physical example ' of the second disk in lemma [ lemma : np ] .
as an application of our main result , we give a geometric multifractal dimension estimate on the size of twisting for a quasidisk .
let @xmath54 be a bounded @xmath289-quasidisk and consider points @xmath58 which twist at a prescribed rate @xmath64 .
the fact that @xmath17 is the image of the unit disk under a global @xmath289-quasiconformal map is equivalent to the fact that ( any ) conformal map @xmath0 has a @xmath38-quasiconformal extension .
the precise relation of the constants involved are @xcite @xmath290 we now recall the notion of rate of twisting from section 2 in terms of the conformal map @xmath0 .
let @xmath291 , @xmath292 and consider the limit @xmath293 if this limit exists and is equal to @xmath64 , then we say that @xmath58 is @xmath67-spiralling ( or twisting ) .
the reader may contrast this with ( * ? ? ?
* corollary 5.4 ) which gives a similar multifractal estimate for twisting points . that bound applies to arbitrary quasiconformal maps , and therefore gives a weaker result when applied to the map @xmath2 . by proposition [ prop : propk ] ,
the maximal pointwise twisting for a quasdisk is @xmath296 .
theorem [ thm : twisting ] is effective in the sense that it shows this extremal twisting can only happen on a set of dimension zero . let @xmath297 be the pre - image of @xmath67-spiralling points and fix an @xmath298
. by definition of we may select for any @xmath299 a radius @xmath300 such that @xmath301 for @xmath302 .
let us denote the arcs of @xmath303 with center @xmath304 and length @xmath305 by @xmath306 .
vitali s covering lemma allows us to select a countable disjoint subcollection @xmath307 such that @xmath308 . for each @xmath309 , we associate a `` top - half '' of a carleson box @xmath310 as follows .
let @xmath311 , where @xmath312 and @xmath313 for short .
now we show that @xmath314 has to be big ( in absolute value ) on each @xmath310 . from koebe distortion and quasisymmetric properties of @xmath2
we have that , for @xmath315 @xmath316 from this , by an argument exactly as in lemma [ lemma : foverz ] , we deduce that the imaginary part also satisfies a similar bound @xmath317 combined with we conclude that @xmath318 with the notation @xmath319 .
we also assume here @xmath320 for simplicity ( the case @xmath321 is similar , we just have to consider @xmath322 below ) . for @xmath323 , @xmath324 summing it up over @xmath310 , we obtain @xmath325 as long as @xmath326 , see corollary [ cor : integrability ] . the sets @xmath327 provide a cover of @xmath328 , each with diameter comparable to @xmath329 .
this shows that @xmath330 . finally , taking the limits @xmath331 and @xmath332 proves the theorem . | consider a conformal map from the unit disk onto a quasidisk .
we determine a range of critical complex powers with respect to which the derivative is integrable .
the results fit into the picture predicted by a circular analogue of brennan s conjecture . 0.4 cm |
last year the selex collaboration reported the first observation of a narrow charmed meson @xmath0 with the mass of @xmath14 mev and decay width @xmath15 mev at @xmath16 confidence level @xcite .
this state has been seen in two decay modes , @xmath17 and @xmath18 , with a relative branching ratio @xmath19 .
this relative branching ratio is rather unusual because the phase space of channel @xmath18 is @xmath20 times larger than that of the channel @xmath17 and according to the su(3 ) flavor symmetry the ratio @xmath21 @xcite if the quark content of @xmath0 is @xmath22 , and the reference @xcite gave an estimate of @xmath23 .
since the @xmath0 is above the threshold for @xmath18 and @xmath17 , and its decays to these final states are okubo - zweig - iizuka ( ozi ) allowed processes , it is also strange that it has a narrow width ( @xmath11 mev ) .
why this state is so narrow and dominated by the @xmath17 decay mode has inspired a lot of theoretical interest @xcite , with many possible explanations , @xmath24 the first radial excitation of @xmath25 , a tetraquark , a hybrid , a two - meson molecule , a diquark - antidiquark bound state , @xmath26 among them the most plausible candidate is the first radial excitation of @xmath25 , the @xmath27 state , since it is the likely conventional meson with component of @xmath28 which has not been found . and
even though the mass of @xmath29 mev is about @xmath30 mev below the traditional potential model prediction ( for example , see @xcite ) , the similar situation has been found in the new narrow charm - strange @xmath31 meson @xmath32 , favored as the conventional @xmath33 state .
furthermore , simonov and tjon @xcite showed a possible mechanism to shift down the meson mass by the coupled - channel analysis method .
chao @xcite gave a reasonable argument that the node structure in the @xmath27 state may explain the small decay width and unusual decay modes .
in this letter , following the references @xcite , we assume @xmath0 as the first radial excitation of vector meson @xmath34
. then we solve the instantaneous bethe - salpeter equation @xcite , and give our prediction of mass for @xmath35 state . at the same time
, we obtain the wave functions of the relevant mesons . by using the reduction formula , pcac relation and low energy theorem
, we write the transition @xmath36-matrix as a formula involving the light meson decay constant and the corresponding axial current transition matrix element between two heavy mesons , which in turn can be written as an overlapping integral of the relevant wave functions .
we give a detailed consideration on the node structure in @xmath37 wave function .
we further try to make a rough estimate on the full decay width and the ratio @xmath38 by assuming @xmath0 as the @xmath39-wave @xmath1 vector state as well as considering the possible @xmath13 wave mixing .
in our previous papers @xcite , we solved exactly the instantaneous bethe - salpeter equations ( or the salpeter equations @xcite ) to describe the behavior of @xmath40 and @xmath1 states . in refs .
@xcite , we have given the mass spectra and wave functions of pseudoscalar @xmath41 , @xmath42 and @xmath43 , and the mass spectra of all these heavy @xmath40 states , which are consistent with experimental data quite well .
for the instantaneous bethe - salpeter equation of @xmath1 states @xcite , we have used the same parameters as used in @xcite , only changing the parameter @xmath44 to fit data because @xmath44 is the only parameter which can be different for @xmath40 and @xmath1 states when their quark contents are the same . in this letter
we first solve the full salpeter @xmath1 equation through fitting the ground state mass at @xmath45 mev ( @xmath2 ) to fix the parameter @xmath44 .
after fixing the parameter @xmath44 , we then give the predictions of the first radial excitation state .
since we are only interested in the leading order estimate , which lets us see the problem clearly , we assume in our calculation that the first radial excitation state @xmath46 is a pure @xmath36 wave state , excluding the possible mixing between @xmath36 wave and @xmath39 wave states . later in this letter
we will also consider the @xmath39 wave contribution .
our prediction for the mass of the first radial excitation state is @xmath47 mev , which is about @xmath4 mev heavier than the experimental data @xmath5 mev .
in this section we give a detailed description for the calculation of the transition matrix element . by using the reduction formula ,
the transition @xmath36-matrix of decay @xmath48 ( similar to the channel @xmath49 ) can be written as : @xmath50 where @xmath51 , @xmath52 and @xmath53 are the total momenta of initial state @xmath54 , final state heavy
meson and final state light meson , respectively ( see figure 1 ) ; @xmath55 is the mass of the final light meson .
the pcac relates the meson field @xmath56 with a current @xmath57 , where @xmath58 is the decay constant of light meson ; @xmath59 is @xmath60 for @xmath61 , @xmath62 for @xmath63 , respectively .
then the @xmath36-matrix becomes @xmath64 @xmath65 and with the low energy theorem this equation can be written approximately as : @xmath66 ( 100,170)(300,435 ) ( 0,0 ) for the decay @xmath67 , the equation is similar , but there is @xmath68 mixing . including the @xmath68 mixing
, the pcac relation reads as @xmath69 @xmath70 @xmath71,\ ] ] where @xmath72 , @xmath73 are the decay constants of octet @xmath74 and singlet @xmath75 , respectively
. then the transition matrix element becomes @xmath76 @xmath77 \langle d_s(p_{f1})|\bar{s}\gamma_{\mu}\gamma_{5}s|d^{+}_{sj}(p_i)\rangle .
\label{pcac2}\end{aligned}\ ] ] note that the decays @xmath78 are ozi - rule - allowed , so we may ignore anomaly of the pcac safely in eqs .
( [ pcac1],[pcac2 ] ) . according to the mandelstam formalism @xcite , at the leading order
, the matrix element @xmath79 can be written as @xcite : @xmath80,\label{eq2}\end{aligned}\ ] ] where @xmath81 and @xmath82 is the mass of @xmath83 and @xmath59 ( @xmath84 ) quark ; @xmath85 is the recoil three dimensional momentum of the final state @xmath39 meson .
@xmath86 , @xmath87 are the momenta of @xmath83 and @xmath88 quark in the initial meson , respectively . and
the relative momentum @xmath89 is defined as @xmath90 , in the center of mass system of the initial meson , and it becomes @xmath91 ; @xmath92 , and @xmath93 are called the positive energy wave function of the initial and final heavy mesons , and @xmath94 .
the positive energy wave function of @xmath42 can be found in paper @xcite : @xmath95@xmath96\gamma_{_5}\;,\ ] ] where @xmath97 and @xmath98 ; @xmath99 , @xmath100 are the radial part wave functions , and their numerical values can be obtained by solving the full salpter equation of @xmath40 state . since the wave function @xmath101 directly relates to the first radially excited state , there is only one node in it .
we note that the node structure can possibly play an important role to explain both the narrow decay width and the dominance of @xmath17 decay mode , to a certain degree .
the value of the radial wave function of @xmath54 state becomes negative as a function of relative momentum @xmath89 when it crosses the node ( denoted as @xmath102 ) , that is when @xmath103 .
therefore , this negative part of the wave function tends to cancel the contribution to the decay width from the positive part of the same wave function , which can be the reason to have a small decay width . in the overlapping integral of the wave functions eq .
( [ eq2 ] ) there is the shift of momentum , @xmath104 , and this will even strengthen the negative contribution because the shift will move the peak of the radial wave function of the final @xmath39 meson close to the node .
we note that the value of the overlapping integral will be significantly affected by the momentum shift as the larger momentum shift will cause the larger negative contribution , and in turn the smaller overlapping integral ( smaller decay width ) . since the recoil momentum @xmath105 of @xmath39 meson is larger than that of @xmath41 , about @xmath106 times , and the momentum shift in case of @xmath107 is @xmath108 times larger than that of the @xmath109 channel , larger negative contribution results in @xmath107 channel than in the @xmath109 channel , which can be the very reason for the small ratio of the branching ratios @xmath110 . as already pointed out in ref .
@xcite , there may be the @xmath13 mixing in @xmath1 state @xcite , so we will try to estimate the possible mixing contributions here too .
we first assume that this @xmath1 state is a pure @xmath36 wave state @xmath111 , and later we will consider the case of @xmath39 wave state . the wave function of @xmath1 can be written as @xcite : @xmath112\frac{(\omega_{c}\omega_{q}+{\vec q}^2-m_{c}m_{q } ) } { ( m_c+m_q){\vec q}^2 } \right.\ ] ] @xmath113 \frac{(m_{q}\omega_{c}-m_{c}\omega_{q})}{(\omega_{c } + \omega_{q}){\vec{q}}^2 } \right\},\end{aligned}\ ] ] and the positive energy wave function is defined as : @xmath114 where @xmath115 is the polarization vector of a @xmath116 state ; @xmath117 , @xmath118 are the energy projection operators for quark and antiquark ; the numerical value of the radial part wave function @xmath119 , @xmath120 will be obtained by solving the exact salpter equation of @xmath1 state .
for our numerical calculations we use the following parameters : @xmath121 mev , @xmath122 mev , @xmath123 mev , @xmath124 mev @xcite , @xmath125 mev @xmath126 mev , @xmath127 mev @xcite and @xmath128 mev @xcite .
the mass of octet @xmath74 can be estimated by the gell - mann - okubo relation @xmath129 , the mass of singlet @xmath75 can be obtained by @xmath130 , and the mixing angle can be estimated by the relation @xmath131 .
[ so our input values are @xmath132 mev , @xmath133 mev , @xmath134 . ]
we estimate the mass of the first radial excitation of @xmath2 by solving the full salpeter equation of @xmath1 state , and our prediction is @xmath47 mev , which is @xmath4 mev heavier than the central value @xmath5 mev measured by selex .
the theoretical uncertainties are estimated by varying all the input parameters simultaneously within @xmath135 , and we choose the largest possible error .
we obtained the wave functions of the corresponding states , and we show radial part wave functions @xmath136 , @xmath137 of @xmath41 in figure 2 ; @xmath136 , @xmath137 of @xmath42 in figure 3 ; @xmath138 , @xmath139 of @xmath111 in figure 4 .
we show the overlapping part of @xmath111 wave function @xmath140 and @xmath41 wave function @xmath141 , when it has the biggest momentum shift @xmath142 in figure 5 .
we also show @xmath140 and @xmath42 wave function @xmath143 in figure 6 .
one can see that in @xmath111 wave function the node localizes at @xmath144 mev , and the biggest momentum shifts are @xmath145 mev for @xmath42 , and @xmath146 mev for @xmath147 .
the decay width for the two body final state can be written as : @xmath148 here the matrix element @xmath149 for @xmath48 and @xmath49 is @xmath150 and for @xmath151 it is @xmath152 \langle d(p_{f1})|\bar{q}\gamma_{\mu}\gamma_{5}s|d^{+}_{sj}(p_i)\rangle .\ ] ] we obtain the following decay widths : @xmath153 @xmath154 @xmath155 and the corresponding ratios are @xmath156 as can be seen , we can not find the dominance of @xmath157 , and the ratio is much different from the experimental ratio @xmath9 .
the sum of those three decay widths is @xmath10 mev , already larger than the observed bound for full width ( @xmath11 mev ) given by selex experiment .
therefore , our assumption of the pure @xmath36 wave vector state can not fit the experimental data . before considering the @xmath13 mixing in @xmath164 vector state ,
let us first assume @xmath0 to be a pure @xmath39 wave state . in this case , we do not solve the full salpeter equation , but instead , we give the leading order estimate for the pure @xmath39 wave explanation . the wave function of a pure @xmath39 wave state ( @xmath165 coupling ) can be simplified as : @xmath166 where @xmath167 . with this wave function as input and by using the same parameters as for the @xmath36 wave state
, we can solve the approximate salpeter equation , @xmath168 , only the positive energy part solution is considered .
the mass of this pure @xmath39 wave state is @xmath169 mev , and the radial wave function of this pure @xmath39 wave state is shown in figure 7 . to estimate the matrix element in this case
, we calculate the overlapping integral of initial and final wave functions .
there is only one term , which can give a non - zero contribution : @xmath173 where @xmath174 ; @xmath83 is a constant ; @xmath175 is a shorthand notation for the terms which are functions of @xmath176 ; @xmath177 is the polarization tensor , which couples with the polarization vector @xmath178 to make the total polarization vector @xmath179 : @xmath180 @xmath181\ , . \;\;\;\;\ ; ( l=2)\end{aligned}\ ] ] then
, we obtain the numerical values for the decay widths when @xmath182 is a pure @xmath39 wave state ( @xmath183 ) : @xmath184 @xmath185 @xmath186 and the corresponding ratios are @xmath187 which show that the pure @xmath39 wave assumption does not fit the experimental data obviously .
we now consider the possibility of the @xmath13 mixing with a mixing angle @xmath192 : @xmath193 and @xmath194 as a function of the mixing angle @xmath192 in figure [ mixing ] .
numerically it is easy to find that when the mixing angle is taken to be @xmath195 , the relative branching ratio @xmath196 can fit the selex data .
we also predict other relative branching ratios @xmath197 with a rather small total decay width @xmath198 which is quite smaller than the experimental bound for the full width ( @xmath11 mev ) .
it is interesting to note that in fact there is another solution for the mixing angle , @xmath199 , which can also fit the data of the relative branching ratio .
for this solution , accordingly , the relative branching ratios @xmath200 and @xmath201 are predicted with a slightly larger total decay width @xmath202 mev .
the two solutions present a different relative aspect on the two decay channels @xmath203 and @xmath49 . as a final note , we only estimated the ozi allowed strong decay widths @xmath204 and @xmath205 assuming that @xmath206 is an @xmath1 vector state . however , there are some other mechanisms through which the @xmath1 vector state can decay to the same final states , which we do not consider here . for example
, there is the ozi forbidden mechanism for @xmath206 decaying to @xmath207 through the anomaly term in pcac , @xmath168 it can decay through two gluons , @xmath208 , @xmath209 , which is a decay similar to @xmath210 ( with decay width about @xmath211 kev ) .
as is well known , the ozi forbidden processes are small and can not improve sufficiently the relative ratio to agree with the selex data .
therefore , we do not include the processes in our analysis .
as summary , according to the data from selex and our estimates made here , it is not likely that the @xmath182 is the pure @xmath36-wave @xmath1 vector state .
only the @xmath13 wave mixing assumption may explain the relative branching ratio @xmath212 , whereas the assumption seems to give a rather small total decay width ( to be compared with the experimental bound @xmath213 mev ) . generally speaking , our results ( based on pcac and low energy theorem ) are qualitatively similar to those of ref.@xcite ( based on @xmath214 model ) .
we find that the node structure for the first radially excited state of the @xmath1 ground state @xmath2 can have a favorable effect on the relative ratio of the decay widths but it can not be so big as indicated by the data .
the @xmath13 wave mixing may help to explain the data , however , more precise measurements on the relative ratio @xmath215 ( and/or @xmath216 ) as well as the total decay width are crucial .
therefore , the explanation of @xmath0 as the first radially excited state of the @xmath1 ground state @xmath2 is too early to conclude . +
* acknowledgements * we would like to thank g. cvetic for careful reading of the manuscript and his valuable comments .
this work of c.h.c . and g.l.w .
was supported by the national natural science foundation of china ( nsfc ) .
the work of c.s.k . was supported in part by chep - src program and in part by grant no .
r02 - 2003 - 000 - 10050 - 0 from brp of the kosef .
the work of g.l.w . was also supported in part by bk21 program , and in part by grant no . f01 - 2004 - 000 - 10292 - 0 of kosef - nsfc international collaborative research grant .
99 selex collaboration , phys .
lett . * 93 * , 242001 ( 2004 ) .
yuan - ben dai , chun liu , y .- r .
liu , shi - lin zhu , jhep 0411 , 043 ( 2004 ) .
t. barnes , f.e .
close , j.j .
dudek , s. godfrey , e.s .
swanson , phys .
b600 * , 223 ( 2004 ) .
simonov , j. a. tjon , phys . rev . *
d70 * , 114013 ( 2004 ) .
kuang - ta chao , phys .
* b599 * , 43 ( 2004 ) .
liu , shi - lin zhu , y.b .
dai , c. liu , phys . rev .
* d70 * , 094009 ( 2004 ) .
yu - qi chen , xue - qian li , phys .
* 93 * , 232001 ( 2004 ) .
ailin zhang , hep - ph/0408124 .
eef van beveren , george rupp , phys .
93 * , 202001 ( 2004 ) .
l. maiani , f. piccinini , a.d .
polosa , v. riquer , phys . rev . *
d70 * , 054009 ( 2004 ) .
maciej a. nowak , e - print archive : hep - ph/0407272 .
stephen godfrey , e - print archive : hep - ph/0409236 .
b. nicolescu , j.p.b.c .
de melo , e - print archive : hep - ph/0407088 .
virendra gupta , e - print archive : hep - ph/0407289 .
f.e . close , e - print archive : hep - ph/0411396 .
s. godfrey and n. isgur , phys . rev .
* d32 * , 189 ( 1985 ) ; e. e. salpeter and h. a. bethe , phys . rev . *
84 * , ( 1951 ) 1232 . c. s. kim and guo - li wang , phys
b584 * ( 2004 ) 285 ; g. cvetic , c. s. kim , guo - li wang and w. namgung , phys
b596 * ( 2004 ) 84 ; c. s. kim , t. lee and g. l. wang , phys .
* b606 * ( 2005 ) 323 .
chao - hsi chang , jiao - kai chen and guo - li wang , hep - th/0406050 .
e. e. salpeter , phys . rev . * 87 * , ( 1952 ) 328 .
s. mandelstam , proc .
london * 233 * , 248 ( 1955 ) .
chao - hsi chang , jiao - kai chen and guo - li wang , hep - th/0312250 .
thorsten feldmann , int .
* a15 * ( 2000 ) 159 .
s. eidelman et al , particle data group , phys
* b592 * ( 2004 ) 1 . | we present a quantitative analysis of the @xmath0 observed by selex mainly focusing on the assumption that @xmath0 is the first radial excitation of the @xmath1 ground state @xmath2 . by solving the instantaneous bethe - salpeter equation ,
we obtain the mass @xmath3 mev for the first excited state , which is about @xmath4 mev heavier than the experimental value @xmath5 mev . by means of pcac and low - energy theorem
we calculate the transition matrix elements and obtain the decay widths : @xmath6 mev , @xmath7 mev , and the ratio @xmath8 as well .
this ratio is quite different from the selex data @xmath9 .
the summed decay width of those three channels is approximately @xmath10 mev , already larger than the observed bound for the full width ( @xmath11 mev ) .
furthermore , assuming @xmath12 is @xmath1 state , we also explore the possibility of @xmath13 wave mixing to explain the selex observation . based on our analysis
, we suspect that it is too early to conclude that @xmath0 is the first radial excitation of the @xmath1 ground state @xmath2 .
more precise measurements of the relative ratios and the total decay width are urgently required especially for @xmath13 wave mixing . |
calibrating measurement instruments is a important problem that engineers frequently need to address .
there exist several statistical methods that address this problem that are based on a simple linear regression approach . in tradition simple linear regression
the goal is to relate a known value of x to a uncertain value of y using a linear relationship .
in contrast , the statistical calibration problem seeks to utilize a simple linear regression model to relate a known value of y to an uncertain value of x. this is why statistical calibration is sometimes called _
inverse regression _ due to its relationship to simple linear regression ( osborne 1991 ; ott and longnecker 2009 ) .
recall in linear regression the model is given as follows : @xmath2 where * y * is a @xmath3 response vector , * x * is a @xmath4 matrix of independent variables with @xmath5 total model parameters , @xmath6 is a @xmath7 vector of unknown fixed parameters and @xmath8 is a @xmath3 vector of uncorrelated error terms with zero mean ( myers 1990 ; draper and smith 1998 ; montgomery _ et al .
it is assumed that the value of the predictor variable * x * = * x * are nonrandom and observed with negligible error , while the @xmath9 error terms are random variables with mean zero and constant variance @xmath10 ( myers 1990 ) . typically , in regression , of interest
is the estimation of the parameter vector ; @xmath6 , and possibly the prediction of a future value @xmath11 corresponding to a new @xmath12 value .
the prediction problem is relatively straightforward , due to the fact that a future @xmath13 value can be made directly by substituting @xmath14 into ( [ eq : linear_matrix_error ] ) with @xmath15=0 $ ] .
+ for the statistical calibration problem let @xmath16 be the known _ observed _ value of the response and @xmath17 be the corresponding regressor , @xmath17 which is to be estimated .
this problem is conducted in two stages : first measurement pairs @xmath18 of data is observed and a simple linear regression line is fit by estimating @xmath6 ; secondly , @xmath19 observations of the response are observed , all corresponding to a single @xmath17 ( zyurt and erar 2003 ) . since @xmath16 is fixed , inferences are different than those in a traditional regression ( or prediction ) problem ( osborne 1991 ; eno 1999 ; eno and ye 2000 ) .
eisenhart ( 1939 ) offered the first solution to the calibration problem , and is commonly known as the @xmath20 estimator to the linear calibration problem .
they assumed that the relationship between @xmath21 and @xmath22 was of a simple linear form : @xmath23 the estimated regression line for the first stage of the experiment is given by @xmath24 where @xmath25 and @xmath26 are the least squares estimate of @xmath27 and @xmath28 , respectively . using the data collected at the first stage of experimentation , eisenhart ( 1939 ) inverts equation ( [ eq : simple ] ) to estimate the unknown regressor value @xmath17 for an observed response value @xmath16 , by : @xmath29 where @xmath30 denotes the @xmath20 estimator for @xmath17 .
since division by @xmath31 is used there is an implicit assumption that @xmath32 .
+ assuming that @xmath32 , brown ( 1993 ) describes the following interval estimate corresponding to eisenhart ( 1939 ) : @xmath33 where @xmath34 @xmath35 krutchkoff ( 1967 ) proposed a competitive approach to eisenhart s ( 1939 ) classical linear calibration solution , which he called the @xmath36 regression calibration method and is written as : @xmath37 where @xmath38 and @xmath39 are the parameters in the linear relationship and @xmath40 are independent identically distributed measurement errors with a zero mean and finite variance . here
@xmath38 and @xmath39 are estimated via least squares .
the unknown @xmath17 can be estimated directly by substituting @xmath16 into the fitted equation : @xmath41 we let @xmath42 denote the @xmath36 estimator of @xmath17 . the @xmath43 confidence interval for @xmath44 can be written as @xmath45 where @xmath46 krutchkoff ( 1967 ) used a simulation study , where he found that the mean squared error of estimation for @xmath17 was uniformly less for this estimator versus the classical estimator .
the inverse approach was later supported by lwin and maritz ( 1982 ) . for criticisms of krutchkoff s ( 1967 )
approach such as bias see osborne ( 1991 ) .
the first noted bayesian solution to the calibration problem was presented by hoadley ( 1970 ) .
his work was motivated by the unanswered question in the frequentist community of whether @xmath28 is zero ( or close to zero ) .
hoadley ( 1970 ) justified the use of the @xmath36 estimator ( krutchkoff , 1967 ) by considering the ususal @xmath47-statistic to test the hypothesis that @xmath48 where @xmath49 , @xmath50 the assumption made by hoadley ( 1970 ) reflects that @xmath17 is random and _ a priori _ independent of @xmath51 , so that the joint prior distribution of @xmath52 .
hoadley ( 1970 ) first assumed that @xmath53 had a uniform distribution , @xmath54 but the prior distribution for @xmath17 was not given .
+ hoadley ( 1970 ) shows for @xmath55 ( one observation at the prediction stage ) , that if @xmath17 has a prior density from a student _ t _ distribution with @xmath56 degrees of freedom , a mean of 0 , and a scale parameter @xmath57 the posterior distribution is @xmath58\right),\ ] ] where @xmath42 is the inverse estimator given by ( [ eq : inveq])@xmath59 and @xmath49 .
+ hunter and lamboy ( 1981 ) also considered the calibration problem from a bayesian point of view and is similar to that of hoadley ( 1970 ) because both assume the prior distribution to be @xmath60 where @xmath61 which is the predicted @xmath16 .
the primary difference between their approach and the approach of hoadley ( 1970 ) is that _ a priori _ they assume that @xmath62 and @xmath53 are independent while hoadley ( 1970 ) assumed _ a priori _ that @xmath17 and @xmath53 are independent .
+ hunter and lamboy ( 1981 ) uses an approximation to the posterior distribution of the unknown regressor @xmath17 by @xmath63 where @xmath64 = \left [ \begin{array}{cc } ( { \bf x}^ { ' } { \bf x})^{-1}\hat{\sigma}^{2 } & { \bf 0 } \\ { \bf 0 } & \nicefrac { \hat{\sigma}^{2}}{m } \end{array } \right],\ ] ] with @xmath30 being the classical estimator given in equation ( [ eq : classeq ] ) , @xmath65 denote the element of the @xmath66 row and @xmath67 column from variance - covariance matrix of the joint posterior density of ( @xmath27 , @xmath28 , @xmath62 ) . + the remainder of this paper
is organized as follows .
section [ sec : dyn_cal ] presents the development of the dynamic approaches to the statistical calibration problem . in section [ sec : sim_study ] the results from the simulation study where the dynamics methods are evaluated along with the static approaches are presented . in section [ sec : application ] the proposed methods are applied to microwave radiometer data . in section
[ sec : future ] future work and other considerations are given .
traditional calibration methods assume the regression relationship is `` static '' in time . in many cases
this is false , for example in microwave radiometry the static nature of the relationship is known to change across time .
a dynamic approach can be created by letting the regression coefficients vary through time , @xmath68 where @xmath69 $ ] and is known as the @xmath70 error .
+ the model may have different defining parameters at different times .
one approach is to model @xmath71 and @xmath72 by using random walk type evolutions for the defining parameters , such as : @xmath73 where @xmath74 and @xmath75 are independent zero - mean error terms with finite variances . at any time @xmath76 the calibration problem is given by : @xmath77 bayesian dynamic linear models ( dlms ) approach of west _ et al . _ ( 1985 ) ; west and harrison ( 1997 ) can be employed to achieve this goal .
recall the dlm framework is : @xmath78\label{eq : observe}\\ \mbox{system equation } : \hspace{1 cm } & { \boldsymbol \theta_{t } } = { \bf g}_{t}{\boldsymbol \theta_{t-1 } } + { \boldsymbol \omega_{t } } , & { \boldsymbol \omega}_{t } \sim n_{d}[{\bf 0 , w}]\label{eq : system}\\ \mbox{initial information } : \hspace{1 cm } & ( { \boldsymbol \theta_{0 } } | d_{0 } ) \sim n_{d}[{\bf m_{0 } , c_{0}}],\end{aligned}\ ] ] for some prior mean @xmath79 and variance @xmath80 with the vector of error terms , @xmath81 and @xmath82 independent across time and at any time .
+ to update the model through time west and harrison ( 1997 ) give the following method : a. posterior distribution at @xmath83 : for some mean @xmath84 and variance @xmath85 , + @xmath86 $ ] .
b. prior distribution at time @xmath76 : @xmath87 $ ] , where + @xmath88 and @xmath89 . c. one - step forecast : @xmath90 $ ] , where + @xmath91 and @xmath92 .
d. posterior distribution at time @xmath76 : @xmath93 $ ] , with + @xmath94 and @xmath95 + where + @xmath96 and @xmath97 .
the dlm framework is used to establish the evolving relationship between the fixed design matrix @xmath98 and @xmath99 by estimating @xmath100 , which is a @xmath101 matrix of time - varying regression coefficients @xmath71 and @xmath72 . for our calibration situation
@xmath99 is a @xmath102 matrix of responses and @xmath103 is a known ( @xmath104 ) system matrix .
the error @xmath81 and @xmath82 are independent normally distributed random @xmath102 matrices with zero mean and constant variance - covariance matrices * e * and * w*. for simplification @xmath103 is set equal to @xmath105 , @xmath106 is set equal to @xmath107 and @xmath108 is @xmath109^{-1}$ ] .
the past information is contained in the set @xmath110 .
+ we specify a prior in the first stage of calibration for the unknown variances and derive an algorithm to draw from the posterior distribution of the unknown parameters , @xmath111 the second stage of the calibration experiment consists of using the joint posterior distribution @xmath112 to derive @xmath113 for each draw of @xmath114 .
the estimator for the parameter of interest , @xmath115 , is defined in a manner akin to eisenhart ( 1939 ) ; hunter and lamboy ( 1981 ) ; eno ( 1999 ) , where @xmath116 in the final stage of the calibration experiment , the posterior distribution summary statistics are gathered at each time point @xmath76 .
the posterior median and credible intervals are taken for each @xmath76 across the draws of @xmath113 .
the result of the dynamic calibration experiment is a time series of calibration distributions across time
. we will be able to observe the distributional changes of the system with respect to the calibration reference .
+ the proposed calibration estimator is developed by first considering the joint posterior distribution @xmath114 .
we let @xmath117 denote the vector of unknown dlm dispersion parameters where @xmath118 .
the prior information for the dispersion parameters is described by a prior density @xmath119 which summarizes what is known about the variance parameters before any data are observed . using the bayesian inferential approach ,
the prior information about the parameters must be combined with information contained in the data .
the information provided by the data is captured by the likelihood functions , @xmath120 and @xmath121 for the observation equation and the system equation , respectively .
the combined information is described by the posterior density using the bayes theorem ( bernardo and smith 1994 ) as @xmath122 for our calibration problem it is believe that @xmath123 . to deal with the variance relationship we specify the following prior distributions : @xmath124 prior distributions ( [ eq : prior1 ] ) and ( [ eq : prior2 ] ) ensures the system variance to be less than the observation variance .
since these are proper prior distributions the resulting posterior distribution will also be proper .
+ in the first stage of calibration , the joint distribution of the observations , states , and unknown parameters is as follows : @xmath125 where the likelihood for the observation equation is @xmath126 and the likelihood for the system equation is@xmath127 given the joint distribution above , the posterior distribution is @xmath128 where @xmath129 and @xmath130 ( i.e. @xmath131 is the cumulative mean of the observations up to time @xmath76 ) and @xmath132 .
samples from the posterior distribution in equation ( [ eq : mod1post ] ) are drawn by implementing the sampling importance resampling ( albert 2007 ; givens and hoeting 2005 ) approach . + the development of the estimator in equation ( [ eq : calmod1 ] ) is deterministic in approach .
we present a fully bayesian approach to dynamic calibration that incorporates the uncertainty in estimation .
the second dynamic calibration model is derived by bayes theorem @xmath133 where @xmath134 is the posterior distribution for @xmath135 .
the prior belief for the calibration values is denoted as @xmath136 with the @xmath137 denoting the likelihood function .
+ the objective of any bayesian approach is to obtain the posterior distribution from which inferences can be made . here
the desired posterior is @xmath138 which must be dynamic through time .
we determine the posterior distribution ( [ eq : cali_post ] ) in a similiar manner as described above in equations ( [ eq : mod1post ] ) and ( [ eq : calmod1 ] ) .
in the first stage of the calibration experiment the data is scaled and centered , therefore setting the @xmath139intercept equal to zero and the reference measurements centered at zero .
centering of the data is used to reduce the parameter space
. the posterior distribution can be thought of as : @xmath140 with @xmath141 representing the transformed calibrated value at time @xmath76 and @xmath142 , where @xmath143 is the cumulative mean of the observations . given this information @xmath144
we define the prior distribution @xmath145 the posterior density in equation ( [ eq : center_post ] ) is defined as @xmath146\right\}\ ] ] where @xmath147 .
applying bayes theorem and completing the square , the posterior distribution is @xmath148 with @xmath149 and @xmath150 where tr ( . )
denotes trace of the one - step forecast variance - covariance matrix .
we derive the posterior in equation ( [ eq : cali_post ] ) by drawing from equation ( [ eq : center_distr ] ) and transforming the data back to the original scale as so : @xmath151 where @xmath152 is the mean of the reference measurements vector and @xmath153 is the standard deviation of the reference measurements vector .
+ the dynamic calibration algorithm is developed for both of the approaches using r ( r development core team , 2013 ) and is conducted as below .
[ dyncal ] * algorithm : * dynamic calibration
a simulation study , mirroring the microwave radiometer example in section [ sec : application ] , considers the performance of the proposed dynamic calibration approaches to the static approaches discussed in section [ sec : introduction ] . for notation ,
the calibration methods are labelled as follows : 1 .
@xmath154 is the first deterministic dynamic calibration model given in equation ( [ eq : calmod1 ] ) ; 2 .
@xmath155 is the bayesian dynamic calibration model given by equation ( [ eq : center_distr ] ) ; 3 .
@xmath156 is the @xmath20 approach of eisenhart ( 1939 ) defined in equation ( [ eq : classeq ] ) ; 4 .
@xmath157 is the @xmath36 approach of krutchkoff ( 1967 ) defined in equation ( [ eq : inveq ] ) ; 5 .
@xmath158 is hoadley ( 1970 ) bayesian approach as defined in equation ( [ eq : hoadeq ] ) ; 6 .
@xmath159 is hunter & lamboy ( 1981 ) bayesian approach as defined in equation ( [ eq : hunteq ] ) .
note that static methods @xmath156 , @xmath157 , @xmath158 , and @xmath159 require that model fitting and the calibration take place after all the data has been collected .
this is in contrast to the dynamic methods that both fit the model and generate calibrated values each point through time and hence provide a near real time calibration . in order to assess the performance of the calibration methods 100 datasets
were randomly generated according to @xmath160 where @xmath161 is a known fixed design matrix of reference values .
the number of references measurements used in the study was two and five .
the reference values at the first stage of the simulation study were equally spaced , covering the interval @xmath162 $ ] .
for the two reference case , the fixed design matrix is @xmath163\ ] ] and for the five reference case the design matrix is @xmath164.\ ] ] the vector of regression parameters , @xmath100 , are randomly drawn from a multivariate normal distribution with mean vector @xmath165^{'}$ ] and variance - covariance matrix , @xmath166^{-1}$ ] for @xmath167 , where @xmath168 . for each @xmath76 ,
the random multivariate error vector is @xmath169\ ] ] where the errors are mutually independent .
the relationship of the values for @xmath170 and @xmath171 will be explained later .
+ the dynamic and static calibration methods are evaluated for three distinct system fluctuations , @xmath172 , on the regression slope calculated in the first stage of calibration . the value @xmath172 is added to @xmath72 , therefore making equation ( [ eq : sim_model ] ) @xmath173 for the @xmath67 calibration references .
the three scenarios for the fluctuations @xmath172 are as follows : 1 . a constant zero ( @xmath174 ) for all @xmath76 , representing a stable system ; 2 . a stable system with abrupt shifts ( @xmath175 in system , with @xmath176 ; and 3 . a constant sinusoidal fluctuation ( @xmath177 ) for all @xmath76
. figure [ fig : gain_fluct ] ) explains the relationship of @xmath172 across time .
+ ; ( b ) @xmath178 with @xmath176 ; ( c ) @xmath179,title="fig:",width=192 ] ; ( b ) @xmath178 with @xmath176 ; ( c ) @xmath179,title="fig:",width=192 ] + ; ( b ) @xmath178 with @xmath176 ; ( c ) @xmath179,title="fig:",width=192 ] the magnitude and relationship of the variance pair @xmath180 influence the dlm and hence to study this influence we set the variances to reflect various _ signal - to - noise ratios_. the true values for @xmath170 and @xmath171 used in the simulation study are @xmath181 and @xmath182 , respectively . petris _
( 2009 ) define the signal - to - noise ratio @xmath0 as follows : @xmath183 the signal - to - noise ratios @xmath0 in the simulation study were examined in two sets .
first , @xmath0 is set equal to 10 , 100 , and 1000 .
next , the ratio @xmath0 was set to equal 2 , 20 , and 200 .
the variety of @xmath0 values allow us to examine the methods under different levels of noise .
each simulation is repeated 100 times for both the 2- and 5-point calibration models , thus providing us with 36 possible models for examination from the settings of @xmath0 .
+ after the data was fit with each of the methods we considered the following measures for assessing the performance of the dynamic methods compared to the familiar static approaches : ( 1 ) average mean square error ; ( 2 ) average coverage probability ; and ( 3 ) average interval width . for each of the simulated data sets , the mean squared error ( @xmath184 )
is calculated as @xmath185 the @xmath184 are averaged across the 100 simulated data sets thus deriving an average mean squared error ( @xmath186 ) as @xmath187 the coverage probability based on the @xmath1 coverage interval is estimated for all of the calibration methods .
the coverage interval for the dynamic and static bayesian approaches is the @xmath1 credible interval and the @xmath1 confidence interval is used for the frequentist methods .
note , that for credible intervals @xmath188 is the 0.025 posterior quantile for @xmath115 , and @xmath189 is the 0.975 posterior quantile for @xmath115 , where @xmath115 is the true value of the calibration target from the second stage of experimentation , then @xmath190 is a @xmath1 credible interval . the coveraged probability ( @xmath191 ) is calculated as such @xmath192 where @xmath193 = \left\ { \begin{array}{ll } 0 & \mbox{if $ x_{0 t } \not\in ( x_{0t}^{l } , x_{0t}^{u})$};\\ & \\ 1 & \mbox{if $ x_{0 t } \in ( x_{0t}^{l } , x_{0t}^{u})$}.\end{array } \right.\ ] ] the average coverage probability ( @xmath194 ) is calculated by averaging across the number of replications in the simulation study , where @xmath195 another quantity of interest to compare the average interval widths ( @xmath196 ) for the methods , where the average interval widths ( @xmath197 ) across the simulated time series is calculated as follows : @xmath198 with the average interval width across the simulation study given as @xmath199 where @xmath200 is the average interval width for the @xmath67 simulation replicate .
the performance of the dynamic calibration approaches will be assess using the average coverage probability ( @xmath194 ) , average interval width ( @xmath196 ) and average mean square ( @xmath186 ) .
+ we consider the performance of the methods under two conditions : interpolation and extrapolation .
interpolation case is of interest to understand how the method performed when the calibrated time series is within the range of the reference values , [ 20 , 100 ] .
extrapolation case also conducted to examine the methods when @xmath115 falls outside of the range of the calibration references , where @xmath201 .
while it not preferable to do extrapolation in the regression case , it is often done in practice in microwave radiometry . +
all simulations were carried out on the compile server running @xmath202 ( r development core team , 2013 ) at virginia commonwealth university .
the compile server has a linux os with 16 cpu cores and 32 gb ram .
each iteration in the study took approximately 15 minutes with a total of 25.63 hours . in the following tables ,
the simulation results for the dynamic and static calibration methods are provided .
the results of simulation studies provide insight into the properties of the calibration approaches .
the results in tables [ tab : constant_inter1 ] and [ tab : constant_inter2 ] indicate that all of the estimators do a good job at approximating the true values of @xmath115 when the gain flucuation @xmath172 is set to 0 . even in this case
we see as the signal - to - noise ratio @xmath0 increases so does the @xmath186 values .
all of the methods have an average coverage probability @xmath194 of 1 or close .
the high coverage rate is of no surprise for a stable system .
there does not appear to be an advantage by including more reference measurements ( i.e 2- or 5-points ) in the model when the system is stable in time .
the clear difference is the @xmath196 values for the dynamic methods compared to the static methods . in tables [ tab : constant_inter1 ] and [ tab : constant_inter2 ] when @xmath203 and @xmath204 , the interval for the dynamic methods is wider than those of the four static methods but as @xmath0 increases the interval width of the dynamic methods remain nearly unchanged as the interval widths for the static methods are 4 to 5 times wider .
+ the simulation results for the stepped gain fluctuations are provided in tables [ tab : stepped_inter1 ] and [ tab : stepped_inter2 ] . clearly
the presence of the stepped @xmath172 has an effect on the fit of the models .
the results in tables [ tab : stepped_inter1 ] show that in nearly all cases , the two dynamic methods @xmath154 and @xmath155 have @xmath186 values smaller than the two static bayesian approaches . the @xmath186 values for the dynamic methods are reasonably lower for @xmath205 .
when @xmath206 , notice the dynamic models @xmath154 and @xmath155 have smaller average mean square errors smaller than the static method @xmath157 .
the average coverage probability @xmath194 is comparable for all of the methods and number of references .
the dynamic methods consistently have shorter interval widths .
the widths of the 95% credible intervals for @xmath154 and @xmath155 is not affected by the increases in @xmath0 .
+ [ comparison of calibration approaches when interpolating to estimate @xmath115 without gain fluctuations ] comparison of calibration approaches when interpolating to estimate @xmath115 without gain fluctuations based on 100 data sets .
avmse is the average mean squared error , avcp is the average coverage probability , and aviw is the average 95% interval width .
the signal - to - noise ratio is denoted as @xmath0 .
ccrcr|rcr|rcr + & & & @xmath207 & & & @xmath208 & & & @xmath206 & + ref . & model & avmse & avcp & aviw & avmse & avcp & aviw & avmse & avcp & aviw + 2 & @xmath154 & 0.0008 & 0.995 & 2.519 & 0.0035 & 0.983 & 2.523 & 0.0307 & 0.939 & 2.517 + & @xmath155 & 0.0012 & 1.000 & 3.782 & 0.0038 & 1.000 & 3.782 & 0.0308 & 1.000 & 3.782 + & @xmath156 & 0.0001 & 1.000 & 1.224 & 0.0012 & 1.000 & 3.868 & 0.0123 & 1.000 & 12.229 + & @xmath157 & 0.0001 & 1.000 & 1.223 & 0.0016 & 1.000 & 3.863 & 0.0335 & 1.000 & 12.168 + & @xmath158 & 0.0002 & 0.997 & 1.182 & 0.0022 & 1.000 & 3.866 & 0.0386 & 1.000 & 12.177 + & @xmath159 & 0.0014 & 1.000 & 1.458 & 0.0139 & 1.000 & 4.606 & 0.1391 & 1.000 & 14.565 + 5 & @xmath154 & 0.0008 & 0.995 & 2.496 & 0.0035 & 0.983 & 2.509 & 0.0307 & 0.941 & 2.514 + & @xmath155 & 0.0013 & 1.000 & 3.983 & 0.0039 & 1.000 & 3.983 & 0.0307 & 1.000 & 3.983 + & @xmath156 & 0.0001 & 1.000 & 1.223 & 0.0012 & 1.000 & 3.865 & 0.0123 & 1.000 & 12.220 + & @xmath157 & 0.0001 & 1.000 & 1.222 & 0.0022 & 1.000 & 3.860 & 0.0813 & 1.000 & 12.113 + & @xmath158 & 0.0002 & 1.000 & 1.223 & 0.0023 & 1.000 & 3.861 & 0.0792 & 1.000 & 12.116 + & @xmath159 & 0.0014 & 1.000 & 1.457 & 0.0139 & 1.000 & 4.604 & 0.1069 & 1.000 & 10.748 + [ tab : constant_inter1 ] [ comparison of calibration approaches when interpolating to estimate @xmath115 without gain fluctuations ] comparison of calibration approaches when interpolating to estimate @xmath115 without gain fluctuations based on 100 data sets .
avmse is the average mean squared error , avcp is the average coverage probability , and aviw is the average 95% interval width .
the signal - to - noise ratio is denoted as @xmath0 .
ccrcr|rcr|rcr + & & & @xmath209 & & & @xmath210 & & & @xmath205 & + ref . & model & avmse & avcp & aviw & avmse & avcp & aviw & avmse & avcp & aviw + 2&@xmath154 & 0.0012 & 0.992 & 2.519 & 0.0041 & 0.981 & 2.520 & 0.0323 & 0.939 & 2.528 + & @xmath155 & 0.0015 & 1.000 & 3.782 & 0.0044 & 1.000 & 3.782 & 0.0325 & 1.000 & 3.782 + & @xmath156 & 0.0001 & 1.000 & 1.230 & 0.0010 & 1.000 & 3.871 & 0.0114 & 1.000 & 12.231 + & @xmath157 & 0.0001 & 1.000 & 1.229 & 0.0012 & 1.000 & 3.866 & 0.0314 & 1.000 & 12.170 + & @xmath158 & 0.0001 & 1.000 & 1.230 & 0.0019 & 1.000 & 3.869 & 0.0371 & 1.000 & 12.179 + & @xmath159 & 0.0190 & 1.000 & 1.155 & 0.0243 & 1.000 & 3.767 & 0.1381 & 1.000 & 14.567 + 5&@xmath154 & 0.0011 & 0.992 & 2.508 & 0.0041 & 0.981 & 2.510 & 0.032 & 0.939 & 2.514 + & @xmath155 & 0.0017 & 1.000 & 3.983 & 0.0045 & 1.000 & 3.983 & 0.032 & 1.000 & 3.983 + & @xmath156 & 0.0001 & 1.000 & 1.228 & 0.0010 & 1.000 & 3.868 & 0.011 & 1.000 & 12.222 + & @xmath157 & 0.0001 & 1.000 & 1.227 & 0.0019 & 1.000 & 3.863 & 0.081 & 1.000 & 12.114 + & @xmath158 & 0.0001 & 1.000 & 1.227 & 0.0021 & 1.000 & 3.864 & 0.082 & 1.000 & 12.118 + & @xmath159 & 0.0013 & 1.000 & 1.462 & 0.0137 & 1.000 & 4.607 & 0.138 & 1.000 & 14.560 + [ tab : constant_inter2 ] [ comparison of calibration approaches when interpolating to estimate @xmath115 with stepped gain fluctuations ] comparison of calibration approaches when interpolating to estimate @xmath115 with stepped gain fluctuations based on 100 data sets .
avmse is the average mean squared error , avcp is the average coverage probability , and aviw is the average 95% interval width .
the signal - to - noise ratio is denoted as @xmath0 .
ccrcr|rcr|rcr + & & & @xmath207 & & & @xmath208 & & & @xmath206 & + ref .
& model & avmse & avcp & aviw & avmse & avcp & aviw & avmse & avcp & aviw + 2&@xmath154 & 0.0191 & 0.961 & 2.509 & 0.0198 & 0.953 & 2.506 & 0.0406 & 0.926 & 2.543 + & @xmath155 & 0.0196 & 1.000 & 3.782 & 0.0201 & 1.000 & 3.782 & 0.0408 & 1.000 & 3.783 + & @xmath156 & 0.0001 & 1.000 & 9.094 & 0.0004 & 1.000 & 9.813 & 0.0094 & 1.000 & 15.209 + & @xmath157 & 0.0046 & 1.000 & 9.065 & 0.0073 & 1.000 & 9.779 & 0.0528 & 1.000 & 15.098 + & @xmath158 & 0.0859 & 1.000 & 9.072 & 0.0838 & 1.000 & 9.786 & 0.1866 & 1.000 & 15.109 + & @xmath159 & 0.1399 & 1.000 & 10.830 & 0.0823 & 1.000 & 11.687 & 0.1836 & 1.000 & 18.115 + 5&@xmath154 & 0.0191 & 0.961 & 2.510 & 0.0197 & 0.954 & 2.511 & 0.0405 & 0.924 & 2.516 + & @xmath155 & 0.0196 & 1.000 & 3.983 & 0.0201 & 1.000 & 3.983 & 0.0405 & 1.000 & 3.983 + & @xmath156 & 0.0001 & 1.000 & 9.087 & 0.0004 & 1.000 & 9.806 & 0.0094 & 1.000 & 15.199 + & @xmath157 & 0.0184 & 1.000 & 9.041 & 0.0267 & 1.000 & 9.749 & 0.1620 & 1.000 & 14.995 + & @xmath158 & 0.0199 & 1.000 & 9.044 & 0.0267 & 1.000 & 9.752 & 0.1559 & 1.000 & 14.999 + & @xmath159 & 0.0706 & 1.000 & 10.826 & 0.0618 & 1.000 & 8.625 & 0.1742 & 1.000 & 15.091 + [ tab : stepped_inter1 ] [ comparison of calibration approaches when interpolating to estimate @xmath115 with stepped gain fluctuations ] comparison of calibration approaches when interpolating to estimate @xmath115 with stepped gain fluctuations based on 100 data sets .
avmse is the average mean squared error , avcp is the average coverage probability , and aviw is the average 95% interval width .
the signal - to - noise ratio is denoted as @xmath0 .
ccrcr|rcr|rcr + & & & @xmath209 & & & @xmath210 & & & @xmath205 & + ref . & model & avmse & avcp & aviw & avmse & avcp & aviw & avmse & avcp & aviw + 2&@xmath154 & 0.0209 & 0.957 & 2.520 & 0.0219 & 0.950 & 2.522 & 0.0436 & 0.921 & 2.526 + & @xmath155 & 0.0214 & 1.000 & 3.782 & 0.0222 & 1.000 & 3.782 & 0.0438 & 1.000 & 3.782 + & @xmath156 & 0.0001 & 1.000 & 9.103 & 0.0003 & 1.000 & 9.822 & 0.0086 & 1.000 & 15.216 + & @xmath157 & 0.0047 & 1.000 & 9.075 & 0.0073 & 1.000 & 9.788 & 0.0511 & 1.000 & 15.105 + & @xmath158 & 0.0084 & 1.000 & 9.081 & 0.0115 & 1.000 & 9.795 & 0.0601 & 1.000 & 15.116 + & @xmath159 & 0.0709 & 1.000 & 10.842 & 0.0826 & 1.000 & 11.698 & 0.2054 & 1.000 & 18.122 + 5&@xmath154 & 0.0209 & 0.957 & 2.509 & 0.0218 & 0.949 & 2.511 & 0.0436 & 0.920 & 2.516 + & @xmath155 & 0.0214 & 1.000 & 3.983 & 0.0221 & 1.000 & 3.983 & 0.0435 & 1.000 & 3.983 + & @xmath156 & 0.0001 & 1.000 & 9.096 & 0.0003 & 1.000 & 9.815 & 0.0086 & 1.000 & 15.205 + & @xmath157 & 0.0185 & 1.000 & 9.050 & 0.0267 & 1.000 & 9.758 & 0.1616 & 1.000 & 15.002 + & @xmath158 & 0.0199 & 1.000 & 9.053 & 0.0281 & 1.000 & 9.761 & 0.1641 & 1.000 & 15.006 + & @xmath159 & 0.0708 & 1.000 & 10.836 & 0.0825 & 1.000 & 11.693 & 0.2052 & 1.000 & 18.114 + [ tab : stepped_inter2 ] [ comparison of calibration approaches when interpolating to estimate @xmath115 with sinusoidal gain fluctuations ] comparison of calibration approaches when interpolating to estimate @xmath115 with sinusoidal gain fluctuations based on 100 data sets .
avmse is the average mean squared error , avcp is the average coverage probability , and aviw is the average 95% interval width .
the signal - to - noise ratio is denoted as @xmath0 .
ccrcr|rcr|rcr + & & & @xmath207 & & & @xmath208 & & & @xmath206 & + ref . & model & avmse & avcp & aviw & avmse & avcp & aviw & avmse & avcp & aviw + 2 & @xmath154 & 4.4088 & 0.628 & 2.657 & 4.4794 & 0.629 & 2.648 & 4.7214 & 0.638 & 2.681 + & @xmath155 & 4.4002 & 0.829 & 3.783 & 4.4708 & 0.825 & 3.783 & 4.7123 & 0.810 & 3.783 + & @xmath156 & 0.0001 & 1.000 & 21.980 & 0.0012 & 1.000 & 22.307 & 0.0123 & 1.000 & 25.206 + & @xmath157 & 0.1541 & 1.000 & 21.665 & 0.1670 & 1.000 & 21.978 & 0.2943 & 1.000 & 24.738 + & @xmath158 & 0.1689 & 0.975 & 20.933 & 0.1868 & 1.000 & 21.994 & 0.3174 & 1.000 & 24.757 + & @xmath159 & 0.4127 & 1.000 & 26.178 & 0.4258 & 1.000 & 26.567 & 0.5531 & 1.000 & 30.020 + 5&@xmath154 & 4.4087 & 0.628 & 2.646 & 4.4793 & 0.630 & 2.648 & 4.7214 & 0.635 & 2.653 + & @xmath155 & 4.3906 & 0.845 & 3.984 & 4.4609 & 0.839 & 3.984 & 4.7023 & 0.824 & 3.984 + & @xmath156 & 0.0001 & 1.000 & 21.964 & 0.0012 & 1.000 & 22.291 & 0.0123 & 1.000 & 25.188 + & @xmath157 & 0.5810 & 1.000 & 21.371 & 0.6218 & 1.000 & 21.671 & 1.0152 & 1.000 & 24.306 + & @xmath158 & 0.5956 & 1.000 & 21.377 & 0.5909 & 1.000 & 21.678 & 0.9628 & 1.000 & 24.314 + & @xmath159 & 0.4123 & 1.000 & 26.166 & 0.3087 & 1.000 & 18.973 & 0.4658 & 1.000 & 25.009 + [ tab : sine_inter1 ] [ comparison of calibration approaches when interpolating to estimate @xmath115 with sinusoidal gain fluctuations ] comparison of calibration approaches when interpolating to estimate @xmath115 with sinusoidal gain fluctuations based on 100 data sets .
avmse is the average mean squared error , avcp is the average coverage probability , and aviw is the average 95% interval width .
the signal - to - noise ratio is denoted as @xmath0 .
ccrcr|rcr|rcr + & & & @xmath209 & & & @xmath210 & & & @xmath205 & + ref . &
model & avmse & avcp & aviw & avmse & avcp & aviw & avmse & avcp & aviw + 2&@xmath154 & 4.4504 & 0.625 & 2.658 & 4.5213 & 0.628 & 2.660 & 4.7643 & 0.6331 & 2.665 + & @xmath155 & 4.4419 & 0.828 & 3.783 & 4.5127 & 0.823 & 3.783 & 4.7553 & 0.8083 & 3.783 + & @xmath156 & 0.0001 & 1.000 & 21.968 & 0.0010 & 1.000 & 22.295 & 0.0114 & 1.000 & 25.196 + & @xmath157 & 0.1538 & 1.000 & 21.653 & 0.1672 & 1.000 & 21.966 & 0.2896 & 1.000 & 24.729 + & @xmath158 & 0.1732 & 1.000 & 21.669 & 0.1867 & 1.000 & 21.982 & 0.3127 & 1.000 & 24.748 + & @xmath159 & 0.4122 & 1.000 & 26.164 & 0.4253 & 1.000 & 26.553 & 0.5518 & 1.000 & 30.008 + 5&@xmath154 & 4.4504 & 0.625 & 2.647 & 4.5213 & 0.627 & 2.648 & 4.7643 & 0.633 & 2.654 + & @xmath155 & 4.4322 & 0.844 & 3.984 & 4.5029 & 0.838 & 3.984 & 4.7451 & 0.823 & 3.984 + & @xmath156 & 0.0001 & 1.000 & 21.952 & 0.0010 & 1.000 & 22.278 & 0.0114 & 1.000 & 25.178 + & @xmath157 & 0.5799 & 1.000 & 21.359 & 0.6206 & 1.000 & 21.660 & 1.0133 & 1.000 & 24.297 + & @xmath158 & 0.5851 & 1.000 & 21.366 & 0.6256 & 1.000 & 21.667 & 1.0183 & 1.000 & 24.304 + & @xmath159 & 0.4119 & 1.000 & 26.152 & 0.4247 & 1.000 & 26.541 & 0.5513 & 1.000 & 29.995 + [ tab : sine_inter2 ] the results provided in tables [ tab : sine_inter1 ] and [ tab : sine_inter2 ] summarize the performance of the methods when the gain fluctuation is sinusoidal noise .
the results for @xmath0 values of 10 , 100 , and 1000 are given in table [ tab : sine_inter1 ] with @xmath211 and 200 given in table [ tab : sine_inter2 ] .
when @xmath172 is sinusoidal , the @xmath186 values for the dynamic methods are consistently larger than any of the static methods .
for all of the chosen @xmath0 values , the @xmath194 is considerably lower than the opposing methods .
the dynamic methods still have average interval widths extremely shorter than any of the static methods .
the @xmath196 is constant across the signal - to - noise ratios .
+ the simulation study shows that methods @xmath154 and @xmath155 do a good job at estimating calibrated values that are interior to the range of reference measurements .
both methods display high coverage probabilities in the presence of drifting parameters .
for the three possible gain fluctuations , the interval widths for the dynamic methods were consistently shorter than the static calibration approaches .
when fitting data where there is a definite linear relationship the dynamic methods are invariant to the number of reference measurements .
when using the proposed methods in this paper , not much will be gained by using more than 2 reference measurements .
overall , when interpolating to estimate @xmath115 , the dynamic methods outperform the static bayesian approaches across the different signal - to - noise ratios . in the following section the performance of the dynamic methods
are assessed when the calibrated values fall outside of the range of reference measurements . at this point in the paper
we examine the calibration approaches when the calibrated values are outside of the reference measurements .
the range of the measurement references is from 20 to 100 .
the true @xmath115 behaved as a random walk bounded between 100 and 110 .
we assessed the performance of the dynamic methods under three possible gain fluctuation patterns .
first , the simulation study is conducted without the presence of additional gain fluctuation ( i.e. @xmath212 ) ; second , the gain @xmath172 is a stepped pattern influencing the time - varying slope @xmath72 over time ; lastly , a sinusoidal @xmath172 is added to @xmath72 . just as the previous results ,
the methods are assessed by the average mean square error ( @xmath186 ) , average coverage probability ( @xmath194 ) , and the average interval width ( @xmath196 ) under different signal - noise - ratios .
+ [ comparison of calibration approaches when extrapolating to estimate @xmath115 without gain fluctuations ] comparison of calibration approaches when extrapolating to estimate @xmath115 without gain fluctuations based on 100 data sets .
avmse is the average mean squared error , avcp is the average coverage probability , and aviw is the average 95% interval width .
the signal - to - noise ratio is denoted as @xmath0 .
ccrcr|rcr|rcr + & & & @xmath207 & & & @xmath208 & & & @xmath206 & + ref . & model & avmse & avcp & aviw & avmse & avcp & aviw & avmse & avcp & aviw + 2&@xmath154 & 0.0018 & 1.000 & 5.255 & 0.0043 & 1.000 & 5.255 & 0.0309 & 1.000 & 5.255 + & @xmath155 & 0.0016 & 1.000 & 3.910 & 0.0042 & 1.000 & 3.910 & 0.0311 & 1.000 & 3.910 + & @xmath156 & 0.0001 & 1.000 & 1.224 & 0.0012 & 1.000 & 3.869 & 0.0123 & 1.000 & 12.234 + & @xmath157 & 0.0001 & 1.000 & 1.223 & 0.0001 & 1.000 & 3.863 & 0.1019 & 1.000 & 12.168 + & @xmath158 & 0.0001 & 1.000 & 1.225 & 0.0008 & 1.000 & 3.867 & 0.1115 & 1.000 & 12.181 + & @xmath159 & 0.0014 & 1.000 & 1.458 & 0.0139 & 1.000 & 4.606 & 0.1391 & 1.000 & 14.565 + 5&@xmath154 & 0.0019 & 1.000 & 5.233 & 0.0043 & 1.000 & 5.233 & 0.0309 & 1.000 & 5.233 + & @xmath155 & 0.0029 & 1.000 & 4.106 & 0.0054 & 1.000 & 4.106 & 0.0323 & 1.000 & 4.106 + & @xmath156 & 0.0001 & 1.000 & 1.223 & 0.0012 & 1.000 & 3.866 & 0.0123 & 1.000 & 12.224 + & @xmath157 & 0.0001 & 1.000 & 1.222 & 0.0027 & 1.000 & 3.860 & 0.5502 & 1.000 & 12.113 + & @xmath158 & 0.0001 & 1.000 & 1.223 & 0.0030 & 1.000 & 3.862 & 0.5566 & 1.000 & 12.120 + & @xmath159 & 0.0014 & 1.000 & 1.457 & 0.0139 & 1.000 & 4.604 & 0.1389 & 1.000 & 14.558 + [ tab : constant_extrap1 ] [ comparison of calibration approaches when extrapolating to estimate @xmath115 without gain fluctuations ] comparison of calibration approaches when extrapolating to estimate @xmath115 without gain fluctuations based on 100 data sets .
avmse is the average mean squared error , avcp is the average coverage probability , and aviw is the average 95% interval width .
the signal - to - noise ratio is denoted as @xmath0 .
ccrcr|rcr|rcr + & & & @xmath209 & & & @xmath210 & & & @xmath205 & + ref . & model & avmse & avcp & aviw & avmse & avcp & aviw & avmse & avcp & aviw + 2&@xmath154 & 0.0031 & 1.000 & 5.253 & 0.0060 & 1.000 & 5.253 & 0.0340 & 1.000 & 5.253 + & @xmath155 & 0.0034 & 1.000 & 3.910 & 0.0064 & 1.000 & 3.910 & 0.0347 & 1.000 & 3.910 + & @xmath156 & 0.0001 & 1.000 & 1.230 & 0.0008 & 1.000 & 3.872 & 0.0107 & 1.000 & 12.236 + & @xmath157 & 0.0001 & 1.000 & 1.229 & 0.0003 & 1.000 & 3.866 & 0.1068 & 1.000 & 12.170 + & @xmath158 & 0.0001 & 1.000 & 1.230 & 0.0010 & 1.000 & 3.871 & 0.1164 & 1.000 & 12.183 + & @xmath159 & 0.0013 & 1.000 & 1.465 & 0.0135 & 1.000 & 4.610 & 0.1376 & 1.000 & 14.567 + 5&@xmath154 & 0.0032 & 1.000 & 5.231 & 0.0060 & 1.000 & 5.231 & 0.0340 & 1.000 & 5.232 + & @xmath155 & 0.0053 & 1.000 & 4.106 & 0.0083 & 1.000 & 4.106 & 0.0365 & 1.000 & 4.106 + & @xmath156 & 0.0001 & 1.000 & 1.228 & 0.0008 & 1.000 & 3.869 & 0.0107 & 1.000 & 12.226 + & @xmath157 & 0.0001 & 1.000 & 1.227 & 0.0035 & 1.000 & 3.863 & 0.5616 & 1.000 & 12.114 + & @xmath158 & 0.0001 & 1.000 & 1.228 & 0.0039 & 1.000 & 3.865 & 0.5680 & 1.000 & 12.121 + & @xmath159 & 0.0013 & 1.000 & 1.462 & 0.0135 & 1.000 & 4.607 & 0.1375 & 1.000 & 14.560 + [ tab : constant_extrap2 ] the results are provided in tables [ tab : constant_extrap1 ] and[tab : constant_extrap2 ] for the statistical calibration methods without gain fluctuations .
the performance of the proposed method is stable across the signal - to - noise ratios .
a point of interest is the reported @xmath196 values for methods @xmath154 and @xmath155 .
we see for @xmath207 and @xmath209 that the @xmath196 is 3 to 5 times wider than those for the static approaches . when @xmath208 and @xmath210 the interval width for all competing methods are relatively close .
the dynamic approaches outperform the static methods in noisy conditions such as @xmath206 and @xmath205 .
the interval widths for the dynamic methods are considerably shorter than the those for the static methods .
the simulation results reveal that when the data is characteristic of having a large signal - to - noise ratio , the dynamic methods , @xmath154 and @xmath155 , will outperform static bayesian approaches and the inverse approach .
+ [ comparison of calibration approaches when extrapolating to estimate @xmath115 with stepped gain fluctuations ] comparison of calibration approaches when extrapolating to estimate @xmath115 with stepped gain fluctuations based on 100 data sets .
avmse is the average mean squared error , avcp is the average coverage probability , and aviw is the average 95% interval width . the signal - to - noise ratio is denoted as @xmath0 .
ccrcr|rcr|rcr + & & & @xmath207 & & & @xmath208 & & & @xmath206 & + ref . & model & avmse & avcp & aviw & avmse & avcp & aviw & avmse & avcp & aviw + 2&@xmath154 & 0.0206 & 1.000 & 5.247 & 0.0210 & 1.000 & 5.247 & 0.0412 & 1.000 & 5.247 + & @xmath155 & 0.0225 & 1.000 & 3.910 & 0.0230 & 1.000 & 3.910 & 0.0435 & 1.000 & 3.910 + & @xmath156 & 0.0001 & 1.000 & 9.097 & 0.0004 & 1.000 & 9.817 & 0.0094 & 1.000 & 15.215 + & @xmath157 & 0.0581 & 1.000 & 9.065 & 0.0656 & 1.000 & 9.779 & 0.3191 & 1.000 & 15.098 + & @xmath158 & 0.0634 & 1.000 & 9.075 & 0.0718 & 1.000 & 9.789 & 0.3361 & 1.000 & 15.115 + & @xmath159 & 0.0707 & 1.000 & 10.830 & 0.0826 & 1.000 & 11.687 & 0.2060 & 1.000 & 18.115 + 5&@xmath154 & 0.0209 & 1.000 & 5.226 & 0.0213 & 1.000 & 5.226 & 0.0412 & 1.000 & 5.226 + & @xmath155 & 0.0268 & 1.000 & 4.106 & 0.0273 & 1.00 & 4.106 & 0.0483 & 1.000 & 4.106 + & @xmath156 & 0.0001 & 1.000 & 9.090 & 0.0004 & 1.000 & 9.809 & 0.0094 & 1.000 & 15.203 + & @xmath157 & 0.2274 & 1.000 & 9.041 & 0.2812 & 1.000 & 9.749 & 1.4628 & 1.000 & 14.995 + & @xmath158 & 0.2307 & 1.000 & 9.047 & 0.2851 & 1.000 & 9.755 & 1.4744 & 1.000 & 15.004 + & @xmath159 & 0.0706 & 1.000 & 10.826 & 0.0825 & 1.000 & 11.682 & 0.2058 & 1.000 & 18.106 + [ tab : stepped_extrap1 ] [ comparison of calibration approaches when extrapolating to estimate @xmath115 with stepped gain fluctuations ] comparison of calibration approaches when extrapolating to estimate @xmath115 with stepped gain fluctuations based on 100 data sets .
avmse is the average mean squared error , avcp is the average coverage probability , and aviw is the average 95% interval width .
the signal - to - noise ratio is denoted as @xmath0 .
ccrcr|rcr|rcr + & & & @xmath209 & & & @xmath210 & & & @xmath205 & + ref . & model & avmse & avcp & aviw & avmse & avcp & aviw & avmse & avcp & aviw + 2&@xmath154 & 0.0242 & 1.000 & 5.245 & 0.0250 & 1.000 & 5.245 & 0.0466 & 1.000 & 5.245 + & @xmath155 & 0.0266 & 1.000 & 3.910 & 0.0275 & 1.000 & 3.910 & 0.0494 & 1.000 & 3.910 + & @xmath156 & 0.0001 & 1.000 & 9.106 & 0.0002 & 1.000 & 9.826 & 0.0080 & 1.000 & 15.222 + & @xmath157 & 0.0620 & 1.000 & 9.075 & 0.0698 & 1.000 & 9.788 & 0.3284 & 1.000 & 15.105 + & @xmath158 & 0.0674 & 1.000 & 9.085 & 0.0760 & 1.000 & 9.799 & 0.3447 & 1.000 & 15.121 + & @xmath159 & 0.0710 & 1.000 & 10.842 & 0.0825 & 1.000 & 11.698 & 0.2048 & 1.000 & 18.122 + 5&@xmath154 & 0.0245 & 1.000 & 5.224 & 0.0254 & 1.000 & 5.224 & 0.0427 & 1.000 & 5.226 + & @xmath155 & 0.0315 & 1.000 & 4.106 & 0.0324 & 1.000 & 4.106 & 0.0485 & 1.000 & 4.106 + & @xmath156 & 0.0001 & 1.000 & 9.099 & 0.0002 & 1.000 & 9.818 & 0.0089 & 1.000 & 14.255 + & @xmath157 & 0.2354 & 1.000 & 9.050 & 0.2902 & 1.000 & 9.758 & 1.1896 & 1.000 & 14.078 + & @xmath158 & 0.2388 & 1.000 & 9.056 & 0.2941 & 1.000 & 9.764 & 1.1995 & 1.000 & 14.086 + & @xmath159 & 0.0709 & 1.000 & 10.836 & 0.0824 & 1.000 & 11.693 & 0.1831 & 1.000 & 16.977 + [ tab : stepped_extrap2 ] [ comparison of calibration approaches when extrapolating to estimate @xmath115 with sinusoidal gain fluctuations ] comparison of calibration approaches when extrapolating to estimate @xmath115 with sinusoidal gain fluctuations based on 100 data sets .
avmse is the average mean squared error , avcp is the average coverage probability , and aviw is the average 95% interval width .
the signal - to - noise ratio is denoted as @xmath0 .
ccrcr|rcr|rcr + & & & @xmath207 & & & @xmath208 & & & @xmath206 & + ref . & model & avmse & avcp & aviw & avmse & avcp & aviw & avmse & avcp & aviw + 2&@xmath154 & 4.4096 & 0.873 & 5.127 & 4.4800 & 0.872 & 5.127 & 4.7214 & 0.866 & 5.127 + & @xmath155 & 4.4410 & 0.833 & 3.904 & 4.5114 & 0.825 & 3.904 & 4.7530 & 0.813 & 3.904 + & @xmath156 & 0.0001 & 1.000 & 21.988 & 0.0012 & 1.000 & 22.315 & 0.0123 & 1.000 & 25.216 + & @xmath157 & 1.8193 & 1.000 & 21.665 & 1.8636 & 1.000 & 21.978 & 2.7760 & 1.000 & 24.739 + & @xmath158 & 1.8602 & 1.000 & 21.688 & 1.9056 & 1.000 & 22.002 & 2.8312 & 1.000 & 24.766 + & @xmath159 & 0.4127 & 1.000 & 26.178 & 0.4258 & 1.000 & 26.567 & 0.5531 & 1.000 & 30.020 + 5&@xmath154 & 4.4105 & 0.872 & 5.106 & 4.4808 & 0.872 & 5.106 & 4.7222 & 0.866 & 5.107 + & @xmath155 & 4.4889 & 0.842 & 4.100 & 4.5593 & 0.835 & 4.101 & 4.8007 & 0.822 & 4.100 + & @xmath156 & 0.0001 & 1.000 & 21.971 & 0.0012 & 1.000 & 22.297 & 0.0123 & 1.000 & 25.195 + & @xmath157 & 6.9539 & 1.000 & 21.371 & 7.2327 & 1.000 & 21.671 & 11.0337 & 1.000 & 24.306 + & @xmath158 & 6.9852 & 1.000 & 21.383 & 7.2650 & 1.000 & 21.684 & 11.0772 & 1.000 & 24.320 + & @xmath159 & 0.4123 & 1.000 & 26.166 & 0.4254 & 1.000 & 26.555 & 0.5526 & 1.000 & 30.007 + [ tab : sine_extrap1 ] [ comparison of calibration approaches when extrapolating to estimate @xmath115 with sinusoidal gain fluctuations ] comparison of calibration approaches when extrapolating to estimate @xmath115 with sinusoidal gain fluctuations based on 100 data sets .
avmse is the average mean squared error , avcp is the average coverage probability , and aviw is the average 95% interval width .
the signal - to - noise ratio is denoted as @xmath0 .
ccrcr|rcr|rcr + & & & @xmath209 & & & @xmath210 & & & @xmath205 & + ref . & model & avmse & avcp & aviw & avmse & avcp & aviw & avmse & avcp & aviw + 2&@xmath154 & 4.4491 & 0.872 & 5.125 & 4.5199 & 0.871 & 5.125 & 4.7626 & 0.866 & 5.126 + & @xmath155 & 4.4809 & 0.828 & 3.904 & 4.5518 & 0.821 & 3.904 & 4.7948 & 0.807 & 3.904 + & @xmath156 & 0.0001 & 1.000 & 21.976 & 0.0008 & 1.000 & 22.303 & 0.0107 & 1.000 & 25.205 + & @xmath157 & 1.8350 & 1.000 & 21.653 & 1.8796 & 1.000 & 21.966 & 2.7956 & 1.000 & 24.729 + & @xmath158 & 1.8759 & 1.000 & 21.676 & 1.9216 & 1.000 & 21.990 & 2.8508 & 1.000 & 24.756 + & @xmath159 & 0.4123 & 1.000 & 26.164 & 0.4250 & 1.000 & 26.553 & 0.5511 & 1.000 & 30.008 + 5&@xmath154 & 4.4497 & 0.872 & 5.10 & 4.5205 & 0.871 & 5.105 & 4.7633 & 0.865 & 5.105 + & @xmath155 & 4.5292 & 0.836 & 4.100 & 4.6000 & 0.832 & 4.100 & 4.842 & 0.814 & 4.100 + & @xmath156 & 0.0001 & 1.000 & 21.958 & 0.0008 & 1.000 & 22.285 & 0.0107 & 1.000 & 25.185 + & @xmath157 & 6.9764 & 1.000 & 21.359 & 7.2560 & 1.000 & 21.660 & 11.0629 & 1.000 & 24.297 + & @xmath158 & 7.0077 & 1.000 & 21.372 & 7.2882 & 1.000 & 21.673 & 11.1064 & 1.000 & 24.311 + & @xmath159 & 0.4119 & 1.000 & 26.152 & 0.4246 & 1.000 & 26.541 & 0.5507 & 1.000 & 29.995 + [ tab : sine_extrap2 ] next , we impose a stepped gain fluctuation @xmath172 to the data generated and wanted to evaluate the behavior of the calibration methods .
the results for the stepped case are given in tables [ tab : stepped_extrap1 ] and [ tab : stepped_extrap2 ] .
we see by the @xmath186 values in both tables that the dynamic methods perform better than most static methods .
if the calibrated values by chance drift outside of the reference range the dynamic methods will do a good job at capturing it with certainty while having a narrower credible interval than confidence intervals of the static methods .
the dynamic approaches outperform all of the static method in terms of @xmath196 .
these results of the simulation study do not change much across the number of references used . once again , when the relationship is assumed to be linear there is no benefit to adding more references .
+ lastly , the study is conducted with a sinusoidal gain fluctuation while extrapolating to estimate @xmath115 .
the results for the sinusoidal case are given in tables [ tab : sine_extrap1 ] and [ tab : sine_extrap2 ] .
the dynamic methods @xmath154 and @xmath155 exhibit the same behavior as before in tables [ tab : sine_inter1 ] and [ tab : sine_inter2 ] with @xmath186 values ranging for 4.4 to 4.8 .
even though the average mean square errors are larger than those of the static methods when using a 2-reference model , the two dynamic methods outperform the static methods @xmath157 and @xmath158 which are based on the inverse approach .
the dynamic models have average coverage probabilities smaller than the static model across all of the signal - to - noise ratios
. we can not fail to point out that once again the @xmath196 are 4 to 6 times shorter than the average widths for the static models .
in this example , we apply the dynamic calibration approaches to the calibration of a microwave radiometer for an earth observing satellite .
engineers and scientist commonly use microwave radiometers to measure the electromagnetic radiation emitted by some source or a particular surface such as ice or land surface .
radiometers are very sensitive instruments that are capable of measuring extremely low levels of radiation .
the transmission source of the radiant power is the target of the radiometers antenna . when the region of interest , such as terrain , is observed by a microwave radiometer , the radiation received by the antenna is partly due to self - emission by the area of interest and partly due to the reflected radiation originating from the surroundings ( ulaby _ et al .
_ 1981 ) such as cosmic background radiation , ocean surface , or a heated surface used for the purpose of calibration . + a basic diagram of a radiometer is shown in figure [ fig : radiometer ] where the radiant power with equivalent brightness temperature ( i.e. the term brightness temperature represents the intensity of the radiation emitted by the scene under observation )
@xmath213 enters the radiometer receiver and is converted to the output signal @xmath214 .
the schematic features the common components of most microwave radiometers .
as the radiometer captures a signal ( i.e. brightness temperature @xmath213 ) , it couples the signal into a transmission line which then carries the signal to and from the various elements of the circuit . in figure [ fig : radiometer ] , a signal @xmath213 is introduced directly into the antenna , then it is mixed , amplified and filtered to produce the output signal @xmath214 .
this filtering and amplification of the signal is carried out through the following components of the radiometer : an amplifier @xmath215 ; pre - detection filter @xmath216 ; a square law detector @xmath217 ; and a post - detection filter @xmath218 .
the output of the radiometer is denoted as @xmath214 .
see ulaby _
et al . _ ( 1981 ) for a detailed discussion .
+ racette and lang ( 2005 ) state that at the core of every radiometer measurement is a calibrated receiver .
calibration is required due to the fact that the current electronic hardware is unable to maintain a stable input / output relationship . for space observing instruments , stable calibration without any drifts
is a key to detect proper trends of climate ( imaoka _ et al .
_ 2010 ) . due to problems such as amplifier gain instability and exterior temperature variations of critical components that may cause this relationship to drift over time ( bremer 1979 ) . during the calibration process
, the radiometer receiver measures the voltage output power @xmath214 , and its corresponding input temperature of a known reference .
two or more known reference temperatures are needed for calibration of a radiometer .
( 1981 ) ; racette and lang ( 2005 ) state that the relationship between the output , @xmath214 and the input , @xmath213 is approximately linear , and can be expressed as @xmath219 where , @xmath220 is the estimated value of the brightness temperature , @xmath214 is the observed output voltage . using this relationship
, the output value , @xmath214 , is used to derive an estimate for the input , @xmath213 ( racette and lang , 2005 ) .
+ traditional calibration methods use measurements taken from known calibration references , for example see figure [ fig : refdata ] . due to possible cost constraints
it is common to use between two and five references .
the reference temperatures are converted to their equivalent power measurement prior to the calibration algorithm .
the radiometer outputs are observed when the radiometer measures the reference temperatures , giving an ordered calibration pair @xmath221 .
the @xmath222 values are observed from the process of the electronics within the radiometer ( see figure [ fig : radiometer ] ) ( ulaby _ et al . _ 1981 ; racette and lang 2005 ) . through the process of calibration
, the unknown brightness temperature @xmath223 is estimated by plugging its observed output @xmath224 into either equation ( [ eq : classeq ] ) or equation ( [ eq : inveq ] ) .
+ it is of interest to develop a calibration approach that can detect gain abnormalities , and/or correct for slow drifts that affect the quality of the instrument measurements . to demonstrate the dynamic approach in terms of application appeal ,
the two dynamic methods were used to characterize a calibration target over time for a microwave radiometer .
the data used for this example was collected during a calibration experiment that was conducted on the millimeter - wave imaging radiometer ( mir ) ( racette et al .
the purpose of the experiment was to validate predictions of radiometer calibration .
+ [ ] time series of mir output voltage measurement data @xmath225 , @xmath226 , and @xmath227 .
[ fig : ex_volt ] the mir was built with two internal blackbody references which will be used to observe a third stable temperature reference for an extended period of time .
the third reference was a custom designed cryogenically cooled reference .
racette ( 2005 ) conducted the mir experiment under two scenarios : the first experiment denoted as @xmath228 examined the calibration predictions when the unknown target is interior ( i.e. interpolation ) to the reference measurements ; the second set of measurements ( denoted as @xmath229 ) where taken when the unknown temperature to be estimated is outside ( i.e. extrapolation ) of the range of calibration references .
+ for demonstration purposes we will only consider the @xmath229 experiment , for details of the @xmath228 experiment see racette ( 2005 ) .
for the @xmath229 run of the experiment , the reference temperatures are as follows : 1 . @xmath230 2 .
@xmath231 with the unknown target temperature that must be estimated denoted as @xmath232 .
each temperature measure has a corresponding observed time series of output measurements ; @xmath225 , @xmath226 , and @xmath227 ( see figure [ fig : ex_volt ] ) .
therefore in this example we only consider a 2-point calibration set - up as we use @xmath233 and @xmath234 as the known reference standards and use @xmath227 to derive estimates of @xmath232 for the first 1000 time periods .
+ the results of the dynamic approaches : @xmath154 and @xmath155 , will be compared to the inverse " calibration method ( krutchkoff 1967 ) implemented by racette ( 2005 ) .
the method considered by racette ( 2005 ) will be denoted as @xmath235 . as in practice , rarely does one know the value of the true temperature to be estimated so the aim of this example is to assess the contribution of the calibration approach to the variability in the measurement estimate .
racette ( 2005 ) analysis did not consider biases that may exist in calibration , continuing in the same spirit , the existence of biases will not be considered in the analysis .
we will apply the @xmath235 , @xmath154 , and @xmath155 approaches to the data to estimate the temperature @xmath232 ; the standard deviation of the estimated time series @xmath236 is used as a measure of uncertainty including the contribution of the calibration algorithm .
+ [ ] time series of calibrated temperature for mir @xmath229 experiment .
black lines indicate the results using the inverse " calibration approach @xmath235 ; the green lines are the results using the dynamic approach @xmath154 with the 95% credible intervals in purple .
[ fig : ex_mod1 ] [ ] time series of calibrated temperature for mir @xmath229 experiment .
black lines indicate the results using the inverse " calibration approach @xmath237 ; the red lines are the results using the dynamic approach @xmath155 with the 95% credible intervals in purple .
[ fig : ex_mod2 ] figures [ fig : ex_mod1 ] shows the time series of the temperature estimates for @xmath232 using krutchkoff s ( 1967 )
inverse " approach @xmath235 and the dynamic approach @xmath154 .
the standard deviations for the @xmath235 and @xmath154 approaches are @xmath238 and @xmath239 , respectfully .
we see the dynamic model @xmath154 improves the estimation process over the static model @xmath235 by observing the corresponding standard deviation values .
the dynamic model decreased the measurement uncertainty by roughly 33% . in figure
[ fig : ex_mod2 ] the time series of the temperature estimates for @xmath232 using the inverse " approach @xmath235 and the dynamic approach @xmath155 is given .
the standard deviations for the @xmath235 and @xmath155 approaches are @xmath238 and @xmath240 . again
, the dynamic approach outperforms the static model @xmath235 . in this case ,
dynamic model @xmath155 decreased the measurement uncertainty by roughly 34% .
two new novel approaches to the statistical calibration problem have been presented in this paper . in
was shown by the simulation results that the use of the dynamic approach has its benefits over the static methods .
if the linear relationship in the first stage of calibration is known to be stable then the traditional methods should be used .
the dynamic methods showed promise in the cases when the signal - to - noise ratio was high .
there is also a computation expense to implementing the dynamic methods compared to the static methods , but in the sense of electronics these methods allow for near real time calibration and monitoring .
+ it is worth noting that the dynamic method shows possible deficiencies when the gain fluctuations is sinusoidal , referring to results in table [ tab : sine_inter1 ] . in figure
[ fig : sine1 ] it is evident the largest source of the error is in the beginning of estimation process , roughly from @xmath241 to @xmath242 .
the mse values for the dynamic approaches ; @xmath154 and @xmath155 were 4.41 and 4.40 , respectively , which was vastly different than those reported for the static methods .
this problem can be addressed by extending the burn - in period .
+ we increased the burn - in period to 200 which allowed the algorithm more time to learn and hence results in a lower mse value . in figure
[ tab : sine2 ] we see that the estimates fit better to the true values of @xmath115 .
the mse decreased from 4.41 to 0.64 for @xmath154 and 0.63 for @xmath155 .
the increased burn - in period improves the coverage probability but the interval width is nt noticeably affected .
the coverage probability increased from 0.628 to 0.722 for @xmath154 and from 0.829 to 0.964 for @xmath155 .
+ [ comparison of calibration approaches when interpolating to estimate @xmath115 with sinusoidal gain fluctuations ] comparison of calibration approaches @xmath154 and @xmath155 when interpolating to estimate @xmath115 with sinusoidal gain fluctuation . for completeness
we consider the behavior of the method if @xmath72 crosses zero .
it is absurd to believe that this would happen in practice because one would test the significance of @xmath72 ( myers 1990 ; montgomery _ et al .
_ 2012 ) for using any method where the possibility of dividing by zero could occur .
we demonstrate this by generating data where @xmath244 for all time and @xmath72 drifts from 1 to -1 over time where @xmath245 ( see figure [ fig : beta_cross ] ) .
figure [ fig : beta_cross_x0 ] shows the dynamic method is close to the true values of @xmath115 until @xmath72 get close to 0 . within the region where the slope crosses the @xmath246
the posterior estimates become _
unstable_. here we define unstable as meaning that we are within a region where there is division by zero .
this instability is only present when @xmath247 , for every @xmath248 . as long as @xmath249 the dynamic method
will perform well when estimating @xmath115 .
+ some calibration problems are not linear or approximately linearly related in @xmath115 and @xmath250 .
future work is to investigate the dynamic calibration methods in the presence of nonlinearity . in such settings
we may not have the ability to use only 2-points as references .
any approach will require more references in order to accurately capture the nonlinear behavior .
another area to be explored is using semiparametric regression which also allow for parameter variation across time and could be implemented in a near real time setting .
imaoka , k. , kachi , m. , kasahara , m. , ito , n. , nakagawa , k. , and oki , t. ( 2010 ) .
instrument performance and calibration of amsr - e and amsr2 . _
international archives of the photogrammetry , remote sensing and special information science_. * 38*(part 8) .
racette , p. , adler , r. f. , wang , j. r. , gasiewski , a. j. , and zacharias , d. s. ( 1996 ) . an airborne millimeter - wave imaging radiometer for cloud , precipitation , and atmospheric water vapor studies .
_ journal of atmospheric and oceanic technology _ , * 13*(3 ) , 610 - 619 . | the problem of statistical calibration of a measuring instrument can be framed both in a statistical context as well as in an engineering context . in the first , the problem is dealt with by distinguishing between the classical " approach and the inverse " regression approach
. both of these models are static models and are used to estimate
exact " measurements from measurements that are affected by error . in the engineering context , the variables of interest are considered to be taken at the time at which you observe it .
the bayesian time series analysis method of dynamic linear models ( dlm ) can be used to monitor the evolution of the measures , thus introducing an _ dynamic _ approach to statistical calibration .
the research presented employs the use of bayesian methodology to perform statistical calibration .
the dlm s framework is used to capture the time - varying parameters that maybe changing or drifting over time .
two separate dlm based models are presented in this paper . a simulation study is conducted where the two models are compared to some well known static calibration approaches in the literature from both the frequentist and bayesian perspectives .
the focus of the study is to understand how well the _ dynamic statistical calibration _
methods performs under various signal - to - noise ratios , @xmath0 .
the posterior distributions of the estimated calibration points as well as the @xmath1 coverage intervals are compared by statistical summaries .
these dynamic methods are applied to a microwave radiometry dataset . |
ever since the discovery of high - temperature superconductivity in cuprates,@xcite understanding the motion of a hole in an otherwise half - filled cuo@xmath0 layer has been a major challenge in condensed matter physics.@xcite if , as most customary , one models this system with a one - band hubbard model,@xcite then in the limit of a large on - site repulsion @xmath1 the half - filled case maps onto an antiferromagnetic ( afm ) heisenberg hamiltonian with spin - exchange coupling @xmath2 , where @xmath3 is the nearest - neighbor ( nn ) hopping.@xcite in this limit , then , the problem reduces to understanding the motion of a hole in a two - dimensional ( 2d ) afm background.@xcite a major reason for the difficulty in dealing with this question is that despite being long - range ordered at @xmath4 , the undoped afm background has a very complicated wavefunction due to spin fluctuations , which make it very unlike the semi - classical nel state .
a simple but more realistic description that could be used for analytical purposes is missing ; as a result , progress has been made primarily through numerical simulations.@xcite here , we present an essentially exact numerical solution , and an approximate but quite accurate analytical approach , for the much simpler issue of a hole moving in an ising afm background,@xcite whose undoped wavefunction is the nel state . the solutions have a variational interpretation , and are shown to be accurate for a wide range of parameters , of up to @xmath5 .
the generally accepted view , which probably explains why this problem has not been solved so far ( to the best of our knowledge ) , is that a hole can not really move in an ising afm , since as it hops away from its initial location it re - shuffles the spins at the sites it visits , creating a string of `` defects '' ( misaligned spins ) . in dimensions
higher than one , the energy cost of this string increases roughly linearly with its length , and as a result the hole is forced to stay in the vicinity of its original position , i.e. its effective mass is infinite.@xcite in this view , spin fluctuations which effectively remove ( or `` heal '' ) pieces of this string of defects are needed to free the hole and allow it to acquire a finite effective mass . of course
, such spin fluctuations are absent in an ising afm .
that the hole is not truly localized even in an ising afm has been known ever since trugman pointed out@xcite that by going twice around a closed loop , the hole can actually acquire a finite effective mass .
we will return to this in more detail below , but the main idea is that while the first circuit along the closed loop creates the usual chain of defects , the second reshuffle of the spins during the second circuit removes all these defects , but also ends with the hole at a different location than the original site . by repeating such processes , the hole can therefore move anywhere on its original sublattice .
even though trugman started from a nel background and identified this mechanism for generating a finite quasiparticle mass in the absence of spin fluctuations , he included spin - fluctuations in his calculation by considering the full heisenberg afm hamiltonian when determine the effective mass of the hole.@xcite the issue of the hole mass in a purely ising afm thus remained unanswered .
here we carry out this calculation , and show that the hole is fairly mobile if @xmath6 . moreover , if a transverse magnetic field is turned on , due to aharonov - bohm interference of the peierls phases associated with these closed loops , the effective hoppings acquire a magnetic field dependence over and above the usual peierls phases , with interesting consequences .
our results reveal that even this seemingly simple problem is actually very interesting and leads to rather non - trivial results .
the paper is organized as follows . in sec .
ii , we specify the model and our notation . in sec .
iii , we describe the analytical calculation in the smallest variational subspace where the hole acquires mass , both with and without magnetic field .
we then explain the generalization for the numerical calculation .
section iv contains our results , and a summary and final conclusions appear in sec .
consider a spin-@xmath7 ising afm on a square lattice @xmath8 = \bar{\jmath } \sum_{\langle i , j\rangle}^ { } \left(\sigma_{i , z } \sigma_{j , z } -1\right)\,,\ ] ] where @xmath9 is the pauli matrix , and @xmath10 .
the ground state is the classical nel state , with all spins on sublattice @xmath11 up , and all spins on sublattice @xmath12 down , @xmath13 where @xmath14 annihilates an electron at site @xmath15 , and for convenience we shifted the energy so that @xmath16{c } \sigma_i^{- } , \mbox { if } i\in a \\ \sigma_i^{+ } , \mbox { if } i\in b \\ \end{array } \right.\,,\ ] ] where @xmath17 are the raising / lowering pauli matrices , and hole creation operators @xmath18{c } c_{i,\uparrow } , \mbox { if } i\in a
\\ c_{i,\downarrow } , \mbox { if } i\in b \\ \end{array } \right.\ ] ] for a hole at the site @xmath15 . with this notation , for example
@xmath19 means that the hole is at site @xmath15 and the spin at site @xmath20 is flipped ; all other sites have their spins in the proper nel configuration .
consider now the motion of the hole , with the no - double occupancy condition enforced .
if the hole is at site @xmath15 , the only possibility is for an electron from one of its four nn sites @xmath21 to hop into @xmath15 , thus moving the hole to site @xmath21 .
if the spin of the electron at @xmath21 was properly oriented , when it moves to @xmath15 it has the wrong orientation ; in other words , a `` defect '' is created at @xmath15 when the hole hops from @xmath22 . on the other hand , if there was a `` defect '' at @xmath21 to begin with , when the electron moves to @xmath15 it will be properly oriented , and therefore the `` defect '' at @xmath21 is removed as the hole hops from @xmath22 .
thus , the hamiltonian that describes the dynamics of the hole in the 2d ising afm is @xmath23 p + { \cal h}_{\rm afm}.\ ] ] the projector @xmath24 enforces the no - double occupancy constraint , as well as the condition that each site has either the hole , or a spin : @xmath25 . in the presence of a uniform transverse magnetic field @xmath12 ,
the hopping integrals @xmath26 include the proper peierls factors ( see below ) .
this hamiltonian is similar to the edwards model,@xcite however , unlike that model it enforces ( i ) the fact that there can be no magnons at the site where the hole is , i.e. states like @xmath27 are forbidden ; ( ii ) the fact that there can be at most one spin - flip per site , i.e. states like @xmath28 are forbidden . finally , ( iii ) unlike the edwards model , where the energy of the defects was assumed to be described by @xmath29 , here we use @xmath30 to calculate the true cost for creating spin flips two neighboring defects
cost less energy than two farther - apart ones , because they only disrupt seven afm bonds , not eight .
such corrections are less important than is imposing ( i ) and ( ii ) , but keeping track of the proper exchange energies is simple enough and we do so .
note that a description of a heisenberg afm would require the addition of terms @xmath31 , describing the xy spin - exchange interaction leading to spin fluctuations . however , since the nel ground - state @xmath32 is not a good approximation for the undoped ground state of this model , it is hard to quantify the meaning of adding these terms .
spin fluctuations could also be introduced with a small magnetic field parallel to the @xmath33-axis , leading to a term @xmath34 , as considered in the edwards model.@xcite as shown there , such terms do significantly lower the effective mass of the hole , but we ignore them here .
we begin by considering the case where no transverse magnetic field is applied . in this case
, the unit cell contains two sites ( one from sublattice @xmath11 , one from sublattice @xmath12 ) and the corresponding magnetic brillouin zone is a square rotated by @xmath36 ( see below ) . for each @xmath37 in this brillouin zone , we define the plane wave @xmath38 and want to calculate the green s function @xmath39 where @xmath40 is the resolvent associated with this hamiltonian , @xmath41 is infinitesimally small , and we set @xmath42 .
@xmath43 denotes the number of unit cells , or the number of sites in each sublattice , and is taken to infinity .
besides producing the single - hole spectrum from its poles , the spectral function @xmath44 is the quantity that would be measured by spin - polarized arpes , assuming that a ( photo)electron with spin - up is removed from the system .
of course , there is a second set of wavefunctions consisting of plane waves involving the sites on sublattice @xmath12 ; their associated green s function would be related to arpes measured when a spin - down electron was ejected from the system .
obviously , the spectral weights are identical for the two cases .
as discussed , as the hole hops away from its initial site , it creates a string of defects as it reshuffles the spins on its path .
these defects increase the energy of the state roughly linearly with the length of the string , because of the broken afm bonds . in the absence of spin fluctuations , there are only two ways to remove this costly string of misaligned spins .
one is for the hole to retrace its path , removing all the defects in this case it ends up at the original site it started from.@xcite such terms can only renormalize the overall energy , but do not generate dispersion .
the second possibility is for the hole to go repeatedly around closed loops : on the first circuit a string of defects is created , while on the second circuit the defects are removed .
these loops generate effective second and third nn hopping terms , as counted in the original square lattice . in fig .
[ fig1 ] we show examples of the shortest sequences of events that result in generating such 2nd ( a ) and 3rd ( b ) nn hoppings.@xcite one can easily convince oneself that longer closed loops can only contribute to one of these two effective hoppings , since the final location of the hole is within two hops of the original one . effective nn hopping is impossible , because of spin conservation : if the hole is on the other sublattice , there must be an odd number of magnons around . ] in perturbational terms , for @xmath45 , the sequence of hoppings depicted in fig .
[ fig1](a ) results in 2nd nn hopping @xmath46 , since the initial and final states , which are both in the no - defect manifold , are connected through 6 hopping processes , and the five intermediary states have energies higher by @xmath47 respectively .
similarly , fig .
[ fig1](b ) gives a contribution to 3rd nn hopping @xmath48 , as the longer strings of defects are more costly .
note that clockwise hopping of the hole around the same loop as shown in fig .
[ fig1](b ) would generate a higher - order contribution to the 2nd nn hopping @xmath49 because some of the intermediary states have slightly different energies .
this allows us to estimate the range where perturbation theory is valid . if we ask that @xmath50 , so that this correction from 5-defect processes to the 3-defect result is small , we find @xmath51 . in other words , because of the relatively long and thus energetically costly loops of flipped spins involved in generating the effective hoppings , perturbation theory extends up to rather large values @xmath52 . for comparison , in cuprates
it is believed that @xmath53 , so they would be just outside the perturbational range ( for these values , @xmath54 ) .
of course , the effects of spin fluctuations can not be ignored in the cuprates .
clearly , then , for @xmath55 the dominant contribution to the hole s dynamics comes from the processes of fig .
[ fig1](a ) .
we therefore begin by calculating @xmath56 in a variational approach , by only allowing such short loops ( with up to 3 defects on 3 corners on a square plaquette ) to be generated .
longer loops are more expensive and the probability to generate them should be lower . of course
, they can be included in a numerical calculation , but this will lead to only quantitative , not qualitative changes ( see below ) .
the method we use is similar to that used in refs . for the 1d , and in ref .
for the 2d edwards model . within this variational approximation
, we generate the equations of motion for the propagator by using repeatedly the dyson identity @xmath57 where @xmath58 is the resolvent for @xmath59 , while @xmath60 is the first term in eq .
( [ m4 ] ) , describing the hopping of the hole .
we then find @xmath61\,,\ ] ] where @xmath62 , @xmath63 is the cost of having only the hole in the system ( first state in fig . [ fig1](a ) ) , and @xmath64 points to any of the four nn sites , i.e. @xmath65 for a lattice constant @xmath66 .
the new propagators are @xmath67 and are related to the amplitude of probability to propagate between a no - defect state and a 1-defect state such as the second state shown in fig .
[ fig1](a ) , with a defect at the original site . in the following ,
for simplicity we will denote @xmath68 as @xmath69 .
an equation of motion for @xmath70 can now be generated .
the hopping @xmath60 can lead to three different outcomes : either the hole hops back to site @xmath15 , removing the defect ; or it hops by another @xmath64 , creating a linear string of the type @xmath71 ; or it hops by one of the two @xmath72 , leading to a state such as the 3rd state shown in fig . [
fig1](a ) .
only the first and last outcomes are allowed within our variational calculation , leading to @xmath73\,,\ ] ] where @xmath74 describes generalized green s functions associated with 2-defect states .
its equation of motion , within our variational space , is @xmath75\,,\ ] ] where @xmath76 describes states with a string of three consecutive defects .
only states such as shown in the middle panel of fig .
[ fig1](a ) are in our variational space , hence the unique @xmath77 term in the equation of @xmath78 . finally , the equation for @xmath77 is now connected to two possible 2-defect states .
the hole could either hop back , linking to @xmath79 , or it could complete the closed loop by hopping to site @xmath15 and removing the defect that was there , resulting in @xmath80 eq .
( [ m11x ] ) follows after a translation by @xmath81 , which maintains the site @xmath15 on the same original sublattice @xmath11 , since @xmath82 . of course , hopping also links @xmath77 to two generalized propagators @xmath83 with 4 defects , however those are ignored in this 3-defect variational space .
thus , we find : @xmath84\,.\ ] ] in other words we have a closed system of linear equations for @xmath85 and the various allowed @xmath86 functions .
this system can be solved analytically .
the final result is @xmath87}\,,\ ] ] where @xmath88 @xmath89 @xmath90 the functions that appear in these definitions are : @xmath91 @xmath92 @xmath93 and @xmath94 the green s function of eq .
( [ m13 ] ) describes a free particle on the @xmath11 sublattice , with an onsite energy @xmath95 , and hoppings @xmath96 and @xmath97 that correspond to nn and 2nd nn for sites on this sublattice .
interestingly , all these effective quantities are functions of @xmath98 , in other words retardation effects are explicitly taken into consideration in this variational calculation .
for @xmath99 we find that indeed @xmath100 , while @xmath101 .
this is precisely twice the value we estimated for fig .
[ fig1](a ) ; the factor of 2 is due to contributions from both clockwise and anticlockwise loops . process generated with only 3-defect strings .
[ fig2 ] ] the appearance of the @xmath97 term at this level of the calculation is surprising at first , since configurations with 5 defects like that appearing in fig .
[ fig1](b ) are not allowed in this variational space .
the answer for how to achieve @xmath97 hopping with only 3-defect loops is shown in fig .
[ fig2 ] : essentially , instead of removing the last defect at the 6th hopping , another 3-defect process is initiated .
it is straightforward to check that in the perturbational limit , the ratio between the expected values for this @xmath97 and the @xmath96 of fig .
[ fig1](a ) is of @xmath102 .
this is precisely the ground - state value of @xmath103 in this asymptotic limit , which further validates this statement .
it is also now clear that this strategy of starting a new loop when only a single defect is left can be repeated an arbitrary number of times , and an infinite sequence of higher order contributions to @xmath96 and @xmath97 can be generated this way .
this explains the @xmath104 denominators in eqs .
( [ m14]),([m14b ] ) , showing that our variational calculation sums contributions from these 3-defect processes to all orders . as argued previously ,
the contribution of 5-defect loops is negligible if @xmath105 .
interestingly , for such values @xmath106 varies from 0 to just above 1 ( however , this ignores the change in @xmath107 because the ground - state energy moves away from @xmath108 , as @xmath109 increases ) . in other words , the denominators in eqs .
( [ m14 ] ) , ( [ m14b ] ) could become very small , implying that significant hopping , and thus light effective masses , may be possible in such systems even for rather small @xmath109 ratios . as a final comment , the fact that @xmath37-dependence appears only in the equation for @xmath77 confirms that the effective hopping terms are due to completely circling around the closed loops . if we truncated the variational space to include only 2-defect states , which would mean setting @xmath110 in eq .
( [ m10 ] ) , the result would be just a renormalization of the on - site energy but no dispersion .
since the dispersion of the hole is due to trugman loops , it is interesting to consider what happens when a transverse magnetic field @xmath12 is applied . because of the large number of elementary hopping processes involved in generating effective longer ranged hoppings , one would expect the phases associated with these effective hoppings to be different from the normal peierls phase
. put it another way , one would expect aharonov - bohm - like interference between clockwise and counterclockwise contributions , over and above the usual peierls phases , and therefore a spectrum that is not simply that of a bare particle with hoppings @xmath112 placed in a magnetic field .
let @xmath113 be the magnetic flux through the elementary square plaquette . because the dressed hole only lives on one sublattice ( sublattice @xmath11 in our calculation ) , the actual relevant flux is @xmath114 , since the unit cell for the sublattice is doubled in size . from elementary considerations of the hofstadter butterfly,@xcite we know that we need to consider the cases when the ratio between the magnetic flux and the elementary flux @xmath115 is @xmath116 , where @xmath117 and @xmath118 are mutually prime integers . in this case
, we expect the spectrum to split into @xmath118 magnetic subbands .
of course , then , the brillouin zone is folded down @xmath118 times .
we are therefore interested in the spectrum of the dressed hole when @xmath119 ( for simplicity of notation , we also introduce @xmath120 ) .
we use the landau gauge @xmath121 , in which case only the hopping integrals in the @xmath122-direction acquire peierls phases .
the magnetic unit cell will contain @xmath123 sublattice @xmath11 sites , and it is drawn , together with the peierls phases , in fig .
each site in the unit cell is labeled by the index @xmath124 which establishes its location in the unit cell , as well as the phase associated with hopping off this site in the positive @xmath122-direction . , where @xmath117 and @xmath118 are mutually prime integers .
the unit cell is marked by the red dashed line , and contains @xmath123 sites , labeled by squares , on the @xmath11 sublattice .
these are indexed by their corresponding @xmath125 .
the value of @xmath126 also appears in the peierls phases for hopping in the direction of the arrow .
[ fig3 ] ] the hamiltonian is identical to that of eq .
( [ m1 ] ) , with the proper phases included in the hoppings .
of course , we assume that the magnetic field is not sufficiently large to overcome the afm coupling and favor a fm ground state . also , we ignore the zeeman term since the spins are reshuffled as the hole hops around , the total @xmath127-axis spin @xmath128 remains constant and therefore the zeeman contribution is an overall constant .
we now need to introduce @xmath123 distinct plane waves , associated with each site of the unit cell @xmath129 where @xmath130 collects all sites on sublattice @xmath11 with the same value of @xmath126 .
the procedure to generate the equations of motion for the propagator @xmath131 is just as before , however now one has to take into consideration the peierls phases .
these can either be included in the definition of the new @xmath132 functions , or can be explicitly pulled out . in either case , the overall structure of the system of linear equations is very similar to that in the @xmath35 case , and can be solved similarly .
this allows us to reduce it to a single equation involving only @xmath85 functions , which reads : @xmath133 \right .
- \tilde{t}_3 \left [ e^{2ik_x } g_{\alpha ' , \alpha-2}({\mathbf{k}},\omega ) + 2 \cos(2k_y-2\alpha\phi ) g_{\alpha'\alpha}({\mathbf{k}},\omega ) + e^{-2ik_x } g_{\alpha',\alpha+2}({\mathbf{k}},\omega ) \right]\right\}\,.\end{gathered}\ ] ] here we used the short - hand notations : @xmath134 @xmath135}{1- 4 \beta^2(\omega)+ 4 \beta^4(\omega ) \sin^2(2\phi)}\,,\ ] ] and @xmath136 remarkably , eq . ( [ n3 ] ) is identical to the equation of motion one would obtain for the propagator associated with a hamiltonian on the sublattice @xmath11 only , with an on - site energy @xmath137 and nn and 2nd nn hopping ( as defined on the sublattice ) given by @xmath138 , plus the proper peierls phases for the applied magnetic field .
the terms in the parenthesis multiplied by @xmath139 come from nn hopping by @xmath140 , with the proper peierls phases accounted for in the phases of the cosine functions ; similarly , the terms in the parenthesis multiplied by @xmath141 come from next - nn hopping by @xmath142 .
only the @xmath143 hopping accumulates a peierls phase , in this case , hence the cosine term appears only for that term .
it follows that @xmath144 and @xmath145 are the effective parameters describing the new quasiparticle ( the dressed hole in the nel afm background plus magnetic field ) .
they depend strongly on the magnetic field , besides the @xmath98-dependence due to retardation effects .
as expected , they have the correct values when @xmath146 , as given by eqs .
( [ m15])-([m14b ] ) .
the @xmath147 and @xmath148 terms in @xmath139 , respectively @xmath149 term in @xmath141 , are due to aharonov - bohm interference between clockwise and counterclockwise loops .
this can be checked directly in the asymptotic limit @xmath150 by counting the accumulated peierls phases for the processes from figs .
[ fig1](a ) and [ fig2 ] .
because of these , we expect the resulting hofstadter butterfly to have an unusual periodicity . that is
, we still expect the appearance of @xmath118 bands when @xmath151 , due to interference between the peierls phases .
however , additional dependence of the hoppings on @xmath152 because of the aharonov - bohm interference will further modulate the overall bandwidth with a period @xmath153 in @xmath152 ( the band structure is invariant if only @xmath96 changes its sign ) .
in contrast , normally , ( i.e. for constant @xmath154 ) the spectrum is an even function of @xmath152 with period @xmath155 , because if the magnetic flux through the unit cell ( @xmath156 , in this case ) , increases by @xmath157 , the spectrum stays unchanged .
thus , we expect a quite different - looking hofstadter butterfly for the dressed hole .
in particular , since at least for small @xmath109 we expect @xmath158 , the spectrum should become extremely narrow every time @xmath159 because of destructive aharonov - bohm interference .
the numerical calculation that we describe in this section has been carried out only for the case @xmath35 .
this is a straightforward generalization of the 3-defect variational calculation discussed above .
the idea is to systematically increase the size of the variational space , to see whether addition of configurations with longer loops of defects changes results significantly .
if it does not , then we know that the calculation is converged and therefore the results are essentially exact .
we use an index @xmath160 to characterize the maximum number of defects allowed within the variational subspace .
the equations of motion are generated as before , starting from just the hole and keeping all configurations linked to by hole hopping , up to those involving @xmath160 defects .
the exception is for longer chains with @xmath161 defects , where we do not allow the chain to `` self - cross '' itself .
for instance , for the 6-defect chain sketched in fig .
[ fig4 ] , the hole is not allowed to hop either left or down .
either of these processes would result in a configuration with the hole having two strings of defects attached to it .
trying to remove all these defects involves a very complicated sequence of trugman loops , and is therefore expected to have a very small contribution to @xmath96 and @xmath97 .
the other role of such configurations is to renormalize energies , but as we will see in the next section , we can argue that their effect must be quite negligible . ]
this is why for a configuration like that of fig .
[ fig4 ] , we allow the hole only to hop back ( towards right ) to the @xmath162 configuration from which it derived ; or to hop up to generate a @xmath163 configuration . in its turn , hopping to the left from this @xmath163 configuration will close the loop and give a new contribution to the effective @xmath96 . new possible closed loops ( and therefore additional contributions to @xmath164 ) appear for @xmath165 once all the allowed configurations have been generated for a given @xmath160 , the resulting ( very sparse ) linear system comprising their equations of motion is solved numerically using pardiso.@xcite we show results with up to @xmath166 , in the following , because here there are already over ten thousand allowed configurations , and , as we will see , this is sufficiently large for convergence to be achieved up to a quite high @xmath109 ratio . finally , note that in the numerical calculation , @xmath167 contains many more configurations than in the @xmath167 analytical calculation , because now all possible 2- and 3-defect configurations ( not just those like in fig . [ fig1](a ) ) are included .
the results of the two methods , therefore , should not be expected to be identical .
, as predicted by the analytic solution ( circles ) and numerically in the variational spaces with @xmath168 ; ( b ) and ( c ) show the dependence on @xmath160 of the gs energy for @xmath169 , respectively 5 . [ fig5 ] ] we begin by looking at the ground - state ( gs ) energy as a function of @xmath109 , using both the analytical and the numerical methods .
results are shown in fig .
[ fig5](a ) for @xmath170 .
as expected , in the limit @xmath171 , all curves converge towards @xmath172 , which is the cost of placing the hole in the lattice ( breaking four afm exchange bonds ) . as @xmath3 increases and the hole acquires a finite mass ,
its energy is lowered . for small @xmath109 ,
all values are in excellent agreement , but for larger @xmath109 they begin to fan out . at @xmath169 , convergence has already been achieved for @xmath166 , as shown in fig .
[ fig5](b ) . for @xmath173
, @xmath166 is not yet fully converged , although further corrections are not expected to be large .
note that we do not show results up to @xmath173 in panel ( a ) because the corresponding increase of the energy range to be displayed makes it even more difficult to distinguish the numerical curves from one another . ,
but for the gs quasiparticle weight instead of gs energy .
[ fig6 ] ] the analytical calculation indeed gives a reasonably accurate value up to @xmath174 .
the fact that the biggest variation is between it and the @xmath167 numerical calculation , means that the 2- and 3-defect configurations ignored in the analytical approximation play a significant role in renormalizing the overall energy .
this is not surprising , since these are some of the least expensive configurations .
however , qualitatively and even quantitatively it is clear that this analytical approximation is quite good up to fairly large @xmath109 values .
the quasiparticle weight @xmath175 defined as the overlap between the lowest eigenstate at a given momentum , @xmath176 and the non - interacting state @xmath177 , is shown at @xmath178 in fig .
it remains quite considerable even for large @xmath109 , showing that a significant part of the wavefunction consists of the hole alone , with no strings of defects .
the effect of increasing the variational space is more visible here , as expected due to normalization : as the wavefunction acquires extra components in a larger variational space , the weight of the hole - only part decreases , even if overall the energy of the state is not much changed .
( upper panel ) and quasiparticle weight @xmath179 ( lower panel ) along the high - symmetry directions of the square lattice brillouin zone , for @xmath169 and @xmath166 .
the inset in the bottom panel shows the upper right quadrant of the full brillouin zone ; the afm brillouin zone is shaded .
lines are guides to the eye .
[ fig7 ] ] the hole band dispersion @xmath180 and quasiparticle weight @xmath179 are shown in fig .
[ fig7 ] , along high - symmetry cuts in the brillouin zone of the original square lattice , for @xmath181 .
the band folding due to the afm order is clearly apparent .
these results are very similar to those found for comparable parameters in the 2d edwards model ( see fig . 10 of ref . ):
the minimum is at @xmath182 , and @xmath183 is a saddle point ( unlike in cuprates , where it is the actual minimum ) .
the quasiparticle weight is fairly constant but with local variations that mimic the shape of the dispersion .
this similarity is not surprising ; even though the edwards model allows more states ( e.g. with more defects at a site ) , those are higher in energy and do not contribute much to the gs .
the mechanism for generating an effective mass is identical in both models , with only small quantitative differences due to the energy assigned to the strings of defects in the two models .
one difference between the two models is illustrated in fig .
[ fig8 ] , where we show the dispersion for @xmath184 and @xmath169 along fewer cuts .
the symbols are data from the numerical simulations , and the lines are fits to an effective dispersion @xmath185,\ ] ] along the @xmath186 line .
using the same parameters along the @xmath187 line is clearly not a good fit .
in contrast , for the edwards model such fits worked well in the entire brillouin zone .
this suggests that the retardation effects may be somewhat stronger in this case : even though we know that only @xmath188 and @xmath189 effective hoppings are generated , these may not be well approximated by constant values @xmath190 if the @xmath98 dependence is considerable within the bandwidth of the dressed hole .
along high - symmetry cuts in the brillouin zone , for @xmath169 and @xmath184 .
symbols are numerical data ( for @xmath166 , same as in fig .
[ fig7 ] ) and lines are results to a fit as in eq .
( [ e1 ] ) , with parameters extracted for best fit on the @xmath191 cut .
dashed lines are fits to extract the effective mass .
[ fig8 ] ] this is less of an issue at smaller @xmath109 , where the bandwidth is significantly narrower . in fact , even for @xmath169 , the results for @xmath167 shown in fig .
[ fig8 ] clearly have a smaller bandwidth than for @xmath192 .
this is not surprising , since longer loops with additional contributions to effective hoppings are included in the latter cases .
even though the decrease in bandwidth is not that large , it is clear that the fit is better for the @xmath167 case .
this may explain why the fit worked well for the 2d edwards model,@xcite where the solution was restricted to @xmath167 and the bandwidths were narrower also due to more costly defects ( equivalent to larger @xmath193 ) than in fig .
[ fig8 ] .
since we can not extract meaningful @xmath190 values , we instead calculate the effective mass @xmath194_{x , y } = \left .
\frac{\partial^2 e({\bf k})}{\partial k_x \partial k_y } \right|_{|{\bf k}|\to 0}\,,\ ] ] with fits as shown by the dashed lines in fig . [ fig8 ] .
as expected , the mass is found to be isotropic .
the results are shown in fig .
[ fig9 ] , on a logarithmic scale .
the effective mass @xmath195 is in units of the band mass @xmath196 , i.e. the bare particle mass if there was no afm background .
we show results up to @xmath173 , even though the values are not fully converged for @xmath197 . at these larger values ,
longer loops than @xmath166 would need to be included . from results such as in fig .
[ fig5](c ) , however , we do not expect those further corrections to be very significant . also , they would decrease @xmath195 , as they would open additional channels for effective hopping , so the values shown in fig .
[ fig9 ] can be taken as an upper bound for @xmath195 . in the limit @xmath198 ,
the effective mass becomes very large .
as discussed , from perturbation theory here we expect @xmath199 .
the corresponding @xmath200 is shown as a dashed line in the inset , in good agreement with the values obtained from the analytical and numerical calculations ( symbols and full lines ) .
the effective mass diverges as @xmath201 since the hole is bound to its original site in this case . as @xmath109 increases , the effective mass decreases significantly , and it reaches @xmath202 for @xmath169 . since this decrease is mirrored by the analytical @xmath167 calculation , it must be due to the summation over repeated 3-defect loops , like those shown in fig .
longer loops further lower the effective mass .
the @xmath203 loops give a significant contribution for @xmath204 , validating the expectations based on perturbation theory .
the contribution of the @xmath166 loops is still quite small , for these values .
we expect @xmath195 to further decrease with increasing @xmath205 .
however , capturing that limit is very difficult if not outright impossible with this variational formulation , given the huge increase in possible configurations with increasing @xmath160 . .
the inset shows the same data , plus the prediction of the weak - coupling perturbation theory ( dashed line ) .
[ fig9 ] ] the main result , thus far , is that a hole in an ising afm is fairly mobile if @xmath3 and @xmath193 are comparable , even though there are no spin fluctuations in this model .
we now proceed to discuss the spectrum of the dressed hole in the presence of a transverse magnetic field @xmath12 .
we will use the @xmath167 analytical eq .
( [ n3 ] ) , and solve it numerically for various values of @xmath207 .
we show results for @xmath169 , even though we know that here the analytical calculation is not sufficient for full convergence , simply because its @xmath35 bandwidth is sufficiently large to make it easier to see the effect of a finite @xmath12 .
we note that the numerical calculation in the variational spaces corresponding to various @xmath160 can be carried out as well , however each previous unknown , corresponding to an allowed configuration of defects , now becomes a @xmath208 matrix of unknowns , corresponding to each of the @xmath123 distinct sites in the magnetic brillouin zone shown in fig .
this huge increase in the size of the linear system to be solved , especially for larger @xmath118 values ( smaller magnetic fields ) makes the implementation of this scheme cumbersome , and unnecessary since we do not expect any qualitative changes . after solving the linear system in eq .
( [ n3 ] ) , we calculate the local density of states ( ldos ) @xmath209 where the sum is over the ( highly folded ) magnetic brillouin zone corresponding to the magnetic unit cell of fig .
this ldos is independent of the site @xmath125 used , hence the lack of a site index .
we first consider the effect of the additional dependence on @xmath12 due to the aharonov - bohm interference , on the hofstadter butterfly expected when the magnetic flux through the unit cell is comparable to @xmath157 .
results for various simple ratios are shown in fig . [ fig10 ] .
in all cases , the upper panel shows the full results , whereas the lower panel shows the results when the aharonov - bohm interference is turned off , i.e. we use the @xmath210 values @xmath211 , @xmath212 and @xmath213 in eq .
( [ n3 ] ) . for the @xmath35 case shown in ( a ) , the results are of course identical .
the density of states resembles that of a 2d square lattice with nn hopping .
this is expected , since the dominant hopping term is @xmath96 , which plays the role of nn hopping for the sublattice on which the dressed hole lives .
the ldos is slightly distorted due to ( small ) contributions from the @xmath97 term and the retardation effects .
these small distortions and asymmetries are observed in all other panels . for a finite flux through the sublattice unit cell @xmath214 , the band splits into @xmath118 subbands , as expected in standard hofstadter butterfly phenomenology .
the bands are not fully separated because we used a broadening @xmath215 which , while small , is still comparable with some of these features bandwidth . despite this , the various subbands are easily identifiable . in figs .
[ fig10 ] ( b ) and ( d ) , the full result ( top panel ) , while quite similar in aspect to the lower panel , shows a somewhat narrower bandwidth .
this is not surprising , since the aharonov - bohm interference terms like @xmath216 etc .
are responsible for a decrease in the value of the effective hoppings , and therefore of the overall bandwidth .
this also explains the apparent `` collapse '' of the butterfly for @xmath217 and @xmath218 , shown in figs .
[ fig10 ] ( c ) and ( e ) , respectively .
the very narrow peak seen in both cases shows the expected subband structure , if a much smaller @xmath219 values is used , as done in the inset : the former case shows the 3 subbands ( these features are so narrow that even this much smaller @xmath220 can only partially resolve them ) , while the later case shows the one band .
the extreme narrowing is due to the aharonov - bohm interference , which at these values of the flux results in a very small or vanishing @xmath221 , see eq .
( [ n5 ] ) , and a bandwidth set by @xmath222 .
as already noted , this additional @xmath12-dependent modulation of the hofstadter structure , coming from the aharonov - bohm interference , is also responsible for an increase in the periodicity of the butterfly .
the lower panel shows that , as expected in models with constant hopping integrals , the butterfly is periodic if the flux through the unit cell increases by a flux quantum ( here , @xmath223 , or @xmath224 ) .
the full result clearly does not have this periodicity . as discussed previously , based on eqs .
( [ n4])([n6 ] ) we expect the periodicity to be @xmath225 for the full aharonov - bohm case ; our simulations confirm this ( not shown ) .
we can imagine that this @xmath153 periodicity survives if one allows longer loops in the calculation , but some care is needed .
for example , loops like those shown in fig .
[ fig1](b ) would bring in factors of @xmath226 in their contribution to @xmath96 , and various other aharonov - bohm phases will be associated with longer loops .
we calculated several of these and all are consistent with the @xmath153 periodicity ; so we believe this @xmath153 period is correct to all orders , but do not have a proof .
finally , we discuss the role of the aharonov - bohm interference at very small magnetic fields . in this case
, one expects the spectrum to separate in a sequence of landau levels ( lls ) separated by the cyclotron frequency .
the appearance of lls in the spectrum leads to quantum oscillations in various transport measurements , such as de haas - van alphen oscillations .
ldos ; ( b ) the location of the lowest ll peak , in the full calculation ( squares ) and with the aharonov - bohm interference turned off ( circles ) .
the dashed line is the analytical prediction of eq .
( [ 27 ] ) .
these results are for @xmath169 and @xmath227 . [ fig11 ] ] the ldos at the lower edge , with ( full lines ) and without ( dashed line ) very small magnetic fields applied , is shown in fig .
[ fig11](a ) .
the @xmath35 ldos shows the jump at the band edge , here smoothed out by a finite @xmath220 value , and the roughly constant 2d density of states above it .
the slight monotonic increase is due to the use of a tight - binding dispersion , as opposed to an approximate parabolic one.@xcite when the small field is turned on , we see that this ldos splits into equally spaced narrow peaks , marking the lls .
the cyclotron frequency , defined by the distance between consecutive lls , increases roughly linearly with @xmath12 , and so does the weight in each lls , as expected because of its larger degeneracy . in fig .
[ fig11](b ) , we plot the energy of the lowest ll against the flux through the unit cell .
squares show the full result , while circles are the result when the aharonov - bohm oscillations are turned off .
the dashed line shows the value of the lowest ll for a particle of constant mass @xmath195 , @xmath228/j\,,\ ] ] where the cyclotron frequency is @xmath229 , and @xmath230 and @xmath195 are the @xmath35 values of the dressed hole ground - state energy and effective mass , respectively .
this prediction is in excellent agreement with the results where the aharonov - bohm interference is turned off . in contrast , the full results show additional quadratic @xmath12-dependence .
this is not so surprising , since for small magnetic fields , the effective hoppings depend quadratically on @xmath12 through the aharonov - bohm interference terms , and one would expect that dependence to be mirrored here through the effective mass .
we do not perform a more quantitative analysis because , as already mentioned , the @xmath167 analytical case is not fully converged at this value of @xmath169 ; for example , the effective mass is @xmath231 , whereas the converged result is @xmath232 this is quite a sizable difference .
going to a smaller @xmath109 is not easy either , because the bandwidth narrows ( or , equally , the effective mass increases ) and it becomes more and more difficult to separate various features .
to conclude , we have investigated the motion of a hole in a 2d square ising afm .
we showed that summation of the contribution of all trugman loops up to a given length @xmath160 can be carried out numerically , and convergence is reached for @xmath166 if @xmath233 .
qualitatively correct and quantitatively quite reasonably accurate results are obtained from a simple analytical approximation for @xmath167 .
we find that the effective mass of the hole can be fairly low , of around 30 40 @xmath196 , if @xmath234 3 5 .
if a magnetic field is turned on , aharonov - bohm interference between clockwise and counterclockwise trugman loops leads to dependence of the effective hoppings on the magnetic field over and above the usual peierls phases .
their most spectacular manifestation is in the `` collapse '' of the hofstadter butterfly structure at fields where the aharonov - bohm interference is destructive . of course , creating large enough magnetic fields to see the hofstadter butterfly band structure in a crystal is still a challenge ( moreover
, one would need a large @xmath193 in order to prevent a transition to ferromagnet order , at such large fields ) .
nevertheless , this is an interesting effect that might be mimicked in some other type of systems . for small magnetic fields , where the magnetic length is large as compared to the unit cell , landau levels form .
here , we find that the effect of the aharonov - bohm phases is to bring additional dependence on @xmath12 through @xmath235 , in the cyclotron frequency .
this is a small effect , but one that might be easier to see through various quantum oscillations - type measurements .
25ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) | we show that a proper consideration of the contribution of trugman loops leads to a fairly low effective mass for a hole moving in a square lattice ising antiferromagnet , if the bare hopping and the exchange energy scales are comparable .
this contradicts the general view that because of the absence of spin fluctuations , this effective mass must be extremely large .
moreover , in the presence of a transverse magnetic field , we show that the effective hopping integrals acquire an unusual dependence on the magnetic field , through aharonov - bohm interference , in addition to significant retardation effects .
the effect of the aharonov - bohm interference on the cyclotron frequency ( for small magnetic fields ) and the hofstadter butterfly ( for large magnetic fields ) is analyzed . |
the @xmath10 @xmath11 nuclei are three neutron holes below the @xmath12 closed shell .
their low - lying positive - parity level structure can be interpreted in terms of the spherical shell model as an interplay between proton - particle and neutron - hole excitations in the @xmath13 orbital .
the possible excitations would then be those belonging to the seniority - three configurations : @xmath9 , which can generate spins up to 21/2@xmath6 and @xmath14@xmath15 , terminating at spin 25/2@xmath6 .
the results of g - factor measurements for the lowest 8@xmath6 state in the @xmath16 isotones @xmath17sr @xcite , @xmath18zr and @xmath19mo @xcite indicate that it is essentially built from the alignment of a @xmath13 neutron pair with a small proton admixture which increases with @xmath20 .
therefore , the neutron aligned @xmath21 state would be expected to be yrast in the level structure of @xmath22 isotones .
the high - spin level structure of @xmath23mo ( @xmath24 , @xmath22 ) has been studied by m. weiszflog et al . @xcite .
the shell model interpretation performed with the code ritsschil @xcite and within the @xmath25 model space indicates that the positive - parity states up to 25/2@xmath6 mainly consist of the proton aligned @xmath14@xmath15 configuration .
a particularly interesting case is the one of the 21/2@xmath6 state .
this state can be generated in the neutron fully aligned @xmath9 configuration but the calculations indicate that this component is as small as @xmath26 .
this interpretation has been confirmed by g - factor measurements of the 21/2@xmath6 isomeric state in @xmath23mo , proving the dominance of the @xmath13 proton alignment @xcite .
the trend observed in the @xmath16 isotones and the measurement in @xmath23mo indicate an evolution from neutron to proton alignment , to generate high - spin states in this mass region .
in particular the 21/2@xmath6 states in the @xmath22 isotone @xmath1ru might reveal a complex structure .
understanding the microscopic structure of these levels should therefore shed light on the competition between the possible seniority schemes for the active @xmath13 protons and neutrons .
several groups have already studied the high - spin level structure of @xmath1ru @xcite .
measurements using @xmath27-decay , @xmath0-@xmath0 and @xmath28-@xmath0 coincidences as well as @xmath0-ray anisotropy ratios have been performed and a level scheme proposed .
however , all the spin and parity assignments were based on indirect evidences , systematics or directional correlations of the @xmath0-rays deexciting oriented states ( dco ratios ) with fairly large uncertainties and had to be considered as very tentative .
the proper way to firmly assign a parity to an excited state is to determine the electromagnetic character of the transition deexciting this particular state . to do this
, it is necessary to measure its linear polarization . when combining the polarization information with the angular distribution measurements , the spins and parities of the excited states can be reliably determined . in recent years , due to its high polarization sensitivity and detection efficiency the ge clover detector @xcite has become a useful tool for the measurement of linear polarization by using compton scattering between adjacent crystals . in the present work we report on the results of linear polarization measurements in @xmath1ru populated in the fusion - evaporation reaction @xmath2ni ( @xmath3ar , @xmath4)@xmath1ru by using the exogam ge clover detector array @xcite . as a result of this work , non yrast ( 21/2@xmath6 ) and ( 17/2@xmath6 ) states
have been observed for the first time and added to the positive - parity structure of @xmath1ru . a theoretical understanding of the structures of the first and second ( 21/2@xmath6 ) and ( 17/2@xmath6 )
levels has been obtained in terms of semi - empirical shell model calculations .
in addition , the polarization sensitivity of the exogam clover detectors , acting as compton polarimeters , have been determined over a wide range of @xmath0-ray energies for the first time .
the paper is organized as follows : a description of the experiment at ganil and the data analysis with a special emphasis on the polarization measurements and the first characterization of exogam as a compton polarimeter will be presented in section ii . in section iii
, we present the results obtained for @xmath1ru while the shell model calculations we performed to understand the microscopic nature of the high - spin states in this nucleus will be discussed in section iv .
excited states in @xmath1ru have been investigated using the fusion - evaporation reaction @xmath2ni(@xmath3ar , @xmath4 ) at a beam energy of 111 mev and with an average intensity of 10 pna .
the beam was provided by the cime cyclotron of ganil , caen , france .
the isotopically enriched ( 99.83% ) @xmath2ni targets used in the reactions had an average thickness of 6.0 mg/@xmath29 , enough to stop the recoiling nuclei .
the @xmath0-rays from the reaction products were detected by the exogam ge clover detector array @xcite , consisting of 11 clover - type ge detectors for this experiment , 7 at an angle of 90@xmath30 and 4 at 135@xmath30 relative to the beam direction .
neutrons evaporated from the compound nuclei were detected using the neutron wall array @xcite composed of 50 organic liquid - scintillator elements , covering the forward 1@xmath31 section of the solid angle around the target position .
the light charged particles ( mainly protons and alphas ) were detected by the diamant detector system consisting of 80 csi scintillators @xcite .
details of the experiment have been described earlier @xcite .
events were collected when at least one neutron was detected by the neutron wall and one @xmath0-ray registered in coincidence in the clover detectors . with these trigger conditions a total of 4 @xmath32 10@xmath33 events were recorded . in the off - line processing ,
coincidence data were sorted into symmetric @xmath0-@xmath0 matrices with different conditions on the number of detected neutrons and charged particles .
these conditions were used to assign new @xmath0-rays to @xmath1ru .
coincidence @xmath0-ray spectra were then obtained by setting gates in these matrices .
examples of coincidence spectra are shown in fig .
[ fig01 ] .
[ fig01](a ) is the total projection of a @xmath0-@xmath0 matrix obtained in coincidence with the detection of two protons and one neutron .
this spectrum is dominated by @xmath0-rays from @xmath1ru with some peaks belonging to @xmath34cr produced in the @xmath35o(@xmath3ar,@xmath4 ) reaction , i.e. in the same reaction channel .
this contamination is removed by setting an additional selection of known @xmath0-rays in @xmath1ru .
[ fig01](b ) shows the spectrum obtained after gating on the 974 kev transition previously known as deexciting the first excited state to the ground state in @xmath1ru .
this spectrum contains only known transitions belonging to @xmath1ru with some additional , unknown @xmath0-rays .
further gating on these new transitions has allowed us to confirm their assignment to @xmath1ru and to position them into the level scheme .
this is what is shown in fig .
[ fig01](c ) which gives a spectrum gated on the new 436 kev transition of @xmath1ru .
finally , with the large statistics obtained during this experiment it is possible to perform a more detailed analysis of the observed transitions .
the geometry of the exogam array allowed the assignments of spins from the information on dco ratios @xcite .
for this purpose , an asymmetric particle - gated matrix was constructed in which @xmath0 events recorded at 90@xmath36 were sorted against those recorded at 135@xmath36 .
the experimental dco ratios ( @xmath37 ) were deduced from pairs of gated spectra according to equation @xmath38 the detection efficiencies of detectors at 90@xmath36 and at 135@xmath36 have the same behaviour with @xmath0-ray energy .
therefore their ratio @xmath39 is a constant ( @xmath40 ) hence no efficiency correction of the dco ratios was needed .
[ fig02 ] shows two projected spectra obtained from the dco matrix .
the spectrum in the upper ( lower ) panel corresponds to @xmath0-rays detected at 135@xmath36 ( 90@xmath36 ) and in coincidence with the 497 kev @xmath41 transition in @xmath1ru observed at 90@xmath36 ( 135@xmath36 ) .
the ratio of the peak intensities in these two spectra provides the @xmath37 values of the @xmath0-rays .
for example , the three most intense transitions in both projected spectra shown in fig .
[ fig02 ] are the 616 , 898 and 974 kev @xmath0-rays .
their intensities in the two spectra are [ 2414(62 ) , 6879(90 ) , 6731(88 ) ] and [ 4821(80 ) , 7134(92 ) , 7062(90 ) ] , respectively .
the deduced dco ratios for these transitions are then [ 0.50(2 ) , 0.96(2 ) , 0.95(2 ) ] .
the dco ratios measured for @xmath0-rays in @xmath1ru and also in @xmath1tc produced in the @xmath42 channel are shown in fig .
[ fig03 ] .
the @xmath37 value for known stretched quadrupole transitions is @xmath43 and is @xmath44 for known pure stretched dipoles , when gating on quadrupole transitions .
if the gate is set on a pure stretched dipole transition , then the @xmath37 value for known quadrupole transitions is @xmath45 and is @xmath43 for known pure stretched dipoles .
based on these assignment criteria , the @xmath37 values obtained in the above example suggest that the 616 kev transition is a @xmath46 dipole transition whereas the 898 and 974 kev transitions have a @xmath47 quadrupole character .
these assignments are consistent with the previous assignments @xcite .
it should be noted that for mixed @xmath48 transitions @xmath37 ratios can vary between 0.6 and 1.0 depending on the @xmath49 multipole mixing ratio of the @xmath0-ray .
a further ambiguity arises for non - stretched ( @xmath50 ) pure @xmath51 ( or @xmath52 ) transitions , where @xmath37 for non - stretched dipole transition with @xmath53 mixing ratio is approximately the same as for a stretched quadrupole transition @xcite .
these ambiguities can be resolved by simultaneously measuring the linear polarization of the @xmath0-ray transitions ( see below ) .
for example , stretched @xmath51 , @xmath54 or unstretched @xmath52 transitions and stretched @xmath52 or unstretched @xmath51 transitions have opposite sign linear polarization values @xcite . in order to determine the multipolarity and the electromagnetic nature of a transition
, both the dco ratio and the linear polarization should be measured .
one of the unique capabilities of the exogam array is the possibility to use the clover detectors as compton polarimeters . in the following
, the measurement of the performances of exogam as compton polarimeter will be described .
the clover detectors placed at 90@xmath30 relative to the beam axis were used to determine the linear polarization of @xmath0-ray transitions , since they are the most sensitive to the polarization @xcite .
we define the emission plane by the direction of the initial @xmath0-ray and the beam axis .
the clover detector is composed of four hpge crystals closely packed in the same cryostat . in compton scattering
, the initial and scattered @xmath0-rays can be detected in adjacent crystals of the same detector and analyzed separately according to whether the scattering has occurred horizontally to the emission plane or vertically to it .
two @xmath0-@xmath0 matrices were created as follows : the first @xmath0-ray corresponds to a single - crystal hit in any clover detector of the array and the second one to the sum of the energy deposited in two crystals within the same clover located at 90@xmath30 ( i.e. the addback energy of events scattering between two adjacent crystals of a clover , this one being positioned at 90@xmath30 ) .
the matrices contain therefore events with either horizontally or vertically scattered @xmath0-rays in a clover at 90@xmath30 on one axis and a single - crystal hit on any of the clover detectors on the other axis .
the number of horizontal ( @xmath55 ) and vertical ( @xmath56 ) scatters for a given @xmath0-ray could be obtained by setting gates on @xmath0-ray transitions in the two asymmetric matrices .
the experimental polarization asymmetry is defined by the ratio @xmath57 , @xmath58-n_{\|}}{\left[a\left(\rm e_{\gamma}\right)n_{\bot}\right]+n_{\|}},\label{eq2}\ ] ] where @xmath59(e@xmath60 ) is the normalization factor corresponding to the asymmetry of the exogam clover detectors , and is defined as @xmath61 the normalization factor is a function of @xmath0-ray energy and has been obtained from the measurement with a standard @xmath62eu radioactive source .
[ fig04 ] shows the variation of @xmath59 with energy e@xmath60 .
it was fitted with the expression @xmath63 , resulting in @xmath64 and @xmath65 , where e@xmath60 is in kev . as is clear from fig .
[ fig04 ] , the value of @xmath59 is almost constant and close to unity , showing nearly ideal symmetry of the four - crystal clover detector acting as the compton polarimeter .
the polarization asymmetry @xmath57 is negative for unmixed stretched magnetic transitions and positive for stretched electric transitions .
it is proportional to the degree of linear polarization @xmath66 , @xmath67 where the quality factor @xmath68 is the polarization sensitivity of the polarimeter . @xmath69 and @xmath70 would indicate completely insensitive and completely sensitive polarimeters , respectively . for a point - like polarimeter
, the polarization sensitivity @xmath68 can be calculated from the klein - nishina formula @xcite , which gives @xmath71 where m@xmath72 is the electron rest mass . for a realistic setup of detectors with finite crystal size
, we have to integrate over a certain range of scattering angles leading to a considerable reduction of the polarization sensitivity .
the effective polarization sensitivity is usually given as @xmath73 according to eqs.([eq4])-([eq6 ] ) , @xmath68 and the two parameters @xmath74 and @xmath75 can be experimentally determined using @xmath0-rays whose linear polarization is well known .
theoretical values of the linear polarization can be deduced from the angular distribution . for @xmath0-rays detected at 90@xmath36 with respect to the beam direction , the polarization of pure electric quadrupole transitions can be calculated from the formula @xmath76 where @xmath77 and @xmath78 are the normalized ( @xmath79 ) coefficients of the legendre polynomials in the angular distribution .
to determine the capability of the exogam array to measure linear polarization , we analyzed the angular distribution for the known pure @xmath54 transitions in the energy range 316 to 1264 kev in the level schemes of @xmath1ru @xcite , @xmath1tc @xcite , @xmath19mo @xcite , and @xmath18mo @xcite .
the angular distribution coefficients , @xmath77 and @xmath78 , for each transition , were extracted from least squares fits of the photopeak areas and summarized in table [ tab01 ] .
the deduced values of the linear polarization @xmath66 and the experimental asymmetry ratio @xmath57 for the known @xmath0-ray transitions are also summarized in table [ tab01 ] , along with the polarization sensitivity @xmath68 of the exogam array .
the experimental values of @xmath68 , together with the results of the fit to the data , are shown in fig .
[ fig05 ] .
the coefficients @xmath74 and @xmath75 were determined by a least squares fit to the values of @xmath68 using the function of eq.([eq6 ] ) ; @xmath80 , @xmath81 .
the quality of a polarimeter depends on both its sensitivity to the polarization and the coincidence efficiency between the scatterer and absorber crystals expressed as @xcite @xmath82 where @xmath83 and @xmath84 are the total number of counts and the photopeak efficiency of the clover considered as a single detector at the energy @xmath85 , measured when the @xmath0-ray has no polarization ( i.e. using a source or when the detector is at @xmath86 with respect to the beam direction ) .
finally it is common to compare polarimeters using a figure of merit defined as @xcite @xmath87 the figure of merit deduced for the exogam clover at 1368 kev is 1.51 @xmath88 which is 4.4 times larger that the one measured for the smaller eurogam clover @xcite . at the same @xmath0-ray energy , the measured polarization sensitivity @xmath89 is 0.135(5 ) ( 0.121(5 ) for the eurogam clover ) which means that the increase is due to the much larger coincidence efficiency .
this increase in efficiency makes exogam an ideal polarimeter for low - intensity @xmath0-rays . | linear polarization measurements have been performed for @xmath0-rays in @xmath1ru produced with the @xmath2ni(@xmath3ar , @xmath4@xmath0)@xmath1ru reaction at a beam energy of 111 mev .
the exogam ge clover array has been used to measure the @xmath0-@xmath0 coincidences , @xmath0-ray linear polarization and @xmath0-ray angular distributions .
the polarization sensitivity of the exogam clover detectors acting as compton polarimeters has been determined in the energy range [email protected] mev .
several transitions have been observed for the first time .
measurements of linear polarization and angular distribution have led to the firm assignments of spin differences and parity of high - spin states in @xmath1ru .
more specifically , calculations using a semi - empirical shell model were performed to understand the structures of the first and second ( 21/2@xmath6 ) and ( 17/2@xmath6 ) levels .
the results are in good agreement with the experimental data , supporting the interpretation of the non yrast ( 21/2@xmath6 ) and ( 17/2@xmath6 ) states in terms of the @xmath7 and @xmath8 members of the seniority - three @xmath9 multiplet . |
it has been over forty years since bekenstein and hawking first disclosed that black hole can be considered as a thermodynamic system , with characteristic temperature and entropy @xcite .
taking into account the fact that black holes have no hair , there are no classical degrees of freedom to account for such thermodynamic properties .
it is a general belief that thermodynamic properties of a system may reflect the statistical mechanics of underlying relevant microscopic degrees of freedom .
but the detailed nature of these microscopic gravitational states has remained as a mystery .
the bekenstein - hawking entropy , @xmath6 , depends on both planck s constant as well as newtonian gravitational constant , implying that thermodynamics of black holes may relate quantum mechanics and gravity .
recently , there have been some progresses on understanding the microscopic degrees of freedom of the black hole entropy , for example in string theory @xcite as well as loop quantum gravity @xcite .
but the accounts of the black hole entropy are not complete and they only work within some particular models and some special domains where string theory and loop quantum gravity can apply .
besides , despite counting very different states , many inequivalent approaches to quantum gravity obtain identical results and it is not clear why any counting of microstates should reproduce the same bekenstein - hawking entropy @xcite .
the statistical mechanical description of black hole entropy is still not elegant . on the other side ,
black hole can be heated or cooled though absorption and evaporation processes . according to boltzmann s insight ,
if a system can be heated , it must have microscopic structures .
recently , in @xcite , possible microscopic structures of a charged anti - de sitter black hole have been studied and some kind of interactions between possible micromolecules have been investigated by an interesting physical tool , the ruppeiner geometry . derived from the thermodynamic fluctuation theory , the ruppeiner geometry @xcite is considered powerful to explore the possible interactions between black hole microscopic structures .
the sign of the ruppeiner invariant @xmath7 ( the ricci scalar of ruppeiner geometry ) was argued to be useful for identifying the physical systems similar to the fermi ( bose ) ideal gas when @xmath8 ( @xmath9 ) or the classical ideal gas when @xmath10 @xcite . besides
, the sign of the ruppeiner invariant @xmath7 can further be used to interpret the type of the dominated interaction between molecules of a thermodynamic system . when @xmath8 , there is a repulsive interaction between molecules , when @xmath9 the interaction is attractive , and for @xmath11 there is no interaction in the microstates rup2,rup3,rup4
. moreover , the magnitude of the ruppeiner invariant @xmath12 measures the average number of correlated planck areas on the event horizon for a black hole system rup05 . for a review on the description of the ruppeiner geometry in black hole systems
, we refer to @xcite and references therein .
further studies on molecular interactions of black holes , based on the ruppeiner geometry , have been carried out in @xcite .
phase transition is another interesting topic in black holes thermodynamics .
davies discussed thermodynamic phase transition of the black holes by looking at the behavior of the heat capacity @xcite .
he claimed that the discontinuity of the heat capacity marks the second order phase transition in black holes .
however , it was argued that physical properties do not show any speciality at this discontinuity point if compared with other heat capacity values , for example the regularity of the event horizon is not lost and the black hole internal state remains uninfluenced @xcite .
thus , it is hard to accept the discontinuity point of the heat capacity as a true physical point of the phase transition . employing the landau - lifshitz theory of thermodynamic fluctuations ll1,ll2 , pavon and rubi gave a deep understanding of the black hole phase transition @xcite .
they found that some second moments in the fluctuation of relevant thermodynamic quantities diverge when the black hole becomes extreme .
this divergence shows that the thermodynamic fluctuation is tremendous and the rigorous meaning of thermodynamical quantities is broken down .
this is exactly the characteristic of the thermodynamic phase transition point . at this phase transition point
, the hawking temperature is zero which indicates that for the extreme black hole there is only super - radiation but no hawking radiation , which is in sharp difference from that of the non - extreme black holes .
black holes phase transition in the context of landau - lifshitz theory have been investigated in @xcite .
recently , further differences in dynamical properties before and after the black hole thermodynamical phase transition has been disclosed in wang2,wang3,wang4 .
a question now arises : how we can further understand this macroscopic thermodynamic phase transition in black hole physics ?
for example whether there is a microscopic explanation of this thermodynamic phase transition .
the ruppeiner geometry is a possible tool we can use to investigate the thermodynamic phase transitions from microscopic point of view .
this method is safer to determine true phase transitions than other methods since regardless of microscopic model , @xmath7 has a unique status in identifying microscopic order ( which is at foundation of phase transitions at microscopic level ) from thermodynamics @xcite .
some attempts , in this direction , have been reported in rupgeo1,rupgeo11,rupgeo2,rupgeo3,rupgeo4,rupgeo5,rupgeo6,rupgeo61,rupgeo7,rupgeo8,rupgeo9,rupgeo10 . in a recent work @xcite
, it was found that the divergence of the ruppeiner invariant coincides with the critical point in the phase transition in a holographic superconductor model .
it is interesting to investigate whether the ruppeiner geometry @xcite can present us further reason to determine which of the thermodynamical discussions mentioned above is valid for describing the thermodynamical phase transition . in particular , we would like to explore whether the davies phase transition conjecture can reflect some special properties in microstructures and be in consistent with the ruppeiner geometry description .
if the davies conjecture does not have the microscopic explanation , we will further think about how to improve the davies conjecture to describe the black hole phase transition .
we will employ the black hole in lifshitz spacetime as a configuration to study our physical problems mentioned above .
this spacetime was first introduced in @xcite , which respects the anisotropic conformal transformation @xmath13 , @xmath14 , where @xmath2 is dynamical critical exponent .
for the lifshitz spacetime , it is necessary to include some matter sources such as massive gauge fields @xcite or higher - curvature corrections @xcite to guarantee the asymptotic behavior of the lifshitz black hole .
it is difficult to find an analytic lifshitz black hole solution for arbitrary @xmath2 , although some attempts have been performed @xcite .
this makes the discussion of thermodynamics for such a black hole difficult .
fortunately , in einstein - dilaton gravity with a massless gauge field , it is possible to find an exact lifshitz black solution for arbitrary @xmath2 @xcite .
this model is suggested in the low energy limit of string theory @xcite . while thermodynamical behaviors of uncharged and charged einstein - dilaton - lifshitz black holes
have been revealed in @xcite and @xcite , respectively , thermodynamics of uncharged gauss - bonnet - dilaton - lifshitz solution has been studied in @xcite . it is also interesting to study lifshitz black hole solutions in the presence of other gauge fields such as the power - law maxwell field @xcite , the logarithmic @xcite and exponential exp nonlinear electrodynamics .
for example , thermodynamics and thermal and dynamical stabilities of einstein - dilaton - lifshitz solutions in the presence of power - law maxwell field have been studied in @xcite . in the context of ads
/ cft @xcite application , the electrical conductivity were explored for exponentially @xcite and logarithmic @xcite charged lifshitz solutions . in the present work , we shall consider the born - infeld ( bi ) nonlinear electrodynamics in the context of einstein - dilaton - lifshitz black holes .
the motivation for considering bi - dilaton action comes from the fact that dynamics of d - branes and some soliton solutions of supergravity is governed by the born - infeld ( bi ) action @xcite . besides
, the low energy limit of open superstring theory suggest the bi electrodynamic action coupled to dilaton field @xcite .
it is surprising to mention that , many years before the appearance of bi action in superstring theory , in 1930 s , this nonlinear electrodynamics was introduced for the first time , with the aim of solving the infinite self - energy problem of a point - like charged particle by imposing a maximum strength for the electromagnetic field @xcite . in this paper
, we will first look for a general ( @xmath15)-dimensional lifshitz black hole solution in the context of einstein - dilaton gravity in the presence of bi electrodynamics .
we will show that the general metric function has different solutions for @xmath0 and @xmath1 cases .
it is important to note that the difference in the metric function has not been observed in the previous studies on lifshitz - dilaton black holes kord2,kord3,kord4 .
based on this general solution , we will study thermodynamics of lifshitz - dilaton black holes coupled to linear maxwell field and bi nonlinear electrodynamics .
we will disclose that the hawking - page phase transition @xcite exists both in the presence of linear and nonlinear electrodynamics .
there are some attempts on study phase transitions of uncharged lifshitz solutions for fixed @xmath2 @xcite or in three @xcite and four @xcite dimensions .
the disclosed hawking - page phase transition in this paper is interesting , since it depends on different values of @xmath2 in different spacetime dimensions in the presence of linear maxwell and nonlinear bi electrodynamic fields .
we will further concentrate our attention to understand the thermodynamic phase transition from microstructures .
we shall examine the relation between the ruppeiner geometry and thermodynamical descriptions of the phase transition such as the davies conjecture and the landau - lifshitz method .
we try to give more microscopic understanding of the thermodynamical phase transitions in the black hole system .
we explore the thermodynamic geometry ( ruppeiner geometry ) for linearly and nonlinearly charged lifshitz solutions separately and disclose the properties of interactions between possible black hole molecules . up to our best knowledge
, there is no study of thermodynamic geometry on lifshitz solutions in literature .
interestingly enough , by studying ruppeiner geometry , we have found that our solutions show the van der waals like behavior near critical point in some cases .
the layout of the paper is as follows . in the next section ,
we provide the basic field equations and obtain the bi charged lifshitz - dilaton black hole solutions . in section [ thermo ] , we first explore the satisfaction of the thermodynamics first law for lifshitz - dilaton black holes in the presence of bi electrodynamics .
then , we study different phase transitions including hawking - page phase transition and phase transition at zero temperature for linearly and bi charged cases . in section [ rg ] , we investigate thermodynamic geometry of the obtained solutions for linearly and nonlinearly bi charged cases by adopting the ruppeiner approach .
we finish with summary and closing remarks in section [ sum ] .
in this section , we intend to obtain exact @xmath16-dimensional dilaton - lifshitz black holes in the presence of bi nonlinear electrodynamics . our ansatz for the line elements of the spacetime is mann , tario@xmath17where @xmath18 is dynamical critical exponent and@xmath19is an ( @xmath20)-dimensional hypersurface with constant curvature @xmath21 and volume @xmath22 . as @xmath23 , the line elements ( [ metric ] ) reduce asymptotically to the lifshitz spacetime , @xmath24on the other side , as it is pointed out above , we would like to consider bi nonlinear electrodynamics . in the absence of dilaton field ,
bi lagrangian density is written as @xcite@xmath25where @xmath26 is the born - infeld parameter related to the regge slope @xmath27 as @xmath28 .
@xmath29 is maxwell invariant in which @xmath30}$ ] where @xmath31 is electromagnetic potential .
one of the effects of presence of dilaton field is its coupling with electromagnetic field .
thus , in the presence of dilaton field we deal with a modified form for bi lagrangian density including its coupling with dilaton scalar field @xmath32 @xcite@xmath33where @xmath34 is a constant .
the lagrangian density of string - generated einstein dilaton model @xcite with two maxwell gauge fields tario in the presence of bi electrodynamics can be written in einstein frame as@xmath35where @xmath36 is ricci scalar and @xmath37 and @xmath38 s are some constants . in lagrangian ( [ lag ] ) , @xmath39 where @xmath40}$ ] and @xmath41 is gauge potential . in the large @xmath26 limit , @xmath42 recovers the einstein - dilaton - maxwell lagrangian in its leading order @xcite@xmath43varying the action @xmath44 with respect to the metric @xmath45 , the dilaton field @xmath32 and electromagnetic potentials @xmath31 and @xmath46 , leads us to the following field equations@xmath47@xmath48where we use the convention @xmath49 . using the metric ansatz ( [ metric ] ) , electromagnetic field equations ( [ fe3 ] ) and ( fe4 ) can be solved immediately as@xmath50where @xmath51 , and @xmath52 can be obtained by subtracting ( @xmath53 ) and ( @xmath54 ) components of eq .
( [ fe1 ] ) and solving the resulting equation .
we find @xmath55substituting eqs .
( [ frt ] ) , ( [ hrt ] ) and ( [ phi ] ) in field equations ( [ fe1 ] ) and ( [ fe2 ] ) , one can solve the equations for @xmath56 to obtain@xmath57 , & \text{for } z = n+1,% \end{array}% \right . \label{f4}\]]where we should set @xmath58so that the field equations are fully satisfied . in the solution ( [ f4 ] ) ,
@xmath59 is a constant which is related to the total mass of black brane as we will see in next section .
the integral of the last term of @xmath56 for @xmath60 can be done in terms of hypergeometric function .
thus , @xmath56 can be written as@xmath61note that solution ( [ fr ] ) obviously satisfies the fact that @xmath62 as @xmath23 ( note that @xmath63 ) .
the behavior of @xmath56 for large @xmath26 is@xmath64which reproduces the result of @xcite for every @xmath2 in linear maxwell case .
the behaviors of the metric function for @xmath0 and @xmath1have been depicted in figs .
[ fig1a ] and [ fig1b ] respectively .
it is notable to mention that in the case of @xmath0 , there is no schwartzshild - like black hole since in this case @xmath56 goes to positive infinity as @xmath65 goes to zero .
however , for @xmath1 , we may have schwartzshild - like black hole ( dash - dotted line in fig .
[ fig1a ] ) in addition to nonextreme ( solid line ) and extreme ( dotted line ) black holes and naked singularity ( dashed line ) .
for nonextreme case , there are two inner ( cauchy ) and outer ( event ) horizons . in both figs .
[ fig1a ] and fig1b , we see that the larger the nonlinearity parameter @xmath26 is , the smaller the distance between two inner and outer horizons is so that for large enough @xmath26 s , we have just one horizon ( extreme case ) or naked singularities .
the schwartzshild - like case occurs for lower @xmath26 s in the case of @xmath1 as fig .
[ fig1a ] shows . as one can see in ( [ lbetaf ] ) , the fourth term in expansions for both @xmath66 and @xmath1 cases reproduce the charge term of @xcite in linear maxwell case as one expects .
the temperature of the black hole horizon can be obtained via@xmath67where @xmath68 is the killing vector and @xmath69 is the radius of event horizon . using ( [ t ] ) , one can calculate the hawking temperature as@xmath70where prime denotes the derivative with respect to @xmath65 and @xmath71 .
temperature has the same formula ( temp ) for both @xmath0 and @xmath1 cases .
one can check that for large @xmath26 , ( [ temp ] ) reduces to the temperature of einstein - maxwell - dilaton lifshitz black holes @xcite , namely @xmath72 the entropy of the black holes can be calculated by using the area law of the entropy @xcite which is applied to almost all kinds of black holes in einstein gravity including dilaton black holes hunt1,hunt2,hunt3,hunt4 .
therefore , the entropy of the black brane per unit volume @xmath22 becomes@xmath73 having eqs .
( [ frt ] ) , ( [ phi ] ) and ( [ constants ] ) at hand , we can find electromagnetic gauge potential @xmath74 in terms of hypergeometric function as@xmath75the large @xmath26 behavior of gauge potential is in agreement with kord2 @xmath76 in next section , we will study thermodynamics of dilaton lifshitz black holes in the presence of bi electrodynamics by seeking for satisfaction of thermodynamics first law through calculation of conserved and thermodynamic quantities .
we also show that our lifshitz solutions can exhibit the hawking - page phase transition .
then , we discuss the inside phase transitions of our lifshitz black holes .
this subsection is devoted to study the thermodynamics first law for lifshitz - dilaton black hole solutions in the presence of bi nonlinear electrodynamics . as the first step , we calculate the fundamental quantity for thermodynamics discussions namely mass . for this purpose
, we apply the modified subtraction method of brown and york ( by ) @xcite . in order to use this method ,
the metric should be written in the form@xmath77for our case , it is clear that @xmath78 and thus@xmath79the metric of background is chosen to be the lifshitz metric ( [ mets ] ) i.e. @xmath80 the quasilocal conserved mass can be obtained through @xmath81 where @xmath82 is the determinant of the boundary @xmath83 metric , @xmath84 is the background extrinsic curvature , @xmath85 is the timelike unit normal vector to the boundary @xmath83 and @xmath86 is a timelike killing vector field on the boundary surface . performing the above modified by formalism , the mass of the space time per unit volume @xmath87 can be calculated as@xmath88where the mass parameter @xmath59 can be obtained from the fact that @xmath89 as@xmath90 & \text{for } z = n+1%
\end{array}% \right . \label{mr+}\]]now , we turn to calculate the electric charge of the solution . using the gauss law
, we can calculate the electric charge via @xmath91where@xmath92are respectively the unit spacelike and timelike normals to the hypersurface of radius @xmath65 . using ( [ chdef ] ) , the charge per unit volume @xmath93 can be computed as@xmath94the electrostatic potential difference ( @xmath95 ) between the horizon and infinity is defined as@xmath96using eqs .
( [ at ] ) and ( [ pot ] ) , one can obtain the electric potential@xmath97which is the same for both @xmath0 and @xmath1 cases . in order to investigate the first law of black hole thermodynamics ,
we should obtain the smarr - type formula for mass ( [ mass ] ) .
( [ mr+ ] ) , ( charge ) and ( [ entropy ] ) at hand , the mass can be written as a function of extensive thermodynamic quantities @xmath98 and @xmath99 in the form of@xmath100 , & \text{for } z = n+1,% \end{array}% \right . \label{m}\]]where @xmath101 .
calculations show that intensive quantities@xmath102coincide with those computed by eqs .
( [ temp ] ) and ( [ elecpot ] ) .
therefore , the thermodynamics quantities satisfy the first law of thermodynamics@xmath103for both solutions for @xmath0 and @xmath1 . in next part of this section
, we will discuss the hawking - page and inside black hole phase transitions for our lifshitz solutions . as it is clear from fig .
[ fig1 ] , there are some parameter choices for which we have extreme black holes and therefore zero temperature . in addition , as one can see from fig .
[ hp1 ] , there are some other choices of parameters that show a non - zero positive minimum for temperature @xmath104 .
the influences of different parameters on @xmath105 can be seen from fig .
when we increase the dimension @xmath3 , @xmath105 increases too , while it decreases with increasing @xmath2 . comparing figs .
[ fighp1a ] and [ fighp1b ] , one finds out that the effect of nonlinearity implies increasing in @xmath105 .
the behaviors illustrated in fig . [ hp1 ] present a hawking - page phase transition for the obtained solutions .
let us have a closer look on fig .
[ hp1 ] . in the first part of @xmath106 curves where we have small black holes (
note that @xmath107 ) , @xmath108 which implies negative heat capacity and therefore small black holes are thermally unstable .
but , in the large black holes part of the curves we have a positive heat capacity and therefore large black holes are thermally stable .
in addition to small and large black holes , we have a thermal lifshitz or radiation solution too . since the small black holes are thermally unstable , system has two choices between large black hole and thermal lifshitz that chooses to be on one of them according to the gibbs free energy .
the gibbs free energy@xmath109can be obtained by using ( [ temp ] ) , ( [ entropy ] ) , ( [ mass ] ) , ( charge ) and ( [ elecpot ] ) .
[ hp2 ] and [ hp3 ] show the behavior of gibbs free energy for some choices of parameters .
the two up and bottom branches correspond to small and large black holes , respectively .
the positive gibbs free energy shows that the system is in radiation phase while there is a hawking - page phase transition at intersection point of bottom branch and @xmath110 .
this fact that the gibbs free energy of large black holes always have the lower energy in comparison to small ones confirms the above arguments about the thermal stability of them .
as one moves rightward on temperature axis in @xmath111 diagram , first experiences radiation regime or thermal lifshitz solution for which @xmath112 . at @xmath110 , the hawking - page phase transition between thermal lifshitz and large black holes occurs and for @xmath113 , we are at large black hole phase .
the temperature at which phase transition occurs is called hawking - page temperature @xmath114 .
effects of change in electric potential @xmath95 , critical exponent @xmath2 and nonlinearity parameter @xmath26 can be seen from figs .
[ hp2 ] and [ hp3 ] .
increase in electric potential @xmath95 and critical exponent @xmath2 makes @xmath114 lower .
also , the lower the nonlinearity parameter @xmath26 is , the lower hawking - page temperature @xmath114 is .
note that lower @xmath26 makes the electrodynamics more affected by nonlinearity .
there are at least three well - known ways to discuss the phase transitions inside the black hole .
two of these ways are based on macroscopic point of view and one of them is based on microscopic viewpoint .
the two macroscopic ways are davies @xcite and landau - lifshitz @xcite methods that discuss , respectively , the behavior of heat capacities and thermodynamic fluctuations .
thermodynamic geometry or ruppeiner geometry @xcite is the microscopic way which discusses the phase transitions in addition to type and strength of interactions . in what follows
, we discuss the relation between the phase transitions predicted by ruppeiner geometry and davies method .
next , we will turn to landau - lifshitz theory of thermodynamic fluctuations . in order to discuss thermodynamic geometry , one should study the divergences , sign and magnitude of ricci scalar corresponding to ruppeiner metric ( usually called ruppeiner invariant ) to determine phase transitions and strength and type of dominated interaction between possible black hole molecules @xcite . to do that
, we define the ruppeiner metric in ( @xmath115 , @xmath99 ) space where the entropy @xmath98 is thermodynamic potential as@xmath116the above metric can also be rewritten in the weinhold form@xmath117the ruppeiner invariant corresponding to ( [ metgtd ] ) can be expressed in a general form as@xmath118where @xmath119 and @xmath120 stand for numerator and denominator of @xmath7 .
the divergences of ruppeiner invariant is determined by roots of @xmath120 which is equal to @xmath121 ^{2}$ ] where @xmath122 is determinant of hessian matrix and @xmath123 .
of course , at these divergence points the numerator @xmath119should be finite .
these divergences show both zero temperature and vanishing @xmath124 .
the root of @xmath124 may show us the boundary between thermal stability and instability . for thermal stability , in addition to positivity of determinant of hessian matrix , @xmath125 and @xmath126 should be positive too @xcite .
it is remarkable to note that at the point where @xmath127 vanishes or equivalently heat capacity at constant charge @xmath128 diverges , we have a thermally unstable system due to negativity of @xmath124 if @xmath129 ( which occurs in many of black hole systems ) .
thus , the heat capacity at constant charge @xmath128 can not be suitable thermodynamic quantity to show phase transition of such systems when we have two changing thermodynamic parameters , for instance @xmath98 and @xmath99 .
there are some works in literature ( for instance @xcite ) in which the correctness of ruppeiner method for recognizing the phase transitions has been judged by comparing the ruppeiner and @xmath128 transition points .
this procedure is of course seems to be incorrect according to what we pointed out above .
also , as we discussed in the introduction , divergences of @xmath7 are safer in order to determine phase transitions . on the other hand , in @xcite and @xcite ,
authors have suggested some suitable thermodynamic quantities to show the phase transitions predicted by ruppeiner invariant .
these quantities are specific heat at constant electrical potential , @xmath130 , analog of volume expansion coefficient , @xmath131 , and analog of isothermal compressibility coefficient @xmath132 defined as@xmath133as one can see in appendix [ app1 ] , these thermodynamic quantities have the forms@xmath134it is obvious that these quantities show the same phase transitions as the ruppeiner geometry because all of them diverge at roots of @xmath135 and @xmath130 vanishes at zero temperature where @xmath7 diverges . to show the coincidence of ruppeiner phase transitions and @xmath130 divergences , some proofs have been presented in @xcite .
the above quantities can be considered as improved davies quantities davies1 which present the phase transitions coincided with ruppeiner ones . in the next part , we study the landau - lifshitz theory of thermodynamic fluctuations to explore the possible signature of black hole phase transitions on properties of black hole radiance . here , we seek for any possible effect of transition on black hole radiance by using landau - lifshitz theory of thermodynamic fluctuations @xcite .
we focus on ( @xmath136)-dimensional linearly charged case .
the extension to higher dimensional or nonlinearly charged cases is trivial and give no novel result .
based on landau - lifshitz theory @xcite , in a fluctuation - dissipative process , the flux @xmath137 of a given thermodynamic quantity @xmath138 is given by@xmath139where dot shows temporal derivative and @xmath140 and @xmath141 are respectively the thermodynamic force conjugate to the flux @xmath137 and the phenomenological transport coefficients .
in addition , the rate of entropy production is expressed by@xmath142where `` @xmath143 '' ( `` @xmath144 '' ) holds for the entropy rate contributions which come from the non - concave ( concave ) parts of @xmath98 . the second moments corresponding to the fluxes fluctuations are ( we set @xmath145)@xmath146where the mean value with respect to the steady state is denoted by the angular brackets and the fluctuations @xmath147 are the spontaneous deviations from the value of steady state @xmath148 .
to guarantee that correlations are zero when two fluxes are independent , the kronecker @xmath149 is put in eq .
( [ smom ] ) . according to @xcite
, the mass @xmath115 , electric potential energy @xmath95 and temperature @xmath150 can be obtained for ( @xmath136)-dimensional linearly charged case as@xmath151@xmath152where@xmath153we know that in extreme black hole case , the hawking temperature on the event horizon vanishes and therefore in this case we have @xmath154 .
( [ sm ] ) , we can obtain the entropy production rate as@xmath155where@xmath156the mass loss rate is given by @xcite@xmath157the first term on the right side of eq .
( [ md ] ) is the thermal mass loss corresponding to hawking radiation which is just the stefan - boltzmann law , with @xmath158 ( we set @xmath159 ) as the radiation constant .
the constant @xmath131 depends on the number of species of massless particles and the quantity @xmath82 is the cross - section of geometrical optics . the second term on the right side of eq .
( [ md ] ) is responsible for the loss of mass corresponding to charged particles .
in fact , it is @xmath160 term which rises in first law of black hole mechanics . with references to what explained and computed above
, one can calculate the second moments or correlation functions of the thermodynamical quantities@xmath161@xmath162 \notag \\ & = & -\frac{\pi zl^{z-1}s^{z/2 + 1}}{2^{z-5}\xi } \left [ \dot{m}+\frac{\pi b^{2z-2}q}{z2^{z-2}l^{z-1}s^{z/2}}\dot{q}\right]\end{aligned}\ ] ] @xmath163 ^{2}}{4s^{2}\xi ^{2}}\left\langle \delta \dot{m}\delta \dot{m}\right\rangle \notag \\ & & + \frac{4^{2-z}\pi ^{2}b^{4z-4}q^{2}\left [ ( z-1)s^{z}+4z^{2}(z+2)l^{-2}s^{z+1}+\pi ^{2}z4^{2-z}b^{2z-2}q^{2}\right ] ^{2}% } { z^{2}l^{2z-2}s^{z+2}\xi ^{2}}\left\langle \delta \dot{q}\delta \dot{q}% \right\rangle \notag \\ & & -\frac{\pi b^{2z-2}q\left [ ( z-1)s^{z}+4z^{2}(z+2)l^{-2}s^{z+1}+\pi ^{2}z4^{2-z}b^{2z-2}q^{2}\right ] \left [ ( z-2)s^{z}+4z^{2}(z+2)l^{-2}s^{z+1}+% \pi ^{2}z(z+2)4^{2-z}b^{2z-2}q^{2}\right ] } { z2^{z-2}l^{z-1}s^{z/2 + 2}\xi ^{2}}% \left\langle \delta \dot{m}\delta \dot{q}\right\rangle \notag \\ & = & -\frac{2^{z-5}}{\pi zl^{z-1}s^{z/2 + 3}\xi } \left\ { \left [ ( z-2)s^{z}+4z^{2}(z+2)l^{-2}s^{z+1}+\pi ^{2}z(z+2)4^{2-z}b^{2z-2}q^{2}\right ] ^{2}\dot{m}\right .
\notag \\ & & \left .
-\frac{\pi b^{2z-2}q}{2^{z-4}zl^{z-1}s^{z/2}}\left [ ( z-1)s^{z}+4z^{2}(z+2)l^{-2}s^{z+1}+\pi ^{2}z4^{2-z}b^{2z-2}q^{2}\right ] % \left [ s^{z}-16\pi ^{2}4^{-z}q^{2}b^{2z-2}z(z+1)\right ] \dot{q}\right\}\end{aligned}\ ] ] @xmath164 } { 2^{z-3}l^{z-1}\xi ^{2}}\left\langle \delta \dot{m}\delta \dot{m}% \right\rangle
\notag \\ & & + \frac{\pi ^{3}b^{4z-4}q^{2}\left [ ( z-1)s^{z}+4z^{2}(z+2)l^{-2}s^{z+1}+\pi ^{2}z4^{2-z}b^{2z-2}q^{2}\right ] } { z2^{3z-8}l^{z-1}s^{z/2}\xi ^{2}}% \left\langle \delta \dot{q}\delta \dot{q}\right\rangle \notag \\ & & -\frac{\pi ^{2}b^{2z-2}q\left [ ( 3z-4)s^{z}+12z^{2}(z+2)l^{-2}s^{z+1}+\pi ^{2}z(z+4)4^{2-z}b^{2z-2}q^{2}\right ] } { 2^{2z-5}\xi ^{2}}\left\langle \delta \dot{m}\delta \dot{q}\right\rangle \notag \\ & = & -\frac{\left [ ( z-2)s^{z}+4z^{2}(z+2)l^{-2}s^{z+1}+\pi ^{2}z(z+2)4^{2-z}b^{2z-2}q^{2}\right ] } { l^{2z-2}s\xi } \left [ \dot{m}+\frac{% \pi b^{2z-2}l^{z-1}q}{2^{z-2}zs^{\frac{1}{2}z}}\dot{q}\right ] .
\notag \\ & & \end{aligned}\]]it is clear that second moments @xmath165 , @xmath166 and @xmath167 diverge for extreme black hole case where @xmath168 vanishes ( see eq .
( [ ttt ] ) ) .
it means that there is a phase transition in this case .
this phase transition is between extreme and nonextreme black holes for which we have a sudden change in emission properties . in nonextreme case
, the black hole can give off particles and radiation through both spontaneous hawking emission and superradiant scattering whereas in extreme case , the black hole can just radiate via superradiant scattering . as one can see from eq .
( [ md ] ) , @xmath169 and @xmath170 are related .
therefore , all of the above second moments can be reexpressed in terms of @xmath171 .
let us calculate @xmath170 for our case .
the rate of charge loss can be stated as@xmath172where @xmath173 is the rate of electron - positron pair creation per four - volume and @xmath174 is charge of electron . according to schwinger s theory @xcite for ( @xmath136)-dimensions , the rate of electron - positron pair creation in a constant electric field @xmath175 is@xmath176 , \label{gam1}\]]where @xmath177 and @xmath59 is the mass of electron . in the presence of linear maxwell electrodynamics ,
the electric field is @xmath178 and therefore@xmath179 .
\label{gam2}\]]combining eqs .
( [ gam1 ] ) and ( [ gam2 ] ) , we arrive at @xmath180 , \label{dq2}\]]where @xmath181 $ ] is incomplete gamma function . when @xmath182 , eq .
( [ dq2 ] ) reduces to @xmath183where we have used@xmath184 \approx \exp \left ( -x\right ) x^{1/(z+1)}\left [ \frac{1}{x^{2}}+o\left ( \frac{1}{x^{3}}\right ) + \cdots % \right ] , \]]in which @xmath185 . in the following section
, we turn to study thermodynamic geometry of our black hole solutions to figure out the behavior of black hole possible molecules and phase transitions .
in this section we study thermodynamic geometry of the lifshitz - dilaton black holes for linearly maxwell and nonlinearly bi gauge fields , separately .
we have introduced this method in subsection [ insphase ] with focus on the study of the phase transitions which occur at divergence of ruppeiner invariant @xmath7 .
in addition to divergences , @xmath186 has other properties which give us information about thermodynamic of the system .
the sign of @xmath7 gives us the information about the dominated interaction between possible black hole molecules while its magnitude measures the average number of correlated planck areas on the event horizon @xcite .
@xmath8 means the domination of repulsive interaction , @xmath9 shows the attraction dominated regime and when * * * * @xmath7 vanishes the system behaves like ideal gas i.e. there is no interaction . in continue ,
we first study thermodynamic geometry in the presence of linear maxwell electrodynamics .
then , we extend our study to nonlinearly charged black holes where bi electrodynamics has been employed .
there is just a necessary comment . as we stated before in subsection [ insphase ] , for thermal stability , @xmath125 , @xmath127 and @xmath187 should be positive stability1,stability2 .
one can show that the positivity of @xmath135 and * * * * @xmath125 ( @xmath127 ) imposes the positivity of @xmath126 ( @xmath125 ) .
therefore , we just turn to study the signs of @xmath135 and * * * * @xmath125 in our following discussions to guarantee the thermal stability . the mass and hawking temperature of black holes in the presence of linear maxwell ( lm ) electrodynamics are @xmath188@xmath189as we mentioned above , for investigating thermal stability we need to check the signs of @xmath125 and @xmath124 . in our case
@xmath190thus , in order to disclose the thermal stability of system , we need to study the sign of determinant of hessian matrix .
we find @xmath191 } } { % 4(n-1)(n+z-3)^{2}l^{2z-2}}\mathfrak{f}(s , q)$ & $ z\neq 2 $ \\ & \\ $ \frac{(n+1)b^{2}2^{(n-5)/(1-n)}s^{2(n-2)/(1-n)}}{(n-1)^{2}l^{4}}$ & $ z=2$% \end{tabular}% \right .
, \label{deno}\]]where @xmath192}+\frac{% z(n+z-3)(n+z-1)s^{2[1+(z-1)/(n-1)]}}{2^{-4/(n-1)}(z-2)(n-2)^{2}l^{2}}-\frac{% \pi ^{2}(n+z-3)q^{2}2^{(n-4z+7)/(n-1)}}{(n-1)(n-2)^{2}b^{2(1-z)}}\right ] , \]]the numerator @xmath119 of ( [ rupmax ] ) is a complicated finite function of @xmath98 and @xmath99 in this case , including long terms that we do not express it explicitly for economic reasons .
however , as it was mentioned in subsection [ insphase ] , one can find the denominator @xmath120 in the form of @xmath193 ^{2 } , \label{deno0}\]]where @xmath194 and @xmath124 have been give in ( [ tmax ] ) and ( [ deno ] ) respectively . having eqs .
( [ deno ] ) and ( [ deno0 ] ) at hand , we are in the position to investigate the divergences of @xmath7 , which play the central role in thermodynamic geometry discussions and also thermal stability of system . as one can see from eqs .
( [ deno ] ) and ( [ deno0 ] ) , for @xmath195 , the divergences occur just in the case of the extremal black holes where @xmath196 . for @xmath197 , in addition to extremal black hole case , @xmath198 diverges in zeros of ( [ deno ] ) . in the latter case
, we can calculate the corresponding temperature by solving @xmath199 for @xmath99 and then putting this @xmath99 in eq .
( [ tmax ] ) to arrive at @xmath200the above temperature is negative for @xmath201 i.e. there is no black hole at this diverging point and therefore the divergences of @xmath7 occur just for extremal black hole case when @xmath201 .
however , for @xmath5 when @xmath202 , we can see an upper limit in entropy and charge of system .
the largest entropy @xmath98 for which @xmath199 ( which we call it critical entropy @xmath203 ) can be calculated by finding the extremum point where @xmath204 as @xmath205at which @xmath206and @xmath207 ^{2-z}<0.\]]one should note that the absolute value of @xmath208 is also the largest charge value which satisfies @xmath199 . another remark to be mentioned is that ( [ sc ] ) imposes an upper limit on the size of black hole too ( see ( [ entropy ] ) ) . at this point
, the corresponding temperature can be obtained as @xmath209for charges greater than @xmath208 , the ruppeiner invariant diverges only in the case of extremal black holes . for @xmath210 , in addition to @xmath211 , we have one other divergence in @xmath7 specified by ( [ sc ] ) and ( [ tc ] ) . for @xmath212 , in addition to @xmath211 , we have at most two other divergences since the order of polynomial in term of @xmath98 is always lower than @xmath213 for @xmath214 and @xmath5 .
one should note that , in latter case , the temperature region between two divergences is not allowed since @xmath215 ( fig .
( [ fig2 ] ) ) .
we have summarized the above discussion in figs .
[ fig3 ] and [ fig4 ]
. these figures also show the sign of ruppeiner invariant for different choices of parameters that determines the type of interaction between black hole molecules @xcite .
[ fig3a ] is depicted for rn - ads case ( @xmath216 , @xmath217 ) . in this case
, it can be seen that for @xmath218 , the ruppeiner invariant diverges only for extremal black holes . as fig .
fig3a shows , there is also a range of @xmath150 for which @xmath9 , namely the dominated interaction between black hole molecules is attractive .
furthermore , the interactions near zero temperature is the same as interactions of fermi gas molecules near zero temperature @xcite . according to fig .
[ fig2a ] , for @xmath218 , @xmath124 is positive ( also @xmath219 ( see ( [ mqq ] ) ) ) , and therefore the system is stable for all @xmath150 region . for @xmath210 , in addition to zero temperature , we have another temperature ( @xmath220 ) , that divergence of @xmath7 occurs at it ( see fig . [ fig3a ] ) . at zero temperature
, the ruppeiner invariant goes to @xmath221 while at @xmath220 , it goes to @xmath222 .
the latter case is similar to the van der waals gas phase transition at critical point in this sense that in phase transition temperature , * * * * @xmath198 goes to @xmath222 @xcite . for @xmath210 ,
@xmath223 becomes positive when we get away from second divergence ( @xmath220 ) on temperature axis . in @xmath210 case
, @xmath124 is positive and just vanishes at @xmath220 ( fig .
[ fig2a ] ) , so , the system is always thermally stable**. * * for @xmath212 , there are three divergences ; one at @xmath224 , one at @xmath225 and one at @xmath226 . in this case , according to fig .
[ fig2a ] , @xmath124 is negative in the temperature region between two roots and show instability .
this not - allowed region is equivalent to the temperature region between two divergences of ruppeiner invariant for * * * * @xmath212 ( fig . [ fig3a ] ) . figures [ fig2b ] and * * [ fig3b ] show the same properties for black holes with different parameters . in this case , @xmath220 is greater than one of previous case while @xmath208 is lower .
[ fig4 ] shows the behavior of ruppeiner invariant for @xmath227 .
as this figure shows , there are just divergences at @xmath224 .
the properties of black hole molecular interactions ( @xmath8 : repulsion , @xmath10 : no interaction and @xmath9 : attraction ) depend on parameters such as dimension of space time and charge , in this case . according to eq .
( [ deno ] ) , for @xmath195 , * * * * @xmath135 is always positive . for @xmath201
, we can find @xmath99 from * * * * @xmath196 and put it in * * * * @xmath124 to receive @xmath228thus , since @xmath124 nowhere vanishes for @xmath201 ( see discussions below ( [ tt ] ) ) and is positive at @xmath224 according to above equation , it is positive throughout the temperature region and therefore system is always thermally stable for @xmath201 . regarding the nature of phase transition occurred at zero temperature where ruppeiner invariant diverges
, we discussed in previous section via landau - lifshitz theory of thermodynamic fluctuations .
however , regarding the phase transitions occurred at divergences of @xmath7 at finite temperatures , we can give some comments here .
we have seen two kinds of phase transitions here for @xmath5 ( see fig . [ fig3 ] ) namely continues ( for @xmath210 where @xmath7 diverges at just one finite temperature or entropy ) and discontinues ( for @xmath212 where @xmath7 diverges at two finite temperatures or entropies and we have a jump between these two points since there is no thermally stable black hole between them ) .
both of these phase transitions can be considered as small / large black holes phase transitions .
the first reason for this argument is that as temperature increases , entropy or equivalently size of black hole increases ( note that @xmath229 ) .
therefore , the left side of phase transition points where temperature is lower , we have small size black holes and the right side where temperature is higher we have large size ones .
this fact can also be seen from the behavior of ruppeiner invariant magnitude in two sides of phase transitions . for small black holes
, we expect the finite correlation between possible black hole molecules ( of course far from phase transition points ) because those are close to each other . for large black holes ,
we expect the correlation between possible molecules to tend to a small value near zero since molecules become approximately free .
these expected behaviors can be seen in fig .
[ fig3 ] . for born - infeld case
, we can calculate the ruppeiner invariant by using eqs .
( [ temp ] ) , ( [ m ] ) and ( [ metgtd ] ) .
the ruppeiner invariant in this case is very complicated due to the presence of hypergeometric functions .
therefore , in this case we discuss the thermodynamic geometry non - analytically and by looking at plots .
we study @xmath5 , @xmath201 , @xmath195 and @xmath66 cases separately .
first , we study @xmath5 case . fig .
[ fig5a ] shows that changing @xmath26 can cause change in dominated interaction .
for instance , in a range of @xmath150 , we have negative @xmath7 ( attraction ) for @xmath230 ( note that in this range system is thermally stable as one can see from figs . [ fig6a ] and [ fig7a ] ) . for @xmath230 ,
the system behaves like fermi gas in zero temperature namely @xmath7 goes to positive infinity at zero temperature @xcite . for @xmath231 ,
ruppeiner invariant diverges at two points that one of them is zero temperature .
according to fig .
[ fig6a ] , for temperatures lower than the second divergence point , @xmath124 is negative and therefore system is thermally unstable . since @xmath219 ( fig .
[ fig7a ] ) , the system is thermally stable just for temperatures greater than the temperature of second divergence for @xmath231 .
[ fig5a ] shows that there is no extremal black hole for @xmath232 i.e. we have a black hole with just single horizon .
the allowed temperature region is temperatures greater than the temperature of divergence according to figs .
[ fig6a ] and [ fig7a ] . in figs .
[ fig5b ] , [ fig6b ] and fig7b , respectively ruppeiner invariant , @xmath124 and @xmath233 are depicted for different choices of parameters .
it is remarkable to mention that , in the case of @xmath234 , fig .
[ fig5b ] shows that the behavior of system looks like van der waals gas at phase transition temperature i.e. @xmath7 goes to negative infinity at this point rup1,rup6 . for @xmath201 , the behavior of @xmath7
is depicted in fig .
[ fig8a ] .
it can be seen that the type of dominated interaction changes for different @xmath26 s and we have negative @xmath7 for some cases . in this case
, we have a behavior like fermi gas at zero temperature for extremal black holes . for @xmath235 , there is a divergence at non - zero temperature that for temperatures lower than it , system is unstable ( figs . [ fig8b ] and [ fig8c ] ) . in the case of
* * * * @xmath195 , @xmath135 and @xmath125 are @xmath236and@xmath237which are always positive and therefore system is always stable and @xmath198 experiences no divergence ( fig . [ fig9 ] ) . in this case , for different values of nonlinear parameter @xmath26 , we have different dominated interaction . for this case , possible molecules of black hole behave like fermi gas at zero temperature . the last case is @xmath0 . in this case
@xmath125 is @xmath238which is positive for all temperatures .
the behavior of ruppeiner invariant and @xmath124 are depicted in figs .
[ fig10a ] and fig10b for this case , respectively .
as one can see the type of interaction is @xmath26-dependent for some temperatures . for @xmath239 ,
@xmath135 is positive just for temperatures greater than the finite temperature of divergence ( fig .
[ fig10b ] ) and therefore system is thermally stable for this range of temperatures .
most of phase transitions discovered above in the presence of bi electrodynamics at finite temperatures , can not be interpreted as small / large black hole phase transitions because in these cases small size black holes are unstable .
further studies to disclose the nature of these phase transitions are called for .
in many condensed matter systems , fixed points governing the phase transitions respect dynamical scaling @xmath13 , @xmath240 where @xmath2 is dynamical critical exponent .
the gravity duals of such systems are lifshitz black holes . in this paper , we first sought for the ( @xmath15)-dimensional born - infeld ( bi ) charged lifshitz black hole solutions in the context of dilaton gravity .
we found out that these solutions are different for the cases @xmath0 and @xmath1 .
we obtained both these solutions and showed that the solution for the case @xmath0 can never be schwartzshild - like .
then , we studied thermodynamics of both cases by calculating conserved and thermodynamical quantities and checking the satisfaction of first law of thermodynamics .
after that , we looked for the hawking - page phase transition for our solutions , both in the cases of linearly and bi charged black holes .
we studied this phenomenon and effects of different parameters on it by presenting the behaviors of temperature @xmath150 with respect to entropy @xmath98 at fixed electrical potential energy @xmath95 and also gibbs free energy @xmath241 with respect to @xmath150 .
then , we turned to discuss the phase transitions inside the black holes . in this part
, we first presented the improved davies quantities that show the phase transition points coincided with ones of ruppeiner geometry .
this coincidence has been proved directly in appendix [ app1 ] .
all of our solutions provided that those are thermally stable at zero temperature show the divergence at this point both from ruppeiner and davies points of view . using landau - lifshitz theory of thermodynamic fluctuations
, we showed that this phase transition is a transition on radiance properties of black holes . at zero temperature
, an extreme black hole can just radiate through superradiant scattering whereas a nonextreme black hole at finite temperature can give off particles and radiation via both spontaneous hawking radiation and superradiant scattering .
next , we turned to study ruppeiner geometry for our solutions .
we investigated thermal stability , interaction type of possible black hole molecules and phase transitions of our solutions for linearly and nonlinearly bi charged cases separately . for linearly charged case , we showed that there are no diverging points for ricci scalar of ruppeiner geometry ( ruppeiner invariant ) at finite temperature for the case @xmath227 . for @xmath5 , it was found that the number of divergences ( which show the phase transitions ) at finite temperatures depend on the value of charge @xmath99 .
we introduced a critical value for charge @xmath208 that for values greater than it there is no divergence at finite temperature , for values lower than it there are at most two divergences and for @xmath210 , there is just one diverging point for ruppeiner invariant . for the case of @xmath212 , there is a thermally unstable region for systems between two divergences at finite temperatures .
so , this phase transition can be claimed as a discontinues phase transition between small and large black holes . for small black holes not close to transition point , we observed finite magnitude for ruppeiner invariant @xmath7 .
this is reasonable since the magnitude of @xmath198 shows the correlation of possible black hole molecules .
also , for large black holes the magnitude of ruppeiner invariant tends to a very small value as expected . for @xmath210 ,
the solutions show a continues small / large black holes phase transition at finite temperature . in the case of bi charged solutions , we investigated the ruppeiner geometry and thermal stability for @xmath5 , @xmath201 , @xmath195 and @xmath0 separately . in some of these cases ,
small black holes were thermally unstable .
so , more studies are called for to discover the nature of phase transitions at diverging points of @xmath7 .
in both linearly and nonlinearly charged cases , for some choices of parameters , the black hole system behaves like a van der waals gas near transition point .
finally , we would like to suggest some related interesting issues which can be considered for future studies .
it is interesting to repeat the studies here such as hawking - page phase transition , ruppeiner geometry and landau - lifshitz theory for black branes to discover the effect of different constant curvatures of ( @xmath20)-dimensional hypersurface on those phenomena .
one can also seek for any signature of different phase transitions discovered here such as hawking - page phase transition , and phase transitions determined by ruppeiner geometry , in dynamical properties of solutions by investigating quasi - normal modes .
some of these works are in progress by authors .
we are grateful to prof .
m. khorrami for very useful discussions .
mkz would like to thank shanghai jiao tong university for the warm hospitality during his visit . also , mkz , ad and as thank the research council of shiraz university .
this work has been financially supported by the research institute for astronomy & astrophysics of maragha ( riaam ) , iran .
in @xcite and @xcite , authors have shown that the divergences of specific heat at constant electrical potential , @xmath130 , analog of volume expansion coefficient , @xmath131 , and analog of isothermal compressibility coefficient @xmath132 are in coincident with the phase transitions specified by ruppeiner invariant . the definition of these quantities
are@xmath242here we will prove that these quantities are exactly suitable ones to characterize phase transitions shown by ruppeiner invariant .
we showed in section [ rg ] that divergences of ruppeiner invariant occurs at roots of determinant of hessian matrix @xmath122 and also zero temperature . in our proof
, we will show that @xmath135 exactly exist at denominator of all above suitable thermodynamic quantities . in the last line of ( [ a1 ] ) , we have used ( [ intqua ] ) .
( [ a1 ] ) shows that @xmath122 is in denominator of @xmath246 and therefore it exactly diverges at the point where ruppeiner invariant diverges . to show this fact for @xmath131
, we should obtain r. tharanath , j. suresh , n. varghese and v. c. kuriakose , _ thermodynamic geometry of reissener - nordstrm - de sitter black hole and its extremal case , _ gen .
_ _ _ _ * * 46 , * * 1743 ( 2014 ) [ arxiv:1404.6789 ] .
g. bertoldi , b. a. burrington and a. w. peet , _ thermal behavior of charged dilatonic black branes in ads and uv completions of lifshitz - like geometries , _ phys . rev .
d * * 82 , * * 106013 * * * * ( 2010 ) [ arxiv:1007.1464 ] .
m. kord zangeneh , a. dehyadegari , a. sheykhi and m. h. dehghani , _ thermodynamics and gauge / gravity duality for lifshitz black holes in the presence of exponential electrodynamics , _ jhep * 1603 , * 037 ( 2016 ) [ arxiv:1601.04732 ] .
o. d. andreev and a. a. tseytlin , _ partition - function representation for the open superstring effective action : cancellation of mbius infinites and derivative corrections to born - lnfeld lagrangian , _ nucl .
b * * 311 , * * 205 ( 1988 ) . | in this paper , we first obtain the higher dimensional dilaton - lifshitz black hole solutions in the presence of born - infeld ( bi ) electrodynamics .
we find that there are two different solutions for @xmath0 and @xmath1 cases where @xmath2 is dynamical critical exponent and @xmath3 is the number of spatial dimensions .
calculating the conserved and thermodynamical quantities , we show that the first law of thermodynamics is satisfied for both cases .
then , we turn to study different phase transitions for our lifshitz black holes .
we start with hawking - page phase transition and explore the effects of different parameters of our model on it for both linearly and bi charged cases .
after that , we discuss the phase transitions inside the black holes .
we present the improved davies quantities and prove that the phase transition points shown by them are in coincident with ruppeiner ones .
we show that the zero temperature phase transitions are transitions on radiance properties of black holes by using landau - lifshitz theory of thermodynamic fluctuations .
next , we turn to study ruppeiner geometry ( thermodynamic geometry ) for our solutions .
we investigate thermal stability , interaction type of possible black hole molecules and phase transitions of our solutions for linearly and bi charged cases separately . for linearly charged case , we show that there are no phase transition at finite temperature for the case @xmath4 . for @xmath5
, it is found that the number of finite temperature phase transition points depends on the value of black hole charge and is not more than two .
when we have two finite temperature phase transition points , there are no thermally stable black hole between these two points and we have discontinues small / large black hole phase transitions . as expected , for small black holes , we observe finite magnitude for ruppeiner invariant which shows the finite correlation between possible black hole molecules while for large black holes , the correlation is very small . finally , we study the ruppeiner geometry and thermal stability of bi charged lifshtiz black holes for different values of @xmath2 .
we observe that small black holes are thermally unstable in some situations . also , the behavior of correlation between possible black hole molecules for large black holes is the same as linearly charged case . in both linearly and bi charged cases , for some choices of parameters ,
the black hole systems behave like a van der waals gas near transition point . |
the quest for superconductivity in layered structures has become the focus of intense research since the discovery of superconductivity in mgb@xmath2 ( t@xmath6 k ) @xcite .
the layered structure of mgb@xmath2 generates one of its most prominent features , namely the b 2@xmath7 orbitals form @xmath3-bands @xcite which are weakly dispersing along the @xmath8 direction and have a marked two dimensional character . in mgb@xmath2
the @xmath9bands are hole doped , but the top of these bands is only @xmath10 ev higher than the fermi level .
the @xmath9bands fermi surface sheets @xcite , two slightly warped cylinders with axis perpendicular to the boron layers , generate a huge electron - phonon coupling along the @xmath8 direction .
the carriers in the @xmath11bands , formed by the b 2@xmath12 orbitals , further enhance the average electron - phonon coupling @xcite .
the formation of @xmath3 and @xmath4 states is typical of graphite - like structures composed by boron or carbon atoms . given the success of mgb@xmath2 it is natural to look for high @xmath5 superconductivity in structures having similar features .
the problem is that , given the boron layers , small variations in the valence or mass of the intercalant or in the structural parameters are sufficient to considerably alter the @xmath3 or @xmath4 bands positions or the shape of their fermi surfaces and consequently destroy superconductivity . for one or some of these reasons alb@xmath2 , zrb@xmath2 , nbb@xmath2 , mob@xmath2 , yb@xmath2 , tab@xmath2 , tib@xmath2 , hfb@xmath2 , vb@xmath2 and crb@xmath2 are not superconducting @xcite .
the hope of finding new superconducting materials in layered structures was recently increased by the discovery of superconductivity in the graphite intercalated compounds , ybc@xmath13 and cac@xmath13 @xcite .
this is particularly promising since a huge number of intercalants are available for graphite @xcite . in cac@xmath13 , despite the layered structure and the existence of @xmath3 and @xmath4 bands originated from the carbon 2@xmath14 orbitals , the electronic structure close to the fermi level is completely different from that of mgb@xmath2 .
the @xmath4 bands , reminiscent of the graphite ones , and an intercalant free - electron - like band @xcite cross the fermi energy .
the intercalant band forms a spherical fermi surface @xcite .
the electron - phonon coupling of cac@xmath13 ( @xmath15 ) is mainly due to coupling of the interlayer band with c vibrations perpendicular to the graphite layers and with ca vibrations .
so , even though missing the @xmath3-bands , cac@xmath13 reaches an interesting 11.5 k @xmath5 .
this temperature is substantially enhanced by pressure ( t@xmath16 k at @xmath17 gpa @xcite ) , contrary to what happens in mgb@xmath2 .
as can be seen from the above examples , even if one restricts to sandwich structures formed by boron or carbon layers , the details of the electronic and phonon spectra and , subsequently , the critical temperature can change dramatically when the intercalant is included . as a consequence ,
a theoretical approach is absolutely necessary to identify the most probable superconductors or at least to exclude the less probable ones .
an attempt in this direction has been recently made in ref .
@xcite , where by using _
ab initio _ methods the authors studied the possible hole - doping of libc , a @xmath18 ev gap semiconductor .
the authors suggested that a @xmath5 of the order of mgb@xmath2 could be reached if the insulating libc is substantially doped with holes to obtain li@xmath19bc .
successive experimental studies have indicated that the structural response to the introduction of holes unfavorably modifies the electronic structure of li@xmath20bc , and so far no high @xmath5 superconductivity has been found in this system @xcite . ideally designing new superconductors _ ab initio _ requires three steps .
the first is the determination of the most stable structures given a set of atomic species .
the second is the calculation of the electronic structure to verify that the given structure is at least metallic or can be made metallic easily .
the third is the determination of the phonon dispersion and of the electron - phonon parameters .
the first point is a daunting task even if one restricts one s search to a specific set of likely candidates@xcite .
a systematic approach to tackle this problem has been recently offered in the way of data mining of _ ab initio _
calculations @xcite . in this method
one uses the informations obtained from _ ab initio _ calculations of many different structures to build a database that can be then used to judge the stability of new structures .
application of this method to intermetallics has led to the identification of new layered lithium monoboride phases which have a good chance to form under proper synthesis conditions @xcite .
once a stable metallic structure is given , a calculation of the phonon spectra and of the electron - phonon coupling needs to be performed to obtain @xmath5 .
indeed , while some qualitative information can be extracted from electronic - structure @xcite , for a quantitative analysis step three is absolutely necessary . in this work
we investigate the superconducting properties of the previously determined metal sandwich ( ms ) lithium monoboride @xcite by calculating its phonon spectrum and electron - phonon parameters .
this system is metallic and , from qualitative arguments , one can infer that @xmath5 is of the same order of that of mgb@xmath2 @xcite .
indeed this system has an electronic structure which is a hybrid between those of mgb@xmath2 and cac@xmath13 , since there are hole - doped @xmath9bands forming cylindrical fermi surfaces and there is an intercalant band crossing the fermi level .
moreover the deformation potential is comparable to that of mgb@xmath2 @xcite . from this point of view , lib is a much more promising material than li@xmath19bc , because even without doping it has a significant density of @xmath3-states at the fermi level . unfortunately , the full electron - phonon coupling calculations performed in this paper indicate that lib should have a @xmath5 in a 10 - 15 k range .
we show that lib captures some of the important physics of mgb@xmath2 , namely the role of the @xmath9bands , but it lacks the contribution of the @xmath4 states to the electron - phonon coupling and it is only a far relative of cac@xmath13 because the interlayer band is very weakly coupled with the phonons . in an attempt to improve the situation we examine what role the hydrostatic pressure and doping can play in determining the critical temperature .
in all our calculations of the layered lithium monoboride we have used the ms1 theoretical crystal structure basing our choice on the following considerations . on the one hand , it has the smallest unit cell of all ms structures , which offers computational efficiency . on the other hand , even though other stacking sequences are possible ( e.g. ms2 , refs .
@xcite ) , ms1 is a good representative model of the layered lithium monoboride because the long - period shifts are expected to have little effect on its superconducting properties @xcite .
ms1-lib has a rhombohedral unit cell with r3 m space group .
there are four atoms in the primitive unit cell with wyckoff positions li@xmath21 , li@xmath22 , b@xmath23 and b@xmath24 .
the fully relaxed parameters are @xmath25 , @xmath26 .
density functional theory ( dft ) calculations are performed using the quantum espresso code @xcite within the generalized gradient approximation ( gga ) @xcite .
we use norm - conserving pseudopotentials @xcite with configuration @xmath27 and non - linear core correction @xcite for li , and configuration @xmath28 for b. the wavefunctions are expanded using a 50 ry cutoff .
the dynamical matrices and the electron - phonon coupling are calculated using density functional perturbation theory in the linear response @xcite . for the electronic integration in the phonon calculation
we use an @xmath29 uniform @xmath30-point mesh and hermite - gaussian smearing from 0.05 ry . for the evaluation of the electron - phonon coupling we use
an @xmath31 monkhorst - pack mesh . for the @xmath32 average over the phonon momentum * q * we use an @xmath33 @xmath34point mesh .
the phonon dispersion is obtained by fourier interpolation of the dynamical matrices computed on the @xmath35 mesh . the pressure- and doping - induced changes in the electronic properties of lib are studied with vienna ab initio simulation package vasp @xcite within the gga @xcite .
we use projector augmented waves ( paw ) @xcite pseudopotentials , in which li semi - core states are treated as valence ; the energy cutoff is set at 30 ryd .
the projected electronic density of states ( edos ) is found by decomposition of the wavefunction within a sphere of the default paw radius of 1.7 a.u .
for the ms1 unit cell , the @xmath36-ms1 and @xmath37-ms2 supercells we use @xmath38 , @xmath39 , and @xmath40 monkhorst - pack * k*-meshes , respectively .
the band structure of li@xmath41b@xmath41 is presented in fig . [
fig : bands ] ( see footnote @xcite for high - symmetry points notation ) . similarly to what happens in mgb@xmath2 @xcite , there are two boron @xmath3-bands crossing the fermi energy @xmath42 . compared to the @xmath3-bands in mgb@xmath2 , these bands are even more two - dimensional ( due to the larger interlayer distance ) and shifted by more than 0.6 ev to higher energies at the @xmath43 point . as in mgb@xmath2 , they generate two cylindrical fermi surfaces ( in our case with axes along the @xmath44 direction , fig . [ fig : fs ] ) .
the boron @xmath4 states in lib resemble more the @xmath4 states of graphite , as they cross exactly at @xmath42 , so that lib is lacking @xmath4 fermi surfaces altogether . in mgb@xmath2
these states cross at about 2 ev above @xmath42 , which leads to the appearance of an extended @xmath4 fermi surface @xcite .
another important difference between the electronic structures of the two borides is the presence of a lithium band at @xmath42 in lib .
the position of this band resembles the intercalant band in cac@xmath13 @xcite , although in lib it has substantial hybridization to boron states close to the @xmath45-point . the corresponding fermi surface ( a compressed sphere )
is depicted in fig .
[ fig : fs ] .
-direction.,height=245 ] the total density of states ( edos ) and the edos projected over atomic orbitals is illustrated in fig [ fig : dos ] .
the main component at @xmath42 is given by boron @xmath46 states .
as in graphite the boron @xmath47 edos at @xmath42 is zero and increases slowly and linearly immediately after @xmath42 .
and b@xmath48 respectively ) and out - of - plane li and b vibrations ( labeled li@xmath49 and b@xmath49 respectively ) .
notation for the phonon modes at the @xmath43-point , given next to the vertical axis , is explained in the text.,width=340 ] the phonon dispersion and the phonon density of states ( phdos ) are illustrated in fig .
[ fig : phonon_dispersion ] .
the phonon modes at the @xmath43-point are decomposed as @xmath50@xcite . to distinguish between modes with the same symmetry but different eigenvectors we use the following notation : @xmath51 @xmath52 $ ]
, @xmath53 @xmath54 $ ] , @xmath55 @xmath56 $ ] , @xmath57 @xmath58 $ ] , @xmath59 @xmath60 $ ] , @xmath61 @xmath62 $ ] , @xmath63 @xmath64 $ ] , @xmath65 @xmath66 $ ] , where in brackets we give the corresponding atoms and vibrations .
for convenience , we label phonon branches everywhere in the brillouin zone using the name of their representation at @xmath43 . except for acoustic modes , a clear separation exists between optical li and b vibrations .
li modes are confined in the 40 - 55 mev region and are not dispersive , meaning that li - vibrations behave essentially as einstein modes .
boron in - plane vibrations are softened along the @xmath44 direction due to coupling to the @xmath3 bands .
the softening at @xmath43 of the @xmath67 phonon branches is approximately 20 mev , to be compared with the almost 40 mev in mgb@xmath2 for the @xmath68 modes @xcite .
this suggests a strong coupling of the @xmath3 bands to the in - plane vibrations in lib @xcite , but not as strong as in the case of mgb@xmath2 .
the three acoustic modes along the @xmath44 direction are substantially softer with respect to the other directions . at the zone border ,
@xmath45 , these modes are formed by ( i ) in - plane ( li+b)@xmath48 vibrations at energies @xmath69 mev and ( ii ) out - of - plane ( li+b)@xmath70 vibrations at @xmath71 mev .
they can be related to the soft modes at the @xmath43 point in ms2-lib , discussed in ref .
@xcite .
the superconducting properties can be understood calculating the electron - phonon coupling @xmath72 for a phonon mode @xmath73 with momentum @xmath74 : @xmath75 , where @xmath76 is the amplitude of the displacement of the phonon , @xmath77 is the kohn - sham potential and @xmath78 is the electronic density of states at the fermi level .
the calculated average electron - phonon coupling is @xmath79 ( @xmath80 and @xmath81 are the previously defined @xmath30-space and @xmath74-space mesh dimensions , respectively ) .
the eliashberg function @xmath82 and the integral @xmath83 are shown in fig .
[ fig : alpha2fdos ] .
as can be seen most of the contribution comes from phonon states in the 60 - 90 mev region and a smaller contribution comes from low energy states .
an estimate of the different contributions of the in - plane ( li@xmath48 and b@xmath48 ) and out - of - plane ( li@xmath70 and b@xmath70 ) vibrations to @xmath32 can be obtained from the relation @xmath84_{i\alpha , j\beta } [ { \bf c_q}^{-1}]_{j\beta , i\alpha}\ ] ] where @xmath85 indices indicate the displacement of the @xmath86 atom in the cartesian direction @xmath87 , @xmath88_{i\alpha , j\beta}=\sum_{{\bf k},n , m}4 { \tilde g}_{i\alpha}^{*}{\tilde g}_{j\beta } \delta(\epsilon_{{\bf k}n } ) \delta(\epsilon_{{\bf k+q}m})/[n(0 ) n_{k}]$ ] , and @xmath89 the off - diagonal terms are small ( but not negligible ) compared to the total @xmath32 .
most of the coupling is to the in - plane b vibration ; contributions from the li and the out - of - plane b vibrations are smaller . since the @xmath3-bands do not couple to the b@xmath70 vibrations and
since there are no @xmath4 fermi surfaces , the coupling to b@xmath70 vibrations is due to the intercalant band .
note that the decomposed values of @xmath32 contain contributions from different modes and are summed over all the @xmath74-points in the brillouin zone .
for example , @xmath90 includes the coupling to the in - plane @xmath67 , @xmath91 , and @xmath92 branches . by examining the integrated eliashberg function @xmath93 in fig .
[ fig : alpha2fdos](c ) and the phonon characters in fig .
[ fig : phonon_dispersion ] one can infer that the @xmath67 branch is the most important of the three : among them it has the highest phdos in the 70 - 100 mev range , in which @xmath32 gains most of its total value .
the soft in - plane @xmath92 branch is far less important , as the net contribution from all the soft modes having energy under 20 mev is only @xmath94 ( fig .
[ fig : alpha2fdos](c ) )
. it is instructive to compare our result with other layered superconductors . in mgb@xmath2
the coupling of the @xmath3-bands to the phonon modes is @xmath95 @xcite , while in lib the corresponding value is less than 0.46 , as discussed above .
this difference can be clarified by noting that the @xmath68 phonon linewidth @xmath96 along @xmath97 in mgb@xmath2 happens to be comparable in magnitude with that of the @xmath67 mode along @xmath44 in lib .
therefore , the reduced electron - phonon coupling in lib is mainly due to the @xmath67 phonon frequency being harder than the @xmath68 one in mgb@xmath2 ( @xmath98 at @xmath43 ) .
this unfortunate result can be linked to the absence of the @xmath4 carriers , which play an important role in softening of the @xmath68 mode in mgb@xmath2@xcite .
we find that lib and graphite intercalated compounds have few similarities in terms of superconducting features .
in particular , in cac@xmath13 the intercalant modes are responsible of @xmath99 of the total electron - phonon coupling , and the rest comes from vibrations of carbon modes in the direction perpendicular to the graphite layers . in cac@xmath13
one has @xmath100 and @xmath101 @xcite . in lib
the overall contribution of b@xmath70 , li@xmath48 and li@xmath49 vibrations is less than half of that of cac@xmath13 , which means that while lib captures some of the physics of mgb@xmath2 , it does not capture the physics of graphite intercalated compounds to full extent .
this is also clear from the phonon spectrum of cac@xmath13 where the intercalant modes are at energies lower than 20 mev and one of the ca modes undergoes a marked softening with a corresponding large electron - phonon coupling ( at point @xmath102 of fig . 2 in ref .
@xcite ) . in lib , on the contrary
, the li modes are much higher in energies ( @xmath103 mev ) mev and dispersionless .
the main reason for this difference comes from the mass of li which is 5.77 times smaller than that of ca leading to frequencies which are on average 2.4 times larger .
the critical superconducting temperature is estimated using the mcmillan formula @xcite : @xmath104 \label{eq : mcmillan}\ ] ] where @xmath105 is the screened coulomb pseudopotential and @xmath106 the phonon frequencies logarithmic average .
we obtain @xmath107 mev leading to @xmath5 of approximately 10 - 15 k for @xmath108 .
this value could be further enhanced by multiband effects .
even though a theoretically - devised from scratch superconductor with @xmath109 k could be considered a success of the materials prediction methodology , the stoichiometric lib compound falls short of expectations to compete with the record - holding binary mgb@xmath2 . in this section
we investigate whether it is possible to favorably modify the electronic properties of lib and achieve higher @xmath5 by doping or applying pressure .
we pay special attention to the evolution of the @xmath4 states , since their reintroduction at @xmath42 may soften the @xmath67 mode and lead to a larger coupling . as has been pointed out previously @xcite , the bonding @xmath4 states are completely filled under ambient conditions . because the band crossing in lib at the fermi level is accidental , it may be possible to move the crossing point with pressure and increase the @xmath4-bands edos at @xmath42 .
figure [ fig : edospdelta](a ) reveals that there is indeed a rapid change in the @xmath3 and @xmath4 edos , followed by a plateau after 5 gpa .
this behavior is a reflection of two distinctly different regimes of the lib structural changes : i ) in the 0 - 5-gpa pressure range the li - li interlayer spacing quickly shrinks and the b - b bond length slightly expands so that at 5 gpa they become about 0.5 and 1.02 of their zero - pressure values , respectively ; ii ) for pressures above 5 gpa the li atoms in the bilayer start experiencing the hard - core repulsion and the compound compresses more isotropically .
the inset in fig .
[ fig : edospdelta](a ) illustrates that by reducing the li - li interlayer distance one forces the charge from the intercalant band ( completely emptied at about 6 gpa ) into the boron @xmath4 and @xmath3 states ( lowered by 0.7 ev at that pressure ) .
once the charge redistribution is complete , no appreciable changes in the edos are seen for the boron states up to at least 30 gpa .
therefore , the peculiar behavior of the nearly free electron intercalant states ( also observed in other systems @xcite ) is the only meaningful factor allowing modification of the lib boron states with pressure .
these simulations demonstrate that the compression of lib does not lead to the desired @xmath4-bands edos values comparable to those in mgb@xmath2 .
moreover , the hydrostatic pressure causes such a quick drop in the @xmath3-bands edos that this will likely negate any possible enhancement in the electron - phonon coupling due to the reintroduction of the @xmath4-states at @xmath42 .
the phonon modes are also expected to harden under pressure , further reducing the electron - phonon coupling in lib @xcite .
and @xmath4 states as a function of pressure ; b ) the same as a function of doping level : the hollow points represent charged lib unit cell in a neutralizing background ; the solid points correspond to substitutional mg and al doping of lib in the ms1 and ms2 superstructures described in the text @xcite . for comparison ,
the @xmath3 and @xmath4 edos in mgb@xmath2 are 0.098 and 0.064 states/(ev@[email protected] . ) , respectively .
the inset shows pressure - induced changes in the position of different states near the fermi level : @xmath3-boron and intercalant ( labeled @xmath111 ) states at the @xmath43 point and crossing of the @xmath4-@xmath112 states along the @xmath113 direction.,width=359 ] lib has plenty of available bonding @xmath3 states , therefore the compound should be easy to electron - dope . a quick examination of the boron edos states around the fermi level ( fig .
[ fig : dos ] or fig . 4 in ref .
@xcite ) gives an idea on what possible changes in the fermi surfaces and , eventually , in the electron - phonon coupling the doping could lead to . at small doping levels ,
when the rigid band approximation normally holds , the edos from the two dimensional @xmath3 bands should only slightly fluctuate until the states are completely filled , which happens at @xmath114n(0)@xmath115 ( e / ev ) ( 1 ev ) = 0.70 e / f.u .
= 0.35 e / boron .
the edos from the @xmath4 bands grows slowly and even at the relatively high electron - doping of 0.35 e / boron it would amount only to about a half of what is observed in mgb@xmath2 .
another way to tweak the electronic structure could be to hole - dope lib as it is done for li@xmath20bc @xcite . the known limitations of this approach are the buckling of the hexagonal layers and the eventual destabilization of the compound upon heavy li depletion @xcite .
we first simulate the electron- ( hole- ) doping using a charged cell with a neutralizing positive ( negative ) background . normally , in this approach one can safely relax the unit cell parameters and obtain valuable information about the bond length variation under small doping
. however , in the case of the electron - doped lib the repulsion between the negatively charged boron layers overcomes the weak binding between the lithium layers , causing the @xmath116-axis to undergo unphysical expansion even at small levels of doping .
therefore , we fix the @xmath116-axis at the zero - doping value and relax only the remaining three parameters .
the set of data , shown as hollow points in fig .
[ fig : edospdelta](b ) , supports our earlier conclusion that the @xmath4-band edos can not be easily increased .
note that the approximations used in this test , i.e. the fixed @xmath116-axis and the use of a neutralizing background , may influence the results to some extent .
for example , the positive electrostatic potential from ionized dopants could bring the delocalized @xmath4-states down ( in addition to the rigid band downshift ) and could potentially be an important factor in increasing the @xmath4 edos . to address these limitations we use a more realistic model of the electron - doped lib by substituting li with mg or al .
small doping levels are obtained only for large unit cells ; we use the hexagonal @xmath117-ms1 and @xmath37-ms2 supercells with 48 and 32 atoms , respectively .
replacement of one or two li atoms in these structures results in the 1/24 , 1/16 , 1/12 , and 1/8 concentrations of dopants per boron , and the level of doping is found in the assumption that they give up all their valence charge . in all the cases
the @xmath116-axis expansion in the fully relaxed unit cells does not exceed 6% .
the resulting averaged boron edos for the mg and al sets are shown in fig .
[ fig : edospdelta](b ) as solid points @xcite .
the scattered presence of dopants in the lattice should cause some dispersion of the local boron properties .
a general trend observed in our supercell calculations is that a downshift of the @xmath4 and @xmath3 states happens only for b layers in direct contact with the dopant .
typical values of the downshift that a single mg ( al ) atom induces in all eight atoms in a neighboring b layer are about 0.2 ( 0.5 ) ev .
it is not easy to isolate the importance of different factors defining the level of b doping , i.e. the simple charge transfer , the electrostatic effect discussed above and the structural changes ( expansion of the @xmath116-axis and contraction of the b - b bond ) .
however , fig .
[ fig : edospdelta](b ) demonstrates that the net effect of the substitutional doping is described reasonably well within the rigid band model . in summary ,
our tests indicate that it is rather difficult to reintroduce a significant amount of @xmath4 states at @xmath42 with hydrostatic pressure or small doping , because the band crossing in lib happens to be exactly at @xmath42 , about 2 ev lower than in mgb@xmath2 . to have a chance of substantially increasing @xmath5 , one should search for more radical ways of modifying the electronic structure of the ms metal borides .
in this work we have investigated electron and phonon properties of the recently theoretically - devised superconductor lib @xcite . by studying in details the phonon properties of this hypothetical material we have found that its critical temperature is of the order of 10 - 15 k. superconductivity is mainly of the mgb@xmath2 kind with planar @xmath46 bands strongly coupled with phonons . differently from mgb@xmath2
, lib has no fermi surface generated by @xmath4 states but an additional intercalant one .
thus , its electronic structure can be seen as a hybrid between mgb@xmath2 and cac@xmath13 .
however , the intercalant electronic states of lib are weakly coupled with the li and b@xmath70 vibrations so that the overall electron - phonon coupling is only @xmath0 . since the discovery of mgb@xmath2 , no other diborides have been found with high @xmath5 ( for a full list , see ref .
if we compare lib with the known diborides , our calculated 10 - 15 k critical temperature is not so low , although it is far from the 39 k of mgb@xmath2 .
nevertheless , the study of lib gives an important understanding .
the common belief is that the main effect for the singular and unique behavior of mgb@xmath2 is given by the presence of almost two - dimensional @xmath3 bands .
lib has even more planar @xmath3 bands and even higher @xmath3 edos at @xmath42 relative to that in mgb@xmath2 @xcite ; however , their contribution to the total electron - phonon coupling turns out to be at least 25% smaller .
this reduction can be attributed to the differences between the in - plane boron vibrations in the two borides , caused mainly by the lack of the @xmath4 carriers at @xmath42 in lib .
namely , the softening of the @xmath67 mode in lib is substantially smaller ( by about a factor of two ) than that of @xmath68 in mgb@xmath2 , which makes the former mode be noticeably harder ( @xmath98 at @xmath43 ) .
therefore , in addition to the direct loss of the @xmath4 states contribution to @xmath32 , their absence at @xmath42 in lib also has a strong indirect negative effect on the overall electron - phonon coupling .
we have investigated whether this somewhat unexpected obstacle , preventing lib to be a truly high @xmath5 superconductor , could be overcome with moderate modifications of the compound s properties .
behavior of electronic features important for the lib superconductivity has been examined as a function of small doping and pressure .
our results indicate that the @xmath4 edos increases very slowly and can not reach the desired values , at least not before the @xmath3 edos is substantially reduced .
thus , small doping and pressure are not expected to significantly improve @xmath5 in lib .
a promising direction to fix the problem would be to find a suitable lib - based ternary alloy ; this question is currently under investigation .
theoretical development of potentially important superconducting materials is a difficult task because @xmath5 critically depends on their band structure features and vibrational properties .
the challenge is even greater if one attempts to design a superconductor from scratch , since one first needs to ensure its thermodynamic stability .
the case of lib shows that it is possible to theoretically predict a compound that both i ) has a good chance to form and ii ) possesses interesting superconducting properties .
study of such promising candidates gives important insights into how to perform a more targeted search for novel superconducting materials . while we were finishing writing this paper , a preprint on the related structure ms2-lib appeared on - line @xcite .
the results of the paper are similar to ours except for some numerical details that can probably be related to the different unit cells and * k*,*q*-point samplings used@xcite . however , our conclusions concerning the possibility of increasing the @xmath5 in lib by doping are rather different , as explained in the previous section .
we acknowledge many fruitful discussions with igor mazin , francesco mauri , michele lazzeri , and roxana margine .
calculations were performed at the san diego supercomputing center and at idris supercomputing center ( project 061202 ) .
a.m. fogg , j.b .
claridge , g.r .
darling and m.j .
rosseinsky , chemical communications * 12 * , 1348 ( 2003 ) ; a.m. fogg , j. meldrum , g.r .
darling , j.b .
claridge , and m.j .
rosseinsky , j. am .
soc . * 128 * , 10043 ( 2006 ) .
for example , the long - period shifts have little effect on the ms - lib energy : ms1 and ms2 have been demonstrated to be nearly degenerate under ambient conditions , so that the synthesized lithium monoboride could be a mixture of the two crystal structure phases @xcite .
the special points are : @xmath118,t@xmath119 , u@xmath120,x@xmath121 , u@xmath122 , l@xmath123,s@xmath124 . given the rhombohedral angle @xmath87 , the parameters @xmath125 and @xmath126 are expressed as @xmath127/12 $ ] and @xmath128/12 $ ] .
we do not observe any significant softening of the in - plane boron phonon mode resulting from the 2% expansion of the boron - boron bond under small ( below 5 gpa ) pressures .
the frequencies of the soft sliding modes have been shown to nearly double at 2 gpa @xcite . in compounds such as mgb@xmath2 , @xmath72 varies substantially all over the brillouin zone .
thus a better q - point sampling than that possible in state - of - the - art _ ab - initio _
simulations is needed to obtain estimates of @xmath32 comparable with that achieved for simple superconductors such as nb .
for this reason , in systems like mgb@xmath2 , the errors in parameters such as @xmath32 or @xmath129 is probably larger than expected . | using electronic structure calculation we study the superconducting properties of the theoretically - devised superconductor ms1-lib ( lib ) .
we calculate the electron - phonon coupling ( @xmath0 ) and the phonon frequency logarithmic average ( @xmath1 mev ) and show that the lib critical temperature is in the range of 10 - 15 k , despite the frozen - phonon deformation potential being of the same order of mgb@xmath2 . as a consequence , lib captures some of the essential physics of mgb@xmath2 but ( i ) the electron - phonon coupling due to @xmath3 states is smaller and ( ii ) the precious contribution of the @xmath4 carriers to the critical temperature is lacking .
we investigate the possible change in @xmath5 that can be induced by doping and pressure and find that these conditions can not easily increase @xmath5 in lib . |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.